2,628 565 155MB
English Pages 1137 Year 2018
Table of contents :
Bioanalytics: Analytical Methods and Concepts in Biochemistry and Molecular Biology......Page 1
Table of Contents......Page 7
Preface......Page 17
Introduction: Bioanalytics - a Science in its Own Right......Page 21
Part I: Protein Analytics......Page 27
1.1 Properties of Proteins......Page 29
1.2 Protein Localization and Purification Strategy......Page 32
1.3 Homogenization and Cell Disruption......Page 33
1.4 Precipitation......Page 35
1.5 Centrifugation......Page 37
1.5.2 Centrifugation Techniques......Page 38
1.6 Removal of Salts and Hydrophilic Contaminants......Page 41
1.7 Concentration......Page 43
1.8.1 Properties of Detergents......Page 44
1.8.2 Removal of Detergents......Page 46
Further Reading......Page 48
Chapter 2: Protein determination......Page 49
2.1 Quantitative Determination by Staining Tests......Page 51
2.1.2 Lowry Assay......Page 52
2.1.3 Bicinchoninic Acid Assay (BCA Assay)......Page 53
2.2 Spectroscopic Methods......Page 54
2.2.1 Measurements in the UV Range......Page 55
2.3 Radioactive Labeling of Peptides and Proteins......Page 57
Further Reading......Page 59
3.1 The Driving Force behind Chemical Reactions......Page 61
3.2 Rate of Chemical Reactions......Page 62
3.4 Enzymes as Catalysts......Page 63
3.6 Michaelis-Menten Theory......Page 64
3.7 Determination of Km and Vmax......Page 65
3.8.1 Competitive Inhibitors......Page 66
3.9 Test System Set-up......Page 67
3.9.3 Detection System......Page 68
3.9.6 Selecting the Buffer Substance and the Ionic Strength......Page 69
3.9.8 Substrate Concentration......Page 70
Further Reading......Page 71
Chapter 4: Microcalorimetry......Page 73
4.1 Differential Scanning Calorimetry (DSC)......Page 74
4.2.1 Ligand Binding to Proteins......Page 80
4.2.2 Binding of Molecules to Membranes: Insertion and Peripheral Binding......Page 84
4.3 Pressure Perturbation Calorimetry (PPC)......Page 87
Further Reading......Page 88
5.1.1 Antibodies and Immune Defense......Page 89
5.1.3 Properties of Antibodies......Page 90
5.1.4 Functional Structure of IgG......Page 92
5.1.5 Antigen Interaction at the Combining Site......Page 93
5.1.6 Handling of Antibodies......Page 94
5.2 Antigens......Page 95
5.3 Antigen-Antibody Reaction......Page 97
5.3.1 Immunoagglutination......Page 98
5.3.2 Immunoprecipitation......Page 99
5.3.3 Immune Binding......Page 110
5.4 Complement Fixation......Page 120
5.5 Methods in Cellular Immunology......Page 121
5.6 Alteration of Biological Functions......Page 123
5.7.1 Types of Antibodies......Page 124
5.7.2 New Antibody Techniques (Antibody Engineering)......Page 125
5.7.3 Optimized Monoclonal Antibody Constructs with Effector Functions for Therapeutic Application......Page 128
Further Reading......Page 132
Chapter 6: Chemical Modification of Proteins and Protein Complexes......Page 133
6.1 Chemical Modification of Protein Functional Groups......Page 134
6.2.1 Investigation with Naturally Occurring Proteins......Page 142
6.2.2 Investigation of Recombinant and Mutated Proteins......Page 146
6.3.2 Photoaffinity Labeling......Page 147
Further Reading......Page 155
Chapter 7: Spectroscopy......Page 157
7.1.1 Physical Principles of Optical Spectroscopic Techniques......Page 158
7.1.2 Interaction of Light with Matter......Page 159
7.1.3 Absorption Measurement and the Lambert-Beer Law......Page 166
7.1.4 Photometer......Page 169
7.1.5 Time-Resolved Spectroscopy......Page 170
7.2.1 Basic Principles......Page 172
7.2.2 Chromoproteins......Page 173
7.3.1 Basic Principles of Fluorescence Spectroscopy......Page 180
7.3.2 Fluorescence: Emission and Action Spectra......Page 182
7.3.3 Fluorescence Studies using Intrinsic and Extrinsic Probes......Page 183
7.3.4 Green Fluorescent Protein (GFP) as a Unique Fluorescent Probe......Page 184
7.3.5 Quantum Dots as Fluorescence Labels......Page 185
7.3.7 Förster Resonance Energy Transfer (FRET)......Page 186
7.3.8 Frequent Mistakes in Fluorescence Spectroscopy: ``The Seven Sins of Fluorescence Measurements´´......Page 187
7.4.1 Basic Principles of IR Spectroscopy......Page 189
7.4.2 Molecular Vibrations......Page 190
7.4.3 Technical aspects of Infrared Spectroscopy......Page 191
7.4.4 Infrared Spectra of Proteins......Page 194
7.5.1 Basic Principles of Raman Spectroscopy......Page 197
7.5.2 Raman Experiments......Page 198
7.5.3 Resonance Raman Spectroscopy......Page 199
7.6 Single Molecule Spectroscopy......Page 200
7.7.1 Linear Dichroism......Page 201
7.7.2 Optical Rotation Dispersion and Circular Dichroism......Page 204
Further Reading......Page 206
8.1 Steps on the Road to Microscopy - from Simple Lenses to High Resolution Microscopes......Page 207
8.2 Modern Applications......Page 208
8.3 Basic Physical Principles......Page 209
8.4 Detection Methods......Page 215
8.5 Sample Preparation......Page 221
8.6 Special Fluorescence Microscopic Analysis......Page 223
Further Reading......Page 231
9.1 Proteolytic Enzymes......Page 233
9.2 Strategy......Page 234
9.4 Cleavage of Disulfide Bonds and Alkylation......Page 235
9.5.1 Proteases......Page 236
9.5.2 Conditions for Proteolysis......Page 241
9.6 Chemical Fragmentation......Page 242
9.7 Summary......Page 243
Further Reading......Page 244
10.1 Instrumentation......Page 245
10.2 Fundamental Terms and Concepts in Chromatography......Page 246
10.3 Biophysical Properties of Peptides and Proteins......Page 250
10.4 Chromatographic Separation Modes for Peptides and Proteins......Page 251
10.4.2 High-Performance Reversed-Phase Chromatography (HP-RPC)......Page 253
10.4.3 High-Performance Normal-Phase Chromatography (HP-NPC)......Page 254
10.4.4 High-Performance Hydrophilic Interaction Chromatography (HP-HILIC)......Page 255
10.4.6 High-Performance Hydrophobic Interaction Chromatography (HP-HIC)......Page 256
10.4.7 High-Performance Ion Exchange Chromatography (HP-IEX)......Page 258
10.4.8 High-Performance Affinity Chromatography (HP-AC)......Page 259
10.5.1 Development of an Analytical Method......Page 260
10.5.2 Scaling up to Preparative Chromatography......Page 262
10.5.3 Fractionation......Page 263
10.6.1 Purification of Peptides and Proteins by MD-HPLC Methods......Page 264
10.6.3 Strategies for MD-HPLC Methods......Page 265
10.6.4 Design of an Effective MD-HPLC Scheme......Page 266
Further Reading......Page 268
Chapter 11: Electrophoretic Techniques......Page 269
11.1 Historical Review......Page 270
11.2 Theoretical Fundamentals......Page 271
11.3 Equipment and Procedures of Gel Electrophoreses......Page 274
11.3.1 Sample Preparation......Page 275
11.3.2 Gel Media for Electrophoresis......Page 276
11.3.3 Detection and Quantification of the Separated Proteins......Page 277
11.3.4 Zone Electrophoresis......Page 279
11.3.5 Porosity Gradient Gels......Page 280
11.3.7 Disc Electrophoresis......Page 281
11.3.9 SDS Polyacrylamide Gel Electrophoresis......Page 283
11.3.10 Cationic Detergent Electrophoresis......Page 284
11.3.12 Isoelectric Focusing......Page 285
11.4.1 Electroelution from Gels......Page 289
11.4.2 Preparative Zone Electrophoresis......Page 290
11.4.3 Preparative Isoelectric Focusing......Page 291
11.5 Free Flow Electrophoresis......Page 292
11.6 High-Resolution Two-Dimensional Electrophoresis......Page 293
11.6.2 Prefractionation......Page 294
11.6.3 First Dimension: IEF in IPG Strips......Page 295
11.6.6 Difference Gel Electrophoresis (DIGE)......Page 296
11.7.1 Blot Systems......Page 298
Further Reading......Page 299
12.1 Historical Overview......Page 301
12.2 Capillary Electrophoresis Setup......Page 302
12.3.1 Sample Injection......Page 303
12.3.2 The Engine: Electroosmotic Flow (EOF)......Page 304
12.3.4 Detection Methods......Page 305
12.4.1 Capillary Zone Electrophoresis (CZE)......Page 307
12.4.2 Affinity Capillary Electrophoresis (ACE)......Page 311
12.4.3 Micellar Electrokinetic Chromatography (MEKC)......Page 312
12.4.4 Capillary Electrochromatography (CEC)......Page 314
12.4.5 Chiral Separations......Page 315
12.4.6 Capillary Gel Electrophoresis (CGE)......Page 316
12.4.7 Capillary Isoelectric Focusing (CIEF)......Page 317
12.4.8 Isotachophoresis (ITP)......Page 319
12.5.2 Online Sample Concentration......Page 321
12.5.3 Fractionation......Page 322
12.6 Outlook......Page 323
Further Reading......Page 325
Chapter 13: Amino Acid Analysis......Page 327
13.1.1 Acidic Hydrolysis......Page 328
13.3.1 Post-Column Derivatization......Page 329
13.3.2 Pre-column Derivatization......Page 331
13.4 Amino Acid Analysis using Mass Spectrometry......Page 335
13.5 Summary......Page 336
Further Reading......Page 337
Chapter 14: Protein Sequence Analysis......Page 339
14.1.1 Reactions of the Edman Degradation......Page 341
14.1.2 Identification of the Amino Acids......Page 342
14.1.3 Quality of Edman Degradation: the Repetitive Yield......Page 343
14.1.4 Instrumentation......Page 345
14.1.5 Problems of Amino Acid Sequence Analysis......Page 348
14.2.1 Chemical Degradation Methods......Page 351
14.2.3 Degradation of Polypeptides with Carboxypeptidases......Page 353
Further Reading......Page 354
Chapter 15: Mass Spectrometry......Page 355
15.1.1 Matrix Assisted Laser Desorption Ionization Mass Spectrometry (MALDI-MS)......Page 356
15.1.2 Electrospray Ionization (ESI)......Page 361
15.2 Mass Analyzer......Page 367
15.2.1 Time-of-Flight Analyzers (TOF)......Page 369
15.2.2 Quadrupole Analyzer......Page 371
15.2.3 Electric Ion Traps......Page 374
15.2.4 Magnetic Ion Trap......Page 375
15.2.5 Orbital Ion Trap......Page 376
15.2.6 Hybrid Instruments......Page 377
15.3 Ion Detectors......Page 381
15.3.1 Secondary Electron Multiplier (SEV)......Page 382
15.4.1 Collision Induced Dissociation (CID)......Page 383
15.4.2 Prompt and Metastable Decay (ISD, PSD)......Page 384
15.4.4 Generation of Free Radicals (ECD, HECD, ETD)......Page 386
15.5.2 Influence of Isotopy......Page 388
15.5.4 Determination of the Number of Charges......Page 391
15.5.7 Problems......Page 392
15.6.1 Identification......Page 394
15.6.3 Structure Elucidation......Page 395
15.7.1 LC-MS......Page 401
15.7.2 LC-MS/MS......Page 402
15.8 Quantification......Page 404
Further Reading......Page 405
16.1.1 Principle of Two-Hybrid Systems......Page 407
16.1.3 Construction of Bait and Prey Proteins......Page 408
16.1.5 AD Fusion Proteins and cDNA Libraries......Page 411
16.1.6 Carrying out a Y2H Screen......Page 412
16.1.7 Other Modifications and Extensions of the Two-Hybrid-Technology......Page 417
16.1.8 Biochemical and Functional Analysis of Interactions......Page 419
16.2 TAP-Tagging and Purification of Protein Complexes......Page 420
16.3 Analyzing Interactions In Vitro: GST-Pulldown......Page 423
16.4 Co-immunoprecipitation......Page 424
16.5 Far-Western......Page 425
16.6 Surface Plasmon Resonance Spectroscopy......Page 426
16.7.1 Introduction......Page 428
16.7.3 Methods of FRET Measurements......Page 429
16.7.4 Fluorescent Probes for FRET......Page 432
16.7.5 Alternative Tools for Probing Protein-Protein Interactions: LINC and STET......Page 434
16.8 Analytical Ultracentrifugation......Page 435
16.8.1 Principles of Instrumentation......Page 436
16.8.2 Basics of Centrifugation......Page 437
16.8.3 Sedimentation Velocity Experiments......Page 438
16.8.4 Sedimentation-Diffusion Equilibrium Experiments......Page 441
Further Reading......Page 442
Chapter 17: Biosensors......Page 445
17.2.1 Concept of Biosensors......Page 446
17.2.2 Construction and Function of Biosensors......Page 447
17.2.3 Cell Sensors......Page 451
17.2.4 Immunosensors......Page 452
17.3 Biomimetic Sensors......Page 453
17.4 From Glucose Enzyme Electrodes to Electronic DNA Biochips......Page 454
Further Reading......Page 455
Part II: 3D Structure Determination......Page 457
18.1 NMR Spectroscopy of Biomolecules......Page 459
18.1.1 Theory of NMR Spectroscopy......Page 460
18.1.2 One-Dimensional NMR Spectroscopy......Page 464
18.1.3 Two-Dimensional NMR Spectroscopy......Page 469
18.1.4 Three-Dimensional NMR Spectroscopy......Page 475
18.1.5 Resonance Assignment......Page 478
18.1.6 Protein Structure Determination......Page 483
18.1.7 Protein Structures and more - an Overview......Page 488
18.2 EPR Spectroscopy of Biological Systems......Page 492
18.2.1 Basics of EPR Spectroscopy......Page 493
18.2.2 cw- EPR Spectroscopy......Page 494
18.2.4 Electron Spin Nuclear Spin Coupling (Hyperfine Coupling)......Page 495
18.2.5 g and Hyperfine Anisotropy......Page 496
18.2.6 Electron Spin-Electron Spin Coupling......Page 498
18.2.7 Pulsed EPR Experiments......Page 499
18.2.8 Further Examples of EPR Applications......Page 505
18.2.10 Comparison EPR/NMR......Page 507
Further Reading......Page 508
Chapter 19: Electron Microscopy......Page 511
19.1 Transmission Electron Microscopy - Instrumentation......Page 513
19.2.1 Native Samples in Ice......Page 514
19.2.2 Negative Staining......Page 516
19.2.3 Metal Coating by Evaporation......Page 517
19.3.1 Resolution of a Transmission Electron Microscope......Page 518
19.3.2 Interactions of the Electron Beam with the Object......Page 519
19.3.4 Electron Microscopy with a Phase Plate......Page 521
19.3.5 Imaging Procedure for Frozen-Hydrated Specimens......Page 522
19.3.6 Recording Images - Cameras and the Impact of Electrons......Page 523
19.4.1 Pixel Size......Page 524
19.4.2 Fourier Transformation......Page 525
19.4.3 Analysis of the Contrast Transfer Function and Object Features......Page 527
19.4.4 Improving the Signal-to-Noise Ratio......Page 530
19.4.5 Principal Component Analysis and Classification......Page 532
19.5 Three-Dimensional Electron Microscopy......Page 534
19.5.1 Three-Dimensional Reconstruction of Single Particles......Page 535
19.5.2 Three-Dimensional Reconstruction of Regularly Arrayed Macromolecular Complexes......Page 537
19.5.3 Electron Tomography of Individual Objects......Page 538
19.6.1 Hybrid Approach: Combination of EM and X-Ray Data......Page 540
19.6.3 Identifying Protein Complexes in Cellular Tomograms......Page 541
19.7 Perspectives of Electron Microscopy......Page 542
Further Reading......Page 543
20.1 Introduction......Page 545
20.2 Principle of the Atomic Force Microscope......Page 546
20.3 Interaction between Tip and Sample......Page 547
20.5 Mapping Biological Macromolecules......Page 548
20.6 Force Spectroscopy of Single Molecules......Page 550
20.7 Detection of Functional States and Interactions of Individual Proteins......Page 552
Further Reading......Page 553
Chapter 21: X-Ray Structure Analysis......Page 555
21.1 X-Ray Crystallography......Page 556
21.1.1 Crystallization......Page 557
21.1.2 Crystals and X-Ray Diffraction......Page 559
21.1.3 The Phase Problem......Page 564
21.1.4 Model Building and Structure Refinement......Page 568
21.2 Small Angle X-Ray Scattering (SAXS)......Page 569
21.2.1 Machine Setup......Page 570
21.2.2 Theory......Page 571
21.2.3 Data Analysis......Page 573
21.3.1 Machine Setup and Theory......Page 575
Acknowledgement......Page 576
Further Reading......Page 577
Part III: Peptides, Carbohydrates, and Lipids......Page 579
22.1 Concept of Peptide Synthesis......Page 581
22.2 Purity of Synthetic Peptides......Page 586
22.3 Characterization and Identity of Synthetic Peptides......Page 588
22.4 Characterization of the Structure of Synthetic Peptides......Page 590
22.5 Analytics of Peptide Libraries......Page 593
Further Reading......Page 595
Chapter 23: Carbohydrate Analysis......Page 597
23.1.1 The Series of d-Sugars......Page 598
23.1.2 Stereochemistry of d-Glucose......Page 599
23.1.5 The Glycosidic Bond......Page 600
23.2 Protein Glycosylation......Page 605
23.2.2 Structure of the O-Glycans......Page 606
23.3 Analysis of Protein Glycosylation......Page 607
23.3.1 Analysis on the Basis of the Intact Glycoprotein......Page 608
23.3.2 Mass Spectrometric Analysis on the Basis of Glycopeptides......Page 614
23.3.3 Release and Isolation of the N-Glycan Pool......Page 616
23.3.4 Analysis of Individual N-Glycans......Page 625
23.4 Genome, Proteome, Glycome......Page 636
23.5 Final Considerations......Page 637
Further Reading......Page 638
24.1 Structure and Classification of Lipids......Page 639
24.2 Extraction of Lipids from Biological Sources......Page 641
24.2.2 Solid Phase Extraction......Page 642
24.3.1 Chromatographic Methods......Page 644
24.3.3 Immunoassays......Page 648
24.3.5 Combining Different Analytical Systems......Page 649
24.4.1 Whole Lipid Extracts......Page 652
24.4.2 Fatty Acids......Page 653
24.4.3 Nonpolar Neutral Lipids......Page 654
24.4.4 Polar Ester Lipids......Page 656
24.4.5 Lipid Hormones and Intracellular Signaling Molecules......Page 659
24.5 Lipid Vitamins......Page 664
24.6 Lipidome Analysis......Page 666
24.7 Perspectives......Page 668
Further Reading......Page 670
25.1.1 Phosphorylation......Page 671
25.1.2 Acetylation......Page 672
25.2 Strategies for the Analysis of Phosphorylated and Acetylated Proteins and Peptides......Page 673
25.3 Separation and Enrichment of Phosphorylated and Acetylated Proteins and Peptides......Page 675
25.4.1 Detection by Enzymatic, Radioactive, Immunochemical, and Fluorescence Based Methods......Page 677
25.5 Localization and Identification of Post-translationally Modified Amino Acids......Page 679
25.5.2 Localization of Phosphorylated and Acetylated Amino Acids by Tandem Mass Spectrometry......Page 680
25.6 Quantitative Analysis of Post-translational Modifications......Page 685
Further Reading......Page 687
Part IV: Nucleic Acid Analytics......Page 689
26.1.1 Phenolic Purification of Nucleic Acids......Page 691
26.1.2 Gel Filtration......Page 692
26.1.3 Precipitation of Nucleic Acids with Ethanol......Page 693
26.1.4 Determination of the Nucleic Acid Concentration......Page 694
26.2 Isolation of Genomic DNA......Page 695
26.3.1 Isolation of Plasmid DNA from Bacteria......Page 696
26.4.1 Isolation of Phage DNA......Page 700
26.4.2 Isolation of Eukaryotic Viral DNA......Page 701
26.6 Isolation of RNA......Page 702
26.6.1 Isolation of Cytoplasmic RNA......Page 703
26.6.2 Isolation of Poly(A) RNA......Page 704
26.7 Isolation of Nucleic Acids using Magnetic Particles......Page 705
Further Reading......Page 706
27.1.1 Principle of Restriction Analyses......Page 707
27.1.3 Restriction Enzymes......Page 708
27.1.4 In Vitro Restriction and Applications......Page 711
27.2 Electrophoresis......Page 716
27.2.1 Gel Electrophoresis of DNA......Page 717
27.2.2 Gel Electrophoresis of RNA......Page 723
27.2.3 Pulsed-Field Gel Electrophoresis (PFGE)......Page 724
27.2.4 Two-Dimensional Gel Electrophoresis......Page 726
27.2.5 Capillary Gel Electrophoresis......Page 727
27.3.1 Fluorescent Dyes......Page 728
27.4.2 Choice of Membrane......Page 730
27.4.3 Southern Blotting......Page 731
27.4.4 Northern Blotting......Page 732
27.4.6 Colony and Plaque Hybridization......Page 733
27.5.3 Purification using Electroelution......Page 734
27.6.1 Principles of the Synthesis of Oligonucleotides......Page 735
27.6.2 Investigation of the Purity and Characterization of Oligonucleotides......Page 737
27.6.3 Mass Spectrometric Investigation of Oligonucleotides......Page 738
27.6.4 IP-RP-HPLC-MS Investigation of a Phosphorothioate Oligonucleotide......Page 740
Further Reading......Page 743
Chapter 28: Techniques for the Hybridization and Detection of Nucleic Acids......Page 745
28.1 Basic Principles of Hybridization......Page 746
28.1.1 Principle and Practice of Hybridization......Page 747
28.1.2 Specificity of the Hybridization and Stringency......Page 748
28.1.3 Hybridization Methods......Page 749
28.2 Probes for Nucleic Acid Analysis......Page 755
28.2.1 DNA Probes......Page 756
28.2.2 RNA Probes......Page 757
28.2.4 LNA Probes......Page 758
28.3.1 Labeling Positions......Page 759
28.3.2 Enzymatic Labeling......Page 761
28.3.4 Chemical Labeling......Page 763
28.4.2 Radioactive Systems......Page 764
28.4.3 Non-radioactive Systems......Page 765
28.5 Amplification Systems......Page 776
28.5.2 Target-Specific Signal Amplification......Page 777
28.5.3 Signal Amplification......Page 778
Further Reading......Page 779
29.1 Possibilities of PCR......Page 781
29.2.1 Instruments......Page 782
29.2.2 Amplification of DNA......Page 784
29.2.3 Amplification of RNA (RT-PCR)......Page 787
29.2.5 Quantitative PCR......Page 789
29.3.1 Nested PCR......Page 792
29.3.4 Multiplex PCR......Page 793
29.3.7 Homogeneous PCR Detection Procedures......Page 794
29.3.10 Other Approaches......Page 795
29.4.1 Avoiding Contamination......Page 796
29.4.2 Decontamination......Page 797
29.5.1 Detection of Infectious Diseases......Page 798
29.5.2 Detection of Genetic Defects......Page 799
29.5.3 The Human Genome Project......Page 802
29.6.3 Helicase-Dependent Amplification (HDA)......Page 803
29.6.4 Ligase Chain Reaction (LCR)......Page 805
29.6.5 Qβ Amplification......Page 806
Further Reading......Page 808
Chapter 30: DNA Sequencing......Page 811
30.1 Gel-Supported DNA Sequencing Methods......Page 812
30.1.1 Sequencing according to Sanger: The Dideoxy Method......Page 815
30.1.2 Labeling Techniques and Methods of Verification......Page 822
30.1.3 Chemical Cleavage according to Maxam and Gilbert......Page 826
30.2 Gel-Free DNA Sequencing Methods - The Next Generation......Page 832
30.2.1 Sequencing by Synthesis......Page 833
30.2.2 Single Molecule Sequencing......Page 839
Further Reading......Page 841
Chapter 31: Analysis of Epigenetic Modifications......Page 843
31.1 Overview of the Methods to Detect DNA-Modifications......Page 844
31.2.1 Amplification and Sequencing of Bisulfite-Treated DNA......Page 845
31.2.2 Restriction Analysis after Bisulfite PCR......Page 846
31.2.3 Methylation Specific PCR......Page 848
31.3 DNA Analysis with Methylation Specific Restriction Enzymes......Page 849
31.4 Methylation Analysis by Methylcytosine-Binding Proteins......Page 851
31.5 Methylation Analysis by Methylcytosine-Specific Antibodies......Page 852
31.6 Methylation Analysis by DNA Hydrolysis and Nearest Neighbor-Assays......Page 853
31.8 Chromosome Interaction Analyses......Page 854
Further Reading......Page 855
32.1.1 Basic Features for DNA-Protein Recognition: Double-Helical Structures......Page 857
32.1.2 DNA Curvature......Page 858
32.1.3 DNA Topology......Page 859
32.2 DNA-Binding Motifs......Page 861
32.3.2 Gel Electrophoresis......Page 862
32.3.3 Determination of Dissociation Constants......Page 865
32.3.4 Analysis of DNA-Protein Complex Dynamics......Page 866
32.4 DNA Footprint Analysis......Page 867
32.4.2 Primer Extension Reaction for DNA Analysis......Page 869
32.4.3 Hydrolysis Methods......Page 870
32.4.4 Chemical Reagents for the Modification of DNA-Protein Complexes......Page 872
32.4.5 Interference Conditions......Page 874
32.4.6 Chemical Nucleases......Page 875
32.4.7 Genome-Wide DNA-Protein Interactions......Page 876
32.5.2 Fluorophores and Labeling Procedures......Page 877
32.5.3 Fluorescence Resonance Energy Transfer (FRET)......Page 878
32.5.5 Surface Plasmon Resonance (SPR)......Page 879
32.5.6 Scanning Force Microscopy (SFM)......Page 880
32.5.7 Optical Tweezers......Page 881
32.6.1 Functional Diversity of RNA......Page 882
32.6.3 Dynamics of RNA-Protein Interactions......Page 883
32.7 Characteristic RNA-Binding Motifs......Page 885
32.8 Special Methods for the Analysis of RNA-Protein Complexes......Page 886
32.8.2 Labeling Methods......Page 887
32.8.4 Customary RNases......Page 888
32.8.5 Chemcal Modification of RNA-Protein Complexes......Page 889
32.8.6 Chemical Crosslinking......Page 892
32.8.8 Genome-Wide Identification of Transcription Start Sites (TSS)......Page 893
32.9.1 Tri-hybrid Method......Page 894
32.9.2 Aptamers and the Selex Procedure......Page 895
Further Reading......Page 896
Part V: Functional and Systems Analytics......Page 899
33.1 Sequence Analysis and Bioinformatics......Page 901
33.2 Sequence: An Abstraction for Biomolecules......Page 902
33.3 Internet Databases and Services......Page 903
33.3.1 Sequence Retrieval from Public Databases......Page 904
33.3.2 Data Contents and File Format......Page 905
33.4.1 EMBOSS......Page 907
33.6 Sequence Patterns......Page 908
33.6.1 Transcription Factor Binding Sites......Page 910
33.6.2 Identification of Coding Regions......Page 911
33.6.3 Protein Localization......Page 912
33.7.1 Identity, Similarity, Homology......Page 913
33.7.2 Optimal Sequence Alignment......Page 914
33.7.4 Profile-Based Sensitive Database Search: PSI-BLAST......Page 916
33.8 Multiple Alignment and Consensus Sequences......Page 917
33.9 Structure Prediction......Page 918
33.10 Outlook......Page 919
34.1.1 Overview......Page 921
34.1.2 Nuclease S1 Analysis of RNA......Page 922
34.1.3 Ribonuclease-Protection Assay (RPA)......Page 924
34.1.4 Primer Extension Assay......Page 927
34.1.5 Northern Blot and Dot- and Slot-Blot......Page 928
34.1.6 Reverse Transcription Polymerase Chain Reaction (RT-PCR and RT-qPCR)......Page 930
34.2.1 Nuclear-run-on Assay......Page 931
34.2.2 Labeling of Nascent RNA with 5-Fluoro-uridine (FUrd)......Page 932
34.3.1 Components of an In Vitro Transcription Assay......Page 933
34.3.3 Template DNA and Detection of In Vitro Transcripts......Page 934
34.4.1 Vectors for Analysis of Gene-Regulatory cis-Elements......Page 937
34.4.2 Transfer of DNA into Mammalian Cells......Page 938
34.4.3 Analysis of Reporter Gene Expression......Page 940
Further Reading......Page 942
35.1.1 Labeling Strategy......Page 943
35.1.3 Labeling of DNA Probes......Page 944
35.1.4 In Situ Hybridization......Page 945
35.2.1 FISH Analysis of Genomic DNA......Page 946
35.2.2 Comparative Genomic Hybridization (CGH)......Page 947
Further Reading......Page 950
36.1.1 Recombination......Page 951
36.1.2 Genetic Markers......Page 953
36.1.3 Linkage Analysis - the Generation of Genetic Maps......Page 955
36.1.4 Genetic Map of the Human Genome......Page 957
36.2.1 Restriction Mapping of Whole Genomes......Page 958
36.2.2 Mapping of Recombinant Clones......Page 960
36.2.3 Generation of a Physical Map......Page 961
36.2.4 Identification and Isolation of Genes......Page 963
36.2.5 Transcription Maps of the Human Genome......Page 965
36.3 Integration of Genome Maps......Page 966
Further Reading......Page 968
Chapter 37: DNA-Microarray Technology......Page 971
37.1.1 Transcriptome Analysis......Page 972
37.1.3 RNA Structure and Functionality......Page 973
37.2.2 Methylation Studies......Page 974
37.2.3 DNA Sequencing......Page 975
37.2.5 Protein-DNA Interactions......Page 977
37.3.1 DNA Synthesis......Page 978
37.3.3 On-Chip Protein Expression......Page 979
37.4.1 Barcode Identification......Page 980
37.4.2 A Universal Microarray Platform......Page 981
37.5.2 Beyond Nucleic Acids......Page 982
Further Reading......Page 983
Chapter 38: The Use of Oligonucleotides as Tools in Cell Biology......Page 985
38.1.1 Mechanisms of Antisense Oligonucleotides......Page 986
38.1.2 Triplex-Forming Oligonucleotides......Page 987
38.1.3 Modifications of Oligonucleotides to Decrease their Susceptibility to Nucleases......Page 988
38.1.5 Antisense Oligonucleotides as Therapeutics......Page 990
38.2.1 Discovery and Classification of Ribozymes......Page 991
38.2.2 Use of Ribozymes......Page 992
38.3.1 Basics of RNA Interference......Page 993
38.3.2 RNA Interference Mediated by Expression Vectors......Page 994
38.3.3 Uses of RNA Interference......Page 995
38.3.4 microRNAs......Page 996
38.4.1 Selection of Aptamers......Page 997
38.4.2 Uses of Aptamers......Page 999
38.5 Genome Editing with CRISPR/Cas9......Page 1000
38.6 Outlook......Page 1001
Further Reading......Page 1002
39.1 General Aspects in Proteome Analysis......Page 1003
39.2 Definition of Starting Conditions and Project Planning......Page 1005
39.3 Sample Preparation for Proteome Analysis......Page 1006
39.4.1 Two-Dimensional-Gel-Based Proteomics......Page 1008
39.4.3 Top-Down Proteomics using Isotope Labels......Page 1012
39.4.5 Concepts in Intact Protein Mass Spectrometry......Page 1013
39.5.2 Bottom-Up Proteomics......Page 1024
39.5.4 Bottom-Up Proteomic Strategies......Page 1026
39.5.5 Peptide Quantification......Page 1027
39.5.6 Data Dependent Analysis (DDA)......Page 1028
39.5.7 Selected Reaction Monitoring......Page 1029
39.5.8 SWATH-MS......Page 1036
39.5.10 Extensions......Page 1038
39.6.1 Stable Isotope Label in Top-Down Proteomics......Page 1039
39.6.2 Stable Isotope Labeling in Bottom-Up Proteomics......Page 1045
Further Reading......Page 1047
Chapter 40: Metabolomics and Peptidomics......Page 1049
40.1 Systems Biology and Metabolomics......Page 1051
40.2 Technological Platforms for Metabolomics......Page 1052
40.3 Metabolomic Profiling......Page 1053
40.4 Peptidomics......Page 1054
40.5 Metabolomics - Knowledge Mining......Page 1055
40.6 Data Mining......Page 1056
Further Reading......Page 1058
41.1 Protein Microarrays......Page 1059
41.1.1 Sensitivity Increase through Miniaturization - Ambient Analyte Assay......Page 1060
41.1.2 From DNA to Protein Microarrays......Page 1061
41.1.3 Application of Protein Microarrays......Page 1063
Further Reading......Page 1065
42.1 Chemical Biology - Innovative Chemical Approaches to Study Biological Phenomena......Page 1067
42.2 Chemical Genetics - Small Organic Molecules for the Modulation of Protein Function......Page 1069
42.2.1 Study of Protein Functions with Small Organic Molecules......Page 1070
42.2.2 Forward and Reverse Chemical Genetics......Page 1072
42.2.3 The Bump-and-Hole Approach of Chemical Genetics......Page 1073
42.2.4 Identification of Kinase Substrates with ASKA Technology......Page 1076
42.2.5 Switching Biological Systems on and off with Small Organic Molecules......Page 1077
42.3.1 Analysis of Lipid-Modified Proteins......Page 1078
42.3.3 Conditional Protein Splicing......Page 1080
Further Reading......Page 1081
43.1 Antibody Based Toponome Analysis using Imaging Cycler Microscopy (ICM)......Page 1083
43.1.1 Concept of the Protein Toponome......Page 1084
43.1.2 Imaging Cycler Robots: Fundament of a Toponome Reading Technology......Page 1085
Acknowledgements......Page 1089
43.2.2 Mass Spectrometric Pixel Images......Page 1090
43.2.3 Achievable Spatial Resolution......Page 1091
43.2.5 Lateral Resolution and Analytical Limit of Detection......Page 1093
43.2.7 Accurate MALDI Mass Spectrometry Imaging......Page 1094
43.2.8 Identification and Characterization of Analytes......Page 1095
Further Reading......Page 1096
Appendix 1: Amino Acids and Posttranslational Modifications......Page 1099
Appendix 2: Symbols and Abbreviations......Page 1101
Appendix 3: Standard Amino Acids (three and one letter code)......Page 1107
Appendix 4: Nucleic Acid Bases......Page 1109
Index......Page 1111
End User License Agreement......Page 1137
Edited by Friedrich Lottspeich and Joachim W. Engels
Bioanalytics
Bioanalytics
Edited by Friedrich Lottspeich and Joachim Engels
Bioanalytics Analytical Methods and Concepts in Biochemistry and Molecular Biology
Editors Dr. phil Dr. med. habil. Friedrich Lottspeich retired from MPI for Biochemistry Peter-Dörfler-Straße 4a 82131 Stockdorf Germany
Prof. Dr. Joachim Engels Goethe Universität OCCB FB14 Max-von-Laue-Straße 7 60438 Frankfurt Germany
Cover credit: Background picture - fotolia_science photo; Circles from left to right: 1st circle - fotolia_T-flex; 2nd circle - Adam Design; 3rd circle fotolia_fibroblasts; 4th circle - The Picture was kindly provided by Dr. Ficner. All books published by Wiley-VCH are carefully produced. Nevertheless, authors, editors, and publisher do not warrant the information contained in these books, including this book, to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate. Library of Congress Card No.: applied for British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at . 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Boschstr. 12, 69469 Weinheim, Germany All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form – by photoprinting, microfilm, or any other means – nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Print ISBN: 978-3-527-33919-8 ePDF ISBN: 978-3-527-69444-0 ePub ISBN: 978-3-527-69446-4 Mobi ISBN: 978-3-527-69447-1 Cover Design Adam Design Typesetting Thomson Digital Printing and Binding Printed on acid-free paper
Table of Contents
Preface
XV
Introduction
XIX
Part I Protein Analytics
1
1 1.1 1.2 1.3 1.4 1.5 1.5.1 1.5.2 1.6 1.7 1.8 1.8.1 1.8.2 1.9
Protein Purification Properties of Proteins Protein Localization and Purification Strategy Homogenization and Cell Disruption Precipitation Centrifugation Basic Principles Centrifugation Techniques Removal of Salts and Hydrophilic Contaminants Concentration Detergents and their Removal Properties of Detergents Removal of Detergents Sample Preparation for Proteome Analysis Further Reading
3 3 6 7 9 11 12 12 15 17 18 18 20 22 22
2 2.1 2.1.1 2.1.2 2.1.3 2.1.4 2.2 2.2.1 2.2.2 2.3 2.3.1
Protein determination Quantitative Determination by Staining Tests Biuret Assay Lowry Assay Bicinchoninic Acid Assay (BCA Assay) Bradford Assay Spectroscopic Methods Measurements in the UV Range Fluorescence Method Radioactive Labeling of Peptides and Proteins Iodinations Further Reading
23 25 26 26 27 28 28 29 31 31 33 33
3 3.1 3.2
Enzyme Activity Testing The Driving Force behind Chemical Reactions Rate of Chemical Reactions
35 35 36
3.3 3.4 3.5 3.6 3.7 3.8 3.8.1 3.8.2 3.9 3.9.1 3.9.2 3.9.3 3.9.4 3.9.5 3.9.6 3.9.7 3.9.8 3.9.9
4 4.1 4.2 4.2.1 4.2.2 4.3
5 5.1 5.1.1 5.1.2 5.1.3 5.1.4 5.1.5 5.1.6 5.2
Catalysts Enzymes as Catalysts Rate of Enzyme-Controlled Reactions Michaelis–Menten Theory Determination of K m and V max Inhibitors Competitive Inhibitors Non-competitive Inhibitors Test System Set-up Analysis of the Physiological Function Selecting the Substrates Detection System Time Dependence pH Value Selecting the Buffer Substance and the Ionic Strength Temperature Substrate Concentration Controls Further Reading
37 37 38 38 39 40 40 41 41 42 42 42 43 43
Microcalorimetry Differential Scanning Calorimetry (DSC) Isothermal Titration Calorimetry (ITC) Ligand Binding to Proteins Binding of Molecules to Membranes: Insertion and Peripheral Binding Pressure Perturbation Calorimetry (PPC) Further Reading
47 48 54 54 58 61 62
Immunological Techniques Antibodies Antibodies and Immune Defense Antibodies as Reagents Properties of Antibodies Functional Structure of IgG Antigen Interaction at the Combining Site Handling of Antibodies Antigens
63 63 63 64 64 66 67 68 69
43 44 44 45 45
VI 5.3 5.3.1 5.3.2 5.3.3 5.4 5.5 5.6 5.7 5.7.1 5.7.2 5.7.3 5.8
6 6.1 6.2 6.2.1 6.2.2 6.3 6.3.1 6.3.2
7 7.1 7.1.1 7.1.2 7.1.3 7.1.4 7.1.5 7.2 7.2.1 7.2.2 7.3 7.3.1 7.3.2 7.3.3 7.3.4 7.3.5 7.3.6 7.3.7 7.3.8 7.4 7.4.1
Table of Contents
Antigen–Antibody Reaction Immunoagglutination Immunoprecipitation Immune Binding Complement Fixation Methods in Cellular Immunology Alteration of Biological Functions Production of Antibodies Types of Antibodies New Antibody Techniques (Antibody Engineering) Optimized Monoclonal Antibody Constructs with Effector Functions for Therapeutic Application Outlook: Future Expansion of the Binding Concepts Dedication Further Reading
71 72 73 84 94 95 97 98 98 99 102 106 106 106
Chemical Modification of Proteins and Protein Complexes 107 Chemical Modification of Protein Functional Groups 108 Modification as a Means to Introduce Reporter Groups 116 Investigation with Naturally Occurring Proteins 116 Investigation of Recombinant and Mutated Proteins 120 Protein Crosslinking for the Analysis of Protein Interaction 121 Bifunctional Reagents 121 Photoaffinity Labeling 121 Further Reading 129 Spectroscopy Physical Principles and Measuring Techniques Physical Principles of Optical Spectroscopic Techniques Interaction of Light with Matter Absorption Measurement and the Lambert–Beer Law Photometer Time-Resolved Spectroscopy UV/VIS/NIR Spectroscopy Basic Principles Chromoproteins Fluorescence Spectroscopy Basic Principles of Fluorescence Spectroscopy Fluorescence: Emission and Action Spectra Fluorescence Studies using Intrinsic and Extrinsic Probes Green Fluorescent Protein (GFP) as a Unique Fluorescent Probe Quantum Dots as Fluorescence Labels Special Fluorescence Techniques: FRAP, FLIM, FCS, TIRF Förster Resonance Energy Transfer (FRET) Frequent Mistakes in Fluorescence Spectroscopy: “The Seven Sins of Fluorescence Measurements” Infrared Spectroscopy Basic Principles of IR Spectroscopy
7.4.2 7.4.3 7.4.4 7.5 7.5.1 7.5.2 7.5.3 7.6 7.7 7.7.1 7.7.2
131 132 132 133 140 143 144 146 146 147 154 154 156
9 9.1 9.2 9.3 9.4 9.5 9.5.1 9.5.2 9.6 9.7
Cleavage of Proteins Proteolytic Enzymes Strategy Denaturation of Proteins Cleavage of Disulfide Bonds and Alkylation Enzymatic Fragmentation Proteases Conditions for Proteolysis Chemical Fragmentation Summary Further Reading
207 207 208 209 209 210 210 215 216 217 218
10 10.1 10.2
Chromatographic Separation Methods Instrumentation Fundamental Terms and Concepts in Chromatography Biophysical Properties of Peptides and Proteins Chromatographic Separation Modes for Peptides and Proteins High-Performance Size Exclusion Chromatography High-Performance Reversed-Phase Chromatography (HP-RPC) High-Performance Normal-Phase Chromatography (HP-NPC) High-Performance Hydrophilic Interaction Chromatography (HP-HILIC) High-Performance Aqueous Normal Phase Chromatography (HP-ANPC) High-Performance Hydrophobic Interaction Chromatography (HP-HIC) High-Performance Ion Exchange Chromatography (HP-IEX) High-Performance Affinity Chromatography (HP-AC)
219 219
10.3 10.4 10.4.1 10.4.2
10.4.4 10.4.5 10.4.6 10.4.7 161 163 163
178 180
181 182 183 189 195 197 205
8.2 8.3 8.4 8.5 8.6
157
160 160
164 165 168 171 171 172 173 174 175 175
Light Microscopy Techniques – Imaging Steps on the Road to Microscopy – from Simple Lenses to High Resolution Microscopes Modern Applications Basic Physical Principles Detection Methods Sample Preparation Special Fluorescence Microscopic Analysis Further Reading
8 8.1
10.4.3
158 159
Molecular Vibrations Technical aspects of Infrared Spectroscopy Infrared Spectra of Proteins Raman Spectroscopy Basic Principles of Raman Spectroscopy Raman Experiments Resonance Raman Spectroscopy Single Molecule Spectroscopy Methods using Polarized Light Linear Dichroism Optical Rotation Dispersion and Circular Dichroism Further Reading
10.4.8
181
220 224 225 227 227 228 229 230 230 232 233
Table of Contents
10.5 10.5.1 10.5.2 10.5.3 10.5.4 10.6 10.6.1 10.6.2 10.6.3 10.6.4 10.7
11 11.1 11.2 11.3 11.3.1 11.3.2 11.3.3 11.3.4 11.3.5 11.3.6 11.3.7 11.3.8 11.3.9 11.3.10 11.3.11 11.3.12 11.4 11.4.1 11.4.2 11.4.3 11.5 11.6 11.6.1 11.6.2 11.6.3 11.6.4 11.6.5 11.6.6 11.7 11.7.1 11.7.2 11.7.3
12 12.1 12.2 12.3 12.3.1 12.3.2
Method Development from Analytical to Preparative Scale Illustrated for HP-RPC Development of an Analytical Method Scaling up to Preparative Chromatography Fractionation Analysis of Fractionations Multidimensional HPLC Purification of Peptides and Proteins by MD-HPLC Methods Fractionation of Complex Peptide and Protein Mixtures by MD-HPLC Strategies for MD-HPLC Methods Design of an Effective MD-HPLC Scheme Final Remarks Further Reading
234 234 236 237 238 238 238 239 239 240 242 242
Electrophoretic Techniques 243 Historical Review 244 Theoretical Fundamentals 245 Equipment and Procedures of Gel Electrophoreses 248 Sample Preparation 249 Gel Media for Electrophoresis 250 Detection and Quantification of the Separated Proteins 251 Zone Electrophoresis 253 Porosity Gradient Gels 254 Buffer Systems 255 Disc Electrophoresis 255 Acidic Native Electrophoresis 257 SDS Polyacrylamide Gel Electrophoresis 257 Cationic Detergent Electrophoresis 258 Blue Native Polyacrylamide Gel Electrophoresis 259 Isoelectric Focusing 259 Preparative Techniques 263 Electroelution from Gels 263 Preparative Zone Electrophoresis 264 Preparative Isoelectric Focusing 265 Free Flow Electrophoresis 266 High-Resolution Two-Dimensional Electrophoresis 267 Sample Preparation 268 Prefractionation 268 First Dimension: IEF in IPG Strips 269 Second Dimension: SDS Polyacrylamide Gel Electrophoresis 270 Detection and Identification of Proteins 270 Difference Gel Electrophoresis (DIGE) 270 Electroblotting 272 Blot Systems 272 Transfer Buffers 273 Blot Membranes 273 Further Reading 273 Capillary Electrophoresis Historical Overview Capillary Electrophoresis Setup Basic Principles of Capillary Electrophoresis Sample Injection The Engine: Electroosmotic Flow (EOF)
275 275 276 277 277 278
12.3.3 12.3.4 12.4 12.4.1 12.4.2 12.4.3 12.4.4 12.4.5 12.4.6 12.4.7 12.4.8 12.5 12.5.1 12.5.2 12.5.3 12.5.4 12.6
Joule Heating Detection Methods Capillary Electrophoresis Methods Capillary Zone Electrophoresis (CZE) Affinity Capillary Electrophoresis (ACE) Micellar Electrokinetic Chromatography (MEKC) Capillary Electrochromatography (CEC) Chiral Separations Capillary Gel Electrophoresis (CGE) Capillary Isoelectric Focusing (CIEF) Isotachophoresis (ITP) Special Techniques Sample Concentration Online Sample Concentration Fractionation Microchip Electrophoresis Outlook Further Reading
13 13.1 13.1.1 13.1.2 13.1.3 13.2 13.3
Amino Acid Analysis Sample Preparation Acidic Hydrolysis Alkaline Hydrolysis Enzymatic Hydrolysis Free Amino Acids Liquid Chromatography with Optical Detection Systems 13.3.1 Post-Column Derivatization 13.3.2 Pre-column Derivatization 13.4 Amino Acid Analysis using Mass Spectrometry 13.5 Summary Further Reading 14 14.1 14.1.1 14.1.2 14.1.3 14.1.4 14.1.5 14.1.6 14.2 14.2.1 14.2.2 14.2.3
Protein Sequence Analysis N-Terminal Sequence Analysis: The Edman Degradation Reactions of the Edman Degradation Identification of the Amino Acids Quality of Edman Degradation: the Repetitive Yield Instrumentation Problems of Amino Acid Sequence Analysis State of the Art C-Terminal Sequence Analysis Chemical Degradation Methods Peptide Quantities and Quality of the Chemical Degradation Degradation of Polypeptides with Carboxypeptidases Further Reading
15 Mass Spectrometry 15.1 Ionization Methods 15.1.1 Matrix Assisted Laser Desorption Ionization Mass Spectrometry (MALDI-MS) 15.1.2 Electrospray Ionization (ESI) 15.2 Mass Analyzer 15.2.1 Time-of-Flight Analyzers (TOF)
VII 279 279 281 281 285 286 288 289 290 291 293 295 295 295 296 297 297 299 301 302 302 303 303 303 303 303 305 309 310 311 313 315 315 316 317 319 322 325 325 325 327 327 328 329 330 330 335 341 343
VIII
Table of Contents
15.2.2 15.2.3 15.2.4 15.2.5 15.2.6 15.3 15.3.1 15.3.2 15.4 15.4.1 15.4.2 15.4.3 15.4.4 15.5 15.5.1 15.5.2 15.5.3 15.5.4 15.5.5 15.5.6 15.5.7 15.6 15.6.1 15.6.2 15.6.3 15.7 15.7.1 15.7.2 15.7.3 15.8
Quadrupole Analyzer Electric Ion Traps Magnetic Ion Trap Orbital Ion Trap Hybrid Instruments Ion Detectors Secondary Electron Multiplier (SEV) Faraday Cup Fragmentation Techniques Collision Induced Dissociation (CID) Prompt and Metastable Decay (ISD, PSD) Photon-Induced Dissociation (PID, IRMPD) Generation of Free Radicals (ECD, HECD, ETD) Mass Determination Calculation of Mass Influence of Isotopy Calibration Determination of the Number of Charges Signal Processing and Analysis Derivation of the Mass Problems Identification, Detection, and Structure Elucidation Identification Verification Structure Elucidation LC-MS and LC-MS/MS LC-MS LC-MS/MS Ion Mobility Spectrometry (IMS) Quantification Further Reading
345 348 349 350 351 355 356 357 357 357 358 360 360 362 362 362 365 365 366 366 366 368 368 369 369 375 375 376 378 378 379
16 16.1 16.1.1 16.1.2 16.1.3 16.1.4 16.1.5 16.1.6 16.1.7
Protein–Protein Interactions The Two-Hybrid System Principle of Two-Hybrid Systems Elements of the Two-Hybrid System Construction of Bait and Prey Proteins Which Bait Proteins can be used in a Y2H Screen? AD Fusion Proteins and cDNA Libraries Carrying out a Y2H Screen Other Modifications and Extensions of the Two-Hybrid-Technology Biochemical and Functional Analysis of Interactions TAP-Tagging and Purification of Protein Complexes Analyzing Interactions In Vitro: GST- Pulldown Co-immunoprecipitation Far-Western Surface Plasmon Resonance Spectroscopy Fluorescence Resonance Energy Transfer (FRET) Introduction Key Physical Principles of FRET Methods of FRET Measurements Fluorescent Probes for FRET Alternative Tools for Probing Protein–Protein Interactions: LINC and STET Analytical Ultracentrifugation
381 381 381 382 382 385 385 386
16.1.8 16.2 16.3 16.4 16.5 16.6 16.7 16.7.1 16.7.2 16.7.3 16.7.4 16.7.5 16.8
16.8.1 16.8.2 16.8.3 16.8.4
Principles of Instrumentation Basics of Centrifugation Sedimentation Velocity Experiments Sedimentation–Diffusion Equilibrium Experiments Further Reading
410 411 412 415 416
17 17.1
Biosensors Dry Chemistry: Test Strips for Detecting and Monitoring Diabetes Biosensors Concept of Biosensors Construction and Function of Biosensors Cell Sensors Immunosensors Biomimetic Sensors From Glucose Enzyme Electrodes to Electronic DNA Biochips Resume: Biosensor or not Biosensor is no Longer the Question Further Reading
419
17.2 17.2.1 17.2.2 17.2.3 17.2.4 17.3 17.4 17.5
Part II 3D Structure Determination 18 18.1 18.1.1 18.1.2 18.1.3 18.1.4 18.1.5 18.1.6 18.1.7 18.2 18.2.1 18.2.2 18.2.3 18.2.4
391 393 394 397 398 399 400 402 402 403 403 406 408 409
18.2.5 18.2.6 18.2.7 18.2.8 18.2.9 18.2.10
19 19.1
Magnetic Resonance Spectroscopy of Biomolecules NMR Spectroscopy of Biomolecules Theory of NMR Spectroscopy One-Dimensional NMR Spectroscopy Two-Dimensional NMR Spectroscopy Three-Dimensional NMR Spectroscopy Resonance Assignment Protein Structure Determination Protein Structures and more — an Overview EPR Spectroscopy of Biological Systems Basics of EPR Spectroscopy cw- EPR Spectroscopy g-Value Electron Spin Nuclear Spin Coupling (Hyperfine Coupling) g and Hyperfine Anisotropy Electron Spin–Electron Spin Coupling Pulsed EPR Experiments Further Examples of EPR Applications General Remarks on the Significance of EPR Spectra Comparison EPR/NMR Acknowledgements Further Reading
Electron Microscopy Transmission Electron Microscopy – Instrumentation 19.2 Approaches to Preparation 19.2.1 Native Samples in Ice 19.2.2 Negative Staining
420 420 420 421 425 426 427 428 429 429
431 433 433 434 438 443 449 452 457 462 466 467 468 469 469 470 472 473 479 481 481 482 482 485 487 488 488 490
Table of Contents
19.2.3 19.2.4 19.3 19.3.1 19.3.2 19.3.3 19.3.4 19.3.5 19.3.6 19.4 19.4.1 19.4.2 19.4.3 19.4.4 19.4.5 19.5 19.5.1 19.5.2 19.5.3 19.6 19.6.1 19.6.2 19.6.3 19.7
20 20.1 20.2 20.3 20.4 20.5 20.6 20.7
21 21.1 21.1.1 21.1.2 21.1.3 21.1.4 21.2 21.2.1 21.2.2 21.2.3 21.3
Metal Coating by Evaporation Labeling of Proteins Imaging Process in the Electron Microscope Resolution of a Transmission Electron Microscope Interactions of the Electron Beam with the Object Phase Contrast in Transmission Electron Microscopy Electron Microscopy with a Phase Plate Imaging Procedure for Frozen-Hydrated Specimens Recording Images – Cameras and the Impact of Electrons Image Analysis and Processing of Electron Micrographs Pixel Size Fourier Transformation Analysis of the Contrast Transfer Function and Object Features Improving the Signal-to-Noise Ratio Principal Component Analysis and Classification Three-Dimensional Electron Microscopy Three-Dimensional Reconstruction of Single Particles Three-Dimensional Reconstruction of Regularly Arrayed Macromolecular Complexes Electron Tomography of Individual Objects Analysis of Complex 3D Data Sets Hybrid Approach: Combination of EM and X-Ray Data Segmenting Tomograms and Visualization Identifying Protein Complexes in Cellular Tomograms Perspectives of Electron Microscopy Further Reading
491 492 492 492 493 495 495 496 497 498 498 499 501 504 506 508 509 511 512 514 514 515 515 516 517
Atomic Force Microscopy Introduction Principle of the Atomic Force Microscope Interaction between Tip and Sample Preparation Procedures Mapping Biological Macromolecules Force Spectroscopy of Single Molecules Detection of Functional States and Interactions of Individual Proteins Further Reading
519 519 520 521 522 522 524
X-Ray Structure Analysis X-Ray Crystallography Crystallization Crystals and X-Ray Diffraction The Phase Problem Model Building and Structure Refinement Small Angle X-Ray Scattering (SAXS) Machine Setup Theory Data Analysis X-Ray Free Electron LASER (XFEL)
529 530 531 533 538 542 543 544 545 547 549
526 527
21.3.1 Machine Setup and Theory Acknowledgement Further Reading
Part III Peptides, Carbohydrates, and Lipids 22 22.1 22.2 22.3 22.4 22.5
Analytics of Synthetic Peptides Concept of Peptide Synthesis Purity of Synthetic Peptides Characterization and Identity of Synthetic Peptides Characterization of the Structure of Synthetic Peptides Analytics of Peptide Libraries Further Reading
23 23.1 23.1.1 23.1.2 23.1.3 23.1.4 23.1.5 23.2 23.2.1 23.2.2 23.3 23.3.1
IX 549 550 551
553 555 555 560 562 564 567 569
Carbohydrate Analysis General Stereochemical Basics The Series of D-Sugars Stereochemistry of D-Glucose Important Monosaccharide Building Blocks The Series of L-Sugars The Glycosidic Bond Protein Glycosylation Structure of the N-Glycans Structure of the O-Glycans Analysis of Protein Glycosylation Analysis on the Basis of the Intact Glycoprotein 23.3.2 Mass Spectrometric Analysis on the Basis of Glycopeptides 23.3.3 Release and Isolation of the N-Glycan Pool 23.3.4 Analysis of Individual N-Glycans 23.4 Genome, Proteome, Glycome 23.5 Final Considerations Further Reading
571 572 572 573 574 574 574 579 580 580 581
24 24.1 24.2
613 613
24.2.1 24.2.2 24.3 24.3.1 24.3.2 24.3.3 24.3.4 24.3.5 24.4 24.4.1 24.4.2 24.4.3 24.4.4
Lipid Analysis Structure and Classification of Lipids Extraction of Lipids from Biological Sources Liquid Phase Extraction Solid Phase Extraction Methods for Lipid Analysis Chromatographic Methods Mass Spectrometry Immunoassays Further Methods in Lipid Analysis Combining Different Analytical Systems Analysis of Selected Lipid Classes Whole Lipid Extracts Fatty Acids Nonpolar Neutral Lipids Polar Ester Lipids
582 588 590 599 610 611 612
615 616 616 618 618 622 622 623 623 626 626 627 628 630
X
Table of Contents
24.4.5 Lipid Hormones and Intracellular Signaling Molecules 24.5 Lipid Vitamins 24.6 Lipidome Analysis 24.7 Perspectives Further Reading 25 25.1 25.1.1 25.1.2 25.2 25.3 25.4 25.4.1
25.4.2 25.5 25.5.1 25.5.2 25.6 25.7
Analysis of Post-translational Modifications: Phosphorylation and Acetylation of Proteins Functional Relevance of Phosphorylation and Acetylation Phosphorylation Acetylation Strategies for the Analysis of Phosphorylated and Acetylated Proteins and Peptides Separation and Enrichment of Phosphorylated and Acetylated Proteins and Peptides Detection of Phosphorylated and Acetylated Proteins and Peptides Detection by Enzymatic, Radioactive, Immunochemical, and Fluorescence Based Methods Detection of Phosphorylated and Acetylated Proteins by Mass Spectrometry Localization and Identification of Post-translationally Modified Amino Acids Localization of Phosphorylated and Acetylated Amino Acids by Edman Degradation Localization of Phosphorylated and Acetylated Amino Acids by Tandem Mass Spectrometry Quantitative Analysis of Post-translational Modifications Future of Post-translational Modification Analysis Further Reading
Part IV Nucleic Acid Analytics 26 26.1 26.1.1 26.1.2 26.1.3 26.1.4 26.2 26.3 26.3.1 26.3.2 26.4 26.4.1 26.4.2 26.5 26.5.1 26.5.2
633 638 640 642 644
645 645 645 646 647 649 651
651 653 653 654 654 659 661 661
663
Isolation and Purification of Nucleic Acids Purification and Determination of Nucleic Acid Concentration Phenolic Purification of Nucleic Acids Gel Filtration Precipitation of Nucleic Acids with Ethanol Determination of the Nucleic Acid Concentration Isolation of Genomic DNA Isolation of Low Molecular Weight DNA Isolation of Plasmid DNA from Bacteria Isolation of Eukaryotic Low Molecular Weight DNA Isolation of Viral DNA Isolation of Phage DNA Isolation of Eukaryotic Viral DNA Isolation of Single-Stranded DNA Isolation of M13 Phage DNA Separation of Single- and Double-Stranded DNA
665 665 665 666 667 668 669 670 670 674 674 674 675 676 676 676
26.6 26.6.1 26.6.2 26.6.3 26.7 26.8
Isolation of RNA Isolation of Cytoplasmic RNA Isolation of Poly(A) RNA Isolation of Small RNA Isolation of Nucleic Acids using Magnetic Particles Lab-on-a-chip Further Reading
27 27.1 27.1.1 27.1.2 27.1.3 27.1.4 27.2 27.2.1 27.2.2 27.2.3 27.2.4 27.2.5 27.3 27.3.1 27.3.2 27.4 27.4.1 27.4.2 27.4.3 27.4.4 27.4.5 27.4.6 27.5 27.5.1 27.5.2 27.5.3 27.5.4 27.6 27.6.1 27.6.2
Analysis of Nucleic Acids Restriction Analysis Principle of Restriction Analyses Historical Overview Restriction Enzymes In Vitro Restriction and Applications Electrophoresis Gel Electrophoresis of DNA Gel Electrophoresis of RNA Pulsed-Field Gel Electrophoresis (PFGE) Two-Dimensional Gel Electrophoresis Capillary Gel Electrophoresis Staining Methods Fluorescent Dyes Silver Staining Nucleic Acid Blotting Nucleic Acid Blotting Methods Choice of Membrane Southern Blotting Northern Blotting Dot- and Slot-Blotting Colony and Plaque Hybridization Isolation of Nucleic Acid Fragments Purification using Glass Beads Purification using Gel Filtration or Reversed Phase Purification using Electroelution Other Methods LC-MS of Oligonucleotides Principles of the Synthesis of Oligonucleotides Investigation of the Purity and Characterization of Oligonucleotides 27.6.3 Mass Spectrometric Investigation of Oligonucleotides 27.6.4 IP-RP-HPLC-MS Investigation of a Phosphorothioate Oligonucleotide Further Reading 28
28.1 28.1.1 28.1.2 28.1.3 28.2 28.2.1 28.2.2 28.2.3 28.2.4 28.3 28.3.1
Techniques for the Hybridization and Detection of Nucleic Acids Basic Principles of Hybridization Principle and Practice of Hybridization Specificity of the Hybridization and Stringency Hybridization Methods Probes for Nucleic Acid Analysis DNA Probes RNA Probes PNA Probes LNA Probes Methods of Labeling Labeling Positions
676 677 678 679 679 680 680 681 681 681 682 682 685 690 691 697 698 700 701 702 702 704 704 704 704 705 706 707 707 708 708 708 708 709 709 709 711 712 714 717
719 720 721 722 723 729 730 731 732 732 733 733
Table of Contents
28.3.2 28.3.3 28.3.4 28.4 28.4.1 28.4.2 28.4.3 28.5 28.5.1 28.5.2 28.5.3
Enzymatic Labeling Photochemical Labeling Reactions Chemical Labeling Detection Systems Staining Methods Radioactive Systems Non-radioactive Systems Amplification Systems Target Amplification Target-Specific Signal Amplification Signal Amplification Further Reading
735 737 737 738 738 738 739 750 751 751 752 753
29 29.1 29.2 29.2.1 29.2.2 29.2.3 29.2.4 29.2.5 29.3 29.3.1 29.3.2 29.3.3 29.3.4 29.3.5 29.3.6 29.3.7 29.3.8 29.3.9 29.3.10 29.4 29.4.1 29.4.2 29.5 29.5.1 29.5.2 29.5.3 29.6 29.6.1
Polymerase Chain Reaction Possibilities of PCR Basics Instruments Amplification of DNA Amplification of RNA (RT-PCR) Optimizing the Reaction Quantitative PCR Special PCR Techniques Nested PCR Asymmetric PCR Use of Degenerate Primers Multiplex PCR Cycle sequencing In Vitro Mutagenesis Homogeneous PCR Detection Procedures Quantitative Amplification Procedures In Situ PCR Other Approaches Contamination Problems Avoiding Contamination Decontamination Applications Detection of Infectious Diseases Detection of Genetic Defects The Human Genome Project Alternative Amplification Procedures Nucleic Acid Sequence-Based Amplification (NASBA) Strand Displacement Amplification (SDA) Helicase-Dependent Amplification (HDA) Ligase Chain Reaction (LCR) Qβ Amplification Branched DNA Amplification (bDNA) Prospects Further Reading
755 755 756 756 758 761 763 763 766 766 767 767 767 768 768 768 769 769 769 770 770 771 772 772 773 776 777
29.6.2 29.6.3 29.6.4 29.6.5 29.6.6 29.7
30 DNA Sequencing 30.1 Gel-Supported DNA Sequencing Methods 30.1.1 Sequencing according to Sanger: The Dideoxy Method 30.1.2 Labeling Techniques and Methods of Verification 30.1.3 Chemical Cleavage according to Maxam and Gilbert
777 777 777 779 780 782 782 782 785 786 789 796 800
Gel-Free DNA Sequencing Methods – The Next Generation 30.2.1 Sequencing by Synthesis 30.2.2 Single Molecule Sequencing Further Reading
XI
30.2
806 807 813 815
31 31.1
Analysis of Epigenetic Modifications 817 Overview of the Methods to Detect DNA-Modifications 818 31.2 Methylation Analysis with the Bisulfite Method 819 31.2.1 Amplification and Sequencing of Bisulfite-Treated DNA 819 31.2.2 Restriction Analysis after Bisulfite PCR 820 31.2.3 Methylation Specific PCR 822 31.3 DNA Analysis with Methylation Specific Restriction Enzymes 823 31.4 Methylation Analysis by Methylcytosine-Binding Proteins 825 31.5 Methylation Analysis by Methylcytosine-Specific Antibodies 826 31.6 Methylation Analysis by DNA Hydrolysis and Nearest Neighbor-Assays 827 31.7 Analysis of Epigenetic Modifications of Chromatin 828 31.8 Chromosome Interaction Analyses 828 31.9 Outlook 829 Further Reading 829 32 Protein–Nucleic Acid Interactions 32.1 DNA–Protein Interactions 32.1.1 Basic Features for DNA–Protein Recognition: Double-Helical Structures 32.1.2 DNA Curvature 32.1.3 DNA Topology 32.2 DNA-Binding Motifs 32.3 Special Analytical Methods 32.3.1 Filter Binding 32.3.2 Gel Electrophoresis 32.3.3 Determination of Dissociation Constants 32.3.4 Analysis of DNA–Protein Complex Dynamics 32.4 DNA Footprint Analysis 32.4.1 DNA Labeling 32.4.2 Primer Extension Reaction for DNA Analysis 32.4.3 Hydrolysis Methods 32.4.4 Chemical Reagents for the Modification of DNA–Protein Complexes 32.4.5 Interference Conditions 32.4.6 Chemical Nucleases 32.4.7 Genome-Wide DNA–Protein Interactions 32.5 Physical Analysis Methods 32.5.1 Fluorescence Methods 32.5.2 Fluorophores and Labeling Procedures 32.5.3 Fluorescence Resonance Energy Transfer (FRET) 32.5.4 Molecular Beacons 32.5.5 Surface Plasmon Resonance (SPR) 32.5.6 Scanning Force Microscopy (SFM) 32.5.7 Optical Tweezers 32.5.8 Fluorescence Correlation Spectroscopy (FCS) 32.6 RNA–Protein Interactions
831 831 831 832 833 835 836 836 836 839 840 841 843 843 844 846 848 849 850 851 851 851 852 853 853 854 855 856 856
XII
Table of Contents
32.6.1 Functional Diversity of RNA 856 32.6.2 RNA Secondary Structure Parameters and unusual Base Pairs 857 32.6.3 Dynamics of RNA–Protein Interactions 857 32.7 Characteristic RNA-Binding Motifs 859 32.8 Special Methods for the Analysis of RNA–Protein Complexes 860 32.8.1 Limited Enzymatic Hydrolyses 861 32.8.2 Labeling Methods 861 32.8.3 Primer Extension Analysis of RNA 862 32.8.4 Customary RNases 862 32.8.5 Chemcal Modification of RNA–Protein Complexes 863 32.8.6 Chemical Crosslinking 866 32.8.7 Incorporation of Photoreactive Nucleotides 867 32.8.8 Genome-Wide Identification of Transcription Start Sites (TSS) 867 32.9 Genetic Methods 868 32.9.1 Tri-hybrid Method 868 32.9.2 Aptamers and the Selex Procedure 869 32.9.3 Directed Mutations within Binding Domains 870 Further Reading 870
Part V Functional and Systems Analytics 33 33.1 33.2 33.3 33.3.1 33.3.2 33.3.3 33.4 33.4.1 33.5 33.6 33.6.1 33.6.2 33.6.3 33.7 33.7.1 33.7.2 33.7.3 33.7.4 33.7.5 33.8 33.9 33.10 34 34.1
Sequence Data Analysis Sequence Analysis and Bioinformatics Sequence: An Abstraction for Biomolecules Internet Databases and Services Sequence Retrieval from Public Databases Data Contents and File Format Nucleotide Sequence Management in the Laboratory Sequence Analysis on the Web EMBOSS Sequence Composition Sequence Patterns Transcription Factor Binding Sites Identification of Coding Regions Protein Localization Homology Identity, Similarity, Homology Optimal Sequence Alignment Alignment for Fast Database Searches: BLAST Profile-Based Sensitive Database Search: PSI-BLAST Homology Threshold Multiple Alignment and Consensus Sequences Structure Prediction Outlook Analysis of Promoter Strength and Nascent RNA Synthesis Methods for the Analysis of RNA Transcripts
34.1.1 34.1.2 34.1.3 34.1.4 34.1.5 34.1.6 34.2 34.2.1 34.2.2 34.3 34.3.1 34.3.2 34.3.3 34.4 34.4.1 34.4.2 34.4.3
873 875 875 876 877 878 879 881 881 881 882 882 884 885 886 887 887 888 890 890 891 891 892 893
895 895
35 35.1 35.1.1 35.1.2 35.1.3 35.1.4 35.1.5 35.2 35.2.1 35.2.2
36 36.1 36.1.1 36.1.2 36.1.3 36.1.4 36.1.5 36.2 36.2.1 36.2.2 36.2.3 36.2.4 36.2.5 36.2.6 36.3
Overview Nuclease S1 Analysis of RNA Ribonuclease-Protection Assay (RPA) Primer Extension Assay Northern Blot and Dot- and Slot-Blot Reverse Transcription Polymerase Chain Reaction (RT-PCR and RT-qPCR) Analysis of RNA Synthesis In Vivo Nuclear-run-on Assay Labeling of Nascent RNA with 5-Fluoro-uridine (FUrd) In Vitro Transcription in Cell-Free Extracts Components of an In Vitro Transcription Assay Generation of Transcription-Competent Cell Extracts and Protein Fractions Template DNA and Detection of In Vitro Transcripts In Vivo Analysis of Promoter Activity in Mammalian Cells Vectors for Analysis of Gene-Regulatory cis-Elements Transfer of DNA into Mammalian Cells Analysis of Reporter Gene Expression Further Reading Fluorescent In Situ Hybridization in Molecular Cytogenetics Methods of Fluorescent DNA Hybridization Labeling Strategy DNA Probes Labeling of DNA Probes In Situ Hybridization Evaluation of Fluorescent Hybridization Signals Application: FISH and CGH FISH Analysis of Genomic DNA Comparative Genomic Hybridization (CGH) Further Reading Physical and Genetic Mapping of Genomes Genetic Mapping: Localization of Genetic Markers within the Genome Recombination Genetic Markers Linkage Analysis – the Generation of Genetic Maps Genetic Map of the Human Genome Genetic Mapping of Disease Genes Physical Mapping Restriction Mapping of Whole Genomes Mapping of Recombinant Clones Generation of a Physical Map Identification and Isolation of Genes Transcription Maps of the Human Genome Genes and Hereditary Disease – Search for Mutations Integration of Genome Maps
895 896 898 901 902 904 905 905 906 907 907 908 908 911 911 912 914 916
917 917 917 918 918 919 920 920 920 921 924 925 925 925 927 929 931 932 932 932 934 935 937 939 940 940
Table of Contents
36.4
The Human Genome Further Reading
942 942
37 37.1 37.1.1 37.1.2 37.1.3 37.2 37.2.1 37.2.2 37.2.3 37.2.4 37.2.5 37.3 37.3.1 37.3.2 37.3.3 37.4 37.4.1 37.4.2 37.5 37.5.1 37.5.2
DNA-Microarray Technology RNA Analyses Transcriptome Analysis RNA Splicing RNA Structure and Functionality DNA Analyses Genotyping Methylation Studies DNA Sequencing Comparative Genomic Hybridization (CGH) Protein–DNA Interactions Molecule Synthesis DNA Synthesis RNA Production On-Chip Protein Expression Other Approaches Barcode Identification A Universal Microarray Platform New Avenues Structural Analyses Beyond Nucleic Acids Further Reading
945 946 946 947 947 948 948 948 949 951 951 952 952 953 953 954 954 955 956 956 956 957
38
The Use of Oligonucleotides as Tools in Cell Biology Antisense Oligonucleotides Mechanisms of Antisense Oligonucleotides Triplex-Forming Oligonucleotides Modifications of Oligonucleotides to Decrease their Susceptibility to Nucleases Use of Antisense Oligonucleotides in Cell Culture and in Animal Models Antisense Oligonucleotides as Therapeutics Ribozymes Discovery and Classification of Ribozymes Use of Ribozymes RNA Interference and MicroRNAs Basics of RNA Interference RNA Interference Mediated by Expression Vectors Uses of RNA Interference microRNAs Aptamers: High-Affinity RNA- and DNAOligonucleotides Selection of Aptamers Uses of Aptamers Genome Editing with CRISPR/Cas9 Outlook Further Reading
38.1 38.1.1 38.1.2 38.1.3 38.1.4 38.1.5 38.2 38.2.1 38.2.2 38.3 38.3.1 38.3.2 38.3.3 38.3.4 38.4 38.4.1 38.4.2 38.5 38.6 39 39.1 39.2 39.3 39.4
Proteome Analysis General Aspects in Proteome Analysis Definition of Starting Conditions and Project Planning Sample Preparation for Proteome Analysis Protein Based Quantitative Proteome Analysis (Top-Down Proteomics)
959 960 960 961 962 964 964 965 965 966 967 967 968 969 970 971 971 973 974 975 976 977 977 979 980
39.4.1 Two-Dimensional-Gel-Based Proteomics 982 39.4.2 Two-Dimensional Differential Gel Electrophoresis (2D DIGE) 986 39.4.3 Top-Down Proteomics using Isotope Labels 986 39.4.4 Top-Down Proteomics using Intact Protein Mass Spectrometry 987 39.4.5 Concepts in Intact Protein Mass Spectrometry 987 39.5 Peptide Based Quantitative Proteome Analysis (Bottom-Up Proteomics) 998 39.5.1 Introduction 998 39.5.2 Bottom-Up Proteomics 998 39.5.3 Complexity of the Proteome 1000 39.5.4 Bottom-Up Proteomic Strategies 1000 39.5.5 Peptide Quantification 1001 39.5.6 Data Dependent Analysis (DDA) 1002 39.5.7 Selected Reaction Monitoring 1003 39.5.8 SWATH-MS 1010 39.5.9 Summary 1012 39.5.10 Extensions 1012 39.6 Stable Isotope Labeling in Quantitative Proteomics 1013 39.6.1 Stable Isotope Label in Top-Down Proteomics 1013 39.6.2 Stable Isotope Labeling in Bottom-Up Proteomics 1019 Further Reading 1021 40 40.1 40.2 40.3 40.4 40.5 40.6 40.7 40.8
Metabolomics and Peptidomics Systems Biology and Metabolomics Technological Platforms for Metabolomics Metabolomic Profiling Peptidomics Metabolomics – Knowledge Mining Data Mining Fields of Application Outlook Further Reading
Interactomics – Systematic Protein–Protein Interactions 41.1 Protein Microarrays 41.1.1 Sensitivity Increase through Miniaturization – Ambient Analyte Assay 41.1.2 From DNA to Protein Microarrays 41.1.3 Application of Protein Microarrays Further Reading
1023 1025 1026 1027 1028 1029 1030 1032 1032 1032
41
42 42.1 42.2 42.2.1 42.2.2 42.2.3 42.2.4
982
XIII
Chemical Biology Chemical Biology – Innovative Chemical Approaches to Study Biological Phenomena Chemical Genetics – Small Organic Molecules for the Modulation of Protein Function Study of Protein Functions with Small Organic Molecules Forward and Reverse Chemical Genetics The Bump-and-Hole Approach of Chemical Genetics Identification of Kinase Substrates with ASKA Technology
1033 1033 1034 1035 1037 1039 1041 1041 1043 1044 1046 1047 1050
XIV
Table of Contents
42.2.5 Switching Biological Systems on and off with Small Organic Molecules 42.3 Expressed Protein Ligation – Symbiosis of Chemistry and Biology for the Study of Protein Functions 42.3.1 Analysis of Lipid-Modified Proteins 42.3.2 Analysis of Phosphorylated Proteins 42.3.3 Conditional Protein Splicing Further Reading 43 43.1 43.1.1 43.1.2 43.1.3 43.2 43.2.1 43.2.2
Toponome Analysis “Life is Spatial” Antibody Based Toponome Analysis using Imaging Cycler Microscopy (ICM) Concept of the Protein Toponome Imaging Cycler Robots: Fundament of a Toponome Reading Technology Summary and Outlook Acknowledgements Mass Spectrometry Imaging Analytical Microprobes Mass Spectrometric Pixel Images
1052 1052 1054 1054 1055
43.2.3 Achievable Spatial Resolution 43.2.4 SIMS, ME-SIMS, and Cluster SIMS Imaging: Enhancing the Mass Range 43.2.5 Lateral Resolution and Analytical Limit of Detection 43.2.6 Coarse Screening by MS Imaging 43.2.7 Accurate MALDI Mass Spectrometry Imaging 43.2.8 Identification and Characterization of Analytes Further Reading
1067 1068 1068 1069 1070
1057 1057
Appendix 1: Amino Acids and Posttranslational Modifications
1073
1057 1058
Appendix 2: Symbols and Abbreviations
1075
Appendix 3: Standard Amino Acids (three and one letter code)
1081
Appendix 4: Nucleic Acid Bases
1083
Index
1085
1051
1059 1063 1063 1064 1064 1064
1065 1067
Preface
This is a book about methods. You may ask: Why do we need a dedicated book about methods and why should I buy it? We can offer at least two good answers. The first answer is of a theoretical nature: the method determines the quality of the scientific finding gained in that manner. Only by understanding a method, its strengths and, more importantly, its weaknesses, is it possible to estimate the general applicability of an observation or hypothesis. The development or improvement of a method is therefore a means to expand and improve the “tentative truth” generated by experimental science. Great value has been placed on describing the material critically and illuminatingly to enable the reader to engage with the material and gain a thorough understanding. This is, in our opinion, the most important reason why methods must be offered for classroom study. However, a deep and broad knowledge of methods is just as important for ongoing experimental work as it is for understanding past experiments. The second answer is the intent – hopefully successful – of this book to make getting to know and understand these methods clear and straightforward in order to make this book an irreplaceable tool for both students and teachers. Our intent results from our conviction, backed by our experience, that today every individual, whether student, teacher, or scientist, is hopelessly overwhelmed by the large number of different techniques currently in use in biological sciences. At the same time, using these techniques is imperative. We proudly undertook this intellectual enterprise to describe these techniques as completely as possible in an up-to-date manner. To the best of our knowledge, no English language textbook exists that is dedicated to these same goals and with the same level of coverage. One might wonder why the most apparent reason to publish this book has not been mentioned: namely, using this book to learn, or hope to learn, methods that are needed directly for ongoing experimental work. We wish to make two things clear: This is not a “cook book”. This means that after digesting a chapter the reader will not be able to go to his or her laboratory bench and apply what has just been read like a recipe – for that to be possible it will be first necessary for the reader to work through the
literature relevant to the topic covered. The reader should be in a position – at least this is our goal and wish – to optimize his approach through the overview and insights acquired. As for the second point of clarification: This book does not see itself as competition for existing laboratory manuals for diverse techniques, such as protein determination or PCR. The intent is much more to use carefully coordinated and complete descriptions of the methods, using frequent cross-referencing of other chapters, either in the text or in a flanking box, to illustrate the connections between apparently unrelated techniques and to show their mutual dependencies. We believe that the reader will profit from these lessons by gaining a sense of orientation and will understand the relationship between different techniques better, or possibly appreciate them for the first time. We do not wish to conceal the fact that for us, the editors, certain methodical relationships only became clear in the course of working through some of the manuscripts. As such, this book intends to provide coverage at a higher level, more than any single method manual or a simple collection of methods could. What is the actual content of this book? The book is titled Bioanalytics, which indicates that it is about analytical methods in biological sciences. This must be qualified. What are the biological sciences? Is it biochemistry or also molecular genetics, or cell and developmental biology, or even medicine? In any case, molecular biology would be included. This matter gets more complicated when one considers that modern medicine or cell biology are unimaginable without molecular biology. This book cannot satisfy all the needs of these sciences. In addition, not all analytical methods are contained within it, instead only those that involve biological macromolecules and their modifications. Macromolecules are most often proteins, but also include carbohydrates, lipids, and nucleic acids like DNA and RNA. Special methods for the analysis of small molecular metabolites are also not included. On occasion, we have crossed over the boundaries we have set for ourselves. For example, methods for the preparation of DNA and RNA are presented, simply because they are so closely and necessarily associated with the subsequent analytical techniques. In addition, many techniques, such as electrophoresis or chromatography, can be used at
XVI
Preface
both analytical and preparative scales. For other techniques it is not easy to distinguish between preparation and analysis if one does not wish to follow the traditional division between the two based solely on the amount of material involved. Is the identification of interaction partners using the two-hybrid system an analytic method, when the final step is based on the labo intensive construction of the corresponding clones, that is, to say based on a method that, at first, does not have anything to do with investigating the interaction? Similar is the case of site-specific mutation of genes for the investigation of gene function, which first requires the construction (and not the analysis) of the mutated sequences in vitro. On the other hand, we intentionally omitted the description of a few techniques that are clearly preparative. The synthesis of oligonucleotides – a clearly preparative technique – and the cloning of DNA were omitted. The latter is, despite being a requirement or goal of a large number of analytical methods, not an analytical method itself. In this case our decision was easy since there are already numerous good introductions and manuals about cloning DNA. In summary, the book describes the analytical methods of protein and nucleic acid (bio)chemistry, molecular biology, and, to a certain degree, modern cytogenetics. In this context, “molecular biology” means those parts of molecular genetics and biochemistry that involve the structure and function of nucleic acids. Methods of (classical) genetics, as well as traditional cell biology, are therefore rarely, if ever, included. We wish to emphasize that chapters that directly relate to the function of proteins and nucleic acids have been collected into a special section of the book, the “Systematic Analysis of Function.” We have gone along with the shift in paradigm from traditional bioanalytics to holistic analysis approaches. In this section many topics are addressed – even though they are sometimes not entirely mature – which are on the cutting edge of science. We are aware of the fact that this area is subject to rapid change and a few aspects could in the near future, perhaps, appear to be too optimistic or pessimistic. However, we believe that discussion of the most modern techniques and strategies at this point in time covers fascinating aspects and hopefully proves to be inspiring. The increasing availability of DNA and protein sequences of many organisms is, on the one hand, the critical fundament for this systematic function analysis and, on the other hand, makes high-throughput analysis and analysis of the data increasingly important. Information gained from the genome, proteome, and metabolome is compared with in silico analysis, which factors in the localization and interaction between biomolecules and unites everything into complex networks. The long-term goal of completely understanding the system can surely only be reached by the incorporation of further areas of expertise that are not yet an accepted component of bioanalytics. Bioanalysts must become, and are becoming, a kind of systems biologists, more interdisciplinary, and more successful in close cooperation together with experts in the fields of informatics, system theory, biotechnology, and cell biology. Who is this book addressed to? What has already been said provides a hint: Primarily biologists, chemists, pharmacists, physicians, and biophysicists. For some (biologists, chemists) the book will be interesting because it describes methods of their own discipline. For the second group (e.g., pharmacists, physicians, and biophysicists) the book is relevant because they can
find the background and fundamentals for much of the knowledge, which they find in their own discipline. Beyond these groups, this book is dedicated to interested readers who would like to know more about the subject matter. The material covered presumes that the user has taken at least an introductory course in the fundamentals of biochemistry or molecular genetics/gene technology, ideally both, or is in the process of doing so. We can imagine that this book would be an ideal supplement to such a course. It can and should especially be consulted when involved in experimental activities. This book is intended to be of equal value to students, teachers, and workers in these fields of science. The organization of the material proved to be one of the most difficult aspects of putting this book together. It is almost impossible to treat the techniques used in such complex fields in the two dimensions paper offers accurately without simultaneously compromising the didactic intentions of the book. We had a choice of two approaches: a more theoretical and intellectually stringent approach or a more practically oriented approach. The theoretical approach would have been to divide the methods exclusively according to type, for example chromatography, electrophoresis, centrifugation, and so on. Under each type of method its use would be divided according to objective and by the differing types of starting materials. This approach is more logical, but harder to comprehend and unrelated to actual practice. The more practical presentation begins with the concrete problem or question and describes the method that answers the question best. This is more intuitive, but inevitably leads to redundancies. A complete deep, “multidimensional” understanding of the material is only possible after the entire book has been absorbed. The approach in this book, for the most part, follows the second, practically oriented, approach. When possible, such as in the section “Protein analysis”, the methods were grouped and presented according to the topic addressed. This includes the fundamentals of instrumental techniques, which is knowledge required for the complete understanding of other sections. We approached the problem of redundancy by cross-referencing the first instance in which a method is described. Sometimes we left redundancies in place for didactic reasons. We leave it to our readers to determine if our choices represent the optimal solution to the problem of structuring the subject matter. An overview of the presented methods and their relationships can be found on the inside back cover. This flowchart should – particularly for readers new to the topic – illustrate how one can employ the analytical approaches, from splitting open the cells down to the molecular dimensions. In the diagram, the natural turbulences of the flow are deliberately sacrificed for the sake of clarity. Hopefully, the expert reader will forgive us! At this point we would like to explain a convention in this book, which is not in general use: the use of the terms in vitro and in vivo. To avoid misunderstandings, we explain here that we use these terms as molecular biologists usually understand them, which means using in vitro for “cell free” and in vivo for “in living cells” (in situ translates literally into “in place” and is used and understood as such). In contrast, pharmacologists and physicians often use the term in vivo to refer to experiments in animals and lack suitable terminology to distinguish between experiments conducted in cell culture and those done in test tubes. In cases
Preface
where the meaning may be unclear, we have used the precise term, “cell free”, “in living cells”, or “in animal experiments.” This first edition in English appears some 18 years after the initial publication of Bioanalytik in German. We are happy about it and finally can follow the repeated wish of the scientific community and use English as the lingua franca of the biological sciences. The sustained interest in this book within the Germanspeaking community has led to our desire to make this book available to a wider international audience. To maintain the same length as the original book, despite the addition of new chapters (calorimetry, sensors, and chemical biology), we have shortened or removed other chapters. The goal was to favor current methods and to reduce method descriptions of a more historical nature. It was sometimes hard to sacrifice some cherished memories to better accommodate the current Zeitgeist. We would be grateful to
XVII
our readers if they would point out any inaccuracies or deficiencies in our presentation, which we may have overlooked. As might be expected, this book involved a great deal of work, but was also a great deal of fun to write! We wish to thank our authors at this point, who through their conscientious and diligent work and their cooperation have been a pleasure to work with. Last but not least, we would like to thank our publisher Wiley-VCH and its dedicated team with Waltraud Wüst and the copyeditor John Rhodes, who, with remarkable enthusiasm and tenacity, were our consistent sources of support during the realization of this book.
Joachim W. Engels and Friedrich Lottspeich Munich and Frankfurt, January 2018
Introduction: Bioanalytics − a Science in its Own Right
In 1975, two publications by O’Farrell and Klose aroused the interest of biochemists. In their work, they showed spectacular images with thousands of neatly separated proteins, the first 2D electropherograms. At that time, a vision emerged in the minds of some protein biochemists: It might be possible to identify complex functional relationships, and ultimately to understand processes in the cell, by analyzing these protein patterns. For this purpose, however, the separated proteins had to be characterized and analyzed a task with which the analytics at that time was hopelessly overtaxed. Completely new methods had to be developed, existing ones drastically improved, and the synergies between protein chemistry, molecular biology, genome analysis, and data processing had to be recognized and exploited, so that today’s proteome analysis is on the threshold of realizing that utopian vision of more than 40 years ago. In 1995, an international consortium with strong support from Jim Watson (HUGO; for Human Genome Organization) decided to sequence the human genome. Even though the scientific community was initially divided on the benefits of this endeavor, those involved showed that it is possible to accomplish such a huge undertaking via international co-operation, completing it even ahead of schedule. The competition between commercial and academic participants certainly contributed to the success. Craig Venter is remembered by many for his appearance and his bold claims regarding shotgun sequencing. The major sequencing groups came from the US and Britain, such as the Sanger Institute in Cambridge, England, the Whitehead Institute in Cambridge, Massachusetts, and the Genome Sequencing Center in St. Louis. With the publication in the journal Nature in October 2004 the gold standard of the human genome was completed. The biggest surprise was that the actual number of genes is much lower than expected. With only about 21,000 genes, humans are nowhere near the top of the list with regard to their number of genes; being surpassed, for example, by parsley.
I.1 Paradigm Shift in Biochemistry: From Protein Chemistry to Systems Biology The human genome project has had a fundamental impact on the entire life sciences. We now know that it is technically possible to perform fully automated high-throughput analysis in bioanalytics, and to process the enormous amounts of data that it generates. The results of the genome projects showed that predominantly datadriven research can provide fundamental insights about biology. All of this initiated a profound change from the classical, targetand function-oriented approach to biological questions to a systems-level, holistic perspective.
I.1.1 Classical Approach Following the classical approach of the pre-genomic era, the starting point for almost every biochemical investigation was (and still is) the observation of a biological phenomenon (e.g., the alteration of a phenotype, the appearance or disappearance of enzymatic activity, the transmission of a signal, etc.). Next, an attempt is made to correlate this biological phenomenon to one or a few molecular structures, most often proteins. Once a protein has been isolated that plays a crucial role in the observed biological context, its molecular structure, including its posttranslational modifications, has to be elucidated using state-ofthe-art protein chemistry, so that finally the gene corresponding to this protein can be “fished”. Thus, the whole arsenal of bioanalytics has to be used for the accurate analysis of an important protein. Molecular biological techniques facilitated and accelerated the analysis and validation enormously, and provided hints on the expression behavior of any proteins that were found. Physical methods such as X-ray crystallography, NMR, and electron microscopy allowed deep insights into the molecular
XX
Introduction: Bioanalytics
a Science in its Own Right
structures that sometimes even led to an understanding of biological processes at the molecular level. However, it was quickly recognized that biological effects are rarely explained by the action of a single protein, but are often due to the sequential actions of different biomolecules. Therefore, it was an essential step in the elucidation of reaction pathways to find interaction partners of proteins under scrutiny. When they were found, the same laborious analysis was carried out on them. It is easy to see that this iterative process was quite timeconsuming so that the elucidation of a biological pathway usually took several years. Despite its slowness, the classical approach was incredibly successful. Virtually all our current knowledge of biological processes has been gained using this strategy. It has nevertheless some basic limitations in that it is extremely difficult to elucidate network-like structures or transient interactions and to gain a complete insight into more complex reaction processes of biological systems. Another principal limitation is that the data it yields are rarely quantitative and usually reflect a rather artificial situation. This is inherent in the strategy itself, in which the complex biological system is successively broken down into modules and subunits, moving further and further away from the biological in vivo situation. During the many separation and analysis steps some of the initial material is inevitably lost, which will affect different proteins in different and unpredictable ways. Thus, it becomes virtually impossible to make quantitative statements, which are extremely important for a mathematical modeling of reaction processes.
I.2 Methods Enable Progress Just as two-dimensional gel electrophoresis, DNA sequencing or the polymerase chain reaction opened up hitherto unthinkable levels of knowledge about biological relationships and at the same time spurred the development in their respective fields, methodical developments regularly are at the roots of truly significant advances in science. In the last decades, the life sciences have developed rapidly and revolutionized the understanding of biological relationships. The speed of this development is closely correlated with the development of separation and analysis methods, as shown in the table below. It is almost impossible to imagine modern biochemistry without one or more of these fundamental methodological achievements. Milestones of bioanalytical methodology 1828
urea synthesis
1866
Mendelian laws
1873
Microscopy
1890
crystallization
1894
key-lock principle
1906
chromatography
1907
peptide synthesis
1923
ultracentrifugation
1926
crystallization of urea
1930
electrophoresis
I.1.2 Holistic Strategy
1935
phase contrast microscopy
1937
raster electron microscopy
Encouraged by the success of the human genome project, conceptually new ways of answering biological questions began to be conceived. Instead of analytically dissecting a biological situation and then selectively analyzing the smallest units, the idea was born to view and examine the biological system as a whole (holistic, Greek holos, whole). The same approach is used very successfully, for example in physics, by deliberately disturbing a defined system and observing and analyzing the reaction of the system. This socalled perturbation analysis (Latin perturbare, to disturb) has the enormous advantage that the response of the system can be monitored without any bias, and any observed changes should be directly or indirectly due to the perturbation. This strategy is ideal for highly complex systems. It is amenable to network-like, transient and, above all, unexpected relationships, and being based on the whole system also very close to the real biological situation. However, to fully exploit the benefits of this strategy, the observed changes must be quantitatively measured. Due to the multitude of components in a biological system, this can be a challenge for highthroughput analytics, data processing, and advanced computing. Nevertheless, the methodological developments in bioanalytics and bioinformatics, driven and motivated by genome analysis, have reached a level that has made this kind of holistic analysis of a biological system feasible. It is seen as an essential enabling technology for systems biology, which aims to mathematically describe complex biological processes.
1941
partition chromatography
1944
EPR/ESR spectroscopy
1946
radioistotope labeling
1946
NMR spectroscopy
1950
protein sequence analysis
1953
gas chromatography
1953
DNA double helix
1959
analytical ultracentrifugation
1959
PAGE
1960
hybridization of nucleic acids
1960
X-ray structure analysis
1963
solid phase peptide synthesis
1966
isoelectric focusing
1967
automated sequence analysis
1972
restriction analysis
1974
gene cloning
1974
HPLC of proteins
1975
2D electrophoresis
1975
Southern blotting
1975
monoclonal antibodies
Introduction: Bioanalytics 1976
DNA sequence analysis
1981
site specific mutagenesis
1981
capillary electrophoresis
1982
transgenic animals
1982
scanning tunneling microscopy
1983
automated oligonucleotide synthesis
1985
CAT assay
1986
polymerase chain reaction
1986
atomic force microscopy
1987
MALDI MS
1987
ESI MS
1988
combinatorial chemistry
1990
phosphoimager
1990
cryo-electron microscopy
1991
yeast two-hybrid system
1993
FISH
1993
differential display
1995
proteome analysis
1995
DNA chip
1996
yeast genome sequence
1998
RNA interference
1999
STED microscopy
2004
human genome sequence
2012
CRISPR/Cas
First, separation methods were developed and their application significantly improved. Starting from the simplest separation procedures, extraction and precipitation, the conditions were created to obtain purified and homogeneous compounds via much more effective methods such as electrophoresis and chromatography. The preparation of pure substances in turn exerted an enormous development pressure on the analytical methods. It soon turned out that biomacromolecules have much more complex structures than the hitherto known small molecules. New methods had to be developed, and old ones adapted to the new requirements. To effect a real breakthrough, the methods had to be implemented instrumentally and the instruments had to become commercially available. Since the 1950s both methods and equipment have been developed at an enormous pace. Today, they are sometimes up to 10,000 times faster and more sensitive than when they were introduced. Thanks to state-of-the-art microprocessor controls, the space requirements of the devices are also orders of magnitude lower than those of their ancestors, and their handling has similarly become easier thanks to software-assisted user guidance. While each of these tools may be quite expensive on their own, their higher throughput has led, in effect, to a tremendous cost reduction. This highly dynamic phase of method developments persists to this day. To cite one example, mass spectrometry entered biology and biochemistry, thereby enabling completely new strategies for answering biological questions, such as proteome analysis. Another important example is the
a Science in its Own Right
XXI
success story of bioinformatics, which is used, inter alia, in the analysis of gene or protein databases and which undoubtedly has enormous potential for deployment and development. The advancement of ever-higher resolution light microscopy (near field scanning optical microscope, NFOM and confocal microscopy, 4Pi) now allows molecules to be observed in action in the cell. The well-known passage from the Bible, “because you have seen me, you have believed”, is also applicable to the scientist. All of this clearly shows that we are at the beginning of a phase of transition in which analytics not only has the task of confirming the data of others as an auxiliary science, but can formulate and answer questions on its own accord as a separate, relatively complex area of expertise. Thus, analytics is changing more and more from a purely retrospective to a diagnostic and prospective science. Typical for modern analytics is the interplay of a wide range of individual processes, in which each method is limited in itself, but whose concerted action produces synergisms that can yield answers of astounding and new quality. However, in order to make this synergy possible, a scientist needs to obtain a fundamental knowledge of the areas of application, possibilities and limits of the various techniques.
I.2.1 Protein Analysis Proteins as carriers of biological function normally must be isolated from a relatively large amount of starting material and separated from a myriad of other proteins. A purification strategy that is optimized for good yield while at the same time preserving biological activity is of utmost importance. The purification of the protein itself is still one of the greatest challenges in bioanalytics. It is often time-consuming and demands from the experimenter a substantial knowledge of the separation methods and properties of proteins. Purification is usually accompanied by spectroscopic, immunological and enzymological assays that identify and quantify proteins among a large number of very similar substances, allowing the purification process to be followed and assessed through various steps. Thorough knowledge of classical protein determination methods and enzymatic activity tests is essential, since these methods often depend on the specific properties of the protein to be measured and can be significantly influenced by contaminating substances. Once a protein is isolated, the next step is to obtain as much information as possible about its primary structure, the sequence of its amino acid building blocks. For this purpose, the isolated protein is analyzed directly with sequence analysis, amino acid analysis and mass spectrometry. Often, the identity of the protein can be ascertained at this stage by a database query. If the protein is unknown or needs to be analyzed more closely, for example, to determine post-translational modifications, it is broken down enzymatically or chemically into small fragments. These fragments are usually separated by chromatography and some of them are fully analyzed. The determination of the full amino acid sequence of a protein with protein-chemical methods alone is difficult, laborious and expensive and is usually restricted to the quality control of recombinant therapeutic proteins. In other cases, a few easily accessible partial sequences are usually sufficient. These partial sequences are used for the preparation
XXII
Introduction: Bioanalytics
a Science in its Own Right
of synthetic peptides, which are used in turn to generate monospecific antibodies, or oligonucleotide probes. These probes are used to isolate the gene of interest, ultimately leading to the DNA sequence through DNA analysis that is orders of magnitude faster and simpler than protein sequence analysis. This is translated into the complete amino acid sequence of the protein. However, posttranslational modifications are not detected in this detour via the DNA sequence. However, since they play a decisive role in determining the properties and functions of proteins, they must be subsequently analyzed with all the available high-resolution techniques on the purified protein. These modifications can as in the case of glycosylations be very complex and their structure elucidation is very demanding. Even if one knows the primary structure of a protein, has determined its post-translational modifications and can make certain statements about its folding (secondary structure), one will rarely understand the mechanism of its biological function at the molecular level. To achieve this, a high-resolution spatial structure, obtained by X-ray structure analysis, NMR, or electron microscopy, must be known. Also, the analysis of different complexes (e.g. between enzyme and inhibitor) can yield detailed insight into molecular mechanisms of protein action. Because of the high material requirements, these investigations generally take place via the detour of the overexpression of recombinant genes. Once the entire primary structure, the post-translational modifications and possibly even the spatial structure have been elucidated, the function of a protein often still remains in the dark. Building on an intensive analysis of molecular interaction data, functional analysis is then used to deduce the functional properties from the structures of the substances studied.
I.2.2 Molecular Biology Throughout their development, methods of biochemistry and molecular biology have mutually fertilized and supplemented each other. While molecular biology was initially synonymous with cloning, it has long become an independent discipline with its own goals, methods and results. In all molecular biological approaches, whether in basic research or in diagnostic-therapeutic or industrial applications, the experimenter deals with nucleic acids. Naturally-occurring nucleic acids exhibit a variety of forms; they can be double- or single stranded, circular or linear, of high molecular weight or short and compact, “naked” or associated with proteins. Depending on the organism, the form of the nucleic acid and the purpose of the analysis, a suitable method for their isolation is chosen, followed by analytical methods for checking their integrity, purity, shape and length. Knowledge of these properties is a prerequisite for any subsequent use and analysis of DNA and RNA. A first approximation to the analysis of the DNA structure is provided by restriction endonuclease cleavage. Only this tool enabled the birth of molecular biology about 50 years ago. The restriction endonuclease cleavage is also the prerequisite for cloning, i.e. the amplification and isolation of uniform individual DNA fragments. It is followed by a variety of biochemical analysis methods, most notably DNA sequencing and a variety of hybridization techniques that can identify, localize, and
quantify a particular, large, heterogeneous set of different nucleic acid molecules. The roughly thirty-year-old, truly Nobel Prizeworthy polymerase chain reaction (PCR) has revolutionized the possibilities of analyzing nucleic acids with a principle that is as ingenious as it is simple. Smallest amounts of DNA and RNA can be detected, quantified and amplified without cloning. The imagination of the researcher seems almost unlimited in PCR applications. Because of its high sensitivity, however, it also contains sources of error, which necessitate special caution by the user. Its evolution into a miniaturized fast and cost-effective standardized method is a good example for the lab-on-a-chip of the future. Of course, PCR has also found its way into the sequencing of nucleic acids, one of the classic domains of molecular biology. Nucleic acid sequencing was the basis for the highly sophisticated, international human genome project. Other model organisms were also sequenced in this context. In 2010 there were about 250 eukaryotes and 4,000 bacteria and viruses sequenced, thanks in particular to modern, massively parallel sequencing methods. Many compare the human genome project with the manned flight to the moon (although it did not require similar amounts of money the budget averaged a mere $ 200 million a year for ten years). Like similarly ambitious goals, it has led to significant technical innovations. The methods developed within the human genome project also had a major impact on biotechnology-related industries, such as medicine, agriculture or environmental protection. An analytical method intertwined with the goals of the human genome project is the mapping of specific chromosomal regions, which is done through genetic linkage analysis, cytogenetics, and other physical processes. Mapping is done for genes (i.e. “functional units”) or DNA loci which literally only exist as sequence units. A new approach has emerged with positional cloning, which used to be called reverse genetics. This “reversed” approach to traditional genetics (first gene, then function = phenotype) has already proved its worth in some cases. The most important diseases such as diabetes, cancer, heart attack, depression and metabolic diseases are each influenced by a multitude of genetic and environmental factors. Although two unrelated humans carry about 99.9% identical gene sequences, the remaining 0.1% can be crucial for the success of a therapy. Finding these differences in the gene sequences responsible for the risks offers a great opportunity to understand complex causes and processes of the disease. It will be interesting to find out which base exchanges in which positions contribute to the fact that an individual can tolerate a drug and that this drug also shows the desired effect. It is precisely this connection between the sequence on the one hand and the effect or function on the other that is the focus of functional gene diagnostics. Array diagnostics and siRNA analysis have proven to be very potent tools. Whereas the former detects the presence of the mRNA by hybridization, the latter can establish the connection between RNA and protein. The result is a high-resolution map of the human genome. siRNAs are small double-stranded RNAs (20 27-mers) that can recognize and switch off complementary mRNA. Since there are only about 21,000 genes in humans, high-throughput siRNA analysis is possible and all genes of an organism can be analyzed. For example, all genes of the nematode Caenorhabditis elegans have been RNAi-inhibited, leading to the first complete functional
Introduction: Bioanalytics
gene mapping. Chemical biology, as a discipline of chemistry at the interface to biology, attempts to find small organic molecules for the modulation of protein interactions. Similar to the siRNA approach, cellular functions can be analyzed, but the small organic molecules have the advantage of generating fast responses that are spatially and temporally reversible. Thus, it is possible to find small molecules for all cellular targets which illuminate the physiological correlations and ultimately help to exploit new therapeutic applications. The analysis of the linear structure of DNA is completed by determining DNA modifications, especially base methylation. They influence the structure of DNA and its association with proteins and affect a variety of biological processes. Base methylation is of particular importance for modulating gene activity. In consequence, humans with their comparatively small number of genes have the ability to regulate transcription by methylating the base cytosine. This phenomenon, known as epigenetics, is responsible for the differential expression of the genes in different cells. Because the specific modifications of the genomic DNA are lost in cloning or PCR amplification, their detection must be done directly with genomic DNA; this requires methods with high sensitivity and resolution.
I.2.3 Bioinformatics Even before the human genome project and other sequencing projects had produced a myriad of data, the trend from wet labs to net labs has been increasing over the past thirty years; that is, the activity of some researchers increasingly shifted from the lab bench to computer-related activities. Initially, these were limited to simple homology comparisons of nucleic acids or proteins, in order to elucidate relationships or to obtain clues about the function of unknown genes. Added to this are mathematical simulations, pattern recognition and search strategies for structural and functional elements, and algorithms for weighting and evaluating the data. Databases familiar to the molecular biologist today include not only sequences but also three-dimensional structures. It is remarkable and pleasing that one has free and sometimes interactive access to this vast amount of data and their processing over the internet. This networked information structure and its management is the basis of today’s bioinformatics.
I.2.4 Functional Analysis We have already seen how bioinformatics opens up systematic functional analysis. In this section, we will cover investigations of the interactions of proteins with each other or with nucleic acids. Researchers looked at protein-DNA interactions early in the history of molecular biology after it became clear that genetic trans-factors are mostly DNA-binding proteins. The binding site can be characterized very precisely with so-called footprint methods. In vivo footprints also allow correlating the occupancy state of a genetic cis element with a defined process - for example, active transcription or replication. This can provide information about the mechanism of activation and also about the protein function in the cell.
a Science in its Own Right
XXIII
Interactions between biomacromolecules can also be detected by biochemical and immunological methods, such as affinity chromatography or cross-linking methods, affinity (far-Western) blots, immunoprecipitation, and ultracentrifugation analysis. In these approaches, an unknown partner that interacts with a given protein usually needs to be subsequently identified by protein chemical methods. In genetic engineering this is easier because the interacting partner must first be expressed by a cDNA which has already been cloned. An intelligent genetic technique developed for this purpose is the two-hybrid technique, which can also be used to study interactions between proteins and RNA. It should be kept in mind, however, that the physiological significance of the identified interactions, however plausible they may appear, must be shown separately. Protein DNA, protein RNA, and protein protein interactions initiate a number of processes in the cell, including the expression of certain – as opposed to all genes. The activity of genes expressed only in specific cell types or under specific conditions can be measured by a variety of methods, such as differential display, which is equivalent to a 1: 1 comparison of expressed RNA species. Having found genes which undergo differential expression, their cis and trans elements in other words, the promoter and enhancer elements and the necessary transactivator proteins which effect the this regulation can be determined. For this purpose, functional in vitro and in vivo tests are carried out. Even though all the aforementioned analyses provide a solid insight into the specific expression of a gene and its regulation, the actual function of the gene its phenotype remains unknown. This is a consequence of the era of reverse genetics, in which it has become comparatively easy to sequence DNA and determine “open reading frames”. Correlating an open reading frame or transcription unit with a phenotype is more difficult. Doing so requires an expression disorder of the gene of interest. This gene disorder can be introduced externally, for example by gene modification, that is, by mutagenizing the region of interest. Until about 25 years ago, site-specific mutagenesis was only possible in vivo in microorganisms, by the application of genetic recombination techniques. Since then, various techniques have been optimized to the point that it is possible to introduce genes modified in vitro into higher cells or organisms and to replace the endogenous gene. However, a disruption of the gene or of the gene function can also be achieved by other methods. In this respect, the methods of translational regulation have proven particularly useful. Replacing earlier antisense or antigenic techniques, in which oligonucleotides complementary to certain regions are introduced into the cell and inhibit the expression of the gene, the advent of RNAi in 1998 has ushered in a new era. By appropriate choice of complementary RNA any mRNA can be switched off. It is important to note that this represents not a gene knock-out, but a knock-down. Thus, crucial genes can be downregulated without killing the organism. Instead of down-regulating, the amount of gene product can also be increased by overexpression. This has its significance for agricultural production through transgenic plants and animals. The latter methods gene modification, antisense and RNAi technique and overexpression have been widely used in medicine and agriculture. The reasons are as manifold as they are obvious. Transgenic animals
XXIV
Introduction: Bioanalytics
a Science in its Own Right
or plants can increase agricultural yields. Clinical expression cloning can open up new possibilities for combating malignant cells that are not recognized by the body’s own immune system without the expression of certain surface antigens. The antisense and RNAi technique can be employed to suppress the activation of undesired genes, for example oncogenes. Along these lines another oligonucleotide-based method with great potential in molecular biology is the CRISPR/Cas9 technology. In prokaryotes, Clustered Regulatory Interspaced Short Palindromic Repeats (CRISPR) and their associated cas genes serve as an adaptive immune system to protect them from infection by bacteriophages. The bacterial system was adapted for use in eukaryotic cells and can now be used for precise genome engineering. Compared to the antisense and RNAi approaches discussed above, the CRISPR/Cas9 technology is extremely efficient and easy to use. In addition, the CRISPR/Cas9 system enables the
full knockout of the target gene whereas alternative methods may only result in the partial knockdown of gene expression. However, because an organism is an infinitely more complex system than a controlled in vitro system or a single cell, the desired effect is not always achieved. By way of example, it should be remembered that some of the therapeutic successes of these techniques had nothing to do with nucleic acid hybridization in vivo but, as was later recognized, rather with a local, nonspecific activation of the immune system due to lack of methyl groups on the CpG dinucleotides of the oligonucleotides used or other protein-mediated effects. Such incidents and other, possibly less harmless complications, in our eyes lend more weight to the already existing duty of the researcher as well as the user to pay close attention to what is happening, and what can happen in their work. To that end, a solid knowledge of the available analytical methods and the interpretation of biological correlations is one of several prerequisites. This book aspires to (also) make a contribution towards that goal.
Part I
Protein Analytics
1
Protein Purification Friedrich Lottspeich Peter-Dörfler-Straße 4a, 82131 Stockdorf, Germany
Investigation of the structure and function of proteins has already kept scientists busy for over 200 years. In 1777 the French chemist Pierre J. Macquer subsumed under the term Albumins all substances that showed the peculiar phenomenon of change from a liquid to a solid state upon warming. Chicken egg white, casein, and the blood component globulin already belonged to this class of substances. Already as early as 1787 (i.e., about the time of the French Revolution) the purification of egg white-like (coagulating) substances from plants was reported. In the early nineteenth century many proteins like albumin, fibrin, or casein were purified and analyzed. It soon became apparent that these compounds were considerably more complicated than other organic molecules known at that time. The word protein was most probably introduced by the Swedish chemist Jöns J. von Berzelius in about 1838 and was then published by the Dutch Gerardus J. Mulder. Mulder also suggested a chemical formula, which at that time was regarded as universally valid for all egg white-like materials. The homogeneity and purity of these purified proteins did not correspond of course to today’s demands. However, it became clear that proteins are different and distinct molecules. At that time purification could succeed only if one could use very simple steps, such as extraction for enrichment, acidification for precipitation, and spontaneous crystallization. Already in 1889 Hofmeister had obtained chicken albumin in crystalline form. Although Sumner in 1926 could already crystallize enzymatically active urease, the structure and the construction of proteins remained unknown up to the middle of the twentieth century. Only by the development of efficient purification methods, which allowed single proteins to be isolated from complicated mixtures, accompanied by a revolution in analysis techniques of the purified proteins, was today’s understanding of protein structures possible. This chapter describes fundamental purification methods and also touches on how they can be used systematically and strategically. It is extremely difficult to look at this subject in general terms, because the physical and chemical properties of single proteins may be very different. However, this structural diversity, which in the end determines also the function of the various proteins, is biologically very meaningful and necessary. Proteins – the real tools and building materials of a cell – have to exercise a plethora of different functions.
1.1 Properties of Proteins Size of Proteins The size of proteins can be very different. From small polypeptides, like insulin, which consists of 51 amino acids, up to very big multifunctional proteins, for example, to the apolipoprotein B, a cholesterol-transporting protein which contains more than 4600 amino acid residues, with a molecular mass of more than 500 000 Dalton (500 kDa). Many proteins are composed of oligomers from the same or different protein chains and have molecule masses up to some millions Daltons. Quite in general it is to be expected that, the greater a protein is, the more Bioanalytics: Analytical Methods and Concepts in Biochemistry and Molecular Biology, First Edition. Edited by Friedrich Lottspeich and Joachim Engels. 2018 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2018 by Wiley-VCH Verlag GmbH & Co. KGaA.
The molar mass (M) – often wrongly called molecular weight – is not a mass but is defined as the mass of a substance divided by the amount of substance: M m=n N A mM The unit is g mol
1
Absolute molecule mass (mM) is the molar mass of a molecule divided by the number of molecules in one mol (= Avogadro constant, NA). mM = M/NA. The unit is g. The relative molecular mass (Mr) is defined as the mass of one molecule normalized to the mass of 12 C (carbon 12), which by definition is equal to 12. Mr 12 m
molecule =m
12 C It is dimensionless, but it has been given the “unit” Dalton (Da) (formerly atomic mass unit).
4
Part I: Protein Analytics
Figure 1.1 Separation methods of proteins and peptides. The separation capacity (i.e., the maximal number of compounds that can be separated in a single analysis) of the various separation methods is very different for different molecular masses of the analyte. Abbreviations: SEC, size exclusion chromatography; HIC, hydrophobic interaction chromatography; IEC, ion exchange chromatography; RPC, reversed phase chromatography; CE, capillary electrophoresis.
Dalton (Da), named after the researcher John Dalton (1766–1844), is a non-SI mass unit. One Dalton is equivalent to the atomic mass unit (u = 1/12 of the mass of 12 C) and corresponds roughly to the mass of one hydrogen atom (1.66 × 10 24 g). In biochemistry the unit kDa (1 kilodalton = 1000 Da) is very often used.
Chromatographic Separation Techniques, Chapter 10
Proteome Analysis, Chapter 39
Detergents, Section 1.8
difficultly its isolation and purification will be. This has its reason in the analytical procedures which show very low efficiencies with big molecules. Figure 1.1 shows the separation capacity (the maximum number of analytes, which can be separated under optimal condition) of individual separation techniques against the molecule mass of the analytes. It is evident that for small molecules like amino acids or peptides some chromatographic procedures are clearly able to distinguish more than 50 analytes in a single analysis. In the area of proteins (Mr > 104 Da) one recognizes that of the chromatographic techniques actually only ion exchange chromatography is able to separate efficiently more complicated mixtures. In the molecular mass area of proteins electrophoretic methods are by far more efficient. That is why in proteome analysis (e.g., the analysis of all proteins of a cell), where several thousand proteins have to be separated, electrophoretic procedures (linear and two-dimensional) are very often used. From the figure is also evident that almost no efficient separation procedures exist for large molecules, for example, for protein complexes with molecular masses greater than 150 kDa, or for organelles. The separation efficiency of a method is not always the relevant parameter in a protein purification. If selective purification steps are available the separation capacity is no longer significant and the selectivity becomes the crucial issue. Consequently, an affinity purification, which is based on the specific binding interaction of a substance to an affinity matrix, for example, an immune precipitation or an antibody affinity chromatography, has a quite low separation capacity of 1, but has an extremely high selectivity. Due to this highly selectivity a protein can easily be isolated even from a complex mixture in a one-step procedure. With the most common purification techniques, electrophoresis and chromatography, the analytes must be present in a dissolved form. Thus, the solubility of the protein in aqueous buffer media is a further important parameter when planning a protein purification. Many intracellular proteins, located in the cytosol (e.g., enzymes), are readily soluble while structureforming proteins like cytoskeletal proteins or membrane proteins most often are much less soluble. Especially difficult to handle in aqueous solutions are hydrophobic integral membrane proteins, which are usually surrounded by lipid membranes. Without the presence of detergents such proteins will aggregate and precipitate during the purification.
1 Protein Purification
Available Quantity The quantity available in the raw material plays a crucial role in determining the effort that must be invested for a protein purification. A protein intended for the isolation is present perhaps only as a few copies per cell (e.g., transcription factors) or as a few thousand copies (e.g., many receptors). On the other hand, abundant proteins (e.g., enzymes) can constitute percentage shares of the total protein of a cell. Overexpressed proteins of proteins are often present in in clearly higher quantities (>50% in a cell) as well as some proteins in body fluids (e.g., albumin in plasma >60%). Purification with higher quantities of a protein is usually much simpler. Especially with the isolation of rare proteins different sources of raw material should be checked for the content of the protein of interest. Acid/Base Properties Proteins have certain acidic or basic properties because of their amino acid composition, properties that are used in separations via ion exchange chromatography and electrophoresis. The net charge of a protein is dependent on the pH of the surrounding solution. At a low pH-value it is positive, at high pH negative, and at the isoelectric point it is zero. Positive and negative charges compensate at the latter pH. Biological Activity The purification of a protein is often complicated by the fact that a particular protein often can be detected and localized among the various other proteins only due to its biological activity and location. Hence, one must take into account at every stage of protein isolation the preservation of this biological activity. Usually the biological activity is based on a specific molecular and spatial structure. If it is destroyed, one speaks of denaturation. This often is irreversible. To avoid denaturation, one must exclude in practice the application of some procedures. The biological activity is often stable to different extents under different environmental conditions. Too high or too low buffer concentrations, temperature extremes, contacts with artificial surfaces such as glass or missing cofactors can change biological characteristics of proteins. Some of these changes are reversible: small proteins in particular are, after denaturation and loss of activity, often able to renature under certain conditions, regaining the biologically active form. For larger proteins, this is rarely the case and often results in only a poor yield. Measurement of the biological (e.g., enzymatic) activity makes it possible to monitor the purification of a protein. With increasing purity a higher specific activity is measured. In addition, the biological activity itself can be utilized for the purification of the protein. The activity often goes hand in hand with binding properties to other molecules, such as enzyme–substrate or cofactor, receptor–ligand, antibody, antigen, and so on. This binding is very specific and can be used to design affinity purifications. These are characterized by high enrichment factors and may achieve great efficiency that is difficult to obtain by other techniques. Stability When proteins are extracted from their biological environment they are often markedly impaired in their stability. They may be degraded by proteases (proteolytic enzymes) or associate into insoluble aggregates, which almost always leads to an irreversible loss of biological activity. For these reasons, protease inhibitors are often added in the first steps of an isolation and the purification is carried out quickly and generally at low temperatures. Considering the diversity of the characteristics of proteins it immediately becomes obvious that a protein separation cannot be performed under a single schematic protocol. For a successful isolation strategy a realistic judgement of the behavior of a protein in different separation and purification methods, a minimal understanding of the solubility and charge properties of the protein to be purified, and a clear vision of why the protein is to be purified are absolutely necessary.
Goal of a Protein Purification Above all the first steps of a purification procedure, the level of purity to be aimed at and also the analytics to be used are highly dependent on the intention behind purifying a certain protein. Thus, far higher demands for cleanness must be made with the isolation of a protein for therapeutic purposes (e.g., insulin, growth hormones, or blood coagulation inhibitors) than for a protein that is used in the laboratory for structural investigations. In many cases one wants to isolate a protein only to make an unequivocal identification or to clarify some amino acid sequence segments. For such purposes usually a tiny amount of protein
Enzyme Activity Testing, Chapter 3
Immune Binding, Section 5.3 Protein Purification with Tags, Section 16.2-16.4
Protein Degradation, Chapter 9
5
6
Part I: Protein Analytics
(usually in the microgram range) is sufficient. With the sequence information one is able to identify the protein in protein data banks or it provides the information needed to produce oligonucleotides to isolate the gene corresponding to the protein. The protein can then be expressed in a host organism in much larger quantities (up to gram quantities) than was present in the original source (heterologous expression). Then, many of the other investigations are carried out not with the material from the natural source but with the recombinant protein. New strategic approaches to the analysis of biological questions, such as proteomics and other subtractive approaches, require completely new types of sample preparation and protein isolation. Here it is essential not to change the quantitative relations of the single proteins. A major advantage of these new strategies is that the preservation of the biological activity is no longer so important. Although each protein purification is to be regarded as a unique case, one can still can find, especially for the first purification steps, some general rules and procedures that have already been applied frequently in successful isolations; they will be discussed in detail below.
1.2 Protein Localization and Purification Strategy The first step in any protein purification aims to bring the protein of interest into solution and remove all particulate and insoluble material. Figure 1.2 shows a scheme for different proteins. For the purification of a soluble extracellular protein, cells and other insoluble components must be removed to obtain a homogeneous solution, which can then be subjected to purification methods discussed in the following sections (precipitation, centrifugation, chromatography, electrophoresis, etc.). Sources of extracellular proteins are, for example, culture supernatants of microorganisms, plant and animal cell culture media, or body fluids such as milk, blood, urine, and cerebrospinal fluid. Often, extracellular proteins are present only in relatively low concentrations and demand as the next step an efficient concentration step. To isolate an intracellular protein, the cells must be destroyed in a manner that releases the soluble contents of the cell and keeps the protein of interest intact. Cell disruption methods differ mainly according to cell type and amount of cells. Membrane Proteins and other Insoluble Proteins Membrane-associated proteins are usually purified after isolation of the relevant membrane fraction. For this purpose, peripheral membrane proteins that are bound loosely to membranes are separated by relatively mild conditions, such as high pH, EDTA addition, or lower concentrations of a non-ionic detergent. This fraction of peripheral membrane proteins often can then be treated like soluble proteins. Integral membrane proteins that aggregate outside their membrane via hydrophobic amino acid sequence regions and become insoluble can only be isolated from the membrane by using high detergent
Figure 1.2 Purification scheme for different proteins. According to localization and solubility different purification steps are necessary before any subsequent selective and highly efficient purification steps.
1 Protein Purification
concentrations. At present, they present probably the greatest challenge to the isolation and purification techniques. Proteins that are insoluble in normal aqueous buffers are in general structural proteins (e.g., elastin). Additionally, they are sometimes also crosslinked via post-translationally attached modifications (e.g., functional groups). Here a first and highly efficient purification step is to remove all soluble proteins. Further steps are usually possible only under conditions that destroy the native structure of the proteins. The further processing is often carried out by cleavage of the crosslinking of the denatured proteins and the use of chaotropic reagents (e.g., urea) or detergents. Recombinant Proteins A special situation occurs in the production of recombinant proteins. A rather simple purification is possible by the expression of recombinant proteins in inclusion bodies. These are dense aggregates of the recombinant product, which are present in a non-native state and are insoluble, because the protein concentration is too high, or because the expressed protein in the host environment cannot be correctly folded, or because the formation of the (correct) disulfide bonds in the reducing environment inside the host is not possible. After a simple purification by differential centrifugation (Section 1.5.2), in which the other insoluble cell components are removed, the recombinant protein is obtained in a rather pure form. However, it still needs to be converted into the biologically active state by renaturation. When the expression of recombinant proteins does not result in inclusion bodies, the protein is present in a soluble state inside or outside of the cell, depending on the vector. Here, further purification is similar to the purification of naturally occurring proteins but with the advantage that the protein to be isolated is already present in relatively large amounts. Recombinant proteins can be easily isolated by using specific marker structures (tags). Typical examples are fusion proteins in which at the DNA level the coding regions for a tag structure and the desired protein are ligated and expressed as a single protein. Such fusion proteins often can be isolated in a rather pure form in a one-step procedure on applying a specific antibody affinity chromatography against the tag structure. Examples are GST fusion proteins with antibodies against GST or biotinylated proteins using avidin columns. Another frequently used tag-structure is multiple histidine residues, which are attached to the N- or Cterminal end of the protein chain and are easy to isolate by immobilized metal affinity chromatography (IMAC).
1.3 Homogenization and Cell Disruption To purify biological components of intact tissues, the complex cell associations must be disrupted in a first step by homogenization. The result is a mixture of intact and disrupted cells, cell organelles, membrane fragments, and small chemical compounds derived from the cytoplasm and from damaged subcellular compartments. Since the cellular components are transferred to a non-physiological environment, the homogenization media should meet several basic requirements:
protection of the cells from osmotic bursting, protection from proteases, protection of the biological activity (function), prevention of aggregation, minimal destruction of organelles, no interference with biological analyses and functional tests.
Normally this is done by isotonic buffers at neutral pH. Often, a cocktail of protease inhibitors is added (Table 1.1). If you want to isolate intracellular organelles, such as mitochondria, nuclei, microsomes, and so on, or intracellular proteins, the still intact cells have to be disrupted. This is accomplished by mechanical destruction of the cell wall. This procedure releases heat of friction and therefore has to be carried out with cooling. The technical realization of the disruption process varies depending on the starting material and location of the target protein (Table 1.2).
Protein Interaction, Section 16.2–16.4
Immobilized Metal Affinity Chromatography, Section 10.4.8
7
8
Part I: Protein Analytics Table 1.1 Protease inhibitors. Substance
Concentration
Inhibitor of
Phenylmethylsulfonyl fluoride (PMSF)
0.1–1 mM
Serine proteases
Aprotinin
0.01–0.3 μM
Serine proteases
ε-Amino-n-caproic acid
2–5 mM
Serine proteases
Antipain
70 μM
Cysteine proteases
Leupeptin
1 μM
Cysteine proteases
Pepstatin A
1 μM
Aspartate proteases
Ethylenediaminetetraacetic acid (EDTA)
0.5–1.5 mM
Metalloproteases
For very sensitive cells (e.g., leukocytes, ciliates) repeated pipetting of the cell suspension or pressing it through a sieve is sufficient to achieve a disintegration by surface shear forces. For the slightly more stable animal cells, the shear forces are generated with a glass pestle in a glass tube (Dounce homogenizer). These methods are not suitable for plant and bacterial cells.
Cells that have no cell wall and are not associated (e.g., isolated blood cells) can be broken
osmolytically by being placed in a hypotonic environment (e.g., in distilled water). The water penetrates into the cells and causes them to burst. In cells with cell walls (bacteria, yeasts) the cell walls must be treated enzymatically (e.g., with lysozyme) before an osmolytic digestion can succeed. Such exposure is very gentle and is therefore particularly suitable for the isolation of nuclei and other organelles. For bacteria repeated freezing and thawing is often used as a disruption method. By changing the aggregate state the cell membrane is deformed so that it breaks and the intracellular content is released. Microorganisms and yeasts can be dried at 20–30 °C in a thin layer for two to three days. This leads to destruction of the cell membrane. The dried cells are then ground in a mortar and can be stored at 4 °C if necessary also for longer periods. Soluble proteins can be extracted with an aqueous buffer from the dry powder in a few hours. With cold, water-miscible organic solvents (acetone, –15 °C, ten-times volume) cells can be quickly drained, with the lipids extracted into the organic phase, and thus the cell walls are destroyed. After centrifugation, the proteins remain in the precipitate, from where they can be recovered by extraction with aqueous solvents. With stable cells such as plant cells, bacteria, and yeasts a mortar and pestle can be applied for cell disruption, although larger organelles (chloroplasts) may be damaged. The addition of an abrasive (sea sand, glass beads) facilitates the disruption. For larger quantities, a rotating knife homogenization can be used. The tissue is cut by a rapidly rotating knife. As this produces considerable heat a way of cooling should be present. For small objects such as bacteria and yeasts, the efficiency of the pulping process is significantly improved by the addition of fine glass beads. Vibration cell mills are used for a relatively harsh disruption of bacteria. These are lockable steel vessels in which the cells are vigorously shaken with glass beads (diameter 0.1–0.5 mm). Again, the heat generated must be dissipated. Cell organelles can be damaged in this decomposition method. Rapid changes in pressure break cells and organelles in a very efficient manner. Therefore, strong pressure changes are produced in the suspension of a cell material with ultrasonic waves in the frequency range 10–40 kHz through a metal rod. Since in this method much heat is released, only relatively small volumes and short sound pulses with a maximal duration of 10 s should be applied. DNA is fragmented under these conditions. In a further disruption method that is particularly suitable for microorganisms, up to 50 ml of a cell suspension are pressed through a narrow opening (1.3
Nuclear membrane
3–12
1.18–1.22
1500g/15 min
Plasma membrane
3–20
1.15–1.18
1500g/15 min
Golgi-apparatus
1
1.12–1.16
Mitochondria
0.5–4
6
2000g/20 min
1.17–1.21
1 × 10 –5 × 10
4 4
4
10 000g/25 min 10 000g/25 min
Lysosomes
0.5–0.8
1.17–1.21
4 × 10 –2 × 10
Peroxisomes
0.5–0.8
1.19–1.4
4 × 103
10 000g/25 min
0.05–0.3
1.06–1.23
1 × 103
150 000g/40 min
1.55–1.58
70–80
1.2–1.7
1–25
3
Microsomes Endoplasmic reticulum
100 000g/1 h
Ribosomes Soluble proteins
0.001–0.01
for the enrichment of particles but also for concentration. Thus, for example, from one liter of bacterial cell culture the cells can be pelleted by centrifugation in 15 min at 2000g and then can be resuspended in a smaller volume. Zonal Centrifugation If the sedimentation rates of molecules do not differ sufficiently, the viscosity and density of the medium can be used to generate selectivity. In the zonal centrifugation a preformed flat density gradient, mostly from sucrose, is used and the sample layered over the gradient (see below). The particles which at the beginning of the centrifugation – in contrast to differential centrifugation – are present in a narrow zone are now separated according to the sedimentation velocity. The density gradient, in addition to the minimization of convection, also has the effect that at increasing density and viscosity those faster particles are slowed down that would otherwise sediment with the increasing RCF caused by increasing distance from the rotor axis. This gives an approximately constant rate of sedimentation of the particles. Zonal centrifugation, which is usually carried out at relatively low speeds with swinging-or vertical
Figure 1.5 Density and sedimentation coefficients of some cell compartments. The figure shows the distribution of different cell components in terms of their density and their sedimentation coefficients. ER, endoplasmic reticulum.
14
Part I: Protein Analytics
rotors, is an incomplete sedimentation; the maximum density of the medium must not exceed the lowest density of the particles. The centrifugation is stopped before the particles pellet. Isopycnic Centrifugation The previously discussed techniques of differential and zonal centrifugation are especially suitable for the separation of particles that differ in size. These techniques are not well suited for particles having a similar size but different densities. For these cases, isopycnic centrifugation (also known as sedimentation equilibrium centrifugation) is used. Here centrifugation is performed for long periods at high speed in a density gradient until equilibration. According to Stokes’ equation, particles remain in the floating state when their density and the density of the surrounding medium are equal (v = 0). Particles in the upper part of the centrifuge tube sediment until they reach the state of suspension and cannot sediment further because the layer below has a greater density. The particles in the lower region rise accordingly up to the equilibrium position. In this type of centrifugation, the gradient density must exceed the density of all particles to be centrifuged. Density Gradient To generate the density gradient, which can be continuous or discontinuous (in stages), various media are used, which have been found for the different application areas as appropriate:
CsCl solutions can be prepared with densities up to 1.9 g ml 1. They are of very low
viscosity, but have the drawback of high ionic strength, which can dissociate some biological materials (chromatin, ribosomes). In addition, CsCl solutions have high osmolality, which makes them unsuitable for osmotically sensitive particles such as cells. CsCl gradients are particularly suitable for the separation of nucleic acids. Sucrose is often used for the separation of subcellular organelles in zonal centrifugation. The inexpensive and easy to prepare solutions are nonionic and relatively inert to biological materials. The low density of isotonic sucrose solutions (0.5% Triton X-100 >0.1% SDS sodium deoxycholate
UV methods
pigments phenolic compounds organic cofactors
2 Protein determination
Additional Methods, not described in detail In addition to the methods described in this chapter, the titrimetric determination of nitrogen by the Kjeldahl method and the ninhydrin assay can be used for the quantification of proteins after acid, thermal degradation of the proteins. Neither method is addressed in detail as they require a great deal of effort, but are discussed here briefly for the sake of completeness. Using the Kjeldahl method, organic nitrogen compounds are oxidized to CO2 and H2O and create an equivalent amount of NH3. Defined conditions are used for the heating with concentrated sulfuric acid and a catalyst (heavy metals, selenium). The NH3 obtained is bound by H2SO4 as (NH4)2SO4 (wet ashing). After the addition of NaOH, ammonia is released, transferred to a distillation apparatus and quantified by titration. The nitrogen content of proteins is approximately 16%. Therefore, multiplying the N-content determined by 6.25 can recalculate the amount of protein. Obviously, the non-protein nitrogen must be have been previously removed. The color assay with ninhydrin is used as a detection method for free amino groups. Therefore, the protein must, firstly, be hydrolyzed into its free amino acids. Exemplarily, this is realized by boiling in 6% sulfuric acid at 100 °C (12–15 h) in fused glass vessels in the absence of oxygen. The ninhydrin reagent is added to the protein hydrolyzate and the resulting purple– blue solution is measured spectrophotometrically at a wavelength of 570 nm. In most cases, Lleucine is used as a standard to generate the calibration curve. However, the color intensities, resulting from the different amino acids of the protein, are not identical. This is one of several sources of error in the ninhydrin method.
25
Ninhydrin Assay, Section 13.3.1
2.1 Quantitative Determination by Staining Tests Protein samples often consist of a complex mixture of different proteins. The quantitative determination of the protein content of such crude protein solutions is usually based on the color reactions of functional groups of proteins with dye-forming reagents. The intensity of the dye correlates directly with the concentration of the groups reacting and can be accurately measured with a spectrophotometer. The basics of spectroscopy (Lambert–Beer law, etc.) and the appropriate equipment are described in detail in Chapter 7. There are several variants for the four following staining methods, which are described in the literature. However, they are all based on the same principles. Spectral Absorption Coefficients Each staining method can only be used in a certain concentration range. Within this range, a constant dependency of the absorption measured on the protein concentration results (at a defined wavelength). The spectral absorption coefficient is determined graphically as the slope in the plot of absorbance versus concentration (abscissa). By default, the absorbance value is related to the path length of the cuvette (in cm) and the concentration of the dissolved protein value in micrograms per milliliter. Alternatively, with a known molecular weight of the protein, the concentration unit mole of dissolved protein per milliliter may be used. Then, a molar spectral absorption coefficient results (formerly molar extinction coefficient) with the units: 1/(moles of dissolved protein per liter) per cm or liters per mole of dissolved protein per cm. The requirements for the staining methods presented (protein concentration ranges, sample volumes) and the approximate resulting spectral absorption coefficient (ml final volume per microgram of protein dissolved per cm), with bovine serum albumin as a standard, are shown in Table 2.2 as an overview. Approximate values are presented in the table because spectral absorption coefficients between 2.3 and 3.2 ml final volume per microgram of protein in solution per cm can be found under apparently identical conditions in the literature (e.g., solely for the biuret-assay)! This is caused by the complexity of influencing factors such as the purity of the chemicals and the water used. Relative Deviations of the Staining Methods Ideally, non-proteinogenic impairment of the assays can be excluded and, apart from a few exceptions, appear under the determination methods presented for one and the same protein, deviating between at least 5% and 20%. The difference is even more dramatic for the quantification of crude protein solutions. It is extremely important
Multiple determinations should be performed in all cases. Triplicate measurements are usually realized and the mean value calculated. The samples are generally measured at the same wavelength against a so-called blank approach, which consists of the same ingredients and volumes as the respective color assay but the protein solution is replaced by distilled water.
Physical Principles of Spectroscopy, Section 7.1
It is very important for all staining methods to specify what the volume (ml) stands for. Several solutions have to be combined in different volumes with the protein sample, depending on the method. The volume specified should always stand for the final volume of the test approach after performing the assay and not the volume of the protein solution used.
26
Part I: Protein Analytics Table 2.2 Overview of the most common staining methods for protein determination. Method
Approx. sample volume required (ml)
Limit of detection (μg-protein ml 1)
Spectral absorption coefficienta) (ml final volume per μg dissolved protein per cm)
Biuret assay
1
1–10
2.3 × 10
4
A550 A650
Lowry assay (modified according to Hartree)
1
0.1–1
1.7 × 10
2
Bicinchoninic acid assay
0.1
0.1–1
1.5 × 10
2
A562
Bradford assay
0.1
0.05–0.5
4.0 × 10
2
A595
a) With the standard protein bovine serum albumin.
when reporting the specific activities of enzymes, antibodies, or lectins, expressed as biological activity per mg of protein, not only to state under which test conditions (e.g., substrate, pH, temperature) the activity was determined but also which method was used for the protein determination.
2.1.1 Biuret Assay
Figure 2.1 The colored protein Cu2+ complex that occurs in the biuret reaction.
This protein determination method is based on a color reaction with dissolved biuret (carbamylurea) and copper sulfate in an alkaline, aqueous environment (biuret reaction). The result is a red-violet colored complex between the Cu2+ ions and two molecules of biuret. The reaction is typical of compounds with at least two CO-NH groups (peptide bonds) and can, therefore, be used for the colorimetric detection of peptides and proteins (Figure 2.1). If tyrosine residues are present, they also contribute significantly to the dye formation by the complexation of copper ions. Thus, the detection is mainly oriented objectively on the peptide bonds and subjectively on the tyrosine residues of proteins. The spectral absorption coefficient given in Table 2.2 was determined at 550 nm. Otherwise, the color intensity can also be measured at 540 nm. Both wavelengths are close to the absorption maximum of the color complex. The absorption maximum varies slightly from protein to protein. The biuret assay is the least sensitive of the color assays (Table 2.2). The protein sample or standard sample is mixed with four parts of biuret reagent and allowed to stand for 20 min at room temperature. Then, the color intensity is directly measured in a spectrophotometer. Ammonium and weak reducing and strong oxidizing agents act especially in a disturbing way (Table 2.1). However, small amounts of sodium dodecyl sulfate (SDS) or other detergents are tolerable. If the solution has to be diluted due to the high absorption, this must done with the sample solution used and not with the final solution after color formation. The color formation reaction must be repeated. This ensures that the required amount of copper ions are present, due to the concentration-dependent equilibrium, which is necessary for complete saturation settings of the complex-forming groups.
2.1.2 Lowry Assay Lowry et al. published a method for the quantitative analysis of proteins in 1951. The method is a combination of the biuret reaction and the Folin–Ciocalteau phenol reagent and is referred to as the Lowry assay. The copper–protein complex mentioned above is formed in alkaline solution. This supports the reduction of molybdate or tungstate, which are used in the form of their heterogenic polyphosphoric acid (Folin–Ciocalteau phenol reagent), by primarily tyrosine, tryptophan, and, to a lesser extent, cysteine, cystine, and histidine of the protein. Here, presumably, Cu2+ of the copper–protein complex is reduced to Cu+, which subsequently reacts with the Folin–Ciocalteau phenol reagent. Due to the additional color reaction, the sensitivity increases enormously compared to the pure Biuret assay. The resulting deep blue color is measured at a wavelength of 750, 650, or 540 nm.
2 Protein determination
27
Various modifications for the Lowry assay are described in the literature. The aim was mostly to improve the relatively high breakdown susceptibility of the Lowry method. The data presented in Tables 2.1 and 2.2 were obtained using a published version of Hartree (1972). The modified method extends, by the same sensitivity, the linear range, compared to the conventional Lowry assays, by 30–40% to about 0.1–1.0 mg ml 1 (Table 2.1). The method shows no problems with dropout salts and used only three stock solutions, which also have better storage stability, instead of the five stock solutions of the original Lowry assay. In this variant, three reagents (parts A : B : C = 0.9 : 0.1 : 3.0) are added successively to one amount of protein sample (1.0): A (carbonate/NaOH solution), B (alkaline CuSO4 solution), and C (diluted Folin–Ciocalteau reagent). After the addition of A and C, the mixture is heated to 50 °C for 10 min each time. Overall, the Lowry assay according to Hartree takes about 30 min. Any necessary dilutions must, as already explained for the biuret assay, be performed with the protein solution. The Lowry method is affected by a wide range of non-proteinogenic substances (Table 2.1). In particular, the usual additives for enzyme purification, such as EDTA, ammonium sulfate, or Triton X-100, are not compatible with the Lowry assay. Compared with the biuret assay, subjective criteria contribute more intensely to dye formation – in particular the individual rates, depending on the protein, of tyrosine, tryptophan, cysteine, cystine, and histidine. Again, the staining is relatively unstable. The measurement of the samples should be carried within 60 min of the last reaction step.
2.1.3 Bicinchoninic Acid Assay (BCA Assay) Smith and colleagues published a highly regarded alternative to the Lowry assay in 1985 that combines the biuret assay with bicinchoninic acid (BCA) as the detection system. Hitherto, BCA had been used for the detection of other copper-reducing compounds, such as glucose or uric acid. Twenty parts of a freshly prepared bicinchoninic acid/copper sulfate solution is added to one part of sample and incubated for 30 min at 37 °C. Similar to the Lowry assay, the method is based on the reduction of Cu2+ to Cu+. The BCA forms a color complex specifically with Cu+ (Figure 2.2). This allows a sensitive, colorimetric detection of proteins at a wavelength of 562 nm (the absorption maximum of the complex). Comparisons with the Lowry assay showed that cysteine, cystine, tyrosine, tryptophan, and the peptide bond reduce Cu2+ to Cu+ and, therefore, allow the color formation with BCA. The intensity of the color formation and the redox behavior of the groups involved depend on, among other things, the temperature. Thus, the BCA assay can be varied between different temperatures to obtain the sensitivity desired. The BCA and Lowry assays are in good agreement for the determination of the concentrations of standard proteins, such as bovine serum albumin, chymotrypsin, or immunoglobulin G. Significant deviations of almost 100% were determined with avidin, a glycoprotein from chicken egg-white. The mechanism of the BCA assay is similar in principle to that of the Lowry assay. However, in no cases are they equivalent. The advantages of the BCA assay over the Lowry assay are the simpler implementation, the ability to influence the sensitivity, and the good stability over time of the color complex formed. The disadvantage is the higher price of the assay, due to the high price of the sodium salt of bicinchoninic acid. The sensitivity of the BCA assay is in the range of the Lowry assay modified by Hartree (Table 2.2). The breakdown
Figure 2.2 The bicinchoninic acid assay: a combination of the biuret reaction with the selective bicinchoninic acid complexation of Cu+.
28
Part I: Protein Analytics
Figure 2.3 Coomassie Brilliant Blue G250 (as sulfonate), the reagent of the Bradford assay.
susceptibility of the BCA assay is also quite high. In addition to the substances listed in Table 2.2, further substances, such as small amounts of ascorbic acid, dithiothreitol, or glutathione, complexing and reducing compounds, interfere with the assay.
2.1.4 Bradford Assay
Protein Detection in Electrophoresis Gel, Section 11.3.3
In contrast to the dyeing methods described so far, no copper ions are involved in this assay. It is named after M.M. Bradford and was published in 1976. The focus is on blue acid dyes, which are called Coomassie Brilliant Blue. In many cases, Coomassie Brilliant Blue G 250 (Figure 2.3) is used. The absorption maximum of Coomassie Brilliant Blue G 250 shifts from 465 to 595 nm in the presence of proteins and in an acidic environment. The reason for this is probably the stabilization of the dye in its unprotonated, anionic sulfonate form by complex formation between the dye and protein. The dye binds fairly nonspecifically to cationic and nonpolar, hydrophobic side chains of proteins. The interactions with arginine, and less so with lysine, histidine, tryptophan, tyrosine, and phenylalanine, are most important. The Bradford assay is also used for staining proteins in electrophoresis gels. It is approximately a factor of two more sensitive than either the Lowry or BCA assay (Table 2.2) and is, thus, the most sensitive quantitative staining assay. It is also the simplest assay, because the stock solution, consisting of dye, ethanol, and phosphoric acid, is added to the sample solution in a ratio of 20-to-50 : 1 and, after 10 min at room temperature, the measurement of the absorbance at 595 nm can be started. Another advantage is that several substances that interfere with the Lowry or BCA assay do not affect the result of the Bradford assay (Table 2.1); this is especially the case concerning the tolerance to reducing agents! On the other hand, all substances that affect the absorption maximum of Coomassie Brilliant Blue are disturbing. This is sometimes difficult to estimate beforehand due to the lack of specificity of the interactions. The biggest disadvantage of the Bradford assay is that equal amounts of different standard proteins can cause significant differences in their resulting absorption coefficient. Thus, the subjectivity of this color assay is considerable and is bigger than that of the three other more complex staining methods.
2.2 Spectroscopic Methods Spectroscopic Bases and Measurement Techniques, Section 7.1
Spectroscopic methods are less sensitive and require higher concentrations of protein than colorimetric methods. These spectroscopic methods should be used with purer or high-purity protein solutions. The spectral absorption or emission properties of the proteins at a defined wavelength in an optical pathway are measured. Therefore, the protein solution (sample solution) is placed in a quartz cuvette (optical pathway in the cell: usually 1 cm). The spectrophotometer is previously calibrated with pure, protein-free solvent in the same quartz cuvette and set at zero as a reference. Subsequently, the value of the sample solution measured will result, either based on literature tables or the calibration curve, in the corresponding protein concentration in mg ml 1. The latter
2 Protein determination
29
Table 2.3 Overview of the most common spectroscopic protein determination methods. Protein component on which the determination is essentially based
Limit of detection (μg-protein ml 1)
Dependence on protein composition
Susceptibility to interference
A280
Tryptophan, tyrosine
20–3,000
Strong
Low
A205
Peptide bonds
1–100
Weak
High
Tryptophan (tyrosine)
5–50
Strong
Low
Method
Photometry:
Fluorimetry:
excitation280 emission320–350
is recommended due to the interference effects of buffer substances, pH values, inaccuracies of devices, and so on. Ideally, a calibration curve is obtained with the pure protein of interest as a standard. An overview of the detailed spectroscopic methods discussed below is given in Table 2.3.
2.2.1 Measurements in the UV Range Absorption Measurement at 280 nm (A280) Warburg and Christian measured the protein concentration of cell extract solutions of different purification degrees at a wavelength of 280 nm (A280) in the early 1940s. The aromatic amino acids, tryptophan and tyrosine, and to a lesser extent phenylalanine, absorb at this wavelength (Table 2.4). Since larger amounts of nucleic acids and nucleotides are present in the protein solutions – this is generally the case after digestion of cells – the values measured at A280 had to be corrected. This is because the nucleic acid bases also absorb at A280. Thus, Warburg and Christian determined a second value at 260 nm (A260), which was correlated with A280 according to the following formula: Protein concentration mg ml 1
1:55A280 −
0:76A260
(2.1)
This relationship can be used up to 20% (w/v) nucleic acids in solution or an A280/A260 ratio of less than 0.6. The A280 measurement alone is sufficient in protein solutions with only a low content of nucleic acids. In accordance with the molar spectral absorption coefficient (ε) (Table 2.4), the A280 method is based essentially on tryptophan, which has an absorption maximum at 279 nm. The two other aromatic amino acids contribute relatively less to the A280 value. Since the content of aromatic amino acids can vary from protein to protein, the corresponding A280 values also vary. At a concentration range of 10 mg ml 1 (A1%), the A280 value of most proteins is between 0.4 and 1.5. However, there are extreme exceptions where A1% is 0.0 (parvalbumin) or 2.65 (lysozyme). An ideal standard protein should have the same level of aromatic amino acids as the protein measured, or should be identical with it. Unfortunately, this is extremely rarely realizable in practice. The A280 method can be used for protein concentrations between 20 and 3000 μg ml 1. It is an easy and fast method and is a lot less disturbed by parallel absorption of non-protein substances Table 2.4 Molar spectral absorption coefficient (ε) at 280 nm and absorption maxima of aromatic amino acids.a) Amino acid
ε × 10
Tryptophan
5.559
219, 279
Tyrosine
1.197
193, 222, 275
Phenylalanine
0.0007
188, 206, 257
a) For aqueous solutions at pH 7.1.
3
(l mol
1
cm 1) at 280 nm
Absorption maxima (nm)
The absorbance value determined (sample or standard) should not exceed 1.0. At a value greater than 1.0, the linear dependency of spectral absorption on the concentration is no longer given. The emission value for fluorescence measurements should not exceed 0.5. If necessary, the sample solution must be diluted and the dilution factor has to be taken into consideration when determining the concentration.
30
Part I: Protein Analytics Table 2.5 Maximum concentration of disturbing additives allowed by the A205 and A280 method. The additives are often used in protein chemistry.a) Additive
A205 method
A280 method
Ammonium sulfate
9% (w/v)
>50% (w/v)
Brij 35
1% (v/v)
1% (v/v)
Dithiothreitol (DTT)
0.1 mM
3 mM
Ethylenediaminetetraacetic acid (EDTA)
0.2 mM
30 mM
Glycerol
5% (v/v)
40% (v/v)
Urea
1 M
KCl
50 mM
100 mM
2-Mercaptoethanol
1 M
NaOH
25 mM
>1 M
Phosphate buffer
50 mM
1M
Sucrose
0.5 M
2M
SDS (sodium dodecyl sulfate)
0.1% (w/v)
0.1% (w/v)
Trichloroacetic acid (TCA)
N↓, the vector points along the +z-axis. The x- and y-components are uniformly distributed on the surface of the double precession cone and do not produce a macroscopic magnetization.
(18.6)
where k is the Boltzmann constant and T is the absolute temperature. Because the energy difference between both levels is orders of magnitude smaller than the thermal energy (kT), both levels are almost equally occupied. For example, for protons at 300 K and a magnetic field of 18.8 T (800 MHz) the excess population in the lower energy level amounts to only 1.3 in 10 000 particles (i.e., N↓ = 0.99987 × N↑). This is the main reason why NMR spectroscopy is so insensitive compared with other spectroscopic methods. Even this tiny difference suffices to give rise to a macroscopic bulk magnetization M0, which results from the combined magnetic moments of the individual nuclear spins. The macroscopic magnetization of spin ½ particles is then given by: M0 N
γ 2 ħ2 B0 I
I 1 γ 2 ħ2 B0 N ; 3kT 4kT
for I
1 2
(18.7)
This shows that the magnitude of the equilibrium magnetization depends on the magnetic field strength B0, the number of spins N, and the temperature T of the sample. Varying each of these parameters can enhance the observable signal (which is the reason for the ongoing development of magnets with increasing field strengths). Importantly, the spectrometer records the time development of exactly this macroscopic bulk magnetization. Moreover, the classical theory of NMR spectroscopy normally considers this macroscopic magnetization as due to its more descriptive behavior than to that of individual spins. In equilibrium, magnetization M0 exists only along the axis of the main field (by convention the z-direction, i.e., Mz = M0). The transverse x- and y-components of the magnetic moments are uniformly distributed and add up to zero (Mx = My = 0) (Figure 18.3). The Bloch Equations According to the Bloch equation, the change of the magnetization vector M over time results from the interaction of the magnetization with an effective magnetic field Beff : dM γ
M Beff ; dt
Beff
ω0 B0 B1 B1 γ |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl}
(18.8)
0
To simplify the mathematical treatment of the magnetization vector, we will introduce a rotating coordinate system, which precesses about the z-axis with the Larmor frequency of the nuclei (ω0 = γB0). In this coordinate system all nuclear spins rotating with the Larmor frequency appear stationary. This concept should be familiar for us because we all live in a rotating coordinate system — the Earth. A person standing on the equator moves for an observer in space with a speed of approx. 1700 km h 1. Consider that this person on earth throws a ball straight up into the air and then sees it falling down again. For him or her the ball moves on a simple straight vertical path; however, for our observer in space this ball would move on a complicated trajectory. In the rotating coordinate system the contribution of the main magnetic field B0 to Beff cancels for nuclei with Larmor frequency ω0. Moreover, if only the main magnetic field B0 is applied, then also Beff is zero and the magnetization vector becomes time-independent. However, if an additional field B1 is applied perpendicular to the main magnetic field B0 , then Beff B1 . The magnetization vector will now precess around the axis of the B1 field if its frequency matches the Larmor frequency of the nuclei (ωrf = ω0, the resonance condition). Thus, the B1 field induces a rotation of the magnetization vector Mz from the equilibrium position to the transverse plane (cross product). Physically, the B1 field is nothing other than a
18 Magnetic Resonance Spectroscopy of Biomolecules
437
short radio frequency pulse. For example, a so-called excitation pulse or 90°-pulse completely converts z- into y-magnetization if the B1 field along the x-axis is of appropriate strength and duration (Figure 18.4). How can we imagine the origin of the y-magnetization for individual spins after a 90°-pulse? (i) Both energy levels are equally populated because Mz = 0. (ii) The magnetization dipoles of individual spins are not uniformly distributed around the z-axis, but small parts precess bundled about the z-axis. It is this state of phase coherence that gives rise to the macroscopic ymagnetization (Figure 18.5). Relaxation The Bloch equation shown above is incomplete because it predicts an infinitely long precession of the magnetization vector once the sample has been excited. However, the transverse magnetization after the 90°-pulse corresponds to a non-equilibrium state from which the system will return to its thermodynamic equilibrium after a short while. Therefore, Bloch defined two different relaxation time constants denoted as longitudinal (T1) and transverse (T2) relaxation time constants. Assuming that the respective relaxation processes follow first-order rate equations, the Bloch equations are modified to: dMx;y γ
M Bx;y dt
Mx;y T2
dMz M 0 Mz γ
M Bz dt T1
(18.9)
Figure 18.4 Effect of a 90°-pulse on zmagnetization. A pulse about the x-axis (bold wavy arrow) rotates the equilibrium z-magnetization (grey arrow) by 90° counterclockwise around the x-axis and creates –y-magnetization (grey arrow).
(18.10)
The equations indicate that due to relaxation the transverse components (Mx, My) decay to their equilibrium value of zero, whereas longitudinal magnetization Mz approaches its equilibrium value M0. In high resolution NMR spectroscopy, the T1 relaxation time constants for protons are in the range of one to several seconds. Usually, T2 is similar to T1 (small molecules); however, for large molecules like proteins T2 is much smaller than T1. From this fact results the wellknown size limitation of NMR spectroscopy to proteins, because the enhanced T2 relaxation in large proteins decreases the resolution and sensitivity of their NMR spectra (Section 18.1.2, Spectral Parameters). Notably, relaxation effects can often be ignored during the radiofrequency pulses, because relaxation times are long compared to commonly used pulse lengths (10–50 μs). Relaxation is caused by different time-dependent interactions (e.g., dipolar couplings) between the spins and their environment (T1) and between the spins themselves (T2). Historically, T1 is also referred to as the spin–lattice and T2 as the spin–spin relaxation time constant. The relaxation time constants depend on several factors including the Larmor frequency (or magnetic field strength) and the molecular mobility of the molecule in solution. The latter is characterized by the rotational correlation time, τ c. We will discuss the measurement of relaxation times in more detail later during the analysis of protein dynamics (Section 18.1.7, Determination of Protein Dynamics). Pulsed Fourier Transformation Spectroscopy Modern NMR spectrometers operate with a technique called pulsed Fourier Transformation NMR (FT-NMR) spectroscopy. This technique replaced older NMR methods (e.g., continuous wave NMR spectroscopy) because it greatly improved the sensitivity and resolution, and also facilitated the development of multidimensional NMR methods (Sections 18.1.3 and 18.1.4). In pulsed FT-NMR spectroscopy all nuclei are excited simultaneously through a radio frequency pulse. The radio transmitter works at a fixed frequency ν0 and would therefore excite only nuclei with this Larmor frequency (resonance condition!). However, the pulse duration is inversely proportional to the frequency bandwidth (and thus the energy of the radiation). If therefore a very short pulse (a few microseconds) is emitted, then the pulse is less “frequency selective.” It contains a broad excitation band around ν0 and excites the Larmor frequencies of all nuclear spins in the sample at once. Strictly speaking, the flip angle through which the pulse rotates the bulk magnetization depends on the offset (or distance) of the Larmor frequencies from the transmitter frequency. For nuclei that are off-resonance, the effective field Beff is not collinear with the B1 field as in the resonant case. As a result, the flip angle for off-resonance nuclei decreases with increasing offset. However, the projection of the transverse magnetization on the y-axis depends on the sine
Figure 18.5 Illustration of transverse magnetization. The identical number of nuclear spins (grey arrows) in both energy levels shows that they are equally populated. Some nuclear spins precess bundled (or in phase) about the direction of the B0 field. Their magnetic moments add up to the macroscopic My-magnetization (bold arrow).
438
Part II: 3D Structure Determination
of the flip angle. For example, even for an 80° flip angle for far off-resonance nuclei the projection is 98.5% of that of a 90°-pulse – a more than acceptable result for NMR spectroscopy. After this excitation pulse, the different nuclei precess with their different Larmor frequencies about the z-axis. According to Maxwell’s equations, a rotating magnetic moment creates a changing magnetic field that induces a current in a wire coil. In the NMR spectrometer, a sensitive receiver coil records this small oscillating current. Through T2 relaxation the induced current decays over time, which is why the recorded data is called free induction decay (FID). Because the current is detected in a time-dependent manner (and not frequency–selective), the acquired signal is a superposition of all frequencies that have been excited with the pulse. The mathematical operation Fourier transformation converts this time-domain signal into the frequency-domain or spectrum. In analogy to other spectroscopic methods one could imagine the resonance phenomenon differently. In the resonant case nuclei absorb the radio radiation if the frequency of the radio pulse matches their Larmor frequency. After the pulse all excited nuclear spins simultaneously emit the absorbed radio radiation, which is then detected. Therefore, the pulsed Fourier transformation method is often compared to the tuning of a bell. In principle, one could determine the individual tones, which make up the sound of the bell, in the fashion of a “continuous wave-experiment.” The bell is sequentially excited with all sonic frequencies through a loudspeaker from the lowest tones to the limit of ultrasound, and the reaction of the bell is measured with a microphone. This procedure is extremely cumbersome and every bell founder knows that it can be done quicker; one simply takes a hammer and strikes! The sound of the bell contains all tones at once and every human being can analyze the sound with his or her ears (an ingeniously “constructed” biological tool for Fourier transformation). Note, however, that no frequencydependent detection occurs on modern NMR spectrometers.
18.1.2 One-Dimensional NMR Spectroscopy The 1D Experiment With these basic theoretical principles we are able to understand the simplest variant of NMR spectroscopy – the one-dimensional (1D) experiment (Figure 18.6). Each 1D NMR experiment consists of two parts: preparation and detection. During the preparation the spin system is brought to a defined state; during the detection the response to the preparation is recorded. Figure 18.6 Schematic illustration of the pulse sequence for a one-dimensional NMR experiment. A 1D experiment consists of two parts, the preparation and the detection. In the simplest case, the preparation consists of a single 90° pulse (black bar). Subsequently, the response of the spin system (FID) to this pulse is recorded during the detection period.
Preparation of the spin system consists in the simplest case of a short, hard excitation pulse (ca. 10 μs) that creates transverse magnetization (compare Figure 18.4). The resulting FID is recorded and saved during the detection period. After a short waiting time (the relaxation delay) that allows the magnetization to return to its equilibrium value through T1 relaxation, the experiment can be repeated multiple times. The individual data of the measurement are then added together, increasing the signal-to-noise ratio. Multiplication of the FID data with window functions can enhance either the sensitivity or the resolution of the spectrum. Additionally, this operation suppresses artifacts that arise from the subsequent Fourier transformation, which converts the FID (time-domain) into the NMR spectrum (frequency domain). Spectral Parameters We will discuss the different NMR spectral parameters (chemical shift, scalar couplings and line width) on the basis of the simple 1D NMR spectrum of ethanol
18 Magnetic Resonance Spectroscopy of Biomolecules
(Figure 18.7). This spectrum contains three signals (or peaks) originating from the methyl (CH3) protons, the methylene (CH2) protons, and the hydroxyl (OH) proton. Because the protons of the methyl group and of the methylene group, respectively, are each equivalent to each other, they each give rise to only one peak. The two peaks appear as so-called multiplets because their signals are split into several lines by scalar coupling. The integral over the respective multiplets yields the number of protons that give rise to these signals. For ethanol one obtains a ratio of the integrals of 3 : 2 : 1 corresponding to the number of protons that contribute to the respective signals. Chemical Shift In a molecule the electrons surrounding the nucleus create a weak magnetic field and shield the nucleus slightly from the main field. This shielding depends on the specific chemical environment (i.e., the structure of the molecule) and influences the Larmor frequencies of the nuclei. The effect is called the chemical shift and is one of the fundamental parameters in NMR spectroscopy, because it determines the distinct positions of individual signals in the NMR spectrum. The chemical shift δ of a signal in ppm (parts per million) is defined as: δ
ωsignal ωreference 106 ppm ωreference
(18.11)
The frequencies are given in ppm instead of Hertz because the former unit is independent of the magnetic field strength. The common reference frequency (ωreference) on which the chemical shift is based is the signal of the methyl groups of tetramethylsilane (TMS). By definition, it has a chemical shift of 0 ppm. For aqueous solutions of proteins and nucleic acids the preferred standards are the methyl signals of 2,2-dimethyl-2-silapentane-5-sulfonic acid (DSS) or trimethylsilylpropanoic acid (TSP). By convention, the chemical shift is plotted on the xaxis of a NMR spectrum from right to left. One often encounters expressions like “a signal appears at high field” (i.e., at low ppm values) or “downfield shift” (i.e., shift towards higher ppm values). These expressions originate from the time when NMR spectra were acquired at constant transmitter frequency through variation of the magnetic field (continuous wave technique). The position of a peak in the NMR spectrum provides substantial information about the origin of the respective signal. Many chemical or functional groups display specific chemical shifts (Figure 18.8); for example, the chemical shift of the hydroxyl group of ethanol differs from that of the methyl group (Figure 18.7). For proteins, the chemical shift alone suffices to distinguish between the signals from the HN, Hα, aromatic and aliphatic protons (Figure 18.12 below). The chemical shift further contains information about the secondary and tertiary structure of a protein, which is very valuable in different stages of structure determination (Section 18.1.6, Determination of the Secondary Structure). Scalar Coupling In the 1D spectrum of ethanol the signals of the methylene and methyl protons (Figure 18.7) appear as multiplets. This line splitting results from the scalar coupling (or indirect
439
Figure 18.7 One-dimensional 1 H NMR spectrum of ethanol. The signal of the methylene group is split into four lines with an intensity ratio of 1 : 3 : 3 : 1 (quartet); that of the methyl group into three lines with a ratio of 3 : 6 : 3. According to the number of hydrogen atoms, one obtains an intensity ratio (= integrals of the signals) of 2 : 3 for the methylene and methyl signals. The hydroxyl proton rapidly exchanges with hydroxyl protons from other ethanol molecules. Therefore, its signal at 2.6 ppm is much broader relative to the other protons. For the same reason, no coupling occurs between the hydroxyl proton and the methylene protons. Hence, neither the hydroxyl signal is split nor does the hydroxyl proton contribute to an additional splitting of the methylene protons.
440
Part II: 3D Structure Determination
coupling) between the protons, and is mediated through the electrons in the atomic bonds connecting the nuclei. Besides the nuclear Overhauser effect (NOE), scalar coupling is the most important mechanism in multidimensional NMR spectroscopy (Section 18.1.3) by which magnetization is transferred between nuclei. Line splitting arises due to the different orientations of a spin ½ to the external magnetic field. Each of the two methylene protons can adopt either a parallel or antiparallel orientation, which corresponds to two different magnetic quantum numbers m. The protons of the methyl group, which are scalar coupled to the methylene protons, therefore, “experience” four possible combinations (↑↑, ↑↓, ↓↑, and ↓↓). The orientations ↑↑ and ↓↓ marginally enhance or attenuate, respectively, the external magnetic field and thus shift the resonance frequency of the methyl protons (Figure 18.9). This gives rise to two lines lying symmetrically left and right of the actual resonance frequency. The orientations ↑↓ and ↓↑ are equivalent and compensate their respective enhancements or attenuations of the main field, thus, an unshifted line with twice the intensity results. The generated splitting pattern for the methyl group is called a triplet.
Figure 18.8 (a) Typical proton chemical shift ranges of different chemical groups. Source: adapted from Bruker Almanac, 1993. (b) Typical proton chemical shift ranges of individual amino acids in proteins. Source: adapted from Wishart, D.S., Sykes, B.D., and Richards, F.M. (1991) J. Mol. Biol., 222, 311–333. (With permission Copyright 1991 Published by Elsevier Ltd.)
18 Magnetic Resonance Spectroscopy of Biomolecules
Figure 18.8 (Continued)
441
442
Part II: 3D Structure Determination
If two nuclei with spin quantum numbers I and S couple with each other, then the resonance of I splits into 2S + 1 lines and the resonance of S into 2I + 1 lines. If the coupling partners of S consist of several identical I-nuclei, then the resonance S splits into 2nI + 1 lines, where n is the number of identical coupling partners (and vice versa).
Figure 18.9 Origin of the triplet: In an AX2 spin system the coupling with two identical nuclei X causes the resonance of nucleus A to split into a triplet. Each of the two X-nuclei can orient itself either parallel or antiparallel to the external magnetic field giving rise to four possible orientations. A parallel orientation enhances, while an antiparallel orientation attenuates, the external magnetic field. Therefore, the line associated with the ↑↑ orientation shifts upfield and the respective line for the ↓↓ orientation shifts downfield. The orientations ↑↓ and ↓↑ are indistinguishable and the respective lines appear at the actual resonance frequency. The single lines of the triplet have an intensity ratio of 1 : 2 : 1.
Figure 18.10 Typical values of different coupling constants (in Hz) in the protein backbone. Only coupling constants larger than 5 Hz are considered. Direct CC- and CN-couplings are shown in black; direct CH- and NH-couplings in grey. Black or grey dashed lines highlight indirect CCand CN-couplings, or indirect CH- and NH-couplings, respectively. Source: adapted from Bystrow, W.F. (1976) Progr. NMR Spectrosc., 10, 41–81. (With permission Copyright 1976 Published by Elsevier B.V.)
The number of individual spin combinations determines the intensity of these lines and follows a binomial distribution that can be illustrated in a Pascal’s triangle. For ethanol, two methylene protons (two I-nuclei) couple with three methyl protons (three S-nuclei). Thus, the signal of the methyl group splits in 2 × 2 × ½ + 1 = 3 lines (triplet), and that of the methylene group in 2 × 3 × ½ + 1 = 4 lines (quartet). The lines of the triplet have intensity ratios of 1 : 2 : 1; the lines of the quartet of 1 : 3 : 3 : 1. The separation of the lines (in Hz) in a multiplet corresponds to the coupling constant J, which is independent of the applied magnetic field strength. In general, only couplings over one bond
1 J , two bonds
2 J , and three bonds (3 J , or vicinal coupling) are observed (Figure 18.10). An important aspect of vicinal coupling constants is that their magnitude depends on the torsion angle between the two protons. The semiempirical Karplus relationship describes this dependence (Figure 18.11): J
ϕ A cos2
ϕ
60
B cos
ϕ
60 C
(18.12)
in which A, B, and C are empirically determined constants that are different for each type of torsion angle (e.g., the ϕ, ψ, and χ angles in proteins). For example, protein structure determination exploits the information about the molecular geometry contained within the 3 J
HN Hα coupling constant to restrain the torsion angle ϕ (HN-N-Cα-Hα). Line Width The line widths of NMR peaks provide direct evidence about the lifetimes of the respective resonances. The longer the lifetime of a resonance is, the narrower is the line width of its peak (and vice versa). The resonance lifetimes are manly determined by T2 relaxation and chemical exchange. As mentioned earlier, the short T2 relaxation time constants of large molecules (proteins > 50 kDa) cause broad line widths and thus result in peaks with low intensity. Similarly, chemical exchange during de- and reprotonation reactions can reduce the lifetime of a proton. For example, the hydroxyl proton of ethanol (Figure 18.7) exchanges with other solvent protons from solvents (in this case other ethanol molecules) and thus possesses a broader line.
Figure 18.11 (a) Relationship between the 3J(HN–Hα) coupling constant and the ϕ angle (Karplus relationship). The torsion angle between COi and COi 1 (b) is plotted against the coupling constant 3J(HNHα). The plot shows that at least two angles correspond to a given coupling constant. The index i in the Newman projection (b) denotes the relative position of the amino acids to each other in the protein sequence.
18 Magnetic Resonance Spectroscopy of Biomolecules
443
Figure 18.12 1D 1 H NMR spectrum of a 14 kDa β-sheet protein with an immunoglobulin fold at 30 °C. The characteristic chemical shift of each proton type facilitates their identification in the different spectral regions. At the left end, at 11 ppm are the peaks of the tryptophan indole protons. The peaks from 10 to 6 ppm are assigned to the amide protons of the protein backbone and of the asparagine and glutamine side chains. Between 7.5 and 6.5 ppm are the aromatic protons, followed by the Hα protons from 5.5 to 3.5 ppm. To the righthand end of the spectrum (2D) NMR spectroscopy. Heteronuclear NMR Experiments Besides protons, biomolecules contain other magnetically active nuclei (the so-called heteronuclei). Multi-dimensional NMR experiments for structure determination rely particularly on the magnetically active isotopes of carbon (13 C) and nitrogen (15 N). Due to the low natural abundances and small gyromagnetic ratios (Table 18.1), the relative sensitivities of 13 C and 15 N are low compared with protons. Therefore, to increase the sensitivity of heteronuclear experiments, two general strategies exist:
First, recombinant expression of proteins in bacteria allows for the production of isotopically enriched proteins. To this end, bacteria are cultivated in minimal media, which contain
18 Magnetic Resonance Spectroscopy of Biomolecules
447
Figure 18.17 (a) Representative 2D NOESY spectrum of a 115 amino acid long protein. The very strong water signal in the center of the spectrum was removed during Fourier transformation. Grey rectangles in (b) schematically highlight the different spectral regions of this spectrum. For each rectangle the observable NOE signals in the respective region are given. The water line is shown as a grey bar.
15
NH4Cl as the sole nitrogen source. For enrichment of 13 C, the sole carbon source is 13 C-glucose. In this manner, singly-labeled (15 N or 13 C) or even doubly-labeled (13 C, 15 N)-samples are produced. Additionally, one can obtain deuterated proteins if D2O instead of H2O is used as the solvent for the culture medium. Second, the signal-to-noise ratio of an NMR experiment depends among other factors on the gyromagnetic ratios of the excited and detected nuclei. Therefore, the direct excitation/detection of heteronuclei is less sensitive relative to excitation/detection of protons. Thus, experiments in general rely on the transfer of the large magnetization of protons to an attached heteronucleus (and vice versa). This achieves an optimal signal-to-noise ratio at only minor magnetization losses. For historical reasons, such experiments are called inverse heteronuclear experiments.
HSQC – Heteronuclear Single Quantum Coherence The HSQC (heteronuclear singlequantum coherence) experiment constitutes the most important experiment that transfers
448
Part II: 3D Structure Determination
Figure 18.18 Pulse sequence of the HSQC experiment. Narrow black bars represent 90° pulses; broad black bars 180° pulses. The upper line displays pulses on the proton (1 H) frequency; the lower line on the nitrogen (15 N) frequency.
Figure 18.19 HSQC spectrum of severin DS111M at 32 °C and pH 7.0. The assignment for each peak is shown as single letter code (the numbers refer to the position in the protein sequence). The nitrogen frequency is plotted on the x-axis; the proton frequency on the y-axis.
18 Magnetic Resonance Spectroscopy of Biomolecules
449
magnetization to a heteronucleus and back (Figure 18.18). The HSQC correlates the frequency (ω1) of a heteronucleus (13 C or 15 N) with that of the directly bound proton (ω2). For example, in a two-dimensional 15N-HSQC each peak represents one proton bound to one nitrogen atom, that is, the spectrum consists essentially of all the amide signals (HN–N) of the protein backbone. Additionally, peaks arise for the aromatic, nitrogen-bound protons of the tryptophan and histidine side chains, and for the side chain amide groups of asparagine and glutamine, respectively (Figure 18.19). In the latter case, two peaks appear at the same nitrogen frequency because two amide protons are bound to the same side chain nitrogen. Under favorable conditions, the nitrogen-bound protons of the side chains of arginine and lysine are also visible. The advantage provided by the additional nitrogen dimension of the HSQC experiment is that it resolves amide proton resonances that often overlap in 1D and homonuclear 2D spectra of larger proteins. Compared with a homonuclear spectrum, the HSQC has of course no diagonal because it correlates completely different types of nuclei during the t1 and t2 times.
18.1.4 Three-Dimensional NMR Spectroscopy The modularity of NMR experiments opens up the possibility for multidimensional NMR simply by introducing further dimensions. For example, we can create a 3D experiment through replacement of the acquisition time after the first mixing period of the 2D experiment (Figure 18.13) with another indirect evolution time and subsequent second mixing period (Figure 18.20). In four-dimensional NMR, a third indirect time follows as well as an additional mixing period. The different indirect times are each incremented individually. The direct data acquisition forms the end of each multidimensional experiment. We start our discussion of 3D NMR with the pulse sequences that are combinations of two 2D experiments because they are conceptually easier. Later, we will describe the so-called triple-resonance experiments that correlate three different types of nuclei (1 H, 13 C, 15 N).
The NOESY-HSQC and TOCSY-HSQC Experiments We have mentioned above that spectral overlap limits the application of 2D spectra (NOESY or TOCSY) for larger proteins. Due to the dispersion of the peaks in a cube instead of a plane, the introduction of a third dimension can greatly resolve this overlap. In general, a heteronuclear coordinate like 15 N or 13 C constitutes the third (vertical) dimension of this cube because the wider frequency range of the heteronucleus provides a better signal resolution than an additional proton dimension. We can create such a 3D experiment simply by combining the pulse sequences for a 2D NOESY and a 2D HSQC, in which instead of the data acquisition the HSQC experiment follows directly after the NOESY experiment. The created experiment is called 13 C- or 15NNOESY-HSQC. In an analogous way, we can convert a 2D TOCSY experiment into a 3D TOCSY-HSQC by combining a 2D TOCSY with a 2D HSQC. The 15N-NOESY-HSQC and 15 N-TOCSY-HSQC represent the basic experiments for the sequence-specific assignment of medium-sized proteins (10–15 kDa). The respective 13 C variants are very useful in assigning the side chains and in identifying NOE signals between the side chain protons. The HCCH-TOCSY and HCCH-COSY Experiments The HCCH-TOCSY and HCCHCOSY experiments are alternatives to the 15N-TOCSY-HSQC, whose sensitivity strongly decreases for larger proteins. Both experiments transfer magnetization exclusively through scalar J-couplings between nuclei. For example, initially the magnetization transfer takes place from the Hα proton to the Cα nucleus (Figure 18.21). From there the magnetization transfer continues to the next carbon nucleus of the side chain, or in the case of the HCCH-TOCSY continues to all carbon nuclei of the side chain. Because the 1JCC-coupling is about 35 Hz, the
Figure 18.20 Schematic illustration of a 3D experiment showing the NOESYTOCSY experiment as an example. Compared with the 2D experiment, a 3D experiment contains additional evolution times and mixing periods. The mixing period of the NOESY transfer step consists of two pulses with a delay time (τm) in between; the mixing period of the TOCSY transfer step consists of a complicated series of pulses called the MLEV mixing sequence (after its inventor Michael Levitt).
450
Part II: 3D Structure Determination
Figure 18.21 Slices of different planes from an HCCH-TOCSY experiment of reduced DsbA from Escherichia coli. All correlation signals of the amino acid Leu185 are shown. The 13 C chemical shift of the peaks is given next to the corresponding plane; the 1 H chemical shift is plotted horizontally along the lowest plane. Individual proton assignments are given next to the respective peak; carbon assignments are next to the slices of the respective plane. Collectively, these peaks produce the typical spin system pattern for a leucine residue also seen in a 2D TOCSY.
mixing time is shorter compared to the homonuclear case (nJHH < 10 Hz) and thus more sensitive for larger proteins. The time duration for the magnetization transfer is calculated according to tmix = 1/(2J). The data acquisition follows after the final magnetization transfer from each carbon nucleus to the directly bound proton. In general, the appearance of an HCCHTOCSY (or HCCH-COSY) spectrum is identical to that of a 13 C-TOCSY-HSQC spectrum. Again characteristic spin system patterns similar to the 2D TOCSY and 2D COSY facilitate the identification of the amino acid type. Triple-Resonance Experiments For proteins larger than 15 kDa, spectral crowding affects even the 3D NOESY-HSQC and TOCSY-HSQC spectra, thus challenging the protein backbone assignment (Section 18.1.5, Analysis of Heteronuclear 3D Spectra). The sequential assignment of large proteins therefore relies on triple-resonance experiments due to their simple appearance. For each amino acid only few peaks appear – often only one. Therefore, the problem with overlapping peaks occurs less frequently in triple-resonance spectra, and enables the sequential assignment of proteins up to 30 kDa. However, for certain nuclei the chemical shift of one residue may accidently match that of a different residue. This so-called degeneracy is especially common for the Cα nuclei. Identification of the correct connection between amino acids with degenerate chemical shifts is, therefore, one of the main obstacles for sequential assignment via triple-resonance experiments (Section 18.1.5, Sequential Assignment from Triple-Resonance Spectra). Because triple-resonance experiments correlate three different nuclei with each other, they require more expensive doubly-labeled 13 C, 15 N or triply-labeled 2 H, 13 C, 15 N protein samples. A further advantage of triple-resonance experiments is their high sensitivity due to an efficient magnetization transfer through the strong 1J- and 2J-couplings (Figure 18.10) between the nuclei (i.e., directly via the covalent atom bonds). The required times for the transfer are, therefore, comparatively short so that relaxation losses are substantially decreased relative to a TOCSY experiment. This high sensitivity is another reason why signal assignment is possible for proteins up to a molecular weight of 30 kDa. For even larger proteins the sensitivity of tripleresonance experiments decreases mostly due to short transverse relaxation times. Especially the Hα protons enhance the relaxation of the Cα and HN nuclei through dipolar interactions. Reducing the number of the aliphatic protons through protein deuteration greatly attenuates this
18 Magnetic Resonance Spectroscopy of Biomolecules
451
type of relaxation and extends the molecular weight of proteins that can be studied by NMR to 50 kDa and beyond (Section 18.1.7, Structure and Dynamics of High Molecular Weight Systems and Membrane Proteins). Nomenclature of Triple-Resonance Experiments A large variety of triple-resonance experiments exists of which Figure 18.22 shows the most important representatives. Even though their nomenclature sounds cryptic, it is very descriptive. The name of the experiment specifies the detected nuclei, the magnetization transfer pathway, and the appearance of the spectrum. To this end, all nuclei through which magnetization is transferred are listed in a row. For example, the experiment denoted HNCO detects three nuclei (1 H, 13 C, 15 N) with the following flow of ´ magnetization: HN
i ! N
i ! C
i 1 (Figure 18.22). Brackets in experiment names mark nuclei that serve only as “relay stations” and whose frequencies remain undetected. For example, the HN (CA)CO detects the same types of nuclei as the HNCO, however, the magnetization transfer α ´ α differs: HN
i ! N
i ! C
i ! C
i , (Figure 18.22). The C nucleus only relays magnetization ´ from nitrogen to the carbonyl (C ) carbon, but its chemical shift is not recorded. Note that in both experiment types the magnetization follows the same pathway back to the amide proton for acquisition (“out-and-back” transfer). The appearance of the two spectra is similar. For each amino acid residue one peak arises that correlates the amide proton and nitrogen with a nitrogenbound carbonyl carbon. However, while the HNCO spectrum shows the correlation with the C´ of the preceding residue (residue i 1), the HN(CA)CO mainly shows the correlation with the intraresidual C´ (residue i). The HNCA Experiment The HNCA constitutes one of the simplest and most useful examples of a triple-resonance experiment. The magnetization transfer pathway is given by α HN
i ! N
i
t 1 ! C
i=
i 1
t 2 ; t1 and t2 specify the indirect dimensions during which the chemical shifts of the heteronuclei are encoded (Figure 18.20). The HNCA utilizes an “outand-back” transfer to detect the amide proton chemical shift in t3. In all cases the magnetization transfer occurs through strong J-couplings (1JHN = 92–95 Hz, 1JNC = 11 Hz). Because the 2JNC
Figure 18.22 Overview of the most important triple-resonance experiments. Only the chemical shifts of the nuclei colored dark grey are detected during the experiment. Nuclei colored light grey serve as transmitters for the magnetization and remain undetected. Arrows mark the magnetization transfer pathway and direction. Under each experiment name all observable correlations are listed, in which subscripts i and i 1 denote the position of the amino acid residues relative to each other. Even though the HCCH-TOCSY is not a tripleresonance experiment, it is included here due to its complementarity with the triple-resonance experiments for the assignment process.
452
Part II: 3D Structure Determination
coupling constant between the nitrogen nucleus and the Cα nucleus of the preceding amino acid is only marginally smaller (7 Hz) than the 1JNC-coupling (11 Hz, Figure 18.10), magnetization transfer from nitrogen occurs to the Cα nuclei of both the amino acid it is part of and the preceding one. Therefore, for each amino acid two peaks arise in the HNCA spectrum: one intra- and one interresidue correlation. The related HN(CO)CA only shows the correlation to the preceding residue. In principle, the intra- and interresidue correlations of the HNCA enable the sequential assignment of the protein backbone. However, in practice one requires additional experiments to identify the cross-peaks of the preceding residue and also to resolve chemical shift degeneracies (Section 18.1.5, Sequential Assignment from Triple-Resonance Spectra).
18.1.5 Resonance Assignment Sequential Assignment of Homonuclear Spectra NMR spectra contain all the necessary information about proton distances and torsion angles to calculate the structure of a protein or nucleic acid, which constitutes one possible aim of biomolecular NMR. To this end, it is necessary to assign each peak observed in the spectrum to the respective proton in the protein. Due to the large number of peaks in COSY, TOCSY, and NOESY spectra, this requires a simple and universal analysis method. Kurt Wüthrich (Nobel Prize in Chemistry 2002) and colleagues developed a method that could exactly achieve this – the sequence-specific assignment. This method exploits the distance information contained in the NOESY spectrum. Because of their direct proximity, an amino acid i + 1 displays specific contacts with amino acid i (in each case i denotes the relative position of the amino acids to each other). For example, due to the molecular geometry, the distances of the amide proton of residue i + 1 to the Hα, Hβ, or Hγ protons of amino acid i, respectively, are almost always less than 5 Å (Figure 18.23). Therefore, at the amide proton chemical shift of amino acid i + 1 (horizontal frequency axis) cross-peaks arise at the chemical shifts (vertical frequency axis) of the respective protons of amino acid i. These interresidual peaks between neighboring amino acids are also called sequential peaks. The method of sequence-specific peak assignment requires a distinction between the interand intraresidual peaks for a given amide proton chemical shift. A simple comparison of the 2D NOESY spectrum overlaid with the 2D TOCSY spectrum of the same sample facilitates this distinction (Figure 18.23). While the sequential cross-peaks provide information about the connectivity to the preceding amino acid, the characteristic pattern of the intraresidual crosspeaks determines the amino acid type. Prolines interrupt this chain of sequential connectivities because they lack an amide proton and therefore show no peak in the amide region of the spectrum. However, proline residues, which adopt the more frequent trans-conformation (Figure 18.24), give rise to sequential Hα
i 1 Hδ
i cross-peaks at the Hα chemical shift or HN Hδ
i cross-peaks at the amide proton
i 1 chemical shift of the preceding residue. Another problem, which arises especially for larger proteins, is that the large number of peaks often results in peak overlap, preventing an unambiguous continuation of the sequential assignment at some positions in the sequence.
Figure 18.23 Overlay of a 2D NOESY (black circles) and a 2D TOCSY spectrum (grey circles) illustrating the schematic peak pattern of two neighboring amino acids. The interresidual, sequential NOESY peaks (filled black circles) on the succeeding amino acid i + 1 form the basis for the sequence-specific assignment. Arrows in the dipeptide on the left-hand side mark the respective protons that give rise to the sequential cross-peaks. For completeness, dashed, open circles signify the intraresidual peaks of amino acid i + 1 and the sequential peaks of amino acid i 1 at the amide proton chemical shift of amino acid i, respectively. For clarity, the symmetrical peaks occurring below the diagonal are omitted.
18 Magnetic Resonance Spectroscopy of Biomolecules
453
Figure 18.24 Conformation of the cis/ trans-isomers of the peptide bond for an amino acid X and a proline residue. The Cα atoms of both residues as well as the connecting bonds that define the torsion angle ω are emphasized.
The first step of the sequence-specific assignment is to identify individual amino acids that serve as starting points for the assignment. Initially, the search is restricted to amino acids such as glycine, alanine, valine, or isoleucine because their characteristic peak pattern differs from other amino acid types. For example, glycine possesses two Hα protons. Detection of these two Hα peaks at the amide proton chemical shift and the appearance of the corresponding Hα1–Hα2 provide definite evidence for glycine (Figure 18.25). The characteristic double peak row of the methyl groups at 0–1.5 ppm (Figure 18.25) signifies valine, leucine, and isoleucine. In addition to the information provided by the amide region of the spectrum, the aliphatic region contains various diagnostic cross-peaks, especially for prolines, that help to validate the spin systems. The next step is to identify the specific sequential contacts in the 2D NOESY spectrum and thus to determine the preceding residue for each of these distinct amino acids. Then, one determines for every newly identified residue the preceding amino acid in an iterative manner. Thus, the initially identified dipeptide is extended into an oligopeptide chain. The information about the amino acid types contained in these fragments enables one to place the fragments into the protein sequence – that is to assign each spin system to the respective amino acid.
Figure 18.25 Schematic illustration of the characteristic peak patterns of (a) valine and (b) glycine. Both residues serve as starting points for the sequence-specific assignment due to their easily recognizable and distinctive patterns in the TOCSY spectrum. The left-hand side shows the structures of the respective amino acids together with the proton designations and typical chemical shift values. For clarity, the symmetrical peaks occurring below the diagonal are omitted.
454
Part II: 3D Structure Determination
Analysis of Heteronuclear 3D Spectra The method of sequence-specific assignment explained above for homonuclear 2D TOCSY and NOESY spectra is also appropriate for the respective heteronuclear 3D spectra 15N-NOESY-HSQC and 15N-TOCSY-HSQC. Every 15Nplane of the spectra contains the NOESY and TOCSY peaks of an amide proton bound to its nitrogen. Again a superposition of the respective 3D TOCSY- and 3D NOESY-HSQC spectra facilitates the distinction between intra- and interresidual correlations. The 15N-plane (of the 3D 15 N-NOESY-HSQC) is, therefore, a kind of sub-spectrum of the respective 2D spectrum (2D NOESY). A major difference from the 2D spectra is the frequency range sampled in the acquisition dimension of 15N-edited 3D NOESY and TOCSY spectra. The experiments select only correlations to protons bound to 15N-nuclei. Therefore, every 3D 15N-NOESY-HSQC and 15 N-TOCSY-HSQC only contains frequencies between 12 and 5 ppm in the acquisition dimension; the side chain region beyond the water signal on the high field side of the spectrum is empty. Selective Amino Acid Labeling 15N-Labeling of selective amino acids constitutes an alternative way to determine the amino acid type. To this end, recombinant Escherichia coli cells are cultivated in a minimal medium containing all 20 naturally occurring amino acids. This medium composition greatly suppresses the cellular de novo synthesis of amino acids because the cells take up the amino acids directly from the medium. Addition of a commercially available 1-15N-L-amino acid to the medium achieves the selective labeling of this residue type. All other amino acids are added to the medium as “normal” 14 N-amino acids. In particular, those unlabeled amino acids need to be added in excess, which can be metabolically synthesized from the selectively labeled amino acid (the so-called scrambling). An HSQC spectrum acquired on a selectively labeled protein shows only the signals from the labeled amino acid type. Therefore, the labeling of different amino acid types facilitates a quick assignment of the NMR signals and further validates the assignment obtained from TOCSY and NOESY spectra. In general, selective amino acid labeling is less expensive than the production of doublylabeled proteins because it requires less protein amounts and the 15N-labeled amino acids are moderately priced (the actual price depends on the specific amino acid type). However, to assign all residues in a protein, different selectively labeled proteins have to be made, which increases the workload compared to the production of a single, uniformly labeled sample. Thus, the production of multiple samples can easily compensate for the previous savings. Sequential Assignment from Triple-Resonance Spectra To illustrate the sequential assignment with triple-resonance experiments, we will restrict ourselves to the most popular
Figure 18.26 Schematic illustration of the sequence-specific assignment for an HNCA spectrum. The 1 H–15 N projection of the 3D spectrum shows one peak for each amino acid (grey peaks). At each amide proton chemical shift two crosspeaks exist, one originating from the correlation with the Cα nucleus of the residue it is part of and one from the correlation with the preceding residue (blue). Starting from the blue signals, one can “walk” via the blue signals stepwise through the amino acid sequence (blue arrows).
18 Magnetic Resonance Spectroscopy of Biomolecules
455
representatives. The HNCA spectrum will serve again as an example to explain the general appearance of triple-resonance spectra and the assignment strategy. We have already shown in Section 18.1.4 (Triple-Resonance Experiments) that the HNCA spectrum consists of three frequency axes (1 H, 15 N, and 13 C). Its 1HN–15 N projection looks like an HSQC spectrum and every peak in this projection corresponds to a single residue. As mentioned earlier, two cross-peaks appear in the 13 Cα dimension at the chemical shift of each amide proton: one intraresidual correlation and one interresidual correlation to the 13 Cα nucleus of the preceding amino acid (Figure 18.26). In principle, this sequential information of the cross-peaks suffices to “walk” through the complete sequence (Figure 18.27). The “sequential walk” requires a clear distinction between intraresidual and sequential crosspeaks. In general, the intraresidue peaks tend to be slightly more intense than the interresidue peaks (due to the more efficient magnetization transfer via the 1JNC coupling). This tendency is, however, fallible because processes like relaxation influence the peak intensity. In contrast, targeted experiments like the HN(CO)CA force the magnetization through the carbonyl carbon and reject the intraresidual pathway, while the spectrum otherwise looks identical to the HNCA. Thus, one obtains exclusively the sequential cross-peaks and resolves the ambiguity that is present in the HNCA.
Figure 18.27 Slices from an HNCA (black contours) and a CBCA(CO)NH spectrum (blue contours) of huMIF. Each stripe corresponds to one amino acid from Phe18 to Lys32 at the respective amide proton (x-axis) and nitrogen chemical shifts (z-axis, not shown). The superposition of both spectra illustrates the “sequential walk” in which black horizontal lines indicate the sequential connectivities.
456
Part II: 3D Structure Determination
Proline residues and chemical shift degeneracies interrupt the assignment procedure (similar to the assignment of homonuclear spectra, Section 18.1.5 (Sequential Assignment of Homonuclear Spectra)). Proline residues lack an amide proton and therefore produce no cross-peaks, while chemical shift degeneracies prevent an unambiguous assignment of the preceding residues because several possibilities exist. As outlined below, other types of triple-resonance experiments correlating the amide group with other carbon nuclei can resolve the chemical shift degeneracy; however, proline residues will always interrupt the sequential assignment in amide proton-detected experiments. Combination of the HN(CA)CO with the HNCO experiment constitutes an independent alternative to validate the connectivities found in the HNCA. The two experiments correlate the amide proton with the intraresidual carbonyl carbon (HN(CA)CO) or that of the preceding residue (HNCO), that is, the sequential assignment is established through C´ instead of Cα connectivities. The superposition of both spectra results in a pattern analogous to the HNCA spectrum. Furthermore, the HNCO spectrum, which is the most sensitive triple-resonance experiment, can be used to resolve accidental signal degeneracies in the HSQC projection. In proteins, each amide proton–nitrogen pair is covalently attached to only one C´ . Therefore, one observes one cross-peak per amide proton frequency. However, if one finds two cross-peaks at the frequency of an amide proton than this means that the signals of two amino acids are degenerate in the HSQC projection. Two further pairs of experiments provide independent assignment strategies. The CBCANH and CBCA(CO)NH experiments give rise to intra- and interresidual correlations of the amide group with the Cα and Cβ resonances, respectively. The closely related HBHA(CBCACO)NH and the HBHA(CBCA)NH experiments provide the correlations with the Hα and Hβ resonances. The Cα and Cβ chemical shifts obtained from the first two experiments are useful to narrow down the amino acid type. Because of the distinct value of the Cα and Cβ chemical shifts, they enable a preliminary identification of alanine, glycine, isoleucine, proline, serine, threonine, and valine residues (Figure 18.28). If the resonances of different amino acids are degenerate in the 15N-HSQC projection, then the HCACO spectrum provides a simple alternative to differentiate these residues. This experiment establishes the correlations between the Hα, Cα, and C´ frequencies of an amino acid and thus allows for the continuation of the assignment.
Figure 18.28 Typical ranges of Cα and Cβ chemical shifts of the 20 amino acids. Source: adapted from: Cavanagh, J., Fairbrother, W.J., Palmer, A.G. III, and Skelton, N. (1996) Protein NMR Spectroscopy, Academic Press.
18 Magnetic Resonance Spectroscopy of Biomolecules
While triple-resonance spectra provide the connectivities between individual spin systems (and thus the sequential assignment) they only yield limited information about the side chain. The side chain assignment therefore relies on HCCH-TOCSY and the HCCH-COSY experiments, and starts from the known frequencies of Cα (obtained from the HNCA) or Hα (obtained from, for example, the 15N-TOCSY-HSQC or the HCACO). Summary: The general strategy to sequentially assign a protein by triple-resonance experiments requires the acquisition of several independent spectra. First, one establishes the sequential connectivity between the spin systems through at least two different nuclei (Cα, Cβ, or C´ ). This strategy minimizes complications of the assignment by chemical shift degeneracies. Secondly, one restricts the amino acid type through the chemical shift information provided by CBCA(CO)NH or CBCANH spectra. Third, one assigns the side chains with HCCH-TOCSY and 13 C-NOESY-HSQC experiments (Section 18.1.4, The NOESY-HSQC and TOCSY-HSQC Experiments). Taking all the information together one can finally place the identified spin systems into the protein sequence.
18.1.6 Protein Structure Determination Constraints for the Structure Calculation Up to now, we have described methods to identify and assign the observable NMR signals to the respective amino acids. Once the assignment of the resonances to individual nuclei has been completed, the next step is to extract structure-defining data from the NMR spectra. Most important are 2D NOESY, 3D 15NNOESY-HSQC, and 3D 13 C-NOESY-HSQC spectra (Section 18.1.4, The NOESY-HSQC and TOCSY-HSQC Experiments), as they provide a wealth of proton–proton distances. For a medium-sized protein (ca. 120 amino acids) one usually observes more than 1000 NOE contacts, which have to be assigned to specific protons on the basis of the previously established sequence-specific resonance assignment. Particularly important are the non-sequential NOEs that define the three-dimensional structure. Medium-range NOEs (less than four amino acids separate the residues involved in the NOE) provide information about the local backbone conformation of the protein and serve to determine secondary structural elements. Long-range NOEs (five or more amino acids separate the involved residues) define the relative orientation of the secondary structural elements to each other and are thus the essential parameters for the determination of the tertiary structure. Because the NOE signal intensity I depends on the distance r between two nuclei i and j according to:
1 I NOEi;j ∝ 6 r ij
(18.13)
the internuclear distances can be obtained through integration of the NOE signal. Alternatively, it is possible to qualitatively estimate their intensity. In both cases, however, the signal intensities have to be calibrated with respect to a NOE signal of known distance (e.g., to known distances in secondary structural elements). Depending on the signal intensities NOEs are classified into different distance groups with fixed distance boundaries (Table 18.2). Table 18.2 Relationship between the intensity of a NOE signal and the respective proton distance. During structure calculations NOE distances act like elastic springs between the atoms. If an atomic distance exceeds the upper limit, then this NOE violation is penalized with an energy factor that depends on the extent of violation. This energy factor forces the respective protons closer together in the next iteration of the structure calculation. Because various effects can reduce the NOE intensity in the NMR spectrum irrespective of the atomic distance, usually no lower limits are used for the calculation; only the “normal” repulsion between the atoms due to their van der Waals radii acts. NOE intensity
Distance (Å)
Upper limit (Å)
Strong
2.4
2.7
Medium
3.0
3.3
Weak
4.4
5.0
Very weak
5.2
6.0
457
458
Part II: 3D Structure Determination
In addition to distance restraints, structure calculations utilize the 3J(HN–Hα) coupling constants provided by COSY or HNCA-J spectra (a variant of HNCA). As described in Section 18.1.2 (Spectral Parameters), 3J(HN–Hα) coupling constants restrain the ϕ torsion angles of the protein backbone through the Karplus relationship (Figure 18.11). Residual dipolar couplings (RDCs) represent a rather new class of NMR restraints that includes chemical shift anisotropy and cross-correlated relaxation. In isotropic solution rotational diffusion averages anisotropic interactions such as the dipolar coupling between two nuclear spins to zero. Dipolar couplings therefore do not result in line splitting in isotropic solution, but act as a relaxation mechanism. However, if the molecular tumbling is anisotropic and certain orientations are preferred or avoided (i.e., a molecular alignment is present), dipolar couplings average incompletely, resulting in scaled down values (relative to the static value) of the dipolar interaction. The corresponding couplings are called residual dipolar couplings. In contrast to NOEs and scalar couplings, which yield distance information about the local geometry, orientational restraints such as RDCs provide long distance restraints for the whole molecule. For example, RDCs determine the relative orientation of secondary structural elements or of individual protein domains to each other. Additionally, RDCs provide dynamic information about slow internal motions of a bond vector. The use of RDCs in structure calculations is, however, more complicated than for NOEs or J-couplings. All RDCs refer to the same reference frame that is fixed to the molecule – the anisotropic alignment tensor σ. Thus, RDCs depend on the orientation of the internuclear vector relative to the molecular reference frame, which is described by two angles (θ, ϕ) (Figure 18.29); θ defines the angle between the internuclear vector and the z-axis of the principle axis frame, and ϕ is the corresponding azimuthal angle. The principle axis frame is defined as that molecular frame in which the alignment tensor σ is diagonal (i.e., with principal values |σz| > |σy| > |σx|). The dipolar coupling DA,B between two nuclei A and B is given by: h 2 DA;B 0:75DA;B max 3 cos θ
Figure 18.29 The orientation of an N H bond vector i relative to the alignment tensor σ determines the magnitude and sign of the residual dipolar coupling. The alignment tensor is fixed to the protein; however, it depends in magnitude and orientation on the orienting medium (such as bicelles and Pf1).
1 σ z sin2 θ cos 2ϕ σ x
σy
i
(18.14)
The magnitude of RDCs scales with the degree of alignment. Experimentally, one prefers a weak alignment of the proteins to determine RDCs with a magnitude of a few Hertz. When the alignment is weak (corresponding to the alignment of 1 out of 1000 molecules) the spectra remain comparable to those in isotropic solution – that is the number of lines does not increase strongly and the NMR signals remain sharp. Several methods exist to achieve a weak alignment of a protein. Originally, particles were studied that aligned spontaneously in the magnetic field (through anisotropic tensors of the magnetic susceptibility). Meanwhile, researchers have resorted to systems that display a liquid-crystalline behavior in the magnetic field even when strongly diluted:
bicelles (flat micelles consisting of two types of lipids with different acyl chain lengths); rod-like virus particles (tobacco mosaic virus or bacteriophage Pf1); poly(ethylene glycol)/alcohol mixtures; purple membranes of halophile bacteria.
In addition, mechanically stretched or compressed polyacrylamide gels also achieve a partial alignment of the solute (protein or nucleic acid). To determine RDCs, mostly heteronuclear experiments, which are also used for measurement of scalar couplings, are recorded. Determination of the Secondary Structure Regular secondary structural elements like α-helices and β-sheets are regions with a defined conformation of the protein backbone, that is, with well-defined torsion angles and fixed proton–proton distances. Therefore, diagnostic NOE signal patterns and 3J(HN–Hα) coupling constants allow for a distinction between the most important secondary structural elements (Figure 18.30). Characteristic for an α-helix are strong N NOE signals between the amide protons HN HN HN
i
i1 and H
i
i2 , and also the very 3 N α specific NOEs between Hα
i HN (Figure 18.30). J(H -H ) coupling constants less than
i3 5 Hz additionally confirm the existence of an α-helix. In contrast, for β-sheets a specific pattern of Hα
i Hα
j and HN HN
i
j NOEs arises between the two parallel or antiparallel strands of the sheet (i and j denote amino acids in different strands of the sheet). The extended structure of the 3 N α protein backbone is further evident from strong Hα
i HN
i1 NOEs and J(H –H ) coupling constants of more than 8 Hz for several consecutive residues.
18 Magnetic Resonance Spectroscopy of Biomolecules
459
Figure 18.30 Characteristic NOEs and typical 3J(HN–Hα) coupling constants for three regular secondary structures. The plot schematically illustrates typical intensities for the HN
i HN
i1 , Hα
i HN
i1 , HN
i HN
i2 , Hα
i HN
i2 , Hα
i HN
i3 , Hα
i Hβ
i3 , and Hα
i HN
i4 cross-peaks. The height of the rectangles reflects the peak intensity.
We will use the DS111M domain of severin to illustrate the identification of secondary structures. This second domain of severin plays a central role in the polymerization and depolymerization of filamentous actin, a constituent of the cytoskeleton. Severin DS111M consists of 114 amino acids and contains three α-helices evident from Hα
i HN
i3 connectiviN ties (Figure 18.31). In addition, strong sequential Hα
i HN HN
i1 contacts and missing H
i
i1 NOEs indicate the existence of five β-sheets. The 3J(HN–Hα) coupling constants (>8 Hz) provide further support for a β-structure in these regions. Moreover, particularly strong Hα
i Hα
j contacts between two strands facilitate the identification of associated strands within the β-sheet (Figure 18.32). These short distances (∼2.2 Å) are easily recognized in 2D NOESY or 3D 13 C-NOESY-HSQC spectra. Because the Hα resonances have chemical shifts close to that of water, it is advisable to record those NOESY spectra on samples dissolved in D2O, in order to reduce the interference from the water signal. Next to the direct structural information provided by distances, orientational restraints and torsion angles, hydrogen/deuterium (H/D) exchange rates of amide protons yield indirect structural information. Hydrogen bonds, which generally stabilize secondary structures, strongly attenuate the H/D exchange. Thus, slowly exchanging amide protons are often an indication for the existence of secondary structure (Figure 18.31). In addition, one obtains clues about the possible position of the residue in the protein structure. Amide protons in the center of the protein possess reduced exchange rates with the solvent due to their reduced accessibility, while amide protons at the surface display higher exchange rates. To experimentally determine the exchange rate, the protein sample is freeze-dried and subsequently dissolved in pure D2O. Over time, those signals disappear that are associated with rapidly exchanging amide protons. Conversely, signals of slowly exchanging amide protons remain visible in the spectrum for longer times (sometimes up to several months). Those amide protons localize almost exclusively to regions of the protein with regular secondary structure (Figure 18.31). The chemical shifts of protein residues provide further evidence for secondary structure. Relative to the disordered state (the random coil shift) the chemical shift changes if an amino acid is in a secondary structural element. This difference is called the secondary chemical shift and reflects the uniformity of the chemical environment in regular secondary structure. For example, Cα or C´ carbons experience downfield shifts (i.e., larger ppm values) in α-helices and
460
Part II: 3D Structure Determination
Figure 18.31 Overview of the sequential and short-range NOEs of DS111M. The height of the bars reflects the NOE peak intensity. The 3J(HN–Hα) coupling constants are given below the amino acid sequence. Filled and open circles denote residues with slow and intermediate, respectively, amide proton exchange rates. Above the amino acid sequence the secondary structure is shown schematically (arrow = β-sheet, spring = αhelix).
Figure 18.32 The complete network of NOE cross-peaks within the five-stranded β-sheet of severin DS111M. The four β-strands β1, β2, β3, and β4 run antiparallel to each other, while β-strands β4 and β5 run parallel. Arrows mark the NOE cross-peaks that occur within and between the backbones of the five β-strands. Solid arrows denote sequential Hα(i)-HN(i+1) cross-peaks, while thick and thin double-headed arrows and dashed arrows denote interstrand HN
i HN
j , Hα
i Hα
j , and Hα
i HN
j cross-peaks, respectively. Additionally, thin dotted lines indicate hydrogen bonds.
18 Magnetic Resonance Spectroscopy of Biomolecules
upfield shifts (i.e., smaller ppm values) in β-sheets (vice versa for Hα and Cβ). The method by which secondary structures are identified from secondary chemical shifts is known as the chemical shift index. Therefore, already in the early stages of structure determination the above-described methods facilitate the identification of regions with secondary structure in the protein. However, the relative spatial positions of these elements to each other as well as the global fold of the protein remain unknown. Calculation of the Tertiary Structure A computer-assisted structure calculation is used to convert the geometric data (distances and angles obtained from the analysis of the NMR spectra) into a three-dimensional structure of the biomolecule. However, the NMR data provide only a limited number of distance and angle restraints between atom pairs (preferentially the protons) that alone are insufficient to determine all atom positions. Luckily, this is only of minor importance because the bond geometries of many chemical groups are well known from X-ray diffraction experiments. This molecular information is contained in the so-called force field, which also includes further general atomic parameters such as van der Waals radii and electrostatic (partial) charges. While NMR data are directly obtained from a target molecule and within error limits are considered as “real,” force field parameters are derived from measurements on reference molecules. It is assumed that the force field parameters depend only on the chemical but not the spatial structure. Therefore, the force field constitutes a reasonable model for the real molecule. Because different possibilities exist to extract force field parameters from experimental reference data, several different force fields were developed for specific applications. Even though structure calculations depend on a force field, the experimental NMR data should determine the result irrespective of the choice of the force field. Additionally, compared to pure molecular dynamics simulations, NMR structure calculations use simplified force fields. For example, the solvent is generally considered only in the form of a fixed dielectric constant. The most important degrees of freedom that determine the 3D structure of a protein constitute rotations about the N–Cα bond (torsion angle, ϕ) and the Cα–C´ bond (torsion angle, ψ). To obtain reliable structures, NMR data should accurately define these angles and not the force field. Especially for loops at the surface, the N- and C-termini, or even the amino acid side chains this is often impossible due to their enhanced mobility. Thus, those regions do not assume a single defined structure, but rather fluctuate between different conformations. Flexible regions of biomolecules are therefore best described by an ensemble of conformations and a single static protein structure may not inevitably be the best representation of the microscopic reality. Practical experience shows that the number of distance restraints (NOEs) is more important than the accuracy with which the distances are determined. Thus, the distance classification introduced in Table 18.2 is sufficiently accurate for structure calculation. In principle, two different methods exist to calculate the structure of a protein in solution (which also can be combined):
The distance geometry method creates matrices from the NMR data and the force field that
contain distance bounds for all atom pairs. Through mathematical optimization methods Cartesian coordinates are calculated for all atoms, which fulfill the distance bounds reasonably well. However, the solution to this problem is ambiguous and one can calculate many independent structures that all reasonably well agree with the NMR data. Because distance geometry takes the covalent geometry only insufficiently into account, all distance geometry structures require further refinement. Simulated annealing is a molecular dynamics (MD) method. Newton’s equation of motion states that a force acting on an atom either accelerates or retards it. By numerically solving this equation, MD simulates the physical motion of atoms. NMR data are included as constraints and guarantee that the protein only adopts conformations that agree with the experimentally determined data. Starting from a template structure, a simulation period at high temperature allows the protein to find a structure that is compatible with the NMR data and the force field. Because at this stage both force field and experimental constraints are comparatively weak compared to the thermal energy, the simulated molecule can perform great conformational changes. The subsequent simulated annealing protocol decreases the simulation temperature and increases the impact of the force field and the experimental NMR constraints. In this way fluctuations in the structure are reduced until, finally, a 3D structure
461
462
Part II: 3D Structure Determination
Figure 18.33 (a) Stereo image of the 20 lowest-energy structures of severin DS111M. Only the heavy atoms of the protein backbone (N, Cα, C´ , and O) are shown. (b) Stereo image of the ribbon model of DS111M.
with minimal energy is obtained. Because the result of simulated annealing can depend on the starting structure, it is necessary to perform multiple calculations with different template structures. NMR structure calculations do not provide a single structure but a family of conformers, which occupy a relatively narrow conformational space. The root-mean-square deviation (RMSD) expresses the variability within this structure family, in which small deviations indicate a narrow conformational space. In general, one determines the RMSD for each structure of this family relative to an average structure (which has to be calculated in advance). Alternatively, one can determine the RMSD by pairwise comparison of two structures of the family and calculating the mean of these deviations. The RMSD differs for individual parts of the protein structure because regions lacking a defined secondary structure show larger deviations due to few NMR constraints. Similarly, flexible regions with an increased internal mobility also give rise to higher RMSD values. Relaxation measurements (Section 18.1.7, Determination of Protein Dynamics) can clarify if the variability within the structural family is due to insufficient NMR restraints or is caused by local dynamics. Figure 18.33 shows the result of a structure calculation for the protein severin DS111M. The individual conformers of the NMR ensemble are almost superimposable apart from a few poorly defined regions (the N-terminus and α-helix 2 at the C-terminus). Figure 18.33b displays DS111M as a ribbon model to enhance the presentation of the β-strands and α-helices. The orientation of the ribbon model is identical to the ensemble in panel (a). These structures were calculated with 1011 distance bounds and 55 ϕ torsion angles, and clearly demonstrate the necessity for a large number of constraints to obtain a well-defined structure.
18.1.7 Protein Structures and more — an Overview NMR spectroscopy constitutes a versatile method with the possibility to investigate multiple chemical and biophysical problems apart from the mere determination of protein structures. The
18 Magnetic Resonance Spectroscopy of Biomolecules
determination of a protein structure is, thus, not the end but rather the start of a structural and biophysical characterization of a protein. Knowledge of the three-dimensional structure allows for a meaningful approach towards functional aspects, such as protein dynamics, interactions with other molecules (e.g., proteins, DNA, or ligands), catalysis mechanism, hydration, and protein folding. An extensive description of these techniques is beyond the scope of this chapter. The following subsections therefore only give a short overview of possible experiments and applications. The bibliography at the end of the chapter contains a selection of important textbooks and review articles, which explain individual subjects more precisely and extensively. Speeding-up NMR Spectroscopy Due to the low sensitivity, NMR spectroscopy is a rather slow and time-consuming method. Even though simple 1D experiments last only a few seconds to minutes, the measurement time for more complex spectra increases dramatically with increasing dimensionality of the experiment. Two-dimensional experiments typically last for tens of minutes to hours; 3D experiments can take up to a few days to finish. Therefore, the acquisition of higher-dimensional spectra (4D or 5D), in which each indirect time dimension is incremented independently (as described above), is impractical. Mainly two factors determine the length of an experiment, the relaxation delay (Section 18.1.2, The 1D Experiment) and the number of increments in the indirect dimension(s) (Section 18.1.3, General Scheme of a 2D Experiment). During the relaxation delay (ca 1–5 s) the magnetization recovers to its equilibrium value through T1 relaxation. Hence, this delay determines how fast individual experiments (e.g., with different t1 increments) can be repeated. The SOFAST- and BEST-type NMR experiments are specifically designed to selectively excite only a subset of the spins (usually the amide protons). The unperturbed protons enhance the longitudinal relaxation of the excited spins through dipole–dipole interactions. As a result one can reduce the relaxation delay to a few hundred microseconds and thus increase the repetition rate five to ten times. The number of increments in the indirect dimension determines the achievable resolution in this dimension. Therefore, high-resolution spectra require more measurement time. Two different approaches exist to reduce the number of increments while maintaining the same resolution (or to increase the resolution in the same experiment time). In normal experiments the indirect time is incremented in constantly spaced intervals Δtn. In contrast, fewer randomlyspaced increments in the indirect dimension (ca 30% relative to conventional sampling schemes) are used in the so-called non-uniform sampling method. Mathematical approaches (maximum entropy or compressed sensing), which are distinct from Fourier transformation, convert the sparsely sampled indirect time-domain data into a conventional spectrum. The second approach relies on the simultaneous incrementation of at least two indirect timedomains to generate a two-dimensional projection of the n-dimensional spectrum. Imagine the three-dimensional cube of an HNCA spectrum with 1 H, 15 N, and 13 C on the x-, y-, and z-axes, respectively. Because of the co-incrementation of the 13 C and 15 N dimension, the resulting projection will intersect the 13 C–15 N plane at a certain angle with respect to the z-axis. Thus, in the projection the signals on the y-axis are the combination of the 13 C and 15 N frequencies, while frequencies on the unaffected x-axis correspond to the amide protons. Even though we are unable to imagine a four-dimensional object, mathematically it is very simple to transfer the described method to 4D experiments (and higher). Suitable projection reconstruction methods facilitate the creation of the n-dimensional spectrum from a limited number of projections. Alternatively, in automated projection spectroscopy (APSY) the chemical shifts for each peak are directly calculated from the projections without any reconstruction of the spectrum. For example, with projection spectroscopy one can acquire 6D experiments in three to four days. Determination of Protein Dynamics Proteins are not rigid, static entities, but rather exist as ensembles of conformations. The internal motions, which give rise to transitions between different structural states, are collectively referred to as protein dynamics and occur on a wide range of time scales (ns–s). Protein dynamics play crucial roles in protein functions particularly for the interaction with binding partners, in enzyme catalysis, and in allosteric regulation. Luckily, many NMR spectroscopic parameters depend both on the mobility of the whole molecule (translational or rotational motions) and on internal motions (transitions between different conformations or bond rotations), providing insight into a wide variety of dynamic
463
464
Part II: 3D Structure Determination
processes ranging from fast fluctuations (lasting picoseconds) to slower conformational changes (lasting microseconds or more). The measurement of relaxation parameters is possible for different nuclei. While 15 N relaxation data of the amide group provide information about the backbone flexibility of each amino acid, relaxation measurements of side chain groups (especially the methyl groups of valine, isoleucine, and leucine) determine their respective mobility. The 15N-relaxation measurements have the advantage that they can be performed on inexpensive, 15N-labeled protein. In addition, a simplified model can be used to analyze 15N relaxation data because it is primarily the directly bound proton that determines the relaxation of the 15 N spin. In this model three parameters characterize the dynamic behavior: the rotational correlation time and the amplitude and the time scale of local motions. The correlation time τc describes the statistic motion of the whole molecule in the form of a rotational diffusion, which for proteins is on the order of several nanoseconds. The 15N-relaxation measurements have been particularly successful in the identification and characterization of excited protein states that have populations as low as a few percent but which are important for protein function and misfolding. Thermodynamics and Kinetics of Protein–Ligand Complexes NMR spectroscopy can elucidate many aspects of the interaction of proteins with small ligands, polypeptides and other proteins, or nucleic acids (DNA and RNA). It characterizes the dynamic, kinetic, and thermodynamic properties of protein–ligand complexes. Even without precise structural information about the ligand, it is possible to identify the residues of a protein involved in binding a ligand. To this end, 15 N- or 13 C-labeled protein is titrated with an NMR-invisible, unlabeled ligand and a HSQC is recorded at each titration point. Due to ligand binding, some peaks display changes in chemical shift and line width, which allow for the identification of the involved residues. Furthermore, analysis of the chemical shifts changes as a function of ligand concentration facilitates the determination of the respective dissociation constant. Detailed binding information can also be obtained when the protein cannot be labeled with 13 15 C/ N. Under suitable conditions (weak binding of a small ligand to a large receptor), so-called transfer NOEs yield the receptor-bound conformation of a ligand even though the receptor is too large for NMR spectroscopy. Additionally, one can detect an interaction between a ligand and a receptor through saturation transfer, which results in intensity changes for the NMR peaks of the ligand. While the ligand is in large excess, the receptor is essentially invisible due to its dilution by a few orders of magnitude. This technique is particularly useful to screen ligands for certain target receptors. Additionally, NMR spectroscopy allows for the determination of the translational diffusion, which depends on the size and shape of the molecule. Therefore, one can determine if a protein is a monomer, a dimer, or if it forms a complex with other proteins. Protein Folding and Misfolding NMR spectroscopic techniques also allow atomic level insight into the folding and misfolding of proteins. In combination with sub-zero temperatures (down to 15 °C) or high pressures (up to 2 kbar) it is possible to determine the 3D structure of partially folded equilibrium intermediates, and to analyze the kinetics of protein folding pathways. In addition, through combinations of H/D exchange and other NMR methods one can follow the formation of hydrogen bonds within a stable secondary structure during the folding process. Because H/D exchange and folding are competitive reactions, one can determine the velocity of the secondary structure formation from the exchange rate of individual protons. Alternatively, a so-called quenched flow apparatus permits H/D exchange reactions only during certain time periods of the folding process. Thus, a time-resolved picture of the formation of secondary and tertiary structures becomes accessible. Furthermore, 15N-relaxation dispersion as well as real-time NMR methods offer unique insight into the processes of folding and misfolding of proteins. Intrinsically Disordered Proteins Intrinsically disordered (also called natively unstructured) proteins (IDPs) lack a well-defined tertiary structure, but still fulfill critical biological functions as part of signal transduction cascades, in spliceosomes, or in cancerassociated processes. The degree of disorder ranges from completely unfolded proteins to mostly folded proteins, which contain disordered regions of 30 residues or more. IDPs possess a very distinct amino acid composition with a high abundance of proline, polar, and charged amino acids (serine, glycine, glutamate, arginine). At the same time, they are depleted in hydrophobic and aromatic amino acids that usually form the hydrophobic core of globular
18 Magnetic Resonance Spectroscopy of Biomolecules
proteins. Due to their dynamic nature, only NMR is capable of providing structural information about IDPs with atomic resolution. The low sequence complexity and the absence of stable structure result in similar chemical environments for each residue. Therefore, IDPs display a narrow chemical shift dispersion (especially the amide protons) that results in severe signal overlap. Chemical shift degeneracy of Cα and Cβ further limits the application of standard triple-resonance experiments (for each amino acid type the chemical shifts are close to the random coil shift) (Section 18.1.4, TripleResonance Experiments). Because the 15 N (and 13 C´ ) dimension provide the highest dispersion in IDPs, sequential connectivities are established with 3D experiments that correlate to two nitrogen spins (e.g., HNN or (H)CANNH). Alternative approaches rely on carbon detection, which correlate the C´ spins with the directly bound nitrogen. For IDPs, the so-called NCO experiments provide improved resolution when compared to 15N-HSQCs, and also allow for the detection of proline residues. Due to the lower gyromagnetic ratio, however, carbon-detected experiments are significantly less sensitive than those that use proton detection. The disordered state imparts IDPs with favorable relaxation properties. Therefore, IDPs are amenable for high dimensional 5D–7D experiments, which provide optimal resolution. In addition, the longer transverse relaxation times of IDPs enable the characterization of proteins significantly larger than 30 kDa. Even without deuteration it was possible to sequentially assign the two microtubule-associated proteins tau and MAP2c, both of which are larger than 45 kDa. IDPs exist as ensembles of rapidly interconverting structures and therefore only provide a few NOE distance restraints (often only short-range). The secondary chemical shifts (Section 18.1.6, Determination of the Secondary Structure) obtained from the sequential assignment allow for the identification of transient secondary structural elements. Often those elements become stabilized upon binding to protein interaction partners. Long-range distance restraints to describe the structural ensemble can be obtained from RDCs and PREs (paramagnetic relaxation enhancements). To measure PREs, a paramagnetic spin label (e.g., a nitroxide) is attached to the IDP. The spin label enhances the relaxation of nearby residues (ca. 25 Å) resulting in line broadening and thus in an intensity decrease. Normalizing the reduced intensity by the intensity of the respective residue in the absence of the spin label allows for the calculation of the distance between the nucleus and the spin label, which is of course an ensemble average. Sophisticated computer programs can utilize this entire structural information together with data from small-angle X-ray scattering to calculate ensemble structures. Structure and Dynamics of High Molecular Weight Systems and Membrane Proteins To extend the application of heteronuclear experiments to even larger proteins or protein complexes of several hundred kilodaltons, Kurt Wüthrich and colleagues developed a technique called TROSY (transverse relaxation optimized spectroscopy) at the end of the last century. The TROSY technique achieves the compensation of two relaxation mechanisms (the dipolar interaction and chemical shift anisotropy). Especially for large proteins, the resulting narrow line widths reduce the signal overlap and improve the sensitivity of the experiments. Due to its modularity, TROSY can be combined with standard 3D and triple-resonance experiments (e.g., NOESY-HSQC or HNCA) to enable a conventional sequential assignment. With this approach the structure of the 81 kDa protein malate synthase G was determined. In addition, the structure of α-helical membrane proteins, which are solubilized in detergent micelles, bicelles, or nanodiscs, can be determined using TROSY techniques. This facilitates pharmacological studies providing detailed insight into the interaction of drugs with membrane receptors. Large protein complexes give rise to vast numbers of peaks with a great potential for chemical shift degeneracies. Therefore, to minimize spectral crowding, TROSY experiments are restricted to the analysis of an isotopically labeled subunit in an otherwise unlabeled complex, or to complexes consisting of symmetrical subunits. To circumvent the problems associated with the multitude of signals in large systems, Lewis Kay and colleagues developed the methyl-TROSY approach. This method combines the advantages of the sharp lines from the TROSY with the limited number of signals provided by selective amino acid labeling (Section 18.1.5, Selective Amino Acid Labeling). Hence, TROSY-type relaxation experiments on protonated methyl groups (isoleucine, leucine, valine, alanine, methionine, or threonine) in an otherwise perdeuterated protein allowed for the functional analysis of the gating mechanism in the proteasome. For very large proteins (>200 kDa) the magnetization transfer through scalar coupling (also used in the TROSY-HSQC) becomes ineffective. For those proteins, the CRIPT and CRINEPT
465
466
Part II: 3D Structure Determination
techniques achieve an efficient magnetization transfer via cross-relaxation. Solid-state NMR represents an alternative to solution NMR methods for the study of large molecular weight systems. As the name implies, proteins are not analyzed in solution but as solid powders or microcrystals. “Magic angle” spinning of the solid protein in specially designed rotors at several thousand Hertz results in line narrowing of the resonances. Theoretically no size limitation exists; however, the complexity of the spectra for large protein restricts their analysis. Solidstate NMR mainly relies on the detection of 13 C and 15 N spins; efforts are underway to directly detect protons at very high spinning speeds (> = 60 kHz). Due to the low sensitivity of the respective nuclei, solid-state experiments are often restricted to two dimensions. However, the development of dynamic nuclear polarization (DNP) techniques to enhance the sensitivity holds the promise of the development of higher-dimensional experiments. In-cell NMR Spectroscopy For structural characterization proteins are highly purified, whereas in the cell the protein coexists with other cellular parts, membranes, and thousands of other proteins at very high concentration (200–300 g l 1). Therefore, it is of great interest to analyze the structure and dynamics of proteins in a cellular context. To address this need, NMR spectroscopy was applied to intact cells, so-called in-cell NMR. Initially, in-cell NMR methods were developed for bacterial cells and the structures of small proteins could be solved. More recently the focus also shifted to mammalian cells. The simplest approach for in-cell NMR is to cultivate the cells in media supplemented with isotopically labeled amino acids and to overexpress the target protein. However, metabolites often produce strong background signals and thus decrease the contrast in the resulting spectra. To obtain clear spectra, protein delivery systems were developed that introduce isotopically labeled proteins (obtained from heterologous bacterial expression) into unlabeled cells. Good delivery efficiencies might be obtained by electroporation or Lipofectamine transfection. Currently, in-cell NMR measurements are limited to only a few hours due to acidification of the buffer and the resulting stress for the cells. Yet, this time enables the measurement of 2D HSQC spectra that can provide information about intracellular protein–protein interactions, post-translational modification, or protein dynamics. To circumvent some of the problems associated with mammalian cells, it may sometimes be more advantageous to work with cell lysates or cytoplasmic extracts. These extracts are easy to make and the reaction conditions can be better controlled, for example, when post-translational modifications are analyzed. Microinjection of proteins into large Xenopus laevis oocytes constitutes another alternative.
18.2 EPR Spectroscopy of Biological Systems Olav Schiemann and Gregor Hagelüken University of Bonn, Institute of Physical and Theoretical Chemistry, Wegelerstr. 12, 53115 Bonn, Germany
Electron paramagnetic resonance (EPR) or electron spin resonance (ESR) is a spectroscopic method that is used to obtain information on the chemical nature, structure, dynamics, and local environment of paramagnetic centers. Such centers are defined by having one or more unpaired electrons. In biological macromolecules these can be metal ions (e.g., Cu(II), Fe(III), Mn(II), Mo(V)), metal clusters (e.g., iron-sulfur clusters or manganese), or organic radicals. Organic radicals are formed, inter alia, as intermediates in electron transfer reactions of proteins (e.g., semiquinone anion, thiyl, or tyrosyl radicals) or they are induced by radiation damage in DNA molecules (e.g., sugar or base radicals) (Figure 18.34). Frequently, these centers are involved in catalytic cycles or in biologically relevant reactions. Diamagnetic biomolecules, in which all electrons are paired, can be made accessible to EPR spectroscopy by spin labeling techniques. In particular, nitroxides can be site specifically and covalently linked to biomolecules. This site directed spin-labeling approach together with EPRbased distance measurements between spin labels can be used to study configuration changes, to obtain coarse-grained structures of a whole biomolecule or to localize paramagnetic centers. Similar to NMR spectroscopy, which was described in the previous section, EPR spectroscopy is a magnetic resonance technique, in which the normally degenerate ms = ±½ levels of an electron spin s = ½ are split by an externally applied magnetic field, and transitions between
18 Magnetic Resonance Spectroscopy of Biomolecules
467
Figure 18.34 Examples of paramagnetic centers in biological systems. The wavy lines and –R represent the peptide chains, oligonucleotides, or cofactors. (a) Cu(II) in plastocyanin, (b) Fe4S4 clusters in ferredoxins, (c) Mo(V) in dimethyl sulfoxide reductases, (d) 3´ -sugar radical in γ-irradiated DNA, (e) thymyl radical in γ-irradiated DNA, (f) benzosemiquinone anion radical, (g) tyrosyl radical in photosystem II, (h) thiyl radical in ribonucleotide reductases, (i) 1-oxyl-2,2,5,5-tetramethylpyrroline-3-acetylene (TPA), (j) 1-oxyl-2,2,5,5tetramethylpyrroline-3-methyl) methanethiosulfonate (MTSL; nitroxide spin label for proteins), (k) 4,4-dimethyl-oxazolidine-3-oxyl-based nitroxide spin label for membranes.
these levels are induced by microwaves. The first continuous-wave (cw) EPR experiment was performed in 1944 by E.K. Zavoisky and the first pulsed EPR experiment in 1961 by W.B. Mims. The high technical requirements for pulsed EPR experiments, however, meant that pulsed EPR spectrometers only became commercially available in the late 1980s, about two decades later than for NMR. Since then, and in conjunction with the ongoing development of high frequency/high-field EPR spectrometers, computer-based EPR simulation programs, and quantum chemical methods for the translation of the EPR parameters into structural data, EPR spectroscopy has become increasingly important.
18.2.1 Basics of EPR Spectroscopy At this point, the physical principles of EPR parameters will be briefly outlined. For a more indepth treatment, a quantum mechanical description is inevitable. For this, the reader is referred to the references in the “further reading” section at the end of this chapter. Electron Spin and Resonance Condition Since the Stern–Gerlach experiment it has been known that an unpaired electron has a quantum mechanical angular momentum, the so-called electron spin s. The length of the vector s is given by: pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi jsj ħ s
s 1
(18.15)
468
Part II: 3D Structure Determination
Here s = ½ is the spin quantum number and ħ is Planck’s constant divided by 2π. As with any charged particle with an intrinsic angular momentum, the electron spin is linked with a magnetic moment μe, which is oriented along the electron spin vector, but opposite to it due to its negative charge: μe ge
e s 2me
(18.16)
Here, e is the elementary charge, me the resting mass of the electron, and ge = 2.0023 is the gfactor of the free electron. In an external magnetic field B, there are only two possible orientations of the spins, parallel or anti-parallel to the external field. The component of the electron spin in the direction of the magnetic field B0 (usually defined as the z-direction) is: sz ms ħ
(18.17)
The magnetic quantum number ms can take values of +½ and ½ for s = ½. Due to Heisenberg’s uncertainty relation, the sx and sy components cannot be simultaneously determined with sz. From the orientation of the electron spin along the z-axis follows a corresponding orientation of the magnetic moment μe. Its z-component μe,z is given by: μe;z ge μB ms
(18.18)
where μB is the Bohr magneton, which itself is given by: μB
eħ 9:274 10 2me
24
JT
1
(18.19)
Thus, the energy E of a magnetic moment that is oriented along the z-axis of the external magnetic field is: E μe;z B0 ge μB ms B0
(18.20)
While, without a magnetic field, the two spin orientations ms = ½ and ms = +½ are energetically degenerate, they are split by applying an external magnetic field (Zeeman splitting). The energies E of the two orientations are given by: 1 E
ms 1 ge μB B0 2 2 E
ms 1 2
1 g μ B0 2 e B
(18.21) (18.22)
The energy difference ΔE between the two energy levels is proportional to the magnitude of B0 (Figure 18.35): ΔE ge μB B0
(18.23)
In a macroscopic sample with N spins, more spins are present in the low energy ms = ½ spin state than in the ms = +½ state. The ratio of the occupation numbers N
ms 1 /N
ms 1 is 2 2 described by the Boltzmann distribution: N
ms 1 2
N
ms 1 2
Figure 18.35 (a) Zeeman splitting for an electron spin in a magnetic field B0. (b) The absorption line, which is obtained when the resonance condition is satisfied, and (c) the first derivative obtained by the field modulation of the absorption line.
e
ΔE kT
e
ge μB B0 kT
(18.24)
For T = 300 K, B0 = 340 mT and k = 1.3806 × 10 23 J K 1 (Boltzmann constant) a ratio of the occupation numbers of 0.999 can be calculated from this equation (NMR: 0.99999). Thus, at room temperature, the lower energy ms = ½ state is populated slightly more. Using electromagnetic radiation, spins from the ms = ½ level can be excited into the ms = +½ level, when the energy of the radiation is equal to ΔE and thereby satisfies the resonance condition: hν ΔE ge μB B0
resonance condition
(18.25)
Here, as in NMR spectroscopy, the magnetic field component of the electromagnetic radiation interacts with the magnetic moment of the electron spin.
18.2.2 cw- EPR Spectroscopy In cw-EPR spectrometers, microwaves at a constant frequency are continuously applied to the sample and the external magnetic field is swept. The experiment is set-up such that at
18 Magnetic Resonance Spectroscopy of Biomolecules
some point during this process the resonance condition is met. Most commonly, cw-EPR spectrometers operating at a microwave frequency of about 9.5 GHz (X-band) are used, so that for a radical with g 2 the resonance absorption occurs at a magnetic field of about 340 mT. Since the ratio of the occupation numbers is close to one, the absorption signal is of relatively low intensity. In cw-EPR spectrometers, the sensitivity is increased by adding a small modulated magnetic field component to the external magnetic field, which enables the lock-in detector to filter-out noise. This field modulation is also the reason that the absorption signal is obtained in the form of its first derivative (Figure 17.35). In addition, the sensitivity can be further improved by making measurements at low temperatures and stronger magnetic fields (increase of the occupation ratio). To indicate the position of the line independently of the magnetic field and the microwave frequency, the position of the absorption line is expressed as a g-value (similar to the chemical shift of NMR spectroscopy): g
ν hν 7:144775 10 2
in MHz=mT μB B0 B0
(18.26)
18.2.3 g-Value
469
Figure 18.36 Splitting scheme for an unpaired electron (blue), which is strongly coupled to a 1 H nucleus (I = ½, black); Aiso is assumed to be positive. The dashed lines show the transitions, which are allowed by the EPR selection rules (Δms = ±1 and ΔmI = 0). Thus, in the EPR spectrum two lines, centered around giso and separated by Aiso, would be observed.
For a free electron only one line would be observed, with g = ge = 2.0023. However, if the unpaired electron resides in a molecule, deviations from ge occur due to spin–orbit coupling. These deviations are characteristic for the electronic state, the bonding situation, and the geometry of the particular molecule. For organic radicals, these deviations are usually small, since their magnetic orbital moment is nearly zero. The nitroxide TPA (Figure 18.34i), for instance, has a g-value of 2.006. For transition metal ions, however, significantly larger deviations can be observed. The Fe+ in an MgO matrix, for example, has a g-value of 4.1304. The g-values can thus be used to characterize and distinguish different paramagnetic species. A quantitative calculation or a translation of observed g-values into structural parameters with, for example, density functional theory (DFT) methods is, however, still complicated, especially for transition metal ions, and is the subject of current research.
18.2.4 Electron Spin Nuclear Spin Coupling (Hyperfine Coupling) In addition to the spin–orbit coupling, the magnetic moment of the electron spin can be coupled to the magnetic moments of nuclei (Figure 18.36) if the nuclear spin I is greater than zero (e.g., 1 H, 14 N, 31 P, 13 C, 17 O, 33 S, 55 Mn, 95 Mo, 57 Fe, 51 V). The coupling of the electron and the nuclean magnetic moments leads to a splitting of the absorption line into M = 2NI + 1 lines (multiplicity rule). Here, N is the number of equivalent magnetic nuclei and I is their nuclear spin quantum number. The magnitude of the splitting, which is called the hyperfine coupling constant Aiso, depends linearly on the magnetic moment μI of the coupling nucleus and the spin density jψ
r 0j2 of the unpaired electron at the nucleus (Fermi contact interaction): Aiso ∼Mi jψ
r 0j2
(18.27)
Since only s-orbitals have a non-zero value at the nucleus but the unpaired electron in most radicals resides in a p-, π, or d orbital, the question arises as to why such radicals show a hyperfine coupling in the first place. The reason for this is spin polarization, a mechanism that generates spin density in low energy s orbitals (Figure 18.37). If an EPR spectrum of a radical is obtained with resolved hyperfine coupling structure, the spin density distribution can be determined from the hyperfine coupling constants and thus statements regarding the nature and structure of the radical can be made. An example is the cw-EPR spectrum of the nitroxide TPA (Figure 18.38a). As in all alkyl nitroxides the unpaired electron resides in a π orbital between the nitrogen and oxygen atom. At both nuclei, spin density is generated through spin polarization. However, since the most abundant oxygen isotope, 16 O, has a nuclear spin of I = 0, no line splitting is induced by this nucleus. In contrast, nature’s most abundant nitrogen isotope, 14 N, has a nuclear spin of I = 1, so the absorption line is split into a triplet with a hyperfine coupling constant
Figure 18.37 Mechanism of spin polarization: the unpaired electron with an α-spin in the p-orbital has its largest probability at a relatively large distance from the nucleus. According to Hund’s rule, an electron in the s orbital has an electron spin (α) parallel to the unpaired electron in the p-orbital, if it is also far away from the nucleus. Due to Pauli’s rule, the second s-electron close to the nucleus must then be set anti-parallel to the first s-electron (β). In this way, excess β-spin density is induced at the nucleus, since the probability for α- and β-s electrons at the nucleus is no longer equal.
470
Part II: 3D Structure Determination
Figure 18.38 (a) cw-X-band EPR spectrum of TPA in liquid solution. (b) Model for an electron transfer chain. The reduction of Co(III) to Co(II) is carried out by an intramolecular electron transfer. The electron is first transferred from the Cr(II) to the pyrazine and only then to the Co(III). The pyrazine radical could be detected by EPR spectroscopy on the basis of the hyperfine signature and the g-value. Often the cw-EPR spectra are shown without an x-axis and only a scale ruler is given. Source: Spiecker, H. and Wieghardt, K. (1977) Inorg. Chem., 16, 1290–1294. With permission, Copyright 1977, American Chemical Society.
Aiso = 1.4 mT = 39.2 MHz. The additionally observed small lines on the low and high field side of each 14 N-line are caused by 13 C hyperfine coupling to one of the directly bonded carbon nuclei (I = ½) or one of the carbon nuclei of the methyl groups. This hyperfine coupling of 0.6 mT splits each of the three lines into a doublet. However, due to the low natural abundance of 13 C (1.1%), the intensity of this triplet of doublets is low, so that the 14 N triplet dominates the EPR spectrum. Actually, each of the three 14 N lines would have to be split by six 13 C nuclei. However, the probability that multiple 13 C isotopes are found within one molecule is so small that these couplings are not observed. Although, the question of the spin density distribution may appear merely academic for a nitroxide, it is important for cofactors in electron transfer proteins. For example, to understand electron transfer mechanisms in proteins, it is crucial to know whether the transported electron can reside on a cofactor in an electron transfer pathway or which part of a co-factor acts as the electron acceptor or electron donor (Figure 18.38b).
18.2.5 g and Hyperfine Anisotropy The considerations in the preceding sections are based on radicals in liquid solution, that is, the radicals rotate so fast compared to the EPR time scale (actually compared to the size of the anisotropy of the interaction) that they do not a have fixed preferential orientation with respect to the external magnetic field B0. Spectra acquired under such conditions are called isotropic. If the sample is frozen, a powder or a single crystal, each molecule has a fixed orientation with respect to B0 and the EPR spectra are characterized by orientationdependent (anisotropic) contributions. Such contributions can be found for g, the hyperfine coupling and the coupling between the unpaired electrons (Section 18.2.6). Although the occurrence of these anisotropic contributions in an EPR spectrum renders its analysis more difficult, they also offer more detailed information about the nature and structure (angles and distances) of the paramagnetic system. For this reason, as well as to increase the sensitivity and to extend the lifetime of short-lived radicals; most EPR experiments on biological systems are performed in frozen solution at very low temperatures (down to, for example, 3.5 K). g Anisotropy If the unpaired electron resides in an s orbital, for which, due to its spherical symmetry, the three spatial directions x, y, and z are equivalent, then gx = gy = gz. For EPR spectra in such a spherically symmetric case and in the absence of hyperfine interactions, one observes one line with a g-value at giso even in the solid state. Anisotropy in g occurs when the unpaired electron resides in an orbital of the molecule that is not spherical. In the axisymmetric case (two equivalent directions in space), a spectrum as shown in Figure 18.39a with two canonical g-values (g? and g|) is obtained. In the orthorhombic case (x ≠ y ≠ z) all three g-values are different (gx ≠ gy ≠ gz) and one obtains a spectrum similar to the one shown in
18 Magnetic Resonance Spectroscopy of Biomolecules
Figure 18.39b. The indices x, y, and z stand for the three canonical directions in space, meaning that g changes in accordance with the orientation of the molecule with respect to B0. In liquid solution, however, when the molecule rotates rapidly, the anisotropic components of g are averaged out such that: giso
1 gx gy gz 3
(18.28)
1 2g? gk 3
(18.29)
471
Figure 18.39 (a) Axial cw-X-band EPR spectrum. The first derivative is given at the bottom and, for clarity, the absorption spectrum at the top. The absorbance at g = gz = g| results from molecules that are oriented with g| parallel to B0, while at gx = gy = g? only those molecules absorb that are oriented with g? parallel to B0. Between these two points, only those molecules contribute to the absorption that have neither g| nor g? parallel to B0. An example of an axially symmetric molecule with resolved g| = 2.00 and g? = 5.67 is the high-spin Fe(III) porphyrin of cytochrome P450. Source: from Woggon, W.-D. et al. (1998) Angew. Chem., 110, 3191–3195. With permission, Copyright 1998 WILEYVCH Verlag GmbH, Weinheim, Germany. (b) Orthorhombic cw-X-band EPR spectra. A system with orthorhombic symmetry and g-values at g1 = 2.196, g2 = 2.145, and g3 = 2.010 obtained from the Ni center in the [Ni -Fe] hydrogenase. Source: Foerster, S. et al. (2005) J. Biol. Inorg. Chem., 10, 51–62. With permission, Copyright 2004, SBIC. The assignment of x, y, and z to the corresponding g-values is possible by measurements on single crystals.
or: giso
High Field/High-Frequency EPR Spectrometer In many cases, especially in organic radicals, the three g-values are not separated at a magnetic field of 340 mT/microwave frequency of 9.5 GHz (X-band) due to the relatively small g-anisotropy. But it is possible to resolve the g-anisotropy by using higher frequencies/fields (Figure 18.40). This exploits the fact that the separation of the g-values (measured in magnetic field units) increases with the magnetic field. In contrast, the size of the line splitting due to the hyperfine coupling is independent of the magnetic field. EPR spectrometers working at higher frequencies and that are commercially available are those working at 36 GHz (Q-band)/1.3 T, 95 GHz (W-band)/
Figure 18.40 Magnetic field dependent splitting of gx, gy, and gz. An example of the cw-EPR spectra of a semiquinone anion radical in X(9.5 GHz/0.34T)- and G (180 GHz/T 6,4)-band. The g-values themselves do not change, only the distance (in mT) between them increases.
472
Part II: 3D Structure Determination
3.4 T, or 260 GHz/9.2 T. Spectrometers with even higher frequencies (360 GHz, 640 GHz, and in the THz range) are also used, but are technically very demanding and not yet commercially available. However, the technical effort is worthwhile, not only for the resolution of the gvalues, but also because of the following advantages:
Superimposed spectra of different radicals can be separated by their different g-values. The sensitivity of the spectrometer increases at higher magnetic field/frequency (due to the Figure 18.41 Illustration of the distance vector r and the angle θ. A and B can be either an electron and a nucleus or two electrons (Section 18.2.6).
larger population difference), which means that smaller sample amounts are needed (e.g., at 180 GHz, 0.1 μl of a 1 mM Mn(II) solution is sufficient). If the same sample is measured at different microwave frequencies, the magnetic-fieldindependent hyperfine splitting constant can be separated from the magnetic field dependent g-splitting (in magnetic field units).
Hyperfine Anisotropy An anisotropic hyperfine coupling component Ai, which is observed in a spectrum, is the sum of the isotropic hyperfine coupling constant Aiso and an anisotropic, purely dipolar component Ai,dip. The subscript i stands for one of the three spatial directions x, y, or z, and means that the hyperfine coupling, depending on the spatial direction or orientation of the radical, varies relative to B0:
Ai Aiso Ai;Dip
(18.30)
The anisotropic portion of the hyperfine coupling arises from the dipole–dipole coupling between the magnetic moments of the electron and the nucleus. It depends both on the distance r between the electron and the nucleus as well as on the angle θ between the distance vector r and the external magnetic field B0 (Figure 18.41). Depending on the symmetry of the paramagnetic center, a distinction of three special cases for the hyperfine coupling is made: spherical symmetry with Ax = Ay = Az, axial with A? (= Ax = Ay) and A| (= Az), and orthorhombic with Ax ≠ Ay ≠ Az (Figure 18.42). In liquid solution the dipolar hyperfine coupling components cancel each other out, so that: Aiso
1 Ax Ay Az 3
(18.31)
1 2A? Ak 3
(18.32)
Aiso
18.2.6 Electron Spin–Electron Spin Coupling If, within a biomolecule, two unpaired electrons A and B are located at a fixed distance r to each other, a coupling νAB between these two electrons can occur. As seen for the hyperfine coupling constant, this electron–electron coupling is the sum of an isotropic and an anisotropic component: νAB J D
(18.33)
The isotropic part, J, is the exchange interaction, which is non-zero when the wave functions of both electrons overlap (R < 10 Å) or if the two electrons are interacting via a conjugated bridge
Figure 18.42 Hyperfine and g anisotropy for the example of the cw-G (180 GHz/6.4 T)-band EPR spectrum of TPA in frozen solution. The Ax and Ay hyperfine couplings are not resolved.
18 Magnetic Resonance Spectroscopy of Biomolecules
473
(super exchange). It can be observed in liquid solution and allows to draw conclusions as to whether the two electron spins are aligned parallel (ferromagnetic) or antiparallel (antiferromagnetic) to each other. The anisotropic component D of the coupling is based on the interaction between the magnetic dipoles of the two electrons and is orientation-dependent. It can only be observed when the biomolecules have a fixed orientation with respect to B0 (frozen solution, powder, or single crystal). In liquid solution, the anisotropic component is averaged out: D νDip 1
νDip
3 cos2 θ
m2B gA gB μ0 1 3 4πh r AB
(18.34)
(18.35)
where: θ is the angle between the distance vector r and the external magnetic field B0 (Figure 18.41); gA and gB are the g-values of the two electrons; μ0 is the permeability in vacuum; mB is the Bohr magneton. The first term in Equation 18.35 (νDip) is therefore a constant. In an anisotropic cw-EPR spectra, the electron–electron dipolar coupling manifest itself as an additional line splitting if it is larger than the line width. If the dipolar coupling constant ν dip can be determined from a spectrum, then the distance between the two unpaired electrons can be calculated. Using cw-EPR spectra, distances of up to 20 Å can be determined in this way. At greater distances, the dipolar coupling is obscured by the intrinsic linewidth and pulsed methods can be used (see also Section 18.2.7). Such distance measurements are important in the determination of the arrangement of paramagnetic centers in biological macromolecules (Figure 18.43). Similarly, the distance between two subunits of a protein can be measured by site-directed spin labeling of the subunits with two nitroxides and concomitant determination of the dipolar coupling between the two spin labels. By varying the spin label positions, and measuring the distances between the respective pairs, it is possible to figure out the spatial arrangement of the subunits. In the same way one can examine whether the binding of ligands (small organic molecules, metal, proteins, RNA, DNA) leads to global structural changes.
18.2.7 Pulsed EPR Experiments With cw-EPR methods, strongly coupled nuclei in the vicinity of the unpaired electron can be characterized. Pulsed EPR methods allow the resolution of weaker couplings to more distant nuclei or distant unpaired electrons, which in the cw spectrum are hidden underneath the broad spectral lines. Other advantages of pulsed experiments are that the pulse sequences allow selection of those contributions to the spectrum that one would like to analyze and that multidimensional experiments are possible. In particular, with experiments such as ESEEM (electron spin echo envelope modulation) and pulse ENDOR
Figure 18.43 Example of a biological system with electron–electron dipolar coupling. (a) Arrangement of the binuclear CuA center and the Mn(II) ion in cytochrome c oxidase from Paracoccus denitrificans. (b) cw-W-band (94 GHz/3.3 T) EPR spectra of the Mn(II) center. The Mn(II) ion (s = 5 =2 and I = 5 =2 ) leads to a W-band cw-EPR spectrum with six lines. In this case, the CuA center has been switched into the diamagnetic s = 0 state, so that it cannot be detected by EPR spectroscopy. If the sample conditions are then adjusted such that the CuA center turns into the s = ½ state, a cwEPR spectrum can be observed in which each manganese line is split into a doublet. From the magnitude of the splitting (2.2 mT) the dipolar coupling constant was determined to be 3.36 mT and with this a CuA–Mn distance of 9.4 Å could be calculated. The paramagnetic CuA center itself was not detected directly. Source: Käß, F. et al. (2000) J. Phys. Chem. B, 104, 5362–5371. With permission, Copyright 2000, American Chemical Society.
474
Part II: 3D Structure Determination
(electron nuclear double resonance) detailed conclusions about the structural environment of the unpaired electron can be made (within a radius of up to 10 Å). With PELDOR (pulsed electron–electron double resonance), even distances of up to 8 nm between two electron spins can be measured. Larmor frequency: Frequency with which an electron or nucleus spins about the z-axis, the direction of an externally applied magnetic field. The name commemorates the Irish physicist Sir Joseph Larmor.
Figure 18.44 (a) Illustration of the spin orientation relative to B0. (b) Illustration of the magnetization M (big arrow). The black arrows represent the magnetic moments μe,z of the electron spins.
Basics The basics of pulsed EPR spectroscopy are comparable to those of pulsed NMR spectroscopy (Section 18.2.1). For EPR, these considerations must simply be transferred to the electron spin. In Section 18.2.1 it was shown that, for a given electron spin, the length of the s-vector and its z-component are defined, while the x and y components are undefined. Note that the magnetic moment of the spins is thus not exactly parallel to B0 (Figure 18.44a) – even if we say so in the following. However, the presence of the magnetic field leads to the electron spins precessing on cones parallel and anti-parallel to the B0 magnetic field axis (Figure 18.44b). The magnetic-field dependent frequency of this precession is known as the Larmor frequency. Since, according to the Boltzmann distribution, more spins are oriented with their magnetic moments μe parallel to B0, the sample experiences a macroscopic magnetization M parallel to B0.
Pulses To generate microwave pulses, the microwave radiation is turned on, and quickly turned off again, after a short period of time. This corresponds to a rectangular microwave pulse. In EPR spectroscopy, such pulses typically have pulse lengths in the range of nanoseconds (NMR: microseconds). Because of the very short pulse lengths, the pulses do not have a singular microwave frequency (such as the microwaves in the cw experiment), but a certain range of frequencies (Heisenberg uncertainty principle). It follows that, at a constant magnetic field, a larger part of the EPR spectrum can be excited. In the following, 90°- or 180°-pulses designate such pulses, which rotate the electron spin by 90° or 180°. In this case, all pulses are irradiated along the x-axis (from x to +x) and the magnetization is rotated clockwise. Both pulse width and pulse amplitude affect the magnitude of the rotation angle. Thus, a 180° pulse has either twice the length or a twofold power of a 90° pulse.
If a 90° pulse is applied to a sample, then this pulse interacts with the spins and rotates M from the z to the +y-axis (Figure 18.45). In Section 18.2.1 it was stated that spins can only align parallel or antiparallel to the field. This raises the question of how the individual spins must be aligned to cause a magnetization in the +y direction. This can be illustrated in the following way: The 90° pulse induces an equal population of spins on both energy levels, making the individual magnetic moments along the zaxis add up to zero. Simultaneously, the spins are no longer uniformly distributed on either of the two precession cones but form spinning packages in the direction of the +y axis. Since the spins in these packages have the same orientation and speed, this phenomenon is called phase coherence (Section 18.1.1 and Figure 18.5). The spinning packages are not static along the +yaxis, but precess at their Larmor frequency around the z-axis. In contrast to the cw experiment in which the absorption of the microwave is detected, this precession of the spins induces a current or a signal in the detector, which can then be recorded. Relaxation In general, relaxation means the return from the excited state to the ground state. Due to random spin–spin interactions, the phase coherence described above is lost with time, so that the spins are again evenly distributed in the two precession cones. This loss of phase coherence is called T2 relaxation. For electron spins, T2 is in the range of nano- to microseconds. At the same time, due to energy exchange processes with the environment, the spins return again into the Boltzmann equilibrium distribution between the two energy states. This relaxation is called T1 relaxation and for electron spins it lies in the range of micro- to milliseconds. Spin Echoes For technical reasons, and for most samples, no free induction decay (FID, Section 18.1.2) can be observed after the 90° pulse in EPR spectroscopy. Instead, as a
18 Magnetic Resonance Spectroscopy of Biomolecules
475
workaround, a so called “spin echo” is detected. Below, a simple two-pulse sequence is described that generates such a spin echo (Figure 18.46). At a time τ after the 90° pulse, the spin packets have traveled different distances from the y-axis due to their differing rotational velocities (Figure 18.45c). The different rotation speeds are the result of the slightly different magnetic environments and thus Larmor frequencies of the spins. If the sample is now irradiated with a 180° pulse, the spins (Figure 18.45d) are rotated by 180° around the x-axis. Since they maintain their direction and speed of rotation, the spins start to re-phase (Figure 18.45e) and form a magnetization or a spin echo along the y-axis after the time 2τ (Figure 18.45f). In honor of its discoverer Erwin Hahn, this echo is called Hahn echo. A different spin echo can be generated with a three-pulse sequence (Figure 18.47). With a 90° pulse, the magnetization is rotated to the x/y plane, then after a short time τ a second 90° pulse follows. By this second 90° pulse, the x/y magnetization is stored along the z-axis. With a third 90° pulse after a time interval T, the magnetization is detected as a so-called stimulated echo. ESEEM – Electron Spin Echo Envelope Modulation If the time interval τ of the pulse sequence shown in Figure 18.46 is gradually increased, and the amplitude of the Hahn-echo is detected at each step, one can observe an oscillating amplitude of the echo, which is known as electron spin echo envelope modulation (ESEEM). The cause of this modulation is transitions between the nuclear spin levels. A plot of the echo amplitude versus τ yields a time trace from which the corresponding oscillation frequencies can be obtained by Fourier transformation. From these frequencies, weak couplings of nuclei, which are too small to be observed by cwEPR, can be determined and thus more detailed information about the structure of the environment of the paramagnetic center can be obtained. A disadvantage of this pulse sequence is that due to the rapid T2 relaxation the line width is very large, resulting in overlapping lines. In addition, higher harmonic frequencies occur, and the observed redundant sum and difference frequencies make it difficult to interpret the spectra. These problems can be avoided by using the stimulated echo based three-pulse ESEEM. In this experiment, the time interval T in Figure 18.47 is gradually increased and the echo amplitude of the stimulated echo is recorded as a function of T. Because the amplitude of the stimulated echo decays with the slower relaxation time T1, a longer time window for monitoring the oscillation is available and thus the obtained frequency lines are narrower than with the twopulse ESEEM. An example of a three-pulse-ESEEM is an experiment on the electron transfer protein bo3 ubiquinol oxidase from Escherichia coli. During the electron transfer reaction, the ubiquinone is converted into an ubisemiquinone anion radical (UB• ), which enabled an EPR investigation into how the UB• is structurally bound. In Figure 18.48 the corresponding three-pulse ESEEM is shown in both the time and the frequency domains. In the Fourier transformed spectrum, there are four lines at 0.95, 2.32, 3.27, and 5.2 MHz, which were assigned to a 14 N-nucleus near the UB• . The occurrence of these four lines can be explained as follows: Since 14 N has a nuclear spin of I = 1, each of the two electron spin states ms = ±½ split into three levels due to the 14 N hyperfine coupling (Figure 18.49). If the nuclear Zeeman interaction (interaction between the magnetic moment μI of the nuclear spin and B0) is half as large as the hyperfine coupling, then both interactions cancel out in one of the two mS levels. The energy separation in this mS level is then determined by the quadrupole interaction that occurs for nuclei with I > ½. This quadrupole interaction is defined by the quadrupole coupling constant Q and the asymmetry parameter η. Both parameters are very sensitive to the distribution of charge in a molecule and the molecular structure. Between the three nuclear spin levels in each of the two ms-levels there are three nuclear spin transitions. The three nuclear spin transitions ν0, ν+, and ν are particularly intense in three-pulse ESEEM. The double-quantum transition νdq from the other mS-level, however, is of low intensity and broad and the two single quantum transitions νsq1 and νsq2 are often not detected. From the frequencies belonging to the four lines, the isotropic 14 N hyperfine coupling constant, the 14 N quadrupole coupling constant, and the asymmetry parameter η could be calculated. With these data and from the two-dimensional spectrum (Figure 18.51b below) and
Figure 18.45 (a) Magnetization M along B0; x, y, and z are Cartesian coordinates and B0 is oriented along z; (b) M after the 90° pulse; (c) dephasing during the time interval τ; (d) inversion of the spins by the 180° pulse; (e) refocusing of the spins in the time interval τ following the 180° pulse; (f) Hahn-echo at time 2τ after the 90° pulse.
476
Part II: 3D Structure Determination
Figure 18.46 Two-pulse sequence to generate a Hahn echo (HE).
Figure 18.47 Three-pulse sequence for generating a stimulated echo (SE).
Figure 18.48 Three-pulse ESEEM spectrum and structural formula of UB• in the QH-binding pocket of the ubiquinol oxidase. (a) Time domain spectrum and (b) frequency domain spectrum after Fourier transform of (a). Source: reproduced with permission from Grimaldi, S. et al. (2001) Biochemistry, 40, 1037–1043. With permission, Copyright 2003, American Chemical Society.
Figure 18.49 Splitting scheme for a coupled spin system with s = ½ and I = 1, for the case that the hyperfine interaction Aiso and the nuclear Zeeman interaction νN cancel each other out (Aiso = 2νN).
isotope labeling it was concluded that the UB• is bound only with the C1 carbonyl group via a strong hydrogen bond to a nitrogen atom in the amino acid backbone while the other carbonyl group is not bound to the protein. This finding of an asymmetric binding of UB• helped in understanding the directional electron transfer in this protein.
Figure 18.50 HYSCORE pulse sequence.
HYSCORE – Hyperfine Sublevel Correlation Experiment The hyperfine sublevel correlation experiment (HYSCORE) is a two-dimensional cross-correlation experiment, which can be used to identify nuclear spin transitions (lines) belonging to the same nucleus. The experiment is carried out such that, between the last two 90° pulses of a stimulated echo sequence, a 180° mixing pulse is introduced (Figure 18.50). Then, both the time interval t1 and the time interval t2 are varied. After a two-dimensional Fourier transform a two-dimensional frequency domain spectrum is obtained. In this spectrum, cross correlations occur between those peaks that belong to nuclear spin transitions in different ms levels but from the same nucleus. In the case of ubiquinol oxidase, HYSCORE was used to determine whether the four lines in the three-pulse ESEEM (Figure 18.48) originate from the nitrogen nucleus. The
18 Magnetic Resonance Spectroscopy of Biomolecules
477
Figure 18.51 (a) Theoretical HYSCORE spectrum. (b) HYSCORE spectrum of UB• in the QH binding pocket of b03 ubiquinol oxidase from Escherichia coli. For a better overview only one of the four quadrants is shown. Source: reproduced with permission from Grimaldi, S. et al. (2001) Biochemistry, 40, 1037–1043; 2003, American Chemical Society.
corresponding HYSCORE spectrum is shown in Figure 18.51b. One can clearly see the crosscorrelations between the four lines indicating that indeed they can be assigned to a single nitrogen nucleus. ENDOR – Electron Nuclear Double Resonance Electron nuclear double resonance (ENDOR) experiments can be used to observe weak and strong hyperfine couplings of near and distant centers. There are several ENDOR variants that can be carried out both as cw- or pulse-experiments. In a Davies ENDOR experiment (Figure 18.52a), the magnetization is inverted with a 180° pulse (from +z to z), and after a mixing time T, the magnetization is detected again with a 90°τ-180°-τ-echo sequence. During the time T, a 180° radio frequency pulse is applied (180° pulse for the nuclear spins) and the echo amplitude is measured as a function of the radio frequency. If nuclear spin transitions are induced by the radio frequency, then lines occur at the respective radio frequencies in the ENDOR spectrum. Another pulsed ENDOR experiment, Mims ENDOR (Figure 18.52b), is based on the stimulated echo sequence, wherein the radio frequency pulse is irradiated after the second 90° pulse. This pulse sequence has a higher sensitivity for small hyperfine couplings. Both pulse sequences are static with respect to the time axis, that is, no time interval is changed during both experiments. Only the frequency of the radio wave is changed while monitoring the amplitude of the detected echo. Figure 18.53b shows a Mims ENDOR spectrum of a 31 P nucleus in a phospholipid. Amongst others, ENDOR experiments are interesting for three reasons: 1. The resolution is very good, which is why even small hyperfine couplings and hyperfine anisotropies can be resolved. 2. The number of peaks is reduced compared to the cw-EPR spectra. If N magnetically inequivalent 1 H nuclei are contributing to a cw-EPR experiment, then the number of lines is 2N, while in the corresponding ENDOR experiments only 2N peaks will be present. In the case of N magnetically equivalent 1 H nuclei a cw-EPR spectrum will show N + 1 lines, whereas an ENDOR spectrum shows only two lines.
Figure 18.52 ENDOR-pulse sequence: (a) Davies sequence and (b) Mims sequence. RF stands for radio frequency and MW for microwave frequency.
Figure 18.53 (a) Structural formula of the phospholipid and (b) Mims ENDOR spectrum of the 31 P nucleus and simulation of the spectrum (below); ν(31 P) is the free Larmor frequency of the 31 P-nucleus (6 MHz). From the splitting of 33 kHz and the line width, the distance between the 31 P-nucleus and the electron spin was determined to be 1 nm. This corresponds to an extended conformation of the lipid. The small signal-to-noise ratio indicates that the distance of 1 nm is at the upper distance limit of the method. Source: adapted from Zänker, P.P, Jeschke, G., and Goldfarb, D. (2005) J. Chem. Phys., 122, 024515, 1-11. With permission, 2005 American Institute of Physics.
478
Part II: 3D Structure Determination
3. In ESEEM experiments, the echo modulation depth tends to go to zero at high microwave frequencies/magnetic fields. In contrast, in ENDOR the nuclei are even better separated at high microwave frequencies/magnetic fields (based on their nuclear Larmor frequencies).
Figure 18.54 Four-pulse PELDOR sequence; νA and νB denote the two different microwave frequencies.
PELDOR – Pulsed Electron Double Resonance With the pulsed electron double resonance (PELDOR) sequence (Figure 18.54), the dipolar coupling between two unpaired electrons A and B can be selectively detected. Here, the detection sequence 90°-τ-180°-t180° is applied to electron A using a microwave frequency νA. This leads to a refocused spin echo, which can be detected after the time interval t-τ after the last pulse. Within the time interval T, a pump pulse with the microwave frequency νB is applied, which inverts the spin of electron B in the same molecule. When the two unpaired electrons are coupled, the inversion of the electron B results in a change of the magnetic field at the electron A and thus a change in the echo amplitude. Moving the pump pulse within the time interval T leads to an oscillation of the observed echo amplitude. The frequency of this oscillation is the frequency of the electron–electron coupling νAB = J + D; J can usually be neglected for distances rAB > 1.5 nm, which in turn makes νAB depend only on D. This dipolar frequency depends only on the distance between the two electrons as well as the orientation-term (1 3cos2θ) (Section 18.2.6). Since the measurement is performed in frozen solution and usually without orienting the sample, all orientations of θ are detected. This yields in the frequency domain the so-called dipolar Pake pattern from which the frequency for θ (90°) can be read off and with which the distance rAB can be directly calculated. With this pulse sequence, distances of up to 160 Å between two spin centers can be measured. An example is shown in Figure 18.55, where the time domain spectrum for a DNA molecule labeled with two nitroxides is presented. The figure also shows the frequency domain spectrum obtained by Fourier transformation (termed Pake spectrum or Pake pattern). The intense line at 6.8 MHz corresponds to the orientation θ = 90°. From this value a distance rAB of 19.5 Å can be
Figure 18.55 (a) Reaction scheme for the covalent attachment of nitroxides to DNA or RNA; (b) PELDOR – time domain spectra for a series of five DNA duplexes in which the distance between the two nitroxides increases from 1 to 5; (c) Fourier transform PELDOR spectrum (Pake spectrum) of DNA 1; (d) correlation of PELDOR distances and molecular dynamics simulations for a number of DNA and RNA duplexes. Source: Schiemann, O. et al. (2004) J. Am. Chem. Soc., 126, 5722–5731. With permission, Copyright © 2004, American Chemical Society.
18 Magnetic Resonance Spectroscopy of Biomolecules
479
calculated (Section 18.2.6), which matches very well with the theoretically expected distance of 19.3 Å. Comparison between PELDOR and FRET Another spectroscopic method, which can be used to measure distances in the nanometer range, is FRET (fluorescence resonance energy transfer). A comparison between FRET and PELDOR shows that the methods are complementary.
FRET, Section 16.7
FRET provides distances of biomolecules in liquid solution. PELDOR, on the other hand, is performed in frozen solution.
For FRET measurements a single molecule is sufficient, while concentrations in the micromolar range are required for PELDOR.
FRET can observe distance changes in a time-resolved manner. PELDOR observes frozen distance distributions.
In a PELDOR experiment, the coupling mechanism between both spin centers can be
resolved and the size of J can be determined. In FRET experiments, the mechanism of fluorescence quenching is not always clear; often, reference measurements are needed. Calculation of the distance from the Pake spectrum is parameter-free. For the analysis of FRET measurements, assumptions about the orientation parameter κ must be made. Different labels are used for FRET and PELDOR. Large chromophores are frequently used in FRET. These are attached to the biomolecules via very flexible linkers. EPR labels are small and can be attached close to the surface, sometimes by rigid linkers (e.g., for DNA or RNA). Thus, it is easier to correlate the measured distances to the structure of the biomolecule. On the other hand, rigid linkers have a higher propensity to change the structure of the molecule under investigation. In reality, for both cases care should be taken to avoid the induction of structural changes.
18.2.8 Further Examples of EPR Applications In the previous sections several examples were presented for the determination of structural elements by EPR methods. This section describes three examples for the determination of mobilities, pH values, and binding constants. Quantification of Spin Sites and Binding Constants The intensity of the EPR signal depends, inter alia, on the number of electron spins in a sample. However, since many other factors influence the signal intensity, the spin number cannot be directly determined from the signal intensity in a straightforward manner. The number of spins can only be determined by comparison with a reference sample. For this, the unknown sample and the reference sample must be measured under exactly identical conditions. For technical reasons it is difficult to do so and thus the error for such experiments is normally about 15%. Nevertheless, in this way the number of spins per biomolecule can be determined if the concentration of the biomolecule is known. In addition, if the EPR spectra of a protein-bound and protein-free paramagnetic center are different, the number of binding sites, and the dissociation constants of the binding sites, can be determined. An example of this approach is described below, namely, the binding of Mn(II) to a catalytically active RNA, the minimal hammerhead ribozyme (Figure 18.56).
Figure 18.56 (a) Secondary structure of the minimal hammerhead ribozyme. The arrow indicates the cleavage site in the phosphodiester backbone. (b) Binding isotherm obtained from the EPR titration. The open circles are the experimental data while the solid line is the fit using the formula given in the graph. Source: Kisseleva, N. et al. (2005) RNA, 11, 1–6. With permission, Copyright 2005 by RNA Society.
480
Part II: 3D Structure Determination
Figure 18.57 (a) Structure of the protonated/unprotonated nitroxide. In the protonated nitroxide, the mesomeric form I is energetically unfavorable due to the repulsion between the two positive charges. Thus, form II predominates, where the electron spin is located on the oxygen. (b) L-Band (1.3 GHz/40 mT) cw-EPR spectrum of the nitroxide at three different pH values. The difference in 14 N hyperfine coupling can be clearly seen. The pKa value of the nitroxide is 4.6. Therefore, at pH = pKa = 4.6 both forms are present in a 1 : 1 ratio. Source: Sotgiu, A. et al. (1998) Phys. Med. Biol., 43, 1921–1930. With permission, copyright 1998 IOP.
When small quantities of Mn(II) are titrated to a buffer solution containing the minimum hammerhead ribozyme, a much smaller cw-EPR signal is obtained than if the same amount of Mn(II) is titrated into a buffer solution without ribozyme. The reason for this is that Mn(II) bound to the ribozyme yields very broad EPR lines, such that the EPR spectrum of the Mn(II)/ ribozyme complex cannot be observed at room temperature. Thus, the concentration of bound Mn(II) can be calculated from the signal intensity originating from Mn(II) free in solution. By plotting the ratio of bound Mn(II) against the amount of free Mn(II), a binding isotherm is obtained. In this way, the dissociation constant for ribozyme-bound Mn(II) was determined to be 4 μM. Local pH Values Nitroxides, which contain an amino group in or on the ring system, are often used as pH probes. The principle of such pH sensors is based on the protonation of the amino group in an acidic solution, whereby a positive charge is produced close to the nitroxide. The positive charge causes a shift in the spin density from the nitrogen to the oxygen atom, and thus a reduction of the 14 N hyperfine coupling and the g-tensor (Figure 18.57a). Depending on the pH, the cw-EPR spectrum is then a superposition of the spectra of the protonated and deprotonated nitroxide. From the intensity ratio of the two spectra, the concentration ratio of the two forms and thus the pH can be obtained. If such a nitroxide is bound to a biomolecule, the pH can be measured in the local environment of the biomolecule. A similar dependency of A and g on the polarity can be used to distinguish the membrane interior from the membrane surface. Mobility Nitroxide spin labels are also frequently used to obtain information on the mobility of biomolecules. This method makes use of the fact that the spectrum of the nitroxide changes depending on its rotational freedom. If the nitroxide is free in its rotational movement, an isotropic three-line spectrum is obtained at X-band (Figure 18.58, upper spectrum). If the rotation is completely frozen, one observes an anisotropic spectrum as shown in Figure 18.58 (bottom) (also compare with nitroxide at 180 GHz, Figure 18.42). Depending on the degree of rotational freedom of the nitroxide, the spectrum changes gradually from isotropic to anisotropic. A measure of the rotational freedom is the rotational correlation time τrot, which can be easily calculated from the EPR line shapes and intensities: τrot 6:5 10 Figure 18.58 Influence of the rotational correlation time on the cw-X-band EPR spectrum of the nitroxide Tempol. In the case of free rotation, the three hyperfine lines of the nitroxide are split by Aiso = 1.7 mT. If the rotation is completely frozen, the nitroxide is dominated by the anisotropic hyperfine coupling constant Az of 3.7 mT. Source: Weber, S., Wolff, T., and von Bünau, G. (1996) J. Colloid Interface Sci., 184, 163–169. With permission, Copyright 1996 Academic Press.
10
sffiffiffiffiffi h0 ΔB h1
! 1
(18.36)
Here, h0 is the intensity of the central line, h1 is the intensity of the low-field line, and ΔB is the width of the central line in Tesla. The Tesla is the unit for the external magnetic field B (actually the magnetic induction); 1 T = 104 Gauss. If a nitroxide is covalently bound to a biomolecule, the freedom of rotation of the former is limited by the freedom of rotation of the latter. The measured rotational correlation time of the nitroxide is thus a measure of the mobility of the biomolecule. However, it is difficult to separate the τrot value of the biomolecule from the measured τrot value, which has a residual contribution from the mobility of the label independent of the host biomolecule. It is simpler to determine
18 Magnetic Resonance Spectroscopy of Biomolecules
481
Figure 18.59 Binding of the TAT protein to the nitroxide-labeled HIV TAR RNA. (a) Reaction scheme for the labeling of RNA with a nitroxide (top) and the secondary structure of the TAR RNA with the nitroxide located at the blue uridine (below). (b) The cw-X-band EPR spectra of TAR RNA spin-labeled at U23 without (black) and with (blue) the bound TAT protein. The broadening of the spectrum and the decrease in the intensity of the low-field line for the TAR-TAT-complex can be clearly seen. By the analysis of multiple spectra, in which the nitroxide is bound to different positions on the RNA, statements about the influence of the dynamics of the RNA on the binding of the protein could be made. Source: Edwards, et al. (2002) Chem. Biol., 9, 699–706. With permission, Copyright 2002 Cell Press. Published by Elsevier Ltd.
relative differences. The binding of ligands to RNA can be followed in this way by EPR spectroscopy (Figure 18.59).
18.2.9 General Remarks on the Significance of EPR Spectra In most cases it is difficult to derive a clear statement or structure from a single EPR spectrum. To obtain reliable results, one can proceed as follows:
vary the sample conditions (e.g., temperature, solvent, etc.); modify the biomolecule biochemically (e.g., protein mutants, spin label positions, isotope labeling);
record cw-EPR spectra at different microwave frequencies to resolve g-values and spectra of
various radicals; this also allows the hyperfine coupling contribution to be separated from gtensor contributions; combine several pulse-EPR/ENDOR-methods to select and assign individual spectral hyperfine contributions; simulate the EPR spectra, to obtain the EPR parameters; translate the EPR parameters into structural data by comparison with results from quantum chemical methods (such as DFT), model systems, and literature data; combine with further spectroscopic methods.
18.2.10 Comparison EPR/NMR As a final point, the two complementary magnetic resonance methods EPR and NMR are compared.
NMR spectroscopy investigates the spin of magnetic nuclei in predominantly diamagnetic
samples and because protons, nitrogen, or carbon nuclei occur everywhere in the biomolecule the method can elucidate the overall structure of the investigated biomolecule on an atomic level. EPR spectroscopy on the other hand detects the spin of unpaired electrons (paramagnetic samples) and observes only the local environment of the spin center. This local restriction of EPR also means that there is no size restriction for the biomolecule to be studied, while NMR is currently limited to biomolecules with a mass of roughly 80 KDa. The sensitivity of EPR spectroscopy is higher (nanomol) than that of NMR spectroscopy (millimol). This is due to the larger magnetic moment of the electron (μe/μH = –1838), which makes the Boltzmann population difference larger for EPR (1.1 × 10 3 compared to 1.1 × 10 6 at 3.4 T and T = 300 K). Due to the larger magnetic moment, the relaxation processes are faster in EPR. This is why the time scale for pulsed EPR experiments is nano- to microseconds as opposed to
482
Part II: 3D Structure Determination
milliseconds to seconds for NMR. This and the need for the use of microwaves also mean that the technical requirements are higher for EPR. The faster relaxation leads to broader lines in EPR (MHz versus Hz). The electron–nucleus spin coupling in EPR is larger than nuclear–nuclear spin coupling in NMR due to the greater magnetic moment of the electrons (MHz versus Hz). For this reason, larger distances between the unpaired electron and a nucleus (to 10 Å) or the electron and another unpaired electron (up to 80 Å) can be determined by EPR. The theoretical effort needed for the simulation of EPR spectra is significantly larger than for high-resolution liquid state NMR spectra due to the anisotropic contributions and the very fast relaxation. The translation of EPR and also NMR parameters into structural conclusions using quantum chemical calculation methods (such as DFT) is still complicated. This is particularly significant for transition metals for which often only trends are obtained. Both NMR and EPR can be carried out in either liquid buffer solutions or in frozen solutions/ powders. This is to say that no single crystals are needed for either method. Both spectroscopic methods can provide insight into the dynamics of biomolecules.
Acknowledgements We thank Yaser NejatyJahromy for careful reading of the manuscript.
Further Reading Section 18.1 Bax, A. (2003) Weak alignment offers new NMR opportunities to study protein structure and dynamics. Protein Sci., 12, 1–16. Blumenthal, L.M. (1970) Theory and Application of Distance Geometry, Chelsea, Bronx, New York. Cavanagh, J., Fairbrother, W.J., Palmer, A.G. III and Skelton, N.J. (1996) Protein NMR Spectroscopy, Academic Press. Creighton, T.E. (ed.) (1992) Protein Folding, Freeman, New York. Croasmun, W.R. and Carlson, R.M. (eds) (1994) Two-Dimensional NMR Spectroscopy, VCH-Verlagsgesellschaft, Weinheim. Derome, A.E. (1987) Modern NMR Techniques for Chemistry Research, Pergamon, Oxford. Dingley, A.J., Cordier, F., and Grzesiek, S. (2001) An introduction to hydrogen bond scalar couplings. Concepts Magn. Reson., 13, 103–127. Ernst, R.R. (1992) Kernresonanz-Fourier-transformationsspektroskopie (Nobel-Vortrag). Angew. Chem., 104, 817–952. Ernst, R.R., Bodenhausen, G., and Wokaun, A. (1987) Principles of Nuclear Magnetic Resonance in One and Two Dimensions, Clarendon Press, Oxford. Evans, J.N.S. (1995) Biomolecular NMR Spectroscopy, Oxford University Press. Fernandez, C. and Wider, G. (2003) TROSY in NMR studies of the structure and function of large biological macromolecules. Curr. Opin. Struct. Biol., 13, 570–580. Friebolin, H. (1998) Ein- und Zweidimensionale NMR-Spektroskopie, VCH-Verlagsgesellschaft, Weinheim. Goldman, M. (1988) Quantum Description of High-Resolution NMR in Liquids, Clarendon Press. Karplus, M. and Petsko, G.A. (1990) Molecular dynamics simulations in biology. Nature, 347, 631–639. Pain, R.H. (ed.) (1994) Mechanisms of Protein Folding, Oxford University Press, Oxford. van de Ven, F.J.M. (1995) Multidimensional NMR in Liquids, VCH-Verlagsgesellschaft, Weinheim. Van Gunsteren, W.F. and Berendsen, H.J.C. (1990) Moleküldynamik-computersimulationen: methodik, anwendungen und perspektiven in der chemie. Angew. Chem., 102, 1020. Wüthrich, K. (1986) NMR of Proteins and Nucleic Acids, John Wiley & Sons, Inc., New York. Weltner, W. (1983) Magnetic Atoms and Molecules, Dover Publications, New York.
Section 18.2 Atherton, N.M. (1993) Principles of Electron Spin Resonance, Ellis Horwood, New York. Berliner, L.J. (ed.) (1998) Spin Labeling: The Next Millennium, Biological Magnetic Resonance, vol. 14, Kluwer Publishing, Amsterdam.
18 Magnetic Resonance Spectroscopy of Biomolecules Dikanov, S.A. and Tsvetkov, Y.D. (1992) Electron Spin Echo Envelope Modulation (ESEEM) Spectroscopy, CRC Press, Boca Raton, FL. Eaton, G.R., Eaton, S.S., and Berliner, L.J. (eds) (2000) Distance Measurements, Biological Magnetic Resonance, vol. 19, Kluwer Publishing, Amsterdam. Kaupp, M., Bühl, M., and Malkin, V.G. (eds) (2004) Calculation of NMR and EPR Parameters, WileyVCH Verlag GmbH, Weinheim. Poole, C.P. (1983) Electron Spin Resonance – A Comprehensive Treatise on Experimental Techniques. Wiley-Interscience, New York. Schweiger, A. and Jeschke, G. (2001) Principles of Pulse Electron Paramagnetic Resonance, Oxford University Press, Oxford. Misra, S.K. (ed.) (2011) Multifrequency Electron Paramagnetic Resonance, Wiley-VCH Verlag GmbH, Weinheim. Weil, J.A., Bolton, J.R., and Wertz, J.E. (1994) Electron Paramagnetic Resonance: Elementary Theory and Practical Applications, Wiley-Interscience, New York.
483
Electron Microscopy Harald Engelhardt Max-Planck-Institut für Biochemie, Am Klopferspitz 18, 82152 Martinsried, Germany
Modern microscopic techniques produce images of small organisms, tissues, single cells, organelles, membranes, macromolecular assemblies, isolated macromolecules, and of small molecules down to atoms (Figure 19.1). The beginning of microscopy dates back to the seventeenth century when Antoni van Leeuwenhoek (1632–1723) in the Netherlands and Robert Hooke (1635–1703) in England built their first simple instruments and initiated the development of light microscopy. The scientific investigation of optical systems and of its resolution limit by Ernst Abbe (1840–1905) in Germany, the improvement of the microscopic illumination by August Köhler (1866–1948), the development of better glass materials, and Robert Koch’s famous microbiological investigations led to a blooming of microscopy in science. In the 1930s, the Dutch physicist Frits Zernike (1888–1966) devised the phase contrast microscope, which allowed
19
Light Microscopy, Chapters 7, 8
Figure 19.1 Biological structures and suitable microscopy techniques. The resolution limits of the human eye, light microscopes, and the transmission electron microscope are indicated. Fluorescence microscopes show the position of fluorophores where the optical near field microscopies (e.g., SNOM), the stimulated emission depletion (STED), photoactivated localization microscopy (PALM), and the stochastic optical reconstruction microscopy (STORM) overcome the classical resolution limit of light microscopes. Scanning electron and scanning probe microscopies provide surface structure information, light and transmission electron microscopy allow three-dimensional imaging.
Bioanalytics: Analytical Methods and Concepts in Biochemistry and Molecular Biology, First Edition. Edited by Friedrich Lottspeich and Joachim Engels. 2018 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2018 by Wiley-VCH Verlag GmbH & Co. KGaA.
486
Part II: 3D Structure Determination
Atomic Force Microscopy, Chapter 20
X-Ray Strucrure Analysis, Chapter 21 NMR Spectroscopy, Chapter 18
researchers to investigate unstained cells and tissue of low contrast and whose basic principle also applies to electron microscopy. Köhler also introduced the first fluorescence microscope (1908), an instrument that gained more significance in cytological research after the development of stronger light sources. Based on the increasing success of molecular genetics and the fusion of the green fluorescent protein to other macromolecules it became possible to identify and localize cellular protein complexes, and this success enormously stimulated the field of cellular biology in the second half of the 1990s. At this time, fluorescence microscopy experienced another revolutionary breakthrough by overcoming the classical resolution limit that the refraction law and the wavelength of light dictate. The new techniques allow the detection of single fluorescence markers in cells with a spatial accuracy of about 30 nm or better. Several instrumental solutions were introduced, such as STED (stimulated emission depletion microscopy), invented by Stefan Hell (∗1962), or PALM (photoactivated localization microscopy), and STORM (stochastic optical reconstruction microscopy). The latter stimulate fluorescence labels locally and calculate the center of the fluorescence signal afterwards. None of these fluorescence microscopes provides genuine structural information, yet, they locate labeled molecules and display their distribution with unprecedented spatial precision (super-resolution). The scanning probe microscopies (SPM) in the optical near-field (scanning near-field optical microscopy, SNOM; scanning near-field infrared microscopy, SNIM) are also techniques that detect objects of “submicroscopic” dimensions and can even record spectroscopic information in the nanometer range independently of the wavelength of the light. The field of “nano-optics” has important applications in material science. The “apertureless” microscopes make use of scanning force microscopy (SFM; atomic force microscopy, AFM) that was introduced by Gerd Binnig (∗1947) in 1986, only a few years after he and Heinrich Rohrer (1933–2013) had invented the scanning tunneling microscope (STM), a novel type of microscopy for imaging surfaces. Several variants of SFM are also useful for biological structure research and are complementary to applications in light and electron microscopy. When Louis de Broglie (1892–1987) realized that electrons can be understood as waves with wavelengths far below one nanometer, he determined the theoretical basis for a microscope operating with electrons. Max Knoll (1897–1969), Ernst Ruska (1906–1988), and Bodo von Borries (1905–1956) developed the first transmission electron microscope (TEM) in 1931, and only two years later they obtained images with much higher resolution than light microscopy. Today it is possible to resolve single atoms in radiation-resistant objects such as alloys. Manfred von Ardenne (1907–1997) built the first scanning electron microscope (SEM) in 1937, an instrument that produces impressive surface images of high depth of field. Biological objects such as proteins that are very radiation-sensitive require preparative and technical efforts before imaging with quasi-atomic resolution (0.3 nm). The two-dimensional (2D) crystalline bacteriorhodopsin from halobacteria and the photosynthetic light-harvesting antenna complex LHII of the photosynthetic membranes from chloroplasts were the first biological specimens whose three-dimensional (3D) atomic structures were solved by means of electron microscopy (electron crystallography). In addition to X-ray crystallography and NMR spectroscopy, electron microscopy is the third method with which we can determine the spatial structure of macromolecules. Electrons interact particularly strongly with the object and, in contrast to the other methods, also provide images of single molecules. Automated data acquisition and novel electron detectors enable us now to collect data of single protein complexes of sufficient quantity and quality to achieve quasiatomic resolution. Single particle electron microscopy is indeed the only way to resolve the 3D structure of macromolecular complexes that are too big for NMR spectroscopy and too flexible for X-ray crystallography and do not form crystals. Wolfgang Baumeister (∗1946) and his team developed an approach that made the 3D structure of native, shock-frozen (vitrified) cells accessible to TEM at macromolecular resolution. The 3D reconstruction of the cytoskeleton in intact cells from the slime mold Dictyostelium discoideum signaled the breakthrough of cryo-electron tomography (CET) in 2002. Meanwhile, CET has granted unprecedented insights into the macromolecular organization of microorganisms and eukaryotic cells and has opened up new and exciting perspectives in structure research in situ. In 2014 Baumeister’s laboratory introduced a new tool (phase plate) for phase contrast enhancement in electron microscopy that has the potential to become standard equipment. Cryo-electron tomography now integrates
19 Electron Microscopy
487
molecular and cellular structure research, which before were separate biological disciplines.
19.1 Transmission Electron Microscopy – Instrumentation TEM received its name from the imaging mode. The electrons that transmit the object are used for imaging, similar to the function of light in optical microscopy. The basic construction of electron and optical microscopes are indeed similar, except that electron magnetic lenses are used instead of glass lenses and the electron microscopes are much bigger (Figure 19.2). The lenses consist of iron-sheathed coils that create a magnetic field upon current flow and direct the field towards the inner space. Electron magnetic lenses are always converging lenses. Correcting spherical and chromatic aberrations is thus challenging, yet, in recent years, correction systems have improved the image formation in high-resolution electron microscopes in particular. The electron source is a cathode that emits electrons. An anode with a potential difference to the cathode of 105 V or more accelerates these electrons. The condenser lens (often two lenses after each other) focuses the electron beam in the object plane where it transmits the object and is expanded by the object lens. Its focus length can be adjusted by variation of the lens current so that changing the magnification does not require other objective lenses. One or more projection lenses between the object and the imaging plane further expand the beam and lead to an increase in magnification. The image is rendered visible on a fluorescence screen (which may be inspected by a binocular loupe with tenfold magnification) and recorded on film plates in the conventional way or, nowadays, by means of special digital cameras. The microscopes allow for magnification between 100-fold and about 106-fold. However, for macromolecules and biological structures a primary magnification of 100 000-fold is sufficient. The electron microscope column must be evacuated so that electrons only interact with object atoms but not with gas molecules. Microscopy in a vacuum necessitates preparation of the biological objects (fixation, dehydration, embedding in plastic material or freezing for
Figure 19.2 Beam path in the light microscope (LM) and in the transmission electron microscope (TEM). The principal system of lenses is the same in both microscopes, the TEM usually contains a second condenser and projection lenses in addition. Condenser and objective apertures limit the illumination and blank out strongly diffracted regions of the beam, thereby enhancing the contrast. The image is observed in the image plane by eye (LM) or on a fluorescence screen (TEM) and can be recorded by a camera. An external energy filter separates electrons of different energy and enables electron energy loss spectroscopy (EELS) analysis and electron spectroscopic imaging (ESI) (see also Figures 19.9 and 19.12). X-Rays that are created by interaction of beam electrons with object electrons are either shielded or can be recorded by a detector.
488
Part II: 3D Structure Determination
cryo-electron microscopy). If not doing so, the specimens would dehydrate in the microscope and become damaged, and the vacuum would break down. The specimens are placed on copper grids (sometimes made of gold or nickel) with meshes of 30–100 μm (Figure 19.3). The grids are covered by a thin (5–10 nm) carbon film, which supports isolated macromolecules or ultrathin sections. These films are deposited on surfaces of freshly broken mica by heat evaporation of graphite in a vacuum chamber. The film is then floated onto the surface of water and transferred to the grids. For cryo-electron microscopy these films contain holes of 0.5–2 μm in diameter. Comparable grids are commercially available ready for use (Lacey® and Quantifoil® grids; Figure 19.3b). Very recently, we saw the introduction of gold grids with regularly perforated (1 μm) gold membranes instead of carbon films; these are extremely stable in the electron beam and work well for highresolution imaging. The holes are filled by water, which becomes thin ice films after freezing, and they thus contain the biological objects (macromolecules, cellular components, or organelles, cells). Carbon-coated grids are hydrophobic and have to be made hydrophilic prior to the application of material dissolved or suspended in water. Ionized gas molecules – produced in evacuated chambers with glow discharge (plasma cleaner) – render carbon film temporarily wettable. As a side effect, the plasma destroys impurities and cleans the grid surface. The grids are mounted on an object holder and inserted into the microscope (Figure 19.3a). Frozen specimens that are examined in the cold ( 180 °C or less, cryo-electron microscopy, see Sections 19.2.1 and 19.3.3) must be continuously cooled by liquid nitrogen in a controlled manner to avoid recrystallization of the amorphous ice above 150 °C. The object holder can be tilted around one axis or different axes (ideally being set perpendicularly to the first one), and the specimen can be inspected from different projection angles, which is necessary for 3D-reconstruction in electron tomography (Section 19.5).
Figure 19.3 Object holder, grids, and plunger for biological cryosamples. (a) Object holder with a mechanical shield to protect the frozen sample from contaminations. The holder can be rotated around its longitudinal axis by about ±70° to record different projections of an individual object for 3D reconstruction (tilt series). (b) Grids (diameter 3 mm) of various designs for TEM. They are usually covered with a 5–10 nm thick carbon film for small biological specimens. (c) Carbon films with holes for cryosamples. A thin ice film containing the biological structures spans the holes. Bar indicates 20 μm. (d) Plunger for vitrification of biological samples for cryo-electron microscopy. Tweezers hold the grid and inject it into a liquid cryogen (ethane) cooled by liquid nitrogen to about 180 °C. The high cooling rate vitrifies water and prevents the growth of ice crystals. The biological structures are preserved in a close-to-live state.
19.2 Approaches to Preparation Biological specimens must be sufficiently thin so that the strongly interacting electrons can pass through them. Intact cells or tissue preparations are usually chemically fixed, dehydrated, and embedded in special resins (Epon® ) or in material that polymerizes in the cold (e.g., Lowicryl® ). Ultrathin sections (100 nm) are cut from these polymer blocks by means of ultramicrotomes. Fixation, staining, and thin sectioning has been the standard procedure in cell biology for about five decades, and we owe most of our knowledge of cell architecture to this technique. We will not go into further details here since shock-frozen and untreated biological samples are currently replacing chemically fixed and stained specimens. The cryotechniques preserve the native structure and organization of macromolecules in the cellular context and open up new perspectives in cytological research (Figure 19.4). Section 19.2.1 describes the preparation of intact cells in amorphous ice and various thinning approaches for cryo-electron microscopy. Electron microscopy of isolated cell wall fragments, membranes, proteins, and other macromolecular complexes does not require thin sectioning; these objects are already thin enough. In addition to cryopreparation, Sections 19.2.2–19.2.4 describe common contrasting and labeling procedures that are used for a quick inspection of soluble macromolecules or for special applications.
19.2.1 Native Samples in Ice If the inner structure of molecules, of macromolecular assemblies, or of intact cells is to be investigated, the object itself must be imaged and not the distribution of staining material. We therefore avoid any chemical fixation and contrast enhancement and look at the native object in aqueous solution that has been physically “fixed” by rapid freezing. Thin objects are directly applied to the grid, blotted to remove excess water until only a thin film is left, and shock-frozen by plunging the grid into a cryogen (liquid ethane or ethane–propane; Figure 19.3d). The high cooling rate (105 K s 1) prevents water from forming ice crystals that would destroy the biological object. The ice remains amorphous; it is vitrified in a similar way as cold molten glass, as Jacques Dubochet (∗1941) showed in 1981. To control conditions such as temperature
19 Electron Microscopy
489
Figure 19.4 Preparation of biological samples. Cells and tissues are either treated by chemical fixation, embedding, and ultrathin sectioning for electron microscopy or for enhanced structural integrity by high-pressure freezing and freeze substitution. A close-to-live preparation is accomplished by vitrification, cryosectioning, or focused ion beam (FIB) milling. Thinner samples (cells, organelles, or isolated macromolecular complexes) are either thinned in the FIB or directly imaged in the TEM. Procedures for vitrified specimens are shaded in gray. The methods of freeze fracturing and negative staining are indicated but are not complete.
or light intensity, one can use a plunger with an integrated incubation chamber for automated blotting. Isolated protein complexes may be re-suspended in solutions of glucose or similar compounds (trehalose, tannic acid) if they have to be stabilized. The frozen grid is transferred to the microscope sample holder under liquid nitrogen and is imaged at 180 °C or less (Section 19.3.3). Since the contrast of protein (specific density 1.4 g cm 3) in ice (1 g cm 3) is low, it is advantageous to use grids with carbon films containing holes in order not to obliterate the weak contrast with other material (Figure 19.3c). However, cryopreparation of isolated protein complexes is standard and has replaced negative staining specimens for 3D structure determination. Some eukaryotic cells, bacteria, archaea, and viruses as well as macromolecular complexes can be vitrified on grids and in most cases also directly inspected in a microscope (Figure 19.4). However, eukaryotic and multicellular specimens are often too thick to be vitrified by simple plunging. The heat transfer in the center of bigger samples (thickness >10 μm) is too slow to prevent the crystallization of water. However, solutions under pressure crystallize less quickly and the slower cooling rate suffices to vitrify samples 200–300 μm thick. High-pressure freezing instruments generate up to 0.2 GPa (2000 atm) at liquid nitrogen temperature. Moreover, the cell suspensions can be supplemented by anti-freeze compounds such as dextran, which is osmotically inert and reduces the ability of water to form ice crystals. The drawback, however, is that these additives mask polysaccharides of the cell surface. Cryofixed cells or tissue may be freeze-fractured, freeze-etched, and metal-coated (replica technique; Figure 19.4) to investigate the cell surface or fracture faces. Alternatively, they can be fixed at about 100 °C, dehydrated, stained, and embedded in material that polymerizes in the cold. In cases were chemical fixation should be avoided, vitrified cells can be left untreated and are sectioned in the cryo-ultramicrotome. The cell material is frozen in a small copper tube and this is mounted on a cryo-ultramicrotome in an atmosphere of liquid nitrogen ( 150 °C). The tube is trimmed so that the ice block is free
Vitrification Glass (vitrum) is an amorphous material that does not form crystals when it solidifies after melting. Water molecules on the other hand form ice crystals whose particular structure and density depend on the temperature and pressure during the freezing process. Currently, we know of 19 different ice forms. Very fast freezing to temperatures below -140 °C at normal pressure lets water assume an amorphous, vitrified state with a density of 0.94 g cm 3 (low density amorphous ice, LDA). The other types of amorphous ice have a rather high density (1.17 and 1.26 g cm 3) and cannot be generated from liquid water. LDA is more similar to liquid water than any other form of ice.
490
Part II: 3D Structure Determination
Figure 19.5 Ultrathin sections of Mycobacterium smegmatis. Preparation by (a) conventional fixation, dehydration, and embedding in epoxide resin (Epon), (b) high-pressure freezing, freeze substitution, and embedding in Lowicryl, and (c) high-pressure freezing and cryomicrotomy without any chemical treatment, (d) a higher magnification of (c). The lipid bilayer is only visible in cryosections; these are compressed in the cutting direction so that the originally round cross section of the bacterial cell becomes oval. Scale bars indicate 100 nm (a)–(c) and 50 nm (d). Source: parts (a) and (b) courtesy of Christopher Bleck, Basel, Switzerland.
from the surrounding copper and thin sections can be produced. Cryosectioning is not a routine approach, and it consists of up to 30% compression of samples in the sectioning direction. Despite several artifacts, cryosections provide insight into details of the cellular architecture that are usually lost during dehydration and plastic embedding (Figure 19.5). A new development for thinning of frozen biological material is focused ion beam (FIB) micromachining. The vitrified samples are mounted in a scanning electron microscope (SEM) that is also equipped with an ion gun (gallium). The focused ion beam removes material from the surface of the object until it is thin enough for imaging in the TEM (300–500 nm). To save time and gallium, for ion milling the specimens should not be thicker than 5–10 μm. Appropriately thinned specimens are free of artifacts and will not be deformed. An alternative approach ablates thin layers of biological material and images the surface of the sample block by SEM. When this procedure is repeated multiple times, the series of images, aligned and consecutively put together produce a 3D cube of the object. This technique is called FIB-SEM or “slice and view” and can be applied to cells and cell assemblies. Another approach uses an integrated microtome to smooth the surface instead of an ion beam; this is particularly applied to large embedded specimens such as tissues (mouse brain).
19.2.2 Negative Staining Negative staining by heavy metal salts is a very simple and quick method to image isolated proteins, fibrillar assemblies, membranes, and similar objects at a resolution of about 2 nm. Negative staining is readily used to inspect preparations in terms of purity and homogeneity and to get a first impression of the object structure. A droplet of the sample (2–5 μl) is applied to a carbon-coated grid that has been made hydrophilic by glow discharge. After 15–60 s most of the liquid is blotted and the grid is washed with pure water, buffer, or salt solution (10 mM) and stained by means of a heavy metal salt. Common stainings are 2% (w/v) solutions of uranyl acetate, phosphotungstic acid, or ammonium molybdate, amongst others. The compounds differ in contrast, radiation sensitivity, the applicable pH range, and their ionic characteristics. The metal salt covers the surface, fills holes and indentations of the macromolecules, and it is the distribution of the metal that is finally imaged (Figure 19.6). The metal coat is much more radiation-resistant than the biological material and preserves the spatial structure after drying with moderate irradiation. Negatively stained specimens may be stored for weeks to months. However, negative staining is usually not suitable for whole cells and bigger objects.
19 Electron Microscopy
491
Figure 19.6 Schematic illustration of different contrasting methods for imaging macromolecules in the transmission electron microscope. Contrasting with a heavy metal leads to different density distributions in the EM image and provides structural information on individual components of a sample. Only images of ice-embedded native specimens show contrast originating from the object itself.
19.2.3 Metal Coating by Evaporation A common method used to investigate intact cells or tissues in a microscope is by freezing, cutting, or fracturing them in the vacuum, sublimation of ice by freeze etching (at 80 °C), and contrasting the surface by evaporating heavy metal at an angle of 30–60°. The metal coat (1–2 nm Pt/C or other) is stabilized by 10–20 nm carbon (90°). Treatment with aggressive acid solutions removes the biological material, and only the remaining replica is inspected in the microscope. The grain size of the evaporated metal limits the resolution to about 2 nm, but complexes of macromolecules in membranes and cellular surfaces are detectable. Freeze fracturing was a common method in cytology and was used to obtain spatial information on cellular surfaces. The approach lost its importance with the introduction of cryo-electron tomography in structural research (Section 19.5.3). Direct metal evaporation onto membranes or isolated protein assemblies that are adsorbed on carbon-coated grids is also possible. Since air-drying would destroy the unprotected biological specimens, they are freeze-dried to gently remove water and then contrasted in the cold, as in the freeze-etching approach. The metal is only deposited on the surface, pointing towards the evaporation source, and it renders the other regions invisible (Figure 19.6). The orientation of membranes or regular objects (2D crystals or S-layers) can thus be evaluated. Images of unidirectional metal-coated objects must be interpreted in a different manner than negatively stained samples since the density distribution resembles a landscape with hills and valleys that is transformed into a pattern of light and shadow (“metal shadowing”). The gray values of the image correspond to the first derivative of the surface function of the object, so that the original function, the surface relief, can be generated by mathematical integration. However, surface relief reconstructions cannot provide the intrinsic 3D structure of objects – this is the domain of tomographic approaches. Two special metal evaporation approaches, that is, rotational “shadowing” and decoration, are particularly suited to identifying regular structures of macromolecular complexes and fibrillar assemblies (e.g., of actin filaments or flagella). Decoration effects occur if only a limited amount of metal is evaporated so that there is no coherent metal coat. Even if the object
492
Part II: 3D Structure Determination
Figure 19.7 Labeling of proteins with antibodies. (a) Localization of the ATP synthase A1 subcomplex in an ultrathin section (after freeze substitution and embedding in Epon) of Ignicoccus hospitalis. The primary antibodies are labeled by secondary ones carrying gold clusters and having been enlarged by silver coating to 30–60 nm in diameter. The antibodies identify the enzyme in the outer membrane of the archaeon. (b) Immunolabeling of α- and (c) β-subunits of the negatively stained 20S proteasome from Thermoplasma acidophilum. The antibodies are clearly visible and identify the α-subunits in the outer and the β-subunits in the inner rings of the enzyme. Scale bars indicate 0.5 μm (a) and 20 nm (b) and (c). Source: part (a) courtesy of Reinhard Rachel, Regensburg, Germany.
Immunological Techniques, Chapter 5
temperature is below 100 °C the evaporated metal clusters can still diffuse and reach preferred locations on the surface. These sites are now decorated and the regular arrangement is clearly detectable. To obtain pure decoration but no shadowing effects it is advisable to deposit the metal at an angle of 90° onto the surface. Noble metals (Ag, Au, Pt) are particularly appropriate for decoration and often find different molecular sites. Metal coating of isolated macromolecules is more complicated than negative staining and less rewarding than electron microscopy of native and vitrified specimens.
19.2.4 Labeling of Proteins Negative staining and metal coating are non-selective contrasting methods. To identify and localize proteins amongst a wealth of other macromolecules we need specific labels such as monoclonal or polyclonal antibodies or other specific compounds that are coupled to gold clusters to achieve contrast in electron micrographs. Thin sections, and particularly cryosections prepared for immunological purposes (method according to K.T. Tokuyasu), may be labeled with secondary antibodies bearing gold clusters (5–20 nm) that bind to the primary antigen-specific antibody and identify its position with about 20–30 nm accuracy (distance between gold cluster and antigen; Figure 19.7). Subsequent silver coating enhances the size and contrast of small gold clusters. Antibody labeling of isolated macromolecules does not require secondary enhancement by gold since the target protein as well as the antibody – and thus the specific contact regions – are visible with much higher spatial accuracy. Such experiments are particularly suited for identifying the position of subunits in heterooligomeric protein complexes (Figure 19.7).
19.3 Imaging Process in the Electron Microscope The appropriate interpretation of images of a biological object does not only depend on its preparation and contrasting history but also on the imaging process in the electron microscope. It is thus useful to get some insight into the principles of image formation. It is sufficient to discuss basic physical concepts here; the comprehensive theory is described in specific textbooks.
19.3.1 Resolution of a Transmission Electron Microscope Louis de Broglie (1892–1987) introduced the relationship between the wavelength λ and the momentum p of a moving mass. His equation can be transformed for electrons with nearly relativistic speed c (in the EM about 200 000 km s 1): 1 λ hc pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2E 0 E E2 E Ub e0 λ 3:7Ub
0:6
(19.1)
19 Electron Microscopy
493
where Planck’s quantum of action h is 6.63 × 10 34 J s, E0 is the rest energy (8.19 × 10 14 J), and E is the kinetic energy of electrons. The latter depends on the accelerating voltage Ub (eV) in the microscope and the elementary charge e0 (1.60 × 10 19 C) as given in Equation 19.1. The wavelength of electrons above 50 000 eV can be assessed by empirical approximation (given in nm). The wavelength amounts to 0.0037 nm for 100 000 eV electrons and is thus smaller than the wavelength of visible light by a factor of 105. Ernst Abbe attributed the resolution limit d of the light microscope to the physical parameters in Equation 19.2, which is also valid for the EM: d
λ 2nsin α
(19.2)
Here the refractive index n is about 1, and α denotes half of the aperture angle (beam width) of the objective lens. This angle identifies the region of the diffracted beam that the objective lens can capture. Only these electrons transmit information about the object; those that do not interact with the specimen are blind to the object and belong to the reference beam. To obtain images with visible structures of size d (Abbe used a periodic pattern with the characteristic distance d) the lens must at least record a beam of the first diffraction order. Since the diffraction angle Θ is related to the inverse of d (Θ λ/d), the limiting angle α for the objective should be as large as possible (α → π/2). The term A = n sin α, that is, the numeric aperture, is of the order of 1 for objectives in light microscopes, but only 0.01 for electron microscopes. Resolution Hermann von Helmholtz (1821–1894) introduced an alternative analysis of the imaging process in optical instruments and derived another definition of resolution. The beam, originating from a luminescent object, is diffracted by a lens of finite size that creates a diffraction disc (Airy disc) in the focus plane instead of a distinct diffraction spot. Two object points are still distinguishable if the diffraction maximum of one point falls into the first minimum of the second one (criterion of John Rayleigh, 1842–1919). Using this approach for a microscope with aperture A = n sin α and replacing the viewing angle by the corresponding distance d, we obtain the formula d = 0.61(λ/(n sin α)). The resolution limits according to Abbe and von Helmholtz differ slightly by a numerical factor, and we also have to take into consideration that the nominator in Equation 19.2 depends on the illumination of the object, that is, parallel or oblique, and can thus achieve values between A and 2A. The determining variables for resolution are wavelength and the numerical aperture. Using a small aperture that increases the image contrast leads to reduced resolution; the microscopist apparently sees better but recognizes less.
Owing to the small wavelength and despite the drawback of a small aperture, electron microscopes have a physical resolution of 0.2 nm with accelerating voltages of 100–300 keV. High voltage transmission electron microscopes (HVTEMs), equipment that is used in material science, are operated at voltages up to 106 eV. The corresponding wavelength of electrons (0.9 pm) thus allows for the resolution of single atoms.
19.3.2 Interactions of the Electron Beam with the Object Imaging and Information Content of Diffracted Electrons in the TEM We distinguish between two kinds of interactions that electrons experience while passing through the object. The two kinds add to the imaging process in different ways and can be separated in certain microscope models. The electrostatic interaction of a beam electron with an atomic nucleus of the object results in a deflection of the electron path. The deflection is strong if the negatively charged electron comes close to the positively charged nucleus, if the nucleus charge is high (high atomic number elements, staining with heavy metal) and if the velocity of the electron is low (the velocity is a function of the accelerating voltage). The energy of the beam electrons remains constant if the deflection angle is not too high and the electrons experience elastic scattering. The mean scattering angle is about 0.1 rad (6°). Part of the electron beam hits the aperture, is excluded from the imaging process, and creates the scattering contrast (Figure 19.8). The drastic suppression of scattered electrons thus creates high contrast in the image but limits the aperture angle and reduces the resolution of structural details (Equation 19.2).
Figure 19.8 Interaction of beam electrons with the nucleus of object atoms. Strongly scattered electrons are shielded by the aperture and give rise to the scattering contrast. Electrons interacting with the object are decelerated by the object potential and show a phase shift of the wave front with respect to the reference beam. Interference of the scattered electrons with the non-interacting reference beam creates the intrinsic phase contrast in the transmission electron microscope (caused by lens aberrations). The scanning transmission electron microscope (STEM) is equipped with an electron detector instead of the objective aperture and records the strongly scattered electrons for imaging. The signal can be used for electron microscopical mass determination of macromolecular structures.
494
Part II: 3D Structure Determination
Figure 19.9 Interaction of beam electrons with the electron shell of object atoms. The beam electrons transfer energy to object electrons and this energy loss causes an increase of the wavelength. The beam electrons can be separated according to their energy in an energy filter and used for elemental mapping and analysis. The energy transfer leads to an excitation of object electrons and the emission of X-rays that are also indicative for the elemental composition of the sample. Emitted (secondary) electrons from (and beam electrons reflected by) the object surface are employed for imaging in scanning electron microscopy.
Mass Spectrometry, Chapter 15
Since the interaction between beam electrons and the object is relatively strong, it is necessary to limit the thickness – or the mass density – of the specimen. This is generally the case with protein complexes, biological membranes, and ultrathin sections of embedded cells. In cryo-electron microscopy voltages of 200 keV are preferred for the examination of isolated macromolecules since they are usually embedded in a thicker layer of ice. Bigger objects such as intact cells (thickness 0.5 μm) in cryo-electron tomography (Section 19.5.3) can only be imaged at higher voltages (300 keV) without massive scattering and shielding of electrons. Interactions of the beam with the electrons of the object have multiple, partly undesirable consequences for imaging. One effect is that the beam electrons are also scattered by Coulomb forces, but the typical scattering angle of 10 5 rad is much smaller than with electron–nucleus interactions. When accelerated electrons hit the electron shell of atoms they lose energy. The kinetic energy is reduced by ΔE and the wavelength increases correspondingly (Equation 19.1; Figure 19.9); this effect is termed inelastic scattering. The formerly almost coherent electron beam becomes incoherent and shows a spectrum of wavelengths after object transition. Since the diffraction depends on the wavelength (Θ = λ/d) the optical system produces differently sized projections of the object and superposes them in the final image (which is equivalent to chromatic aberration of glass lenses). The structures become blurred and reduced in contrast, which is particularly problematic for objects with a high mass density, such as large frozenhydrated macromolecular complexes, viruses, cell organelles, or intact cells. In general, thick ice layers considerably increase the proportion of inelastically scattered electrons. It is, however, possible to separate electrons of lower energy from the elastically scattered ones by means of an energy filter (Section 19.3.5). Electron Energy Loss Spectroscopy The energy loss of beam electrons (ΔE 2000 eV) correlates with the energy uptake by object atoms and thus contains information on the interacting elements. The spectral analysis of inelastically scattered electrons reveals the elementary composition of the object, and since these electrons can also be used for imaging we can record images of the element distribution (elementary map). Spectra are obtained by electron energy loss spectroscopy (EELS) and images by electron spectroscopic imaging (ESI) or electron spectroscopic diffraction (EDI). For this purpose the microscope must be equipped with a magnetic prism that sorts electrons according to their energy and filters those that are not required for imaging. EELS is usually applied to thin sections of cells or tissues; it is not suitable for single molecules and vitrified specimens. Mass Determination in the Scanning Transmission Electron Microscope The strongly elastically scattered electrons that are shielded by the object aperture also contain information about the biological object since they were deflected according to the number of protons in the nucleus and the amount of corresponding atoms in the object. These electrons can be utilized for imaging if we replace the aperture with an electron detector (Figure 19.8) in the scanning transmission electron microscope (STEM). Provided the elementary composition of the biological specimen is known, which is the case for proteins to a good approximation, we can interpret the signal intensity as a measure for the mass of the object and determine the molecular mass of protein complexes and other biological structures. The sample must not be embedded in any other material and is investigated in a native, freeze-dried state. It is sufficient to record the signal of several hundred (or more) particles for a reasonable statistical analysis with an accuracy of about ±5%. STEM mass determination of isolated proteins is thus not superior to mass spectroscopy. However, the method does not depend on the size and structure of the material and it is possible to determine the mass of large and arbitrarily formed heterooligomeric and multimeric complexes of macromolecules. Typical tasks for STEM mass determination are the stoichiometric analysis of protein complexes, the mass per length of fibrillar structures such as flagella and microtubules, and the mass per area or per structural feature of membranes, 2D crystals, or other 2D assemblies. Mass mapping is only possible with the STEM technique. Scanning Electron Microscopy and Analytical Electron Microscopy Any interaction of electrons between beam and object retains the energy and momentum of the entire system. This means that the energy loss of beam electrons increases the energy of object electrons. They are excited and occasionally even expelled from the electron shell. This process is accompanied by
19 Electron Microscopy
emission of electromagnetic radiation, heat, and electrons from the irradiated material. These secondary electrons are used for imaging in the SEM. The electrons originate from the surface of the object and create images of cells and small organisms with impressive depth of sharpness. The resolution power of the conventional SEM is limited (10 nm) and is usually too low for imaging protein complexes. However, low voltage scanning electron microscopes (LVSEMs) are able to record small structures. Electrons having been removed from the electron shell leave behind a gap that is filled by electrons from a higher energy level. The energy difference is emitted by radiation where electron gaps in the K shell entail particularly energy-rich radiation, that is, X-rays. These contain information on the elementary composition of the material, similar to the EELS spectra of the beam electrons. Energy dispersive X-ray spectrometers (EDSs) detect signals from elements above the atomic number 10 (EELS spectra are sensitive down to atomic number 4). The method of X-ray microanalysis is usually applied with ultrathin sections and inorganic samples.
19.3.3 Phase Contrast in Transmission Electron Microscopy Thin biological specimens that mainly consist of elements of low atomic number (H, C, N, O) behave as weak phase objects in the microscope. The object potential (equivalent to the refractive index in light microscopy) decelerates scattered electrons, and the electron wave experiences a small phase shift Δϕ with respect to the non-interacting reference beam when leaving the object (Figure 19.8). The intensity remains (almost) unmodified. Our eye, digital cameras and photo-emulsions cannot record phase differences and the object would actually be invisible. Looking at the difference between the weakly phase-shifted and the reference wave, we obtain a wave of identical wavelength, smaller amplitude, and a phase shift of about π/2 or λ/4 with respect to the reference. If it were possible to shift the reference wave by π/2 too, so that the amplitudes would be aligned, the waves would interfere and produce a detectable amplitude modulation. In light microscopy, the diffracted and the reference beams are conducted through a glass plate (phase plate) that is of different thickness for the reference beam (the beams are separated from each other in the back focal plane where the phase plate is located) and creates the desired phase shift. The non-diffracted beam is, moreover, attenuated, which enhances the amplitude modulation, that is, the phase contrast. The situation is more complicated in the TEM. Here, it is sufficient to mention that the spherical aberration of the objective lens (characterized by the parameter CS) generates phase contrast as a function of the scattering angle and that it can be adjusted by varying the focus. The ideal adjustment is a weak underfocus, known as the Scherzer focus after Otto Scherzer (1909–1982) who carried out the theoretical calculations. However, the phase shift and thus the contrast, is neither constant nor ideal over the complete scattering range, and it is particularly low with Scherzer focus conditions. The focus-dependent contrast transfer function (CTF) describes the contrast contributions for structural details of size d (corresponding to spatial frequencies related to 1/d; Section 19.4.3). The phase contrast may be strong or weak, become zero, or even change its sign. EM images contain more or less completely or correctly transmitted object information as a function of focus. Thus, the appropriate interpretation of images always requires analysis and ideally the correction of the CTF (Section 19.4.3).
19.3.4 Electron Microscopy with a Phase Plate If there was a phase plate for electron microscopes analogous to the one in light microscopy, one could adjust ideal focus conditions at maximum contrast. The physicist Hans Boersch (1909–1986) had already introduced the basic ideas in 1947, but its realization was hampered by technical problems. Only recently have technical developments changed the situation. The basic principle is to apply an electric potential to the reference beam (or to the scattered one) to create the required phase shift of π/2. This is obtained by an electrostatic potential in the center of the back focal plane for the unscattered electrons (Boersch phase plate) or by a thin carbon film with a small central hole, leaving the reference beam unchanged (Zernike type phase plate; Figure 19.10a). While both solutions and variants thereof were realized, they did not find their way into everyday applications. In particular, biological EM has not used
495
496
Part II: 3D Structure Determination
Figure 19.10 Phase contrast by phase plates in transmission electron microscopy. (a) Scheme of the optical path in the microscope. The undiffracted (zero) beam is focused in the back focal plane below the object lens. The Zernike type phase plate possesses a central hole so that the zero beam can pass through whereas the diffracted beam crosses the phase plate material (carbon film) and experiences a phase shift due to the positive inner potential of the material. The ideal result is a phase shift of π/2 compared to the zero beam (positive phase contrast). The Volta phase plate is a continuous (carbon) film heated to >200 °C to avoid contamination during irradiation. The focused zero beam induces a negative surface (vacuum) potential (Volta potential) that overcompensates for the positive inner potential of the material and effectively shifts the phase by ideally π/2. The object is imaged with positive phase contrast. (b) Electron microscopic projection of part of a native frozen-hydrated and unstained worm sperm taken without a phase plate and (c) with a Volta phase plate illustrating the increase of contrast. The black dots are gold markers. Bar indicates 200 nm. Source: courtesy of Maryam Khoshouei, Martinsried, Germany.
electrostatic phase plates, and the thin film Zernike-type phase plate suffers from charging and produces fringing in images, an unavoidable diffraction effect of the sharp edge of the central hole. The effect can be lessened by image processing, but it still affects the image quality. A novel type of thin film phase plate, the Volta potential phase plate, exploits the electric (surface or vacuum) potential created by the high intensity of the reference beam in a contamination-free carbon film (Figure 19.10). This principle has only recently been discovered, and the formation of the electric potential is not yet fully understood. However, this tool avoids fringing (there is no hole), contamination (by heating), is stable, reusable, and does not require complicated alignments – an important advantage for routine and frequent applications in cryo-EM and cryo-electron tomography. The resulting phase contrast is remarkable, so that this technique will likely become a standard application in biological EM of unstained and frozen material (Figure 19.10b).
19.3.5 Imaging Procedure for Frozen-Hydrated Specimens Cryo-electron microscopy is the most important development in microscopic structure research of biological material. The technique enables us to examine molecules and cellular components at high resolution in a close-to-native state. However, chemically untreated and unstained samples are very radiation-sensitive and are rapidly destroyed upon irradiation. The electrons create molecular radicals that readily react with other molecules and cleave chemical bonds, leading to mass loss and eventually to structural destruction. The low temperature of liquid nitrogen ( 196 °C) lessens the reaction rate and renders the biological object six times more resistant to radiation than at room temperature. Cooling with liquid helium (4 K) promises an even higher cryoprotection factor but unfortunately presents us with problems created by the loss of friction in frozen hydrated specimens. Resolution-limiting radiation damage is indicated by bubble formation in the frozen sample (Figure 19.11). To avoid this effect the total electron dose should not exceed 100 e Å 2, whether it is applied to a single projection or to a complete tilt series for 3D reconstruction (Section 19.5). In the latter case, the tolerable electron dose must be shared by all projections and adjustment procedures. Although it is technically possible to keep the dose arbitrarily low for each projection and to protect the specimen from any damage, one needs a sufficiently high
19 Electron Microscopy
signal-to-noise ratio for subsequent image analysis and processing (Section 19.5.3). The only way between Scylla and Charybdis here is to use most of the electron dose for data recording and do all the adjustments at other object sites. In cryo-electron microscopy this process is automated. The microscopist avoids direct inspection of the object on the fluorescence screen and instead records images by means of a camera. The appropriate procedure is to search the area of interest at low magnification, recording a test image with the desired magnification close to the object site, and determining the focus and other parameters offline. The program calculates the actual values for the object site and adjusts the imaging parameters of the microscope accordingly. The micrograph is recorded automatically, and the electron beam is deflected immediately after in order to minimize irradiation. In cases that require many (hundreds to thousands) images of single particles, the program selects a new site and thereby scans large areas of the grid. If using different projections from a single object, the program actuates the sample holder, turns it by a given angle, centers the object site again, adjusts the focus conditions, records the image, and continues accordingly until the series of tilt angles (tilt series) is complete. In this way, almost the entire electron dose is available for image recording. Projections of thick, ice-embedded samples contain a significant amount of information originating from inelastically scattered electrons that attenuates the image contrast considerably (Section 19.3.2). The use of energy filters in the mode of zero-loss filtering excludes undesirable electrons that experienced energy loss. Only elastically scattered ones can pass, resulting in improved image quality (Figure 19.12). The contrast enhancement is physically independent of the contrast created by phase plates. The techniques are complementary and they both contribute to optimizing image contrast and quality.
19.3.6 Recording Images – Cameras and the Impact of Electrons The last step in image formation is the detection of the imaging electrons. In recent years, digital cameras have replaced film plates, and they are now the usual medium for recording electron micrographs. CCD (charged coupled device) cameras record electrons in a scintillator layer that emits photons instead. By repeating this process, these photons produce a shower of new photons and thus increase the signal while also broadening it at the expense of signal intensity of high-resolution information. The camera imprints another characteristic on the recorded image, namely, the modulation transfer function (MTF), which modulates (attenuates) the signal of structures according to its spatial frequency. Recently, the construction of direct electron detectors yielded two important advantages. Firstly, the incoming electron is recorded in the respective pixel of the detector and not spread over a couple of neighboring pixels, that is, the spatial resolution is high. The MTF is considerably increased for higher spatial frequencies, and small structural details can be detected much better. Secondly, the detector is very fast and allows images to be read out in milliseconds, much faster than with CCD cameras. Recording movies (a number of frames) instead of one single (final) image revealed that the objects are moving by beam-induced effects. These movements blur images taken by conventional
497
Figure 19.11 Cryo-electron microscopy of vitrified cells of the archaeon Pyrodictium abyssi imaged with (a) low and (b) high cumulative electron dose. The cells are embedded in ice between the bars of the carbon support. One of the cells contains a protein crystal (enlarged and displayed together with its power spectrum (PS) in the inset). The heavily irradiated image shows clearly attenuated spots in the PS and bubbles (bright) within and outside of the cell. They indicate massive beam damage of the biological material and ice. The cumulative dose must therefore be kept below a critical threshold to prevent detectable beam damage in the object. Source: courtesy of Stephan Nickell, Martinsried, Germany.
Energy filter Electrons of various energy states differ in their frequency and wavelength (“color”). They are individually diffracted by the electron lenses and project into slightly different positions in the final image. The projected structures are thus of varying size and attenuate the sharpness and contrast upon superposition in the final image. Electromagnetic energy filters widen the electron beam according to the spectrum of electron energies, and an adjustable aperture selects “monochromatic” electrons of identical wavelength for imaging or analysis. Energy filters correspond to color filters in optical microscopy.
498
Part II: 3D Structure Determination
Figure 19.12 TEM images of vitrified lipid vesicles ((a) and (b)) and of frozenhydrated enzyme complexes tripeptidyl peptidase II from Drosophila melanogaster, (c) and (d). (a) Electron micrograph without energy filtering; elastically and inelastically scattered electrons contribute to the image. (b) Only elastically scattered electrons formed the phase contrast image. Inelastically scattered electrons with lower energy were filtered out. The multiply nested vesicles are now clearly visible. (c) The enzyme complexes are poorly detectable in the original micrograph because of the low contrast and signal-to-noise ratio. (d) The macromolecules in different orientations with higher magnification. Classification of equivalent projections are shown in Figure 19.19. Source: part (b) courtesy of Rudo Grimm, Martinsried, Germany; part (d) courtesy of Beate Rockel, Martinsried, Germany.
cameras and also destroy high-resolution information. But the shifts can now be corrected by aligning the frames of a series and adding the frames to a well-resolved and unblurred image. The new detectors push the quality of appropriate single-particle reconstructions to quasiatomic resolution, a “quantum step” in electron microscopic imaging. The first step in electron microscopy – recording images of the biological sample – is now complete. The second part deals with data analysis and image processing for 3D reconstruction.
19.4 Image Analysis and Processing of Electron Micrographs Electron micrographs contain the recordable signal of the object, although obliterated by unstructured noise and modulated by the contrast transfer function of the microscope and the signal transfer characteristics of the camera. The smaller the objects and the higher the expectations for the resolution of molecular details are, the more the contributions of noise and hardware effects affect the visibility of structures. Noise and effects of transfer functions must thus be determined and ideally separated from the desired signal. This is the business of image analysis and the processing of electron micrographs.
19.4.1 Pixel Size Digital EM images are usually 2048 × 2048 (“2k”), 4096 × 4096 (“4k”), or more pixels in size. The beneficial primary magnification of the microscope depends on the pixel size of the camera and the desired resolution of the image. Harry Nyquist (1889–1976), Claude E. Shannon (1916–2001), and others described the minimal condition for the resolution of structural details
19 Electron Microscopy
in digital images. A structural element can be regarded as being resolved if it is defined by at least two pixels (in one dimension). If Pc is the pixel size of the camera and M the primary magnification of the microscope (including corrections for the optical distance to the camera) the pixel size on the object level is Po = Pc/M and the resolution limit d = 2Po. The pixel should not exceed 1=3 or 1=4 of the desired resolution to compensate for some loss upon interpolations (Equation 19.3) and to cope with the limited sensitivity of cameras (Section 19.3.6): Po
Pc d M 3
(19.3)
The visibility of tiny structures depends on the physical resolution of the microscope (Equation 19.2), on the parameters of image recording, the type and quality of the camera, and on the delicate preparation of the biological object (fixation, staining). The shape of negatively stained protein molecules can be resolved to 2 nm. Parts of this size consist of about 30 amino acids (NAA) with a mass of 3400 Da (MPROT). These values, as well as the domain volume Vd (nm3), can be assessed with Equation 19.4. The factors 3.9 or 7.6 (nm 3) and 430 or 840 (Da nm 3) derive from the protein density (1.41 g cm 3) and the average mass per amino acid (110 Da) in proteins. The average protein density depends on the protein size and increases to about 1.5 g cm 3 as the protein size changes from 30 to 10 kDa. Often, the molecular water layer on proteins is part of the volume determinations and reduces the average density to 1.37 g cm 3, a value that is commonly applied in calculations: NAS 3:9d 3 7:6V d MPROT 430d 3 840V d
(19.4)
Vitrified proteins yield structural information to 0.5 nm or even better after image processing. Thus, the secondary structure and in ideal cases even the density of amino acid residues is resolved. Cameras with a typical pixel size of 15–20 μm suggest a primary magnification M of 30 000–100 000. Higher magnifications are usually not necessary and would increase the radiation burden, which is related to M2.
19.4.2 Fourier Transformation Many operations for image analysis and processing are not performed with real data but with its Fourier transform (FT). It is used, for example, to judge the quality of the contrast transfer function (CTF) and to analyze structural regularities of objects. The FT is also required for improving the signal-to-noise conditions in low contrast images and for certain steps in averaging and 3D reconstruction. It is helpful to know what characteristics a Fourier-transformed image has, but it is not necessary here to introduce the mathematics developed by Joseph Fourier (1768–1830). Let us consider a one-dimensional image, that is, a single pixel line of a micrograph, representing the intensity curve of an object. Fourier’s insight was that almost all continuous curves can be represented by the sum of (infinitely many) basic sine functions of varying frequencies (Figure 19.13). To distinguish the sine oscillations along a distance in space (e.g., along the x-direction of an image) from the oscillations in time (frequency) we use the term spatial frequency. The inverse or reciprocal of the spatial frequency, the wavelength, identifies a certain distance d in the real image, and the amplitude of the sine wave is related to the density oscillation along d. Mathematically, the sine function has the amplitude 1. To adjust it to the corresponding density value in the image we need to multiply it by a factor. These factors describe the density distribution (i.e., the gray values) of the object’s structure and are known as structure factors. Sine functions with small spatial frequencies (large wavelengths or values of d) belong to large object structures, and functions with high spatial frequencies (small values of d) to small structural details. Obviously, the macrostructure (e.g., the body of a hedgehog) is characterized by large structure factors and the fine-structure (e.g., the stingers) by much smaller amplitudes of the corresponding sine functions. If an image is to be analyzed to maximum resolution the very small structure factors have to be separated from contributions of the superposed high-frequency noise. The amplitudes of sine waves are, however, not sufficient to exactly describe the curve of a structural density. Moreover, we have to know the exact location of the origin of each sine wave. The position of the corresponding density curves defines where the object (the hedgehog) is located and the substructures (the stingers) are arranged. Since the trigonometric function is
499
500
Part II: 3D Structure Determination
Figure 19.13 Fourier analysis (Fourier transformation) of a one-dimensional object (a, curve in black). The superposition of three sine functions (broken lines in red) approximately traces the original structure (continuous line in blue). The object could be exactly reconstructed by (infinitely) many sine functions with increasing spatial frequencies, appropriate amplitudes, and phase shifts. Plotting the amplitudes and phase shifts of all constructive sine functions against the spatial frequency produces the Fourier transform of the original image (b). The transform has basically the same size as the original image, but it contains equivalent Fourier data symmetrically to the center that are related according to Friedel’s law (Georg Friedel, 1865–1933). This relation is clearly defined and so it is sufficient to show only one half of the Fourier transform.
periodic, the largest difference between the origin of a particular sine wave and a common reference point (e.g., the origin of a coordinate system in the center of an image) is the wavelength of the corresponding sine function. This deviation is called the phase shift (–π Δϕ π), often (but incorrectly) referred to as phase. The object function is thus fully described by the sum of elementary sine waves with (continuously) increasing frequency νi and the corresponding individual amplitudes ki and phase shifts Δϕi. A description of an image with an infinite number of sine functions is impractical. Since we usually record digital images of n pixels in size (e.g., in the x-direction) we subdivide the object function F into n pixels with n/2 spatial frequencies, that is, basic sine waves. The smallest frequency ν possesses a period of exactly 1 with a wavelength λ = n pixels corresponding to the image size S (S = n pixels where ν = S/λ). The highest frequency (Nyquist frequency) is characterized by the smallest possible wavelength with λ = 2 pixels. The object function F is the sum of all possible (finite) sine functions (Equation 19.5). The pixel number s indicates a position in the image of size S, and x denotes the relative position (x = s/S). The value F(x) describes the density (gray value) of a pixel at curve position x and is calculated by: F
x s x S
P
ki sin
2πv i x Δϕi
(19.5)
The Fourier analysis of an image is nothing other than the calculation of all the structure factors ki and phase shifts Δϕi that are necessary to scale the sine waves with frequencies νi and to place them with respect to the origin of the coordinate system (the image). Fourier transforms are therefore data (“images”) that consists of two parts. One contains the amplitudes (structure factors), the other the phase shifts of the sine functions plotted against the spatial frequency ν. Since ν is related to the reciprocal of distance in real space, Fourier transforms are representations of images in reciprocal (Fourier or frequency) space. The Fourier transform F(x,y) of a two-dimensional image requires two-dimensional sine functions (x,y) of course, but they are essentially operated the same way (Figure 19.14).
19 Electron Microscopy
501
Figure 19.14 (a) Examples of two-dimensional sine functions (organized in a 2D lattice) with lower and higher spatial frequencies, varying orientations, and different amplitudes. The right-hand column shows the real images of the functions (bright areas indicate positive, dark areas negative values); the left-hand column shows the corresponding power spectra (PS). The spots in the PS characterize the spatial frequencies and the orientations of the lattices in the x–y-direction. (b) Examples of the synthesis of images through superposition of basic sine waves. The upper image contains the information from the first two images in (a), the other images increasingly more sine functions with different orientations and higher spatial frequencies. (c) Examples of symmetric and antisymmetric (non-periodic) images (column on the left), their Fourier transforms (FT, central columns), and power spectra (column on the right). The FTs consist of a symmetric (real) and an antisymmetric (imaginary) part each (see Equation 19.6). The Fourier data represent negative (dark) and positive values (bright); zero corresponds to a medium gray. The FT of the symmetric image possesses Fourier coefficients unequal to zero only in the real part, the FT of the antisymmetric Yin Yang symbol only in the imaginary part. The image in the middle contains symmetric and antisymmetric features that can be separated by setting one or the other of the two FT parts to zero prior to back-transformation.
Usually, the FT is given in an alternative form, which is obtained by a simple trigonometric operation (Equation 19.6): ki sin
2πv i x Δϕi ai cos
2πv i x bi sin
2πv i x ai ki sin
Δϕi and bi ki cos
Δϕi
(19.6)
The amplitudes and phase shifts can be calculated from the factors ai and bi. The FT of an image is again a two-part representation of the factors ai and bi as functions of spatial frequency. This representation has the advantage of possessing analytical properties. Since the cosine is a symmetric function with respect to the origin while the sine is an antisymmetric one, the part of the FT containing the factors ai characterizes the symmetric features and the other part with factors bi the antisymmetric features of the structure. The Fourier transform is a complex mathematical function (and usually written in an exponential form); the symmetric part is thus also named the real part and the antisymmetric one the imaginary part of the FT (Figure 19.14). Image processing systems make use of the fast Fourier transform (FFT), which is an efficient algorithm for transformation of digital (discrete) data of certain dimensions. It is often sufficient to know the structure factors (intensity, power) and their distribution in reciprocal space to get some idea of object characteristics and image quality. These are provided by the power spectrum (PS), that is, the product of the FT with its complex conjugate FT∗. The PS is the transform of the auto-correlation function of an image or the “cross-correlation” function of an image with itself (Section 19.4.4). The PS misses the phase information and contains the squared structure factors only. It corresponds to the light-optical diffractogram that is created by diffraction of coherent light (e.g., of a laser beam) passing through a micrograph of a structure (Figure 19.14). Calculations of Fourier transforms, power spectra, and correlation functions are standard operations in program system for the analysis, processing, and 3D reconstruction of images.
19.4.3 Analysis of the Contrast Transfer Function and Object Features The electron microscopic image is a function of the projected object structure and the contrast transfer function of the microscope that depends on the electron optical characteristics and the
502
Part II: 3D Structure Determination
actual adjustments of imaging parameters. The object structure (the density function) is convolved with the transfer function during the imaging process. Mathematically, this means that the functions are multiplied in Fourier space, and this means they can be analyzed in Fourier transforms of images. Contrast Transfer Function We already know that the diffraction of the electron beam by the object results in positive and negative phase contrast, depending on the diffraction angle, and that regions in-between are missing contrast, that is, the structural factors of corresponding spatial frequencies are (close to) zero (see Section 19.3.2). The transfer function is particularly clear in projections of a thin amorphous carbon film. Depending on the focus adjustments the power spectrum of such an image shows several bright rings separated by dark gaps. The bright regions represent the squares of the structure factors of transferred spatial frequencies. The dark rings denote gaps in information transfer (contrast). The intensity of corresponding spatial frequencies is very low, or even zero, and thus eliminated. This pattern is known as Thon rings (Figure 19.15). The structure factors of the object that falls into these gaps are eliminated as well, and the corresponding information is thus missing in the electron micrograph. Structure details represented by spatial frequencies falling into the region just beyond the first gap are imaged with inverted contrast and so on. Such effects can hardly be identified by eye in projections of biological specimens, but a series of images taken with different focus values illustrates the effects (Figure 19.15). There is an ideal focus level (moderate defocus) that guarantees a continuous transfer of contrast to a spatial frequency of 1 nm 1 and that is advantageous for biological electron microscopy of (stained) protein complexes. This focus level is close to the absolute contrast minimum and it requires some training to prefer it to the allegedly clear, contrasty but heavily defocused adjustment. If we try to obtain reconstructions with maximum resolution we have to correct for the contrast reversal of certain spatial frequencies afterwards. The Fourier coefficients between the first and the second gap (and third and fourth gap, etc.) are “flipped” by multiplication by 1, and images with different focus levels fill the missing information of the unavoidable gaps (producing different CTFs). Two typical imaging aberrations, that is, astigmatism and drift, are easily identified in power spectra (and should encourage the microscopist to discard those data). Astigmatism is the effect of different focus states in perpendicular image directions. It stems from distortions of the electromagnetic field in the lenses of the microscope. The Thon rings become ellipses and hyperbolas close to in-focus (“Gauss focus”) situations. The gapless transfer of the signal to a certain spatial frequency (casually referred to as resolution) may therefore change dramatically in different directions of the image. Such micrographs typically exhibit streak artifacts. Objects or specimen holders that drift while the micrograph is being recorded produce blurred images in the direction of movement and Thon rings with partially reduced intensities. It is an essential part of quality control to evaluate the transfer function and then select suitable electron micrographs for further processing. Object Characteristics The Fourier transform of an amorphous structure, for example, of single protein molecules, is not very informative for the interested observer (Figure 19.14). However, the situation differs for regularly arrayed macromolecules in 2D lattices. Here, the same amplitudes and phase shifts occur as often as molecules exist in the 2D crystal, and they accumulate to a significant spot (reflex) at corresponding spatial frequencies in the PS. Moreover, the frequencies describing the regular structure must be related to each other, that is, the higher frequencies are always whole-number multiples of the lowest frequency since all the sine waves must have identical relative positions with respect to all molecules in the lattice. Sine functions that do not hit identical reference positions of each molecule with the same phase (and thus with the same intensity) are not suited to describe the structure of the crystal and are blanked out; their structure factors are zero. This is the case for most frequencies and only a few constructive ones are left, which are identified by a number of regularly arrayed diffraction spots in the PS (Figure 19.14). Irregularly distributed data inbetween originate from other, non-crystalline contaminants and noise. The arrangement of diffraction spots yields information on the crystal structure such as the orientation and type of the lattice (i.e., tetragonal, hexagonal, and others), the lattice constant (the periodic distance between molecules in the crystal), and the characteristic angle between the lattice vectors
19 Electron Microscopy
503
Figure 19.15 (a) Focus series of a negatively stained protein complex (diameter 12 nm). The images were taken with strong underfocus (maximum contrast), moderate underfocus, in focus condition (Gauss focus with minimum contrast), and strong overfocus. The grain pattern is particularly dominant in strong defocus conditions. (b) The corresponding power spectra illustrate the effect of the contrast transfer function that varies with focus. Dark gaps between the bright Thon rings denote missing structure information. These gaps already occur in regions of low spatial frequencies, that is, with relatively large object structures, in cases of strong defocus. The first gaps in the series are located at spatial frequencies of (3.2 nm) 1, (1.4 nm) 1, (60° or to < 60°). Rotational symmetry also reduces the region of missing data from a missing wedge to a missing pyramid (Figure 19.20). Two-dimensional crystals of macromolecules are usually not ideally regular. It is thus profitable to average the unit cells by correlation methods and to then combine the averages in Fourier space. Averaging and interpolation of lattice lines efficiently remove superimposed noise. Biological objects with several or many identical macromolecular units in regular arrangements – for example, capsids of viruses and phages or in helical structure such as flagella, pili, and other protein filaments (Table 19.1) – are suited for averaging approaches. If the geometrical positions of repetitive units are known, for example, the helical arrangement and the number of units per helical turn, it is possible to define the projection geometry for all units. The great advantage of these macromolecular structures is that, in principle, a single projection is sufficient to calculate the 3D reconstruction since one micrograph already contains various projections of the macromolecule. The same goes for virus capsids. In addition, it is of course possible to combine a couple of 3D data sets for a final, better-defined reconstruction (Figure 19.22).
Tomography All tomographic reconstruction approaches share the same principles, that is, projection of the object density by transmission imaging, combining projections from different directions, and calculation of a three-dimensional data cube. The methods are non-invasive and provide insight into an intact object. Essentially, all electron microscopical 3D reconstruction approaches are tomographic ones. However, it has been vernacularized to speak of tomographic reconstructions of individual, noncrystalline objects (complex protein assemblies or intact cells) that were projected under different angles in a tilt series.
19.5.3 Electron Tomography of Individual Objects Individual biological objects such as macromolecular assemblies, membranes, amorphous viruses, ultrathin sections of cells, cell organelles, or intact cells require tilt series data for 3D reconstruction. This application is known as electron tomography, with reference to computer tomography of macroscopic specimens or probands. Three-dimensional reconstructions of single molecules and regular structures, such as described in the previous sections, always include averaging steps of identical particles to increase the quality and completeness of data for the final 3D model. But the reconstruction process itself – that is, filling the Fourier space with data and back-projection or equivalent approaches – is independent of averaging. It is also possible to reconstruct non-repetitive, individual objects (Figure 19.20b). Indeed, the specific feature of electron microscopy is its ability to image individual structures down to the subnanometer range. To do so, (cellular) cryoelectron tomography has to solve two incompatible challenges. On the one hand, we should record many projections over the whole angular range with maximum signal-to-noise ratio to achieve the best resolution of small structures. On the other hand, it is necessary to minimize the
19 Electron Microscopy
513
Figure 19.22 Three-dimensional reconstruction of a negatively stained 2D protein crystal, the surface layer (S-layer) from the Gram-positive bacterium Sporosarcina urea. (a) Averages of the tilt series projections with non-equidistant tilt angles from 0° to 78°. (b)–(g) Horizontal sections through the 3D reconstruction in contour line presentation. The 3D volume contains four unit cells with the lattice constants a = b = 13 nm and the lattice angle of 90°. The sections show the protein density from the outer surface of the layer (b) to the inner surface (g) in positions at 2.6, 1.2, 0.3, 0.3, 1.2, and 2.5 nm with respect to the central plane. (h) and (i) Surface-rendered 3D models showing the inner and the outer surface of the protein layer. For vertical sections through the central protein domain see Figure 19.20.
total electron dose so as not to just destroy the structures that we want to see in the reconstruction! Fractionation of the total dose over all micrographs of the tilt series results in a corresponding loss of the S/N ratio with the consequence that structures may not be detectable among the obliterating noise. The S/N ratio of individual projections must be so high that they are aligned with each other (Section 19.3.3). The mechanical accuracy of the goniometer is not good enough to avoid image shifts upon tilting. These unavoidable errors have to be corrected for afterwards. Small gold clusters (colloidal cold) that are added to the sample before freezing and provide high contrast anchor points facilitate the correlation alignment. Now, the S/N of the structures of interest is not limiting for alignment and may even be low. The two conditions – maximally allowed cumulative electron dose and minimal contrast (especially in images taken without a phase plate) – limit the possible number of projections of a tilt series and thus the achievable resolution of an object of thickness D. Cellular objects should not be thicker than 0.5 μm, which means that cryoelectron tomography applications are restricted to viruses, a number of prokaryotes, and flat areas of eukaryotic cells in the first place. However, cryo-microtomy and ablation of material by ion beam treatment (FIB) offer ways to investigate sections or subvolumes of bigger biological cells in a close-to-native state (Section 19.2.1). Reconstructions of individual objects miss the opportunity to minimize noise by averaging. Other criteria are required to separate structures from uncorrelated noise. One approach is nonlinear anisotropic diffusion. This algorithm exploits the moderate variation of voxels belonging to a continuous structure, which is different from the randomly varying and uncorrelated values of noise. The missing wedge of data also creates problems in tomograms (Section 19.5.2). One effect is that flat (thin) structures such as membranes become blurred or even invisible in the zdirection. Accordingly, the reconstructed top and bottom regions of cells are incomplete. To reduce the region of missing data in Fourier space one could rotate the grid by 90° in the microscope after recording a tilt series and take another series. The missing wedge then becomes a missing pyramid and the data is more complete (Figure 19.20). The final result of a 3D reconstruction is a data cube that contains the density distribution of the reconstructed cell or section. The interpretation of these complex and detail-rich structures requires special approaches for analysis and visualization that allow the researcher to identify and visually enhance distinct structures (Figure 19.23 and Section 19.6).
514
Part II: 3D Structure Determination
Figure 19.23 (a) Original projection and reconstructed 3D data of the virus Herpes simplex, (b) of Mycobacterium bovis BCG, and (c) of a part of the eukaryotic cell Dictyostelium discoideum. All the organisms were imaged by cryo-electron tomography without any chemical fixation and staining and reconstructed by the approach of filtered back projection. The images show the 0° projection, one (central) x–y slice of the reconstruction and the surface-rendered 3D model. The model of the mycobacterium shows the lipid bilayer structure of the inner and outer membranes and material in the periplasmatic space, that is, cell wall polymers. (d) Surface model of the tripeptidyl peptidase II from Drosophila melanogaster as obtained from singleparticle cryo-electron microscopy; the protein dimers from X-ray structure determination are fitted into the electron density. This hybrid approach allows us to calculate a quasi-atomic model of the entire 40-meric enzyme complex. Source: Courtesy of Kay Grünewald (a), Ohad Medalia (c), and Beate Rockel (d), Martinsried, Germany.
19.6 Analysis of Complex 3D Data Sets 19.6.1 Hybrid Approach: Combination of EM and X-Ray Data Cells contain large and heterogeneous protein complexes that may be purified without losing their integrity but that do not crystallize for X-ray structure determination. Examples are the 26S proteasome, the spliceosome, polysomes, the bacterial flagellar motor, the nuclear pore complex, centriole structures, cilia, the cytoskeleton network, and many others. However, in many cases it is possible to isolate the subunits and to determine their atomic structures individually. These subunits can then be fitted into the density of the intact complex as obtained
19 Electron Microscopy
515
from single particle or tomographic reconstructions and can be used to calculate a pseudoatomic model. This hybrid approach often is the only successful way to get deeper insight into the structure and functional conformations of protein complexes; the enzyme TPP II is a typical example (Figures 19.21 and 19.23). There are two approaches for fitting, namely, rigid body docking and flexible fitting. Rigid body docking is suitable for moderate resolution structures (1–3 nm) where the best place and orientation of subunits is determined by correlation methods. Higher resolution structures often reveal local discrepancies between the atomic structure and the reconstructed model that derive from real conformational differences. Molecular dynamics calculations allow adaptation of the atomic structure in such a way that it fits into the structure of the whole complex. Meanwhile, there are explorations of approaches to model structures consisting of many parts or structurally unknown components, such as the nuclear pore complex. Here, the combination of biochemical data and information from other sources yields additional criteria to define restricting conditions for modeling.
19.6.2 Segmenting Tomograms and Visualization Tomograms of cells or cellular segments contain the signatures of many (theoretically all) macromolecules and cellular structures one would like to analyze and identify in 3D presentations. Due to the high protein concentration in living cells (80–400 mg ml 1), creating the phenomenon of macromolecular crowding, it is difficult or even impossible to visually assign density data in tomograms to distinct structures. It is thus necessary to demarcate structures of interest such as membranes, filaments, or macromolecular protein complexes from each other. They are segmented and usually highlighted by colors in 3D models. Figure 19.23 shows EM data and the corresponding 3D models of segmented tomograms from a virus, a bacterial cell, and a segment of a eukaryotic cell. The simplest means of segmentation is to define a gray value that represents the border of a molecular surface and to blank out all voxels below this threshold. This is the common method of rendering 3D models of single molecules (Section 19.6.1), but it is not suitable for complex structural assemblies such as biological cells. Here, we need criteria to identify voxels that are correlated and belong to a coherent structure. There are automated procedures that recognize membranes and filaments even in noisy 3D data, but it will sometimes be necessary to refine segmentations by hand. Segmentation of single protein complexes in the crowded cytoplasm is not possible and so we need more powerful approaches.
19.6.3 Identifying Protein Complexes in Cellular Tomograms If protein complexes are located closely together in cells or if they are part of a supramolecular structure we use the approach of template matching to identify and localize them in tomograms (Figure 19.24). For this purpose, we need a 3D model of the protein complex of interest. Models from X-ray crystallography or single particle reconstructions are suitable data. Cross-correlation of the tomogram with the template yields the position and the orientation of the molecules of interest. The correlation is a function of the two 3D data sets and the six degrees of freedom for the position (x,y,z-coordinates) and the three Euler angles. The required computing power is challenging. However, the approach is feasible; experiments have localized and identified different molecules in phantom cells (lipid vesicles) (Figure 19.24). These experiments showed that a resolution of at least 2 nm is necessary to identify the target molecules reliably and to minimize false positive hits. This is a challenging proposition, but the new electron detectors and improved imaging conditions promise fruitful scenarios. The 3D distribution and arrangement of ribosomes in bacterial cells have already been investigated (Figure 19.24), the native structure of polysomes (poly-ribosomes) and inactive pairs of hibernating ribosomes have been studied in situ, and recent research has identified active and inactive conformations of the 26S proteasome in neuronal cells. Tomograms of cells are actually unique structures, but they contain several redundant proteins that may be extracted from the 3D data set after localization and identification and that
Macromolecular crowding The high concentration of macromolecules in cells means that about 30% of the cytoplasmic volume is occupied by protein complexes and other big biological molecules. The characteristic distance between neighboring macromolecules is only 10–20 nm, that is, the size of many protein complexes. Big complexes can thus not occupy these intermolecular zones – only a rather limited volume remains available for them. There is literally a shortage of space in crowded cells. This situation causes dramatic equilibrium shifts for reactions that increase or decrease the occupied volume (including the surrounding space) of macromolecules. Amongst these processes are isomerization and folding reactions or oligomerization and dissociation of protein complexes. This is the reason why supramolecular complexes and protein assemblies are more stable in cells than in the diluted environment in vitro. Some of these (hypothetical) complexes can thus only be observed in intact cells.
516
Part II: 3D Structure Determination
Figure 19.24 (a) Detection of macromolecule complexes in a 3D data set (tomogram) of a cell is performed by cross-correlating the reconstructed data with a 3D template (template matching). The correlation determines the spatial position (x, y, z) and orientation (three Euler angles) of the molecules. The elaborate process is calculated in a parallel computer. (b) Positions and orientations of the protein complexes proteasome (bright) and thermosome (gray) in a “phantom cell” (lipid vesicle). (c) Projection of the vitrified bacterium Spiroplasma melliferum, (d) slice of the tomogram that shows large protein complexes in the cytoplasm, and (e) image of the correlation function obtained from the tomogram and the template 70S ribosome; bright spots reflect the identified position and orientation of ribosomes in the 3D data set of the cell. (f) 3D model of the cell containing models of the ribosome at places and in orientations as determined by template matching. Size of image 600 nm. Source: parts (c)–(f) courtesy of Julio Ortiz, Martinsried, Germany.
may be averaged after classification (subtomogram averaging). The appeal and the scientific potential of this process is that the complexes remain in the natural environment, in a native and untouched state, and that their functional interactions with surrounding proteins can be visualized. The nuclear pore complexes are one example: they are immobilized by freezing in different situations of translocating the cargo. By cumulating many subtomograms and sorting them into the right order it was possible to obtain a “movie” of the translocation process.
19.7 Perspectives of Electron Microscopy About 70 years after its invention in 1931 electron microscopy opened up new perspectives for molecular structural biology and cytology by establishing advanced cryotechniques, 3D reconstruction, and visualization methods. About a decade later, two technical (r)evolutions mark another “quantum step” in electron microscopy – new electron detectors, improved resolution, and an employable phase plate ameliorated contrast. Two lines of applications delineate the fascinating future of electron microscopic structure research. Single particle analysis of protein complexes deals with many projections, reaching up to millions. With the use of direct electron detectors it is now realistic to aim for the atomic resolution of complexes, especially of those that cannot be tackled by other methods of structure research. While it is possible to obtain atomic models of rigid protein complexes with less than
19 Electron Microscopy
105 single particles, an even more fascinating perspective is to collect as many projections of flexible complexes in various conformational states as possible, to then classify them, and to sort the different structures into a consecutive order of transformation. The result is a quasitime- or process-resolved series obtained from “four-dimensional” electron microscopy. The first examples were the translocation of tRNA between different binding sites in ribosomes and the visualization of the flexibility of the nuclear pore complex in native nuclei. Investigations at higher resolution showed conformational changes of the 26S proteasome that led to an atomic model of functional transitions. The development of cryo-electron tomography paved the way for 3D models of intact cells or sections thereof in a close-to-life state. Even at a moderate resolution of 3–5 nm we can detect intracellular structures that cannot be observed in conventionally fixed and embedded preparations. This immediately shows the significance of CET for the investigation of pathological effects in cells. The improved detectors, the correction of beam-induced movements, and the impressively enhanced image contrast created by a phase plate mark a further quality step. We can realistically expect to interpret cellular tomograms at the level of about 1 nm in the near future. A single tomogram contains the signatures of hundreds of proteins and supramolecular complexes, that is, a wealth of 3D information of the cellular proteome. We assume that many proteins interact in the cytoplasm and temporarily form supramolecular complexes under the conditions of macromolecular crowding in cells. Such complexes tend to dissociate in diluted environments and cannot be isolated as stable structures. Only cryo-electron tomography enables us to visualize macromolecular aggregates in situ and to investigate the structural network of different macromolecules. Once the proteins in tomograms are identified and their positions and orientations known, it is possible to dock atomic models into the electron densities and to create a pseudo-atomic map of macromolecular structures and their interactions in individual cells. There is still some way to go until medium-sized protein complexes or even small ones can be unambiguously identified, but we have taken the first steps.
Further Reading Frank, J. (2006) Electron Tomography: Methods for Three-Dimensional Visualization of Structures in the Cell, 2nd edn, Springer, Berlin. Frank, J. (2006) Three-Dimensional Electron Microscopy of Macromolecular Assemblies: Visualization of Biological Molecules in their Native State. Oxford University Press, Oxford. Hawkes, P.W. and Spence, J.C.H. (2007) Science of Microscopy, vol. I, Springer, New York. Reimer, L. and Kohl, H. (2008) Transmission Electron Microscopy. Springer Series in Optical Sciences, vol. 36, Springer, New York. Williams, D.B. and Carter, C.B. (2009) Transmission Electron Microscopy. A Textbook for Materials Science, 2nd edn, parts 1 to 4, Springer, New York.
A public domain program for the analysis and processing of microscopic images: NIH, ImageJ: Image Processing and Analysis in Java. http://imagej.nih.gov/ij/
517
Atomic Force Microscopy Daniel J. Müller1 and K. Tanuj Sapra2 1 2
ETH Zürich, Biosystems Science and Engineering, Mattenstrasse 26, 4058 Basel, Switzerland Department of Biochemistry, University of Zurich, Winterthurerstrasse 190, 8057 Zurich, Switzerland
20
20.1 Introduction The year 1986 ushered in a new era of imaging and manipulation of hard and soft matter at the nanoscale, turning theoretical (or undreamed of) possibilities into practical realities. The invention of a conceptually simple yet very powerful device, the atomic force microscope (AFM), provided a tool to zoom into the molecular scale thereby revolutionizing nanotechnology. The last three decades have witnessed great strides in AFM technology; it is now a routine to map objects with a resolution of up to a few angstroms (Å), manipulate them with high-precision, and at the same time quantify their physical, chemical, and biological properties. The AFM belongs to the family of scanning probe microscopes (SPMs), which utilize a sharp probe as a scanning “stick” and a handle to manipulate (e.g., pick, drop, remove) objects at the nanoscale. The SPM microscopy technique relies on specific interactions between the probe and the object; in many cases the interactions can be tailored to a specific sample or the application. The detection system in SPM microscopes includes optical signals in scanning near field microscopy (SNOM), tunnel currents in the scanning tunneling microscope (STM), ion currents (scanning ion conductance microscope, SICM), or magnetic interactions in the magnetic force microscope (MFM). The detection versatility has enabled the development of more than 20 different measurement applications for the scanning probe microscopy of inorganic and organic samples. As explained in Section 20.2, the AFM detects interaction forces between a sharp atomic or molecular probe and the object. A major advantage of the AFM is that sensitive biological samples can be investigated in their natural aqueous milieu under defined conditions; for example, specific pH, ion compositions, and temperature, which simulate the physiological conditions. The signal-to-noise ratio of the AFM compares superlatively to the optical and electron microscopes. However, a thorough understanding of the molecular interactions between the AFM cantilever tip and the sample (Section 20.3) and optimum sample preparation (Section 20.4) are imperative prerequisites for achieving a signal with low noise to enable molecular mapping of soft biological matter at sub-nanometer resolution. Because the energy imparted by a cantilever tip while scanning a surface (Section 20.3) is of the order of thermal energy (3.5kBT), individual biomacromolecules can be scrutinized with high precision in vitro or in situ without compromising their structural and functional integrity. In contrast, photons of wavelength 300 nm possess energies 3150kBT, which is sufficient to break covalent bonds of organic molecules and is detrimental to protein structure. Section 20.5 provides a few examples of sub-structural (high-resolution) AFM imaging of biological cells, single proteins, nucleic acid polymers, and sugar chains under native conditions. The fast recording of imaging sequences allows direct observation of the molecular machineries at work, and the dynamics of the macromolecular complexes in a cell. Imaging Bioanalytics: Analytical Methods and Concepts in Biochemistry and Molecular Biology, First Edition. Edited by Friedrich Lottspeich and Joachim Engels. 2018 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2018 by Wiley-VCH Verlag GmbH & Co. KGaA.
Electron Microscopy, Chapter 19
520
Part II: 3D Structure Determination
with AFM extends beyond simple topographic mapping; information on the physical and chemical properties of a system can also be obtained by functionalizing the cantilever probe. Of direct relevance to studying biological mechanisms is the monitoring of various biological interactions, for example, cell–cell adhesion, interaction forces between individual receptor– ligand complexes, chemical groups, or even observing the convolution and unfolding processes of individual proteins (Section 20.6). In addition, the AFM is gaining importance as an analytical tool; for example, to determine the mechanical properties (flexibility, stability) of biological and synthetic polymers, and also biological macromolecules. It is becoming increasingly evident that the AFM offers unprecedented high spatial resolution unmatched by any bioanalytical or biophysical approach. Thankfully, the potential of scanning probe microscopy in bioanalytical applications is far from being exhausted.
20.2 Principle of the Atomic Force Microscope Inarguably, the most crucial functional element of the AFM is a sharp probe (or tip) made of silicon or silicon nitride (Si3N4) and engineered at the ventral end of a microscopic spring lever (also known as the cantilever); the whole system is designed to have a low spring constant (k 0.1 N m 1) (cantilevers with higher spring constants are produced for applications in material sciences). For the cantilever to function as a force sensor, the dorsal or the top surface of the cantilever is usually metal-coated onto facilitate and maximize reflection of the laser beam to a quadrant photodiode (Figure 20.1). The quadrants of the photodetector maintain a voltage difference, which to a first approximation depends linearly on the applied force. The inclusion of a piezoelectric transducer optimizes the raster scanning capabilities of the tip, which is able to inspect each defined point of the biological surface while detecting the interaction forces at those points (Figure 20.1). The cantilever acts as a force amplifying arm, that is, upon sensing a force the cantilever deflects and the signal is registered on the location-sensitive photodetector. The interaction forces consist of repulsive and attractive contributions, and are typically in the range 0.01–100 nN (Table 20.1). Table 20.1 Forces between the AFM tip and object. Force
Figure 20.1 Schematic representation of the atomic force microscope. An immobilized sample is moved by means of a triaxial (x, y, z) piezoelectric element under a sharp scanning cantilever probe. During this raster movement, the bending of the cantilever spring is measured by a laser beam reflected onto a photodetector. The voltage difference between the upper and lower segments of the photodetector (V = (A + B) – (C + D)) is a direct measure of the cantilever spring deflection, which is used to calibrate the spring constant and quantify the miniscule interaction forces.
Direction of force
Range
Pauli repulsion
Repulsive
Extremely short (0.2 nm)
Van der Waals interactions
Attractive
Very short (a few nm)
Electrostatic interactions
Attractive/repulsive
Short (nm to μm)
Capillary (probe in water)
Attractive
Long (μm to mm)
20 Atomic Force Microscopy
So far, several methods have been developed for scanning force microscopy. In the most commonly used scanning method – the contact mode – the cantilever tip is maintained in contact with the sample surface at a constant user-defined force. This is achieved by keeping the cantilever deflection (i.e., the force) constant by continuously monitoring and changing the distance of the sample surface from the cantilever tip. To ensure the integrity of the biological samples and that the surface structures are not irreversibly deformed, a scanning force of 0.1 nN is generally used (see Section 20.3). By regulating the height required for a constant contact force, the surface features of the sample are mapped point-by-point resulting in a surface topography (Figure 20.2a). Besides forces normal to the plane of the sample surface, lateral forces can deform or scratch away a soft biological object from its base during raster scanning in contact mode (Figure 20.2c). The undesirable sample distortions by normal and lateral interactions can be minimized by setting low contact forces (see Section 20.3). Dynamic imaging methods (e.g., TappingTM mode) offer an alternative approach to reducing the impact of the scanning tip on the object. In the TappingTM mode, the cantilever spring is stimulated to a sine-wave oscillation close to its natural resonance frequency. The oscillation is a means to ascertain that the tip touches the biological object only at the lower end of each cycle (Figure 20.2b), as a result minimizing the contact time and lateral interactions. Dynamic imaging procedures are therefore particularly suited for imaging weakly immobilized biological objects (e.g., cells or fibrillar structures). A crucial criterion for obtaining an unperturbed topography is to maintain the oscillation amplitude of the cantilever by controlling the distance between the cantilever tip and the sample surface.
521
Figure 20.2 Imaging with atomic force microscopy. (a) In the contact mode, the tip is in continuous contact with the biological sample and, therefore, follows the surface features of the sample even upon encountering an obstacle, resulting in deflection of the cantilever spring. To protect the sample from excess force, the contact force (measured by cantilever deflection) is maintained by changing the Z-position of the sample with the aid of a feedback loop. The error signal is used to map the sample topography. (b) In the dynamic mode, a cantilever is oscillated close to its natural resonance frequency with an amplitude of a few nanometers. Thus, the object is touched only briefly during the downward travel of the cantilever avoiding lateral interactions. The feedback control loop in this case is used to maintain the oscillation amplitude. (c) Lateral scanning of the AFM tip can deform a soft biological object. This can be prevented by precise adjustment of the contact force, the scan speed, and the feedback parameters. (d) Elasticity is determined by moving the cantilever tip against a soft object (e.g., cells). The ratio of the distance travelled by the cantilever into the sample and the cantilever deflection associated with this gives a measure of the sample elasticity.
Topography Three-dimensional illustration of a surface.
20.3 Interaction between Tip and Sample Akin to its siblings of the SPM family, the AFM can deliver resolution up to atomic dimensions. Albeit the achievable resolution depends crucially on the tip sharpness and the surface corrugations of the object, the finite radius of the cantilever tip limits the contouring of sharp and delicate features of an object. As shown in Figure 20.3a, the tip can broaden the lateral dimensions of an object. In contrast, structural periodicities with less defined edges can be reproduced with a higher accuracy (Figure 20.3b). In all these cases, the measured topography represents a nonlinear superposition of the object under investigation, the strength of the pronounced details being dependent on the corrugation of the object and the tip dimensions. Owing to the considerably soft nature of biological objects compared to the cantilever spring, contact forces of 1 nN exerted by a cantilever tip are sufficient to deform and denature proteins. As explained above (Section 20.2), minimizing vertical and lateral physical forces is key for scrutinizing biological samples reliably at a high resolution. It is therefore of utmost importance to understand the different interaction mechanisms between the tip and the surface. Consequently, the selection of suitable cantilevers, imaging procedures, and the optimization of regulating parameters (e.g., feedback loop, scanning
Figure 20.3 Superposition effects between the tip and the sample can distort the AFM topography. (a) Owing to its finite diameter (10 nm in most cases), a cantilever tip cannot trace very sharp edges. Therefore, the imaged topography (gray line) is a superposition between the tip and the sample surface. (b) In many cases, however, the periodicity of the biological structures can be resolved correctly irrespective of the tip diameter.
Corrugation waviness of a surface.
522
Part II: 3D Structure Determination
speed, image size) are key experimental parameters that require meticulous tuning. For example, the type of ions (i.e., monovalent or divalent) and their concentrations play a decisive role in increasing the resolution of a map; they are to be selected such that a repulsive (0.05 nN) and long-range (several nm) interaction is generated between the tip and the object. This strategy increases the contact force marginally with only a small fraction acting locally on the protein structures thus preventing sample deformation in most cases. However, this tuning of the interactions is associated with an increase in the gross force of the coupled cantilever–sample system, further suppressing the natural thermal resonance of the cantilever spring (Section 20.5). Advantageously, this reduction in the thermal noise increases the signal-to-noise ratio, resulting in a high resolution of the AFM topography.
20.4 Preparation Procedures Because the general working principle of the AFM is based on sensing and detecting molecular interaction forces, it is not necessary to metal coat or fluorescently label biological macromolecules or cells to identify them. The most important and often the only sample preparation step required is to immobilize the biological sample on an atomically flat support. This is a mandatory prerequisite for high-resolution imaging because the corrugation of an atomically flat surface hardly superimposes with the morphology of the soft biological sample and enables precise spatial control of the cantilever tip on the specimen. The sample preparation therefore requires finding a balance between the need to anchor the biological object while at the same time minimizing the interactions between the object and the support surface to ensure the mapping of an unaltered, native state (conformation) of the sample (protein). Sample supports used in optical and electron microscopy are also utilized for AFM applications. Examples include glass, muscovite mica, graphite, and metal (such as goldcoated) surfaces. Based on the desired application, each of these sample carriers can be tuned to have unique physical and chemical properties (e.g., surface charge or roughness). For example, mica, which is characterized by a layered crystalline structure, is ideally suited for the immobilization of proteins and nucleic acids. An adhesive tape can be used to peel off crystalline layers from the underlying surface to provide a relatively chemically inert, negatively charged, and an atomically flat surface. The most commonly applied immobilization strategies are based either on physical interactions between the biological sample and a chemically inert surface or on the covalent attachment of the sample to a functionalized (reactive) support. Physical adhesion is mainly achieved by shielding the repulsive electrostatic interactions between the biological object and the sample carrier. Thus, increasing the ion concentration leads to an attractive interaction between the sample and the support surface, which is sufficient to immobilize the biomolecules. For this purpose, the buffer solution containing the biological sample is applied directly onto a freshly split mica surface, the macromolecules allowed to adsorb firmly for a few minutes, and their surfaces mapped with a cantilever tip. For chemical coupling, the sample support is first functionalized with a chemical moiety, for example, glass with silanes or gold with thioalkanes, and in a final step the biomolecules are reacted with the chemically active surface. Because biological structures require water molecules for their structure–function dynamics, drying biological samples and imaging in air should be avoided as far as possible. During the drying process, biological molecules are exposed to tensile forces generated by the surface tension of water often leading to the collapse of the samples. Irrespective of the fact that drying artifacts can be minimized or avoided by certain vacuum sublimation procedures, biological samples, whenever possible, should be prepared and imaged in aqueous solutions to preserve their native structural and functional integrity.
20.5 Mapping Biological Macromolecules Using the AFM, surface structures and dynamic processes of a range of different biological samples can be observed under native conditions. These include biological macropolymers (e.g., proteins, DNA, RNA, and polysaccharides), supramolecular complexes (e.g., metaphase
20 Atomic Force Microscopy
chromosomes), as well as bacterial or cellular associations, and tissues of higher organisms. Currently, the highest resolution is achievable on isolated macromolecules immobilized on an atomically flat surface. The technique is capable of providing high-resolution topography of individual membrane proteins in their native lipid bilayer environment, while at the same time the resolution is high enough to monitor and discern subunit dynamics such as associated with helix and polypeptide loop movements during the opening and closure of membrane channels. Figure 20.4 shows the cytoplasmic surface of the purple membrane of the archaebacterium Halobacterium salinarum. This purple membrane consists of lipids and the seven transmembrane α-helical protein bacteriorhodopsin, a light-driven proton pump. The AFM topography of the native membrane surface clearly shows the natural crystalline arrangement of bacteriorhodopsin – individual bacteriorhodopsin molecules assemble as trimers, which further form a two-dimensional hexagonal lattice. The close packing allows the maximum density of proton pumps for light-driven energy production. The AFM topography evidently portrays the molecular (polypeptide loops and termini) variability of each bacteriorhodopsin assembly. Nevertheless, the statistical average of the individual particles reveals a representative structural snapshot of bacteriorhodopsin, and the calculated standard deviation is a measure of the conformational variation of the population. In addition, it is useful to compare the averaged AFM topography with the structural information obtained from complementary structural biology methods (e.g., X-ray crystallography, electron microscopy, or NMR). For example, superimposing the mean contour from AFM on the three-dimensional X-ray structure makes it possible to assign the surface details to the secondary structures, and to further study the structural dynamics of the protein under different conditions. (Figure 20.4) The AFM is also used to characterize the stoichiometry and the formation of membrane proteins in functional complexes. Figure 20.5a shows the Na+-ion driven rotors of the FOF1ATP synthase from Ilyobacter tartaricus; the number and arrangement of the subunits of the functional rotor can be unambiguously recognized. The high-resolution tool provided a straightforward approach to illustrate that the cation-fueled (H+ or Na+) FOF1-ATP synthases from different organisms are constructed of different numbers of subunits. An attractive although debatable speculation is whether the different ATP synthases evolved to adapt the number of rotor subunits to the membrane potential of the cell and thus regulate ATP synthesis for energy consumption. Thus, elucidating the cellular mechanism that controls rotor stoichiometry will advance our understanding of ATP synthesis. AFM is also applied for observing dynamic processes such as DNA transcription, protein diffusion, conformational changes of individual proteins, and even the emergence of molecular networks. The time required to record an AFM topograph is a decisive factor in determining the temporal resolution of the dynamic process. Depending on the system being investigated and the AFM setup, the recording time is between 1 s and 15 min. The so-called “gap junctions” or the communication channels from the rat liver cells are shown in Figure 20.5b and c. The individual connexins of the gap junctions show a nearly perfect hexagonal packing. The central pores of the individual hexamers are clearly discernible in the unprocessed topographs. In the presence of a signaling molecule such as calcium in the buffer solution, reversible closing of the channels can be directly observed (Figure 20.5c). The average surface structures of the channels
523
Figure 20.4 Cytoplasmic surface of the native purple membrane of Halobacterium salinarum. (a) The AFM topograph clearly reveals the assembly of single bacteriorhodopsin molecules into trimers; the trimeric units are further organized in a hexagonal lattice. The membrane topography was recorded in a physiological buffer at room temperature. (b) The diffraction pattern of the topography extends up to the 11th order (drawn circles), which suggests a lateral resolution of 0.49 nm. (c) The average topography (top) and the corresponding standard deviation (bottom) of the bacteriorhodopsin trimer allows correlation to structural data obtained by electron crystallography. Superimposed is the outline of bacteriorhodopsin molecules, as well as the positions of the seven transmembrane α-helices (A–G).
X-Ray Structure Analysis, Chapter 21 Electron Microscopy, Chapter 19 NMR Spectroscopy of Biomolecules, Chapter 18
524
Part II: 3D Structure Determination
Figure 20.5 Determining the protein complex assembly and function by AFM. (a) Na+-driven rotors of FoF1 ATP synthase from Ilyobacter tartaricus; the high resolution topography allows to characterize the stoichiometry and arrangement of the subunits of a functional rotor. (b) The extracellular surface of purified communication channels (gap junctions) from rat liver epithelial cells clearly shows hexameric proteins with an open central channel. The average and the standard deviation (SD) allow insights into the structure and its conformational flexibility. While the average profile reveals the channel entrance, the SD map assigns an increased structural flexibility to the central channel. (c) In the presence of 0.5 mM Ca2 + (at neutral pH) the central channel of the hexamer is closed. This mechanism is evident in the average hexamer structure, and the associated SD map depicts a loss in flexibility of the channel entrance in the closed state. All topographies were recorded in a physiological buffer at room temperature.
and their standard deviations provide insight into the structural variability and map the flexible regions of the protein. An open communication channel demonstrates structural variability and flexibility; these characteristics are lost upon channel closure. This observed relationship between flexibility and functional conformational change of the channel is not an exception, and has been determined previously for other proteins. Thus, to a large extent it can be generalized that the flexibility of structural features often correlates with their ability to perform functionally related conformational changes. The advent of the ultrafast AFM, capable of taking several hundred pictures per second, is a major breakthrough in AFM development, and will allow real-time observation of various dynamic processes under native conditions. In contrast to the sub-nanometer imaging of individual molecules, a maximum resolution of 50 nm is attained for AFM imaging of cells and tissues. The low resolution is partially owing to the flexibility and dynamic motion of living cells, but also due to significant roughness of the cell surface. Although the AFM still provides important physical insight into cellular function, the low resolution of cell surface imaging hinders identification of the observed structures on cells. To overcome this shortcoming, AFMs are often combined with modern light microscopy techniques. The combination makes it possible to correlate topographic information of a cell with its superficial structures (e.g., vesicles or the cytoskeleton, which need to be fluorescently labeled for simultaneous AFM and optical microscopy). In addition to the determination of structural features, the AFM permits the determination of various other physical parameters (e.g., the elasticity of a biological object). The cantilever spring deflects when the tip is pushed against a sample; this signal is analyzed to characterize the elastic properties of the sample. Importantly, an unambiguous interpretation of cell elasticity maps requires the identification of different structural components that contribute to the mechanical stability of the cell. From a morphological standpoint, the cytoskeleton thereby plays a crucial role, can be directly traced by tip-induced deformation of the flexible cell membrane. Alternatively, it is also useful to analyze the phase changes of an oscillating cantilever spring during the dynamic mapping of a cell. Phase modulation reflects changes in the elastic properties, charge, and roughness of a cell surface.
20.6 Force Spectroscopy of Single Molecules The ability of the AFM to detect forces with pN sensitivity helps to characterize the strength of biological and chemical bonds, and the behavior of individual molecules under mechanical stress. An attractive straightforward measurement is between ligands and receptors. This is achieved by first functionalizing the cantilever tip with the protein or the small molecule ligand, and then using the probe to interrogate the binding partner on a sample support. Using this approach it has been possible to detect specific binding and quantify molecular interactions
20 Atomic Force Microscopy
525
Table 20.2 Estimated rupture forces of chemical and biological bonds. Bond
Rupture force (pN)a)
Covalent
1000–2000
Cell–cell interaction
500
Biotin–avidin
200
Antibody–antigen
OH3 > OH4. HPAEC-PAD does not require any derivatization of the monosaccharides and thus is simpler and superior to other methods of monosaccharide analysis (Figure 23.19). With mammalian cell glycoproteins, acid hydrolysis in 2 N trifluoroacetic acid (4 h, 100 °C) with subsequent detection and quantification of the monosaccharides via HPAEC-PAD has proven optimal. Trifluoroacetic acid (TFA) has, compared to HCl or H2SO4, the advantage that it is volatile and removed during lyophilization or in a SpeedVac. At the given acid hydrolysis conditions, Nacetyl sugars (as GlcNAc or GalNAc) will be N-deacetylated and detected as the corresponding amino sugars (GlcNH2 or GalNH2). Since N-acetylneuraminic acid decomposes under these hydrolysis conditions, Neu5Ac must be released in a special and milder hydrolysis or enzymatically (e.g., by means of neuraminidase (sialidase)), and determined separately (see below). As is obvious from the N-glycan structures (Figures 23.9–23.11), defined monosaccharide molar ratios may be concluded from the regular biantennary, triantennary, and tetraantennary N-glycans with complete sialylation. The same applies for the high-mannose type N-glycans (Table 23.4). In some cases the molar GlcNAc/mannose ratio may even enable us to conclude the actual structure type: A GlcNAc/Man < 0.5 ratio generally points towards the presence of high-mannose type N-glycans, whereas at a ratio between 1 and 2 complex type N-glycans may predominantly (or exclusively) be expected. For rhEPO the quotient GlcNAc/Man = 2.0 indicates the presence of complex type N-glycans, where a high proportion of tetraantennary N-glycans may be expected. The finding of mannose (9 mole mole 1) points towards the presence of three glycosylation sites, as complex type N-glycan structures always carry three mannose units per N-glycan. Likewise, the GlcNAc content (18 mole mole 1) and Gal content (14 mole mole 1) indicate the presence of three tetraantennary N-glycans.
587
23 Carbohydrate Analysis Table 23.4 Molar ratios of monosaccharide components of N-glycans.
Complex type
High-mannose type
Erythropoietin
Tetraantennary
Fuc
GalNAc
GlcNAc
Gal
Man
Neu5Ac
GlcNAc/Man
Neu5Ac/Gal
0–1
—
6
4
3
4
2.00
1.00
Triantennary
0–1
—
5
3
3
3
1.67
1.00
Biantennary
0–1
—
4
2
3
2
1.33
1.00
Man5
—
—
2
5
0.40
Man6
—
—
2
6
0.33
Man7
—
—
2
7
0.29
Man8
—
—
2
8
0.25
Man9
—
2
9
0.22
3
0.8
18
14
9
11
The fact that approximately 1 mole per mole of GalNAc is detected for rhEPO indicates the presence of a single O-linked glycosylation site. Accordingly, 1 mole per mole of galactose should be attributed to the O-glycan (Section 23.2.2), which means that 13 mole per mole of galactose should still remain for the three tetraantennary N-glycans. The detection of GalNAc generally indicates the presence of O-glycans, because GalNAc is only in exceptional cases part of N-glycans. In the case of positive GalNAc findings, the presence of an O-glycan should be confirmed using another method. On the other hand, a negative GalNAc result always proves the lack of any O-glycosylation. Analysis of Sialic Acid Assessment of the sialylation status is of great interest, especially with biopharmaceutical glycoproteins such as EPO, because this determines the half-life of the glycoprotein in the blood circulation and thus its biological activity. Incompletely sialylated glycoproteins are recognized by receptors in the liver and removed from the blood circulation (clearance), and their biological activity is thus reduced. The production of therapeutic glycoproteins therefore demands confirmation of the intact and consistent glycosylation and sialylation from batch to batch for pharmacological safety reasons. In the thiobarbiturate method according to Warren (the so-called Warren test) the sialic acid residues are released from the glycoprotein by acid hydrolysis, oxidized by periodate, and stained by the addition of thiobarbituric acid. The color may likewise be generated by the reaction of neuraminic acid with resorcinol. The resulting colored complex is extracted from the aqueous medium with cyclohexanone or n-butanol/butyl acetate (1 : 5). In both cases the result is quantified via sialic acid standards carried in the reaction. The absorption of the respective color complex measured at 549 and 580 nm is, over a certain concentration range, proportional to the sialic acid content of the sample. Nowadays, these two staining methods have been largely replaced by the simpler and clearer analysis via HPAEC-PAD (see above). The bound sialic acid residues are liberated from the glycoprotein by mild acid hydrolysis (e.g., in 0.1 N H2SO4 or 2 N acetic acid for 0.5–1 h at 80 °C), or by incubation with the enzyme neuraminidase. The actual analysis is then performed via HPAEC-PAD, which not only enables the exact quantification using a calibration curve but also a simple differentiation between N-acetylneuraminic acid (Neu5Ac) and N-glycolylneuraminic acid (Neu5Gc) (structural formulas Figure 23.3). Compared to the N-acetyl group, the N-glycol group carries an additional OH group that is acidified by the adjacent CO group. This causes for Neu5Gc an enhanced interaction with the anion-exchange matrix and thus a greatly increased retention time. Neu5Gc does not occur in natural human glycoproteins, but does, however, for example, in bovine glycoproteins or monoclonal antibodies derived from mouse cell lines. The NeuGc content of human glycoproteins recombinantly expressed in animal cells is very low (in the case of rhEPO (CHO) less than 3%); nevertheless, for biopharmaceutical glycoproteins this provides an important quality parameter that needs to be controlled from batch to batch. Under the alkaline conditions of routine HPAEC-PAD, O-acetylated neuraminic acid residues lose their O-acetyl groups (ester hydrolysis). Its detection therefore requires neutral or weakly acidic eluents or (better) a mass spectrometric analysis of the corresponding glycopeptides (an example is given below in Figure 23.22c).
2.0
0.8
Man/Fuc
3.0
The data of a monosaccharide components analysis should not be overestimated! In general, the situation will be much more complicated, and usually no information on the nature of the oligosaccharide side chains is obtained from monosaccharide composition analysis.
588
Part III: Peptides, Carbohydrates, and Lipids
23.3.2 Mass Spectrometric Analysis on the Basis of Glycopeptides
Mass Spectrometry, Chapter 15
Figure 23.20 Amino acid sequence of EPO with GluC cleavage products E1–E13 (N-glycosylation at Asn24, Asn38, and Asn83; O-glycosylation at Ser126; disulfide bond in the two peptides E1 [(1–8)SS(160–165)] and E5).
Figure 23.21 Total ion current (negative ion mode) of the (glyco)peptides of an endo-GluC digest of EPO (CHO) separated by means of RP-HPLC/ESI-MS. The peaks are detected as single or double charged species. Peaks c, d, g, and i correspond to the glycopeptide fragments of the four glycosylation sites; the observed peak broadening or peak splitting results from the microheterogeneity of glycosylation. Source: according to: Kawasaki, N. et al. (2000) Anal. Biochem., 285, 82–91. With permission, Copyright 2000 Academic Press.
Direct mass spectrometric characterization of a glycoprotein is complicated by the microheterogeneity of glycosylation. Multiply and heterogeneously glycosylated proteins (such as EPO) lead to a broad, unresolved mass peak. However, on the basis of glycopeptide fragments, mass spectrometric methods (MALDI-MS Section 15.1, ESI-MS Section 15.2) in combination with HPLC are essential. They enable quick insight into the glycan composition per glycosylation site (microheterogeneity) and provide important sequence information via fragmentation experiments (MS/MS) (Section 23.3.4). For this purpose, the glycoprotein to be analyzed is digested with an appropriate protease (e.g., trypsin, chymotrypsin, endoproteinase LysC, endoproteinase GluC; cf. Section 9.5.1). Direct characterization of the (glyco)peptide pool via MS, however, is limited by a masking effect (ion suppression) the peptide signals exert on the glycopeptide signals. Therefore, the (glyco)peptide pool is usually separated via RP-HPLC and the individual peaks are analyzed via mass spectrometry, either online (LC/MS) or, after fractionation, offline (MALDI-MS, ESIMS). MS analysis of glycopeptides plays an important role, particularly in the characterization of O-glycans, since the free O-glycans are not readily accessible as such (Section 23.2.2). Trypsin cleaves peptide bonds of polypeptides C-terminal from lysine (K) and arginine (R). If, for example, EPO is digested with trypsin, two glycosylation sites (Asn24 and Asn38) remain on one and the same fragment (T5, amino acid residues 21–45). Endoproteinase GluC cleaves peptide bonds each C-terminal from glutamate (E) and occasionally aspartate (D). If you digest EPO with endo-GluC, the three N-glycosylation sites and the Oglycosylation site will be located on four different fragments (E5, E6, E10, and E12, Figure 23.20). The resulting 13 endo-GluC fragments can be separated via RP-HPLC (Figure 23.21) and allow the determination of the site-specific glycosylation (glycosylation per glycosylation site). Mass spectrometric analysis of the (glyco)peptides separated via RP-HPLC (see example of EPO, Figure 23.21) allows identification of the various glycosylation sites: The mere peptides can be assigned by comparing the experimental masses to the masses calculated from the respective amino acid sequence; the non-attributable peptide masses refer to an existing glycosylation (see endo-GluC digest of EPO, Table 23.5). The mass spectra of the glycopeptides particularly enable us to deduce the glycosylation per glycosylation site, which allows, via the molecular masses detected, assignment of the sitedirected microheterogeneity (see example of EPO, Figure 23.22). Thus, the mass analysis of
23 Carbohydrate Analysis
589
Table 23.5 Theoretical and experimental masses of EPO GluC peptides; fragments E5∗ (Asn24), E6∗ (Asn38), E10∗ (Asn83) and E12∗ (Ser126) are glycosylated. Source: according to Kawasaki, N. et al. (2000) Anal. Biochem., 285, 82–91. Peak in Figure 23.21
Glu-C peptide number
A
E8
B
E2
Amino acid residue (cf. Figure 23.20)
Average theoretical mass
M
56–62
729.4
728.4
9–13
602.4
601.5
m/z 1-
C
E6
∗
38–43
D
E5∗
22–37
E
E1
(1–8)S-S(160–165)
1502.7
1502.1
F
E3
14–18
692.1
691.5
G
E12∗
108–136
H
E7
44–55
1571.8
1571.1
63–72
1114.6
1113.9
I
E10
J
E9
∗
M2-
750.6
73–96
K
E13
137–159
2835.6
L
E11
97–107
2211.3
1417.0 2210.6
Figure 23.22 ESI-MS mass spectra of the four glycopeptides of GluC digestion of EPO (CHO) from Figure 23.21: peak c (a), peak d (b), peak g (c), and peak i (d). The assigned N-glycan structures are shown in Table 23.6. Source: according to: Kawasaki, N. et al. (2000) Anal. Biochem., 285, 82–91. With permission, Copyright 2000 Academic Press.
590
Part III: Peptides, Carbohydrates, and Lipids Table 23.6 Theoretical and experimental masses of EPO GluC glycopeptides; E5 (Asn24), E6 (Asn38) and E10 (Asn83) (peaks c, d and i of Figure 23.21). Source: according to Kawasaki, N. et al. (2000) Anal. Biochem., 285, 82–91. Ion in Figure 23.21 c
Amino acid residue (cf. Table 23.5)
N-Glycan structure (each with core-fucose)
E6 (38–43)
Asn38
Average theoretical mass
M
m/z 2-
c1
BiLac1NA2, TriNA2
3375.2
1686.8
c3
TriNA3
3666.5
1831.8
c4
BiLac2NA2, TriLac1NA2, TetraNA2
3740.6
1869.1
c6
TriLac1NA3, TetraNA3
4031.8
2014.5
c8
TetraNA4
4323.1
2160.2
TetraLac1NA4
4688.4
2343.2
c10 d
E5 (22–37)
M3-
1561.8
Asn24
d1
BiNA1
3750.7
1875.1
d2
BiNA2
4042.0
2020.3
1346.2
d4
BiLac1NA2, TriNA2
4407.3
2202.7
1468.2
d6
TriNA3
4698.6
2348.7
1565.6
d7
BiLac2NA2, TriLac1NA2, TetraNA2
4772.6
2385.7
1590.7
d8
TriLac1NA3, TetraNA3
5063.9
1687.3
d10
TetraNA4
5355.2
1784.5
TetraLac1NA4
5720.5
1905.7
d13 i
E10 (73–96)
Asn83
i1
BiLac1NA2, TriNA2
5388.5
1795.1
i3
TriNA3
5679.8
1892.2
i4
BiLac2NA2, TriLac1NA2, TetraNA2
5753.9
1917.6
i5
TriLac1NA3, TetraNA3
6045.1
2014.2
i7
TetraNA4
6336.4
2111.1
i10
TetraLac1NA4
6701.7
2233.4
i13
TetraLac2NA4
7067.1
2355.8
HPLC peak c of Figure 23.21 (GluC cleavage product E6 of EPO with glycosylation at Asn38) reveals 11, peak d (E5 with Asn24) 13, and peak i (E10 with Asn83) 14 different glycostructures, the major masses of which are shown in Table 23.6. Thereby, the individual ion peaks may cover different glycan isomers that cannot be identified in this glycopeptide analysis, but may only be assigned on the basis of the free N-glycans (cf. Section 23.3.5). Only for peak g (E12 with O-glycosylation at Ser126) do the m/z values refer to the structures NeuAc-GalGalNAc and NeuAc2-Gal-GalNAc (Figure 23.22c).
23.3.3 Release and Isolation of the N-Glycan Pool The procedure for glycan release depends on the glycoprotein to be analyzed and the degree of the desired information. Depending on the target, different enzymatic as well as chemical methods may be applied. Thus, the asparagine-linked sugar chains of glycoproteins may be released by glycopeptidases (PNGase F, PNGase A) or endoglycosidases (Endo-H, Endo-F) or by heating in anhydrous hydrazine for several hours. The N-glycans separated from the peptide backbone may be analyzed in chromatographic or electrophoretic manner or by MS or a combination of these methods.
23 Carbohydrate Analysis
591
Note: For the release of O-glycans, there is currently only a single enzyme, O-glycanase from Diplococcus pneumoniae – but this enzyme cleaves, because of its high substrate specificity, only Galβ1,3GalNAcα-Ser/Thr. In addition, an ideal chemical release process does not exist. However, Oglycans can usually be released by reductive alkali-catalyzed β-elimination (alkali borohydride reaction); the free glycans are reduced in situ to avoid the formation of degradation products. But this reaction is not specific so that in parallel 10–20% of the available N-glycans will be released. Thereby, the free glycans lose their reactive reducing ends, so that these are no longer available for the easy introduction of a UV or fluorescent label. However, the reduced O-glycans may be easily identified by mass spectrometry, so that their analysis today is usually carried out by MS.
The enzyme peptide-N4-(N-acetyl-β-glucosaminyl) asparagine amidase (PNGase F from Flavobacterium meningosepticum) practically cleaves all Nglycosidically linked sugar chains (except those bound to the amino or carboxy terminus of a polypeptide). (Also excluded are those N-glycans that at the proximal GlcNAc carry α1,3 (instead of α1,6)-linked fucose, as found in plant and insect cells but not in animal cells; such N-glycans are released by PNGase A). However, the enzymatic process sometimes requires optimized reaction conditions – such as a prepend proteolytic digestion of the glycoprotein (e.g., with trypsin or chymotrypsin). The peptide fragments thus formed are conformationally more flexible, and thus enable easier access of the enzyme to the individual glycosylation sites. The addition of an appropriate detergent (e.g., Triton X-100, Tween 20, CHAPS, or sodium dodecyl sulfate) facilitates the cleavage of the sugars (the detergent will unfold the protein and thus likewise facilitate the access of the enzyme to the glycosylation sites). PNGase F cleaves the N-glycosidic bond between the “proximal GlcNAc” (the anchor sugar) and the amino acid asparagine (Figure 23.23). While here the N-glycans are released in unaltered form, the anchor molecule asparagine (Asn) will be modified to aspartic acid (Asp). The removal of the sugar residues by PNGase F will usually cause a decrease in molecular weight, which is evident in SDS-PAGE (see example of EPO, Figure 23.15). The released sugar chains can be extracted with (freezer-cooled) ethanol (10%). Phenolic extraction should be repeated to obtain complete denaturation of the proteins until no interphase is detectable. Since phenol is in part soluble in water, the phenol dissolved in the aqueous phase is removed by extracting the aqueous phase with a chloroform/isoamyl alcohol mixture. For purification of RNA, the phenol should be buffered in water or low pH buffers (“acidic phenol”). DNA contaminants are much more soluble in acid phenol and can be removed more efficiently. The nucleic acids purified by phenolic extraction can be precipitated by ethanol as described below (Section 26.1.3).
26.1.2 Gel Filtration Gel Filtration/Permeation Chromatography, Section 10.4.1
Reversed Phase Chromatography, Section 10.4.2
Gel filtration methods can also be used for the purification of DNA (and RNA) solutions (the most common are Sephadex G50 or G75 and Bio-Gel P2). The purification effect is based on size exclusion allowing the separation of certain nucleic acid contaminations. Huge DNA molecules elute much faster (usually in the void volume of the column) than smaller, low molecular weight contaminants that become trapped in the pores of the column material and therefore elute at later time-points of the chromatography. The nucleic acid containing solution is loaded onto the gel filtration column and the column is eluted with buffer. The eluate is collected in fractions and can be tested for nucleic acid content. Gel filtration columns can be self-made (Figure 26.2) using a glass Pasteur pipette, which can be sealed with glass wool or a glass beads. When purifying only very small amounts of DNA, the glass pipette should be treated with silane prior to purification as DNA sticks to the glass. Many companies offer suitable gel filtration or purification columns at relatively low cost. Loading volume and elution volume are predefined by the supplier. In contrast to regular gel filtration columns, spin columns use centrifugal forces to apply and elute the fractions, and are therefore much faster. Instead of gel filtration, the principle of reversed-phase chromatography can also be applied. The nucleic acid solution is loaded onto the column material at low salt concentrations and is eluted with high salt buffer (e.g., Elutip columns).
26
Isolation and Purification of Nucleic Acids
667
Figure 26.2 Gel filtration for purification of DNA solutions. (a) Gel filtration columns can be made out of Pasteur pipettes and are filled with equilibrated column material. Depending on the size of the DNA, Sephadex G50, Sephadex G25, or Sephacel materials are used. (b) Using the molecular sieve effect small molecules, contaminating nucleotides, or salts are withheld, while large DNA molecules are not detained and will elute first from the column. A tentative column profile is depicted. The fractions containing the DNA can be analyzed by OD determination, ethidium bromide staining, or when purifying radioactive DNA by radiation analysis. The maxima are closer together when small DNA molecules are purified.
26.1.3 Precipitation of Nucleic Acids with Ethanol The most common method for the concentration and further purification of nucleic acids is by precipitation with ethanol. In the presence of monovalent cations, DNA or RNA form an ethanol-insoluble precipitate that can be isolated by centrifugation. Monovalent cations are usually supplied by addition of sodium acetate or ammonium acetate. Ammonium acetate is used to reduce co-precipitation of free nucleosides. However, ammonium ions can inhibit the activity of certain enzymes, for example, T4 polynucleotide kinase. For some applications, RNA is precipitated in the presence of lithium chloride. Lithium chloride is soluble in ethanol and will not be precipitated together with the nucleic acids. Chloride ions can act as inhibitors for several reactions, therefore precipitation with chloride ions should only be used in certain circumstances. In the laboratory, the nucleic acid solution is adjusted to the desired salt concentration using a higher concentrated stock solution of, for example, sodium acetate. To this solution the 2.5–3fold volume of ethanol is added and incubated, depending on the nucleic acid to be precipitated, at room temperature or 80 °C and centrifuged (Figure 26.3). The precipitated salt can be
Figure 26.3 Precipitation of nucleic acids with ethanol. (a) To the aqueous nucleic acid solution a 2.5-to 3-fold volume of absolute ethanol (or in the case of isopropanol 0.5–1-fold volume) is added and the nucleic acids are precipitated by centrifugation. A colorless pellet is usually visible on the bottom of the tube. (b) High molecular weight genomic DNA can be precipitated on the phase interface by cautious overlay of the aqueous phase with ethanol. The DNA can be visualized by winding it on a sterile rod.
668
Part IV: Nucleic Acid Analytics
The carrier has to be chosen so that the material does not interfere with the subsequent reactions or applications. For example, tRNA will also be phosphorylated by T4 polynucleotide kinase and should not be used as carrier if the subsequent reaction involves phosphorylation reactions. Glycogen can interact with DNA–protein complexes.
removed by washing the pellet with 70% ethanol. In contrast to the precipitated DNA, most salts are soluble in 70% ethanol. The nucleic acid pellet is dried briefly and re-solved in buffer or water. Nucleic acids can also be precipitated by the addition of 0.5–1 volume of isopropanol. This protocol is advantageous if the volume of the reaction should be kept to a minimum. Sodium chloride is precipitated better in isopropanol. Isopropanol is less volatile than ethanol; the nucleic acid pellet needs to be washed carefully with 70% ethanol. Precipitation of Small Amounts using Carrier Material As very low concentrations of RNA or DNA (150 kb) is not precipitated but purified using dialysis or extraction with 2-butanol. Extremely high molecular weight DNA can be analyzed by melting the cells or tissue to be analyzed into agarose and all the following steps are then performed in these agarose blocks. Lysis of Cell Membranes and Protein Degradation A fundamental step during isolation of genomic DNA is the proteolysis of cellular proteins by proteinase K. Simple extraction of the proteins using phenol is not sufficient. In addition, genomic DNA is complexed with histones and histone-like proteins, which cannot be removed completely by phenolic extraction. The optimal incubation temperature for proteinase K is between 55 and 65 °C. The enzymatic performance is optimal at 0.5% sodium dodecyl sulfate (SDS). Incubation of the starting material with proteinase K containing buffer is in many cases sufficient to disrupt and lyse the cell membranes. The addition of RNAse removes the contaminant RNA. In many cases, the cells or tissue need to be disrupted mechanically before protease digestion. Homogenizers disrupt the cells with blades that rotate at high frequencies. A so-called French press or ball mills can also be used. A French press uses high pressure to disrupt the cells. Ball mills contain very fast moving small beads made of glass or steel. If the DNA is obtained from tissue, the tissue is shock frosted using liquid nitrogen and then pulverized to achieve a homogenous mixture. The lysis of bacterial walls is achieved by lysozyme and yeast cell walls are disrupted by zymolase or lyticase specifically degrading yeast cell walls. The enzymes are inactivated during protease K treatment. Table 26.2 Enzymes and lysis reagents for isolation of genomic DNA. Nucleic acid origin
Lysis by
Subsequent treatment
Eukaryotic cell cultures
Sodium dodecyl sulfate (SDS)
Proteinase K
Tissue
Sodium dodecyl sulfate/proteinase K
Proteinase K
Plants
SDS or N-laurysarkosin
Proteinase K
Yeast (Saccharomyces cerevisiae; Schizosaccharomyces pombe)
Zymolyase or lyticase
Proteinase K
Bacteria (Escherichia coli)
Lysozyme
Proteinase K
669
670
Part IV: Nucleic Acid Analytics
Purification and Precipitation of Genomic DNA
The proteinase K is inactivated and removed by phenolic extraction. The precipitation of genomic DNA after addition of ethanol can be easily observed: the DNA precipitates at the interphase between the water and ethanol and can be rolledup as filaments on a sterile rod (Figure 26.3). Genomic DNA should be dried very cautiously to avoid solubility problems later on. The genomic DNA can be dissolved by incubation for several hours at 4 °C. Genomic DNA isolated by ethanol precipitation is of sufficient purity for most applications. Phenolic extraction and subsequent ethanol precipitation will result in an average molecular size of around 100–150 kb. This size is sufficient for the generation of DNA libraries using the bacteriophage λ-vectors and for Southern blot analysis. The construction of cosmid libraries requires DNA fragments of at least 200 kb, and so extraction with organic solvents cannot be used. The proteinase K and remaining proteins are denatured using formamide and removed by dialysis using collodion bags. This method avoids shear forces and results in high molecular DNA fragments (>200 kb). A fast and easy isolation method is the lysis of cells and denaturation of the proteins with guanidinium hydrochloride. The DNA is isolated by ethanol precipitation. This method yields genomic fragments with an average size of 80 kb and can be used for Southern blot or PCR analysis. Additional Purification Steps Genomic DNA can be purified using CsCl gradient centrifugation. During centrifugation RNA contaminations are pelleted and removed completely. The CsCl density gradient centrifugation is described in Chapter 1 and in Section 26.3. Commercially available kits use the anion exchange method for the isolation and purification of genomic DNA. The kits are available with columns containing the anion exchange material using gravity or spin-columns using centrifugal forces. This purification method does not need any organic extraction; however, the DNA is subject to shear forces so that high molecular weight DNA cannot be isolated using this method. The genomic DNA isolated by the anion exchange columns can be used for Southern blot, PCR, and next generation sequencing. Polysaccharide contaminants can be removed by treatment with CTAB (cetyltrimethylammonium bromide, Figure 26.5). This purification step is essential when genomic DNA is isolated from plants or bacteria, as they contain high levels of polysaccharides. CTAB complexes polysaccharides and removes the remaining proteins. By addition of chloroform/ isoamyl alcohol the complexed polysaccharides are precipitated in the interphase. An important factor is the NaCl concentration: if the concentrations is below 0.5 M, the genomic DNA will also precipitate in the presence of CTAB.
Figure 26.5 CTAB (cetyltrimethylammonium bromide or hexadecyltrimethylammonium bromide). The quaternary ammonium salt can act as cationic detergent.
26.3 Isolation of Low Molecular Weight DNA 26.3.1 Isolation of Plasmid DNA from Bacteria Plasmids, that is, extrachromosomal mostly circular DNA occur naturally in microorganisms. Plasmids consist of 2 to more than 200 kb and fulfill various different genetic functions. In daily laboratory business, plasmids consist of defined genetic elements (replication origin, resistance gene, and polylinker – Figure 26.6). These so-called plasmid vectors are essential tools for a huge variety of applications. The methods described below deal exclusively with the isolation of bacterial plasmid vectors. Plasmids are grown in bacteria using antibiotic selection. The plasmids contain at least one selection gene for resistance to certain antibiotics, for example, the bla-gene coding for β-lactamase enables bacteria that carry the gene to grow in ampicillin-containing media. In addition, plasmids contain a bacterial origin of replication for propagation of the plasmid in the bacterium. The kind of origin of replication determines the copy number of a plasmid in the bacterium (Table 26.3.).
26
Isolation and Purification of Nucleic Acids
671
Table 26.3 Origins of replication of commonly used plasmid vectors and copy numbers. Plasmid
Origin of replication
Gene of resistance r
Copy number
r
pMB1
Amp , Tet
pUC
pMB1
Amp
r
500–700
pBluescript
pMB1
Ampr
300–500
pGEM
pMB1
Amp
r
300–400
pVL 1393/1392
ColE1
Ampr
pBR322 and derivatives
pACYC pLG338
15–20
>15 r
r
Chloramphenicol , Tet
p15A
r
Kan , Tet
pSC101
r
10–12 ca. 5
Plasmids are classified as low copy (copy number < 20) and high copy plasmids (copy number > 20). The copy number of the plasmids is a major determinant for the yield of a plasmid from a bacterial culture. Most plasmids contain a mutated version of the pMB1 origin of replication derived from ColE1 multi-copy plasmids of the Enterobacteriaceae family. The isolation steps can be divided into growth and lysis of the bacteria and isolation and purification of the plasmid DNA. Bacterial Culture For the isolation of plasmids, derivatives of the Escherichia coli strain K12 are used. The strain is considered biologically safe as it is missing pathogenic genes (e.g., factors relevant for adhesion and invasion, toxins, and certain surface molecules). Not all Escherichia coli K12 strains are equally useful for plasmid production. Good host strains are, for example, DH1, DH5a, and XL1 Blue. Certain strains like HB101 and JM100 express a high amount of endonucleases and carbohydrates that are detrimental to the plasmid isolation procedure. To avoid mutations and unwanted DNA recombinations, strains deficient of recombinase A (recA-), like XL1 Blue and its derivatives, are preferred. Bacteria are grown in liquid culture using autoclaved Luria-Broth (LB, contains yeast extract, Bacto tryptone, and sodium chloride) in the presence of the antibiotic. The amount and quality of the added antibiotic is important for the plasmid yield. Ampicillin is temperature sensitive and should not be added to hot autoclaved medium. According to good microbiological practice, the broth is inoculated with a single bacterial colony, which is first grown in a small volume of media and then diluted to the needed volume. According to the amount of culture volume the DNA preparations are classified as “mini-” (1–10 ml), “midi-” (25–100 ml), or “maxi-” (>100 ml) preparations. The yield of low copy plasmids can be increased by the addition of chloramphenicol to the media (see below). Chloramphenicol is also used for the isolation of high copy plasmids as it keeps the number of bacteria and thus the amount of bacterial debris low. Plasmid containing a ColE1 origin of replication can be amplified selectively compared to the bacterial genome. During the logarithmic growth phase an inhibitor of translation (e.g., chloramphenicol) is added to the bacterial culture. Chloramphenicol inhibits the synthesis of the Rop (repressor of primer) protein. This protein accounts for the control of the copy number of plasmids. Inhibition of this protein results in an increased replication of the plasmid (relaxed replication).
Lysis of Bacteria Many different methods are available for the lysis of the plasmid containing bacteria (Table 26.4). The method of choice depends on the type and use of the plasmid to be Table 26.4 Methods used to reveal bacteria. Method of digestion
Means of analysis
Comment
Alkaline
SDS/NaOH
Koch lysis
Lysozyme/100 °C
Quick and easy. The most suitable method for large plasmids and low-copy plasmids Endonuclease A is inactivated completely
Lithium method
LiCl/Triton X-100
Quick and efficient, not suitable for large plasmids (>10 kb)
SDS lysis
Lysozyme/SDS
Frequently used for large plasmids (>15 kb)
Figure 26.6 Composition of a typical plasmid vector for cloning and amplification of DNA fragments. The fragment of interest is cloned to the artificial multiple cloning site (MCS). T7 and T3 are promoters that are recognized specifically by RNA polymerases of the T7 and T3 bacteriophages and are used for RNA synthesis of the cloned fragments. AmpR depicts a gene for selection that renders bacteria containing the vector resistant to ampicillin: the bacteria can grow in ampicillin containing medium. The replication of origin (ori) is necessary for the autonomous replication of the plasmid. The ori region enables double-stranded replication of the plasmid whereas a second origin of replication (e.g., f1 single-stranded phages) permits singlestrand replication (ss-ori).
672
Part IV: Nucleic Acid Analytics
Figure 26.7 Principle of the alkaline lysis of bacteria for the isolation of plasmid DNA. (1) The bacteria are lysed using SDS and the DNA is denatured by NaOH. (2) The solution is neutralized by the addition of sodium acetate. Denatured proteins and chromosomal DNA are precipitated together with the potassium salt of the dodecyl sulfate. Low molecular plasmid DNA remains in solution and renatures. (3) The insoluble complexes are separated by centrifugation and the plasmid DNA can be isolated. Source: adapted according to: Micklos, D.A. and Freyer, G.A. (1990) DNA Science. A First Course in Recombinant DNA Technology, Cold Spring Harbor Laboratory Press and Caroline Biological Supply Company, Cold Spring Harbor.
It is important that the RNase A does not contain any contaminating DNases. This can be achieved by incubation of the RNase A solution at 95 °C. RNase H is a very stable enzyme that renatures after heat treatment to yield an active enzyme whereas DNases are permanently inactivated.
isolated. The most common method is alkaline lysis (Figure 26.7). The bacterial culture is centrifuged and the pellet resuspended in a buffer containing EDTA. EDTA complexes bivalent cations (Mg2+, Ca2+) that are important for the structural integrity of the bacterial walls. The buffer can also contain RNase A to degrade most of the bacterial RNA at this first step. The bacterial suspension is lysed completely by the addition of SDS and NaOH. SDS functions as detergent, solubilizing the phospholipids and proteins of the bacterial cell walls. Sodium hydroxide denatures proteins, chromosomal, and plasmid DNA. The timespan of the incubation of the solution under alkaline conditions is important for the quality of the plasmid DNA. Too long an incubation time leads to irreversible denaturation of the plasmid; too short an incubation results in incomplete lysis of the bacteria and low plasmid yield. Completely denatured plasmid DNA can be detected using gel electrophoresis on an agarose gel. Denatured plasmid DNA has a higher mobility than superhelical plasmid DNA and is stained less by ethidium bromide. The lysate is neutralized with potassium acetate buffer. Potassium dodecyl sulfate has a much lower solubility in water than sodium dodecyl sulfate and precipitates at high salt concentrations present in the lysate. Denatured proteins, high molecular weight RNA and denatured chromosomal DNA, and cellular debris form insoluble complexes in the presence of potassium dodecyl sulfate and will be co-precipitated with the potassium dodecyl sulfate. The smaller plasmid molecules remain in solution and renature upon neutralization of the solution. The insoluble debris is centrifuged and the supernatant can be processed further. For some applications the purity of the DNA solution is sufficient and the plasmid DNA can be precipitated with ethanol or isopropanol and washed with 70% ethanol.
26
Isolation and Purification of Nucleic Acids
This quick and easy method is useful for the preparation of many plasmids simultaneously and is used to check cloning efficiency. Many different single bacteria colonies are inoculated in a small volume of media and successful cloning is checked by digestion of the plasmid DNA with restriction enzymes if the plasmid contains the desired insert. Commercially available kits are based on the alkaline lysis principle. The plasmid DNA is purified as described below by anion exchange chromatography before precipitation with ethanol. As well as alkaline lysis, bacteria can be lysed thermally (boiling lysis). The bacterial cell walls are broken down by addition of lysozyme and lysed bacteria are heated for a short time. The debris is pelleted and the plasmid DNA can be isolated by ethanol precipitation. This method does not inactivate completely endonuclease A present in some E. coli strains (HB101, endA+). The plasmid DNA should therefore be purified by phenolic extraction prior to precipitation. Other (more uncommon) methods are incubation with the non-ionic detergent Triton X100 (Figure 26.8) in the presence of lithium chloride or lysis by SDS and lysozyme. The latter method (without the addition of NaOH) is used when high molecular plasmids need to be isolated. High molecular weight plasmids cannot be renatured completely in the presence of NaOH.
673
Figure 26.8 Non-ionic detergent Triton X-100.
Lysozyme: Abundant hydrolase found in saliva and tear fluid. Lysozyme hydrolyzes the 1,4-β-linkages between N-acetylmuramic acid and N-acetyl-D-glucosamine present in bacterial cell walls.
Purification of DNA by Anion Exchange Chromatography
In general, commercially available columns are used for purification by anion exchange. The positive charge is provided by protonated diethylammoniumethyl (DEAE) groups. The negatively charged DNA is bound at lower salt concentrations (750 mM) to the column material. Proteins and degraded RNA are not bound to the column material under these conditions. The column material is washed with buffer containing a higher salt concentration (1 M) to elute traces of bound protein or RNA. DNA does not elute from the column under these conditions. The DNA is eluted at even higher salt concentrations (1.25 M). The exact buffer conditions are dependent on the column material and supplier. Table 26.5 depicts an overview of expected yields using anion exchange columns. Several purification protocols have been set up for the removal of endotoxins prior to the purification by anion exchange. The lipopolysaccharides adherent to the bacterial membranes are treated with detergents (n-octyl-β-D-thioglucopyranoside, OSPG) to remove binding proteins. Then, the lipopolysaccharides are removed using columns loaded with polymyxin B. This antibiotic binds lipopolysaccharides very efficiently. Ultrapure DNA with very low toxin content can be obtained by repeated purification using a CsCl density gradient. Purification of DNA by Density Gradient Centrifugation
Ultrapure, high yield DNA can be obtained by centrifugation using a CsCl density gradient. Due to the significantly increased quality of the commercially available anion exchange kits, the method of density gradient centrifugation has lost its relevance and will therefore only be summarized briefly. The isopycnic centrifugation of DNA molecules within a CsCl density gradient is performed in the presence of ethidium bromide. The mechanism and thermodynamic aspects of ethidium bromide intercalation will be discussed in Chapter 27.2. Plasmid DNA and chromosomal DNA
Table 26.5 Approximate DNA yield after anion exchange purification. The yield of high copy plasmids is approx. 2–5 μg ml − 1, of low copy plasmids 0.1 μg ml − 1. Vector
Plasmid type
Bacterial culture (ml)
Yield (μg)
pUC, pGEM
High copy
25
50–100
pUC, pGEM
High copy
100
300–500
pBR322
Low copy
100
50–100
pBR322
Low copy
500
100–500
Anion Exchange Chromatography, Section 10.4.7
With the described purification method it is possible to co-purify certain lipopolysaccharides present in almost all Gram-negative bacteria. The presence of these so-called endotoxins is critical when transfecting DNA in sensitive cells or cell lines. Endotoxins can reduce transfection efficiency and can result in stimulation of protein synthesis or activation of the innate immune or the complement system.
674
Part IV: Nucleic Acid Analytics
Figure 26.9 Purification of plasmid DNA by CsCl density gradient centrifugation in the presence of ethidium bromide. Using vertical rotors, the gradient is generated parallel to the axis of the rotor. After stopping the rotor and centrifugation, the gradient tips but keeps its layers so that the plasmid is visible as circular bands. Due to the different densities, superhelical form I and forms II and III (open form, linear form) can be separated within the gradient. RNA–ethidium complexes are pelleted on the wall of the centrifuge tube. To isolate the plasmid, the sealed tube is ventilated using a needle. The plasmid is obtained by puncturing the tube with a second cannula below the plasmid band. The DNA containing solution is aspirated with a syringe. If genomic DNA was present, it can be found above form I because of its lower density. For ultrapure DNA, the CsCl density gradient purification is repeated.
can be distinguished by their different densities with intercalated ethidium bromide. Ethidium bromide intercalates into double-stranded DNA, preferentially linear or nicked plasmid DNA, and to a lower extent into covalently closed-circular plasmid DNA. The resulting differences in density are used to separate the different molecular forms of DNA (Figure 26.9). The buoyant density of RNA is higher than the maximal density of CsCl, resulting in RNA/Ethidium bromide pellets. RNA separation is achieved using Cs2SO4. The ethidium bromide is removed by repeated extraction of the DNA solution with n-butanol. Any remaining traces of ethidium bromide are completely removed by phenolic extraction. The high concentration of CsCl can be removed by dialysis of the DNA against TE buffer or water. The DNA can also be diluted to low concentrations and then precipitated using ethanol.
26.3.2 Isolation of Eukaryotic Low Molecular Weight DNA Yeast Plasmids The isolation of ultrapure yeast plasmids is difficult. In practice, total DNA is isolated. Since yeast plasmid contains a yeast origin of replication and a bacterial origin of replication pure yeast plasmids are isolated after re-transformation into E. coli as contaminating chromosomal yeast DNA cannot be replicated in bacteria. Hirt Extraction Low extrachromosomal DNA like plasmid DNA or viral DNA from cell or tissue cultures is isolated using a protocol established by B. Hirt 1967 for the isolation of polyomavirus DNA from murine cells. The cells are lysed using 0.5% SDS and adjusted to 1 M NaCl. The mixture is incubated over night at 0 °C and centrifuged. The supernatant contains low molecular weight DNA and can be purified using proteinase K and phenolic extraction.
26.4 Isolation of Viral DNA 26.4.1 Isolation of Phage DNA Bacteriophage λ and others are widely used as vectors for phage display, as reporter phages, or for cloning of genomic libraries. No other cloning system allows insertion of high molecular weight DNA fragments (10–20 kb); it can also be used very conveniently for high throughput screening. It may be necessary to isolate and analyze the DNA fragment inserted into the phage genome.
26
Isolation and Purification of Nucleic Acids
Proliferation of Phages Bacteriophages are proliferated using liquid culturing of E. coli. The choice of host strain is dependent on the bacteriophage strain. The bacteria are grown in the log phase using maltose. Maltose induces expression of the bacterial receptor (lamB) for bacteriophage λ. The bacteria are harvested and the culture is adjusted to a certain density using a buffer containing Mg2+ (λ-diluent, SM media). The bacterial cell number is determined photometrically: absorption of the culture is measured at 600 nm (blank is pure media). 1 OD equals approx. 8 × 108 bacteria. The Mg2+ ions stabilize phage particles and Mg containing media is used for phage proliferation (NZCYM media). For optimal proliferation of bacteriophages, the initial ratio of phages to bacteria is important. If the number of phages outweighs the number of bacteria in the initial phase, the bacteria are lysed completely and no phage proliferation can occur, and the yield will be very low. If the bacterial culture is infected initially with a low number of phages, the bacteria will overgrow a possible phage infection and the phage yield is also low. The initial ratio needs to be determined for phage and bacterial strain. A good ratio is found when complete lysis takes more than 8 h. Complete lysis of a bacterial culture can be detected by the sudden clearance of the turbid bacterial culture and the sudden occurrence of lysed bacterial debris.
Isolation of Phage Particles To remove bacterial RNA and DNA, RNases and DNase is added to the bacteriophage cultures. The phage DNA will remain intact, protected by the intact phage capsid. The phage particles are isolated by ultracentrifugation (100 000g). The purity of the phages is, for most applications, sufficient. Phages form a colorless to light brown pellet that is resuspended in TE. Phage particles are very sensitive to complexing agents that decrease the Mg ion concentration. Resuspension of the pellet in EDTA containing buffers destabilizes the capsid and facilitates later lysis. In some protocols the particles are precipitated using poly(ethylene glycol) (PEG). If a higher purity of the phages is desired they can be purified by CsCl density gradient centrifugation. Isolated and purified phage particles are lysed by proteinase K and the protein components of the capsid are degraded. The DNA is purified by phenolic extraction or anion exchange chromatography (Section 26.1). Phage DNA is a high molecular linear DNA (45–50 kB) that should be handled with care. For many applications, phage DNA can also be obtained with commercial kits and the needed fragments amplified by PCR.
26.4.2 Isolation of Eukaryotic Viral DNA The diversity of eukaryotic viruses requires several additional adapted strategies for isolation of their nucleic acids. Two general purification methods can be distinguished. In infected cells the viral DNA is present as extrachromosomal DNA (e.g., adenoviruses, polyomaviruses, SV40, papillomaviruses, baculoviruses). The viral DNA can be isolated from the infected cells via Hirt extraction (Chapter 26.3.2) in high yield and sufficient purity for many applications. Because some viruses contain high molecular weight DNA the same precautions should be taken as with any other high molecular weight DNA. The Hirt extraction method does not yield highly pure virus DNA and the viral DNA might be modified differently than in the virus particle (proteins bound, covalent modifications, circular or non-covalently closed DNA). Highly pure, native viral DNA can be obtained by purification of the viral particles. In most cases, infected cell release newly synthesized virus particles into the medium. The viral particles can be pelleted by ultracentrifugation (approx. 100 000g) and purified using CsCl gradient centrifugation. The viral shell is lysed specifically depending on the virus type. Usually, viruses are incubated with proteinase K followed by phenolic extraction. Using this method, proteins bound to viral DNA, like the terminal protein bound to adenoviral DNA or chromatin-like structures of Polyoma or SV40 nucleic acids, are destroyed. In some cases, mild alkaline lysis is sufficient to isolate native viral DNA. Commercial kits use silica-membrane based matrices or anion exchange chromatography. The viral DNA is isolated using cell free liquids (supernatants of blood plasma) as the methods do not allow the separation of viral and cellular DNA. Using spin columns or 96-well filter plates blood or samples can be analyzed by (RT-)PCR for the presence of viral DNA (or RNA), for example, of HBV, HCV, and HIV on a high throughput basis.
675
DNA library (genomic or cDNA library): Genomic DNA libraries contain the whole genome of an organism but split into smaller fragments that can be handled and cloned. The genome is fragmented enzymatically and then cloned into suitable vector systems, like the bacteriophage λ genome. Beside the genomic libraries, cDNA libraries represent the mRNA spectrum of a cell or organism. The cDNA is generated by reverse transcription of the mRNA.
676
Part IV: Nucleic Acid Analytics
26.5 Isolation of Single-Stranded DNA 26.5.1 Isolation of M13 Phage DNA Filamentous phages like M13, f1, or fd possess single-stranded covalently closed circular DNA (approx. 6.5 kb). Cloning of foreign DNA into the phage genome allows the isolation of singlestranded DNA of the desired sequence with high yield. The phage M13 infects exclusively E. coli (e.g., JM109, JM197) by intrusion using the sex pili of the bacteria that are coded on the F episome. The phage is converted into the replicative form (RF), a double-stranded version of the phage. Infection of the bacteria with M13 does not result in lysis as with bacteriophage λ infections but only in diminished growth rates. The single-stranded version of the M13 genome is isolated by isolation of the phage particles; the replicative form can be purified from the bacterial pellet. M13 phages are isolated by poly(ethylene glycol) precipitation or by anion exchange chromatography. Commercial kits are available for the isolation of M13 to allow high throughput isolation. The purification is based on silica-gel membranes. At high salt conditions, single-stranded DNA binds with higher affinity to this material than do double-stranded DNA or proteins.
26.5.2 Separation of Single- and Double-Stranded DNA Single-stranded and double-stranded DNA can be separated from a complex mixture by hydroxyapatite chromatography. Hydroxyapatite, a crystalline form of calcium phosphate (Ca5(PO4)3(OH)), is bound preferentially by double-stranded DNA and with much lower affinity by single-stranded DNA or RNA. Binding of double-stranded DNA is performed using phosphate containing buffer at high temperatures (60 °C). At these conditions the singlestranded DNA does not bind to the column and will be found in the void volume. The doublestranded DNA can be eluted by increasing the phosphate content of the buffer. A critical factor of this purification method is the high phosphate content of the obtained nucleic acids as this interferes with nucleic acid precipitation. The fractionated nucleic acids are concentrated first with sec-butanol and then desalted by gel filtration.
26.6 Isolation of RNA Working with RNA requires even more care than working with DNA. RNases are in contrast to DNases very stable, do not need any co-factors, and cannot be inactivated completely by autoclaving. Only ultrapure buffers should be used for isolation of RNA. RNases can be inactivated by treating the buffers with diethyl pyrocarbonate (DEPC) (Figure 26.10). DEPC inactivates RNAses by covalent modification of the histidine residue in the active center of the enzyme. Buffers containing free amino groups cannot be treated with DEPC. DEPC is toxic due to its modifying properties. Excess DEPC needs to be inactivated by autoclaving as DEPC present during RNA purification will modify the bases of the RNA (carboxyethylation of the adenines and, seldom, guanines). DEPC is degraded to carbon dioxide and ethanol. The use of gloves and sterile, RNase free plastic ware is essential for handling RNA. Glass ware should be decontaminated by heat treatment at 300 °C. For many experiments, RNase inhibitors can be added but these inhibitors can only inactivate low contents of RNases (Table 26.6). In addition
Figure 26.10 Chemical formula and mechanism of DEPC (diethyl pyrocarbonate) treatment. DEPC inactivates RNases by covalent modification of amino groups and histidines. DEPC degrades to ethanol and carbon dioxide upon heating and autoclaving.
26
Isolation and Purification of Nucleic Acids
Table 26.6 Frequently used RNase inhibitors. RNase inhibitor RNasin
Diethyl pyrocarbonate
Vanadyl ribonucleoside complexes SDS, sodium deoxycholate
Protein from human placenta Forms non-covalent equimolar complexes with RNases Cannot be used at denaturing conditions Used for buffer treatment Covalent modification Needs to be inactivated Transition state analog that binds RNases and inhibits their activity Cannot be used for cell free translation systems Denaturation
β-Mercaptoethanol
Reduction
Guanidinium thiocyanate
Used in connection with cell lysis Denatures RNases reversibly Used in denaturing agarose gels Covalent modification
Formaldehyde
to these commonly used RNase inhibitors, optimized protein- or antibody-based inhibitors specific for certain RNases are commercially available.
26.6.1 Isolation of Cytoplasmic RNA In contrast to DNA localized in the nucleus, most RNA molecules are located in the cytoplasm. Cytoplasmic RNA is composed of various RNA species, such as classical, long-known ribosomal RNA, transfer RNA, and messenger RNA. With recent new technologies like deep sequencing and tiling arrays, new RNA species have been identified. A majority of the human genome is transcribed while only an estimated 2% of these RNAs will be translated to proteins. The so-called non-coding RNAs constitute a new group of RNA molecules with various functions, many of which have yet to be discovered. Non-coding RNAs with sizes above 200 nt are classified as long ncRNAs, whereas miRNA (micro), piRNA (PIWIinteracting), and siRNA (small interfering) belong to the group of small ncRNAs. Due to the very heterogeneous nature of the RNAs, various isolation and purification protocols and commercial kits are available. For some applications, like Northern blots or RTPCR or ribonuclease protection assays, the isolation of cytoplasmic RNA is sufficient. Minor contamination of the RNA preparation with genomic DNA can be excluded by the use of proper controls. A RT-PCR reaction, for example, can be performed without the use of reverse transcriptase. The PCR result should be negative if no genomic DNA is present. In the presence of contaminating genomic DNA the PCR will be positive even without reverse transcriptase treatment. The use of intron/exon spanning primers is recommended. With these primer pairs, only spliced mRNA will yield fragments of the correct size. For some applications it can be useful to enrich or purify the mRNA out of the total RNA. Cultivated Cells The plasma membranes of the cells are lysed with a non-ionic detergent (Nonidet P40) while keeping the cell nuclei intact. The nuclei are separated and the proteins in the cytoplasmic fraction are degraded using proteinase K. The RNA can be purified by phenolic extraction. If cells have been transfected with plasmid DNA, the cytoplasmic RNA can be contaminated with episomal DNA, which can be removed by digestion with RNase-free DNases. Tissue and Cultivated Cells The nuclease activity of a tissue can be very high. Therefore, the tissue is frozen immediately in liquid nitrogen. Cells are lysed and proteins completely denatured using the chaotropic salt guanidinium thiocyanate. β-Mercaptoethanol and L-lauryl-sarcosine (Figure 26.11) are added to prevent degradation of the RNA. Cells or tissue are also often lysed using phenol. Since RNases are not completely inactivated by pure phenol, a mixture of acidic phenol: chloroform: isoamyl alcohol (Section 26.1) is used. Most methods and kits use the combination of both reagents for efficient and more convenient denaturation of proteins and inactivation of RNases. The RNA can be purified using anion exchange in a similar manner to DNA purification. For RNA, adapted buffer conditions are used to bind and elute the RNA from the column.
677
678
Part IV: Nucleic Acid Analytics
Figure 26.11 (a) N-Lauryl-sarcosine and (b) guanidinium thiocyanate.
Contaminating DNA can be removed by digestion (also on-column) with DNases. RNA can also be purified using CsCl density gradient centrifugation. RNA–ethidium bromide complexes pellet due to their higher density and can so be separated from genomic DNA. If the RNA needs to be purified as a band, higher density gradients (using Cs2SO4) need to be performed (Section 26.3.1). Most commercial kits are based on silica technology use solid phase extraction (SPE). RNA (and also DNA) can be bound to filters or columns consisting of silica particles, glass fibers, or glass beads (Section 26.7). The RNA is bound to the silica-material in the presence of high-salt chaotropic buffers (in most cases this buffer is provided during lysis of cells using guanidinium thiocyanate buffer). The RNA is washed and eluted from the matrix with low salt buffers; in many commercially available kits, the eluent is RNase free water. These recent technologies enable researchers to obtain high quality RNA simply and quickly in high throughput quantities. All RNA isolation kit sellers offer specialized protocols and kits for all kind of RNA sources and applications. For some applications, for example, next generation sequencing, it is useful to remove the major part of the ribosomal RNA before sequencing. This will reduce material cost due to unnecessary sequencing of contaminating ribosomal RNA. Ribosomal RNA depletion kits are based on hybridization of the ribosomal RNA to oligonucleotide probes specific for ribosomal RNA. The hybridized ribosomal RNA:DNA strands are then bound to beads and removed from the solution, for example, by magnetic separation (Section 26.7).
26.6.2 Isolation of Poly(A) RNA Nearly all eukaryotic mRNA species contain long adenine-rich regions on their 3´ termini. These poly(A) tails are used to purify mRNA from cytoplasmic RNA. Column or bead material are coupled to single-stranded thymidine rich short DNA fragments (oligo(dT)). The poly(A) tails hybridize to the oligo(dT) strands and are bound to the column material (Figure 26.12). Contaminating non-poly(A) containing RNAs can be easily removed by washing the column.
Figure 26.12 Isolation of poly(A) RNA using a oligo(dT) column. Total RNA (cytoplasmic RNA) is loaded to the column. RNA with a poly(A) tail is bound by hybridization of the adenines to the oligo (dT) residues to the column whereas all other molecules are collected in the flow through. The poly(A) RNA is eluted using conditions that destabilize the dT:rA hybrids.
26
Isolation and Purification of Nucleic Acids
679
To ensure optimal hybridization and loading of the column, the starting RNA material needs to be denatured. For optimal yield, the starting material can be applied to the column several times. The poly(A)-RNA is bound to the column at high salt concentrations (500 mM NaCl or LiCl) and purified poly(A) RNA is eluted with water. These conditions destabilize the dT:rA hybrids. Low cost oligo(dT) columns can be prepared by coupling of oligonucleotides (dT12–18) to activated column material. For more convenience, commercial kits are available in different formats.
26.6.3 Isolation of Small RNA In recent years significant research has focused on small non-coding RNAs like miRNAs, siRNAs, or piRNAs with sizes lower than 200 nt. These RNAs are purified from tissues, cells, or extracellular vesicles like exosomes. Many of the RNA isolation and purification protocols developed for longer RNAs are not optimal for small RNAs, for example, ethanol precipitation is not very efficient for small RNAs and many protocols need to be adapted. It is important for a good recovery of the small RNAs to include an acid phenolic extraction at the beginning of the isolation protocol. Only if tissue or cells or exosomes are denatured completely using acidic phenol : chloroform : isoamyl alcohol is the yield of small RNAs sufficient. Individual purification protocols depend on the column material and kit used and specific enrichment of small RNAs can be achieved by a combination of different separation techniques and buffer conditions. Tailored isolation kits are accessible to the research community for the purification of small RNAs from different sources
26.7 Isolation of Nucleic Acids using Magnetic Particles In recent years, the demands on nucleic acid purification protocols have increased dramatically regarding speed, costs, yield, purity, and format. A lot of scientific questions require isolation of a huge number of samples simultaneously, for example, for expression profiling or SNP analysis (single nucleotide polymorphism). The development of automated high throughput isolation protocols was mandatory. Certain protocol steps cannot be transferred easily to automated liquid handling systems (e.g., centrifugation). New protocols needed to be developed. The isolation of nucleic acids can easily be automated using magnetic particles. Beads with paramagnetic (will be magnetized by an external magnetic field) or magnetic properties are used. Applications of this technique are very general and have advantages compared to the conventional separation protocols. The material is not subject to shear forces as no centrifugation steps are necessary and the use of organic reagents is obsolete. The magnetic beads are loaded with the nucleic acids and brought to an external magnetic field. The beads and bound nucleic acids are retained in the magnetic field while decontaminating material can be washed away (Figure 26.13). If used manually, the beads are often retained in a column placed in a magnetic field. In automated liquid handling systems, the magnetic field is usually provided by a magnetic plate, on which the 96-well plate containing the beads is placed. For the isolation of DNA, silica coated magnetic beads are used as DNA binds in the presence of chaotropic reagents to glass surfaces (Section 26.5). Using solid phase reversible immobilization (SPRI), DNA is loaded reversibly to magnetic beads modified with carboxyl-groups in the Figure 26.13 Principle of magnetic bead isolation. The nucleic acids in cell or bacterial lysates are bound specifically to the magnetic particles. By applying a magnetic field the beads are fixed and the contaminations can be washed away. After the washing steps the nucleic acids are eluted from the magnetic beads. All protocol steps can be performed on automated systems. The isolation protocol and the kind of bead depend on the type of nucleic acid to be purified.
680
Part IV: Nucleic Acid Analytics
presence of high salt concentration and poly(ethylene glycol) (PEG). The PEG is important for the binding of the DNA to the bead surface. Streptavidin coated beads are used for the isolation of very low amounts of mRNA. The beads are coupled with biotinylated oligo(dT) primer and the mRNA is coupled and isolated. This principle of binding biotinylated nucleic acids to streptavidin beads can be applied to a huge variety of isolation methods (e.g., isolation of DNA binding proteins).
26.8 Lab-on-a-chip Not only the format, time, and throughput of nucleic acid purification protocols have improved significantly in recent decades – with the lab-on-a-chip (LOC) system it is possible to isolate DNA in a miniaturized fashion. The LOC is part of the microelectromechanical systems (MEMSs) and is based on a chip that is between square millimeters and centimeters in size. The volume can be as low as 1 picoliter. The concept is to include all techniques, starting from the isolation of the nucleic acids (from blood or tissue) to the analysis of the nucleic acids, on the same chip. The systems are also part of the micro total analysis systems, μTAS. Similar to the isolation methods for automated liquid handling systems, protocols cannot be based on centrifugation or phenolic extraction. It is also important to achieve a certain concentration of the DNA for the subsequent analysis steps. The SPE methods can be transferred in part to the chip technology. The silica based isolation methods where DNA binds to the solid phase in the presence of chaotropic reagents are suitable. In addition, SPRI methods are applied. Additional suitable materials, like poly(methyl methacrylate) (PMMA), are used to enlarge the active surface on the chip.
Further Reading Ausubel, E.M., Brent, R., Kingston, R.E., Moore, D.D., Smith, J.A., Seidman, J.G., and Struhl, K. (1987) Current Protocols in Molecular Biology, John Wiley & Sons, Inc., New York. Farrell, R.E. (2010) RNA Methodologies: A Laboratory Guide for Isolation and Characterization, 4th edn, Academic Press, Elsevier. Glasel, J.A. and Deutscher, M.E. (1995) Introduction to Biophysical Methods for Protein and Nucleic Acid Research, Academic Press, New York. Green, M.R. and Sambrook, J. (2012) Molecular Cloning: A Laboratory Manual. 4th edn, Cold Spring Harbour Press, Cold Spring Harbor. Krieg, P.A. (ed). (1996) A Laboratory Guide to RNA: Isolation, Analysis and Synthesis, Wiley-Liss, New York. Kües, U. and Stahl, U. (1989) Replication of plasmids in Gram-negative bacteria. Microbiol. Rev., 53, 491–516. Levinson, P., Badger, S., Dennis, J., Hathi, P., Davies, M., Bruce, I., and Schimkat, D. (1995) Recent developments of magnetic beads for use in nucleic acid purification. J. Chromatogr. A, 816, 107–111. Micklos, D.A. and Freyer, G.A. (1990) DNA Science. A First Course in Recombinant DNA Technology, Cold Spring Harbor Laboratory Press and Carolina Biological Supply Company, Cold Spring Harbor. Perbal, B. (1998) A Practical Guide to Molecular Cloning, John Wiley & Sons, Inc., New York. Price, C.W., Leslie, D.C., and Landers, J.P. (2009) Nucleic acid extraction techniques and application to the microchip. Lab Chip, 9, 2484–2494. Tan, S.C. and Yiap, B.C. (2009) DNA, RNA and protein extraction: the past and the present. J. Biomed. Biotechnol., article ID 574396. Wen, J., Legendre, L.A., Bienvenue, J.M., and Landers, J.P. (2008) Purification of nucleic acids in microfluidic devices. Anal. Chem., 80, 6472–6479. Zähringer, H. (2012) Old and new ways to RNA. LabTimes, (2), 52–61.
Analysis of Nucleic Acids
Nucleic acids, isolated from different sources, different tissues from different organisms or cell or tissue cultures, subsequently appear as a compact, high molecular bulk of, especially in case of genomic DNA, unspecific fragments, which are hard to analyze in this status. For processing it is necessary to determine purity, conformation, fragment size and last but not least the sequence of these nucleic acid fragments. In this chapter we summarize basic analytical methods available for nucleic acids processing. The presented methods result in a basic characterization and/or are necessary for more detailed and final characterization or manipulation of nucleic acids. For example the transformation of a high molecular bulk of nucleic acids into specific molecular fragments by restriction analysis, which can easily be further characterized and manipulated for example by cloning. Fragments can be separated by gel electrophoresis, visualized by staining, isolated from the gel matrix or transferred by “blotting” to a specific carrier material for more specific characterization by “hybridization”. Most of these techniques are basic and daily routines when working with nucleic acids.
27.1 Restriction Analysis Ute Wirkner1 and Joachim W. Engels2 1
German Cancer Research Center, Clinical Cooperation Unit Translational Radiation Oncology, Im Neuenheimer Feld 400, 69120 Heidelberg, Germany Goethe University Frankfurt, Institute of Organic Chemistry and Chemical Biology, Department of Biochemistry, Chemistry and Pharmacy, Max-von-Laue Straße 7, 60438 Frankfurt am Main, Germany
2
Restriction analysis is used for the characterization, identification, and isolation of doublestranded nucleic acids and, thus, is a basic tool in nucleic acid analysis. Cloning of DNA molecules is almost unthinkable without restriction analysis. Even if it is possible to clone DNA by PCR without the need of restriction, mostly restriction analysis is used to prepare the DNA fragments and the vectors for cloning and to identify the resulting cloning product. In addition, for any other kind of DNA manipulation, such as mutagenesis or amplification by PCR, restriction analysis is the tool of choice to identify the desired product. To initially determine the crude structure of any DNA, from small fragments to whole genomes, establishing a restriction map is a useful step on the way to complete sequencing. Restriction analysis of genomic DNA to detect mutations or restriction fragment length polymorphisms (RFLPs) is used for genetic mapping, to identify and isolate disease genes, or, for example, in criminalistics to identify individuals.
27.1.1 Principle of Restriction Analyses The basis for restriction analysis is the activity of restriction enzymes, which bind and cut double-stranded DNA molecules at specific recognition sequences. These are mainly so-called Bioanalytics: Analytical Methods and Concepts in Biochemistry and Molecular Biology, First Edition. Edited by Friedrich Lottspeich and Joachim Engels. 2018 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2018 by Wiley-VCH Verlag GmbH & Co. KGaA.
27
682
Part IV: Nucleic Acid Analytics
Hybridization Section 28.1 PCR, Chapter 29
type II restriction enzymes. DNA-fragments resulting from this activity have a specific length, defined by the positions of the recognition sites, and can be separated according their size by gel electrophoresis (Section 27.2.1). The analysis of a DNA molecule results in a specific band pattern of restriction fragments. By comparison with a respective size standard each fragment can be assigned to its approximate size. Depending on the size of the initial DNA molecule(s) the detection of the fragments is carried out: if the original molecule is comparatively small, as with most cloning products or vectors (plasmids, lambda phages, cosmids around 3–50 kb), unspecific detection by staining all nucleic acids in the gel by, for example, ethidium bromide is sufficient (Section 27.3.1). If a certain area within a complex genome is to be analyzed, detection by specific Southern blot hybridization (Section 27.4.3) has to be performed or the fragment has to be amplified in vitro by PCR before undergoing a restriction analysis. Thus restriction analysis can be performed on any type and size of double-stranded DNA and is comparatively easy and quick to perform. Another undoubted aspect of the broad application spectrum is the large variety of restriction enzymes and respective restriction sites.
27.1.2 Historical Overview Before restriction enzymes were discovered it was almost impossible to characterize or isolate certain genes or genomic areas. Genomic DNA isolated from cells or tissues is a mass of large, chemically monotonous molecules that in principle only allow separation according to their size. Since a functional unit such as a gene does not exist as a single molecule in a cell but is part of a much larger DNA- molecule, specific breakdown of this large molecule is required to separate and isolate an interesting section (e.g., a gene). Even if DNA–molecules can be sheared at random sites by mechanical forces the result is again a heterogeneous unspecific mixture of DNA molecules and no defined DNA-fragments can be isolated. It was only with the discovery and isolation of restriction enzymes by Arber, Smith, and Nathans in the late 1960s that the opportunity first arose to manipulate DNA in a defined way, namely, to degrade it to specific fragments of defined length, which can be specifically separated and isolated. Hence, a first detailed characterization of DNA was possible and the basis for isolating and amplifying DNA by cloning was laid. By the implementation of restriction enzymes and consecutive cloning, hybridization, and other enzymatic DNA-manipulations, DNA has shifted from the least accessible to the easiest manipulable and analyzable macromolecule.
27.1.3 Restriction Enzymes Restriction enzymes are endonucleases that occur mainly in bacteria but they also exist in viruses and eukaryotes. They cleave phosphodiester bonds of both strands in a DNA molecule by hydrolysis and differ in recognition sequence, cleavage site, organism of origin, and structure. Several thousand restriction enzymes with several hundred different recognition sequences are known. Biological Function The biological function of restriction enzymes is to protect the organism of origin from infiltration by foreign DNA (e.g., from phages by cleaving and inactivating and, thus, “restricting” the growth of the phage). Its own DNA and every DNA synthesized in the cell is protected from this attack by modifications, mostly methylation. This restriction/modification (R/ M) system is specific to its organism of origin and is a protection mechanism, a kind of immune system. The host specificity of bacteriophages is based on this system; they only can infect bacteria efficiently that have the same methylation pattern as their bacteria of origin. Classification of Restriction Enzymes Primarily, three types of restriction enzymes (I, II, and III) are distinguished, whose properties and differences are summarized in Table 27.1. Type I restriction enzymes possess restriction as well as methylation activity and have a defined recognition sequence. If both strands of the recognition sequence are methylated, DNA will not be cleaved. If only one strand is methylated, the sequence is recognized and the second strand will be methylated. If both strands are methylated the sequence is recognized as well and the DNA is cleaved around 100 bp away from the recognition sequence. Restriction enzymes that are most
27 Analysis of Nucleic Acids Table 27.1 Classification of restriction enzymes (REases). Type I
Type II
Type III
Function
Endonuclease and methylase
Endonuclease
Endonuclease and methylase
Recognition sequence
Two parts, asymmetric
4–8 Bases, most palindromic
5–7 Bases, asymmetric
Cleavage site
Unspecific, often >1000 bp distance to the recognition sequence
Within or close to recognition sequence
5–20 Bases in front of the recognition sequence
ATP needed
Yes
No
Yes
frequently used in analytics are usually type II restriction enzymes. In addition, most well-known restriction enzymes belong to type II. In contrast to type I and type III restriction enzymes, type II restriction enzymes usually possess only restriction activity and cleave the DNA within the recognition sequence, which results in DNA fragments of defined length and defined ends. Type III restriction enzymes, like type I, have restriction activity as well as methylation activity. They cleave the DNA at sites at a distinct distance from the recognition sequence so that the resulting fragments have a defined length and variable ends. So-called homing endonucleases, like I-PpoI, in their native form have longer recognition sequences (>15 bp) and initiate the insertion of their own genes, so-called homing endonucleases genes (HEGs). They are selfish elements that colonize genomes and occur in different animal kingdoms from bacteria to eukaryotes. These endonucleases are mainly used for gene targeting and they are engineered to alter target site specificity. Nickases represent a small group of nucleases that have double-stranded recognition sequences but cleave only one strand of the DNA. Nomenclature of Type II Restriction Enzymes (REases) The nomenclature of type II restriction enzymes is based on the organism of origin. For example, restriction enzyme EcoRI was isolated from a resistance (R) factor of Escherichia coli strain RY 13. Here “I” stands for the first restriction enzyme isolated from this strain. Analogous BamHI was the first enzyme isolated from Bacillus amyloliquefaciens strain H. The scientific community agreed in this unique nomenclature in 2003. Here the terms restriction enzyme and restriction endonuclease were denoted synonymously and the abbreviation REases was introduced. Since type II enzymes compose by far the largest group of restriction enzymes, and since in addition there are members that differ from classical recognition features, the type II group was segmented into subgroups, which are described below in Table 27.3. All type II enzymes do not depend on ATP, they mostly do not form a complex with the respective methylase, they recognize a specific DNA sequence, and they cut within or close to the recognition sequence. The resulting DNA fragments have 5´ phosphate and 3´ -OH groups. Recognition Sequences Thousands of type II restriction enzymes with hundreds of recognition sequences have been characterized, and the number is constantly increasing. Comprehensive compilations can be found in regularly updated databases, company catalogues, or books of molecular biological methods. The recognition sequences of these restriction enzymes span 4–8 nucleotides and are, most often, palindromic. Table 27.2 lists representative examples of some restriction enzymes and their recognition sequences. In accordance with convention the sequence is given in the direction 5´ to 3´ . The cleavage site is usually located within the recognition sequence and thereby the resulting restriction fragments have defined ends, which is among other factors relevant for cloning. But there are also restriction enzymes like FokI (Table 27.3), whose cleavage site is a few bases away from the recognition site. As shown in Figure 27.1, the cleavage of DNA with restriction enzymes can result in blunt ends or in cohesive or “sticky” ends. Sticky ends can have either an overhanging 5´ or 3´ end, depending on which strand of the DNA forms the overhang. Usually, DNA fragments resulting from restriction enzyme activity have a 3´ hydroxyl and a 5´ phosphate group.
683
684
Part IV: Nucleic Acid Analytics Table 27.2 Specification of some type-II restriction enzymes (type-II-REases). Restriction enzyme
Recognition and cleavage sitea)
Organisme of origin
Isoschizomers
BamHI
G/GATCC
Bacillus amyloliquefaciens H
Bstl
Bstl
G/GATCC
Bacillus stearothermophilus 1503-4R
BamHI
EcoRI
G/ÁATTC
Escherichia coli RY13
Fokl
GGATGN9/ CCTACN13/
Flavobacterium okeanokoltes
HindII
GTPy/PuAC
Haemophilus influenzae Rd
HindII
HindIII
A/AGCTT
Haemophilus influenzae Rd
HpaII
C/CGG
Haemophilus parainfluenzae
Mspl
Mspl
C/CGG
Moraxella species
HpaI
NotI
GC/GGCCGC
Nocardia otitidiscaviarum
Sadl
GAGCITC
Streptomyces achromogenes
Sau3A
/GATC
Staphylococcus aureus 3A
Smal
CCC/GGG
Serratia marcescens Sb
Xmal
Xmal
C/CCGGG
Xanthomonas malvacearum
Smal
Mbol, Ndell
a) Py: pyrimidine (C or T); Pu: purine (A or G); N: A, C, G, or T.
The frequency of a recognition sequence depends mainly on its length but also on its own base composition and the composition of the DNA that is restricted. Assuming a random composition a 4 bp recognition sequence statistically occurs approximately every 44 bp (256 bp), a 6 bp or 8 bp recognition sequence respectively every 46 bp (4096 bp) or 48 bp (65 536 bp). However, different organisms possess different base compositions of their Table 27.3 Subtypes of Type-II restriction enzymes (REases). Subtypea)
Characteristics
Example
Recognition and cleavage site
A
Asymmetric recognition sequence
Fokl Acrl
GGATG (9/13) CCGC (3/-1)
B
Cleavage on both sides of the recognition sequence
Bcgl
(10/12) CGANNNNNNTGC (12/10)
C
Symmetric or asymmetric recognition sequence; R- and M-function in one polypeptide
Gsul HaelV Bcgl
CTGGAG (16/14) (7/13) GAYNNNNNRTZ (14/9) (10/12) CGANNNNNNTGC(12/10)
E
Two copies of the recognition sequence, one is cleaved, the other serves as an allosteric effector
EcoRII NaeI
↓CCWGC GCC↓GGC
F
Two recognition sequences, both are cleaved in coordination
SfiI SgrAl
GGCCNNNN↓NGGCC CR↓CCGGYG
G
Symmetric or asymmetric recognition sequence; depend on AdoMet
Bcgl Eco571
GTGCAG (16/14) CTGAAG (16/14)
H
Symmetric or asymmetric recognition sequence, gene structure similar to type I REases
Bcgl AhdI
(10/12) CGANNNNNNTGC(12/10) GACNNN↓NNGTC
M
Subtype IIP or IIA; recognize only methylated recognition sequences
DpnI
Gm6A ↓TC
P
Symmetric recognition and cleavage site
EcoRI PpuMI BslI
G↓AATTC RG ↓GWCCY CCNNNNN↓NNGG
S
Asymmetric recognition and cleavage site
FokI MmeI
GGATG (9/13) TCCRAC (20/18)
T
Symmetric or asymmetric recognition sequences; heterodimers
BpuI0I BslI
CCTNAGC ( 5/-2)b) CCNNNNN↓NNGG
a) Not all subtypes are exclusive! For example, BslI is subtype P and T. b) Abbreviation means the following cleavage: 5´ CC ↓T N AG C 3´ G GA N T ↓C G
27 Analysis of Nucleic Acids
685
Figure 27.1 DNA ends generated by restriction enzyme cleavage. Depending on the applied restriction enzyme three kinds of DNA ends occur: cohesive ends (sticky ends) arise by, for example, cleavage with BamHI and Sacl, whereby BamHI creates 5´ -overhanging and Sacl 3´ -overhanging ends. Blunt ends are created by, for example, SmaI.
genomes The A/T and accordingly the G/C content is rarely 50% and the dinucleotide CpG occurs less frequently in eukaryotes than do the other dinucleotides. Consequently, a recognition sequence containing CpG will occur less frequently in eukaryotic genomes than calculated according its length. Restriction enzymes with an 8 bp recognition sequence are, for example, applied to establish restriction maps from whole chromosomes. The resulting, very long DNA fragments are separated and detected by pulse-field gel electrophoresis (Section 27.2.3). Most frequently used restriction enzymes recognize 6 bp sequences since the length of the resulting fragments is good for separation and isolation. However, if a partial restriction is to be performed, for instance to establish a genomic library, restriction enzymes with 4 bp recognition sequences are selected. Isoschizomeres Isoschizomeres (Table 27.2) are restriction enzymes that have identical recognition sequences but originate from different organisms. The cleavage site might be identical (e.g., BamHI and Bstl) or different (e.g., Smal and XmaI). Isoschizomers with different cleavage sites are termed neoschizomers. The enzymes might also differ in their sensitivity towards methylation: for example, HρaII and Mspl have identical recognition sites but HρaII does not cleave if the second cytosine is modified to 5-methylcytosine (5m C), while Mspl will cleave despite this methylation.
27.1.4 In Vitro Restriction and Applications In a restriction enzyme reaction mixture, the DNA to be analyzed is incubated with the desired restriction enzyme under defined buffer conditions at a defined temperature for a certain time. The restriction buffer usually contains Tris-buffer, MgCl2, NaCI, or KCl as well as a sulfhydryl reagent (dithiothreitol (DTT), dithioerythritol (DTE) or 2-mercaptoethanol). A divalent cation (mostly Mg2+) is necessary for enzymatic activity, as well as the buffer, which provides the correct pH, mostly between pH 7.5 and pH 8. Some restriction enzymes are sensitive towards ions such as Na+ or K+, while others are active within a wide concentration range. Sulfhydryl reagents stabilize the enzyme. The optimal temperature for most restriction enzymes is 37 °C, but it may vary depending on the enzyme, with respect to the organism of origin, to higher (e.g., 65 °C for Taql) or lower (e.g., 25 °C for Smal) temperatures. Complete Restriction For most purposes complete restriction of DNA is intended. For this, optimal conditions for the respective restriction enzyme are selected and a sufficient amount of enzyme for the DNA to be cleaved.
The amount of restriction enzyme is given in units: one unit of a restriction enzyme is the amount needed to cleave one microgram of substrate DNA under optimal conditions within one hour. As a general rule bacteriophage lambda DNA is used as substrate for this definition.
686
Part IV: Nucleic Acid Analytics
Incomplete or Partial Restriction For some purposes, like restriction mapping or the preparation of a genomic DNA-library, partial restriction is desired. This means that statistically not all of the restriction sites are cleaved. This is achieved by an under optimization of the reaction conditions, such as a lower amount of restriction enzyme, shorter reaction time, or change of buffer conditions (e.g., reduced MgCl2 concentration). Multiple Restriction This involves the restriction of DNA with several restriction enzymes. The DNA might be incubated either simultaneously or one after another with the desired restriction enzymes. The crucial criterion here is the compatibility of the reaction conditions. Multiple restriction, among others methods, is applied to establish restriction maps.
Restriction Mapping To establish a restriction map recognition sequences of one or several restriction enzymes are localized within a DNA-molecule. Thus the restriction map is a crude physical map of the DNA-molecule to be analyzed. The perfect physical map is the complete nucleotide sequence of the DNA. Consequently, restriction mapping is applied to identify known sequences, for example, to verify successful cloning of a known DNA-fragment or as the first step of projects that aim to identify a complete nucleotide sequence. Therefore, restriction analysis of DNA fragments integrated in cloning vectors (e.g., plasmids, cosmids, or lambda phages) is performed. Before the introduction of next generation sequencing, DNA had to be cloned before elucidating the nucleotide sequence, which was usually performed by Sanger sequencing. Restriction maps of these sequencing clones are established, overlapping clones are identified by comparing their restriction maps, and finally the map of the originally cloned DNA can be elucidated. Often it is not necessary to isolate the fragments after the first restriction, instead it is sufficient to compare the fragment pattern of the single digest with that of the double digest to determine the order of the restriction fragments. For this approach it is important to use restriction enzymes that produce at least a few overlapping fragments. Consequently, it might be necessary to test some restriction enzymes.
Radioactive Systems, Section 28.4.2
Probes for Nucleic Acid Analysis, Section 28.2 Analysis of Epigenetic Modifications, Chapter 31
Combination of Multiple Restriction Enzymes By this method the relative position of recognition sequences of different restriction enzymes is determined and from this their absolute position on the originally analyzed DNA fragment. To do so, the DNA fragment to be analyzed is first restricted with each single restriction enzyme in one reaction and fragments are analyzed by gel electrophoresis. Ideally these fragments are isolated and restricted with the second restriction enzyme and these double restrictions again analyzed by gel electrophoresis. By comparing the lengths of the resulting DNA fragments after single and double restriction, overlapping parts can be identified and the relative order of the fragments can be determined. This is shown in Figure 27.2 using the example of a 5 kb DNA-fragment. Partial Restriction By this method the order of recognition sequences of a single restriction enzyme can be identified. The DNA fragment to be analyzed is digested once completely and once incompletely with the same restriction enzyme and both reactions are analyzed by gel electrophoresis. By comparing the pattern of the resulting restriction fragments the completely restricted fragments can be allocated to the incompletely restricted fragments and thus the order on the original DNA fragment can be determined. This method is shown in Figure 27.3 on a 5 kb DNA-fragment. In the case of a complex restriction pattern, for example, when analyzing a long DNA fragment or using a very frequently cutting enzyme, it is advisable to use the method shown in Figure 27.4. Hereby, the DNA-molecule is labeled at one end before partial digestion by, for example, incorporation of a labeled nucleotide. After gel electrophoresis these labeled fragments can be detected selectively (e.g., by autoradiography). The size of a detected fragment corresponds here to the distance of a cleavage site to the labeled end of the DNA molecule.
Restriction Analysis of Genomic DNA When carrying out restriction analysis of large eukaryotic genomes there is the problem that too many restriction fragments are generated. After gel electrophoresis no single bands are visible but instead there is a smear of DNA, which consists of specific DNA fragments with many sizes. By selecting certain hybridization probes, a fragment in the analyzed genome containing DNA complementary to the selected probe can be detected. This is done by so-called Southern blot analysis (Section 28.4.3). This analysis enables, for instance, the restriction analysis of a gene whose transcript has been cloned as cDNA and can be used as hybridization probe. There are other objectives for which restriction analysis is helpful, such as the detection of a methylation pattern that is lost by cloning and
27 Analysis of Nucleic Acids
687
Figure 27.2 Restriction mapping by multiple restriction. A 5 kb long, linear DNA fragment was cleaved by restriction enzymes A and B in single reactions and in a double reaction. (a) Separation of the restriction fragments in an agarose gel. Fragment sizes determined by comparison to the size standard are given. Cleavage with enzyme A results in restriction fragments with lengths of 2500 bp (fragment A2500), 1300 bp (A1300), and 1200 bp (A1200). The corresponding nomenclature for enzyme B fragments and double restriction fragments is shown. Restriction fragments from single reactions were isolated and cleaved with the respective second restriction enzyme: A-fragments with enzyme B and B-fragments with enzyme A. (b) Electrophoretic separation of these secondary cleavage products. By comparison of the restriction pattern, overlapping fragments can be identified and, as shown in (c), can be aligned: The 1900 bp fragment from the double digest is contained in A2500 and B2100; consequently, A2500 and B2100 overlap in this area. In addition, A2500 contains a 600 bp fragments that is also present in B1400, and B1400 contains a 200 bp fragment that overlaps with A1200. After analysis of all fragments the restriction map of the 5 kb DNA fragment can be generated.
688
Part IV: Nucleic Acid Analytics
Figure 27.3 Restriction mapping by partial digestion. A 5 kb DNA molecule was cleaved both completely and partially with restriction enzyme A. (a) Gel electrophoretic separation of the resulting restriction fragments. By comparing complete and partial cleavage 5000, 3800, and 3700 bp fragments can be identified as partially cleaved, of which the 5000 bp fragment is the original fragment. (b) The 3800 bp fragment can only be composed of the 2500 and 1300 bp fragments and the 3700 bp fragment composed of the 2500 and 1200 bp fragments. Accordingly, the restriction map can be established.
DNA sequencing. For other applications, for example, to compare restriction patterns in several individuals, cloning is too laborious and analysis is done directly on genomic DNA. Alternatively, interesting areas may be amplified by polymerase-chain-reaction (PCR) before amplification products are then analyzed by restriction. This reaction can then be analyzed again by normal gel electrophoresis, without any specific labeling. Detection of Methylated Bases
Since there are isoschizomers like HρaIl and Mspl (see above) that differ in their sensitivity towards a methylation within their recognition sequence, methylated bases can be detected by them. As an example so-called CpG islands are found in several promoter regions of eukaryotic genes. They are sections of DNA where dinucleotide CpG is overrepresented. If a gene is transcriptionally inactive this is often connected to the methylation of cytosines in the CpG island of the gene. If in a CpG island not all restriction sites cleaved by MspI are cleaved as well by HpaII this is an indication of methylation and thus transcriptional inactivity of the respective gene. Restriction analysis has to be performed directly on genomic DNA and is detected by Southern blot analysis (Section 28.4.3). The difficulty of DNA-methylation analysis is discussed in detail in Chapter 30.
Detection of Mutations and Restriction Fragment Length Polymorphisms (RFLP) Individuals within a population differ in the composition of their genomes. There exist highly conserved areas that are of high relevance for the carrier and which are nearly unchanged within the population or even among species (e.g., globin genes). A mutation of such a region may cause illness or the death of the carrier (e.g., sickle cell anemia as a mutation in globin genes). On the other hand, there are areas with several variants within a population, so-called polymorphisms.
27 Analysis of Nucleic Acids
689
Figure 27.4 Partial restriction and end labeling. A 5 kb DNA molecule is labeled at one end, partially restricted by enzyme A, and the reaction products are separated by electrophoresis. Only the endlabeled fragments are detected (a). (b) The size of every fragment corresponds to the distance between a restriction site of enzyme A and the labeled end. The label is shown as a dot. The resulting restriction map is shown.
These differences in DNA sequence can be exchanges, deletions, or insertions of single bases or sections of DNA. These mutations can cause a change in length of a restriction fragment, or a restriction sequence can be deleted or inserted. If a polymorphic region can be detected by change of a restriction pattern this is called a RFLP. Hereby restriction analysis is either performed on genomic DNA in combination with Southern blot analysis (Section 27.4.3), and the interesting region is used as hybridization probe, or the region is amplified by PCR in vitro and restriction analysis is performed on the PCR product. Since every individual has two homologous copies of every DNA section, in the case of heterozygosity two kinds of restriction pattern will be detected when analyzing this RFLP, one representing the paternal and one the maternal allele (Figure 27.5). Figure 27.6 shows the heredity of a RFLP is over three generations. Genetic Fingerprint A genetic fingerprint is based on the detection of highly variable RFLPs, which result in a restriction pattern that is highly characteristic for each individual. The basic causes for this are short, mostly two to three base pairs long, highly repetitive sequences, whose number of repetition is highly variable. This is helpful in identifying individuals, for example, as proof of paternity or in criminalistics (compare Section 27.2.1). Restriction Fragment Length Polymorphisms in Genetic Mapping In genetic mapping it is not the nucleotide sequence that is evaluated but the relative order of so-called genetic markers towards each other. This is done by gene linkage analysis. Possible genetic markers are blood groups and disease genes and also RFLPs. This is discussed in detail in Chapter 36.
RFLPs as Genetic Markers, Section 36.1.2
690
Part IV: Nucleic Acid Analytics
Figure 27.5 Detection of a RFLP by Southern blot analysis. (a) Homologous chromosomal sections of an individual containing a polymorphous region. Restriction sites are indicated by arrows. The area detected by the hybridization probe in Southern blot analysis (Section 27.4) is marked. (b) The respective result after restriction, gel electrophoresis, and Southern blot analysis. One restriction site is missing in the maternal allele, which results in detection of a longer restriction fragment. The shorter fragment corresponds to the paternal allele. Thus the respective restriction fragment is polymorphic in this individual.
Figure 27.6 Heredity of a restriction fragment length polymorphism over three generations. In the family analyzed four alleles occur for the polymorphic region: allele A, B, C, and D. The heredity is in accordance with Mendel’s laws. Most individuals are polymorphic for the restriction fragment analyzed, others have the same allele in both homologous areas.
27.2 Electrophoresis Marion Jurk Miltenyi Biotec GmbH Friedrich-Ebert-Straße 68 51429 Bergisch Gladbach Germany
Electrophoretic Techniques, Chapter 11
Electrophoresis is a most important method by which to analyze nucleic acids. Its advantages are obvious: electrophoresis can be performed in a very short time frame with low amounts of material. The necessary equipment and detection methods are in most cases very cheap and are easily available in every laboratory. The underlying theoretical principles and the hands on work have similarities to, but also significant differences from, the electrophoretic separation of proteins. Like proteins, the separation of nucleic acids in an electric field is performed in a solid carrier material such as agarose or polyacrylamide. In contrast to proteins, nucleic acids are negatively charged within a very broad pH range. The negative charges are carried by the phosphate groups on the backbone of
27 Analysis of Nucleic Acids
691
Figure 27.7 Theories explaining the movement of nucleic acids in the gel matrix. The Ogston theory (a) postulates a globular sphere for the nucleic acids. Its radius is defined by the length of the molecule and thermal agitation. The molecules migrate through the pores of the gel matrix if the diameter of the nucleic acid smaller than the average pore size. According to the reptation theory (b) the nucleic acids align themselves along the electric field and move snake-like through the gel matrix. Source: adapted according to Martin, R. (1996) Gel Electrophoresis: Nucleic Acids, Bio Scientific Publishers Limited, Oxford.
the nucleic acids. The migration of nucleic acids in the electric field towards the anode is therefore pH independent. Another notable difference from proteins is their constant charge density, meaning that the ratio of molecular weight to negative charge remains unchanged. There is no need to generate homogenous charge surfaces by SDS like it is the case with proteins. The electrophoretic mobility, that is, the velocity of migration in the electric field (Chapter 12), is equal for all nucleic acids in free solution independent of their molecular weight. Differences in their mobility can only be measured in a solid gel matrix. The differences in migration velocity are caused solely by different molecule sizes. The movement of nucleic acids in an electric field can be described by two theories (Figure 27.7). The migration of nucleic acids in reality could be seen as a “mixture” of these two theories. The Ogston sieving effect is based on the assumption that nucleic acids in solution have a globular, spherical structure. The size of nucleic acids is described by the radius of the sphere that is theoretically occupied by the nucleic acid. The bigger the sphere the more often collisions with the gel matrix will occur. Migration of the nucleic acids in the field will then be slowed down. Very small fragments will not be slowed down by the pores of the gel matrix. Small fragments cannot be separated. According to the Ogston sieving theory, very big molecules with sphere sizes bigger than the pores of the gel should not be able to migrate at all. A second theory, the reptation theory aims to explain the migration of big nucleic acids in the electric field. The theory assumes that big nucleic acids can abandon their globular structure and align themselves in the electric field. The migration of the molecules occurs by moving one end of the molecule ahead through the matrix pores (end-to-end migration). The theory is called reptation owing to the snake-like movement of the nucleic acids. Size selection occurs because bigger molecules need more time to move than smaller ones. Both theories together can explain most of the phenomena observed in the electrophoresis of nucleic acids with sizes of 10 kb. The behavior of very large molecules cannot be explained by these theories and requires new model theories (Section 27.2.3).
27.2.1 Gel Electrophoresis of DNA Agarose Gels The choice of carrier material depends mainly on the size and kind of nucleic acid to be analyzed. Agarose, a linear polysaccharide polymer, is the most important electrophoresis material for nucleic acids. The migration velocity of DNA molecules is determined by several factors. The effective size of a nucleic acid is not only determined by its absolute mass but also depends on its form: superhelical (form I), open-circular (form II), double-stranded linear (form III), or single stranded. Separation of Linear, Double-Stranded DNA Fragments Gel electrophoresis of linear DNA fragments (form III DNA) can be used to determine the size of the DNA reproducibly
692
Part IV: Nucleic Acid Analytics
Figure 27.8 Relationship between the migration distance and fragment length at various agarose concentrations. The semi-logarithmic curves were created using length standards. The size of a fragment can be determined by its position. Buffer: 0.5 TBE/0.5 μg ml 1 ethidium bromide. electrophoresis at 1 V cm 1; 16 h.Source: adapted according to Maniatis, T., Fritsch, F.F., and Sambrook, J. (1989) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor.
with good accuracy. There is a linear relation between the logarithm (log10) of the size (in bp) of the fragment and the migration distance (measured in cm in reference to the total distance) in an agarose gel (Figure 27.8). The migration velocity of linear DNA fragments is dependent on the concentration of the agarose, the applied voltage, composition of the running buffer, and the presence of intercalating dyes. Linear DNA fragments can be separated by agarose gels spanning a broad range of fragment lengths (Table 27.4). Very small fragments (100 bp) migrate in 1–1.5% containing agarose gels at the same speed because the pores of the gels are bigger than the fragments. Separation of these small fragments becomes possible by increasing the agarose concentration. Small DNA fragments and oligonucleotides are usually separated using 2–3% agarose gels. The migration speed of the fragments is proportional to the applied voltage. Big fragments migrate into the gel slowly if the voltage is too high – bigger fragments should therefore be separated at lower voltages. Good separation of fragments (2 kB) takes place when the applied voltage is less than 5 V cm 1, with the distance between the electrodes being the influential parameter, not the length of the gel. For the separation of DNA molecules a running buffer with Tris acetate (TAE) or Tris borate (TBE) is used. Fragments separated in TAE buffer can be better isolated from an agarose gel. The bands are usually sharper. A disadvantage of TAE buffer is its lower buffering capacity and lower stability during electrophoresis. If long electrophoresis times or high field forces are necessary, TBE buffer is used. Linear fragments migrate faster in TBE buffer (approx. 10% faster) than in TAE buffer. The separation capacity is similar in both buffer systems; however, superhelical DNA can be better separated in TBE buffer. The ion concentration of the running buffer is of importance as well. Too low a concentration causes minimal electric conductivity and consequently the speed of migration of the nucleic acids is low. Too high a concentration results in a very high electric conductivity, which causes heating of the buffer. The DNA could possibly be denatured and the agarose melted. The presence of intercalating dyes influences the speed of the nucleic acids as well. The principle of intercalation is described in Section 27.3.1. Addition of ethidium bromide decreases the migration velocity of linear double-stranded DNA fragments about 15%. Separation of Circular DNA The migration velocity of circular DNA form I (superhelical) or form II (open) depends mainly on the consistency of the agarose gel. Superhelical DNA migrates faster than linear DNA. Relaxed DNA molecules (form II) are slower than linear or superhelical DNA (Figure 27.9). The migration velocity of these three forms is influenced by the running conditions, concentration of the agarose, applied voltage, and choice of running buffer. The different forms can be identified by ethidium bromide staining.
Practical Considerations Agarose gels can be poured as vertical or horizontal gels. In most laboratories, the more practical, vertical gels are used. According to the size of the gel, there are mini, midi, and maxi gels. Mini gels have a very short distance for separation (6–8 cm) and are not suited for size determination of DNA fragments. They are used for a quick analysis of the quality of the DNA and for control of restriction digestion. Midi and maxi gels (approx. 20 or Table 27.4 Coarse separation of DNA fragments at different agarose concentrations. Source: according to Sambrook, J. and Russell D.W. (eds) (2001) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Press, Cold Spring Harbor. Agarose concentration (% w/v)
Optimal separation range of linear double-stranded DNA fragments (kb)
0.3
5–60
0.6
1–20
0.7
0.8–10
0.9
0.5–7
1.2
0.4–6
1.5
0.2–3
2.0
0.1–2
27 Analysis of Nucleic Acids
30–40 cm) are used for accurate DNA size determination and for the isolation of fragments. The separation distance and loading capacity of the gel are much higher. The DNA is loaded onto the gel using a so-called loading buffer. They increase the density of the DNA solution (using Ficoll, glycerol, or saccharose). The DNA solution sinks into the gel pockets and does not diffuse to the running buffer. The loading buffer usually contains negatively charged dyes indicating the migration during electrophoresis. The most commonly used dyes are bromophenol blue and xylene cyanol. Bromophenol blue migrates in an agarose gel, depending on the exact conditions, like a linear DNA fragment of approx. 300 bp. An important means for the determination of DNA fragment sizes are DNA standards. DNA standards are commercially available and contain DNA fragments of defined sizes. They are separated together with the DNA fragments of interest. Using the known sizes of the DNA standard, the unknown size of the DNA fragment can be determined. For an accurate size determination it is very important that the DNA standard and unknown fragment are loaded as similar amounts and with similar buffer conditions. DNA standards can be produced by restriction cleavage of plasmid or phage DNA. A common DNA standard is the 1 kb DNA ladder (Figure 27.10). There are also standards available for small size DNA fragments, for example, the 100 bp ladder consisting of DNA fragments that differ exactly by 100 bp. The choice of standard depends on the size of the expected DNA fragments.
693
Figure 27.9 Migration behavior of superhelical, open, linear, and denatured DNA in an agarose gel. The migration properties of superhelical DNA can be influenced by the ethidium bromide concentration.
Denaturing Agarose Gels
Single-stranded DNA forms intramolecular secondary structures and intermolecular aggregates very easily. These structures influence the migration behavior in a gel. For the size determination of single-stranded DNA, denaturing agarose gels are employed where the electric mobility of the DNA depends solely on the molecular size.
Figure 27.10 Various commonly used DNA length standards. The λ-DNA marker can be generated by digestion with certain restriction enzymes (Section 27.1). Source: Photography by Dr. Marion Jurk.
694
Part IV: Nucleic Acid Analytics
Alkaline agarose gels are used to determine the synthesis efficiency of first and second strand synthesis of cDNA and to test the nicking activity of enzyme preparations. Sodium hydroxide is used as denaturing agent. The agarose needs to be dissolved in water because the addition of hot sodium hydroxide hydrolyzes its polysaccharide structures. Since ethidium bromide does not bind to the DNA at high pH, the electrophoresis is performed in the absence of ethidium bromide. Low Melting Agarose and Sieving Agarose Derivatization of agarose by the introduction of hydroxyethyl groups into the polysaccharide chains results in an agarose moiety with different properties. This low melting agarose is also heated and will gel when cooling down. However, its melting point is reduced. This property is used when isolating DNA fragments out of agarose gels (Section 27.5.3). The migration velocity of DNA in low melting agarose gels is higher, the separation and loading capacity is lower. The properties of sieving agarose are similar to those of low-melting agarose. Sieving agarose is used especially for the separation of small DNA fragments. Both types of agarose should not be used at concentrations below 2% for reasons of stability. Recommended concentrations are 2–4%.
Gel Media for Electrophoresis, Section 11.3.2
Polyacrylamide Gels The properties of polyacrylamide and the definitions of concentration and degree of crosslinking have been introduced in Section 12.3.2. Electrophoresis of DNA in polyacrylamide gels (abbreviated as PAGE, polyacrylamide gel electrophoresis) can be performed under native or denaturing conditions depending on the scope of application. Polyacrylamide gels (these gels as well as agarose gels are colloquially called slab gels) are poured as horizontal gels between two glass plates. Advantages and disadvantages of polyacrylamide and agarose gels are listed in Table 27.5. Non-denaturing Gels for Analysis of Protein–DNA Complexes Native, non-denaturing gels are used for electrophoretic mobility shift assays (EMSAs). With this method, protein– DNA complexes can be separated from free DNA. Large DNA–protein complexes are retarded in the gel by the cage effect. This method is described in detail in Chapter 32. Non-denaturing PAGE of Double-Stranded DNA Native polyacrylamide gels yield a higher resolution than agarose gels (Table 27.6) together with a higher loading capacity. This is used for the purification and isolation of double-stranded DNA fragments (1000 bp) Loading capacity higher without loss of resolution
DNA can easily be isolated
Purification yields DNA of high quality
DNA can easily be stained Capillary and vacuum blotting Disadvantages Bands are more diffuse and broader
Difficult to pour, greater technical effort
Resolution of smaller fragments is lower
No capillary or vacuum blotting possible
Isolated DNA fragments can contain impurities
Lower separation range
27 Analysis of Nucleic Acids
695
Table 27.6 Separation range of native polyacrylamide gels. The ratio of acrylamide to N,N´ methylene-bisacrylamide is 29 : 1. Acrylamide concentration (%)
Separation range (bp)
Migration of bromophenol blue in native gels (bp)
3.5
100–2000
100
5.0
100–500
65
8.0
60–400
45
12.0
50–200
20
15.0
25–150
15
20.0
5–100
12
(number of base pairs) under certain electrophoresis conditions. This effect is thought to be due to conformational changes in the DNA, such as kinks or bends. The abnormal migration behavior is more pronounced at higher polyacrylamide concentrations, higher Mg2+ concentrations, or at lower temperatures. An increase in temperature or concentration of Na+ ions results in opposite effects. Native PAGE of Single-Stranded DNA (SSCP) These gels are usually used to analyze changes within genomic DNA for certain disease indications. For the determination of various genetic mutations, methods are needed that allow many patient samples to be run at the same time. Sequencing of the individual genomic DNA would be far too expensive and time consuming. The commonly used SSCP (single-stranded conformational polymorphism) method is based on the observation that single-stranded DNA molecules with different sequences assume different conformations. The double-stranded DNA fragments to be analyzed are denatured using formaldehyde and are loaded onto a native polyacrylamide gel. The isolated single-stranded DNA strands will assume individual conformations resulting in different migrational behavior (Figure 27.11). Gene sections of individuals
Figure 27.11 Schematic principle of SSCP analysis. The DNA fragments are denatured in the presence of formamide by heat. The resulting single-stranded molecules assume a certain conformation according to their sequence and base composition. The single-stranded molecules are separated using native polyacrylamide gel electrophoresis. The migration properties differ according to the different conformations. From the characteristic gel pattern, homozygous and heterozygous individuals with genes containing certain point mutations in certain genes can be identified. Source: adapted according to Martin, R. (1996) Gel Electrophoresis: Nucleic Acids, Bio Scientific Publishers Limited, Oxford..
696
Part IV: Nucleic Acid Analytics Table 27.7 Separation of oligonucleotides in denaturing polyacrylamide gels. The ratio of acrylamide to N,N´ -methylene-bisacrylamide 19 : 1. Acrylamide concentration (%)
Separation range (in nt; nt = nucleotide)
20–30
2–8
15–20
8–25
13.5–15
25–35
10–13.5
35–45
8–10
45–70
6–8
70–300
containing point mutations will have a different conformation and consequently result in a different migration speed. It is essential to perform a native PAGE as denaturing gels would result in a uniform separation according to the size of the fragments but not according to sequence. Using SSCP various different DNA samples can be analyzed for point mutations simultaneously. The first SSCP analysis was performed using restriction generated fragments of DNA following Southern blot analysis (Section 27.4.3). In more recent approaches, the gene sections of interest are amplified using PCR and radioactive labeled with no follow up detection necessary. The analysis is performed under the assumption that mutations will behave differently than the original fragment. A negative result is not irrevocable proof that there are no point mutations present in the analyzed gene section. Denaturing PAGE of single-stranded DNA or RNA is used in various fields of applications due to the very exact separation of the single-stranded molecules (Table 27.7). The most common denaturing agent is urea, but formaldehyde is also used. Alkaline reagents cannot be used with polyacrylamide gels because the gel matrix will be destroyed. The gels are usually polymerized in the presence of 7 M urea, the running buffer is TBE. The loading buffer is usually formaldehyde to denature the probes (Section 27.2.2). Single-stranded DNA or RNA migrates in this type of gel independent of its sequence and, therefore, the separation of DNA molecules differing in size by only in one nucleotide is possible. These gels are therefore used for sequencing, S1 nuclease analysis, and RNAse protection experiments. The gels are also used for the DNA fingerprinting method. DNA Fingerprinting DNA fingerprinting is applied for the lineage analysis of genomic DNA. The technique is used in forensic analysis and for zoological studies, and also for paternity testing. DNA minisatellites, repetitive variant repeats in the genomic DNA of 10–100 bp, are inherited to a similar extent from both parents. The distribution and cleavage behavior are unique for each individual. The genomic DNA is cut by restriction enzymes and separated on denaturing polyacrylamide gels. The gels are blotted to a membrane (Southern blotting, Section 27.4.3) and hybridized using specific probes that recognize the minisatellite DNA. Fingerprinting can also be performed using PCR with random primers. Short DNA fragments are synthesized by PCR in the presence of radioactive labeled nucleotides. Two individuals will differ in their spectra of synthesized DNA fragments when run on a denaturing polyacrylamide gel with high resolution. The electrophoretic methods have been replaced but will eventually be completely replaced by the next generation of sequencing methods. Oligonucleotide Purification Denaturing polyacrylamide gels are often used for the purification of synthetic oligonucleotides or single-stranded DNA. Oligonucleotides with n bases can be separated from oligonucleotides with n 1 bases, yielding a population of nucleic acids with uniform length. The oligonucleotides are detected by fluorescent quenching. The gel is laid on a thin-layer chromatography (TLC) plate and irradiated with UV light (long wavelength). The TLC plate fluoresces upon excitation, except for the parts where the oligonucleotides are located. The oligonucleotide can be identified as dark bands and excised from the gel. If the oligonucleotides need to be isolated from the gel, the excitation time should be kept to a minimum to avoid damage to the nucleic acids or crosslinking to the gel matrix.
27 Analysis of Nucleic Acids
697
When this method is chosen for oligonucleotide purification the oligonucleotides should not contain modifications that interact with the polyacrylamide matrix.
27.2.2 Gel Electrophoresis of RNA Similar to single-stranded DNA, single-stranded RNA forms secondary structures by intra- and intermolecular base pairing. These different conformations behave differently during gel electrophoresis. An exact and reproducible analysis of RNA is only possible using denaturing gel electrophoresis. In denaturing gels the hydrogen bridges are destroyed and all RNA molecules will be separated according to their molecular weight. The electrophoresis of complex RNA mixtures (e.g., for Northern blotting, Section 27.4.4) is performed using denaturing 1–1.5% agarose gels. Smaller RNA fragments are separated like oligonucleotides using denaturing PAGE. For a rapid analysis of the RNA (e.g., for reasons of quality control), non-denaturing TBE gels can be used. The denaturing reagents employed are usually dimethyl sulfoxide/glyoxal or formaldehyde for agarose gels, and urea for polyacrylamide gels. Formaldehyde Gels The denaturing effect of formaldehyde is based on the formation of socalled Schiff bases between the aldehyde functional group and the amino group of the adenine, cytosine, and guanine bases. Consequently, the amino groups of the nucleobases cannot form hydrogen bonds for the formation of secondary structure or aggregates. The agarose gel usually contains 1.1 % formaldehyde. For longer separations (overnight), the formaldehyde content must be increased. As formaldehyde is toxic, electrophoresis should be performed under a fume hood. Since formaldehyde also interacts with the amino groups of Tris (Tris(hydroxymethyl) aminomethane), a different running buffer has to be used. This is usually a mixture containing 3-N-morpholino-1-propane sulfonic acid (MOPS) and sodium acetate. The RNA has to be denatured before loading onto the gel in the presence of formaldehyde using formamide and MOPS. The formamide destroys the base pairing of the RNA, allowing the formation of Schiff bases between the formaldehyde and the amino groups. Formamide can be contaminated by ions like ammonium formate, which can hydrolyze the RNA. Formamide is therefore deionized using ion exchange chromatography. Since MOPS possesses a very high buffering capacity, there is no need to replace or recycle the running buffer during the run as it is the case with glyoxal gels. If the gel needs to be blotted afterwards, the formaldehyde needs to be removed before blotting. Otherwise the amino groups of the bases will not be available for hybridization with the probes.
Glyoxal gels yield sharper RNA bands than formaldehyde gels, which is an advantage for blotting. Glyoxal binds to the guanine residues at neutral pH and prevents base-pairing of the RNA. In contrast to formaldehyde gels, glyoxal is only added before loading and is not added to running or loading buffer. The RNA is denatured in the presence of 1 M glyoxal and sodium phosphate and dimethyl sulfoxide (DMSO) at 50 °C. Sodium phosphate acts as buffer and DMSO destroys the inter- and intramolecular hydrogen bonds, enabling the glyoxal to react with the guanine residues. Glyoxal is easily oxidized to glyoxylic acid, which hydrolyzes RNA. Contaminating glyoxylic acid has to be removed by ion exchange before use of glyoxal in gel electrophoresis. Glyoxal reacts with ethidium bromide; consequently, the separation is performed in the absence of the intercalating reagent. Above pH 8.0, the glyoxal dissociates from the RNA. To avoid pH gradients during electrophoresis, the running buffer has to be replaced or recycled using a pump. RNA Standards Cytoplasmic RNA of eukaryotic cells consists of approx. 95% of ribosomal RNA (rRNA). Ribosomal RNA consists of 28S, 18S, and 5S rRNA. RNA preparations of high quality display two sharp, clearly separated bands in an agarose gel (Figure 27.12) that can be used as internal standards. The exact size of the ribosomal RNA depends on its origin: for human rRNA the lengths were determined to be 5.1 kb for 28S and 1.9 kb for 18S rRNA. Other length standards are commercially available or can be generated by in vitro transcription of DNA fragments of defined length.
Since RNA is subject to nuclease digestions and hydrolysis by acids or bases, the experimental set-up for RNA electrophoresis has to be modified compared to DNA electrophoresis. The same precautionary measures as for isolation of RNA have to be taken for the electrophoresis of RNA. For example, the electrophoresis chamber needs to be cleaned carefully and only RNAse-free water should be used.
Schiff bases are generated by the reaction of primary amino groups with aldehydes and water is released. An imine bond is formed.
698
Part IV: Nucleic Acid Analytics
Figure 27.12 Migration of cytoplasmic RNA. High quality RNA preparations should generate clearly visible bands of the 28S and 18S of ribosomal RNA and should barely be degraded. These bands can be used as internal length standards. Source: Photography by Dr Marion Jurk.
27.2.3 Pulsed-Field Gel Electrophoresis (PFGE) Principle High molecular weight nucleic acids cannot be separated by regular gel electrophoresis. They all possess the same so-called limiting mobility. This effect cannot be explained by the reptation theory; several other theories have been put forward to explain the observed phenomenon. One model postulates that the DNA molecules act as rigorous entities, whereby no separation effect can be obtained. Another model describes the movement of high molecular weight DNA as similar to movements in solution, where no separation effect can occur. The formation of loop structures could also explain the limiting mobility effect. Pulsed-field electrophoresis (PFGE) uses, instead of a continuous electric field, a pulsed field with changing directions of the electric fields. DNA molecules assume a relaxed globular shape in free solution (without an electric field). When an electric field is applied, the molecules align themselves according to the electric field and will move towards the anode (according to reptation theory). On removing the electric field, the molecules will again resume the relaxed globular state. By applying an electric field with a different direction, molecules will realign according to the new field. If the direction of the electric field is changed again, the molecules have to realign again. Relaxation and alignment of the molecules according to the electric field depends on the size of the molecule. Larger molecules need more time for relaxation and alignment than smaller molecules. The time needed for movement along the electric field is shorter for larger molecules than for smaller ones. The sum of all applied field vectors yields the direction of movement of the DNA molecules. The separation principle of large DNA molecules here is based on the time the molecule needs to align according to the applied electric fields. Within PFGE, several techniques are condensed, most of them differing in direction and sequence of the electric pulses. Field inversion gel electrophoresis (FIGE) uses two electric fields with opposing directions. Migration is achieved because the duration and amplitude of the forward pulse is larger (Figure 27.13). The method can separate molecules within a broad size range with high resolution. The CHEF (contour-clamped homogenous electric field) method is a more commonly used PFGE method. The electrodes are arranged in a hexagon around the agarose gel (Figure 27.14).
Figure 27.13 Principle of field inversion gel electrophoresis (FIGE). Two alternating electric fields with directions differing by 180°. The migration direction to the anode is determined by a longer or stronger pulse in this direction.
27 Analysis of Nucleic Acids
699
Figure 27.14 Principle of the contour-clamped homogenous electric field (CHEF) method. The pulses are applied in different directions. The migration of the DNA is determined by the sum of all applied field vectors and is, as displayed, a zigzag pattern.
The electric field is applied in such a way that the field vectors are aligned at angles of 60° and +60° relative to the vertical axis of the gel. The resulting movement of the nucleic acids resembles a zigzag pattern. The angle, field strength, and duration of pulses can be varied. With this method molecules up to 2000 kb can be separated. An improved version of the CHEF method is PACE (programmable autonomously controlled electrode). Twenty-four hexagonally arranged electrodes can perform any desired pulse sequence. Improvements of pulse sequences result in optimal resolution and separation properties of the gel. Applications For the separation of high molecular weight DNA the integrity of the nucleic acids is of upmost importance. To avoid any destruction of the DNA, the material is packed into agarose blocks before lysing the cells with detergents and proteinase K. High molecular weight DNA is isolated within the agarose block by incubating the agarose in the respective buffers. The agarose blocks are then applied to the gel pockets of the PFGE gel (usually 1% agarose gels). With PFGE, higher voltages are applied, resulting in a temperature increase of the running buffer due to the higher currents. The running buffer is diluted TBE (0.5 × TBE) with the addition of glycine. Glycine increases the mobility of DNA without influencing the current. To avoid pH and temperature differences, the buffer is recycled during electrophoresis. The electrophoresis is usually performed in the absence of ethidium bromide, but when separating molecules with sizes smaller than 100 kB the addition of ethidium bromide can increase the resolution efficiency because the dye influences the reorientation of the DNA molecules. The length and type of pulse sequences is very variable for the different types of PFGE and need to be optimized according to the individual conditions. Pulses are between 5 and 1000 s, field strength is usually between 2 and 10 V cm 1. Running times can vary from 10 h up to several days.
Length standards can be high molecular nucleic acids like the genomic DNA of bacteriophage T7 (40 kb), T2 (166 kb), or G (756 kb). Ligation of bacteriophage lambda DNA yields an optimal length standard (Figure 27.15) with multiples of the lambda-DNA (48.5 kb)
Figure 27.15 Common length standards for PFGE gels. The λ-DNA ladder can be generated by ligation of λ-phage DNA. Source: with kind permission of Bio-Rad, Munich.
700
Part IV: Nucleic Acid Analytics
PFGE is commonly used for the analysis of pathogens from food or clinical isolates. Different strains of certain bacteria (e.g., Listeria monocytogenes) can be analyzed according to its origin. Although with PFGE a resolution of 5 Mb is possible, PFGE cannot resolve human chromosomes (>50 Mb). However, using restriction enzymes, mapping and analysis of the human genome is possible. Rare cutters (e.g., NotI, NruI, MluI, SfiI, XhoI, SalI, SmaI (see Section 27.1)) are used. The PFGE gels are then blotted and hybridized. Physical maps of the human genome by PFGE are used for genomic fingerprinting and to analyze chromosomal deletions or translocations. The whole genome mapping method is also used for subtyping pathogenic strains.
27.2.4 Two-Dimensional Gel Electrophoresis Two-dimensional gel electrophoresis (2D gel electrophoresis) is necessary when information obtained by one-dimensional electrophoresis is not sufficient or clear-cut. The high resolution of 2D gels is achieved by repeating the electrophoresis under completely different conditions. Complex nucleic acid mixtures can be separated that cannot be achieved by a single electrophoresis. The nucleic acid mixture is first separated through a standard electrophoresis where the nucleic acids are separated by their molecular weight (first dimension). The gel lane containing the separated nucleic acids is cut out of the gel and applied to a second gel where the nucleic acids are separated under different conditions (second dimension, Figure 27.16). Usually, the electric field of the second dimension is perpendicular to that of the first dimension. Twodimensional electrophoresis can be applied in the analysis of RNA and DNA. Two-Dimensional Electrophoresis of RNA The electrophoresis conditions for RNA differ regarding the concentration of urea, polyacrylamide, and the pH of the two dimensions (Table 27.8). By performing an electrophoresis in the presence (first dimension) or absence (second dimension) of urea (urea shift), the nucleic acid is first separated according to its size (first dimension) and then according to its conformation (second dimension). A change in concentration of polyacrylamide within the two dimensions can separate the RNA molecules by interaction with the gel matrix. Molecules with different conformations can display a similar migration behavior at a given polyacrylamide concentration but will differ when the pore size of the gel is changed. In addition, the concentration of urea, pH, and pore size can be changed simultaneously. The net charge of the RNA molecules will be influenced by lower pH, so that not all RNA molecules are negatively charged. As certain bases are protonated more easily at lower pH the net charge of the whole RNA molecule depends on the sequence. The second dimension is then performed under conditions that separate the nucleic acids according to their molecular weight. Two-Dimensional Electrophoresis of DNA The 2D electrophoresis of DNA can be used for genome mapping, whereby the DNA is first cut with one restriction enzyme and separated. The separated fragments are then cut with a second enzyme and a second electrophoresis is performed. Fragments that are not cut by the second enzyme will be found on the diagonal of the gel; only
Figure 27.16 Principle and practical application of two-dimensional gel electrophoresis. The gel lane is isolated after first dimension electrophoresis and loaded to a second dimension gel. The direction of the second electrophoresis is rectangular with respect to the first dimension.
27 Analysis of Nucleic Acids
701
Table 27.8 Experimental conditions for the 2D gel electrophoresis of RNA molecules. First dimension %PAAa) Urea-shift
Concentration shift
pH/urea/concentration
pHb)
Second dimension Urea (M)
%PAAa)
pHb)
Urea (M)
X
Neutral
5–8
X
Neutral
0
X
Neutral
0
X
Neutral
5–8
X
Neutral
0
2X
Neutral
0
X
Neutral
4
2X
Neutral
4
X
Acidic
6
2X
Neutral
0
a) X represents a certain polyacrylamide concentration. b) The pH range of neutral electrophoresis is 4.5–8.5. Acidic electrophoresis occurs at a pH below 4.5. Neutral gels are usually run at pH 8.3, acidic gels at pH 3.3. A typical polyacrylamide concentration is 10–15%.
fragments cut by the second enzyme will run differently. With new methods, like next generation sequencing, these methods are becoming less important. Two-dimensional electrophoresis has also been applied in mapping the origins of replication and in the analysis of topoisomers of superhelical DNA. The curvature of DNA can also be analyzed by 2D gel electrophoresis. Temperature Gradient Gels A distant variant of the 2D gels is temperature gradient gel electrophoresis (TGGE), where the second dimension is temperature. The electrophoresis is performed in one direction and a temperature gradient is applied perpendicular to the gel (Figure 27.17). For this method one sample has to be loaded to the whole width of the gel. With increasing temperature the DNA is converted into the denatured state. Melting of DNA is a cooperative process accompanied by a drastic reduction of electric mobility. The process is strongly dependent on the sequence of the DNA fragments, since A/T rich regions melt at lower temperatures. Temperature gradients are applied for mutational analysis as this method can resolve single base changes. Using parallel TGGE (temperature gradient parallel to the electric field), many different samples can be analyzed simultaneously. For denaturing gradient gel electrophoresis (DGGE), the temperature gradient is replaced by a chemical-based gradient with increasing concentrations of denaturing reagents in the opposite direction to the electric field. Double-stranded DNA fragments are separated according to their melting properties. Both methods, TGGE and DGGE, are applied in heteroduplex analysis and for the analysis of microorganisms in environmental analytics.
27.2.5 Capillary Gel Electrophoresis Capillary gel electrophoresis (CGE) is mainly applied for analysis of nucleic acid. The advantages of this method lie in its rapidity, small sample volume, higher sensitivity, and high resolution. The theory of CGE is covered in Chapter 12. The separation principle is the migration of negatively charged nucleic acids in an electric field. The separation is performed in a capillary (50–100 μm in diameter, approx. 20–50 cm long) utilizing the sieving effect of the gel matrix. An important difference to the already described slab gels is that in CGE noncrosslinked gels can also be used as sieving material. Crosslinked gels (also referred to as chemical gels) usually consist of polyacrylamide and can be used in the capillary for 30–100 runs. Non-crosslinked gels (referred to as physical gels) can be easily replaced after each run from the capillary by applying pressure, allowing each run to be performed under reproducible conditions with fresh material. Polymers used for physical gels are hydroxypropylmethyl cellulose (HPMC), hydroxyethyl cellulose (HEC), poly(ethylene oxide) (PEO), polyvinylpyrrolidone (PVP), or linear polyacrylamide. The running buffer is similar to that used with slab gels, TBE. For denaturing conditions urea is added. The samples (1–2 μl) are loaded by electrokinetic injection or with pressure (Chapter 12). Injection of the sample and migration in the capillary are strongly dependent on the salt concentration – in most cases the probes are desalted before loading. During electrophoresis high voltages are applied (1–30 kV). The nucleic acids can be detected by UV light in the presence of fluorescent dyes (OliGreen® ,
Topoisomers forms of DNA molecules of identical length and sequence that differ only in their linking number.
Figure 27.17 Principle of temperature gradient gels. The DNA is loaded onto the whole gel width and a temperature gradient is applied perpendicularly to the electric field.
702
Part IV: Nucleic Acid Analytics
SYBR® Green, Section 27.3). The UV-opaque polyimide layer stabilizing the capillary has to be removed at the detection site (Chapter 12). An important application of CGE is the quality control of synthetic oligodeoxy- and oligoribonucleotides as CGE allows resolution of one nucleotide difference. Oligonucleotides of up to 100 nucleotides can be separated and failure sequences (n 1, n 2, etc.) can be detected. Using the area under the curve method, the ratio of the separated nucleic acids and hence the purity of the material can be determined. Figure 27.18 shows an electropherogram of an oligonucleotide. CGE is also used in nucleic acid sequencing. The sequence reactions labeled with four different dyes are separated by CGE and detected with laser-induced fluorescence (LIF). Conformational polymorphisms (SSCP and HA) can also be analyzed by CGE. Figure 27.18 CGE electropherogram of an oligonucleotide. An oligodeoxyribonucleotide 23 bases long was analyzed by denaturing capillary electrophoresis. The main peak is the full-length product and the lower peaks are contaminations with failure sequences (n 1, n 2). By determination of the area under the curves, the ratio of the different products can be calculated. Source: with kind permission of Dr. Bernhard Noll, Qiagen GmbH.
Figure 27.19 Chemical structure of ethidium bromide. Ethidium bromide intercalates preferentially into double-stranded DNA and interacts with the planar heterocyclic rings of the nucleobases. Newly developed fluorescent dyes that irreversibly stain the DNA are TOTO-1 and YOYO-1.
27.3 Staining Methods 27.3.1 Fluorescent Dyes Ethidium bromide Ethidium bromide (3,8-diamino-5-ethyl-6-phenylphenanthridinium bromide) is an organic dye with a planar structure that intercalates to DNA (Figure 27.19). The aromatic rings can interact with the heteroaromatic rings of the nucleobases. Single-stranded DNA or RNA also intercalates ethidium bromide but to a much lesser extent. The intercalated dye is excited by UV light (254–366 nm) and emits orange–red light (590 nm). Binding of the dye to DNA increases the fluorescence (increased quantum yield) so that the staining of the DNA is also
27 Analysis of Nucleic Acids
703
Figure 27.20 Changes of the geometric properties of circular DNA by intercalation of ethidium bromide. The intercalation of ethidium bromide into negative supercoiled DNA is energetically preferred compared to intercalation into the relaxed form of DNA as positive supercoils have to be introduced.
visible in the presence of unbound ethidium bromide. Ethidium bromide can be added to the gel and running buffer during electrophoresis, making post-staining of the gel unnecessary. The fluorescent ethidium cation migrates to the cathode during electrophoresis. When performing longer electrophoresis runs, the running buffer should contain ethidium bromide since smaller, faster migrating fragments will otherwise be stained only weakly. For certain applications (e.g., blotting of the gel), the gels are stained after the electrophoresis. The blotting efficiency of RNA is diminished in the presence of ethidium bromide. Lanes for staining are cut from the gel and stained separately to control quality and size standard.
In agarose gels, approximately 10–20 ng of double-stranded DNA can still be detected. DNA with intercalated dye has a reduced mobility in the gel (approx. 15%). Since ethidium bromide intercalates into the DNA; it is a strong mutagen and should be handled with extreme care. Influence on DNA Geometry Ethidium bromide changes the superhelical density of circular DNA molecules (form I) through a reduction of the negative supercoiling. Topoisomers with negative superhelical density turn into the relaxed form (increase of entropy). This conversion is favored over intercalation of ethidium bromide to give linear DNA fragments for thermodynamic reasons. Further intercalation of ethidium bromide will induce positive supercoiling. This process is less favorable compared to linear DNA (decrease of entropy, Figure 27.20). For CsCl gradient density centrifugation the ethidium bromide concentration should be saturated. All supercoiled DNA is transformed into the conformation with positive supercoiling with a lower amount of ethidium bromide than for the relaxed forms. The positive supercoiled DNA has a lower density than linear (chromosomal) DNA and can be separated from the chromosomal contamination. The intercalation of ethidium bromide can also be used to analyze the conformation of circular DNA. Other Fluorescent Dyes In recent years various intercalating dyes on the basis of asymmetric cyanine substance classes have been developed. These dyes are highly sensitive and less
704
Part IV: Nucleic Acid Analytics
mutagenic than ethidium bromide. A commonly used dye is SYBR® Green with an excitation maximum of 492 nm and a second absorption maximum at 284 nm. The emission maximum is at 519 nm. The dye can be used in fluorescence reader instruments, allowing exact quantification of the nucleic acid solution by comparison with standards. Newly developed variants of this substance class display a higher affinity to singlestranded DNA or RNA (e.g., SYBR® Green II, OliGreen® ). The dyes do not specifically bind to single-stranded nucleic acids but the quantum yield is drastically increased when interacting with single-stranded RNA. Other dyes, like TOTO-1 and YOYO-1, bind with much higher affinity to DNA than ethidium bromide. This can be used for certain applications when higher sensitivity is a decisive factor. Some of the new dyes (e.g., TOTO-3, YOYO-1, JOJO-1) cannot be excited with UV light and can only be used with laser induced fluorescence (LIF).
27.3.2 Silver Staining Silver stains can only be performed with polyacrylamide gels as agarose gels yield too high a background staining. The gels need to be poured using high quality reagents and should be handled with extreme care as all protein or nucleic acid contamination will result in high background staining.
Silver staining is a less commonly applied method for the detection of nucleic acids. The advantage of this method, as for proteins, is its sensitivity. Very small amounts of nucleic acids (up to 0.03 ng mm 2) can be detected. There is no need for mutagenic or radioactive detection reagents. The method is time consuming and background staining can be high. Silver staining is based on the change of redox potential in the presence of nucleic acids (or proteins). The reduction of silver nitrate to silver is catalyzed. The metallic silver precipitates on the nucleic acids if the redox potential is higher than that in the surrounding solution. These conditions can be achieved through the choice of buffer and reagents. Recent analysis found that the purine bases account for this reaction.
27.4 Nucleic Acid Blotting 27.4.1 Nucleic Acid Blotting Methods Electroblotting, Section 11.7
For further analysis of the nucleic acids they are separated by gel electrophoresis or transferred to a membrane. The fixed nucleic acids can be identified and analyzed by hybridization with labeled probes of known sequence. The nucleic acids can be transferred to a membrane by various methods: capillary blotting, vacuum blotting, and electroblotting. Figure 27.21 displays the principle underlying each method. While blotting and subsequent hybridization with certain labeled probes was once the first choice to analyze nucleic acids (e.g., to detect chromosomal rearrangements, mutations, etc.), newer techniques (e.g., PCR, next generation sequencing) are nowadays applied more frequently than nucleic acid blotting, although blotting remains a useful, fast, and inexpensive way for nucleic acid analysis. Capillary blotting can be performed with the least technical effort. The nucleic acids are transferred by capillary forces to the membrane using paper towels, through which the blotting buffer is soaked through the gel and membrane. Vacuum blotting is performed using a vacuum chamber with the membrane attached and the nucleic acids are transferred through the gel. Capillary and vacuum blotting can only be performed with agarose gels. The nucleic acids will not move from polyacrylamide gels due to the lower pore size. For polyacrylamide gels, electroblotting systems are used. The nucleic acids are transferred to the membrane using an electric field. Electroblotting can be performed using a tank filled with buffer or by semi-dry blotting where the membrane and the filter are in contact with wet filter paper.
27.4.2 Choice of Membrane For nucleic acid blotting two types of membranes are used: nitrocellulose and nylon membranes. Nitrocellulose has long been used but is increasingly being replaced by nylon (or poly
27 Analysis of Nucleic Acids
705
Figure 27.21 Schematic drawing of the different blotting techniques.
(vinylidene fluoride), PVDF) membranes with improved handling and binding properties. Table 27.9 gives an overview of membranes commonly used and their properties. The nucleic acids are bound covalently to the surface of the nylon membrane, fixing them better to the material. The filters can be used several times. The binding to nitrocellulose is non-covalent. The advantages of nylon membranes are manifold: higher stability, higher binding capacity, and better fixation. Nitrocellulose is harder to handle and is more fragile.
Nylon and PVDF membranes can yield a strong background signal, which can be reduced by the choice of suitable blocking reagents. If the membrane needs to be hybridized more than once, nylon membranes are the material of choice.
27.4.3 Southern Blotting In 1975, E. Southern was the first to immobilize DNA separated by gel electrophoresis to a nitrocellulose membrane. Since then, the transfer of DNA from gels to a membrane has been called Southern blotting. Table 27.9 Properties of different blotting membranes. Source: according to Ausubel, F.M., Brent, R.E., Moore, D.D., Smith, J.A., Seidman, and J.G., Struhl, K. (1987–2005) Current Protocols in Molecular Biology, John Wiley & Sons, Inc., New York. Property
Nitrocellulose
Improved nitrocellulose
Neutral nylon membrane
Positively charged nylon membrane
Application
ssDNA, RNA, proteins
ssDNA, RNA, proteins
ssDNA, dsDNA, RNA, proteins
ssDNA, dsDNA, RNA, proteins
Binding capacity (μg nucleic acid cm 2)
80–100
80–100
400–600
400–600
Type of nucleic acid binding
Non-covalent
Non-covalent
Covalent
Covalent
Size restrictions for transfer
500 nt
500 nt
50 nt or bp
50 nt or bp
Resilience
Bad
Good
Good
Good
Multiple hybridizations
Bad (become brittle)
Bad (loss of signal intensity)
Good
Good
706
Part IV: Nucleic Acid Analytics
Southern blotting can be described by three main steps: preparation of the gel, transfer, and immobilization of the DNA on the membrane. The efficiency of the transfer can be strongly improved by partial depurination of the DNA in the gel. The gel is treated with dilute hydrochloric acid, whereby purines are separated from the DNA backbone. When the DNA is denatured, either within the gel or during blotting, the phosphodiester bonds on the apurinic nucleotides are broken leading to a fragmentation of the DNA during transfer. The procedure is necessary especially for large DNA fragments (>10 kB). However, the fragments should not be too small because small fragments are not fixed efficiently to the membrane. The denaturing step is dependent on the type of nucleic acid to be transferred. For genomic DNA denaturing is essential. Usually, the gels are not stained with ethidium bromide to avoid gel and separation artifacts. If nitrocellulose is used for blotting, the transfer buffer has to be high in salt for efficient binding to the membrane. Usually, 20 × SSC buffer is used containing sodium chloride and sodium citrate. If a denaturing step was performed, the gel needs to be neutralized before blotting as DNA does not bind to nitrocellulose at above pH 9. For nylon membranes, the DNA can also be transferred using 20 × SSC buffer (with a previous denaturing step) or the denaturing step can be performed during blotting using an alkaline transfer buffer (e.g., 0.4 M sodium hydroxide) and sodium chloride. Capillary blots are usually transferred overnight whereas vacuum blots are performed within 1–2 h. The DNA then needs to be immobilized on the membrane. The crosslinking of the DNA to the membrane can be performed using UV light. The thymidine bases are covalently linked to the amino groups of the nylon membrane. The duration and strength of the UV crosslinking needs to be optimized as crosslinking that is too strong makes most of the thymidine bases unavailable for hybridization whereas weak crosslinking leads to loss of signal intensity. If the transfer is performed using alkaline buffer, immobilization of the DNA to the nylon membrane is not necessary. With nitrocellulose membranes, the DNA is non-covalently bound by incubation of blotted membranes by temperature (not higher than 80 °C and not longer than 2 h, as the nitrocellulose can ignite). For the transfer of DNA out of polyacrylamide gels, only electroblotting can be performed using nylon membranes. Southern blotting can be used for genomic analysis where the genomic DNA is cut with different restriction enzymes, separated, and blotted. The DNA can be analyzed for a specific hybridization pattern using known gene probes. It can also be used to detect gene families or single-copy genes. With increasing knowledge of the human genome by deep sequencing, genomic analysis by Southern blotting is becoming less common. Southern blotting techniques are and have been used to detect similarities between different species (“Zoo blot”). The genomic DNA of different species (or strain subtypes) is cut with a restriction enzyme and hybridized with the probe of interest.
27.4.4 Northern Blotting According the nomenclature of Southern blotting for the transfer of DNA to membranes, the transfer of RNA to membranes is termed Northern blotting (and the transfer of proteins is called Western blotting). Like electrophoresis methods, blotting techniques need to account for the different properties of DNA and RNA. Since RNA is often separated in denaturing gels, the denaturing step preceding the blotting is not necessary. However, the denaturing reagents used during electrophoresis need to be removed by soaking of the gel in dilute sodium hydroxide solution or by incubation of the filter at higher temperatures. If high molecular weight RNA molecules need to be transferred, the gel is incubated briefly in sodium hydroxide (0.05 M) to partially hydrolyze the RNA for easier transfer. Nitrocellulose membranes are more often used for Northern blotting than for Southern blotting. RNA gels are usually blotted in 10× or 20 × SSC buffer. An alkaline transfer is possible but only with a very low concentration of sodium hydroxide (7.5 mM). RNA gels are blotted by vacuum or capillary blotting, but the transfer requires more time (usually two days for a capillary blot). The RNA is immobilized similarly to DNA on the membrane. Northern blots are used to study the expression of certain genes in different cells or tissue using total RNA or poly-adenylated mRNA purified by oligo d(T) affinity chromatography. With low expression genes, the use of purified poly(A) mRNA is mandatory. To detect
27 Analysis of Nucleic Acids
707
Figure 27.22 Scheme of the setup of a slot and dot blotting unit.
differences in the expression levels of mRNA of a certain gene, it is necessary to load equal amounts of RNA to the gel. To control the amount of RNA loaded and blotted to the gel, the blot is hybridized again with a probe for a gene that is equally expressed in all tissue or cells (housekeeping genes, usually glucose-6-phosphate dehydrogenase, β-tubulin, or β-actin mRNA). The signal strength obtained with the housekeeping probe should be similar for each of the different samples. A similar analysis is performed with RT-PCR (Chapter 29) and is usually the first choice method as Northern blots are not as quantitative as RT-PCR and require more hands-on time.
27.4.5 Dot- and Slot-Blotting Dot- and slot-blotting are simple applications of membrane hybridization. The nucleic acids to be analyzed are blotted to the filter without prior separation by electrophoresis. The transfer is performed in dot-blotting units (Figure 27.22). With this set-up a large number of samples can be analyzed simultaneously. Dot- and slot-blots are used to analyze a large sample number for the presence or absence of a certain nucleic acid sequence.
27.4.6 Colony and Plaque Hybridization A variant of the blotting technique is the generation of so-called colony or plaque filters for screening of cosmid or phage libraries. Colonies of bacteria grown on agar plates (colony hybridization) or phages (plaque filters) are transferred to membranes for hybridization with a certain probe. A huge number of colonies or phages can be screened for the presence of the sequence of interest. For colony or plaque hybridization, the stable nylon membranes are used. The membranes are transferred to the colony plate and colonies are lifted to produce exact copies of the agar plate. The direction of plate and filter need to be marked exactly to ensure that the colonies on the membrane can be allocated later to the colonies on the plate. Bacteria and phages are lysed using a denaturing solution (containing sodium hydroxide and sodium
708
Part IV: Nucleic Acid Analytics
chloride). The DNA is fixed to the membrane and the remaining, contaminating RNA hydrolyzed. Filters are neutralized and treated with UV light. To avoid hybridization artifacts, each agar plate is lifted twice and only those hybridization signals that can be detected on both membranes are considered positive.
27.5 Isolation of Nucleic Acid Fragments To isolate pure DNA fragments, the fragments need to be separated with high resolution. This can be achieved by optimization of agarose concentration and electrophoresis conditions. The gel should not be overloaded as resolution will decrease. Chaotropic agents Certain ions can disrupt hydrogen bonding and destroy chemical structures. These ions are negatively charged with large diameters and low charge density. Examples are I , ClO4 , and SCN .
DNA fragments of a defined sequence are the basis for various methods. Isolation methods for DNA fragments depend on the type of application they are needed for. Isolation of DNA from polyacrylamide gels will yield very pure material but the separation range and isolation efficiency are lower than for agarose gels. Isolation of fragments from agarose gels can result in contamination with polysaccharides, depending on the quality of the agarose. Very large DNA fragments (>5 kb) as well as small amounts of DNA are only isolated from an agarose gel with low efficiency. For preparative approaches (e.g., isolation of plasmid vectors), the DNA digested with restriction enzymes is loaded to several gel pockets. After separation the DNA fragment is cut out from the gel using a scalpel. The fragments are detected by ethidium bromide staining and UV light. For isolation of the fragments it is important to use long wavelength UV light and to keep the exposure time as short as possible to avoid damage of the DNA and crosslinking to the gel matrix. The described methods can also be used to isolate or purify fragments that have not been separated previously by slab gels.
27.5.1 Purification using Glass Beads Most commercially available DNA fragment isolation kits use glass beads. DNA can be bound to the glass surface in the presence of chaotropic salts (e.g., lithium acetate or sodium perchlorate). Hydrogen bridges within the agarose polymer are destroyed by high concentrations of chaotropic salts and the gel matrix is dissolved. At these salt concentrations, the DNA will adsorb to the silica surface of the glass beads. The adsorption is strongly pH dependent. Some protocols use pH indicators to ensure the correct pH (must be below 7.5 for optimal binding). The glass beads are either centrifuged or handled in columns. The adsorbed DNA is eluted from the glass beads after several washing steps to remove residual agarose and salt using low salt buffer at high pH (TE usually works well). With this method, fragments with sizes between 40 bp and 50 kB can be isolated from 0.2–2% agarose gels. Higher molecular weight fragments (>4 kB) need more time and higher temperatures for isolation from the glass beads. The fragment yield will depend on the size of the DNA fragment. If the fragments are isolated from TBE gels, monosaccharides need to be present to chelate the borate anions. The method can also be used with some modifications for the isolation of DNA from polyacrylamide gels.
27.5.2 Purification using Gel Filtration or Reversed Phase The principle of gel filtration has been discussed in Chapter 26.1.2. For fragment isolation this method is used to separate (radioactive labeled) nucleotides from reactions or to desalt the nucleic acid solutions. The choice of column material is dependent on the fragment size. The method is also frequently used for the purification of PCR fragments, to remove primers, nucleotides, and fluorescent dyes. Gel filtration methods are available for 96-well format.
27.5.3 Purification using Electroelution This method is based, similarly to electrophoresis, on the migration of nucleic acids within the electric field. The method requires a higher instrumental effort and is less frequently used than commercially available kits, but if an instrument is available electroelution is cheaper than gel filtration columns. With the simplest experimental set up, the agarose gel is captured in a dialysis tube containing electrophoresis buffer and submitted to an electric field. The DNA will migrate
27 Analysis of Nucleic Acids
709
according to the field out from the agarose piece and is trapped in the dialysis tube. Dependent on the type of electroelution instrument, the technical set up can vary; however, the DNA will always concentrate in the direction of the anode.
27.5.4 Other Methods Oligonucleotides can be purified using denaturing polyacrylamide gels (Section 27.2.1). It is possible to separate n-mers from failure sequences (n 1) or (n 2). The oligonucleotide band can be visualized using fluorescence quenching. Efficient elution of the cut bands can be achieved by incubation of the small cut polyacrylamide gel pieces in sodium acetate solution. The oligonucleotides will diffuse into the solution. If DNA fragments of low purity and yield are needed (e.g., for simple cloning experiments), agarose gel pieces can be centrifuged through silanized glass wool. The agarose matrix will be disrupted by the centrifugal forces and held back by the glass wool. Simple centrifugation of the gel pieces in a tube can also yield enough DNA fragment in the supernatant above the agarose pellet. These methods are not suitable if high purity and high yield fragments are needed.
27.6 LC-MS of Oligonucleotides Markus Weber1 and Eugen Uhlmann2 1 2
Currenta GmbH & Co. OHG, Chempark Q18, 51368 Leverkusen, Germany iNA ImmunoPharm GmbH, Zentastrasse 1, 07379 Greiz, Germany
27.6.1 Principles of the Synthesis of Oligonucleotides Synthetic oligonucleotides and their derivatives are important tools in molecular biology and in the development of new types of drugs, in particular antisense oligonucleotides, siRNAs, aptamers, antagomirs, and CpG adjuvants. These days their synthesis takes place on gram and kilogram scales, primarily by the phosphoramidite method on a solid phase (Figure 27.23). The step-by-step, computer-controlled synthesis takes place in the 3´ to 5´ direction. The first nucleoside residue is bound to the solid phase support (organic polymer or controlled pore glass) by its 3´ -hydroxyl group and a base labile succinic acid. Orthogonal protective groups, like the acid labile 5´ -O-dimethoxytrityl (DMT) protective group and base labile protective
Figure 27.23 Schematic representation of the reaction cycle of oligonucleotide synthesis according to the phosphoramidite method on a solid phase (TBDMS: tbutyldimethylsilyl, DMT: dimethoxytrityl).
710
Part IV: Nucleic Acid Analytics
Figure 27.24 Possible by-products of the oligonucleotide (phosphorthioate) synthesis.
Protective groups Temporarily introduced groups that enable a molecule with several reactive functions to selectively react at only one of these functions. Diastereomers Substances with several chiral centers that are not mirror images of one another and are therefore different compounds with different physical properties. For example, the phosphorothioate modified oligonucleotides have, besides the chiral β-Ddeoxyribose, a further chiral center on the phosphate.
Figure 27.25 Depurination reaction (iBu: isobutyryl).
groups on the nucleobases and on the phosphate residue, allow the targeted exposure of reactive functions. In the first step, the DMT group is removed by treatment with diluted tri- or dichloroacetic acid. The free hydroxyl group is converted into a trivalent phosphite ester in a condensation reaction with 5´ -O-DMT-nucleoside-3´ -phosphoramidite, catalyzed by tetrazole. These esters can then be oxidized to a phosphotriester with iodine or to a thiophosphotriester with a sulfurizing agent like the Beaucage reagent. After complete chain synthesis and removal of the protective groups, the oligonucleotides can be phosphodiesters, phosphorothioates, or mixed backbone analogs, depending on the reagents employed. Since the substitution of an oxygen atom with a sulfur atom creates a chiral phosphate, in the course of making phosphorothioates a mixture of 2n diastereomers results, where n is the number of internucleotide bonds. For example, a 20-mer oligonucleotide with 19 phosphorothioate modifications on the internucleotide bonds consists of 524 288 diastereomers. An acylation capping reaction is used to prevent excess 5´ -hydroxy components from reacting in subsequent cycles of the coupling reaction. After multiple repetitions of the reaction cycle, corresponding to the length and composition of the desired sequence, the oligonucleotide is removed from the solid-phase column with concentrated ammonia and the protective groups are removed. RNA synthesis differs from DNA synthesis only in as far as an additional protective group is required for the 2´ hydroxyl group. Often, the t-butyldimethylsilyl (TBDMS) protective group is used, which is stable during the synthesis and can be removed in the very last step with triethylammonium fluoride. Although the cycles of the phosphoramidite method operate with a very high yield of 98–99%, by-products can be present that are either the product of failed reactions during chain assembly or result from the final reactions to remove the protective groups (Figure 27.24). Since the coupling reactions do not operate with 100% efficiency, not only are oligonucleotides of the full length (N) present but also those of shortened lengths (N 1, N 2, N 3, etc.), which are missing one or more nucleotides. Interestingly, the reactions can also result in a nucleotide of greater than the expected length (N + 1). These arise from a double addition during the tetrazolecatalyzed coupling reaction as a result of a minor cleavage of the acid labile DMT group, either on the monomer or on the growing chain. This side reaction happens most frequently during the condensation of deoxyguanosine, whose 5´ -DMT group is the most labile of the four bases, due to the slightly acidic nature of the tetrazole catalyzed coupling reaction. In the case of incomplete sulfurization during the synthesis of phosphorothioates, reaction products contain a phosphodiester bond (mono-phosphodiester) in addition to the expected phosphorothioate. Another side reaction during the synthesis of purine-containing sequences is depurination (Figure 27.25), which takes place in an acidic environment. This refers to the hydrolysis of the N-glycoside bond between the nucleobase and deoxyribose due to protonation of the purine base at the 7-position. Possible side reactions of the deprotection are either the incomplete removal of the protective groups, such as the isobutyryl protective group on the exocyclic amino function of guanine, or the production of acrylonitrile adducts. The latter can come about
27 Analysis of Nucleic Acids
711
Figure 27.26 Formation of cyanoethyl adducts during the removal of the cyanoethyl protective groups with concentrated ammonia.
after β-elimination of the 2-cyanoethyl phosphate protective group and subsequent base-catalyzed addition of the acrylonitrile to the N3 of the thymine base (Figure 27.26). The by-products of oligonucleotide synthesis, due to their complexity, can only be partially separated by subsequent ion exchange or reversed-phase chromatography and are therefore only detectable by suitable analytical methods, such as the LC-MS method described in the following section.
27.6.2 Investigation of the Purity and Characterization of Oligonucleotides While traditionally oligonucleotides with the natural phosphodiester internucleotide bond were of primary interest, more recently modified oligonucleotides now play a greater role, particularly in the area of therapeutic applications. The demands on the capabilities of the analytical methods have increased dramatically due to the increasing use of these modified oligonucleotides in recent years. Methods like polyacrylamide gel electrophoresis (PAGE), capillary gel electrophoresis (CGE), and anion exchange high performance liquid chromatography (HPLC) have been the mainstay of analytical techniques in the past, but the investigation of synthetic oligonucleotides, in particular, makes the use of HPLC-MS methods increasingly important. The online coupling of HPLC with a detection method based on mass spectrometry (MS) results in extremely powerful and conclusive HPLC-MS methods. A successful LC-MS analysis requires the best possible separation of the analytes with a HPLC method. Furthermore, it must be ensured that the HPLC method is compatible with the subsequent MS methods, which represents a significant hurdle in the development of new LC-MS techniques. Electrospray ionization mass spectrometry (ESI-MS) is the method of choice. It, however, requires the use of volatile buffer systems in the preceding HPLC. The direct infusion of the oligonucleotide to be investigated without prior HPLC purification is greatly complicated by the formation of cation adducts that are a product of the high affinity of oligonucleotides for sodium and potassium ions. Greig and Griffey have shown that the addition of strong bases like triethylamine (TEA) or piperidine strongly reduces the formation of adducts and thereby increases the sensitivity of ESI detection. For the investigation of complex mixtures by ESI-MS, a separation of the analytes by HPLC is essential. It turned out, however, that the mobile phases that led to a good separation of the analytes inhibited the ionization of the electrospray. As a result of the weakly hydrophobic character and the polyanionic nature of oligonucleotides they are ill suited to conventional reversed-phase (RP) HPLC. Ion pair reagents, which strengthen the interaction between the analytes and the stationary phase of the column, are therefore employed during the HPLC separation of oligonucleotides (Figure 27.27). Triethylammonium acetate (TEAA) and tetrabutylammonium bromide (TBAB) are two IP reagents that are frequently used during the separation of oligonucleotides by IP RP-HPLC. TBAB is, however, not volatile and can therefore not be used in combination with electrospray ionization. Although TEAA is a volatile ion pair reagent, it negatively impacts the sensitivity of the MS detection. The concentration of TEAA normally required for efficient separation leads, in general, to a significant loss of sensitivity of the MS detection. Apffel et al. were the first to use hexafluoroisopropanol/
712
Part IV: Nucleic Acid Analytics
Figure 27.27 Schematic representation of the mechanism of separation of ion pair reversed-phase HPLC. Reversed phase chromatography means that the stationary phase is less polar than the eluent mixture. Typical stationary phases are porous silica gels, which have alkyl groups of various lengths bound to their surface. The chain length of the alkyl residues determines the hydrophobicity of the stationary phase. Most often C18 or C8 alkyl chains are used.
triethylamine (HFIP/TEA) as an ion pair reagent and thereby achieved a high efficiency of the HPLC separation while maintaining the high sensitivity of MS detection and low adduct formation. Apffel attributed the increased MS sensitivity of the HFIP/TEA buffer to the different boiling points of HFIP, acetic acid, and TEA. Since acetic acid (boiling point 118 °C) has a higher boiling point than TEAA (89 °C), TEA evaporates preferentially, leading to a decrease in the pH of the HPLC eluent during the electrospray process. The drop in pH leads to protonation of the negatively charged oligonucleotide and therefore to a decrease in the sensitivity of the MS detection. During the desolvation of a HPLC eluent from HFIP/TEA buffer and analytes, in contrast, the HFIP preferentially evaporates, which leads to an increase of the pH value and therefore a deprotonation of the phosphate groups in the oligonucleotide backbone. The resulting negatively charged oligonucleotide can be evaporated into the gas phase during the ESI process and can be detected with high sensitivity. Gilar et al. further optimized the HFIP/TEA buffer system originally introduced by Apffel et al. for the separation of oligonucleotides by HPLC and thereby facilitated its broad adoption in the research and development of oligonucleotides. Depending on the nature of the column and eluents, even nonpolar and polar analytes can be separated by RP-HPLC. If the polar characteristics are very pronounced, such as with oligonucleotides, a separation by normal RP methods is not possible. To increase the reactivity and thereby also the affinity of the polar substances to RP phases, ion pair reagents are generally used. These are characterized by undergoing a hydrophobic interaction with the RP phase and a charged interaction with the analyte. Ion pair reagents are used for the separation of oligonucleotides, which use alkyl residues to attract ammonium ions to increase the interaction between the analytes and the RP phase. Besides charged interactions, hydrophobic interactions between the RP phase and the hydrophobic bases of the oligonucleotides occur, which contribute to the total retention of the analytes. The separation efficiency of an ion pair method for the separation of oligonucleotides is primarily determined by the lipophilic nature of the ammonium cation. In addition, the counter-ion also has an impact on the separation. Gilar explains the high separation efficiency of the HFIP/TEA buffer as a result of the decreased solubility of protonated TEA molecules in HFIP, compared to acetic acid, which increases the surface concentration of the cation on the RP phase.
27.6.3 Mass Spectrometric Investigation of Oligonucleotides HPLC systems coupled to mass selective detectors use electrospray ionization in which the analytes, dissolved in the separation buffer used to elute them from the HPLC column, are injected via a capillary into the ion source. Under normal atmospheric pressure, an electric field of several kilovolts is applied to the LC capillary as the ion source passes through it to form a
27 Analysis of Nucleic Acids
713
Figure 27.28 Schematic representation of the LC-MS analysis of a complex mixture of three components, of which two (2a and 2b) cannot be separated by chromatography.
finely dispersed spray of highly charged solvent droplets with a diameter of a few micrometers. The analysis of oligonucleotides, which due to their phosphate backbone form very slightly negatively charged molecular ions, takes place in negative ion mode, in which the LC capillary receives a positive charge. The ionization is particularly effective when the oligonucleotide is already in a deprotonated form due to the use of a suitable HPLC buffer system. This can be achieved by the use of buffers that have an alkaline pH value during the electrospray. The ions in the solvent droplets move into the gas phase through the process of desolvation, during which, dependent on the molecular weight of the oligonucleotide, primarily multiply charged ion molecules are formed. The charge distribution is determined mainly by the molecular weight of the analyte, but may also be influenced by the type of HPLC buffer, as well as the device parameters. A fragmentation of the analytes is not observed due to the low thermal load during the ionization as part of the electrospray procedure. After the transfer of the analytes into the gas phase, the mass analysis of the ions takes place in an ion mass spectrometer. For the analysis of oligonucleotides, HPLC coupling offers the advantage that complex substance mixtures can be investigated in a relatively simple manner. In addition, the chromatographic purification offers the possibility of an almost complete removal of salts, which would otherwise inhibit the electrospray process. By coupling HPLC with the ESI-MS one receives, in addition to the UV chromatogram, further chromatographic data, the so-called total ion current (TIC) chromatogram, which usually correlates well with the UV chromatogram detected at 260 nm. The mass spectrometric detection allows the visualization of a mass spectra at every point in the TIC chromatogram (Figure 27.28). The electrospray ionization of oligonucleotides generally leads to the formation of a series of multiply charged ions, which carry a variable number of negative charges in their backbones. Therefore, in the mass spectra of an oligonucleotide there are always a series of ion signals that differ by exactly one charge, z. As a result, for a molecule of mass m, a series of different values of m/z are always detected. Since ESI-MS does not detect the mass directly, but instead the ratio m/z, multiply ionized molecules with a relatively high mass can still be measured. The intensity of an ion signal always depends on the statistical probability that the corresponding ion is formed during the electrospray process. In the ideal case the intensity distribution forms a Gaussian distribution curve. The actual form of the intensity distribution and the position of the maximum are, however, dependent on the choice of MS parameters, since these can strongly influence the transmission of individual ions. The formation of ion series prevents the direct determination of the molecular weight from the mass spectra of the oligonucleotides. The molecular weight can be obtained, however, by deconvolution of the data using the measured m/z values of the individual charged states. The mass spectra recorded by an ion trap in full scan mode can be used to generate an extracted ion chromatogram (EIC). This involves using the individual m/z values of an ion series of a compound to calculate a chromatogram trace, which can be used to see at what time point a particular component is eluted. An EIC can be used to generate
714
Part IV: Nucleic Acid Analytics
Gauss (or normal) distribution – a symmetrical distribution in the form of a bell curve, with which many random processes in nature can be described.
chromatogram traces of co-eluting substances of differing molecular weights. These can then be used like any normal UV or TIC chromatogram to quantify substances; this means that even substances that cannot be separated by HPLC can be quantified, provided they are of differing molecular weights.
27.6.4 IP-RP-HPLC-MS Investigation of a Phosphorothioate Oligonucleotide This section explains the investigation of a synthetic oligonucleotide using IP-RPHPLC-MS by using the example of a 20-mer phosphorothioate oligonucleotide. There are no other chemical modifications besides the phosphorothioate modification of the oligonucleotide backbone. Table 27.10 shows the sequence of the phosphorothioate. The objective of IP-RP-HPLC-MS based LC-MS analysis is to identify by-products of the oligonucleotide synthesis based on their molecular weight and therefore to draw conclusions about the purity of the main product of the synthesis. The UV or TIC chromatograph shown in Figure 27.29 provides an overview of the number of contaminants present. In addition to the Table 27.10 Identification of the by-products from the synthesis of the 20 mer phosphorothioate oligonucleotide based on their molecular weight. The by-products, which make up less than 0.5% of the total UV trace, are not discussed here. ID
MW (Da)
AMW (Da)
MWBer. (Da)
Identificationa)
% UV
A
6665.6
0
6665.4
N
78.1
B
6532.7
132.9
6532.2
N-Guanine + H2O
C
6649.8
15.7
6649.3
Nox
D
6320.3
345.3
6320.1
N
E
7010.8
+345.2
7010.6
N+G
5.3
F
6719.1
+53.5
6718.4
N + CE
1.8
G
5974.9
690.7
5974.8
N
2G
2.4
H
5629.6
1036.0
5629.5
N
3G
0.8
a) Nox: Mono phosphodiester. N-Guanine + H2O: depurinated. N + CE: cyanoethyl adduct. Sequence of the desired oligonucleotide: G∗G∗G∗G∗G∗A∗G∗C∗A∗T∗G∗C∗T∗G∗G∗G∗G∗G∗G∗G. where ∗ represents a phosphorothioate-internucleotide bond.
Figure 27.29 Ion pair RP-HPLC-separation of a 20-mer phosphorothioateoligonucleotide. The total ion current (TIC, - - -) of the ESI detection corresponds to conventional UV (___) detection at 260 nm.
G
11.6
27 Analysis of Nucleic Acids
715
Figure 27.30 ESI spectra of the main components (A) of the chromatographic separation. The molecular weight of components (A) is determined by deconvolution of the ion series (A).
contaminants separated by HPLC, other contaminants that are not separable by chromatography can be present, which cannot be identified on the basis of the UV or TIC traces. Provided they are of differing molecular weight, they can be detected by ESI mass spectrometry and their content estimated via the EIC method described earlier. With the aid of mass spectrometric detection, which is carried out in addition to conventional UV detection, it is possible to determine the molecular weight of compounds eluted from the HPLC into the mass spectrometer. Figure 27.30 shows a typical example of a measured MS spectrum of the desired main component (A). It shows a series of signals of the negatively charged ions typical of oligonucleotides. In this case, five signals were measured, which can be attributed to species with a differing number of negative charges (6–10) in the phosphorothioate backbone. Depending on the measurement conditions, a different number of charged states may be found in the primary spectrum. By deconvolution the molecular weight of the analyte can be calculated from the m/z values of an ion series. Here a molecular weight of 6665.6 Da was measured, which is sufficiently close to the calculated mass of the desired oligonucleotide of 6665.4 Da. The molecular weight of two analytes (compare compounds E and F in Figure 27.29) incompletely separated by HPLC chromatography are easily determined based on their mass spectra (Figure 27.31). The series of the MS signals of the components of E correlate due to their higher signal intensity with the signal of the higher UV intensity of the double peak (E), (F) in the UV lane. Using the higher intensity signals of series E, a molecular weight of 7010.8 Da is calculated, which corresponds to an oligonucleotide extended by one nucleotide (N + G) (Table 27.10). The signals of the series F result in a mass of 6719.1 Da after deconvolution, which correspond to the cyanoethyl adduct of the oligonucleotide (N + CE). The signals E and F, which could not be completely separated chromatographically, can be unequivocally assigned to the two by-products N + G (E) and N + CE (F) (Table 27.10). On the basis of these simple considerations, it is possible to determine the molecular weight of components A and D to H without great difficulty (Figure 27.29). The by-products B and C of the synthesis cannot be separated from the main product A of the reaction by HPLC; this means that all three compounds appear in the UV or TIC detection traces as a single signal. In this case it is, however, possible to use the mass-dependent detection of these components to identify them. Just ahead of the main components in the mass spectra, beside the signal from the main component A, there are two more ion series: B and C (Figure 27.32). Deconvolution allows determination of the molecular weights of the by-products B and C, which coelute with the target compound A. In this case, besides the desired oligonucleotide N of mass 6665.4 Da, two
716
Part IV: Nucleic Acid Analytics
Figure 27.31 ESI spectra of the chromatographically incompletely separated components (E) and (F).
other by-products of masses 6532.7 Da (B) and 6649.8 Da (C) are detected, which correspond to the depurinated product (B) and the mono phosphodiester Nox of the oligonucleotide (C). The m/z value of the differentially charged analytes of the ion series A, B, and C can also be used to extract an ion chromatogram. The EIC method is particularly helpful in this case, since the by-products of the synthesis (ion series B and C) cannot be separated chromatographically from the main product A. In this manner the by-products B and C, which coelute with the main product, can be quantified (Figure 27.33). The mass spectrometrically determined molecular weight of the main product A can be used to make a comparison with the expected molecular weight to confirm its identity. In addition, in many cases the difference in molecular weight between the main product and by-products can
Figure 27.32 ESI spectra from the leading edge of the main components detect three components of differing molecular weight.
27 Analysis of Nucleic Acids
717
Figure 27.33 Extracted ion chromatograms (EIC) of the components A, B, and C. By integration of the chromatogram traces, the relationship of the co-eluting compounds A, B, and C to one another can be determined.
be used to determine the identity of the by-products. (Table 27.10). In this example, what appears in the UV-HPLC chromatogram as a single uniform peak (A, B, C) (Figure 27.29) is revealed by the EIC trace to be a mixture of the desired main product A (88%), as well as the depurinated product B (7.5%) and the mono phosphodiester C (4.5%) (Figure 27.33).
Further Reading Section 27.1 Ausubel, F.Μ. Brent, R., Kingston, R.E., Moore, D.D., Smith, J.A., Seidman, J.G., and Struhl, K. (1987–2005) Current Protocols in Molecular Biology, John Wiley & Sons, Inc., New York. Roberts, R.J. et al. (2003) A nomenclature for restriction enzymes, DNA methyltransferases, homing endonucleases and their genes. Nucleic Acids Res., 31, 1805–1812. Roberts, R.J., Vincze, T., Posfai, J., and Macelis, D. (2010) REBASE – a database for DNA restriction and modification: enzymes, genes and genomes. Nucleic Acids Res., 38, D234–D236.
Section 27.2–27.5 Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Smith, J.A., Seidman, J.G., and Struhl, K. (1987–2005) Current Protocols in Molecular Biology, John Wiley & Sons, Inc., New York. Chrambach, A., Dunn, M.J., and Radola, B.J. (eds) (1987) Advances in Electrophoresis, Volume 1, VCHVerlagsgesellschaft, Weinheim. Darling, D.C. and Brickell, E.M. (1994) Nucleinsäure-Blotting. Labor im Fokus, Spektrum Akademischer Verlag, Heidelberg. Glasel, J.A. and Deutscher, M.E. (1995) Introduction to Biophysical Methods for Protein and Nucleic Acid Research, Academic Press, New York. Grossman, L. and Modave, K. (eds) (1980) Nucleic Acids: Part I, Methods in Enzymology, vol. 65, Academic Press, New York. Hafez, M. and Hausner, G. (2012) Homing endonucleases: DNA scissors on a mission. Genome, 55, 553–569. Krieg, E.A. (ed.) (1996) A Laboratory Guide to RNA: Isolation, Analysis and Synthesis, Wiley-Liss, New York. Martin, R. (1996) Elektrophorese von Nucleinsäuren, Spektrum Akademischer Verlag, Heidelberg.
718
Part IV: Nucleic Acid Analytics Miller, J.M. (2013) Whole-genome mapping: a new paradigm in strain-typing technology. J. Clin. Microbiol., 51 (4), 1066–1070. Nassonova, E.S. (2008) Pulsed field gel electrophoresis: theory, instruments and application. Cell Tissue Biol., 2 (6), 557–565. Nowacka, M., Jockowiak, P., Rybarcyk, A., Magaz, T., Strozycki, P.M., Barciszewski, J., and Figlerowicz, M. (2012) 2D-PAGE as an effective method of RNA degradome analysis. Mol. Biol. Rep., 39, 139–146. Rickwood, D. and Hares, B.D. (eds) (1990) Gel Electrophoresis of Nucleic Acids: A Practical Approach, IRL Press, Oxford. Salieb-Beugelaar, G.B., Dorfman, K.D., van den Berg, A., Eijkel, J.C.T. (2009) Electrophoretic separation of DNA in gels and nanostructures. Lab Chip, 9, 2508–2523. Sambrook, J. and Russell, D.W. (2001) Molecular Cloning: A laboratory Manual, 3rd edn, Cold Spring Harbor Press, Cold Spring Harbor.
Section 27.6 Apffel, A., Chakel, J.A., Fischer, S., Lichtenwalter, K., and Hancock, W.S. (1997) New procedure for the use of high-performance liquid chromatography-electrospray ionization mass spectrometry for the analysis of nucleotides and oligonucleotides. J. Chromatogr. A, 777, 3–21. Engels, J.W. (2013) Gene silencing by chemical modified siRNAs. New Biotechnol., 30 (3), 302–307. Gilar, M. (2001) Analysis and purification of synthetic oligonucleotides by reversed-phase high-performance liquid chromatography with photodiode array and mass spectrometry detection. Anal. Biochem., 298, 196–206. Gilar, M., Fountain, K.J., Budman, Y., Holyoke, J.L., Davoudi, H., and Gebler, J.C. (2003) Characterization of therapeutical oligonucleotides using liquid chromatography with on-line mass spectrometry detection. Oligonucleotides, 13, 229–243. Gilar, M., Fountain, K.J., Budman, Y., Neue, U.D., Yardley, K.R., Rainville, E.D., Russell, R.J. II, and Gebler, J.C. (2002) Ion-pair reversed-phase high-performance liquid chromatography analysis of oligonucleotides: retention prediction. J. Chromatogr. A, 958, 167–182. Greig, M. and Griffey, R.H. (1995) Utility of organic bases for improved electrospray mass spectrometry of oligonucleotides. Rapid Commun. Mass Spectrom., 9 (1), 97–102. Kusser, W. (2000) Chemically modified nucleic acid aptamers for in vitro selections: evolving evolution. Rev. Mol. Biotechnol., 74, 27–38. Martin, R. (1996) Elektrophorese von Nucleinsäuren, Spektrum Akademischer Verlag, Heidelberg. Matteucci, M.D. and Caruthers, M.H. (1981) Synthesis of deoxynucleotides on a polymer support. J. Am. Chem. Soc., 103, 3185–3191. Uhlmann, E. (2000) Recent advances in the medicinal chemistry of antisense oligonucleotides. Curr. Opin. Drug Discovery Dev., 3, 203–213. Uhlmann, E. and Vollmer, J. (2003) Recent advances in the development of immunostimulatory oligonucleotides. Curr. Opin. Drug Discovery Dev., 6, 204–217. Warren, W.J. and Vella, G. (1995) Principles and methods for the analysis and purification of synthetic deoxyribonucleotides by high-performance liquid chromatography. Mol. Biotechnol., 4, 179–199.
Techniques for the Hybridization and Detection of Nucleic Acids Christoph Kessler1 and Joachim W. Engels2 1
PD Christoph Kessler, Consult GmbH, Icking, Schloßbergweg 11, 82057 Icking-Dorfen, Germany Goethe University Frankfurt, Institute of Organic Chemistry and Chemical Biology, Department of Biochemistry, Chemistry and Pharmacy, Max-von-Laue Straße 7, 60438 Frankfurt am Main, Germany 2
The last two decades have seen the development of many new assay techniques for the detection and analysis of sequences of DNA or RNA. These highly specific and sensitive methods have become standard methods in molecular biology in a short span of time. Today they are used for:
diagnosis of infective diseases: viral or bacterial identification; tissue and organ tolerance diagnostics: histocompatibility genes; cancer diagnosis and risk analysis: gene mutation analysis; diagnosis of inheritable diseases, pre-implantation diagnostics: gene and chromosome analysis; paternity testing, forensic medicine, animal breeding: DNA profiling; plant breeding: analysis of gene transfer, patterns of resistance; crop and wine analysis: tests for pathogens, resistance, or marker genes linked to new or modified gene products; molecular archeology and anthropology: gene analysis of mummies, archeological finds; production of recombinant human, pharmaceutically active proteins: quality control and specificity analysis; safety surveillance of genetic laboratories: contamination tests.
Other important fields of application are the elucidation of certain genetic changes, such as point mutations, deletions, and insertions, or triplet repeats, which cause disease in the fields of oncology, genetic disease, and chronic infection. Knowledge of the precise sequence changes responsible for disease is a prerequisite for the development of genetic diagnosis tests and gene therapy approaches. The Human Genome Project has provided a standard of reference for these efforts. Begun in the early 1990s and projected to last 15 years, the Human Genome Project elucidated completely the molecular structure and sequence of the human genome. The first sets of data of the complete human genome were published at the beginning of 2001 and updated by the Human Genome Consortium in October 2004. In 2008 the sequence of eight human genomes were published and at present efforts are underway to sequence 1000 genomes in the USA and 10 000 genomes in the UK (UK10K). Disease can result from defined gene defects, chromosome aberrations, such as translocations, chromosome number abnormalities, or sub-chromosomal aberrations, such as amplifications or deletions. A well-known example is Down’s syndrome, which is caused by the presence of three copies of chromosome 21. Due to the great variety of potential gene changes, the methods used for the analysis of nucleic acids span a broad range. The analysis must be able to detect monogenetic mutations, those in single positions of the genome, as well as polygenetic mutation patterns, which involve a number of mutations. In many cases the mutations are polymorphic, meaning that more than one type of mutation is known, and new, spontaneous mutations may also arise. The type of Bioanalytics: Analytical Methods and Concepts in Biochemistry and Molecular Biology, First Edition. Edited by Friedrich Lottspeich and Joachim Engels. 2018 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2018 by Wiley-VCH Verlag GmbH & Co. KGaA.
28
720
Part IV: Nucleic Acid Analytics
Fluorescent in Situ Hybridization in Molecular Cytogenetics, Chapter 35 DNA Sequencing, Chapter 30
mutation determines the assay type: while defined mutations or simple mutation patterns are primarily detected with hybridization techniques, the analysis of variable, complex mutations, such as polymorphisms and spontaneous mutations, is increasingly performed with the new sequencing methods. The sample type also influences the choice of assay type. Different techniques are required depending on whether the nucleic acids are isolated, amplified, or chromosome preparations, or whether they are fixed to solid membranes, in solution, or in cells, tissue, or organisms. An extreme case is Drosophila embryos fixed to glass. The analysis of sequence fragments or amplified DNA or RNA sequences involves in vitro nucleic acid analysis. In situ analyses are used in molecular cytogenetics to detect chromosome aberrations or to analyze endogenous or exogenous sequences in cells, tissues, or organisms. Different hybridization and sequencing methods are used in these techniques, depending on the nature of the changes in the nucleic acids. In this chapter we will discuss the most common methods of in vitro hybridization and detection. The nucleic acid sequencing will be explained in Chapter 30.
28.1 Basic Principles of Hybridization
Isolation of Plasmid DNA from Bacteria, Section 26.3.1
Figure 28.1 Complementarity of base pairing.
The complementary bases of DNA, A and T or C and G, are bound to one another via hydrogen bonds (Figure 28.1). Hydrogen bonds are non-covalent bonds of middle to low strength, depending on the hydrogen donor acceptor atom and the distance between them. This complementarity of base pairing is the basis of hybridization. In 1961, Julius Marmur and Paul Doty discovered that the DNA double helix could be separated into two strands when denatured by heating it above its melting temperature. Allow a mix of single strands long enough to cool and they will hybridize back into double strands. Based on this observation, Sol Spiegelman developed the technique of nucleic acid hybridization. Spiegelman investigated whether newly synthesized bacterial mRNA is complementary to T2 DNA after an infection by T2 phages. In his experiment, he mixed 32 P-marked T2 mRNA and 3 H-marked T2 DNA. After denaturation of the double-stranded DNA into single strands and subsequent reassociation, he analyzed the nucleic acid mix by density gradient centrifugation. Since RNA has a higher density than DNA, it is possible to separate them in a caesium
28
Techniques for the Hybridization and Detection of Nucleic Acids
721
Figure 28.2 C0t (cot) curves represent the renaturation of different thermally denatured DNA species. The left-hand curve, mouse satellite DNA, contains many repetitive sequences and therefore renatures quickly. Source: adapted according to Britten, R.J. and Kohne D.E. (1968) Science, 161, 1530.
chloride (CsCl) density gradient. A measurement of the distribution of the radioactivity showed that a third band of RNA–DNA hybrids had appeared, in addition to the single-stranded RNA and the DNA double helices. In further experiments with T2 mRNA and DNA from unrelated organisms, no such hybrids were observed. This experiment showed that the correct sequence of complementary bases in the antiparallel nucleic acids is necessary for the hybridization reaction to take place. Information about the complexity of a particular DNA sequence can be obtained from hybridization experiments. Sequences that appear frequently in the genome renature faster than those that only appear once. Eukaryotic DNA can be divided into four classes depending on its frequency within the genome: DNA molecules immediately renature to double helices when made up of inverse repeats (palindromes), which pair by folding back on themselves, and hairpin (or stem) loops. Highly repetitive sequences reform the helices somewhat slower. Then come the less repetitive sequences, and finally the unique sequences, which, under normal circumstances, are last to rehybridize. The complexity of the DNA is expressed in the cot value: if c0 is the concentration of singlestranded DNA at time point t = 0 and c(t) the concentration of single-stranded DNA at time point t: c
t 1 c0 1 kc 0 t
(28.1)
where k is the association constant, a kinetic constant. The function c(t)/c0 is the proportion of double-stranded DNA at a particular time point, t (Figure 28.2). At a specific time, tϕ, 50% of the DNA strands are hybridized: c(t)/c0 = 0.5; the value of c0 × tϕ is the cot value. Besides conclusions about the complexity of DNA, hybridization experiments can also be used to identify particular DNA sequences in mixtures. A known sequence, a probe, is labeled, either radioactively or non-radioactively, and is hybridized with the DNA to be analyzed. Identification results from the detection of the labeled hybrids.
28.1.1 Principle and Practice of Hybridization One thing all hybridization techniques have in common is that the detection of the nucleic acid target molecules results from the sequence-specific binding of complementary labeled nucleic acid probes (Figure 28.3). In general, the nucleic acid mixture to be analyzed is blotted onto a membrane or other solid support, such as onto the surface of a microtiter plate, or left in solution. Under carefully controlled, stringent conditions (Section 28.1.2), the nucleic acids are mixed with a solution containing the labeled probe and left to incubate at a fixed temperature. Labeled probes can be made from oligonucleotides, DNA fragments, PCR (polymerase chain reaction) products (amplicons), in vitro RNA transcripts, or artificial probes like PNA (Section 28.2.3). Single-stranded RNA and DNA hybridize with one another, such that all three possible double helices are formed: DNA:DNA, DNA:RNA, and RNA:RNA. The probe hybridizes with the complementary target sequence. After completing the incubation, stringent washes are carried out
722
Part IV: Nucleic Acid Analytics
Figure 28.3 Nucleic acid detection by hybridization. Southern blots (named after Edwin Southern) are often used for this process. The polynucleotides, such as DNA fragments or PCR products, are incubated in a plastic bag for several hours with labeled probes at 60–70 °C. This temperature allows the DNA probes to hybridize with their complementary sequences on the blot. After washing, the position of the DNA segments complementary to the labeled probe is revealed by detection of the signal from the label.
to wash away unspecifically adsorbed probes. Techniques without wash steps, called homogeneous assays, have also been described. The sought target sequence is identified by measurement of the specific binding of the labeled probe; this specific binding is visualized by autoradiography or non-radioactive methods, which will be described in Sections 28.2 and 28.3.
28.1.2 Specificity of the Hybridization and Stringency The specificity of hybridization is dependent on the stability of the hybrid complex formed, as well as the stringency of the reaction conditions. The stability of the hybrid correlates directly with its melting point (Tm). The Tm value is defined as the temperature at which half of the hybrids have dissociated. It is dependent on the length and base composition of the hybridizing section of sequence, the salt concentration of the medium, and the presence or absence of formamide or other helix-destabilizing agents, as well as the type of hybridizing nucleic acid strands (DNA:DNA, DNA:RNA, and RNA:RNA). For DNA:DNA hybrids the following formula applies to a first approximation: T m 81:5 °C 16:6log c
Na 0:41
% G C
500 n
(28.2)
In this equation, c (Na+) is the concentration of Na+ ions and n the length of the hybridizing section of sequence in base pairs. Because G/C hybrids contain three hydrogen bonds instead of two, like A/T hybrids, they increase the melting temperature more, which needs to be taken into account. The expression 500/n does not apply for a sequence longer than 100 bp. The melting point of DNA:RNA hybrids is 10–15 °C higher, that of RNA:RNA hybrids around 20–25 °C higher, than for DNA:DNA hybrids given by the formula above. Base pair mismatches lower the melting point. The kinetics of hybridization are decisively influenced by the diffusion rate and length of the duplexes; the diffusion rate is highest for small probe molecules. As a result, hybridization with oligonucleotide probes is usually complete in 30 min to 2 h, while hybridization with longer probes is typically carried out overnight. A disadvantage of oligonucleotide probes, however, is that their sensitivity is not as high as with longer probes, since both the length of the hybridizing sequence and the number of labels that can be incorporated are limited. Nonetheless, the sensitivity can be increased by the use of oligonucleotide cassettes (Section 28.2.1) and through terminal tailing with multiple labels. Repetitive sequences hybridize faster since the number of potential matches for any given sequence is greater. Reaction accelerators, such as the inert polymers dextran sulfate or
28
Techniques for the Hybridization and Detection of Nucleic Acids
poly(ethylene glycol), coordinate water molecules and thereby increase the effective concentration of the nucleic acids in the remaining solution. The proportion of correctly paired nucleotides in a hybridized duplex molecule is determined by the degree of stringency with which the hybridization was conducted. Stringent conditions are those in which only perfectly base-paired nucleic acid double strands are formed and remain stable. When, under given conditions, oligo- or polynucleotide probes will only pair with the desired target nucleic acid in a mix of nucleic acids (i.e., no cross-hybridization with other nucleic acids takes place) the hybridization is defined as selective. An example of the use of this with oligonucleotide probes in specific hybridization is the detection of differences between almost identical sequences, with only a single base pair difference, such as ras wild-type/mutant at position 12 or the difference between Neisseria gonorrhoeae and Neisseria meningitidis, which only differ by a single base. Factors that influence the stringency are mainly ion concentration, the concentration of helix-destabilizing molecules, such as formamide, and the temperature. While monovalent cations, usually Na+, and mutually repulsive, negativelycharged phosphates coat the helix backbone and thereby increase the stability of the double helices, formamide inhibits the formation of hydrogen bonds and thus weakens helix stability. Temperature is of considerable importance. For example, the melting point, Tm, of a DNA segment composed of 50% (G + C), at an ion concentration of 0.4 M NaCl, is 87 °C. Hybridization takes place between 67 and 72 °C in this case. Adding 50% formamide lowers the melting point of the DNA helix to 54 °C, so hybridization can take place at 37–42 °C. This decrease in temperature is used in in situ hybridization, for example, because cellular structural integrity is lost at typical hybridization temperatures. Temperature is what defines the stringency of hybridization at defined formamide and Na+ concentrations. The Tm of a duplex molecule decreases up to 5 °C for every 1% base pair mismatch; higher temperatures serve to allow only perfectly complementary sequences to pair (high stringency). By reducing the temperature, hybrids with unpaired bases are also tolerated (low stringency). Use of high stringency conditions restricts the detection of hybridizing sequences to those that find a perfectly complementary match. After hybridization, the wash steps are conducted at only 5–15 °C below the Tm (destabilization of hybrid complexes) and in a buffer containing a low ion concentration (0.1 × sodium saline concentration (SSC), which corresponds to 15 mM Na+). A precise differentiation of mismatched base pairing is easiest using a PNA hybridization probe (Section 28.2.3). These artificial nucleic acid analogs, with an uncharged, peptide-like backbone, show more pronounced differences in stability than RNA or DNA probes between wild-type and mutant hybridizations, which results in better discrimination of base pair mismatches (also see LNA, Section 28.2.4).
28.1.3 Hybridization Methods Heterogeneous detection systems employ a detection reaction subsequent to washing off the remaining probe. Homogeneous detection systems carry out detection without separating the remaining probe, which usually involves a change in probe state to turn a signal on when hybridized. Heterogeneous Systems for Qualitative Analysis In addition to the Southern blots already mentioned, there are also dot, reverse dot, and slot blots methods for the qualitative analysis of DNA. The same methods can be used for RNA except that the Southern blot is called a Northern blot instead. Bacteria are detected with the aid of colony hybridization assays and viruses with plaque hybridization assays. Targets for in situ hybridization are chromosomes, cells, swabs, tissue, or even entire small organisms, such as Drosophila, on slides. Heterogeneous Systems for Quantitative Analysis Heterogeneous systems for the quantitative analysis of nucleic acids include sandwich assays using capture and detection probes, replacement assays of a short detection probe from the complex, or special amplification methods, in which the labeling of the detection complex takes place by incorporation of a dNTP or a primer during amplification (e.g., with DIG; Section 28.4.3). The labeled amplicon is subsequently hybridized to a biotinylated capture probe and the complex immobilized on a streptavidin-coated membrane. Alternatively, reverse dot blots capture the target using a probe
723
In Situ Hybridization, Section 35.1.4
The higher the stringency, the more specific the hydrogen bonding between the complementary strands along the entire length of the hybridizing sequence. With oligonucleotide probes, it is possible to differentiate single mutations under stringent conditions, which is essential, for example, for the specific detection of point mutations, such as the single base pair difference in sickle cell anemia, or the detection of RNA sequences from certain pathogenic species of bacteria, such as Neisseria gonorrhoeae.
724
Part IV: Nucleic Acid Analytics
covalently bound to the membrane. After washing away the excess label, the amount of bound, DIG-labeled amplicon reflects the original concentration of the target. In amplification assays, the principle is turned around and the primer is labeled with biotin, while detection is via hybridization with a labeled detection probe (Figure 28.4).
Figure 28.4 Principle of heterogeneous amplification systems. The capture marker attached to the solid support (F) is incorporated during amplification. The amplicon strand attached to the solid support is detected by hybridization with an oligonucleotide probe (D: detection marker).
28
Techniques for the Hybridization and Detection of Nucleic Acids
Homogeneous Systems for Quantitative Analysis Homogeneous systems are much more difficult to design and develop but their convenience and large dynamic range gives them significant advantages over heterogeneous systems. In combination with efficient amplification techniques like PCR, homogeneous amplification techniques have enabled quantitative and reproducible detection of femto- to attogram amounts (10 15 to 10 18 g), which suffices to detect as little as a few copies of the target sequence. At this extreme level of sensitivity, statistical limitations in collecting the samples begin to limit the overall sensitivity of detection of the target sequences. Homogeneous systems allow measurement of the amplification products during the course of the amplification reaction, without requiring the separation of reaction educts before addition of the detection reagent. The risk of contamination can be greatly reduced by carrying out the reaction in a closed system of sealed vessels, which also allows the direct detection of fluorescence signals at any time during the amplification reaction. The use of glass capillary tubes, for example, allows detection of fluorescent signals directly through the wall of the tube. The formation of the amplicons can be followed in real time. Another advantage of the homogeneous systems is their larger dynamic range, up to eight–to-ten orders of magnitude, in comparison to heterogeneous systems. In homogenous systems, the probes bind to target sequences between the primers during the amplification, generating a detectable signal. This results either from digestion of the primer by the elongation enzyme, in the case of TaqMan probes, or through the binding itself, for LightCycler rapid PCR and 3´ - and 5´ end-labeled probes (HybProbes), which hybridize next to one another. Molecular beacon-labeled probes open their branched structure on hybridization to the target sequence. This results from labeling the probes with a fluorescence-quencher set or two FRET pairs. Rhodamine or fluorescein are often used as the fluorescent marker in these systems and rhodamine derivatives, such as dabcyl or cyanine dyes, are used as the quencher. A particularly good quencher is the black hole quencher, which almost completely absorbs the emitted light. The type of label pair is tuned to the specific system. TaqMan or 5´ Nuclease Methods A modern homogeneous detection system is the TaqMan or 5´ nuclease amplification detection principle, also known as the 5´ nuclease assay. In this wellknown homogeneous detection system the probe is equipped with a marker pair, consisting of a fluorescent marker and a quencher. The distance between the pair is chosen such that the incident primary light stimulates release of fluorescent light from the marker, which is absorbed by the quencher as long as the probe is free, preventing release of a detectable signal. After hybridization of the detection probe to the amplified target molecules, the 5´ -3´ exonuclease activity of the Taq DNA polymerase frees the fluorescent nucleotide from the probe, which diffuses away from the quencher and can now emit unquenched light (Figure 28.5). Prior to degradation by the 5´ nuclease catalysis, the probe’s fluorescent signal is inactivated by the proximity of the quencher. As a result the free probe is inactive and need not be separated to specifically detect newly formed hybrid complexes. Since the quenched fluorescent nucleotide comes exclusively from hybridized complexes, the amount of fluorescence measured as a result of the decoupling of the fluorescent probe from the quencher is a direct measurement of the amount of the hybrid complex. This allows measurement of the intended target molecules without the need to separate the excess probes. Measurement of the signal increase allows the quantification of the amplified targets formed (Figure 28.6a). Figure 28.7 shows the results of a TaqMan measurement: The amount of amplification product is not measured after completion of the amplification reaction, as in end point measurements, instead the formation of amplification products is measured continuously during the course of the PCR cycles, which is why it is called a real time measurement. The unit of measurement, the cT value, gives the PCR cycle during which a signal is first seen above the threshold of detection. Plotting the cT value semi-logarithmically against the initial copy number, prior to amplification, results in a linear relationship between the cT value and the original copy number of the target sequence in the sample. By correlating the results with corresponding curves from external controls, or co-amplification of the target sequence with internal standards of known copy number, the copy number of the target sequence in the sample can be quantitatively determined. Internal standards are constructed such that they can use the same primers but contain a different probe-binding sequence. If the probe for the desired target sequence is labeled with one dye, and the probe for the control with another, filters can be used to separate the two signals measured simultaneously.
Amplification of DNA, Section 29.2.2
Instruments, Section 29.2.1
725
726
Part IV: Nucleic Acid Analytics
Figure 28.5 Principle of the 5´ nuclease reaction format (TaqMan). As long as the fluorescent detection marker (D) and quencher (Q) are linked in close proximity, no signal is emitted. The 5´ nuclease activity eliminates the linkage, the detector diffuses away and is no longer quenched, resulting in a signal proportional to the number of targets created by the Taq DNA polymerase.
28
Techniques for the Hybridization and Detection of Nucleic Acids
727
Figure 28.6 Homogeneous detection systems: (a) TaqMan/5´ nuclease system; (b) FRET system with HybProbes; (c) molecular beacon system; (d) intercalation system.
Figure 28.7 Coupled amplification and detection in the TaqMan system. The higher the cT value, the lower the copy number of the sample shown by the curve.
728
Part IV: Nucleic Acid Analytics
Figure 28.8 FRET principle. The absorption and emission spectra of the two FRET components are different. The emission spectrum of the energy donor, however, overlaps with the absorption spectrum of the energy acceptor, so that an energy exchange can take place. The detected secondary light emitted by the FRET pair is of a longer wavelength than the primary light input to induce the signal.
Förster Resonance Energy Transfer (FRET), Section 7.3.7
FRET System In the FRET system, probes are used that carry two fluorescent detection markers, with different but overlapping absorption and emission spectra. This allows energy to be transferred from one to the other, and also allows distinguishing between the emitted light signals. The two components form a fluorescence energy resonance transfer (FRET) pair. One component takes up the primary light and transfers it, in the form of light, to the second component, if it is close enough (0.5–10 nm). The second component emits the absorbed energy in the form of longer waved secondary light. The increase in wavelength allows the output signal to be measured without interference from the primary light (Figure 28.8). HybProbes are probe pairs in which each of the two probe components are marked individually with one of the two FRET components. The upstream binding probe is marked on the 3´ terminal, the second probe, which binds immediately downstream, is labeled on the 5´ end. In solution, the two component probes do not result in a FRET signal since they are not in physical proximity. But if the two probes hybridize to the newly synthesized target sequence during the annealing phase of the PCR reaction they are now in direct proximity to one another. Since the FRET components are labeled on the directly proximal ends, the FRET effect between the components results in a signal. The more probe pairs that bind to the new amplicons, the stronger the signal becomes. Measuring the signal increase allows quantification of the amplified sequences (Figure 28.6b). In the LightCycler, the FRET probes and samples are contained in thin glass capillary tubes, which allows rapid PCR temperature cycles. Probes used in the LightCycler are HybProbes, which have a fluorescent reporter and quencher pair. The tiny volume and high surface area of the capillaries allows for a very rapid temperature exchange. Each cycle is short, which reduces the total amplification time significantly. Molecular Beacon System The use of molecular beacons is another way to measure the formation of products during amplification. In this case, in contrast to the HybProbes, only a single probe is used. The probe has a reporter–quencher pair bound to each end. An example of such a molecular beacon is the combination of fluorescein as the reporter and dabcyl as the quencher. The probe sequence is so chosen that the probe displays a pronounced stem-loop structure, which brings the photosensitive components on the two ends into immediate proximity of one another, leading to quenching. As a result, no fluorescent signal is observed. The other end of the stem-loop structure is covalently held together by the loop. The loop contains the sequence that can hybridize with the target sequence. When it does so, the probe is unfolded and the two ends are separated, which prevents the quencher from blocking the fluorescent signal of the detection marker. Thus, molecular beacons are also a type of fluorescence dequenching assay. In this homogeneous system, the signal becomes stronger as more probe pairs bind to the formed amplicons, which directly reflects the amount of the amplified products. By measuring the increase in signal, this method allows quantification of the formation of amplified sequences (Figure 28.6c). Intercalation Assay In dye intercalation assays the signal is generated by the intercalation of dyes like SYBR Green into the newly synthesized double strands of the amplification products. In
28
Techniques for the Hybridization and Detection of Nucleic Acids
729
contrast to other assay types, no detection probe is hybridized to the amplicon, instead a sequenceindependent, quantitative measurement is allowed by the intercalation of fluorescent dye into the amplification products (see also Section 28.3.1). Comparison of shifted LightCycler profiles of different products generated in a single run is possible if the amplicons differ in length or base composition and thus have different melting points. Other Homogeneous Systems Other homogeneous systems are used primarily for the quantitative analysis of nucleic acids of bacteria or viral infections. Examples are the activation of inactive β-galactosidase by a complementary α peptide in enzyme complementation assays or the measurement of changes in mass of the complex through changes in fluorescence depolarization. These homogeneous assays will not be discussed in more detail.
In Situ Systems In situ assays are used to analyze nucleic acids in fixed cells, tissues, chromosomes (metaphase chromosomes or prophase nuclei), or complete organisms such as whole mount embryos. Such molecular cellular analysis supplements the nucleic acid analysis to detect sequence or genetic aberrations in isolated nucleic acids discussed up to this point. In situ detection begins with the mounting of the biological material on slides or cover slips. Cell walls must be lysed enzymatically to allow hybridization with a labeled probe in the cells, whole mounts, or to the fixed chromosomes. The resulting optical (e.g., DIG-AP plus BCIP/ NBT) or fluorescent signal (e.g., directly coupled fluorescein, rhodamine) creates the signal. An example of sequence detection in whole mounts is the detection of mRNA expression from early developmental genes in Drosophila embryos. In cells from higher organisms, DNA in the nucleus is in the form of chromatin or, during the metaphase of cell division, chromosomes. Cell division of many lymphoid cells can be stimulated with the application of phytohemagglutinin, a plant hormone, and cultivation for 2–3 days at 37 °C. The resulting metaphase chromosomes can be isolated by treating the cells with the spindle inhibitor colchicine; this freezes the cells in the middle of cell division. After pipetting the cells onto a glass slide, the cells are lysed with a hypotonic solution to create a chromosome spread. After fixing, the chromosomes can be observed under a microscope. Fluorescence in situ hybridization (FISH) is the best-known system for karyotype analysis of chromosomes with the aid of fluorescence-tagged probes. Specific regions of the chromosomes can be detected with fluorescent signals. The following sections go into the details of nucleic acid detection with a primary focus on hybridization assays and non-radioactive labeling and detection. Radioactive labeling is also discussed. Staining is presented in Section 28.4.1
28.2 Probes for Nucleic Acid Analysis Labeled oligonucleotides or nucleic acid fragments play a central role in strategies for sequence-specific nucleic acid detection by both sequencing and hybridization by serving as primers for nucleic acid synthesis. Either radioactive or non-radioactive probes are employed. The repertoire of methods employed to generate and use non-radioactively labeled probes has expanded greatly in the last few years, making them the method of choice, particularly for standard methods of nucleic acid analysis. While radioactive methods were originally the only option, problems relating to contamination of sophisticated and expensive instrumentation and disposal problems have greatly favored the development of non-isotopic assays. Although the routinely used isotopes 3 H, 14 C, 32 P, 33 P, 35 S, and 125 I have the advantage that the chemical structure, and therefore the hybridization characteristics, of the probes remains unchanged during nucleic acid analysis, the use of isotopes has the following serious disadvantages:
limited half-life and therefore detection opportunity: for example, the frequently used 32 P isotope has a half-life of only 14.3 days;
necessity of internal standards for quantitative analyses; the molecular damage to the probe itself caused by its radioactive emissions;
In Situ Hybridization, Section 35.1.4
Application of FISH and CGH, Section 35.2
730
Part IV: Nucleic Acid Analytics
need to repeat probe labeling for longer experiments; necessity for a special safety laboratory with expensive safety precautions; necessity of disposing the radioactive material; increased planning and logistics; potential health risks.
These disadvantages make the use of isotopes, particularly with the increasing availability of non-radioactive methods with at least comparable sensitivity and range of use, more problematic. Since, however, many laboratories in the research field still have the equipment for radioactive work, isotopes remain in use for blot hybridization and manual sequencing. It is, however, to be expected that with increasing standardization and automation of analytical methods the radioactive methods will be used less and less often. Probe Types DNA and RNA probes, short single-stranded DNA oligonucleotides or longer, double-stranded DNA, or single-stranded RNA probes, are used in the analysis of nucleic acids. Cloned probes contain vector fragments, unless the vector sequences are removed in additional steps. Vector fragments can lead to undesired cross-hybridization; for example, unspecific crossreactions of pBR vector sequences with genomic human DNA have been described. These undesirable effects can be avoided by using vector-free probes, which can be synthesized by PCR amplification, in vitro synthesis, or chemical synthesis. A recently invented alternative to DNA oligonucleotides is peptide nucleic acid (PNA) oligomers, which contain the same base-specificity and hybrid geometry, attached to an uncharged peptide-like synthetic backbone. Due to the lack of the mutually repelling phosphate groups, hybrids of PNA probes and nucleic acid sequences have a higher melting point, allowing the use of higher hybridization temperatures and increasing the specificity of the hybridization. A further advantage of PNA is an increased ability to discriminate mismatches.
28.2.1 DNA Probes
Polymerase Chain Reaction, Chapter 29
The main types of DNA probes used in nucleic acid analysis are genomic probes, cloned DNA probes or corresponding restriction fragments, (i.e. DNA probes), PCR-generated amplicon probes, and synthetic oligonucleotides probes. For many years, cloned cDNA or genomic fragments were the most commonly used DNA probes used to detect complementary DNA or RNA sequences in Southern or Northern blots. The probes were usually between 300 bp and 3 kb in length. The sensitivity depends on the length of the hybridizing region and the density of the labeling: as a result, genomic probes are more sensitive than cDNA probes, since the cDNA probes can only hybridize with the exon sequences of the nucleic acids to be detected, while genomic probes often encompass the extensive intron sequences. Both probe types, however, have the disadvantage that their cloning and subsequent plasmid isolation is time consuming. The production of vector-free probes requires an additional restriction cleavage and fragment separation. Even then, repetitive sequences within the section of the probe can lead to cross-reactions with eukaryotic DNA or amplified eukaryotic genes or cellular total mRNA. This can lead to unspecific bands. The possibility of creating hybridization probes by PCR amplification with Taq DNA polymerase led to an enormous and rapid increase in the availability of probes. This procedure has several advantages:
Cloning and plasmid isolation are no longer necessary, so the probes do not contain vector sequences.
Both DNA and RNA can serve as templates to create probes (RNA must first converted into cDNA with reverse transcriptase in RT-PCR).
The probes are of a defined length, allowing easy adjustment of the stringency of the hybridization.
Probe design is extremely flexible, since the probe length and position can be easily controlled by the selection of the primers.
Only the sequences where the primers bind need to be known for the amplification, thus it is possible to generate probes for new, unknown sequences between the primer binding sites.
28
Techniques for the Hybridization and Detection of Nucleic Acids
731
Similarly, probes for mutants with sequence variations in the probe area are readily available. The probes can be labeled during generation by using marked labeled nucleotides or primers, leading to a uniform density of labeling.
By using new DNA polymerases or blends of polymerases, probes that are kilobases in length are possible, instead of the 100–1000 bp length from standard Taq DNA polymerases. These advantages have made PCR the method of choice for the production of DNA probes. The probes can often be used directly in hybridizations. To avoid co-hybridization with unspecific amplification products, the amplification products are usually purified. Long PCR products can have secondary structure effects, resulting in lower sensitivity or unspecific hybridization signals; these can be avoided by additional restriction digestion. Besides long PCR probes, synthetic oligonucleotides are increasingly used for hybridization probes. Modern oligonucleotide synthesis machines can create oligonucleotides up to a length of around 150 nucleotides with a defined sequence or with targeted sequence changes in any position. Oligonucleotide probes are well suited to detect point mutations. Oligonucleotides between 17 and 40 bp in length are used for this purpose, which allows the optimization of the hybridization and wash steps. Base pair mismatches are easiest to recognize when they are located in the middle of the hybridizing region; mutations in flanking sequences are not as well discriminated. A further advantage of short oligonucleotide probes is that they hybridize faster than longer probes. A disadvantage is their lower sensitivity (Section 28.1.2). Nevertheless the strength of oligonucleotide probes does not lie in the detection of single copy genes nor low copy mRNA, but instead in the mutation analysis of PCR-amplified genes, strongly expressed mRNAs, or rRNA species amplified 103–104-fold. Oligonucleotide arrays are currently in development as hybridization tools for oligonucleotides and cDNA. These arrays contain a large number of oligonucleotides or sDNA capture probes with differing specificity. With these chips, a large number of mutations in different positions of the target amplicon can be analyzed in mutation/polymorphism analysis or from differing cells, tissues, or organs in expression pattern analysis.
28.2.2 RNA Probes Single-stranded RNA probes are produced by run off in vitro transcription of sequences cloned into vectors containing bacteriophage SP6, T3, or T7 promoters (Figure 28.9). DNA fragments or PCR products are cloned into the multiple cloning site immediately downstream of the promoter. The recombinant vector is then cleaved directly at the 3´ end of the insert, which creates a fixed termination point for the run-off transcription. The strong promoter selectivity and the fixed
Figure 28.9 Synthesis of labeled RNA probes.
DNA Microarray Technologies, Chapter 37
732
Part IV: Nucleic Acid Analytics
Template DNA and Detection of in Vitro Transcripts, Section 34.3.3
termination point lead to transcripts of identical length. The transcription cycle terminates and reinitiates 100–1000 times resulting in a high yield of the probe. Radioactive ribonucleotides or hapten-modified dUTP can be added during transcription to label the RNA probes in the process of transcription, in the same way as for PCR probes. New vectors contain two different promoters in opposite orientations on either side of the cloning region, which allow the transcription of complementary RNA strands with opposite polarities (sense/antisense RNA). To avoid unspecific hybridization signals from the vector segments, the in vitro transcripts are treated with RNase-free DNase. The main advantage of RNA probes is that DNA:RNA or RNA:RNA hybrid complexes havea higher meltingpointthan DNA:DNA hybrid complexes. This provides increased sensitivity, so that low abundance mRNA in Northern blots or in situ can also be detected. However, RNA probes are vulnerable to ubiquitous RNases, so that all solutions and equipment must be sterilized with chemical additives like diethyl pyrocarbonate and heat treatment prior to use. When used for in situ experiments, the run off transcripts are treated with a limited amount of RNase because full-length transcripts often penetrate the cell wall or membrane poorly; the shortened molecules penetrate better, which leads to more probe in the nucleus available for hybridization and, thus, increased sensitivity.
28.2.3 PNA Probes Concept of Peptide Synthesis, Section 22.1
Figure 28.10 Structural comparison of PNA and DNA.
Synthetic peptide nucleic acids (PNAs) contain a peptide-like backbone which has many advantages relative to DNA and RNA probes (Figure 28.10). PNA probes can be produced in peptide synthesis machines as well as DNA synthesis machines. Boc synthesis chemistry is used for peptide analog synthesis; Fmoc synthesis chemistry is used for DNA analog synthesis. In both Boc and Fmoc chemistry protective groups are used – only the structural elements of the backbone differ. The solubility of PNA oligomers can be increased by the introduction of charged terminal or internal side chains (e.g., Glu, Lys) or charged groups in the labeling group spacers, so that the synthesis of up to 30-mers is possible. PNA oligomers have a range of advantages relative to DNA oligonucleotides:
Greater hybrid stability; therefore higher temperature, and correspondingly more stringent, hybridization conditions can be used.
Shorter oligonucleotides have higher diffusion rates and faster reaction kinetics. Ion concentration-independent hybridization allows low salt concentrations: double-stranded PCR products hybridize without denaturing.
The hybridization at low salt concentrations opens potential secondary structures within the target molecules.
The Tm difference between matched and mismatched base pairs is more pronounced with PNA probes than with DNA or RNA probes. This allows better mismatch discrimination.
The discrimination of mismatched bases is optimal throughout the entire length of the probe, except for the terminal three or four bases on the flanks.
PNA probes are more resistant to nucleases and proteases due to the artificial structure of their backbones and base linkage, which increases the stability of the probes.
The solubility of PNA can be increased by the use of charged amino acids (e.g., Lys, Glu) in the backbone or by charges in the marker linkers. These advantages make PNA oligomers an attractive alternative to oligonucleotide probes for point mutation analysis. These advantages are particularly relevant to array systems on chips, since the selective detection of mismatches and avoiding secondary structure effects are of central importance in these systems. Also critical is the solubility of the PNA probes in these systems, which is achieved with a long linker molecule. PNA capture probes allow the selective isolation of target nucleic acids through the formation of very stable triplexes on certain target sequences or duplex formation in mixed target sequences.
28.2.4 LNA Probes Locked nucleic acids (LNAs) are a new class of bicyclic DNA analogs in which the 2´ and 4´ position is coupled into a furanose ring by an O-methyl group (LNA: stabilization by bicyclic
28
Techniques for the Hybridization and Detection of Nucleic Acids
733
bridged ribose; Figure 28.11). The binding of these analogs to complementary nucleic acids is the strongest of the known DNA analogs. LNA probes are more hydrophilic, and therefore nuclease-resistant, than PNA, which makes them a desirable alternative for some purposes. In addition:
LNA probes are more soluble than PNA probes, which makes their hybridization characteristics similar to DNA probes.
The highest Tm values of any oligonucleotide analog pair are seen in LNA:DNA mixed
oligonucleotides. This allows the use of shorter oligonucleotide probes for the determination of point mutations, which increases the discrimination between wild-type and mutation targets, similarly to PNA. The O-methyl bridge can also be replaced by a thio or amino bridge.
28.3 Methods of Labeling Non-radioactive modifications can be incorporated into probes in enzymatic, photochemical, or chemical reactions. Isotopes are usually incorporated into probes by enzymatic reactions. The labeling positions and type of label differs between methods, depending on whether isotopes or non-radioactive reporter groups are used. DNA, RNA, and oligonucleotides can be labeled by means of the enzymatic incorporation of labeled nucleotides, resulting in high sensitivity probes that are densely labeled. The photolabeling of DNA and RNA results in less strongly labeled probes, but avoids damage caused by radioactive labeling, which can affect the length of the probe; thus, it is the most suitable method for the synthesis of labeled size standards. Chemical labeling was initially used to label DNA fragments and its use is increasing for the labeling of DNA or PNA oligomers. Figure 28.12 shows an overview of the most common non-radioactive labeling reactions, which are discussed in this section. Enzymatic labeling uses either 5´ labeled primers (PCR labeling) or labeled nucleotides instead of, or in addition to, unlabeled nucleotides. Non-radioactive labeling uses nucleotide analogs modified with haptens like digoxigenin fluorescein or conjugates like biotin (Section 28.4.3). Hapten-dUTP or hapten-dCTP can be used as the enzyme substrate for RNA, DNA, and oligonucleotide labeling; the latter two can also make use of hapten-cATP. The labels can interfere with one another if they are too close to one another; consequently, maximum sensitivity requires a certain distance between them. The optimum distance needed to achieve highest labeling density and the necessary minimum separation is hapten-specific. The optimal labeling density is achieved with hapten-specific mixes of hapten-dNTP and non-modified dNTPs (e.g., 33% DIG-dUTP/67% dTTP). High sensitivity requires that the modifications are not buried in the helix structure. Therefore, a spacer between the nucleic acid strand of the probe and the modifying group is critical. In the case of the haptens previously mentioned, the spacers are at least 11 atoms long and are often composed of oxycarbonyl elements, which are coupled via ester or amide bonds. The N and O atoms make the linker sufficiently hydrophilic.
28.3.1 Labeling Positions Radioactive labeling involves exchanging stable natural isotopes for unstable radioactive isotopes, depending on the nature of the isotope, on different positions of nucleoside triphosphates. Exchanging isotopes does not change the chemical structure, so that the labeled molecules have the same chemical properties as the natural substances. This means that the reaction conditions do not have to be changed for enzymatically incorporating labeled nucleotides into probes, nor do the hybridization conditions require adaptation. The most commonly used labels are 32 P or 33 P phosphates, exchanged for either the α- or γ-phosphate residue in 2´ -deoxyribo-, 3´ -deoxyribo- (cordycepin) or 2´ -ribonucleotides. Labeling with 35 S replaces an oxygen atom of the α-phosphate with the radioactive sulfur (Figure 28.13a). The 32 P- or 35 S-labeled α position remains attached to the nucleoside when probes are homogeneously labeled by polymerases (e.g., random-primed labeling, nick translation, reverse transcription, or PCR amplification), whereas end labeling, such as with T4 polynucleotide kinase, involves transfer of the labeled γ-phosphate from ATP to the free 5´ terminal OH of the
Figure 28.11 Structural comparison of LNA and DNA; X = O, S, NH.
734
Part IV: Nucleic Acid Analytics
Figure 28.12 Schematic representation of the enzymatic, photochemical, and chemical labeling reactions, which are discussed in the following sections. Source: from Kessler, C. (1992) Non-radioactive Labeling and Detection of Biomolecules, Springer, Berlin, Heidelberg; and Kessler, C. (ed.) (2000) Nonradioactive Analysis of Biomolecules, Springer, Berlin, Heidelberg.
28
Techniques for the Hybridization and Detection of Nucleic Acids
735
Figure 28.13 Exchange positions for radioactive labels: (a) nucleoside triphosphate; (b) positions of the base rings. For the residues R1 to R3, see the inside front cover; X = N, CH (7-deaza purine).
probe. The 3 H isotope is usually used for in situ applications, due to its lower radiation scatter and longer half-life. It is incorporated into various positions of the base ring. Labeling with 125 I is at the C5 position of cytosine (Figure 28.13b). The low radiation intensity and very long halflife, with its attendant disposal problems, of 14 C isotopes have made their use rare. The chemical structure of the labeled nucleotides and the labeled probe is changed by the non-radioactive modification of probes with reporter groups. This means, for example, that the reaction conditions for enzymatic incorporation of the label need to be adapted to the altered substrate characteristics. Unspecific binding to the modifications must be blocked with suitable blocking reagents (e.g., milk proteins); in the case of blot hybridization, the entire membrane surface must be blocked; for in situ hybridization, the cell or tissue surface requires blocking. The correspondingly modified protocols are, however, well established and do not limit the use of non-radioactive probes.
The most commonly labeled position for sequencing primers, PCR primers, or hybridization oligomers (DNA, PNA) is the 5´ -terminal. This preserves the characteristics of the probe and the formation of hydrogen bonds during hybridization is not influenced. Modifications are introduced through bifunctional linear diamino bonds of variable length. Nucleotides are usually labeled by base modifications; however, labeling of the 2´ position of ribose has also been described. The base modifications are chosen such that they do not interfere with the formation of hydrogen bonds during hybridization. The most common modification position is the C5 position of uracil or cytosine; in the case of deoxyuridine, the modification group imitates the methyl residue of the thymidine base of deoxythymidine (compare Figure 28.13b). In both cases the introduction of the modifying group is not direct, instead it is introduced attached to a spacer. Other suitable positions for the introduction of non-radioactive modifying groups are the C6 of cytosine and the C7 of deaza-guanine and deaza-adenine (Figure 28.13). The amino groups of cytosine, adenine, or guanine, which were frequent targets in the past, are less well suited, since these positions are involved in hydrogen bonding.
28.3.2 Enzymatic Labeling Many of the enzymatic labeling reactions are analogous when using radioactive or non-radioactive labels. The difference is, however, that in the case of radioactive labeling enzyme substrates are structurally identical isotope-labeled nucleotides, while non-radioactive labeling involves additional modification of the nucleotides, which may require adaptation of the reaction conditions.
736
Part IV: Nucleic Acid Analytics
Enzymatic Labeling of DNA Homogeneous DNA labeling involves random priming with the large fragment of the Escherichia coli DNA polymerase I (Klenow enzyme), nick translation with Escherichia coli DNA polymerase I (Kornberg enzyme) or via PCR amplification with Taq DNA polymerase. The density of the labeling is around one label per 25–36 base pairs. Oligonucleotides can be labeled with the help of the terminal transferase reaction; depending on the substrate, one to five labels per oligonucleotide are attached (for an overview see Figure 28.12). Random Priming In the random priming protocol, double-stranded DNA is denatured and rapidly cooled and high concentrations of primer are added to prevent the rehybridization of the target strands. The primers are a mix of all 4096 conceivable hexanucleotides (thus “random” primer), so that statistically every target sequence is covered and the hybridization can occur at any point in the sequence. The Klenow enzyme, the large subtilisin fragment of the Escherichia coli DNA polymerase holoenzyme, extends the primer in a template-dependent reaction. During the elongation reaction, unlabeled dNTPs and hapten-modified dUTPs are built in. Since the template strand has been replicated, strand displacement leads to a new round of synthesis, which leads to a yield of over 100% of the input template DNA. Since statistically several primers bind per target, each primer elongation only replicates part of the sequence; the result is a mix of probes of variable probe length. The partial sequences are all target-specific, however, and carry a homogeneous labeling. Random primed DIG-labeled probes reach a high detection sensitivity in the sub-picogram range. Nick Translation Nick translation uses the Escherichia coli DNA polymerase holoenzyme (Kornberg enzyme) and very small amounts of pancreatic DNase I. This technique requires the 5´ -3´ exonuclease activity of the DNA polymerase, as well as its polymerase activity. The DNase catalyzes the creation of single strand nicks; a very precisely controlled, small amount of DNase is required to ensure that the nicks remain limited. Both ends flanking the single-strand break function as templates, either for the 3´ -5´ polymerase activity or the 5´ -3´ exonuclease activity. The 5´ -3´ exonuclease activity successively removes 5´ -phosphorylated nucleotides, while the 3´ 5´ polymerase activity simultaneously fills in the holes with new, labeled nucleotides. By this concerted action, the nick moves in the 5´ -3´ direction, referred to as nick translation. It involves a DNA replacement synthesis, the yield though remains below 100%, which is less than random priming. Besides the lower labeling, getting the relationship between the amounts of DNase I and Escherichia coli DNA polymerase exactly right is a critical factor. As a result, nick translation is going out of fashion for labeling probes.
Polymerase Chain Reaction, Chapter 29
PCR Amplification
As previously mentioned, PCR amplification with Taq DNA polymerase is a useful method for the creation of vector-free probes. The amplification reaction consists of 30–40 temperature cycles with three partial reactions – denaturation, primer binding, and primer elongation – each at different temperatures. Besides this three-step protocol, two-step protocols have also been described, in which primer binding and primer elongation take place in a single step. Labeled primers or labeled dNTPs are used in the creation of labeled products as hybridization probes. PCR amplification is explained in detail in the next chapter.
Reverse Transcription Labeled DNA probes can also be synthesized from RNA targets with viral reverse transcriptases (e.g., AMV RTase, MMLV RTase). After first and second strand cDNA synthesis, unlabeled dNTPs and hapten-labeled dNTPs are added to the reaction mix to label the probe. Terminal Transferase Reaction DNA oligonucleotides can be labeled enzymatically by the template-independent attachment of labeled dNTPs, sometimes referred to as tailing, with the enzyme terminal transferase. If a mix of labeled and unlabeled dNTPs are employed, tails with multiple labels are created in a template-independent reaction. Using labeled cordycepin triphosphate (3´ -dATP) or 2´ ,3´ -ddNTPs allows the attachment of only a single labeled nucleotide, since the reduced 3´ position can no longer be extended. (see Figure 28.12: enzymatic oligonucleotide 3'-end labeling, oligonucleotide 3'-tailing).
Template DNA and Detection of in Vitro Transcripts, Section 34.3.3
Enzymatic Labeling of RNA
RNA can also be labeled using the previously described in vitro run off transcription with bacteriophage-encoded SP6, T3, or T7 RNA polymerases (Section 28.2.2). Owing to the re-initiation of transcription, high synthetic yields are achieved (up to
28
Techniques for the Hybridization and Detection of Nucleic Acids
737
20 μg transcript from 1 μg recombinant vector). The density of the labeling for enzymatic DNA labeling is around one label for every 25–36 nucleotides.
28.3.3 Photochemical Labeling Reactions There is only one photochemical DNA and RNA labeling reaction: aryl azide-activated haptens are reacted with nucleic acids under long wavelength UV light. Light excitation leads to the release of elementary nitrogen (N2); the short-lived reactive nitrogen radical reacts with different positions of the DNA or RNA and thus covalently bonds the hapten to the nucleic acid. The labeling density is relatively low, however, at around one label per 200–400 base pairs. A range of photoactive substances intercalates into nucleic acids and can be subsequently covalently bound to nucleic acid bases in a photoreaction. These substances include coumarin compounds (psoralen, angelicin), acridine stains (acridine orange), phenanthroline (ethidium bromide), phenazine, phenothiazine, and chinone (Figure 28.14). For photolabeling the most important of these is the bifunctional psoralen, which forms either mono- or bis-adducts with pyrimidine bases after intercalation and photoactivation.
Figure 28.14 Intercalating, photoactive substances for the detection of nucleic acids. The compound first intercalates and a subsequent photoreaction bonds it to the nucleic acid strand covalently.
28.3.4 Chemical Labeling Chemical labeling reactions are used mainly for the labeling of DNA, LNA-oligonucleotides, and PNA-oligomers. Labeling takes place during solid-state oligonucleotide synthesis by the direct incorporation of modified phosphoramidites. Alternatively, labeling can take place after synthesis by coupling the modifying group and the 5´ terminal phosphate of the oligonucleotide with the help of bifunctional diamino compound linkers. Protected phosphoramidites containing the desired hapten are commercially available. Such labeled oligonucleotides can often be used directly after removal of the protective group, without need for further HPLC purification. A second possibility is the incorporation of uracil or cytosine phosphoramidites carrying protected allyl- or propargylamine residues onto the C5 position or deaza-purines onto the C7 position during nucleic acid synthesis (Figure 28.15). After removal of the protective group, they react with N-hydroxysuccinimide-activated haptens or other coupling-capable reporter
Principles of the Synthesis of Oligonucleotides, Section 27.6.1
Figure 28.15 Examples of modified phosphoramidites used in oligonucleotide synthesis. Shown are a pyrimidine and a purine nucleoside as DNA and RNA building unit. R1 and R2 are protecting groups, R3 = H, OH, and X = hapten, reporter molecule.
738
Part IV: Nucleic Acid Analytics
molecules. In this case, subsequent purification by gel electrophoresis or HPLC is often necessary, owing to the often non-quantitative synthetic yield. Oligonucleotides can be coupled directly to labeling enzymes like alkaline phosphatase or horseradish peroxidase with bifunctional linking reagents. Direct detection of the enzyme is possible after hybridization with such enzyme-tagged probes; however, the hybridization temperature and duration are limited by loss of enzyme stability. After synthesis and removal of the protective groups, the chemical labeling of PNA involves coupling of the 5´ terminal amino group with the hapten with the aid of the previously described bifunctional diamino compounds.
28.4 Detection Systems Radioactively and non-radioactively labeled probes, primers, or nucleotides are the central components of nucleic acid detection systems for sequencing and hybridization. The hybridization formats allow greater flexibility in the use of labels and detection components than sequencing protocols. This is because direct and indirect systems, in which a particular modifying group can be detected and measured in multiple ways, can be used. This allows great variety in the application of non-radioactive detection systems, not only in blot formats but also in the full range of qualitative and quantitative systems. If the detection of particular sequences is unnecessary and the pattern of restriction fragment patterns is of interest, staining with intercalating compounds like ethidium bromide is the method of choice. The formation of nucleic acids can also be detected immediately in scintillation counters after incorporation of radioisotopes and separation of unincorporated labeled nucleotides; while these methods were once in widespread use, they are now only rarely used.
28.4.1 Staining Methods Determination of Nucleic Acid Concentration, Section 26.1.4
The absorption of UV light can be used to determine the concentration of nucleic acids; doublestranded DNA can also be detected by the intercalation of fluorescent dyes or by silver staining. Although silver staining is more sensitive, the visualization of DNA fragments in gels is routinely done by intercalation with ethidium bromide and illumination with UV light to generate a fluorescent signal. The sensitivity of the ethidium bromide dye depends on the length of the fragments, like all staining methods, and lies in the nanogram range. The sensitivity of silver staining is in the sub-nanogram range.
28.4.2 Radioactive Systems The β-emitters 32 P, 33 P, and 35 S are usually used for blotting procedures with radioactive probes. Although 35 S isotopes have approximately tenfold less energy, they have the advantage of a longer half-life. 33 P isotopes are more expensive, but offer more energy than 35 S and better resolution than 32 P, making them a good compromise in some situations. 3 H isotopes are also β-emitters but have tenfold less energy than 35 S. Owing to its high stability and its low scattering, this isotope is used in situ and in tissues, though these techniques are increasingly switching to non-radioactive methods. Scattering refers to the tendency of high energy emitters to expose film, even when striking the film at an oblique angle, which causes the signals to become so-called diffuse reflections that appear blurry. Trichloroacetic acid (TCA) precipitations have been used to employ 14 C, but today this isotope is only of historical significance. Table 28.1 lists the key facts for probe labeling and sequencing. The highest energy isotope in common use is 32 P, which makes it the most sensitive. However, its short half-life and high scattering limits the use of this isotope in sequencing and the analysis of complex fragment patterns, since the resolution of closely packed bands is limited. The legibility of sequence gels is also limited towards the origin. When high resolution is more important than high sensitivity, 33 P or 35 S are better choices. Examples are enzymatic sequencing and the analysis of DNA- or RNA-binding proteins in gel shift assays, due to altered
28
Techniques for the Hybridization and Detection of Nucleic Acids
Table 28.1 Characteristics of radioisotopes used to label probes. Isotope particle
Emax (MeV)
Half-life
Application
Characteristics
β
0.0118
12.3 years
In situ
Low sensitivity, high resolution
S
β
0.167
87.4 days
Filter hybridization, sequencing, in situ
Medium sensitivity, good resolution
125
I
γ
0.035
60.0 days
In situ
Low sensitivity, high resolution
P
β
1.71
14.2 days
Filter hybridization, sequencing
Highest energy, highest sensitivity, medium resolution due to reflection
3
H
35
32
mobility of the protein-bound fragments in gels. Besides 3 H, the low radiation intensity of 35 S and 125 I make them suitable for in situ applications, where limiting β-scatter is important to determine exact cellular localizations. Radioisotopes are commercially available in the form of nucleotides already labeled at the desired position. The aqueous solutions contain stabilizers to inhibit the degradation of the biologically active substances by the ionizing radiation. The isotopes are stored at –20 or –70 °C to balance thermal heat decay. Owing to the steady degradation and formation of radicals, radioactive nucleotides should be used as soon as possible, even when the half-life allows a longer use. The most important factors with respect to the storage of radioactively labeled probes are:
the half-life of the isotope; the specific radioactivity of the probe; probes with a high labeling density are very sensitive but are subject to rapid degradation;
the position of the radioactive atom in the molecule; internal labeling leads to strand breaks more easily than terminal labeling. The detection of radioactive nucleic acids takes place in blot form by autoradiography with Xray film, which can be stored permanently to document the results. There are different means of detection, depending on the isotope and the required sensitivity:
Direct autoradiography: The radiating surface (membrane, gel, cell layer or tissue slice) is
brought into direct contact with the X-ray film. This method applies to all β-emitters. Film without a protective layer is required for 3 H, so that the energetically weak electrons can penetrate to the photoactive layer. Fluorography: the radiating surface is treated with fluorescing chemicals, which convert the radiation energy into fluorescence; the most common fluorophores are 2,5-diphenyloxazole (PPO) and sodium salicylate. Indirect autoradiography with intensifier screens: High energy β-emissions are absorbed by phosphate residues of the intensifier screen and converted into visible light by illumination with a laser. Fluid emulsions for cytological or cytogenetic in situ applications: The low to mid energy 3 H or 35 S decay products require direct contact with the detection medium; the solid emulsion is melted at 45 °C and the slide is dipped into it. After drying, exposure occurs for days to months at 4 °C in the absence of light. Pre-exposed X-ray film for direct autoradiography and fluorography: A short pre-exposure activates the silver grains, which then require fewer photons to generate a signal. The preexposure can only be used for fluorography or light intensifiers (light processes).
The right method of detection depends on the characteristics of the isotope in use, such as its type, specific radioactivity, and total radioactivity, as well as the requirements of the experiment, such as the required sharpness, and maximum feasible exposure time.
28.4.3 Non-radioactive Systems The non-radioactive labeling and detection systems are divided into direct and indirect indicator systems (Figure 28.16).
739
740
Part IV: Nucleic Acid Analytics
Figure 28.16 Direct and indirect detection systems.
The two types of reactions differ in the number of components and the reaction steps, as well as the flexibility of use. While direct systems are primarily used in standardized processes (e.g., for the labeling of universal sequencing primers), the more flexible indirect systems are used to selectively detect nucleic acid sequences. In direct systems, the probes are directly and covalently coupled to the signal-generating reporter group; detection is done in two reaction steps:
hybridization between the nucleic acid target and the directly labeled probe; signal generation by the directly coupled reporter group. The advantage of direct systems is that only hybridization between the target and the probe is needed. The disadvantage is, however, that for every hybridization sequence each probe must be covalently coupled to the detection label. Therefore, this detection method is used primarily for the labeling of standard sequences with easily coupled fluorescent dyes. , e.g. with sequencing primers. In indirect systems, the probes are not directly labeled, instead they are detected by an additional, non-covalent interaction between a low molecular weight tag and a universal detector. Thus, indirect systems first require the enzymatic, photochemical or chemical incorporation of the modifying group into the probe (Section 28.3). These incorporation reactions are easy; the corresponding protocols are well-established. A universal detection unit binds to the tag, independent of the type of probe and its specificity. The detector, which contains a binding unit as well as the reporter, couples specifically and with high affinity to the tag of the probe. The indirect detection takes place in three reaction steps:
hybridization between the nucleic acid target and the modified probe; specific and high affinity, non-covalent interaction between the modifying groups of the probe and binding components of the universal detection unit;
signal generation by the indirectly bound reporter component.
Although an additional reaction step is required, the high flexibility in the generation of the probes and the coupling with diverse types of detectors are significant advantages for indirect systems. As a result, simple and fast reactions can be used to build different tags into different types of probe; in addition, the tags can be detected with a large set of alternative, universal detectors, dependent on the application. The additional non-covalent interaction makes many combinations possible that allow a broad use of the non-radioactive reporter systems in basic research and the applications previously described at the beginning of the chapter. Direct Detection Systems The most commonly used non-radioactive reporter groups in direct detection systems are fluorescent or luminescent reporter groups, as well as reporter enzymes. Goldlabeling is used for in situ applications; additional use of colored latex beads or silver staining amplifies the signal up to 104-fold. Box 28.1 shows an overview of important reporter groups.
28
Techniques for the Hybridization and Detection of Nucleic Acids
741
Box 28.1 Important direct, non-radioactive reporter groups. Fluorescent label
Metal label
Direct fluorescence
Gold- labeled antibodies
Fluorescein
Enzyme labels
Rhodamine
Direct enzyme coupling
Hydroxy-coumarin
Alkaline phosphatase
Benzofuran
Horseradish peroxidase
Texas Red
Microperoxidase β-Galactosidase
Bimane Ethidium/TB
3+
Urease
Time-resolved fluorescence 3+
3+
Lanthanide (Eu /Tb ) micelles/chelates Fluorescence energy transfer (FRET)
Glucose oxidase Glucose-6-phosphate dehydrogenase Hexokinase
Fluorescein (FAM)
Bacterial luciferase
Rhodamine (TAMRA)
Firefly luciferase
Cy 3 Cy 3,5 CY 5 CY 5,5 CY 7
Enzyme channeling Glucose oxidase: horseradish peroxidase Enzyme complementation Inactive β-galactosidase: α-peptide Polymer labels
Luminescence labels
Latex particles
Chemiluminescence
Polyethyleneimine
(Iso-)luminol derivatives Acridinium esters Electroluminescence Ru2+ (2,2´ -bipyridine)3 complexes Luminescent energy transfer Marker enzymes are such as alkaline phosphatase (Jablonski, E. et al. (1986) Nucl. Acids Res., 14, 6115–6128), horseradish peroxidase (Renz, M. and Kurz, C. (1984) Nucl. Acids Res., 12, 3435–3444; as well as fluorescent tags such as fluorescein or rhodamine (Kessler, C. (1994) J. Biotechnol. 35, 165–189), and Kessler, C. (ed.) (2000) Non-radioactive Analysis of Biomolecules, Springer, Berlin, Heidelberg.
Bacterial alkaline phosphatase (AP) is mainly used for the direct labeling of oligonucleotides and horseradish peroxidase (POD) for the direct labeling of fragments. The use of marker enzymes requires an additional substrate reaction (see below). Coupling alkaline phosphatase to oligonucleotides is a single-step reaction using a bifunctional linker. Direct AP-coupled oligonucleotides are useful in standard assays with a fixed sequence. For example, AP-coupled primers are employed in sequencing, direct blotting electrophoresis (DBE), and as universal amplifiers in signal amplification systems, like probe brushes (Section 28.5.3). The use of POD labeled fragment probes is limited, since POD is increasingly unstable above 42 °C, which limits the maximum temperature to a range that is not suitable for all purposes. Well-known fluorescent labels are fluorescein, rhodamine, and coumarin derivatives. Higher sensitivity is achieved with phycoerythrins or fluorescein lattices; in these cases the coupling
742
Part IV: Nucleic Acid Analytics
FISH Analysis of Genomic DNA, Section 35.2.1
reactions are more complex. Besides use as sequencing primers, fluorescent markers are mainly used for fluorescent in situ hybridization (FISH). The detected fluorescence can contain unspecific signals due to unspecific background light or fluorescent reaction components. For example, hemoglobin fluoresces, which causes interference in any experiments carried out using serum. This can be avoided with time-resolved fluorescence measurements with europium or terbium complexes, chelates, or micelles coupled directly to the probe via a linker, since the emission of the secondary light is delayed in these cases. Direct luminescence markers can be grouped according to their activation type, which can be chemical, electrochemical, or biochemical. Well-known chemically activatable markers for the direct measurement of nucleic acids are acridinium esters, which are activated by H2O2/alkali, as well as the protein aequorin from the jellyfish Aequorea, which is activated by Ca2+ ions. Acridinium esters glow, as they release photons over a long time frame. Aequorin flashes a short, intense pulse of light that is very specific due to the extremely low background, leading to high sensitivity. Electrochemical luminescence markers are stimulated to emit photons by electrochemical reactions. Corresponding markers are [Ru2+(bipyridyl)3] or phenanthroline complexes. The ruthenium ions are oxidized on a gold electrode (Ru2+ → Ru3+) while the subsequent reduction of the Ru3+ by tripropylamine (TPA) creates a chemiluminescent signal. The resulting Ru2+ ion is then available to begin the next reaction cycle. Gold particles can be used in blot or in situ formats to directly visualize targets. An additional silver staining can increase the sensitivity of the detection. In this case, the original gold particles have silver layered on them, increasing their size and making them easier to see. Indirect Detection Systems Several different couplings are available to indirect detection systems due to the additional specific interaction between the modified nucleic acid hybrid and the universal detector carrying the reporter group. Table 28.2 shows the coupling groups commonly used for indirect nucleic acid detection. Most systems use antibodies or the biotinbinding proteins avidin or streptavidin to recognize special modifying groups attached to nucleic acids, but other less widespread systems exist that use specific sequences in the hybrid or specific hybrid conformations as binding components (operators or promoters) to bind to proteins like repressors or RNA polymerases. The DIANA concept rests on the binding of lac repressor β-galactosidase conjugates to hybrid-coupled lac operator tag sequences. In addition, non-sequence specific binding proteins like the single-stranded binding (SSB) protein or histones have been employed as tags. Conformation-specific antibodies are examples of conformation-recognizing binders. Tagging probes with metal ions or to poly(A)-coupled systems were also used in the early days of the development of non-radioactive reporter systems. Of the various systems, only the biotin (BIO) system and antibody systems with digoxigenin (DIG), fluorescein (FLUOS) and 2,4-dinitrophenol (DNP) have sensitivity in the sub-picogram range. Thus, these systems have become standards for the non-radioactive detection of nucleic acids, while the other systems are described more out of historical interest. Enzymatic labeling takes place with hapten-labeled nucleotides; as examples of these labels, the structures of DIG-, FLUGS- and BIO-labeled nucleotides are shown in Figure 28.17a. The labeling of Table 28.2 Interaction pairs for indirect, non-radioactive detection systems. Source: from Kessler C. (ed.) (1992) Non-radioactive Labeling and Detection of Biomolecules, Springer, Berlin, Heidelberg; for further inetraction pairs see Kessler, C. (ed.) (2000) Non-radioactive Analysis of Biomolecules, Springer, Berlin, Heidelberg. DNA modification ↔ binding partner
Examples
Vitamin ↔ binding protein
Biotin ↔ streptavidin
Hapten ↔ antibodies
Digoxigenin ↔ anti-digoxigenin antibody
Protein A ↔ constant region of IgG
Protein A ↔ IgG
DNA/RNA-hybrid ↔ DNA/RNA specific antibody
DNA/RNA ↔ anti-DNA/RNA antibody
RNA/RNA-hybrid ↔ RNA/RNA specific antibody
RNA/RNA-hybrid ↔ anti-RNA/RNA antibody
Binding protein-DNA-sequences ↔ binding protein
T7-promotor ↔ Escherichia coli RNA polymerase
Heavy metal ↔ sulfhydryl-reagent
Hg2+ ↔ HS-TNP ↔ (TNP/DNA)-specific antibody
Polyadenylation-polynucleotide phosphorylase/ pyruvate kinase
ATP-coupled red firefly luciferase reaction
28
Techniques for the Hybridization and Detection of Nucleic Acids
Figure 28.17 (a) Structure of hapten and biotin-labeled dNTPs. (b) DNP-modified phosphoramidites.
743
744
Part IV: Nucleic Acid Analytics
Figure 28.18 Structure of digoxigenin. Digoxigenin is a steroid with the formula C23H35O5. The A/B rings are cis isomers, the B/C rings are trans isomers, and the C/D rings are cis isomers.
FISH Analysis of Genomic DNA, Section 35.2.1
oligonucleotides takes place mainly through modified phosphoramidite; an example, the DNPmodified phosphoramidite, is shown in Figure 28.17b. Mixtures of differently labeled probes are used for the parallel detection of different fragments on blots (DIG, BIO, FLUGS: rainbow detection) or in situ for the detection of different chromosomal segments or different chromosomes (DIG, BIO, DNP: multiplex FISH, chromosome painting). The Digoxigenin System Digoxigenin is the chemically synthesized aglycone of the cardenolide lanatoside C (Figure 28.18). The digoxigenin:anti-digoxigenin (DIG) system is based on the specific interaction between digoxigenin and a high affinity, DIG-specific antibody covalently bound to a reporter group. The DIG system can specifically detect sub-picogram amounts
Table 28.3 Important optical, luminescent, and fluorescent detection types. Source: from Kessler, C. (ed.) (1992) Non-radioactive Labeling and Detection of Biomolecules, Springer, Berlin, Heidelberg; for further detection types see Kessler, C. (ed.) (2000) Non-radioactive Analysis of Biomolecules, Springer, Berlin, Heidelberg. Format
Specific detection Optical
Blots
Luminescence
Fluorescence ®
®
AP/BCIP, NBT
AP/AMPPD, Lumiphos , CSPD , CDPstar
AP/naphthol-AS-azo-dye, diazonium salt
β-Gal/AMPGD
Fluorescein, rhodamine, hydroxycoumarin, AP/AMPPD, fluorescein/rhodamine/hydroxycoumarin β-Ga/AMPGD, fluorescein/rhodamine/hydroxycoumarin
AP/p-NPP
AP/AMPPD, LumiphosTM, CSPD® , CDPstar®
AP/4-MUF-P
β-Gal/CPRG POD/ABTSTM
β-Gal/AMPGD, POD/luminol, isoluminol
β-Gal/4-MUF-β-Gal, POD/homovanillic acid-o-dianisidine/H2O2
GOD:POD-pair/ABTSTM
Xanthine-oxidase/cyclic dihydrazide
Aromatic peroxalate compounds/H2O2
Hexokinase: G-6-PHD-pair ABTSTM
G-6-PDH/phenazinium salt
Lanthanoid-(Eu3+Tb3+)-complex,-micelles-, chelates
TM
HRP/TMB, immunogold Solution
Hydrolysis of acridinium ester, rhodamine:luminol pair, AP/D-luciferin-O-phosphate: firefly-luciferase/ATP/O2 Renilla-luciferase green fluorescent protein, Ru2+(bpy)3 In situ
ABTS: AMPGD: AMPPD: BCIP: CPRG: CSPD:
AP/BCIP, NBT
Fluorescein
POD/TMB
Rhodamine
Immunogold
Hydroxycoumarin
2,2-azino-di(3-ethyl)-benzothiazole sulfate 3-(4-methoxyspiro(1,2-dioxetan-3,2´ -tricyclo[3.3.1.13,7]-decan)-4-yl)phenylβ-galactoside 3-(4-methoxyspiro(1,2-dioxetan-3,2´ -tricyclo[3.3.1.13,7]-decan)-4-yl)phenylphosphate, Na2 5-bromo-4-chloro-indolylphosphate chlorophenol-red β-galactoside 3-(4-methoxyspiro(1,2 dioxetan-3,2´ ,5´ -chloro)-tricyclo[3.3.1.13,7]-decan)4yl)-phenylphosphate Na2
Naphthol-AS: Diazoniums salt: GOD: MUF: NBT: NPP: POD: TMB:
2-hydroxy-3-naphtholicacid-anilide Fast blue B, Fast Red Tr, Fast Brown RR glucose oxidase methylumbelliferyl phosphate nitro-blue tetrazolium salt p-nitrophenyl phosphate horseradish peroxidase 3,3´ ,5,5´ -tetramethyl-benzidine
28
Techniques for the Hybridization and Detection of Nucleic Acids
745
Figure 28.19 Digoxigenin detection system with its detection marker alternatives: fluorescent, luminescent, and optical.
of DNA or RNA. Alkaline phosphatase is often used as the reporter group, which is covalently conjugated to the antibody. Alkaline phosphatase catalyzes the conversion of optical or chemiluminescent substrates (BCIP/NBT, AMPPD; Table 28.3 below). The DIG system is shown in Figure 28.19. Enzyme-linked immunosorbent assays (ELISAs) employ detection with universal antibody-marker enzyme conjugates. The high specificity of detection and the low background level of the DIG system are due to the fact that digoxigenin is naturally only present in Digitalis plants (foxglove) in the form of lanatoside compounds; the antibodies employed thus do not recognize any cellular component from other biological materials. This is particularly important for in situ hybridizations. DIG-specific antibodies have very few unspecific cross-reactions with cellular components. Only human serum contains components that are known to interact with digoxigenin-specific antibodies; these serum components can be specifically removed, however, by pretreatment of the polyclonal anti-DIG antibodies with serum.
Due to the high specificity of the digoxigenin antibodies used, even the structurally similar steroids are only recognized to a very limited extent or not at all; examples are the bufadienolide K-strophanthin (cross-reactivity < 0.1%) and the steroids estrogen, androgen, or progestogens (cross-reactivity < 0.002%). The important difference between digoxigenin and the sexual hormones is that the C and D rings are in the cis isomer conformation, rather than trans. Digoxigenin is coupled to the nucleotide via the –OH group to the C3 position of the cardenolide frame with a linear spacer. This does not interfere with antibody binding so detection with antibody conjugates is still possible after incorporation of digoxigenin into the probe. Digoxigenin is isolated from leaves of the plant Digitalis lanata or Digitalis purpurea by the cleavage of three digitoxose and a glucose unit of the natural substance deacetyl-lanatoside C. To further reduce unspecific effects, only the Fab fragment of the antibody is used, rather than the complete antibody. The Fab fragment is isolated by papain cleavage of the constant Fc fragment; it contains just the short antibody arms with the highly variable binding sites. The complete antibody is only used in coupled systems with secondary reporter antibodies, which recognize the Fc portion. For example, a secondary antibody from mouse, which recognizes the Fc portion of the DIG-specific sheep antibody, serves to amplify the signal. The reporter groups are not coupled to the primary DIG-specific antibody, but instead to the Fc-specific secondary antibody. The Biotin System With the biotin:avidin (or streptavidin) (BIO) system, the ubiquitous biotin, also known as vitamin H, is used as the tag (for the structure of biotin see Figure 28.17a). Coupling
Immune Binding, Section 5.3.3
746
Part IV: Nucleic Acid Analytics
employs linear spacers attached to the terminal carboxy group of the biotin side chain. Avidin, from egg white, or streptavidin, from Streptomyces avidinii bacteria, have four high affinity binding sites for biotin; their binding constants are among the highest known natural affinities, at 1015 mol–1. Avidin or streptavidin is covalently coupled with a reporter group to allow detection. After binding to the biotin-modified hybrid complex, secondary biotinylated detection components can be coupled to the remaining free biotin binding sites to amplify the signal. Although the BIO system is as sensitive as DIG, the disadvantage of the system is that biotin is ubiquitously present in biological material. As a result, unspecific binding of free endogenous biotin is possible with almost all cellular material. This results in a high background, particularly for in situ formats. Another problem with this system is the tendency of the two binding proteins to stick to membranes, even after the proteins have been saturated with blocking reagents. This stickiness is caused by two factors: the high basicity of avidin and the presence of multiple tryptophan residues in the binding pockets of avidin and streptavidin. Both factors lead to unspecific polar or hydrophobic interactions with proteins of the membrane blocking reagents. In the case of avidin, unspecific binding is also caused by glycan chains on the surface, which can bind to sugar-binding proteins on the surface of cells (for in situ procedures) or the membrane blocking reagents (in blotting procedures). Since streptavidin is isolated from bacteria, this factor plays no role, since there is no protein glycosylation in prokaryotes. These unspecific background reactions can be reduced in several ways:
Acetylation or succinylation of the lysine chains or complex formation between avidin and the acidic protein lysozyme reduces the basicity of avidin and thus its unspecific polar interactions.
Deglycosylation of avidin reduces unspecific adsorption by binding to sugars. Pre-incubation of the blocked membrane in a high ion concentration buffer reduces unspecific protein interactions.
Despite the high binding affinity and the other measures to reduce the unspecific background, a poor signal-to-background ratio is observed at low target concentrations, which limits the sensitivity of the entire system. While this may matter when using BIO in a detection system, for other applications this is less important. The biotin–streptavidin binding system is often employed in the isolation of nucleic acids by hybridization with biotin capture probes. Samples are coupled to a streptavidincoated surface (e.g., beads or microtiter plates) and the unbound analytes are washed away. The sequence-specifically bound nucleic acids can subsequently be amplified and detected, for example, after incorporation of DIG, specifically and without interference. The Fluorescein System The fluorescein:anti-fluorescein (FLUOS) System (Figure 28.17a) is another antibody-based system. The hapten, fluorescein, is coupled by amide binding between the spacer and the free carboxy group of the hapten. The sensitivity of the fluorescein-specific antibody is also high. Since fluorescein is light-sensitive, however, exposure to light during storage causes loss of sensitivity. This can be avoided by storage in a cooled place protected from light. The Dinitrophenol System The 2,4-dinitrophenol:anti-2,4-dinitrophenol (DNP) system is also based on antibody binding (Figure 28.17b). The hapten is bound to the aromatic C1 position, for example, by chemical conversion of 1-fluoro-2,4-dinitrobenzene with a protected amino terminal spacer into DNP-labeled phosphoramidites. These are then built into oligonucleotides during chemical synthesis. Since DNP is a synthetic compound it also has few unspecific interactions with biological material. DNP-labeled oligonucleotides are often used for more complex in situ investigations, like multiplex FISH.
Non-radioactive Detection in Indirect Systems In indirect detection systems, the entire repertoire of detection systems can be conjugated to the component that binds the tag on the hybridizing nucleic acid, depending on the format. Besides the reporter groups used in direct detection systems (fluorophores, luminescent dyes, gold atoms, see above), marker enzymes are often coupled to the tag-binding component, which create an optically visible luminescent or fluorescing reaction product through a catalytic substrate reaction. Well-known marker enzyme systems are bacterial alkaline phosphatase (AP), horseradish peroxidase (POD), β-galactosidase (β-Gal), luciferase, and urease.
28
Techniques for the Hybridization and Detection of Nucleic Acids
747
The alternatives for detection can be summarized as follows:
optical systems: mixtures of indolyl derivatives and tetrazolium salts or diazonium salts for coupled redox reactions;
chemiluminescent systems: dioxetane derivatives, (iso-)luminol derivatives, acridinium
esters, aequorin; electrochemiluminescent systems: Ru2+ or phenanthroline complexes; bioluminescent systems: luciferase derivatives; fluorescent systems: Attophos, Eu3+ or La3+-complexes, -micelles, or -chelates; FRET systems: fluorescein and rhodamine derivatives; fluorescence quencher systems: fluorescein (FAM) and rhodamine (TAMRA) derivatives as the fluorescing component, Black Hole Quencher (BHQ), TAMRA-, dabcyl-, cyaninederivatives as quenchers. metal precipitating systems: silver deposition on antibody-bound gold atoms (immunogold); electrochemical systems: urease-catalyzed pH changes; in situ detection systems: fluorescence in situ hybridization (FISH), primed in situ hybridization (PRINS), chromosome painting, multiplex FISH (M-FISH with SKY probes), comparative genome hybridization (CGH).
In addition, the following amplification reactions can be coupled to the primary signal generator (Section 28.5):
probe crosslinking: generation of crosslinked structures (e.g., Christmas trees, probe brushes);
crosslinking of binding components: examples include polystreptavidin, polyhapten, PAP, APAAP;
poly-enzymes: crosslinked marker enzymes, for example, poly-AP; signal cascades: for example, NAD+/NADH + H+ cycles coupled to a redox color reaction that recycles the substrate for the next round. Table 28.3 summarizes the most important systems for optical, luminescent, and fluorescent detection. Optical Detection Optical enzymatic detection systems are based on the conversion of colored substrates, coupled with a change in the wavelength of absorption. Either colored precipitates (blot or in situ applications) or colored solutions (for quantitative measurements) are used. The most important substrates for colored precipitates are mixtures of 5-bromo-4-chloro-indoxyl phosphate (BCIP, Figure 28.20) or β-galactosidase (X-gal) and nitro-blue tetrazolium salt (NBT, Table 28.3), giving rise to a deep blue–violet precipitate (indigo/diazonium salt) after cleavage of the phosphate in a coupled redox reaction. Luminescence Detection Luminescent systems are based on the chemical, biochemical, or electrochemical activation of substrates, which emit light as the atoms return to their ground state. Chemiluminescence Common substrates like AMPPD or AMPPG (Table 28.3) are used for the detection of chemiluminescence; they form intermediate dioxetanes that decompose under chemiluminescence. After enzymatic cleavage of the phosphate or β-galactoside residue, the dioxetane coupling becomes labile. The AMPD anion is created in an unstable, excited state and it decays by emitting light at a wavelength of 477 nm (Figure 28.21).
Figure 28.20 The coupled optical redox reaction BCIP/NBT.
748
Part IV: Nucleic Acid Analytics
Figure 28.21 Mechanism of the dioxetane chemiluminescence reaction. The target nucleic acid is anchored to a solid support by a biotin–streptavidin (SA) interaction, thus marking it with the DIG/ AP system. AP then activates the chemiluminescent substrate.
Several derivatives differ in terms of the stabilizing moiety and the stabilizers in the substrate solutions (e.g., LimiphosTM, CSPD® , CDPstar® ). The different formulations lead to different rates of decay or to different light intensities. The light emission can also be modulated by the addition of fluorescent detergents, which surround the dioxetane molecules like micelles. Depending on the type of additive, blue, red, or green secondary fluorescence (aquamarine, ruby, emerald) is created. POD catalyzes the oxidation of luminol. The sensitivity of the chemiluminescence is increased in enhanced chemiluminescence by the addition of certain phenols (e.g., p-iodophenol), naphthols (e.g., 1-bromo-2-chloronaphthol), or amines (e.g., p-anisidine). In addition to their use in direct detection systems, Ru2+ complexes (Section 28.4.3) are also used in indirect systems, after coupling with hapten-specific antibodies, for chemiluminescent detection after electrochemical excitation. Electrochemiluminescence
Bioluminescence Owing to their high sensitivity, luciferase enzymes from fireflies (Photinus pyralis) or bacteria are used. The eukaryotic enzyme catalyzes the conversion of luciferin into oxyluciferin. In an AP-coupled indicator reaction, D-luciferin-O-phosphate is the substrate for AP. After cleavage of the phosphate, luciferase converts the formed D-luciferin in the presence of ATP, O2, and Mg2+ ions into oxyluciferin, thereby emitting light (Figure 28.22).
Figure 28.22 Mechanism of the luciferin bioluminescence reaction. Analogous to the chemiluminescence reaction in Figure 28.21, the target nucleic acid is anchored to a solid support by a biotin: streptavidin (SA) interaction marking it with the DIG:AP system. AP then activates the bioluminescent substrate.
28
Techniques for the Hybridization and Detection of Nucleic Acids
Alternatively, a combination of the enzyme glucose-6-phosphate-dehydrogenase, NAD(P) H-FMN-oxidoreductase, and bacterial luciferase can be used. The FMNH2 formed in the redox reaction is oxidized in the presence of decanal and O2, resulting in the emission of light. Renilla luciferase, from sea anemones, catalyzes the bioluminescent oxidation of coelenterazine. In the presence of green fluorescent protein (GFP) this leads to green secondary fluorescence at a wavelength of 508 nm. The Renilla enzyme is often used indirectly as a secondary reporter molecule in the form of a biotin conjugate.
749
Green Fluorescent Protein (GFP), Section 7.3.4 and 34.4.3
Fluorescence Detection Fluorescent light is created by absorbing primary light (broadband or monochromatic laser light); the excited fluorescent molecule returns to its ground state by emitting longer wavelength secondary light. Because of the possible overlap of the incident primary light and emitted secondary light, background signals can occur. Background effects are reduced by the geometry of the detector (perpendicular detector orientation) or by timeresolved fluorescence (TRF) with Eu3+- or Tb3+-complexes, micelles, or chelates (Section 28.3.3). Coupling several of these TRF fluorophores to the four binding sites via indirect coupling by biotin:streptavidin amplifies the resulting signal. Indirect fluorescence can be detected by coupling the previously described direct fluorescence markers (e.g., fluorescein or rhodamine) to hapten binding antibodies (e.g., anti-DIGfluorescein or anti-DIG-rhodamine conjugates). FRET Detection Probes marked with interacting fluorescent components are used for FRET detection. The two components form a fluorescence energy resonance transfer (FRET) pair. One component absorbs the primary light and transfers it in the form of energy to the second component, if it is in immediate physical proximity (Figure 28.8). The second component emits the absorbed energy in the form of longer wavelength secondary light. The longer wavelength light is selectively measured to separate the signal from the input light used to excite the FRET pair. A modern homogeneous detection system is the 5´ -nuclease system (TaqMan). In this wellknown homogeneous detection system the probe is labeled with a marker pair consisting of a fluorescent marker and a quencher (Figure 28.5). For HybProbes two directly adjacent probes are used, marked with a FRET pair at their vicinal ends. In the case of molecular beacons one probe carrying a FRET pair on both ends (termini) is sufficient; see Figure 28.23 for FRET components: fluorophores, and quencher (c, D).
Förster Resonance Energy Transfer (FRET), Section 7.3.7
In Situ Detection Fluorescein and rhodamine derivatives are usually employed as the fluorescent markers for in situ detection. These are directly or indirectly bound to the probes. DIG, biotin, or dinitrophenol (DNP) are used as the interacting components for indirect binding. For SKY probes, direct or indirect probe labeling with different marker combinations (Orange, Texas Red, Cy5, Spectrum Green, and Cy5.5 or DIG:Cy5.5 or FITC, Cy3, Bio-Cy3.5, Cy5, DIG-Cy7) as fluorescent markers can create up to 24 different colors.
Figure 28.23 Examples of FRET components: (a) 6-carboxyfluorescein (FAM), (b) tetramethylrhodamine (TAMRA), (c) Cy5, and (d) dabcyl.
750
Part IV: Nucleic Acid Analytics
28.5 Amplification Systems Often, detection is coupled with nucleic acid amplification procedures. Three types of amplification formats are known:
target amplification: amplification of the nucleic acid to be detected; target-specific signal amplification: signal amplification coupled with target hybridization; signal amplification: amplification of the signaling components. Table 28.4 gives an overview of various amplification reactions. Target amplification procedures have many advantages. In most cases, they lead to an exponential amplification and Table 28.4 Overview of amplification reactions. Source: Kessler, C. (1994) Non-radioactive analysis of biomolecules. J. Biotechnol., 35, 165–189; and Kessler, C. (ed.) (2000) Non-radioactive Analysis of Biomolecules, Springer, Berlin, Heidelberg. Type of amplification
Examples
Target amplification Replication Temperature cycles
PCR: polymerase chain reaction
cDNA synthesis/temperature cycles
RT-PCR: PCR connected with cDNA synthesis
Isothermal cycles
PCR: in situ PCR
Transcription Cyclical isothermal cDNA synthesis
NASBA: nucleic acid sequence based amplification TMA: transcription mediated amplification TAS: transcription based amplification system 3SR: self-sustaining sequence replication
Increased rRNA copy number
16S/23S rRNA probes
Target specific signal amplification Replication Isothermal replication cycles
Qβ replication
Ligation Temperature cycles
LCR: ligase chain reaction
Replication/ligation Temperature cycles
RCR: repair chain reaction
Probe hydrolysis Isothermal cyclic RNA hydrolysis
CP: cycling probes
Displacement of indicator probes on flap structures by cleavase enzymes and amplification of indicator probes
Invader technology
Signal of amplification Tree structure Probe network
Hybridization trees, branched probes
Network of indicator molecules
PAP: peroxidase: anti-peroxidase complex APAAP: alkaline phosphatase:anti-alkaline Phosphatase complex
Enzyme catalysis Enzymatic substrate turnover
ELISA: enzyme linked immunosorbent assay
Polyenzyme
Enzyme gel conjugate
Coupled signal cascades Cyclic NAD+/NADP+ redox reaction
Self redox cycle
28
Techniques for the Hybridization and Detection of Nucleic Acids
therefore to high amplification rates (106–109). In addition, the complexity of the detection system is significantly decreased, since only the targets to be detected are amplified and not unspecific sequences. An example of a target amplification is the polymerase chain reaction (PCR), in which the section of nucleic acid to be detected is enzymatically reproduced in a primer-dependent reaction. This reduction in complexity is absent in plain signal amplifications; consequently, for complex genomic DNA, like the human genome, they are only used in combination with ceding PCR target amplification procedures. An example of a signal amplification combined with target amplification is the enzymatically catalyzed conversion of dyes or luminescencegenerating substrates (ELISA). In addition, plain signal amplification only leads to linear signal amplification and therefore lower amplification rates (10–103); the result is lower sensitivity of the detection reaction. Since unspecific hybridizations are also amplified, this often results in a less favorable signal-to-noise ratio, which decreases the specific detection of the target sequence. The sensitivity reached in systems without target amplification, but with ELISA signal amplification, lies in the range of picograms (10–12 g) to femtograms (10–15 g). In combination with target amplification systems like PCR, the sensitivity goes up to the attogram (10–18 g) range. Such reactions allow the detection of single molecules. In this range the sensitivity reached in practice is no longer limited by the sensitivity of the detection system, instead it is limited by statistical effects relating to preparation of the test sample.
Polymerase Chain Reaction, Chapter 29 ELISA, Section 5.3.3
While signal amplification is an important component of protein or glycan analysis, only nucleic acid analysis allows amplification of the target. This enables the combination of both types of amplification for the detection of nucleic acids, which leads to very high detection sensitivities.
28.5.1 Target Amplification Target Elongation Target elongations are thermocyclic reactions in which both strands of the DNA to be detected are replicated. The most important method for target elongation is PCR. RNA must first be transcribed into cDNA with reverse transcriptase (RT PCR) prior to use in a PCR. In homogeneous detection systems, PCR amplification is often carried out in TaqMan or LightCycler format and fluorescence detection is employed. Strand displacement amplification (SDA) is another means to amplify DNA, in this case under isothermal conditions. In the case of in situ PCR, these amplification reactions can be carried out in fixed tissues or cells for the specific amplification of target sequence fragments. Target Transcription The amplification reactions involved in transcription are isothermal and cyclic through reverse transcription and transcription of an intermediately formed transcription unit. The starting material is target RNA, which is converted into cDNA with the help of promoter-containing primers, forming intermediate double-stranded DNA molecules containing a T7, T3, or SP6 promoter. These intermediary transcription units are transcribed into the starting RNA. There are different ways to carry out the transcription amplification. The reactions differ, among other things, in that only one or both primers carry promoter sequences. One variant, nucleic acid sequence-based amplification (NASBA), is described in the following section. An important alternative is transcription-mediated amplification (TMA).
Nucleic Acid Sequence-Based Amplification (NASBA), Section 29.6.1
In Vivo Amplification Special systems for the detection of bacteria exploit the fact that ribosomal RNA includes species-specific sequences and in vivo are present at a copy number of 103 to 104, so they can be easily detected. The indicator sequences are the variable regions of the rRNA or intergenic spacer regions, i.e. segments between the rRNA genes.
28.5.2 Target-Specific Signal Amplification This type of reaction does not involve the amplification of the target itself, instead a nucleic acid tag or oligonucleotide, which hybridizes with the target, is replicated or changed. These reactions have several disadvantages relative to direct amplification of the target: For one thing, it involves a pure detection reaction, since no new target sequences are created. For another, these detection reactions have a limited specificity, since a target-coupled signal is increased and therefore unspecific amplification products cannot be filtered out as, for example, they would be with a subsequent hybridization with a target-specific probe. An example for target-specific signal amplification is the ligase chain reaction (LCR), which will be briefly described in the next chapter. Another example is the cleavase reaction, which
751
Ligase Chain Reaction (LCR), Section 29.6.4
752
Part IV: Nucleic Acid Analytics
involves cleavage of a mismatched end of a probe, a flap, by the enzyme cleavase, which recognizes and cuts such ends. This leaves an extra base on the end of the cleaved free end of the oligonucleotide, which then allows the amplification of a signal sequence after binding to a complementary indicator probe, which allows elongation of its 3´ end.
28.5.3 Signal Amplification This sort of amplification includes reactions to boost the resulting signal, instead of the number of targets, and brings only limited increases in sensitivity (10–103). Unspecific reactions mentioned for the target-independent amplifications can lead to unspecific signals. Nevertheless, the signal amplifications are of great value, particularly when combined with target amplification reactions. Branched Structures Branched structures are built up from forked probes, which contain target-specific sequences as well as sequences for the generation of signal (Christmas trees, probe brushes). The branched probes are complemented by universal detection probes, which are linked to a marker enzyme, such as AP. The primary, bivalent probes are composed of sequences that have specific sequences complementary to the branched secondary probes. The secondary branched probes are composed of sequences complementary to the primary probes as well as branched structures complementary to a third probe. The third components are universal, enzyme-labeled AP probes. By hybridization between the target, primary (bivalent), secondary (branched), and tertiary (AP-labeled) probes, complex structures are built up that lead to amplification of the signal. An example is the use of branched DNA (bDNA) for the identification of hepatitis C virus (Figure 28.24). Additional sensitivity can be achieved through the cassette-like attachment of several branched structures to a target. A disadvantage of this system is, besides the limited amplification rates, the complex set up and therefore the limited ability to control the branched structures, as well as the possibility of the occurrence of unspecific hybridization. These can lead to unspecific background reactions and therefore to a poor signal-to-noise relationship, which limits its sensitivity. Enzyme Catalysis The amplification of signals through the enzymatic conversion of substrate with the help of marker enzymes is used in direct and indirect detection systems (e.g., AP, POD, β-Gal; Section 28.4.3). Enzyme catalysis as part of the detection reaction
Figure 28.24 Branched structures for signal amplification. The structural components are the capture probe, the bivalent primary detection probe, the branched secondary probes, and the universal AP detection probes. The capture probe can also be covalently bound directly to a membrane.
28
Techniques for the Hybridization and Detection of Nucleic Acids
753
Figure 28.25 Signal amplification by coupled cyclic ADH: diaphorase redox reaction. ADH: alcohol dehydrogenase; DP: diaphorase.
leads, depending on the reaction format and type of marker system, to up to a 103-fold increase in sensitivity. The use of polyenzymes leads to an additional increase in sensitivity by a factor of 3–5. In this context, coupling of PCR and enzyme catalysis with luminescent substrates is often the best choice. For example, the DIG hapten (Section 28.4.3) is often incorporated into the amplicon by the primers or nucleotides during PCR; the DIG-marked amplicon is subsequently detected with high sensitivity with the help of AP-antibody conjugates by the catalytic conversion of indolyl or dioxetane substrates. By the combination of amplification types, sensitivities of up to the detection of single molecules are achieved. Coupled Signal Cascades
For this reaction type, a primary substrate conversion is coupled to a secondary enzymatic reaction. If the secondary reaction is cyclic, a signal cascade results. An example of a signal cascade is the self-redox cycle shown in Figure 28.25. The primary reaction is catalyzed by the marker enzyme alkaline phosphatase (AP) that is linked, either directly or indirectly, with the hybridization probe. Primary NADP is dephosphorylated by AP to NAD. The formed NAD activates a secondary redox cycle, in which it is reduced to NADH by the enzyme alcohol dehydrogenase (ADH), coupled with the oxidation of ethanol to acetaldehyde. The secondary reaction cycle is completed by the enzyme diaphorase (DP), which re-oxidizes the NADH to the original NAD in a further coupled reaction and at the same time reduces the dye NBT-violet to deeply colored formazine. Through the enzyme diaphorase, the secondary reaction cycle is closed. The soluble formazine can be quantitated photometrically (λ = 465 nm); the signal amplification leads to a 10- to 100-fold increase in sensitivity or, alternatively, to much shorter reaction times. Traces of NAD contamination in the primary substrate NADPH, which accumulate during extended storage, generate a background, which can critically compromise the sensitivity.
Further Reading Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G., Smith, J.A., and Struhl, A. (1987– 2005) Current Protocols in Molecular Biology, vol. 1–4, Suppl. 1–69, Greene Publishing Associates and Wiley-Interscience, New York. Clarke, J.R. (2002) Molecular diagnosis of HIV. Expert Rev. Mol. Diagn. 2, 233–239. Griulietti, A., Overbergh, L., Valckx, D., Decallonne, B., Bouillon, R., and Mathieu, C. (2001) An overview of real-time quantitative PCR: applications to quantify cytokine gene expression. Methods, 25, 386–401. Hares, B.D. and Higgins, S.J. (1995) Gene Probes, vol. 1 and 2, IRL Press, Oxford.
754
Part IV: Nucleic Acid Analytics International Human Genome Sequencing Consortium (2004) Finishing the euchromatic sequence of the human genome. Nature, 431, 931–945. Keller, G.H. and Manak, M.M. (1993) DNA Probes, 2nd edn, Stockton Press, New York. Kessler, C. (1991) The digoxigenin (DIG) technology – a survey on the concept and realization of a novel bioanalytical indicator system. Mol. Cell. Probes, 5, 161–205. Kessler, C. (ed.) (1992) Non-radioactive Labeling and Detection of Biomolecules, Springer, Berlin, Heidelberg. Kessler, C. (ed.) (2000) Non-radioactive Analysis of Biomolecules, Springer, Berlin, Heidelberg. Kidd, J.M. et al. (2008) Mapping and sequencing of structural variation from eight human genomes. Nature, 453, 56–64. See also http://www.hgsc.bcm.tmc.edu/ Kricka, L.J. (1995) Nonisotopic Probing, Blotting, and Sequencing, Academic Press, San Diego. Kricka L.J. (2002) Stains, labels and detection strategies for nucleic acids assays. Ann. Clin. Biochem,. 39, 114–129. Lander, E.S., Linton, L.M., and Birren, B. et al. (2001) Initial sequencing and analysis of the human genome. Nature, 409, 860–921 and 411, 720 and 412, 565. Lee, H., Morse, S., and Olsvik, O. (1997) Nucleic Acid Amplification Technologies, Eaton Publishing, Natick. Marshall, A. and Hodgson, J. (1998) DNA chips: an array of possibilities. Nat. Biotechnol., 16, 27–31. McPherson, M.J., Quirke, E., and Taylor, G.R. (1996) PCR – A Practical Approach, vol. 1 and 2, IRL Press, Oxford. Nuovo, G.J. (1997) PCR In Situ Hybridization: Protocols and Applications, 3rd edn, Lippincott-Raven Press. Nussbaum, R.L., McInnes, R.R., and Willard, H.F. (2001) Thompson & Thompson Genetics In Medicine, 6th edn, W.B. Saunders, Philadelphia. Ørum, H., Kessler, C., and Koch, T. (1997) Peptide nucleic acid, in: Lee, H., Morse, S., and Olsvik, O. (eds) (1997) Nucleic Acid Amplification Technologies, Eaton Publishing, Natick. Persing, D.H., Smith, T.F., Tenover, E.C., and White, T.J. (1993) Diagnostic Molecular Microbiology: Principles and Applications, American Society for Microbiology, Washington. Ramsay, G. (1998) DNA chips: state-of-the art. Nat. Biotechnol., 16, 40–44. Reischl, U., Wittwer, C., and Cockerill, F. (eds) (2002) Rapid Cycle Real-Time PCR – Methods and Applications, Springer-Verlag, Berlin, Heidelberg. Tyagi, S., Bratu, D., and Kramer, E.R. (1997) Multicolor molecular beacons for allele discrimination. Nat. Biotechnol., 16, 49–53. Venter, J.C., Wittwer, C.T., and Herrmann, M.G. et al. (2001) The sequence of the human genome. Science, 291, 1304–1351; Venter, J.C., Wittwer, C.T., and Herrmann, M.G. et al. (2001) Science, 292, 1838.
Polymerase Chain Reaction Nancy Schönbrunner,1 Joachim Engels,2 and Christoph Kessler3 1
Roche Molecular Systems, Inc., 4300 Hacienda Dr., Pleasanton, CA 94588, USA Goethe University Frankfurt, Institute of Organic Chemistry and Chemical Biology, Department of Biochemistry, Chemistry and Pharmacy, Max-von-Laue Straße 7, 60438 Frankfurt am Main, Germany 3 PD Christoph Kessler, Consult GmbH, Icking, Schloßbergweg 11, 82057 Icking-Dorfen, Germany 2
The polymerase chain reaction (PCR), a method for the amplification of nucleic acids, is one of the greatest scientific discoveries of the recent past. Without exaggeration it can be said that this discipline, though still young, has revolutionized molecular biology. The possibilities of PCR appear to be almost unlimited. The number of articles in scientific journals about improvements, new applications, and breakthroughs in the areas of basic and applied research, as well as medicine, diagnostics, and other areas, grows daily. The history of its discovery, during a night drive through the mountains of California, is portrayed in an impressive article by Kary B. Mullis, its inventor. He writes: “The polymerase chain reaction was not the result of a long development process or a lot of experimental work. It was invented by chance on a late evening in May, 1983 by the driver of a gray Honda Civic during a drive along Highway 130 through the mountains between Cloverdale and Anderson Valley in California.” Since the discovery of PCR in 1983, around 800 000 publications in diverse journals and magazines have appeared as of March 2014. PCR applications lie in the most diverse areas: molecular biological basic research, cloning of defined sequence segments, generation of samples, genetics, medicine, genome diagnostics, forensics, food sector, plant breeding, agriculture, environment, and archeology, to just name a few. The PCR reaction has been integrated into the most diverse processes: PCR amplification with subsequent electrophoretic separation of the amplification products, cloning and sequencing of the amplified sequences, allelic PCR for the elucidation of mutations, analysis of the PCR products in blot detection formats, coupling with quantitative heterogeneous and homogeneous detection methods, or in situ PCR for the amplification of particular target sequences directly in tissues or cell cultures. For biological samples an important requirement is for the upfront sample preparation to extract and concentrate nucleic acid from the biological matrix and remove inhibitors. Alternatives to PCR have also been established, with NASBA (nucleic acid sequence-based amplification) and TMA (transcription-mediated amplification) amplification procedures being the best known. While PCR begins with DNA as the starting nucleic acid and the amplification is accomplished by temperature cycles, NASBA and TMA amplifications begin with RNA as the starting nucleic acid and the amplification cycles are carried out at a constant temperature, such as 42 °C.
29.1 Possibilities of PCR The PCR method as a means to amplify any given nucleic acid segment relies on an idea as brilliant as it is simple. To estimate the value of such an amplification, we will begin by considering conventional analysis procedures. For example, gel electrophoresis has a lower Bioanalytics: Analytical Methods and Concepts in Biochemistry and Molecular Biology, First Edition. Edited by Friedrich Lottspeich and Joachim Engels. 2018 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2018 by Wiley-VCH Verlag GmbH & Co. KGaA.
29
756
Part IV: Nucleic Acid Analytics
Figure 29.1 Schematic of the polymerization of DNA. In the presence of a primer with free 3´ -OH ends and free nucleotides (dNTPs) DNA polymerases convert single-stranded (ssDNA) into double-stranded DNA.
limit of detection of about 5 ng DNA. If you calculate the number of molecules it ends up being 1010 for a fragment 500 base pairs (bp) in length. To increase the sensitivity, the DNA in such gels can be transferred to solid supports in order to detect them with radioactive or nonradioactive probes. This increases the sensitivity such that approximately 108 molecules can be detected; however, for many diagnostic purposes this is not even close to being sufficient. In viral diagnostics titers of under 1000 particles per milliliter blood are often present. Even the most sensitive current analysis procedures do not come close to reaching the sensitivity of PCR: it is theoretically capable, under optimal conditions, of generating up to 1012 identical molecules from a single nucleic acid segment in a few hours, which then are available for diagnostics or other analytical method (Section 29.5). How is this accomplished? PCR takes advantage of the ability of DNA polymerases to duplicate DNA. A requirement for this is a short segment of double stranded DNA with a free 3´ OH end, which can correspondingly be extended (Figure 29.1). Mullis realized that such a short segment can be created artificially by adding DNA fragments of about 20 nucleotides in length, also referred to as oligonucleotides or primers. These bind, or anneal, to the ends of the DNA strand to be amplified and can now be extended by the polymerase. If the newly synthesized double strand DNA is denatured by increasing the temperature, new primer molecules can bind upon cooling and the process can begin again. If two primers are added to the sample, one of which binds to the sense strand, the other the antisense strand, after each cycle of new synthesis and denaturation a doubling of the segment between the primers takes place. PCR leads to an exponential amplification, since the new strands are available as templates for the next round of amplification (Figure 29.2). And something else was recognized by Mullis’s group: if one uses a temperature stable DNA polymerase, such as those found in organisms that live in hot springs, it is possible to run the reaction without interruption.
29.2 Basics 29.2.1 Instruments The required reagents and tools for PCR are quite simple. In the course of time they have been steadily improved in terms of data security, throughput, and user comfort. The first thermocyclers consisted of three water baths set to different temperatures and the samples were moved by hand and with the aid of a stopwatch from one bath to the next. Later, robotic arms took over this task. Today, they are relatively compact devices that hold the PCR samples in a 96-well plastic micro-volume plate in a metal block, which is systematically heated and cooled. A significant difference of modern thermocyclers is their heating technology, which employs either Peltier elements or works with the aid of liquids. The newest developments in this field are aimed at a drastic increase in speed through miniaturization – PCR in a glass capillary tube with a very low volume (LightCycler® ) or the amplification and detection in a single device (PCR combined with simultaneous FRET detection (TaqMan®)). Devices of this sort allow real time detection of a PCR product during the amplification (Section 29.7).
29 Polymerase Chain Reaction
757
Figure 29.2 Schematic of PCR. The number of DNA segments can theoretically double in each cycle of a process of denaturation, primer annealing, and primer extension. The first two cycles are shown. The number of the copies grows exponentially with each round.
758
Part IV: Nucleic Acid Analytics
29.2.2 Amplification of DNA Probe Preparation
If the starting material for PCR comes from biological material, the nucleic acids must first be released from, for example, virus particles, bacteria, or cells and separated from interfering components like proteins, lipids, or inhibitors, such as hemoglobin degradation products in blood. Probe preparation, besides releasing and purifying the samples, leads to the concentration of the nucleic acids. Various methods are used: A common method uses the property that DNA binds to glass in the presence of chaotropic salts. If the glass particles have a magnetic core, they can be captured by placing a magnet on the vessel wall, separated from interfering substances by wash steps, and the surface bound nucleic acids eluted and concentrated into a small volume of solution by suitable elution buffer. The advantage of this method is that it is generic, thus it can be used on different target sequences. Another method uses hydroxyapatite columns. If the goal is to purify particular target sequences for PCR, capture beads, which have capture probes on their surface, can be used. The capture probe binds to the particle by the binding pair streptavidin (adsorption to the particle surface) and biotin (tagging the capture probe). Complementary target sequences bind to the capture probe, and if the particles employed are also magnetic then separation from interfering substances can take place by applying a magnetic field. Cycles A typical PCR run consists, as a rule, of three stages at different temperatures. This is particularly easy to visualize as a temperature–time profile (Figure 29.3). The reaction is started by heating to 92–98 °C. This step serves to denature DNA into its single strands. Since the initial DNA is present in a complex, high molecular weight structure, a time of 5–10 min is chosen to ensure that even GC-rich sequences are denatured. The second step of the reaction is annealing of the primer. For this to take place, the reaction must be cooled to a probe-specific temperature. Annealing of the primer to a single strand of the target sequence critically influences the specificity of the PCR. The annealing has a critical influence on the specificity of the PCR. After the annealing step the temperature is increased to 72 °C, the optimum temperature of the enzyme used, Taq DNA polymerase. For both the annealing and extension steps a time of less than a minute is usually sufficient. Only for very long PCR products of over a kilobase is the extension time lengthened in order to be sure that the complete strand is synthesized. This is important, since only completely extended DNA strands can function as templates in the next cycle of PCR. In the next step of the reaction it is again heated to 92–95 °C, in order to separate the product doublestranded DNA into single strands. Since in the ideal case only the newly synthesized segments are
Figure 29.3 Temperature/time profile of PCR (two and three step).
29 Polymerase Chain Reaction Table 29.1 Summary of PCR Master mixes (100 μl end volume). Reagent
Final concentration
Taq DNA polymerase
2–5 units 1×
10 × Taq-Buffer (100 mM Tris/HCI pH 8.3; 500 mM KCI)
1×
10 mM nucleotide mixture (dATP, dCTP, dGTP, dTTP)
0.2 mM
MgCI2
0.5–2.5 mM
Primer Ι
0.1–1 μΜ
Primer II
0.1–1 μΜ
Η2O
Variable
Templates
Variable
present as double-strands, after the second cycle a much shorter denaturation time of 10–60 s is necessary. Newer protocols combine annealing and extension into a single step, usually at 62–72 °C. This makes a two-step PCR out of a three step one. For most applications, enough product for further analysis is present after 30–35 cycles. Amplification only requires 40–50 cycles in exceptional situations or for nested PCR (Section 29.3.1). Enzyme
The most important requirements for DNA polymerases in PCR are to have a high processivity, the ability to synthesize long stretches of DNA, and/or rapid binding and extension kinetics at 72 °C, as well as a very high temperature stability at 95 °C. The polymerases that have these characteristics are Taq, Tth, Pwo, and Pfu DNA polymerases. Taq DNA polymerase is the most commonly used in standard protocols. Tth DNA polymerase, like Taq DNA polymerase, has a high 5´ -3´ polymerization activity, but in addition Tth possesses reverse transcriptase activity under certain conditions. This is explained in more detail in Section 29.2.3. Pwo and Pfu DNA polymerases have, in addition to their polymerization activity, a 3´ exonuclease activity. This activity is referred to as proofreading activity, since these enzymes are capable of recognizing and removing an incorrectly incorporated nucleotide so that the mistake can be corrected with a new round of polymerization. Besides the single polymerases, mixtures of the enzymes are often offered commercially. Examples of the reagents needed for amplification according to standard protocols are listed in Table 29.1. Buffer
The buffer conditions need to be set according to the requirements of the polymerase. The ion concentration of the buffer, usually provided with the enzyme, is very important, since it influences the specificity and processivity of the total reaction. The buffers usually provided with Taq DNA polymerase usually come with and without magnesium chloride. For the optimization of a new PCR (Section 29.2.4), buffers without magnesium chloride are better, since the range of conditions is much greater with an additional magnesium chloride solution. Other possible additives are bovine serum albumin (BSA), Tween 20, gelatin, glycerol, formamide, and DMSO. This can lead, in some cases, to stabilization of the enzyme and to optimized primer annealing.
Nucleotides The concentration of the four deoxynucleotide triphosphates (dATP, dCTP, dGTP, dΤΤP) is usually in the range 0.1–0.3 μΜ. All four dNTPs should be present in an equimolar amount. The only exception is when other nucleotides are used for amplification, for example dUTP. Nucleotide analogs are often used in excess, mostly 3 : 1, since they are not incorporated as well by Taq DNA polymerase. Primers
Important prerequisites for an optimal PCR, measured on the basis of specificity and sensitivity, is the selection of the primers. There are four basic types of primers:
sequence-specific primers, degenerate primers, oligo(dT) primers (only for RNA), short random primers (usually for RNA).
759
760
Part IV: Nucleic Acid Analytics
Oligo(dT) primers and random primers (hexanucleotides) are usually used for the amplification of RNA. They are described in the following section. The use of degenerate primers is limited to particular questions and will be treated separately in Section 29.3.3. The primers used most often, by far, are sequence-specific primers. To ensure that they actually bind to their target sequences, there are several rules that must be followed. Some of the most important are mentioned briefly here.
Primer requirements 1. 2. 3. 4. 5. 6. 7. 8.
Specificity of the Hybridization and Stringency, Section 28.1.2
at least 17 nucleotides long (usually 17–28 nt); balanced G/C to A/T content; melting point between 55 and 80 °C; melting point of the forward and reverse primers should be as close as possible; no hairpin structures, particularly on the 3´ -end (Figure 29.4); no dimerization: neither by itself nor with the second primer (Figure 29.4); no G/C nucleotide on the 3´ -end, if possible, since this increases the danger of mispriming; no “strange” base sequences like poly(A), more than four Gs, or long G/C stretches, if possible.
Today diverse computer programs support the user in the search for suitable primer sequences and, in addition, show their melting points. This can be calculated with various formulas. The simplest of these formulas calculates a temperature of 2 °C for each A or T and for every G or C a temperature of 4 °C. For a primer of 20 nucleotides in length, a 20-mer, with a balance number of A/T and G/C, it calculates a melting point of 60 °C. Templates (Genomic DNA, Plasmids, Viral DNA) The most important influence of the target on the success of PCR is the length of segment to be amplified, the sequence of the primer binding sites, and the number of input molecules. A microgram of human genomic DNA contains 3 × 105 target sequences, provided it is a single copy gene and not a repetitive element. The same mass of a plasmid of 3 kb DNA contains 3 × 1011 molecules. In other words 1 μg genomic DNA contains as many molecules as 1 pg of plasmid DNA. This needs to be taken into account when using different templates, particularly when used for preparative purposes. The maximum amplification length of a DNA is primarily determined by the processivity of the polymerase used. Today there are enzymes and enzyme mixtures that allow amplification of fragments up to 40 kb. In such cases the extension time must be much longer, up to 30 min per cycle. In general, short segments of 0.1–1 kb in length are favored, since these can be optimally amplified with PCR. Besides the length and the number of molecules, the primer binding sites also determine whether PCR is successful. To avoid mispriming, repetitive sequences should be excluded and instead a single copy site should be selected for the primers.
Figure 29.4 Secondary structures of primers. For the selection of primers it is important to avoid secondary structures. Complementary segments between the sense and antisense primers must also be factored into the analysis.
29 Polymerase Chain Reaction
761
29.2.3 Amplification of RNA (RT-PCR) Many methods to analyze RNA exist in molecular biology, such as Northern blots, in situ hybridization, RNAse protection assays, and nuclease S1 analysis, to name just a few. All these methods have the disadvantage, however, that they are time consuming and often not sensitive enough. This is particularly true for the analysis of low copy number transcripts or for viral RNA, which is only present at very low starting concentrations. Adaptation of PCR technology to allow the amplification of RNA led to many new discoveries and more sensitive diagnostics. It can be used to investigate gene expression in cells or, with the aid of quantitative RT-PCR, to determine the amount of a specific mRNA or viral RNA. In addition, with oligo(dT) priming (see below), complete cDNA banks can be created, which enables an overview of tissue-specific expression.
In situ Hybridization, Section 35.1.4 Ribonuclease Protection Assay (RPA), Section 34.1.3 Nuclease S1 Analysis of RNA, Section 34.1.2
Enzymes Since the starting RNA cannot be directly used as a temple by Taq DNA polymerase, the RNA must first be transcribed into DNA to be amplified. There are several enzymes, called reverse transcriptases (RTases) or RNA-dependent DNA polymerases, for this purpose. The newly synthesized strand is termed complementary DNA (cDNA) and the step in which this cDNA is created is called reverse transcription (RT). The complete reaction of RT and amplification is therefore called RT-PCR (Figure 29.5). Several different reverse transcriptases can be used for this purpose. MMLV RTase The enzyme comes from Moloney murine leukemia virus, has an optimum temperature of 37 °C, and is able to synthesize cDNA up to a length of 10 kb due to its high processivity. The optimum pH is 8.3. AMV RTase Isolated from avian myeloblastosis virus (AMV) from birds, its optimum temperature is 42 °C and it has a similarly high processivity as MMLV RTase. Its optimum pH is 7.0.
Figure 29.5 Schematic portrayal of RTPCR. Since RNA cannot be amplified directly by PCR, it must first be transcribed into cDNA. Enzymes that catalyze this step are MMLV RTase, AMV RTase, and Tth polymerase.
762
Part IV: Nucleic Acid Analytics
Tth DNA Polymerase This heat stable enzyme comes from the bacteria Thermus thermophilus. In contrast to the other two enzymes, Tth DNA polymerase possesses two activities: In the presence of manganese ions it has both RT and DNA polymerase activity. Since the Tth DNA polymerase comes from a thermophile, just like Taq polymerase, it has a temperature optimum of 60–70 °C. It is the only enzyme capable of carrying out both steps of an RT-PCR under the same buffer conditions. For the RT step a high concentration of manganese ions is optimal, but it tends to inhibit DNA polymerase. This affects the processivity of the enzyme. For this reason, Tth DNA polymerase is only able to synthesize 1–2 kb of cDNA.
Procedure After using different enzymes, there are also different possibilities to carry out the reaction for a RT-PCR. The first RT step is carried out in a relatively small volume (to increase the sensitivity). This has the advantage that the reaction conditions can be set optimally for the RT enzyme used. After the RT step, the entire reaction, or an aliquot, is removed and a “normal” PCR is carried out. The disadvantage of this procedure is an increased danger of contamination (Section 29.4), since an additional pipetting step is required. For such a two-step process, AMV or MMLV RTase is used for the RT and Taq DNA polymerase for the PCR.
In Two Reaction Tubes
In a Single Reaction Tube For the reason mentioned above, it is advantageous to carry out the whole RT-PCR reaction in one tube without the need to pipette from one tube into another. This is possible, in principle, with all three reverse transcriptases, partially in combination with Taq polymerase; however, Tth DNA polymerase is particularly well suited. In almost all cases, the lower processivity is an acceptable price to pay, since the length of the segment to be amplified is less than 2 kb. The main advantage, however, of using Tth DNA polymerase is the possible increase of the reaction temperature of the RT step to 60 °C. This temperature helps to resolve secondary structures and thereby eases annealing of the primer.
Primer
For RT-PCR, three different types of primers can be used (Figure 29.6):
Sequence-specific primers anneal specifically in both the RT step and the following
Figure 29.6 Different priming methods in RT-PCR.
amplification to the same site of the RNA or cDNA. These are most often used for diagnostic tests for the detection of viral RNA. Oligo(dT) primers consist of a segment of 12–18 dTs, which semi-specifically bind to the polyA tail of eukaryotic mRNAs. They are only useful for the RT step. For the following amplification steps further, sequence-specific primers are needed. Short random primers are a mixture of hexanucleotides of different sequences. They bind “randomly” to the RNA and lead to a pool of cDNAs of various lengths, which, like the oligo (dT) primers, are then amplified with a second set of sequence-specific primers.
29 Polymerase Chain Reaction
763
29.2.4 Optimizing the Reaction One of the most important and time-consuming parts of the establishment of a new PCR is optimization of the reaction. Among the analytical aspects, the sensitivity is particularly important. This is critical in forensic medicine or the detection of very low amounts of DNA or RNA in infectious disease. In cases where the amplification is for preparative purposes, such as by the synthesis of probes or templates for sequencing, the yield, or amount of the PCR product formed, is of primary importance. This section will explain a few of the important adjustments that can be made and a few strategies for the optimization. DNA Amplification Choice of the Primers The selection of primers requires a substantial investment of time. If the sequence segments allow, several primers pairs should be tested, since secondary structure of the target sequence is, in principle, always to be expected. If no PCR product is synthesized, it can be helpful to successively lower the annealing temperature. However, one must pay particular attention to the possibility of nonspecific amplification. Magnesium Ions An important factor, which determines the processivity and overall activity of Taq polymerase, is the concentration of magnesium ions. In the first experiments a concentration of between 0.5 and 5 mM should be explored. Additives
Many additives to the reaction solution can help stabilize Taq polymerase or annealing of the primers. These include glycerol, BSA, and PEG. Denaturation is aided by the addition of DMSO, formamide, Tween 20 Triton X-100, or NP40. Detergents may also stabilize the DNA polymerase.
Hot Start PCR
If nonspecific amplifications are a problem, in some cases a hot start PCR can help: To suppress polymerization of nonspecifically hybridized primers at low temperatures, the activity of Taq polymerase is controlled such that it only begins at higher temperatures. One possibility is to add the enzyme only after the sample has been heated. There are also commercially available antibodies against Taq polymerase that only denature at higher temperatures and set the enzyme free. There are also chemically modified versions of Taq DNA polymerase (Taq GOLD), which is inactive below 60 °C; the chemical adducts are hydrolyzed under the specific buffer conditions at elevated temperatures. Finally, some DNA polymerases are supplied with specific aptamers that inhibit activity below the melting temperature of the aptamer.
Templates
The quality of the template is also very important for a successful PCR. Sample preparation should ensure that PCR inhibitors are efficiently removed. Of particular significance are degradation products of hemoglobin, for blood preparations, and ethanol, which is often used to precipitate DNA.
RNA Amplification In addition to the points mentioned above, there are a few more aspects to factor in for RT-PCR: Single-stranded RNA often has more secondary structure than DNA. Since the formation of secondary structure is still a very complex and poorly understood process, the current computer programs are only of limited use in this aspect of primer selection. Even in cases where primer design appears optimal, the synthesis of the PCR product may fail due to secondary structures that either prevent the annealing of the primers or block extension. In this context, the use of Tth DNA polymerase has often proven advantageous, since the RT reaction takes place at 60 °C, which helps to melt such secondary structures. In some cases, adding an RNase inhibitor is recommended, since RNA is fundamentally a much more vulnerable template than DNA.
29.2.5 Quantitative PCR The quantification of nucleic acids with PCR or RT-PCR has become an essential component of diagnostic questions. This is particularly true for the diagnosis and monitoring of infectious diseases. Two examples of great significance are the AIDS-causing HIV (human
RT-PCR is principally less efficient than PCR. Even under optimal conditions only about 10–30% of the RNA present is transcribed into cDNA, which is then available for further amplification.
764
Part IV: Nucleic Acid Analytics
Figure 29.7 Typical course of a PCR. The kinetics of the PCR runs through an exponential phase into a plateau phase. The plateau comes about as a result of the build-up of inhibitors, competition between strand re-annealing with primer annealing, and because the reagents become limiting.
immunodeficiency virus) and the liver inflammation-causing hepatitis C-virus (HCV). In addition, in oncology there is interest in the quantitative measurement of mRNAs. The quantification is complicated by the fact that PCR is not a linear amplification. The exponential nature of the amplification means that small differences in reaction efficiency, such as in the presence of inhibitors in a particular sample, have profound effects on the yield of amplicon. The following equations demonstrate this: N N 0 2n
(29.1)
where N is the number of amplified molecules, Ν0 the number of molecules prior to amplification, and n the number of cycles. The number of molecules doubles under these (idealized) conditions with each cycle. In practice, however, the following formula applies: N N 0
1 E n
(29.2)
where E is the efficiency of the reaction and has a value between 0 and 1. This value depends very strongly on the degree of optimization of the PCR. Experimentally, values for an optimized PCR have been found to be between E = 0.8 and E = 0.9. The quantitative measurement is also made more difficult by the fact that towards the end of the amplification the exponential phase begins to plateau; this means that the value of E changes during the PCR (Figure 29.7). The maximum amount of product that can be generated during PCR is around 1013 molecules, but can deviate relatively strongly downwards. Many methods for quantitative measurement have been developed and continuously improved over the years:
Limiting dilution: A reference standard of known concentration is diluted in multiple steps,
amplified, and the concentration is determined that was required in the preceding PCR to generate a barely detectable amplification. A sample to be measured is then also diluted in multiple steps to measure this same point. The number of dilutions then allows conclusions to be drawn about the starting concentration of the solution. External standard: The concentration of a sample to be measured is determined by comparing the signal it generated with that of a standard of known concentration
Both methods are not able to recognize internal interference in the amplification efficiency of an individual sample. The next generation of quantitative tests was focused on internal controls and standardized amplification reactions. There are two types of standardization:
Internal endogenous standard: quantification of a so-called “house-keeping” gene. Competitive (RT) PCR: quantification using mimic fragments, which are added to the reaction and amplified along with the actual target sequence. The following explains the last three options for quantification in more detail. External Standard Samples of known concentrations are used to generate a curve to provide an external standard. The standard should be relatively similar to the intended
29 Polymerase Chain Reaction
765
Figure 29.8 Quantification using an external standard. Shown is the measurement of a sample against a standard curve that was created with a known concentration of HIV from a cell line (geq, genome equivalents).
target sequence and should use the same primers for amplification. Very suitable, for example, is an HIV cell line that contains a known number of proviral genomes. This cell line is serially diluted, processed, amplified, and the signal plotted compared to the starting concentration (Figure 29.8). After amplifying the unknown sample, the signal is looked up on the standard curve to determine the starting concentration. The disadvantage is, as mentioned, the lack of an internal control of the reaction, ensuring that it ran properly and like the other samples. It is easy to imagine that even a low level of inhibition can lead to dramatic under-quantitation. Internal Standardization DNA amounts in the samples are calibrated to internal sequences of the genome to provide internal standardization. The β-globin gene is usually used for DNA measurements (e.g., HIV provirus genomes). This requires the use of two primer pairs in a multiplex PCR (Section 29.3.4). One pair amplifies the test DNA, the other a segment of the β-globin gene. Since the amount of the β-globin gene is known and the signal after amplification is constant, this allows conclusions to be drawn about the amount of the target DNA. In contrast to external standards, this procedure allows the detection of inhibitory substances, as long as they have the same influence on both PCRs and are not sequencespecific. For the quantification of RNA, the signal can be calibrated against the signal from socalled “house-keeping” genes. These are genes that are thought to be expressed in all cells and tissues at the same level at all times. An important aspect that makes the quantitative measurement of RNAs significantly more difficult is the great variation in the efficiency of reverse transcription, in particular when cDNA is synthesized from two different RNAs. A possible way to limit this variance is by the use of artificial standards, so-called mimic fragments. Competitive (RT) PCR
An artificial, cloned standard of known concentration and containing the same primer binding sites is added to the reaction in this procedure. Since it is coamplified with the same primers as the target sequence and thus “mimics” the target it is called a mimic fragment and the reaction is called competitive (RT) PCR. Ideally the amplified mimic fragment is the same size as the target sequence. After amplification, the two are separated, either by a differential hybridization or by different restriction sites, and analyzed. Alternatively, differentially labeled probes are used to detect the target and mimic products by TaqMan PCR (Section 29.2.1). Competition between the two target sequences occurs whenever the starting amount differs by more than about three to four orders of magnitude (Figure 29.9). The starting sample for measurement is divided into about four equal aliquots and an increasing amount of RNA mimic fragment is added to each. After amplification the two signals are compared to the starting concentration of the RNA mimic fragment and at the intersection of the two curves, the starting amount of the test sample is extrapolated (Figure 29.9).
In general RNA mimic fragments, and not DNA mimic fragments, should always be used, since the greatest variance comes from the RT step, as mentioned above. In addition the RNA mimic fragments should be added at the beginning in order to control each step (sample preparation, amplification, and detection) along the way.
766
Part IV: Nucleic Acid Analytics
Figure 29.9 Competitive (RT) PCR: The sample to be measured is aliquoted and spiked with an increasing, known amount of mimic fragment. After the amplification, the same pieces of the amplicon are hybridized with the corresponding probe and the signals are compared to the starting concentrations of the mimic fragments. The starting concentration is given by the point of equivalence of the sample.
29.3 Special PCR Techniques 29.3.1 Nested PCR Nested PCR involves the use of two sets of primer pairs, an outer and an inner pair (Figure 29.10). The advantage of these methods is the increased specificity and sensitivity of the entire reaction, since the outer pair is first used to synthesize a larger amplicon (PCR
Figure 29.10 Nested PCR: The starting DNA is successively amplified in two separate PCRs. First an out primer pair produces a somewhat larger segment, the first amplicon, then an inner primer pair binds. This pair amplifies a smaller internal segment, the second amplicon, in a further 20–25 cycles. Nested PCR can lead to significantly increased sensitivity and specificity of the PCR.
29 Polymerase Chain Reaction
product) and in a second reaction the inner primer pair is used to amplify the first amplicon, which serves to eliminate the by-products of the first amplification. In general the inner pair is added to the mix after 15–20 cycles and the reaction is allowed to continue for another 20–25 cycles. The serious disadvantage of this method, however, is the drastically increased chance of contamination (Section 29.4), caused by the pipetting of the amplicon. The possibility of avoiding this danger exists with one tube nested PCR, which involves adding all four primers to the reaction at the beginning. The outer primer pairs must have a higher melting point than the inner pair so that at first only the outer pair can anneal under the reaction conditions in use. After the correct number of cycles, the annealing temperature is lowered so that the inner pair can now anneal and produce the inner product.
29.3.2 Asymmetric PCR When one of the two primers is present in excess relative to the other primer this is referred to as asymmetric PCR. These conditions lead to a selective amplification of one of the two strands. This technique is used for, among other things, sequencing PCR products (Section 29.3.5). If the goal is to hybridize the PCR product with a labeled probe after amplification, it can be advantageous to use asymmetric PCR. The strand that hybridizes to the probe is preferentially amplified, which creates a more favorable situation in the competition between renaturation of the two amplified strands and hybridization with the labeled probe. This comes at the price of the amplification no longer being exponential, but instead it quickly become linear.
29.3.3 Use of Degenerate Primers Degenerate primers are a mixture of individual molecules that differ at certain points in their sequence. They are used whenever the sequence of the amplification target is not exactly known or the sequences diverge from one another. The first case comes into play when, for example, a gene segment from a different species, whose sequence is only known for other species, needs to be amplified. Homology searches allow variable positions to be identified and corresponding degenerate primers synthesized that contain all the expected nucleotide variations. They are then used to attempt to amplify the corresponding segment from the desired species. Another use is when only the amino acid sequence of the protein is known. In such cases the amino acid sequence can be used to narrow down the possibilities of the base sequence enough so that degenerate primers can be created. Degenerate primers are also often used when the targets sequences vary from one another. This can occur in the amplification of different HIV subtypes. Even in regions that are otherwise very strongly conserved, single nucleotides can differ from subtype to subtype. The more degenerate the primers are, the greater the danger of nonspecific amplification. This is a fundamental disadvantage of this approach.
29.3.4 Multiplex PCR Multiplex PCR refers to the procedure of using multiple specific primer pairs to generate the same number of amplicons, all in one tube at the same time. It is apparent that the throughput increases drastically and the amount of work decreases at the same time. Classic uses for multiplex PCR are especially routine diagnostics. For example, the disease cystic fibrosis (CF) is due to certain mutations in the CFTR gene. However, over 100 mutations of this gene are known today, which can be spread across all 24 exons. Multiplex PCR is able to amplify many of the exons at the same time, in order to then investigate the products for point mutations. The situation is similar for other inherited diseases, such as familial hypercholesterolemia, Duchenne muscular dystrophy, polycystic kidneys, and many more.
767
768
Part IV: Nucleic Acid Analytics
Another very attractive indication is the diagnosis of several viral infections (HBV, HCV, HIV) at the same time from a single blood sample, which is of particular interest for blood banks. Similarly to the use of degenerate primers, however, it can also be difficult here to get the high complexity of the total reaction under control, which often leads to nonspecific amplification. The newest multiplex protocols have succeeded in measuring up to five viral parameters (HIV-1M, HIV-10, HIV-2, HBV, HCV) at the same time, without nonspecific by-products. Another example of a multiplexing application is a new test for sepsis, which uses primers that bind to the spacer region of rRNA of pathogenic bacteria to detect and differentiate between a whole palette of pathogenic organisms. This allows the rapid selection of a suitable antibiotic that would otherwise require days with conventional tests, such as selective cultivation of the bacteria, by which time life threatening complications such as septic shock or multiple organ failure may have arisen. The test detects Gram negative (e.g., Klebsiella pneumoniae), Gram positive pathogenic bacteria (e.g., Staphylococcus aureus), and pathogenic fungi (e.g., Candida albicans). The high specificity is demonstrated by the lack of nonspecific cross reactions with over 50 closely related bacteria.
Multiplex PCR is also used for DNA analysis in forensics and paternity testing. Up to 16 PCR amplicons are generated at the same time, which can be separated into different peaks on the basis of different primer fluorescence markers and length after gel electrophoresis. The individual heterogeneity of the amplified repetitive sequence segments show PCR multiplex patterns specific to the individual. These patterns can be stored in binary data banks and rapidly identified with the aid of search programs.
29.3.5 Cycle sequencing
Sequencing According to Sanger: the Did, eoxy Method Section 30.1.1
To sequence a PCR product, the sequence need not necessarily first be cloned into phages (M13) or plasmids; instead it can be analyzed directly. Either the product is sequenced subsequent to the PCR or sequencing takes place during the amplification reaction. The latter is referred to as cycle sequencing. Since only one primer is used in each reaction tube, the amplification is linear instead of exponential. As in Sanger sequencing, the chain termination method is usually used, with the aid of dideoxynucleotides. The sequencing takes place, therefore, in four reaction tubes, which differ in the corresponding termination mix (ddATP, ddCTP, ddGTP, DDTTP). The reaction can be started with very small amounts of DNA, which is a decisive advantage of the cycle sequencing method compared to older sequencing methods. In addition, any sort of double- or single-strand DNA can be used as the template. Cycle sequencing is used particularly often for mutation analysis, since it allows the easy investigation of certain genome segments without the need to first clone the region. A disadvantage of this method, however, is that a polymerization error of the Taq polymerase that occurs in an early cycle of the linear amplification can be interpreted as a suspected “mutation” and lead to false conclusions. In such cases the opposite strand should always be sequenced. After the reaction is complete, the products are electrophoretically separated on the basis of their differing lengths and the sequence determined. Depending on the label used for the primer, the reaction can be radioactive or non-radioactive. Modern approaches to cycle sequencing utilize differentially fluorescently labeled ddNTPs so that only a single tube reaction needs to be carried out.
29.3.6 In Vitro Mutagenesis PCR is ideally suited to introduce mutations into DNA strands in vitro and produce sufficient amounts of mutations to be useful for many purposes. The creation of mimic fragments for competitive PCR (Section 29.2.5) with such a substitution mutagenesis involves the exchange of nucleotide sequences. Mutagenic PCR can also be used to generate a diversity library that can be screened for desired features.
29.3.7 Homogeneous PCR Detection Procedures To avoid contamination (Section 29.4), real time procedures are increasingly used in closed systems, in which amplification and detection take place in a single step, without the need to
29 Polymerase Chain Reaction
open the reaction vessel. This avoids contamination during pipetting steps or by the formation of aerosols by opening of the reaction vessels. The use of closed systems and direct detection of the fluorescent signal through the thin glass wall of the capillary tubes or the tops of film-sealed microwell plates at any time during the amplification reaction has greatly reduced the danger of contamination. The formation of amplicons is followed in real time. Another advantage of this homogenous format is its high dynamic range of up to 8–10 orders of magnitude in comparison to heterogeneous formats (see Chapter 28).
29.3.8 Quantitative Amplification Procedures Besides the quantitative TaqMan format (5´ nuclease assay) with TaqMan, HybProbe, or molecular beacon probes, homogenous amplifications are also carried out with the aid of fluorescent intercalating dyes. Different amplicons can be detected in multiplex procedures by creating amplicons of different lengths, and thus different melting points, by measuring the resulting hysteresis curves during the PCR cycles. Another quantitative homogeneous amplification technique is the measurement of amplification products by fluorescence depolarization. In this format, detection takes place through an increase in the polarization, which is a result of binding of the single-strand probes to the generated amplicon. The hybridization of the detection probe to the generated amplicon increases the molecular weight of the resulting complex and thereby the polarization of monochromatic light (e.g., xenon light with a monochromator at 495 nm) shone through it. This technique is also used in association with strand displacement amplification (Section 29.6.2).
29.3.9 In Situ PCR Recently protocols for in situ PCR, amplification within cells (e.g., on histological slices) have been published. The difficulty in this procedure is to stabilize the tissue structure on the slide by suitable fixation such that the structure remains intact through the thermal cycles. This is accomplished with special protocols that fix the paraffin-embedded tissue structures with a 10%, buffered formalin solution. In situ PCR leads to a large increase in the detected signals in histological slices. There are already thermocyclers on the market that directly heat the slides.
29.3.10 Other Approaches Besides the special PCR approaches described above, there are many other techniques that are often implemented to answer specific questions or are used for preparative purposes. Only a brief overview can be presented here:
Digital PCR: An alternative approach to quantitate starting copy DNA is to perform limiting
dilutions of the sample, followed by hundreds or even thousands of PCRs, such that each reaction contains at most one template copy. Poisson statistics are applied to the fraction of positive reactions to extrapolate the starting concentration without the need for a standard curve. Sophisticated microfabricated devices partition the reaction automatically either into droplets or chambers on a chip, coupled with sophisticated post-PCR detection formats. Such dilution of sample also enables the differentiation and detection of rare mutant sequences. RACE PCR: rapid amplification of DNA ends is a method to amplify and clone the 5´ ends of cDNAs, in particular those of long mRNAs that were not completely synthesized during in vitro transcription. Inverse PCR: An important method for the amplification of unknown DNA sequences. Primers binding in opposite directions to a known section of sequence are synthesized and used to amplify the DNA in both directions. The products are digested with a restriction enzyme and self-ligated to form circular DNA sequences, which can be amplified and sequenced with the starting primers. Vectorette PCR: this is used frequently for the characterization of unknown DNA segments.
Hybridization Methods, Section 28.1.3
769
770
Part IV: Nucleic Acid Analytics
Alu PCR: Alu elements are short repetitive elements that are more or less evenly distributed
cot Curves, Basic Principles of Hybridization, Section 28.1
throughout the genome of primates. By amplification with primers that bind to the Alu repeats, a characteristic pattern of bands is created that is so distinctive it can be regarded as a genetic fingerprint of an individual. cot Curves: DOP PCR: Degenerated optimized PCR (see also Section 29.3.3) is used for the analysis of micro-amplifications and deletions in comparative genome hybridization. In this method, the entire genetic material of a test cell (e.g., cancer cell) and a control cell are amplified and different detection markers are incorporated, such as rhodamine/digoxigenin- and fluorescein/ biotin-labeled, Chapter 28). After pre-hybridization of the chromosomes of the target cell with cot DNA, to saturate the repetitive sequences, the mix of digoxigenin and biotin labeled amplicons is hybridized with the chromosomes of the target cell. Detection is accomplished by use of the different filters of a fluorescence microscope for fluorescein and rhodamine. Overlapping the signals reveals a fluorescence pattern which marks those spots where the fluorescein or rhodamine signal predominates due to a micro-amplification or deletion. PRINS PCR: Primed in situ PCR is precursor of in situ PCR, in which a primer is extended once after in situ hybridization to the target DNA in a fixed target cell.
29.4 Contamination Problems The high analytical sensitivity of PCR represents, for obvious reasons, enormous progress for science and diagnostics. However, the ability to create many millions of molecules from just a few in a very brief span of time creates an enormous danger of false positive results, since each molecule is an optimal template for further amplification. In addition, the danger of contamination is particularly great when laboratories frequently work with the same primers and always amplify the same target with them. Since PCR will become increasingly common in routine laboratory use in coming years, we will examine the problem in more detail here. There are three basic types of contamination with DNA:
cross contamination from sample to sample during isolation of the DNA; contamination with cloned material that contains the amplification target sequence; cross contamination with already amplified DNA. In general the greatest danger is presented by aerosols. Table 29.2 demonstrates what sort of contamination is probable from aerosols, which are fine drops of liquid suspended in air. Beginning with the assumed size of such aerosols, the volume of such particles are given. For an amplification reaction with a volume of one hundred microliters, the table shows that already a picoliter contains 10 000 amplifiable molecules that could potentially cause contamination. Each of these amplicons is a perfect target for a new amplification.
29.4.1 Avoiding Contamination Avoiding contamination should have the highest priority in diagnostic as well in research laboratories, even before decontamination. To accomplish this, it is important to be clear about the possible sources of contamination: Table 29.2 Danger of contamination by aerosols. Type of contamination
Size
Volume
Amplicon per volume
100 μΙ
1012
∼1 μΙ
1010
∼100 μm
∼1 nΙ
107
∼10 μm
∼1 pl
104
∼1 μm
∼1 fl
10
Splashes
Aerosol
29 Polymerase Chain Reaction
aerosol formation by centrifugation, ventilation, uncontrolled opening of the sample and PCR tubes;
transfer by contaminated pipettes, disposables, reagents, gloves, clothes, hair, and so on; splashes while opening tubes or pipetting liquids. Numerous measures that can help to minimize the risk of contamination can be inferred from these points.
General measures to minimize the risk of contamination:
aliquot frequently used reagents and samples; only use autoclaved reagents, pipette tips, and reaction tubes; minimize manual steps to the greatest extent possible; avoid pipetting; avoid strong drafts; clean and decontaminate devices and pipettes from time to time with dilute bleach solutions; if possible, avoid the use of nested PCR, since pipetting the samples drastically increases the risk of contamination (Section 29.3.1).
Sample handling:
open tubes with cotton wool, a cloth, or something similar to avoid contamination of your gloves; if the tube has liquid in the lid, centrifuge briefly; work as closed as possible, which means only one tube should be open at a time; only use pipette filter tips or positive displacement pipettes; change gloves frequently; pipette slowly and with care; open tubes slowly and with care.
Waste disposal:
inactivate used (contaminated) pipette tips with HCl or bleach (sodium hypochlorite); do not dispose of remaining samples and amplicons in the regular trash, dispose of separately after inactivation;
close PCR tubes before disposal. Separation of work areas:
strict division into three areas:
– Area 1: preparation of the amplification mixes; for this purpose a laminar flow hood can be used; – Area 2: sample preparation; – Area 3: amplification and detection, must be in a separate room; under all circumstances, separate clothing must be worn in areas 1 or 2 and area 3; separate hardware (pipettors, tips, etc.) for each area; Sample flow should always be one way: Area 1 → Area 2 → Area 3 → autoclave or waste.
In general, samples and amplified material should be handled with the same amount of care as infectious or radioactive material. In addition, a suitable number of negative controls should be taken through each and every step (sample preparation, amplification, detection) in order to detect contamination early on.
29.4.2 Decontamination Decontamination includes two different types of measures:
Chemical or physical measures to clean equipment or the laboratory. This includes substances that destroy DNA directly or at least inactivate it such that it can no longer be amplified. Examples are HCl, sodium hypochlorite and peroxide.
771
772
Part IV: Nucleic Acid Analytics
Measures that are integrated into the routine operation of the tests and take place before or after every amplification. These can be further subdivided into physical, chemical, and enzymatic measures. Physical Measures UV irradiation: bombarding amplicons after PCR with UV light at a wavelength of 254 nm leads to the formation of pyrimidine dimers (T-T, C-T, C-C) within and between the DNA strands of the amplicon. Such inactivated DNA is no longer usable by Taq polymerase as a template. This measure has, however, a few disadvantages. There is a correlation between DNA length and the efficiency of the irradiation: The shorter the amplicon, the less effective the UV light is. In addition, the decontamination is less effective for GC-rich templates than for AT-rich templates. Chemical Measures Isopsoralens are intercalating dyes, which lead to crosslinking of the two strands when irradiated with long wavelength UV light (312–365 nm). This also blocks the polymerase activity. 3´ -Terminal ribonucleotides in primers create a base-sensitive position in the amplicon. A subsequent treatment with base hydrolyzes the primer binding sites. Enzymatic Measures
Restriction digestion, DNase I digestion, exonuclease III digestion, UNG system.
Digestion with uracil-N-glycosylase (UNG) is the most efficient method for the decontamination of previously amplified DNA. This measure relies on the incorporation of dUTP instead of dTTP during amplification. The resulting PCR product contains uracil bases in both strands and is therefore different than all starting DNAs to be amplified. UNG is an enzyme that cleaves the glycosidic bond between uracil and the sugar phosphate backbone of the DNA. Through subsequent heating or base treatment such abasic DNA hydrolyzes into small fragments and thus can no longer be amplified. The UNG system is particularly effective for two reasons: first, every newly amplified molecule contains uracil bases and, thus, is a substrate for UNG. Second, UNG decontaminates before a new amplification, when possible contaminations are smallest. Many other decontamination measures have the disadvantage that they either occur after the amplification and then need to be quantitatively effective or they require additional steps, which bring with it an additional risk of contamination. Since uracil bases are only substrates for UNG in single- or double-stranded DNA, but not as individual nucleotides or in RNA, the enzyme can be added directly to the amplification mixture and is suitable for use in RT-PCR.
29.5 Applications Many applications of PCR have already been mentioned in Sections 29.2 and 29.3. Most of these address particular questions in a research laboratory. In the following section, applications in medical diagnostic laboratories will be described in more detail and the possibilities of PCR in genomic analysis will be sketched out.
29.5.1 Detection of Infectious Diseases The detection of disease causing agents is an ideal application for PCR, since many bacteria and viruses can either not be cultivated at all or only very slowly and conventional tests are not nearly as sensitive as PCR (Figure 29.11). Consequently, such tests have entered into the routine of molecular laboratories in food quality control, as well as veterinary and human medicine. Examples are the viruses HCV (hepatitis C virus), HIV (human immunodeficiency virus), HBV (hepatitis B virus), and CMV (cytomegalovirus), as well as the bacteria Chlamydia, Mycobacterium, Neisseria, and Salmonella. By the detection of such pathogenic organisms with PCR, there are three critical aspects: A sufficient specificity of the reaction to avoid false positive results, a very high but also clinically
29 Polymerase Chain Reaction
773
Figure 29.11 Schematic of the detection of the trinucleotide (CAG-) expansion typical for Huntington’s disease. The amplification takes place with specific primers that flank the CAG-repeat. The size of the repeats in a polyacrylamide gel provides the diagnostic result (see Figure 29.12).
relevant sensitivity of the test, and a clear, verified result. The challenge of the specificity of the complete reaction results from the question of how specifically a primer pair needs to bind, in order to only amplify HIV but at the same time to recognize all the subtypes, for example. The sensitivity is decisively influenced by the method of sample collection and the volume of the sample. The former must guarantee an efficient separation from inhibitors, to enable an undisturbed reaction. This is particularly important for difficult sample material like sputum, stool, and urine. In addition, the amount of sample naturally has an effect on the sensitivity of the reaction. With ultrasensitive tests such as for the diagnosis of HIV it is often necessary to enrich the virus in the samples. This usually involves ultracentrifugation. In this way a sensitivity of around 20 genome equivalents per milliliter can be achieved. On the other hand, in considering sensitivity, it is always important to factor in the clinical relevance. For example, when a dose of more than 105 bacteria from the Salmonella group are necessary to trigger an acute gastroenteritis, there is no need for an ultrasensitive test. The target sequence also plays a decisive role in the accuracy of the test. HIV, like all retroviruses, replicates its genome via a DNA intermediate, the proviruses in the host genome. Only during an acute infection can replicating RNA be found in the blood of the host organism. Besides the relatively qualitative yes–no answer of PCR, for certain parameters, such as HCV and HIV, more and more quantitative tests are gaining prominence. These allow monitoring of the success of a therapy and thereby help to recognize early on the success of certain therapeutic measures on the course of disease.
29.5.2 Detection of Genetic Defects In the area of molecular medicine, PCR has provided the prerequisite to diagnose many genetic or acquired diseases on the level of DNA or RNA, prior to the appearance of symptoms. Many methods for this purpose have been developed and refined. In general this is a very new and innovative field that is subject to rapid changes. This section will only provide a general overview of current methods and highlight a few examples. The detection of known genetic defects can be classified based on the type of mutation, with the exception of translocations, into point mutations or length variations, such as insertions, deletions, and expansions. It is also important to differentiate between simple single site mutations or diseases that are caused by complex mutation patterns. For example, over 300 mutations are known for cystic fibrosis and familial hypercholesterolemia, while Huntington’s disease (see below) is caused by a single mutation. Each of these mutation types requires a different method. Length Variation Mutations Mutated and wild-type alleles of this mutation type can be differentiated on the basis of the length of the PCR product. A well-known example is the detection of trinucleotide expansions for a few neurodegenerative diseases. The causal mutation in Huntington’s disease (HD) is the expansion of a trinucleotide repeat (CAG) in the affected HD allele in the IT15 gene. The normal allele already varies in length, with up to 32 repeats being common. Research has shown, however, that a repeat length above 36 CAGs can be described as a positive result. The principle of the test is shown in Figure 29.12. Since the
Figure 29.12 Detection of the expansion of the Huntington gene locus. Shown is the trinucleotide expansion (n = 55) of an affected woman and her pre-symptomatic children. While her son’s expansion is also 55 CAG repeats, transmission of the allele to the first daughter reduced the repeat length (n = 51), while the second daughter experienced an expansion (n = 59). The healthy father is homozygous, with each allele 19 CAG repeats in length.
774
Part IV: Nucleic Acid Analytics
disease behaves in an autosomal dominant manner and homozygous carriers practically do not exist, a second, healthy gene is always found. After amplification of both alleles, the PCR products are separated on an electrophoretic gel and the corresponding repeat length is determined (Figure 29.12). Larger length variations, such as those found in Fragile X Syndrome, can also be detected in Southern blots after hybridization with specific probes.
Sequencing, Chapter 30
In vitro Restriction and Application, Section 27.1.4
Specificity of the Hybridization and Stringency; Section 28.1.2
Figure 29.13 Reverse dot blot: The detection of known mutations makes use of allele-specific hybridization probes (wt = wild type; mut = mutant), which are immobilized on a membrane. The position of hybridization is visualized by the label and allows genotyping.
Point Mutations Sequencing The surest way to identify and characterize known, as well as unknown, mutations is to sequence the PCR product (Section 29.3.5). Since this is technically demanding and laborintensive, it is not suitable for screening procedures. Restriction Fragment Length Polymorphisms RFLPs can be used for analysis whenever mutations have led to the creation or loss of a restriction site. After amplification, the PCR product is cleaved with the corresponding restriction enzyme and the fragments are separated by gel electrophoresis. This method is only suitable for the detection of known mutations. Reverse Dot Blot (Allele-Specific Hybridization) Reverse dot blots involve immobilizing allele specific probes to the surface of a membrane and hybridizing the PCR product to it. Only in the case of a perfect match between the probe and the amplicon is hybridization allowed. This means that mutated alleles are not bound by the wild-type probe and the wild type does not hybridize to the mutant-specific probe. The location of the hybridization is visualized with specific labels and reveals the gene type. The principle is shown in Figure 29.13. To avoid unspecific binding, the stringency of the hybridization needs to be precisely adjusted (salt concentration, time, temperature). Exact knowledge of the mutations is also required to apply this method.
29 Polymerase Chain Reaction
775
Figure 29.14 OLA (Oligonucleotide Ligation Assay) technique: After a PCR, allelespecific oligonucleotides (wt, mut) bind to the single-stranded amplicon. Only when the 3´ ends hybridize perfectly can the oligonucleotide be ligated to another universal, labeled oligonucleotide. Since the mut oligonucleotide and the wt oligonucleotide differ in length, the ligated oligonucleotides can be separated electrophoretically and detected by their label.
If the 3´ end of a primer cannot bind to a template, the amplification is inhibited, since Taq polymerase only extends a hybridized 3´ -OH end efficiently. Allele-specific PCR takes advantage of this fact and uses two different forward or reverse primers. The amplification of the DNA to be investigated takes place in two separate PCR tubes with one of the two sets of primers in each. This allows the elegant characterization of the genotype (homologous wild type, heterozygous, homologous mutants). The exact mutations must be known for design of the primers. Allele-Specific PCR
OLA Technique The oligonucleotide-ligation assay (OLA) also takes advantage of the fact that only perfectly hybridized, neighboring oligonucleotides can be ligated together. To analyze a known mutation, oligonucleotides are created that differ in length or labeling, and bind to the PCR product in an allele-specific manner. Depending on the presence of the mutation, the ligase connects one or the other allele-specific oligonucleotides with the universal oligonucleotide. This provides the information for the genotyping (Figure 29.14). Single-Strand Conformational Polymorphism (SSCP) Single-stranded DNA (ssDNA) forms unpredictable intramolecular secondary structure under renaturing conditions, which is strictly dependent on the primary sequence. Since a mutated allele carries a different sequence, it will also adopt a different conformation upon renaturation. In single-strand conformational polymorphism (SSCP) analysis the PCR product is denatured after amplification and immediately loaded onto a renaturing gel. Even the smallest changes in conformation of the single strands leads to a different mobility in the gel, which is observed as a band shift (Figure 29.15). With this method, unknown
Gel Electrophoresis of DNA, Section 27.2.1
776
Part IV: Nucleic Acid Analytics
Figure 29.15 SSCP analysis: After amplification the PCR products are denatured and loaded onto a renaturing sequencing gel. Owing to the different refolding, the mobility of the mutated allele (mut) is different from that of the control DNA (wt). In general four bands appear, since the single strand of each allele refolds differently based on its sequence.
mutations or polymorphisms can be recognized, but not characterized. That requires subsequent sequencing. Denaturing Gradient Gel Electrophoresis (DGGE) This approach is based on a very similar principle to that of SSCP. The double-stranded amplicon is loaded onto a gel that contains a gradient that is more and more denaturing. Depending on the sequence of the allele, denaturation of the mutated spot takes place before or after wild type. The altered mobility is also observed here as a band shift.
29.5.3 The Human Genome Project
Physical and Genetic Mapping of Genomes, Chapter 36 Generation of a Physical Map, Section 36.2.3
Identification and Isolation of Genes, Section 36.2.4
In October 2004 the sequence of the human genome was published in the journal Nature: The result of 13 years of work involving more than 2800 scientists. An analysis of the data and the 2.85 billion base pairs revealed the presence of 20 000 to 25 000 genes. The standard of quality applied required 99% of the gene-containing sequences to be included and the accuracy was given as 99.999%. Sequence identification was only the first step, now understanding the function of the genes is the focus. PCR also strongly drove the development of the Human Genome Project. PCR allowed the introduction and use of sequence tagged sites (STSs), which were a tremendous aid to the mapping work. STSs are specific DNA segments on chromosomes, which are defined by the sequence of two corresponding primers. Such information is readily available through databanks and directly available for use by every researchers involved in mapping the human genome or cloning genes. If an STS is part of an expressed sequence, it is referred to as an expressed sequence-tagged (EST) site. A particular form of STS are short tandem repeat polymorphisms (STRPs), short dinucleotide repeats, usually CA, which can vary in length from individual to individual. STRPs allow the determination of recombination frequencies and therefore conclusions about the separation of such markers. They also make the characterization of haplotypes possible and the isolation of genes by a positional cloning approach. The first gene that was cloned using a positional cloning approach was the CFTR gene, which is responsible for cystic fibrosis, in 1989. The modern DNA/RNA sequencing methods (next generation-, deep-, sequencing) would not have been possible without the developments in the field of PCR. The parallelization and miniaturization of PCR were decisive. The use of ever smaller volumes (down to picoliters), allows use of less sample and fewer reagents. Speed and precision also increase. The current focus of development is the commercialization of microfluidics, in which preparative amounts
29 Polymerase Chain Reaction
of nucleic acids can be created for sequencing purposes. By suitable choice of primers, tags and barcodes can be incorporated into PCR amplification. They can then, depending on the sequencing protocol, be further amplified by, for example, emulsions PCR or bridge PCR.
29.6 Alternative Amplification Procedures Besides PCR, other amplification procedures exist for the multiplication of nucleic acids. Even if such procedures have found a use in individual laboratories in the past, it is becoming increasingly clear that a broad, routine application can only be accomplished through the use of PCR. Therefore, only a few of the most important alternatives will be discussed in this book.
29.6.1 Nucleic Acid Sequence-Based Amplification (NASBA) This procedure involves an isothermal amplification of nucleic acids at 42 °C. Enzymatic components of the NASBA reaction are reverse transcriptase, RNase H, and T7 RNA polymerase. Another special aspect is the incorporation of a T7 promoter by means of a primer (primer A). This occurs when the promoter is attached to the specific nucleotide sequence of the primer. The reaction is started with the binding of Primer A to the RNA template. Reverse transcriptase (RTase) converts the template into cDNA and the RNA of the hybrid is digested by the RNase H. Primer B then binds to the opposite strand and the DNA-dependent, DNA polymerase activity of the RTase completes synthesis of the strand. This leads to the replication of the promoter. T7 polymerase, as the third enzymatic component of the mixture, binds to the promoter and synthesizes around 100 new RNA molecules (dependent on the length of the amplicon), and the process repeats from the beginning (Figure 29.16). Self-sustained sequence replication (3SR) is based on the same principle. NASBA and 3SR can be slightly modified and DNA templates transcribed into RNA prior to beginning the reaction to allow use of DNA templates. Transcription-Mediated Amplification (TMA) This procedure is an alternative isothermal amplification of RNA at 42 °C. Enzymatic components of the TMA reaction are a reverse transcriptase and T7 RNA polymerase. RNase H is replaced by the partial RNase H activity of the reverse transcriptase in this reaction. This amplification procedure also involves the incorporation of a T7 promoter attached to a primer (Primer A) into the amplicon.
29.6.2 Strand Displacement Amplification (SDA) This method of nucleic acid replication also involves an isothermal reaction. It is based on the ability of DNA polymerases to begin new synthesis at single-strand breaks and displace the old strand in the process. Amplification involves a cyclical cleavage of single strands and the subsequent strand replacement. Since restriction enzymes normally completely cleave double-stranded DNA, the creation of the single strand breaks, called nicks, is done by incorporating a nucleotide analog into the opposite strand. After cleavage the single stranded (inactive) remaining sequence of a restriction site is made double-stranded by extending a primer. New synthesis does not employ the four natural dNTPs, which would lead to a double-stranded reaction site subject to complete cleavage, but occurs instead in the presence of three of the natural dNTPs and a thio-deoxy nucleotide. This results in the formation of a double-stranded hybrid between the normal and the newly synthesized sulfurcontaining strand. This site cannot be completely cleaved by the restriction enzyme, which leads to the desired single strand nicks. The principle is illustrated in Figure 29.17.
29.6.3 Helicase-Dependent Amplification (HDA) As an alternative to strand displacement, DNA helicase is employed and the resulting single stranded DNA is protected from reassociation with single-strand DNA binding proteins. In the next step two primers are used, as for PCR, and DNA polymerase generates the two daughter strands. These strands then become available for the helicase and the next round of amplification is begun. In a sense, it is just like PCR at a constant temperature. One advantage is that it
777
778
Part IV: Nucleic Acid Analytics
Figure 29.16 Nucleic acid sequence-based amplification (NASBA): The starting point of the amplification is a single-stranded RNA that binds to Primer A. This primer contains a T7 promoter sequence on its 5´ end. A cDNA is synthesized by reverse transcriptase and the RNA in the hybrid is digested immediately by RNase H. Primer B now binds to the single-stranded DNA and synthesizes the opposite strand, in so doing creating a functional T7 promoter. The T7 RNA polymerase recognizes its promoter and synthesizes around 100 RNA molecules, dependent on amplicon length, which then trigger the cyclical phase of the NASBA reaction, in which the described steps are repeated. The complete reaction takes place at a constant temperature and in a single amplification buffer.
29 Polymerase Chain Reaction
779
Figure 29.17 Strand displacement amplification (SDA): This is a cyclical process consisting of synthesis, restriction digestion, and strand displacement. The primers contain the recognition sites for the restriction enzyme. Since new synthesis is carried out in the presence of a thio nucleotide, nicks result, since such thiobonds are resistant to restriction enzymes. SDA, like NASBA, runs under isothermal conditions. Source: according to Persing, D.H. et al. (1993) Diagnostic Molecular Microbiology: Principles and Applications, American Society for Microbiology, Washington D.C.
does not require a thermocycler. A disadvantage is, however, that the choice of primers and optimization of reaction conditions involves a more complicated search.
29.6.4 Ligase Chain Reaction (LCR) The ligase chain reaction (LCR) does not lead to increasing the amount of the actual target sequence, instead it leads to an amplification of two oligonucleotides ligated together, which are complementary to the original strands (Figure 29.18). After the initial hybridization of two immediately neighboring oligonucleotides, they are linked together by a thermostable ligase. These form the target for two complementary oligonucleotide pairs, which also hybridize and are linked by the ligase. In about 30 cycles LCR reaches a similar sensitivity to that of PCR. To increase the amplification specificity, LCR protocols have been developed in which the two inner 5´ ends of the oligonucleotides are selectively phosphorylated, to avoid unspecific ligation.
780
Part IV: Nucleic Acid Analytics
Figure 29.18 Ligase chain reaction (LCR): Theoretically each round of the cycle, consisting of denaturation, annealing of four oligonucleotides, and ligation, leads to the doubling of the number of the oligonucleotides linked together. Analogous to PCR, LCR also leads to exponential amplification.
Repair Chain Reaction (RCR)
The repair chain reaction is related to the ligase chain reaction. In contrast to the ligase chain reaction, the two complementary oligonucleotide pairs do not abut, but instead are separated by a gap of one or more nucleotides (Figure 29.19). The gap is selected such that the addition of dGTP and dCTP or dATP and dTTP, a polymerase and the ligase repairs the missing nucleotides of the gap in a double strand-dependent reaction. This combined, limited, elongation and ligation increases the specificity of the amplification, since the two oligonucleotide pairs cannot be ligated together without gap filling taking place.
29.6.5 Qβ Amplification Qβ amplification does not involve the elongation of a primer (like PCR or NASBA/TMA), instead new synthesis is triggered by the structure of the Qβ genome. Copying the Qβ structures leads to (–) and (+) copies in each reaction cycle and therefore leads to exponential amplification. The proliferation takes place isothermally, after the triggering Qβ are coupled to a parameter-specific probe, which hybridizes with the target sequence (Figure 29.20). A disadvantage of this method is that it is a signal amplification, which can also lead to the amplification of falsely hybridizing probes, causing false positive results.
29 Polymerase Chain Reaction
781
Figure 29.19 Repair chain reaction (RCR): In contrast to LCR, RCR does not lead to the annealing of the two primers next to one another, instead it leaves a gap of several nucleotides. Ligation only becomes possible after a polymerase has filled in the gap in the presence of dGTP and dCTP or dATP and dTTP, which increases the specificity of the reaction. The gap is chosen such that only dG and dC or dT and dA are contained in the opposite strand of the target DNA in the gap.
Figure 29.20 Qβ amplification: Qβ amplification involves a structure-initiated replication of the Qβ indicator sequence by the enzyme Qβ replicase at an isothermal temperature. The amplification is exponential and leads to a rapid increase in the number of amplification products through (–) and (+) strand intermediates. In contrast to PCR or other target amplifications, Qβ amplification leads to the proliferation of the Qβ indicator sequence; this is a signal amplification of a coupled, target-independent indicator sequence, not the target sequence itself.
782
Part IV: Nucleic Acid Analytics
29.6.6 Branched DNA Amplification (bDNA) Signal Amplification, Section 28.5.3
The branched DNA amplification (bDNA) method is a more recent means by which to amplify signals. The target nucleic acid to be detected is coupled to a solid surface with specific capture probes. Other nucleotides, extenders, then hybridize to the nucleic acids. The extenders bind target-independent amplifier molecules, which then stick out from the original target like antenna. These amplifiers bind many more oligonucleotides, which carry an alkaline phosphatase label. After several washes, the phosphatase substrate is added and chemiluminescence is used to detect the nucleic acids. The bDNA method allows the detection of around 105 target molecules. A disadvantage of this method, however, is that it involves a signal amplification, which likewise amplifies the signal resulting from a falsely hybridizing probe and therefore can lead to false positive results.
29.7 Prospects PCR has become a central bioanalytical method of molecular research laboratories. If current trends continue, it will become increasingly important in routine diagnostics. There are many reasons why PCR is one of the cornerstones of modern molecular biology. Increased automation, not only of PCR, but also of the critical preceding sample preparation, plays an important role. The enzymatic contamination control systems have been particularly important for routine diagnostic use. In addition, closed tube detection formats, such as TaqMan, allow routine use with fully automated PCR analysis devices with sensitivity in the range of a few copies. RT-PCR has allowed application to retroviral diagnostics (e.g., for HIV and HCV) and expression profiling. Another trend is shortening reaction times with faster thermocyclers and minimizing the reaction volume, as has already been realized in the LightCycler® . One also hopes to reduce amplification times down to minutes by miniaturizing the PCR reaction vessel in the form of chips (lab on a chip). The continuous development of microfluidics technology is promising for future routine analytics. Digital PCR offers the possibility of precise absolute quantitation, rare mutation detection, and quantitation of small differences in copy number. It can be expected that further automation of the entire workflow together with miniaturization will enable different and more challenging sample types, such as whole blood, sputum, urine, and spinal fluid to be amenable to PCR. Future applications that will take advantage of further automation and miniaturization of both sample preparation and PCR are on the horizon. In particular, point of care PCR devices that allow minimally trained operators to perform PCR from sample to answer in what is currently relegated to sophisticated laboratories with highly trained technicians is particularly exciting. Such rapid, simple RT-PCR would facilitate the spread of cost-effective DNA diagnostic methods for diseases such as Ebola and HIV in countries with limited financial resources and the detection of foodborne contaminants and fraudulently labeled foodstuffs on-site.
Further Reading Dieffenbach, C.W. and Dveksler, G.S. (2003) PCR Primer. A Laboratory Manual, 2nd edn, Cold Spring Harbor Laboratory Press, New York. Logan, J., Edwards, K., and Saunders, N. (eds) (2009) Real-Time PCR: Current Technology and Applications, Caister Academic Press. International Human Genome Sequencing Consortium (2004) Finishing the euchromatic sequence of the human genome. Nature, 431, 931–945. Kessler, C. (ed.) (2000) Non-radioactive Analysis of Biomolecules, Springer, Berlin, Heidelberg. Larrick, J.W. and Siebert, E.D. (1995) Reverse Transcriptase PCR, Ellis Horwood, London. Yang, S. and Rothman, R.E. (2004) PCR-based diagnostics for infectious diseases: uses, limitations, and future applications in acute-care settings. Lancet Infect. Dis., 4 (6), 337–348. Rabinow, P. (1996) Making PCR: A Story of Biotechnology, University of Chicago Press. Espy, M.J., Uhl, J.R., Sloan, L.M., et al. (2006) Real-time PCR in clinical microbiology: applications for routine laboratory testing. Washington Clin. Microbiol. Rev., 19, 165–256.
29 Polymerase Chain Reaction Millar, B.C., Xu, J., and Moore, J.E. (2007) Molecular diagnostics of medically important bacterial infections. Curr. Issues Mol. Biol., 9, 21–40. Reischl, U., Wittwer, C., and Cockerill, F. (2002) Rapid Cycle Real-Time PCR. Methods and Applications, Springer, Berlin, Heidelberg. Saunders, N.A. and Lee, M.A. (eds) (2013) Real-Time PCR: Advanced Technologies and Applications, Caister Academic Press. Hugget, J.F., Foy, C.A., et al. (2013) The digital MIQE guidelines: minimum information for publication of quantitative digital PCR experiments. Clin. Chem., 59 (6), 892–902. Niemz, A., Ferguson, T.M., and Boyle, D.S. (2011) Point-of-care nucleic acid testing for infectious diseases. Trends Biotechnol., 29, 240–250. UNITAID (2014) HIV/AIDS Diagnostic Landscape, 4th edn, World Heath Organization. http://www .unitaid.eu/images/marketdynamics/publications/UNITAID-HIV_Diagnostic_Landscape-4th_edition. pdf.
783
DNA Sequencing Jürgen Zimmermann and Jonathon Blake EMBL Heidelberg, Meyerhofstraße 1, 69117 Heidelberg, Germany
In 1975, Fred Sanger laid the foundation for the most powerful tool for the analysis of the primary structure of DNA, with the development of an enzymatic sequencing method. At that time, neither the far-reaching implications for the understanding of genes or whole genomes nor the rapid development of this method could have been foreseen. Back then, Fred Sanger was happy about the sequencing of five bases in one week, as he himself noted in retrospect at a reception in the Sanger Center (Cambridge, England) in 1993. In comparison to these five bases, genome sizes reach astronomical dimensions. The average length of a small viral genome is in the range of 105 base pairs (bp). With the increasing complexity of organisms, further magnitudes are quickly exceeded: Escherichia coli already reaches 4.7 × 106 bp, Saccharomyces cerevisiae 1.4 × 107 bp, Drosophila melanogaster 1.8 × 108 bp, and humans 3.2 × 109 bp. Genomes of plants and even lower organisms can achieve even greater lengths: wheat (Triticum aestivum) 1.6 × 1013 bp and Amoeba dubia 1.2 × 1014 bp. The actual number of bases to be sequenced can easily reach hundred times the genome size depending on the strategy used. In this consideration, the needs of diagnostic DNA sequencing, which are increasingly gaining in importance, are not yet included. In addition, even though DNA sequencing analyzes only small fragments, it processes these in large numbers. At the same time as the development of the Sanger method, cloning methods became available in M13 phages, which allowed both the biological amplification of DNA fragments in a size range of up to two kilobase pairs as well as the generation of “easily” sequenceable singlestranded DNA. A maximum read length of 200 bp could be achieved. Therefore, only a fraction of the entire sequence could be determined during a sequencing run. This disparity forced the development of sequencing strategies that – at a reasonable expense – enabled the reconstitution of the whole sequence. Equipped with the tools of the Sanger method, it began with the analysis of whole genomes in the 1970s. In 1977, Sanger and his coworkers published the 5386 bp DNA sequence of the phage phiX174. In 1982, the complete sequence of the human mitochondrion with a length of 16 569 bp was determined. By 1984, they had already achieved, with the sequence of the Epstein–Barr virus, a length of 172 282 bp. Only 25 years after the breakthrough of Sanger, in 2001, the first sequence of the human genome was published. The cost of the Human Genome Project (HGP) exceeded the limit of one billion US dollars by far. The project made one thing clear: a cost-effective and rapid sequencing of larger genomes would have hardly been possible with existing technologies, and a lot of questions remained unanswered, such as the dynamics of genomes, even if the data constituted a milestone and the results made many procedures possible, such as microarray analysis. It would take many years for massively parallel sequencing (MPS) approximation methods, also known as next generation sequencing (NGS), to become available. The gross output of 300 Gb and more per experiment and the associated – and substantial – cost reduction created the chance for new procedures to allow for new objectives covering all aspects of genomes, which were previously not possible (Table 30.1). Bioanalytics: Analytical Methods and Concepts in Biochemistry and Molecular Biology, First Edition. Edited by Friedrich Lottspeich and Joachim Engels. 2018 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2018 by Wiley-VCH Verlag GmbH & Co. KGaA.
30
786
Part IV: Nucleic Acid Analytics Table 30.1 Application areas in the field next generation sequencing (NGS). Structural genomics
De novo sequencing Resequencing
SNPs, structural variants, exome, tumor/ normal tissue, personalized medicine, GWAS
Metagenomics Pharmacogenomics
Functional genomics
Transcriptome sequencing sRNA
miRNA
Protein–DNA interactions
ChIP
Epigenetic modifications
DNA methylation
The dynamics of the new sequencing systems is found in the arbitrarily increasing coverage/ depth, with which the areas are analyzed. In a microarray experiment, this is limited and it is difficult to find rare transcripts in a sample. With a sequencer, it is in principle easier to detect rare transcripts by manipulating the variables mentioned above. Most of these methods are still based on the Sanger method in their basic approaches, and despite their substantial improvement in throughput previous sequencing processes are still used in their own right, for example, for sequencing individual genes to validate expression constructs and PCR products. In addition, it should not go unnoticed that the new methods have their own systematic errors: enzymatic amplification is necessary in most cases to obtain enough products that can be detected in an instrument later on. These include preferences in the amplification techniques used, a guarantee that the generation of sequencing libraries did not occur fully randomized (e.g., negative GC-selectivity in the generation of sequencing libraries) or even preferences for sequencing enzymes that distort the results. This may result in a non-representative occurrence of all sequence regions; hence a complete picture is questionable. Even in methods that are not based on amplification the instruments themselves suffer from selectivity. While the MPS method is reaching an increasing methodological maturity, the first single molecule sequencing devices are already available and other methods are in development. In 1996, more than 300 mega-base (Mbp) of sequence information was newly recorded in the EMBL Nucleotide Sequence Database. This is nearly as much as had been registered in the 13 years since the founding of the institution. In June 2005, 95 giga-bases (Gb) in 54 × 106 entries could be published. This is the same as the daily production of an MPS machine. As of March 2011, there are 206 × 106 entries and 319 × 106 bp. These figures illustrate both the technological advances of the processes used as well as the increasing use of these techniques. As great as the numbers may seem, they represent sequences from different organisms, different versions, and also sequence fragments of small size, of which the location is not known in all cases. The path to a final single and complete sequence as the representation of a genome is not to be underestimated and requires a considerable effort. Recently the first projects for “Platinum genomes” have been started.
30.1 Gel-Supported DNA Sequencing Methods As already mentioned, gel-supported methods are still broadly used in DNA sequencing of smaller portions and diverse objectives such as the screening of expression constructs, PCR products, or structures that do not parallelize for use on a MPS (massive parallel sequencing) device, have overly low length, or cannot be resolved on the new sequencers at all until now. The original shotgun method was based on the statistical reduction of the genome into small fragments of 1–2 kb in length, their cloning, and sequencing as well as the assembly of the individual sequences like a puzzle. The overall picture of the sequence thus arises only at the end of a project. This uncertainty during the project was addressed by developing orderly and
30 DNA Sequencing
purposeful methods. Primer walking, nested deletions, Delta restrictions cloning, chromosome walking or combinations of methods allow, even during sequencing, a localization of the obtained information. Hybrid procedures combine the high initial data rate of accumulation of a random strategy with the reduction of the sequencing effort of a directional strategy. The starting point for genome-wide DNA sequencing is a correspondingly precise physical map that acts as a rough guide. Positions and relations of individual clones are reviewed with fingerprinting methods at a large scale and with fine-mapping they are reviewed in detail. The entire process of cloning and sequencing is based on statistical events that lead to an unpredictable sequence representation of individual results. Accordingly, gaps between sequence contigs are to be expected. These may have their origin both in sequence gaps (e.g., a too short read length) as well as in physical gaps (section missing from the corresponding clone library). Sequence gaps are usually closed by primer walking on the cloning, in which the neighboring but not associated contigs come to rest. Physical gaps can only be closed with the aid of a second clone library, one that was created using a different cloning vector, size selection, and fragment generation. Using an independent library is usually necessary to bypass possible instability of target sequences in a vector/host system. For identification of the missing sequences, the second library is sampled with oligonucleotides (PCR) whose sequences correspond to the ends of the previously associated contigs. A PCR product can arise only where the two end sequences come to rest. The corresponding clone thus contains the missing sequence to connect the two contigs in question. Sequence repeats (repeats, inverted repeats), which exceed the reading length of a read, can also make it difficult to reconstruct the original sequence. Only the use of additional map information of different resolutions as well as the examination of 5´ and 3´ border sequences can help in such cases.
Primer walking Primer walking is a directed DNA sequencing strategy. In a first step, the DNA fragment that is to be sequenced will be sequenced from both ends in one reaction. Starting from the obtained information, a new primer is placed in the same reading direction at each end. In this way a prolonged sequence section can be determined in each step. In addition, primers are placed in the opposite reading direction to determine the sequence of the complementary DNA strand. At each point of a primer walking project, the position and the running direction of the sequence is uniquely determined. The redundancy of the obtained sequence reaches a value close to two in the optimal case. The sequence information, however, can only be generated serially and in steps and is dependent on the synthesis of a new primer for each reaction.
Nested deletions The generation of unidirectional nested deletions was developed by Steven Henikoff in 1984. This sequencing strategy allows an ordered sequence extraction using merely a standard primer. The starting material uses plasmids with a known priming sequence, which must have a certain structure: Between the priming point and the cloned insert, there must be singular recognition sites for two different restriction endonucleases whose sequence is not reflected in the insert itself. The cutting site, which is closer to the direction of the priming point, must generate a 3´ overhang. The second, closer to the direction of the insert, must generate a 5´ overhang. After a double digestion with both restriction endonucleases, the linearized DNA molecule is treated with exonuclease III (Exo III). Exo III attacks the 5´ overhanging end of a DNA double strand, where it acts as 3´ -5´ exonuclease and successively removes the 3´ recessive DNA. At specific time points, an aliquot is removed and processed parallel in the next steps. In the next step, the 5´ overhang is removed by S1 nuclease to produce a blunt end. After reparation of ends, recircularization and transformation plasmids occur, whose inserts are shortened to a certain number of bases. The sequence shortening takes place right next to the binding site for the sequence primer. Thus, after the primer, a section of unknown sequence begins. Fragments with a maximum length of about 3 kb can be processed. A second series of deletions must be generated to determine the sequence of the complementary DNA strand.
Delta restrictions cloning Delta restrictions cloning allows for the generation of numerous sub-clones – whose position to each other is known – through simple digestion with polylinker restriction endonucleases. In the first step, the
787
Contig A continuous (contiguous) DNA sequence that is generated during the mathematical assembly from overlapping DNA fragments in the computer.
788
Part IV: Nucleic Acid Analytics
Cosmids Circular DNA molecules such as plasmids. The name cosmid refers to the DNA sequences that are referred to as cos and are derived from the phage lambda. These sections make it possible to add larger genes to cosmids than to plasmids. BAC Bacterial artificial chromosome (BAC) comes from the single-copy F-plasmid of the bacterium Escherichia coli and permits stable cloning of longer inserts of more than 300 kbp in bacteria.
Possibilities of PCR, Section 29.1
fragment that needs to be sequenced is cut and analyzed on an agarose gel. For all clones, which have an internal restriction site for a polylinker enzyme, a piece is removed by this digestion. Simple recircularization of the fragments results in clones that have a deletion relative to the position of the primer and thus deliver de novo sequences during the reaction with standard primers.
Transposon-mediated DNA sequencing This method enables the introduction of primer binding sites in a simple enzymatic step, which in turn allows the sequencing of longer DNA sections without subcloning or primer walking. As means of support transposons are used, which provide both the primer binding and encode a selectable marker (e.g., kanamycin, tetracycline) and allow for bidirectional sequencing. The insertion is carried out by simple incubation with the corresponding transposase and the DNA target mixture. After transformation into Escherichia coli, the selected clones can be selected and sequenced. In principle, the systems have no sequence-specific preference. However, in some cases accumulation in the area of certain hotspots can occur.
While the sequencing of whole genomes or genes was carried out with classical sub-cloning strategies as a shotgun with a framework of directional clones until the turn of the century, in phylogenetic analyses and in clinical diagnostics direct template production through PCR has prevailed. The use of PCR products as sequencing templates only requires a simplified purification (e.g., silica material or magnetic particles) prior to the actual DNA sequencing reaction. Parallel to the development of sequencing strategies was the development of sequencing techniques. While the first sequencing reactions were still performed with radioactive markers, current techniques make use of fluorescent markers. In addition, the increased understanding of DNA polymerases led to the use of other and new enzymes. After initial work with DNA polymerase I and the Klenow fragment and the use of genetically modified T7 DNA polymerases, thermostable DNA polymerases or even mixtures are now being used. Meanwhile, read lengths of 1000 bp and more can be achieved. The classic retrieval of RNA/DNA sequences occurs primarily in six steps: 1. Isolation and purification of nucleic acid: Genomic DNA is extracted with an adapted method for the target organism(s) and/or the target region. The direct sequencing of RNA is used rarely nowadays. As a rule, a cDNA copy is generated and subsequently sequenced through reverse transcription. 2. Cloning or PCR amplification: The DNA obtained in the first step is in several respects not suitable for DNA sequencing: The length of the genomic DNA is too large to be able to process with a conventional procedure. At present, a sequence of ∼1 kb can be produced in one sequencing run but only in favorable cases. During the analyses of, for example, a human gene of 50 kb, one only gets a fraction of the total information. For this fragment, it is still necessary to arrange its position and reconstruct it to the original sequence with the other generated fragments. For all types of DNA isolates, the number of copies contained in a preparation is not sufficient for sequencing. The automated DNA sequencing systems that are currently available have a detection limit of about 10 000 molecules, which cannot be achieved in a simple DNA isolate. To obtain a sufficient output amount for the DNA sequencing, it is subject to amplification. For the reconstitution of longer DNA sequences (>2 kb) this process must be conducted in vivo in cloning systems. The existing limitations of this process can be avoided with appropriate sequencing strategies. For shorter sequence segments that are frequently analyzed in medical diagnostics, PCR reactions are sufficient. 3. DNA purification for the sequencing reaction: To obtain optimal DNA sequencing results, a further purification is required. Contaminating proteins, carbohydrates, and salts can influence the set environment in an uncontrolled manner and can lead to dramatically reduced read lengths. For a more detailed description of the procedures, please refer to the relevant chapters of this book. 4. DNA sequencing and electrophoresis: The reaction products of the sequencing reaction of a DNA fragment are fractionated by gel electrophoresis, the generated band patterns are recorded online or offline and subjected to the following analysis. 5. Reconstitution of the original sequence information: As mentioned above, the sequence generated in a sequencing reaction is in most cases smaller than the entire sequence that must
30 DNA Sequencing
789
be determined. From many fragments generated by sequence reactions, the original image must be restored in the form of a genetic puzzle. For this, computerized and automated methods are used. 6. Error correction and sequence data analysis: The obtained sequence information is subjected to quality control. To suppress possible errors of sequencing from individual experiments, the DNA sequences that are to be determined become multiply redundant and, in addition, both complementary strands are sequenced. Subsequently, the sequence is examined for possible contamination by vector, bacterial, or foreign DNA in a first step and, where appropriate, removed from the sequence. With the aid of codon usage tables, potential ORFs (open reading frames), other reference sequences as well as various other tools, possible sequencing errors can be detected. These steps ensure the correction during the sequencing and the subsequent assembly of errors that have arisen. The final generated sequence can be compared to the DNA sequences of already known databases and can undergo a detailed sequence analysis. The overall process, as outlined in the list above, requires the combined use of a multiplicity of methods that have been described in detail elsewhere in this book. The following sections of this chapter are limited to the actual processes of gel-supported DNA sequencing methods, as well as their labeling and detection methods. Gel-supported DNA sequencing methods are mainly based on the production of base-specific terminated DNA populations that are separated according to their size in a subsequent, denaturing polyacrylamide gel/linear acrylamide gel electrophoresis. It is therefore basically an endpoint analysis. Up to 96 samples can be processed parallel and read lengths of up to 1000 bp per sample can be achieved. These populations can be generated in two different ways: The reaction products can be prepared by synthesis of a DNA strand (dideoxy sequencing, Sanger method) or by base-specific fission (chemical fission, Maxam–Gilbert method). Gel electrophoresis is performed in capillaries filled with linearly polymerized acrylamide gels.
30.1.1 Sequencing according to Sanger: The Dideoxy Method The dideoxy method (also known as the chain termination method, terminator method) is based on the enzyme-catalyzed synthesis of a population of base-specific terminated DNA fragments that can be separated by gel electrophoresis according to their size. From the resulting band patterns of a denaturing polyacrylamide gel, the sequence can be reconstructed. The basic principle will be described in the following section. Starting from a known start sequence, the synthesis of a complementary DNA strand is initiated by adding a sequencing primer (a short DNA oligonucleotide of about 20 bp), a nucleotide mix, and a DNA polymerase. To detect the reaction products, they must be labeled with either radioactive isotopes or fluorescent reporter groups (Section 30.1.2). On one hand, the use of a primer is necessary to obtain a defined starting point for sequencing, and on the other hand it is necessary for the formation of the initiation complex and thus to initiate the start of synthesis of the DNA polymerase. The reaction is started simultaneously in four parallel aliquots, which differ only by the use of different nucleotide mixes. Reactions labeled A, C, G, and T contain a mixture of the naturally occurring 2´ deoxynucleotides and in each case only one type of synthetic 2´ ,3´ -dideoxynucleotides, the so-called terminator (Figure 30.1). During strand synthesis through stepwise condensation of nucleoside triphosphates in the 5´ 3´ direction, two different reaction events can occur: During the condensation of the 5´ triphosphate group of 2´ -deoxynucleotides (dNTP) with the free 3´ -hydroxyl end of the DNA strand, an elongated DNA molecule is produced with the release of inorganic diphosphate (pyrophosphate), which in turn has a free 3´ -hydroxyl group. The synthesis can therefore be continued in the next step (Figure 30.2a). If, however, a condensation reaction occurs between the free 3´ -hydroxyl end of a DNA strand and the 5´ -triphosphate group of a 2´ ,3´ -dideoxynucleotide, an extension of this strand is no longer possible, because a free 3´ -hydroxyl group is no longer available. The strand is terminated (Figure 30.2b). The characteristics of the DNA polymerase used and the structure of dideoxynucleotide determine the mixing ratio, leading to the production of a population of base-specific terminated
Figure 30.1 Structural comparison of a 2´ ,3´ -di-deoxynucleotide and a 2´ -deoxynucleotide. The illustrated 2´ ,3´ -di-deoxyCTP nucleotide analogue differs from its naturally occurring analogue 2´ -deoxyCTP by the lack of the hydroxy group at C3´ of the sugar.
790
Part IV: Nucleic Acid Analytics
Figure 30.2 Synthesis of a DNA strand with incorporation of a 2´ -deoxynucleotide (a) and a 2´ ,3´ -dideoxynucleotide (b). In case of the latter, further polymerization is no longer possible.
Gel Electrophoresis of DNA, Section 27.2.1
Figure 30.3 Principle of the chain termination method according to Sanger. In a primed DNA synthesis reaction catalyzed by a polymerase, base-specific terminated DNA fragments of different lengths are synthesized. These fragments produce a particular band pattern in the gel electrophoresis that is used for reformation of the base sequence.
reaction products that in each case differ only in one base. At a molar ratio of dNTP : ddNTP of 200 : 1 termination events are relatively rare in the case of catalyzed T7 DNA polymerase reactions. Long reaction products of up to 1000 bp in length are then created. As mentioned above, the reaction is carried out in four aliquots. Each of these partial reactions contains only one type of terminator (ddATP, ddCTP, or ddGTP, ddTTP), which statistically replaces each one of the naturally occurring nucleotides in the synthesized DNA chain. In every reaction vessel products are generated that only end on one base type, such as A. The reaction products are separated by electrophoresis according to their size in a denaturing polyacrylamide. Gels of the labeled reaction products in all four partial reactions create a “ladder” of bands that each differ by one base (level). From this series of rungs of reactions A, C, G, and T, the base sequence can be read from the bottom (position 1) upwards (Figure 30.3).
30 DNA Sequencing
791
Figure 30.4 Autoradiogram of a sequencing run. In each case four adjacent tracks represent the partial reactions A, C, G, and T of a sample. According to the running direction of the gel, the smaller reaction products are at the lower end of the figure. The sequence is read from bottom to top. Each band is different, ideally by one base from the previous one. Source: adapted according to Nicholl, D.S.T. (1994) An Introduction to Genetic Engineering, Cambridge University Press.
Figure 30.4 shows a classic, old-fashioned but instructive, autoradiography of a sequencing gel with radioactively labeled reaction products. The other figures in this section show pseudo chromatogram and fluorescently labeled reaction products. In this illustration, the bands of a track are connected through a cut line in the running direction of the gel and the corresponding band intensities are determined. In this way, the data is reduced by one dimension, creating a representation known as trace data, which illustrates the intensity curve according to time. This reaction principle has remained unchanged since the developments of Sanger from 1977 until today. The discovery and modification of DNA polymerases led to the refinement of the method described above to protocols in which T7 DNA polymerase is used, and to the development of cyclical process that combine signal amplification and sequencing into one reaction. T7 DNA Polymerase-Catalyzed Sequencing Reactions T7 DNA polymerase belongs to the class I DNA polymerases, such as Escherichia coli DNA polymerase I and Taq DNA polymerase. Nevertheless, they differ in their function, their properties, and in their structure. T7 DNA polymerase is the replicating enzyme of T7 phage genome, while the other polymerases are responsible for repair and recombination. Accordingly, the native enzymes contain exonuclease activity: T7 DNA polymerase has a 3´ -5´ exonuclease activity, DNA polymerase I 3´ -5´ and 5´ -3´ exonuclease activity, while Taq DNA polymerase shows only a 5´ 3´ exonuclease activity. Through N-terminal deletion, the 3´ -5´ exonuclease activity could be deleted in T7 DNA polymerases and some thermostable DNA polymerases, as it would otherwise lead to deterioration in the sequencing results. T7 DNA polymerase is distinguished from other non-modified enzymes by their significantly lower discrimination against modified nucleotides. This has the advantage of lower costs and improved sequencing results. It is the ideal model DNA polymerase for DNA sequencing, the stability at higher temperatures is low, however. T7 DNA polymerase, because of its higher and better distribution of signal intensity of sequence patterns and the low signal background, has made the Klenow fragment, which was
792
Part IV: Nucleic Acid Analytics
Figure 30.5 Trace (raw) data of a T7 DNA polymerase-catalyzed sequencing reaction. A plasmid was alkaline denatured, neutralized, and subjected to Sanger reaction with fluorescently labeled primers. The data were generated on an automated DNA sequencing apparatus and analyzed.
used as the original DNA sequencing enzyme, largely obsolete. In Figure 30.5, the results of a T7 DNA polymerase-catalyzed reaction are exemplarily represented with fluorescent marking margins. The representation is different from than the top view on a sequencing gel as selected in Figure 30.4. Figure 30.5 shows a longitudinal section through the four lanes of a DNA sequence ladder. The individual bases (marks) can be reproduced in color. Specifically, a sequencing reaction is composed with a dsDNA molecule from the following partial steps: denaturation and neutralization, primer annealing, strand synthesis, reaction stop, and final denaturation, which will be considered below in more detail. Denaturation and Neutralization DNA sequencing reactions can be carried out both by double-stranded (ds) and single stranded (ss) DNA. They differ only in the denaturation as carried out in the first step. A denaturation in the presence of an added primer is a prerequisite for the subsequent enzymatic DNA synthesis. Single-stranded DNA is denatured by the action of heat, but only short term, and for further reaction it is brought to the optimum reaction temperature of DNA polymerase (37 °C.). In the case of the double-stranded DNA, as described here, the protocol begins with an alkaline denaturation, since heat incubation alone is not sufficient for the complete denaturation. Only combination with a strong alkaline agent such as NaOH can bring about the desired strand separation. The subsequent neutralization allows for adjustments of the reaction conditions required for the strand synthesis.
Basic Principles of Hybridization, Section 28.1
Primer Hybridization However, the ordered synthesis initiation only takes place if the DNA used is converted into a linear, single-stranded form and subsequently a primer is hybridized at the designated location. The kinetics of oligonucleotide hybridization is described by the formula derived by Lathe. It provides a connection between the hybridization time t1/2 in which 50% of
30 DNA Sequencing
793
the oligonucleotides hybridize to a template, as well as the length, size, and the concentration of oligonucleotides: t 1 2
Nln2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3:5 105 L Cn
(30.1)
where: t1/2 = the hybridization time, N = number of base pairs in a non-repetitive sequence seconds, L = structure length of the oligonucleotides, Cn = oligonucleotide concentration in mol l 1. If one now uses the most common parameters occurring in a sequencing reaction for a primer length of 18–25 base pairs and a concentration of 10 7 M, hybridization times of between 3 and 5 seconds will occur. Strand Synthesis and Pyrophosphorolysis The synthesis of DNA catalyzed by a DNA polymerase is an equilibrium reaction. The equilibrium reaction is shifted to the side of the condensation products, that is, the reverse reaction runs considerably slower than the forward reaction. With increasing reaction times and depending on the amount of DNA used, however, the reaction can run essentially backward. In these cases there may be a reduction of terminal dideoxynucleotides. This effect, which is called pyrophosphorolysis, manifests itself in disappearing sequencing bands in the sequencing gel (Figure 30.6). Pyrophosphorolysis therefore seems to be sequence-specific at selected locations in both T7-catalyzed as well as cycle sequence reactions. However, the removal of diphosphate from the reaction equilibrium can suppress the
Figure 30.6 Comparison of a sequencing run with diphosphatase (a) and without diphosphatase (b). The complete lack of reaction products (position 91, 92) and the reduced signal strengths at specific positions (position 68, 86, 106) are easy to detect.
794
Part IV: Nucleic Acid Analytics
reverse reaction almost quantitatively. This can be achieved by adding a diphosphatase (pyrophosphatase), which cleaves the diphosphate into monophosphates. Strand Synthesis and Cofactors The process of enzyme-catalyzed DNA polymerization is Mg2+-dependent. It is speculated that Mg2+ ions are required during catalysis for the stabilization of the α-phosphate group of the incorporated nucleotide. Sequencing reactions that contain Mg2+ as counter-ions for nucleotides are characterized by strong variations in the signal levels of the individual sequencing bands. The partial substitution of Mg2+ by Mn2+ in T7 DNA polymerase reactions leads to homogenization of the signal intensities and thus facilitates reading of sequences, especially regions in which the resolution of the sequencing gel decreases. This effect, however, cannot be applied to thermostable DNA polymerases as Mn2+ inhibits the entire reaction in all the ten cases that were studied. Strand Synthesis and Additives For additives in DNA sequencing reactions, different groups of substances are considered, such as proteins or detergents. The addition of one protein from single strand binding Escherichia coli (SSB) or the T4 gene 32 product could in fact stabilize single-stranded DNA structures. However, the addition shows only marginal improvements in the reaction with a high concentration necessary. We have already discussed the combination of DNA polymerase and diphosphatase above. Further possibilities include combinations of DNA polymerases with different characteristics. In a mixture, a polymerase for the DNA brand marking can be combined with another polymerase to synthesize the sequencing products. The reduction of background signals with the addition of DMSO in T7 polymerase reactions is attributed to its denaturing effect. The addition of formamide, and detergents such as Triton X-100, reduces the background of thermostable enzymes of catalyzed sequencing reactions. Strand Synthesis and Nucleotide Analogs In dideoxy sequencing reactions the nucleotide analogues C7 deaza-dGTP (Figure 30.7a) and dlTP (Figure 30.7b) are used. Both analogues are accepted by DNA polymerases and built into the polymerized strand. However, their derivatization prevents the formation of hydrogen-bonds. This effect is necessary in areas or structures with high GC content. Otherwise, an effect may occur that is referred to as compression. In sequence gels a zone is then observed, in which the band gap is continuously shortened and is eventually cumulated in a considerably broader zone. After this compression, the next bases occur again only after a significantly wider, empty zone. In the compression zone, more than one sequencing product can be located within a visible band. The usually appropriate spacing pattern
Figure 30.7 Structure of deoxynucleotide analogues in dideoxy sequencing reactions: (a) C7 deaza-dGTP and (b) dITP.
30 DNA Sequencing
795
for a base is severely disrupted. This compression is caused by the interaction of highly GCcontaining, complementary sections. It can result, for example, from hairpin structures, which differ drastically from regular sequences in their mobility. Final Denaturation For exact sizing of DNA fragments a complete denaturation is necessary to prevent a sequence-dependent folding of DNA molecules or the formation of conglomerates among multiple DNA molecules and thus an uncontrollable runability. For the denaturing agent urea is usually used at a concentration of 7–8 M. Because of their polar properties, the carbonyl group and the amino groups of urea compete with the individual bases for the formation of hydrogen bridge bonds and can thus prevent the formation of base pairs. The additional use of formamide – after completion of secondary sequencing reactions and subsequent heat treatment – already results in a broad denaturation in the reaction vessel before applying the sample onto the sequencing gel. Chelation of the divalent metal ions (Mg2+, Mn2+) present in the reaction environment through EDTA leads to the dissolution of the DNA polymerase complexes.
Cycle Sequencing with Thermostable DNA Polymerases The information obtained with T7 DNA polymerase about the structure and function of the enzyme could be transferred to thermostable DNA polymerases in wide ranges. By way of genetic engineering it was thus possible to completely remove 3´ -5´ exonuclease activity. The positive feature of T7 DNA polymerase by which it discriminates only slightly between dNTPs and ddNTPs could be attributed to the tyrosine residue 526 of the nucleotide binding point. An exchange of the corresponding phenylalanine residue at position 667 of the Taq DNA polymerase also allowed for a lower discrimination. Today, thermostable DNA polymerases are available that match the properties of T7 DNA polymerase in essential points, but have considerable advantages because of their thermal stability. Mixtures of polymerases, additives, and modifications that can mediate, for example, hot-start properties make cycle sequencing the method of choice. The application of thermostable DNA polymerases allows for amplification and an additional sequencing in DNA sequencing analogous to PCR (Figure 30.8). This process is called cycle sequencing. Unlike PCR there is only one primer in the reaction, so it will be amplified only linearly and not exponentially. The repeated heat incubation is also sufficient for the denaturation
Figure 30.8 Principle of a cycle sequencing reaction. In a cyclically (about 30 times) accumulated sequencing reaction, a mixture is incubated consisting of primers, template, thermostable DNA polymerase, dideoxynucleotides, and the deoxynucleotides at 97, 60, and 72 °C. Because of the high temperatures, primer and template are thermally dissociated. At the lowest temperature occurring in the process, primer and complementary DNA section are associated, to be extended and terminated at a medium temperature through DNA polymerase. Source: adapted according to Strachan, T. and Read, A.P. (2005) Human Molecular Genetics, 3rd edn, Oxford University Press, Heidelberg.
796
Part IV: Nucleic Acid Analytics
Figure 30.9 Comparison of enzyme processivity in highly repetitive sequence sections. The shown repeat has a degree of repetition of 200(!). The structure is clearly resolved with T7 DNA polymerase (a), while a thermostable DNA polymerase (b) shows an unspecific break after the first repeat.
of double-stranded DNA. In a mixture of DNA templates, primers, thermostable DNA polymerase, and a dNTP/ddNTP mixture, a thermal profile, consisting of primer denaturation, primer annealing, and DNA synthesis, is repeated up to 30 times, thereby repeating a sequencing reaction up to 30 times. A correspondingly large amount of sequencing fragments is produced. The use of genetically modified enzymes, detergents, and diphosphatase allows for the production of sequencing data that reach the quality of T7 polymerase reactions (Figure 30.9). Despite the now improved cycle sequencing conditions there are structures, in particular repetitive sections, that cannot yet be accurately determined through these protocols. Figure 30.10 gives an example of this. Even today, DNA sequencing requires the use of different methods, since each individual method cannot cover all areas satisfactorily. Even the expensive methods of chemical fission of Maxam and Gilbert (Section 30.1.3) are rarely but still in use in difficult cases. Figure 30.10 Introduction of fluorescent markings in DNA sequencing reactions: (a) labeled primers, (b) labeled deoxynucleotides and (c) marked dideoxynucleotides. In cases (a) and (b), further deoxynucleotides can be fused after installation of the selected group.
DNA Probes, Section 28.2.1
30.1.2 Labeling Techniques and Methods of Verification In the following, the incorporation of marker groups in the sequencing products and their detection in automated systems is described. Isotopic Labeling The use of radioactive isotopes in DNA sequencing is based on the fact that DNA polymerases do not discriminate between different isotopes and the incorporation into a DNA strand does not lead to changes in mobility in the gel. The isotopes 32 P and 35S are used. The radiation from 32 P is of higher energy than that of 35S. The exposure times in the autoradiography
30 DNA Sequencing
797
that follows the gel electrophoresis can therefore be shorter with 32 P. However, the spatial resolution is significantly lower, due to the greater energy-related blackening areas. Therefore, especially for longer DNA sequencing runs, 35 S is preferred. The label may be introduced by a radioactively labeled primer or during the sequencing reaction itself. Radioactive labeling of DNA sequencing primers is performed by phosphating with γ-32 P-ATP and polynucleotide kinase. The primers produced by chemical synthesis have free 5´ OH groups, which can be phosphorylated directly, without prior dephosphorylating. The mark during the dideoxy sequencing reaction is performed by adding α-32 P-dATP. The isotope is incorporated into the synthesized DNA strand, which thus is detectable. Detection is carried out either by autoradiography or the use of image plates. The autoradiograms can be evaluated manually or semi or fully automatically with a digitizer or scanner. This method is now nearly outdated. The introduction of fluorescent markings is more difficult than that of radiotracers. The fluorescent groups have a considerable size and are only marginally integrated in many cases by these enzymes due to steric conditions. If the detection groups are accepted, however, it is necessary to prevent statistical multiple incorporation. Owing to the other charge ratios and the additional mass they would inevitably lead to a change in the gel electrophoretic mobility and thus exclude a sequence determination. Labeling can be carried out with a labeled primer, internal labeling, or labeled terminators (Figure 30.10). Fluorescein, NBD, tetramethylrhodamine, Texas Red, and cyanine dyes are used for labeling. The dyes can be linked to the appropriate components (nucleotides, amidites) and are accepted as either 2´ -deoxynucleotides or 2´ ,3´ -di-deoxynucleotides of DNA polymerases. Through laser excitation in detection systems, they are only slightly bleached and are sufficiently stable under coupling, sequencing, and electrophoresis conditions. Fluorescent Labeling and Online Detection
Combinations of fluorescent dyes are also used. These so-called energy transfer group systems are based on the idea of being able to excite individual dyes for fluorescence whose excitation spectrum does not correspond to the excitation wavelength of the system or to use the emission wavelengths of a color that does not interact with other dyes used in the system (minimization of spectral overlap). The laser-induced emission of a dye is used to excite the actual fluorescent marker (Figure 30.11).
Energy Transfer Dyes
The use of a 5´ fluorescently labeled primer in a DNA sequencing reaction is not critical. The unlabeled products that exist in each preparation of a primer are not visible in the analyzers and therefore do not disturb the results. A rarely occurring self-priming of the template DNA, caused by partial self-complementarity, would not be visible. Non-specific termination, which is generally due to inadequate reaction conditions or the structure of the DNA that is to be sequenced, can, however, be recognized as a non-readable structure. The marking of the primers generally takes place by linking a fluorescent amidite in the last step of the synthesis (at the later 5´ end). An amino link can also be added through another amidite and these are linked after the conclusion of the synthesis with the fluorescent dye. However, the efficiency of this process is less Primer Labeling
Figure 30.11 Energy transfer dyes: (a) 5- (and 6-) carboxytetramethylrhodamine succinimidyl ester (5(6)-TAMRA, SE); (b) 5-(and 6-)carboxy-X-rhodamine succinimidyl ester (5 (6) -ROX, SE); (c) 5-(and 6-)carboxyfluorescein succinimidyl ester (5(6)-Fam, SE).
798
Part IV: Nucleic Acid Analytics
than the simple linking and requires further purification steps. The labeled primer is set in the reaction similarly to a radioactive primer. Internal Labeling with Labeled Deoxynucleotides For the internal label, fluorescently labeled 2´ -deoxynucleotides are used (e.g., fluorescein-15-dATP), and a DNA polymerase. The sequencing process must run in two stages. In a first step, with a low concentration of the labeled 2´ -deoxynucleotides, a maximum of one is added to each strand. A prerequisite is that the position following the primer allows the installation of this nucleotide. The concentration of marker component should be kept so low that in the next step of the reaction no further nucleotides of this type are installed. Another random incorporation would lead to uncontrolled mobility changes. The actual sequencing reaction is carried out according to the classical Sanger principle. This technique offers the advantage of using inexpensive unlabeled primers for a reaction. However, undesirable reaction products that are the result of self-priming, can also be labeled and complicate the sequencing results by overlays of multiple sequences. This method has become increasingly obsolete with the broader usage of better cycle sequencing methods. Labeled Terminators
Fluorescently labeled terminators have prevailed as a standard and offer the possibility of using non-labeled primers in DNA sequencing reactions. In the reaction mixtures, the labeled 2´ ,3´ -dideoxynucleotides replace the completely unlabeled analogues. The marking is done through the simple incorporation of a labeled dideoxynucleotide. Another advantage is that reaction products that are not properly terminated with a dideoxynucleotide, due to the lack of fluorescence group, are not detectable. Such false, sequence-dependent reaction products are visible with labeled primers or deoxynucleotides in the system. The availability of base-specific color-labeled dideoxynucleotides allows the execution of the reaction and gel electrophoresis in one reaction vessel and a single track. The disadvantage, however, is that the DNA polymerases hardly accept these modified nucleotides, so that a high working concentration and thus a subsequent purification of the reaction products is required. Furthermore, the dyes in gel electrophoresis have different mobilities, which must be corrected by software. For selfpriming products the same as was said for deoxynucleotides applies. Duplex DNA Sequencing
Online Detection Systems In 1986 and 1987, online DNA sequencing systems were developed based on laser-induced fluorescence. Significant developments were made by L. Hood in the USA and W. Ansorge in Europe. Online detection systems consist mainly of a vertical electrophoresis system, an exciting laser, a detector, and a recording computer system (Figure 30.12). The laser is either linked transversely to the longitudinal sides perpendicular to the detector or at a certain angle from the front or rear side in the sequencing gel. The locally resolved bands of the classic method are now seen as time-resolved band patterns. The first system of Smith, Hood, and their colleagues was based on the detection of reaction products, which were produced with the use of labeled primers in different fluorescent colors in the dideoxy method. All products of a reaction were applied to the gel on one track. The fluorescence dyes used, such as fluorescein, Texas Red, tetramethylrhodamine, and NBD, have a sufficient spectral distance to enable a secure distinction of the bases. In a further development fluorescence labeled terminators were available, which made the complex primer marker virtually superfluous. The mobility differences between the dyes used make a manual analysis of the primary data impossible, but can be corrected automatically by software programs. The differing absorption spectra also require excitation with two different wavelengths. For observation of a gel in its entire width, a scanning mechanism has been developed. The original electrophoresis systems were based on planar gels and were later replaced by capillary electrophoresis systems, which allow a higher speed and better resolution at lower thermal load. Automated DNA sequencing systems support a much higher throughput of sequencing reactions than traditional methods (Figure 30.13). While an average of four to six reactions take place on a radioactive gel, online systems now have a payload capacity of 96 clones and read lengths up to 1000 bp per sample. Automated Sample Preparation Automation strategies are often based on a flexible liquid handling system, which allows various applicative adaptations and accessories and thus a very wide range of applications with particular effectiveness. There are also devices available that have been specifically designed and optimized for a specific chemical reaction sequence. These
30 DNA Sequencing
799
Figure 30.12 Principle of an online DNA sequencing device. A vertical gel electrophoresis apparatus is observed at the lower end of a detector. The excitating laser beam is coupled at the level of the detector into the gel (not shown here). The signals obtained from the detector are sent to an analyzing computer. The spatially resolved band patterns, derived from radioactive DNA sequencing, are replaced by a time-dissolved listed banding pattern marked by the detection finish line. Source: adapted according to Smith, L.M. et al. (1986) Nature, 321, 674–679.
Figure 30.13 Automated capillary DNA sequencer. Source: adapted according to Perkel, J.M. (2004) The Scientist, 18, 40–41 (now LabX Media Group).
800
Part IV: Nucleic Acid Analytics
different automation strategies are derived both from the different sample throughput as well as from the level of complexity of the process. Flexible automation strategies are used in the field of small and medium numbers of samples ( A: A-cleavage reaction: After methylation of the adenine residue N3 with dimethyl sulfate, the N-glycosidic bond is cleaved. The added piperidine leads to the elimination of the base and simultaneous β-elimination of both phosphates.
30 DNA Sequencing
803
Figure 30.17 (a) Thymine- and cytosine-specific fission reaction. Hydrazine leads to a ring opening of the base between C4 and C6 and their fission. The added piperidine leads to the simultaneous β-elimination of both phosphates. (b) C-cleavage reaction. Hydrazine opens the ring between C4 and C6 and splits off the base residue. The added NaCl suppresses the reaction of thymine. The added piperidine leads to elimination of the base and simultaneous β-elimination of both phosphates.
804
Part IV: Nucleic Acid Analytics
G > A: The methylated DNA is first heated at neutral pH for 15 min at 90 °C. This leads to
the elimination of all methyl-A and all methyl-G. The A-depurination occurs four- to sixtimes faster than the G-depurination. Because of the low adenine methylation degree, however, the G-depurination outweighs the A-depurination. Hot alkaline treatment (piperidine or NaOH) of the partially depurinated DNA therefore leads to a strengthening of the Gcleavage and weakening of the A-cleavage (G > A-pattern). The A-part of the G > A-reaction is shown in Figure 30.16b. Single-stranded DNA is also methylated at the N1-atom of adenine and at C3 of cytosines. Methylation on N1 of adenine does not lead to cleavage. At C3 methylated cytosines are cleaved and are visible as bands with about 1/10 to 1/5 of the intensity of the G-bands. A > G: The methylated DNA is treated at 0 °C with dilute acid (e.g., 0.1 M HCl) for 2 h. This leads to a preferential hydrolysis of the glycosidic bond of methyladenines. Hot alkali treatment of the depurinated DNA thus results in a stronger A- and in a weaker G-fission (A > G pattern). A > C: To achieve this specificity, the DNA is treated with strong alkali (1.2–1.5 M NaOH) at 90 °C for 15–30 min. This opens the adenine and cytosine rings. Hot piperidine treatment leads to the elimination of these bases and to A > C cleavage patterns. A + G: Limited treatment of DNA with formic acid leads to unspecific depurination. Alkali treatment generates the A = G or A + G pattern. C + T: Cleavage at the pyrimidine bases is performed by modification with aqueous hydrazine and subsequent cleavage with alkali. The chemical reactions of the thymine part of this combined reaction can be seen in Figure 30.17a. Piperidine reacts with all glycoside products generated by hydrazine. C: Inclusion of 1–2 M NaCl in the previous reaction suppresses the reaction of hydrazine with thymine. The steps of the cytosine reaction can be seen in Figure 30.17b. Piperidine reacts here with all the glycoside products of the hydrazine.
Except in the case of the G-reaction, NaOH (0.1 M) or piperidine (1 M) can be used as alkali. Piperidine is preferred because it can then be removed very easily by evaporation. When using NaOH, the DNA must be precipitated with ethanol. More base-specific reactions of DNA are presented in Chapters 31 and 32. However, for the methods discussed here, they are not equally important for the determination of the DNA sequence. Basically, the reactions G, G+A, C+T, and C are sufficient to be able to read a sequence clearly. Solid Phase Process The chemical cleavage could be greatly facilitated by the development of a solid phase process and its adaptation to the automatic DNA sequencing that works with fluorescence. After binding of the DNA template to an activated membrane surface, all cleavage reactions and required washing procedures up to the elution of the reaction products can be performed on a solid surface.
Figure 30.18 Structure of α-thionucleotides.
α-Thionucleotide Analogues The incorporation of α-thionucleotides (Figure 30.18), for example during a PCR reaction, allows a simple chemical cleavage reaction with 2,3-epoxy1-propanol and has the additional advantage that a DNA strand is not attacked by exonuclease activities; compared to the standard Maxam–Gilbert method, 5´ -O-1-thiotriphosphate shows good incorporation characteristics with major DNA polymerases. However, the band patterns generated only allow a reliable sequence determination in short areas. Multiplex DNA Sequencing Multiplex DNA sequencing (Figure 30.19) is based on the above chemical cleavage of DNA. While in the above case detection can take place immediately, it requires sophisticated hybridization patterns in multiplexing. The goal is the simultaneous performance of 50 sequences in a reaction, thereby minimizing the workload of the elaborate cleavage reactions and gel operations. The 50 DNA fragments to be sequenced are cloned in 50 different vectors. These vectors differ in the fragments of left and right flanking linker sequences. This vector-specific oligonucleotide follows a standardized restriction site. After the excision of the DNA fragments they are enclosed by the above-mentioned sequence regions. All fragments are united in a cleavage reaction, separated by gel electrophoresis, and transferred to a nylon membrane (blotting). The reaction products are detected by successive hybridizations of the
30 DNA Sequencing
805
Figure 30.19 Multiplex DNA sequencing.
fragments with labeled oligonucleotides that are complementary to the flanking regions. In this way, a filter can be recycled up to 50 times. The variant of the multiplex walking dispenses with the elaborate cloning at the beginning of the process, and instead fragments a cloned DNA and then essentially follows the above procedure. The first hybridization begins at the start point specified by the vector. From the obtained sequence a new oligonucleotide is then synthesized and hybridized again. By repeating the process, new starting points are generated that make it possible to travel along the fragment (primer walking). RNA Sequencing Typically, mRNA is not sequenced directly, but converted by reverse transcription into cDNA and enzymatically sequenced by Sanger’s method. For sequencing of rRNA, chemical cleavage is sometimes used. In chemical RNA sequencing, we are talking about a chemical cleavage by a Maxam and Gilbert method analogue. Owing to the lower stability of the phosphodiester bond and other chemical compositions of RNA, modified cleavage reactions are used. A 3´ -terminal marked RNA fragment is subjected to four parallel batches of base-specific modification reactions and the RNA strand is split with aniline instead of NaOH or piperidine:
G-reaction: base methylation with dimethyl sulfate, followed by reduction with sodium borohydride and cleavage of the modified RNA strand in the ribose bond with aniline.
A> G-reaction: ring opening of the bases with diethyl dicarbonate (diethyl pyrocarbonate) at N7 and subsequent strand cleavage with aniline.
U-reaction: Treatment with hydrazine results in cleavage of the base by nucleophilic addition to the 5,6-double bond. The modified RNA strand is cleaved by aniline.
806
Part IV: Nucleic Acid Analytics
C> U-reaction: Treatment with anhydrous hydrazine in the presence of NaCl leads to preferred release of cytosine. Again, the strand is cleaved by aniline.
30.2 Gel-Free DNA Sequencing Methods – The Next Generation Gel-free sequencing methods brought in a new wave of sequencing technologies starting in 2005 with a version of pyrosequencing technology named 454. One particular feature of this and the following technologies is that they produce large amounts of sequencing data, but it is made up of generally much shorter reads than the standard Sanger sequencing method. To obtain the mass of sequencing data that characterizes the Next generation sequencing (NGS) wave, all the techniques rely on a high level of parallelization of their sequencing method. Massively parallel sequencing (MPS) is another term used for the NGS sequencing. With the start of the NGS wave it became apparent that sequencing of individual genomes was within reach, with all the impact that this may have on health care and personalized medicine. As a result many different NGS methods have been proposed, and following on from them several attempts have been made to commercialize various methods (for a review of methods proposed and the relevant time line see Reuter, J.A., Spacek, D.V., and Snyder, M.P. (2015) Mol. Cell, 58 (4), 586–597). As a result of the intense method development and rapid commercialization many of the newer sequencing technologies are referred to by their commercial name – that is, Illumina sequencing and not the underlying method – sequencing by synthesis. The gel-free methods eliminate gel electrophoresis as the throughput and resolution limiting factor in DNA sequencing, but cannot hide their similarities to the Sanger sequencing method. Many are based on the incorporation of labeled nucleotides that have the label removed after detection and are converted into an actively extendable polynucleotide string. The detection occurs in situ bound to a solid surface instead of a measured end point in a gel. Therefore, it is possible to achieve an extreme size reduction in the reaction volume and the related detectors. The use of activated surfaces makes a dense packing of the reacting molecules possible and thereby a huge parallelization of the sequencing and a huge sequence output. As almost every single incorporation event is detected (HiSeq, 454), the throughput is mainly governed by the microfluidics of the system in question – how quickly can the sequencing cycles be captured versus the buildup of fluorescent noise in the system with time. However, there are already new systems available that are mainly governed by the progressivity of the polymerase (PacBio RS). The next generation of sequencing devices, those that use nanopores or other methods, are capable of measuring single molecules and are already to some extent available. Since the next generation of sequencing devices came to the market the amount of sequencing data has expanded hugely, as has the size of sequencing projects that are being attempted (e.g., 1000 genomes and 1000 cancer genomes). The rate of sequence data production continues to increase with systems such as the x10 from Illumina promising $1000 dollar human genomes. Further increases in the sequencing output are predictable, however, and the data produced from the latest sequencing instruments can measure over a terabase produced in less than 4 days. This has meant that not only are there new opportunities appearing for biological research but also new challenges appear dealing with the large amounts of data produced. New methods in bioinformatics for analysis of the data such as aligning large numbers of small sequences have had to be developed and IT infrastructure that holds and processes these data has had to be expanded. NGS data analysis can last for days or weeks even on a powerful cluster. It can have very high memory requirements (1 terabyte of RAM) and the transfer of 100 GBs of result data can push the boundaries of conventional WAN (wide area network) technology. In some aspects the preparation of probes has become easier, the classic cloning steps are removed that can lead to an unwelcome bias, although other biases are added such as PCR sensitivity to base content. In other aspects the process does lead, however, to a much greater effort in library preparation. The available systems have a detection level of around 25 000 molecules and therefore require amplification of the sequencing product to be able to detect it. The sequencing methods are also very dependent on the size distribution of the molecules to be
30 DNA Sequencing
807
sequenced, therefore shearing and accurate size selection of the input material is required to reduce the size of the starting molecules to a few hundred bases long.
30.2.1 Sequencing by Synthesis Classic Pyrosequencing Through analysis of the byproducts of a polymerization reaction one can determine the nucleotide sequence (pyrosequencing, Figure 30.20). As mentioned above, every polymerization reaction results in a free diphosphate and these can be detected by another chemical reaction. The enzyme sulfurylase catalyzes the conversion of the diphosphate into ATP, which when hydrolyzed with luciferase results in luciferin and oxyluciferin. The polymerization can be detected by adding single nucleotides to a polymerization mixture in a reaction chamber. A successful polymerization produces the expected emission of light. Through successive cycles of nucleotide addition, detection of polymerization by emitted light, and removal of the reaction products, the sequence of a nucleic acid can be determined. Light is only emitted when the correct nucleotide for the sequence is added. The presence of homopolymer stretches results in a larger signal based on the number of nucleotides incorporated although the increase in signal cannot be relied upon to be a linear representation of homopolymer length. In general, stretches of more than eight identical nucleotides cannot be reliably quantified. Each cycle time takes a few minutes per base and read lengths of around 60 bases are achievable, although in this form pyrosequencing has been generally regarded as a mini-sequencing technique. 454-Technology (Roche) The 454-system was the first second-generation MPS device to the market and uses pyrosequencing as its basis although the sequence read lengths are longer than in classic pyrosequencing and the high level of parallelization of the process gave a huge increase in sequence data output at the time. The sequencing process (Figure 30.21) is made up of three steps: 1. Creation of a library of DNA molecules in the range 300–800 bp. The starting material that is longer than the acceptable range, such as genomic DNA, is sheared by nebulization, purified, the overhanging ends repaired and then phosphorylated (Figure 30.21a, 1). Afterwards two adapters, A and B, are ligated to the target molecules (2). The adapters are each 44 bp long themselves and are not phosphorylated but carry the target sequences for amplification and sequencing. A 4 bp long key sequence identifies the library sequences to the system and facilitates the calibration of base recognition and calling. This step is left out when sequencing PCR products as the necessary adaptors should have been added during the PCR amplification itself. The B adaptor has an extra 5´ -biotin modification. This facilitates a cleaning step using streptavidin-covered paramagnetic beads (3). The library fragments without biotin are simply washed away. The gaps between the adapter and the fragment are filled by a strand-displacement DNA polymerase catalyzed reaction. During the cleaning procedure alkali denaturing of the one sided modified fragments and occasionally the unmodified and complimentary strand are set free for further steps, while double labeled fragments spontaneously reattach under the chosen conditions. 2. The DNA being sequenced has to undergo clonal amplification to give a clear signal above the detection system’s lower limit (4, Figure 30.21a). A bead dilution is performed so that ideally only one capture bead binds to one DNA molecule. The beads carry matching capture primer, allowing them to bind to a DNA fragment. The beads are then mixed in an oil and water emulsion, where an emulsion-PCR reaction (emPCR) takes place and the sequences are amplified (5). A further biotinylated primer is used to allow downstream purification of clonally amplified sequences. After the PCR reaction the emulsion is broken and the beads binding DNA, byproducts, and empty beads are freed. Only the beads with amplified sequences are bound by streptavidin beads and magnetically filtered out of the mixture. Through the breaking of the emulsion and following alkali denaturing a bead is produced that has a bound, single-stranded DNA molecule that can be sequenced. 3. For DNA sequencing (Figure 30.21b) the sequencing primer is hybridized to the bound template in a reaction mixture containing sequencing primers, DNA polymerase, and required cofactors. The mixture is transferred to a pico-titer plate with the aim of having just one bead occupying one well of the plate. Each well has a diameter of 44 μm and has a single
Figure 30.20 Principle of pyrosequencing: An immobilized DNA probe is incubated with a reaction mixture containing only one nucleotide (A, C, G, or T) in cycles. If the correct nucleotide is currently present in the mixture it is incorporated in the polymerizing nucleotide chain and a diphosphate is released which is converted into ATP by ATPsulfurylase. This is in turn used by luciferase to convert luciferin into oxyluciferin along with the emission of light. This light emission is used to measure the incorporation of the base or bases in this cycle. Source: Ronaghi, M. et al. (1996) Anal. Biochem., 242, 84–89. With permission, Copyright 1996 Academic Press. All rights reserved.
808
Part IV: Nucleic Acid Analytics
Figure 30.21 454-System Workflow: (a) Library preparation: 1, fragmenting the nucleic acid to a length between 300 and 800 bp; 2, ligation of A and B adaptors; 3, binding of template to a capture bead; 4, bead after clonal amplification; 5, bead after the emulsion is broken and clean-up. (b) Bead bound pyrosequencing. Source: adapted after Roche Diagnostics Corporation.
Phred: The calculation of Phred quality scores dates back to the automatic gel sequencing methods of the 1990s. The value is based on the error probability for each base in a read and is calculated using the following formula: Q 10log10 P
A Phred score of 40 represents an error probability of 1 in 10 000. The Phred scores are presented as a string of characters where the ASCII encoding value for the character minus a particular offset, 33 in most cases, is the calculated Phred score.
engraved glass fiber within the plate housing assigned to it. The fiber in each well transfers its signal to a point on a CCD sensor, allowing the pattern of sequencing to be detected. The dimensions of the well prevent more than one capture bead occupying it. Added enzyme beads contain the two enzymes luciferase and sulfurylase, which generate emitted light in the presence of PPi. The process follows the above-mentioned classic pyrosequencing; however, 400 000 reaction wells are simultaneously active and the achieved read length is significantly longer. During the runtime the detector pictures are analyzed and the signal intensities calculated from the pixel data. These data are reduced to give the intensity values assigned to the particular positions on the pico-titer plate. The series of single pictures is used to finally calculate the quality values for each base (Phred encoded probability of error) and the read sequence is presented as a fluorogram. Yields of the 454 high-throughput variant of this system are around 700 MB with an average read length of 700 bases in around 23 h. The 454 system was a major breakthrough in the world of sequencing technology.
30 DNA Sequencing
809
Illumina Sequencing by Synthesis (HiSeq, MiSeq, NextSeq500, ×Ten) The Illumina technology is based on the integration of fluorescent markers with a reversible terminator dNTP and the result of sequencing result is based on the optical detection of incorporated fluorescent markers through sequencing cycles. As with other sequencing methods a library preparation step is required here as well. Ultrasound is used to produce a set of DNA fragments of the correct length, targeting 200–500 bp. A mixture of T4-DNA-polymerase and Klenow fragment removes the 3´ -overhangs and fill up the 5´ overhangs. T4-polynucleotide kinase phosphorylates the blunt ends of the fragments and incubation with a 5´ -3´ -exo-deletion mutant and dATP leads to a simple adenylation of the ends. Finally, both the differing end adaptors possessing the necessary 5´ T overhang can be ligated (1 in Figure 30.21). The P5 adapter possesses a region complementary to the sequencing primer and the P7 adaptor attaches a complementary sequence to allow binding of the fragment to the flow-cell where sequencing will actually take place. Size selection and separation from unligated adaptor sequences are performed by gel electrophoresis followed by gel elution, or magnetic bead purification, and PCR can be used to further enrich the sample being sequenced. For the DNA to be actually sequenced it must be bound to the sequencing flow-cell and clonally amplified to allow detection. This is done through a process known as clustering (Figure 30.22).
Figure 30.22 Illumina sequencer workflow: 1, fragmentation of the nucleic acid and adapter ligation; 2, loading of the template library on to the flowcell; 3, strand initialization; 4, template preparation by denaturation; 5, clonal amplification; 6, incorporation of fluorescently labeled reversible terminators; 7, detection of the incorporated bases using the scanned images.
810
Part IV: Nucleic Acid Analytics
Clustering occurs on the transparent flow-cell where the sequencing takes place. The flowcell can have up to eight channels (HiSeq) allowing eight separate DNA libraries to be run without multiplexing with index barcodes. The surface of the flow-cell is coated with adaptors with complementary sequences to the P5 and P7 adaptors added in the library preparation above. The library material is denatured and added to the flow cell as single-stranded DNA. After hybridizing with surface oligonucleotides the template strand is copied (3). The product is denatured again and the newly synthesized strand stays bound to the surface of the flow-cell while the other material is washed out (4). In the next step the template can bind a nearby oligonucleotide on the surface and hybridize to this, allowing a further strand synthesis to occur. In this way initialization of the solid phase cluster PCR is achieved. A bridging PCR as shown above results in the required clonal amplification of the sequence template (5). On most Illumina sequencing systems the cluster formation is randomly distributed over the flow-cell and the density, be it over clustered and therefore unreadable or under clustered producing a poor yield, is highly dependent on the concentration of the DNA library added. The size selection of the fragments is also essential to ensure that the PCR products stay within the formed clusters, otherwise the sequence data cannot be unambiguously read. A general rule of thumb would be a cluster density of 610–678 K mm 2 for the HiSeq 2000 platform although other sequencers of the Illumina family require different densities. Illumina has produced patterned flow-cells for their later sequencing platforms HiSeq x10 and HiSeq 4000 that allow higher loading of the flow-cells by restricting cluster formation to fixed well positions on the flow-cell. Once prepared the flow-cell is loaded onto the sequencer. In the first sequencing step all DNA molecules with polymerization capable ends are blocked to stop unspecific primer extension. The sequencing process is then initialized by adding a mix of sequencing primers, polymerases, and four different fluorescent markers, reversible protected dNTPs. Owing to the extensionprotected nucleotide each cluster has just one base incorporated complementary to the template strand (6, Figure 30.22). Finally, at the end of each sequencing cycle the flow-cell is scanned, producing four pictures, one for each of the fluorophore laser excitement frequencies representing one for each base. Afterwards the fluorophores are cleaved off and the end of the sequence strands reactivated for further polymerization. The sequence is called based on the positional information relating to the cluster and the sequence of base specific fluorescence (7). The maximum output from this system is governed by the flow-cell surface available to cluster and the read length, which is itself dependent on the speed at which the fluorescent sequencing cycles can be performed. Illumina has various different models that offer higher throughput rates or longer read lengths. Currently, highest throughput occurs on a HiSeq 4000 with 2× 150 base reads giving around 1.5 terabase of data in 3 days. Longer read lengths but much less data is produced by the MiSeq, giving read lengths of over 2 × 300 bases. Adaptations to Library Preparation and Sequencing The amount and properties of the sequence reads produced by sequencing such as read length and read number are governed by the capabilities of the sequencing instrument and the chemistry it uses. This often means that sometimes the read sequence is too short to be used in certain steps such as spanning over a repeat region, or that there are not enough reads to cover a whole genome. Certain adaptations to the library preparation and sequencing techniques have been developed to deal with this. The examples given below are taken from the Illumina platform but can be applied to other platforms as well. Paired-End-Sequencing When the template is sequenced from both ends (Figure 30.23), the reactive group mentioned above is cleaved after the first end is read. The reverse strand can then cluster via bridging PCR and then be sequenced in the same way the first read is sequenced. The advantages of paired end sequencing are (i) the extra coverage of a second read and (ii) the positional information provided by having two anchor reads of known distance apart. Paired end reads are often used in genomic sequencing, allowing structural variations in the sequenced genome to be identified, and assist genomic assembly considerably (Figure 30.24). Figure 30.23 Pair-end sequencing: adapters A1 and A2 allow bridge amplification from either end of the strand and sequencing primers SP1 and SP2 allow sequencing to be started from each end sequentially.
Mate-Pair Library Creation This refers to the generation of a specific type of library to obtain sequence anchors spanning a wider sequence distance than can be handled with normal paired end reads. Hence the libraries are formed with a large inner distance between the sequence fragment end, circularized, and the shorter outer fragment is used to form paired end reads. Here the goal is to obtain positional information from separated anchor reads and is often
30 DNA Sequencing
811
Figure 30.24 Alignment of paired-end-sequences allows both ends of a sequenced fragment to be aligned. This allows unique alignment of one of the pairs outside a repeat when the other end aligns in a repeated region, allowing repetitive regions to be unambiguously aligned and making assembly of repeated sequences easier.Source: adapted after Illumina, Inc.
used in scaffolding of de novo assemblies (Figure 30.25). For this type of library the first step is to fragment the DNA to 5 kb or greater instead of 200–500 bp. This gives a large spanning distance between the read that can be used to span over repeated regions in the genome. As fragments of this length cannot be used for clustering on the flow-cell the ends are biotinylated and the fragments circularized. The circularized DNA is once more fragmented (400–600 bp) and the biotinylated region captured. The captured fragments are then sequenced as normal in the paired end fashion. Indexed Libraries As the number of sequence reads per sequencing run increases, the question arises: How deep does one have to sequence a particular sample? If it is the case that a desired sequence can be covered with less depth than would be delivered by one lane or run of the sequencing several samples can be run together with sequence barcodes included during library preparation (Figure 30.26). In the case of Illumina sequencing a library is created as described above. After adaptors are ligated that already carry the sequencing read primer 1 the fragments are PCR amplified. Here the index barcode is introduced to the library fragments. One of the PCR primers has the index sequence and the sequencing primer for the second read. A second primer contains the P5 structure for attachment to the flow-cell surface. A third primer carries the P7 attachment section and the index sequence. The first set of index barcodes consisted of 12 modified P7-adapter sequences to be used so several different samples could be run in the same flow-cell lane. The index is separately read through an index read and can be up to eight bases long. Now, the use of dual indexing barcodes of 16 bases is possible. Other methods involve the integration of the index after the standard sequencing primers. However, with this method there is the disadvantage that there is a loss in sample sequence length as the index is read at the start of the first sequencing run. Target (Exome) Enrichment When it is uneconomical to sequence a whole genome or genomes of individuals or if only a particular genomic region is required to be sequenced, target enrichment can be used to extract only the desired regions of the genome for sequencing. The most frequent example of this is exome enrichment or exome capture. This method reduces the sequencing to the coding regions of the genome and allows comparison of the exome sequences from a range of individuals. Illumina technology is also used as an example here (Figure 30.27). After denaturing, the fragments of the DNA library are hybridized to a capture library of biotinylated probes and then bound to magnetic streptavidin-coated beads and eluted. The enrichment process is repeated and the eluted capture product is amplified before sequencing. The produced inserts cover up to 460 bases around the center of the probe and therefore can also include exon flanking regions. The methods of other companies are based on similar principles but often use other probe sequences to capture a different subset of the genome.
Semiconductor Sequencing (Ion Torrent) This method is also based on DNA sequencing by synthesis. The material is also handled similarly to the above processes (creation of the library and clonal amplification). The technology here, however, is based on a different form of
Figure 30.25 Mate-pair sequencing uses biotinylation, circularization, a second fragmentation, and then ligation of paired end sequencing adaptors to the ends of the original molecule. This allows the sequencing of the ends of a fragment, which can be in the range of 5 kb long, giving spatial information about the genome. Source: adapted after Illumina, Inc.
812
Part IV: Nucleic Acid Analytics
Figure 30.26 Indexed libraries: (a) creation of the library. 1, Rd1-SP-adapter ligation; 2, PCR to add the P5 and index-SP and Rd2-SP and P7 sequences with index barcode sequence; 3, structure of the sequencing template. (b) During sequencing three independent read cycles are carried out: 1, read 1; 2, index read after the removal of the first strand; 3, read 2 of the library molecule. The index read is a separate read and in this way does not reduce the length of either read 1 or 2.
detection. With every DNA polymerase catalyzed extension of a nucleotide chain a hydrogen ion (proton) is released that leads to a change in the pH of the reaction volume. In other systems the pH changes are buffered by the reaction mixture, in this method they are used to detect an incorporation of a base. The design of the system transfers a great deal of the sequencing device onto a disposable silicon based sequencing chip. The sequencing reaction takes place on this microfluidic chip in which several cavities are loaded with DNA bound beads. Each nucleotide base is flooded in cycles into the reaction chambers in series individually so that a change of pH measured in the presence of a particular nucleotide indicates which base was incorporated. The charge changes on the sensory surface at the base of the chamber are measured as a voltage change and are used to produce a fluorogram of the sequence. The level of voltage change is used to measure how many bases were incorporated (Figure 30.28). Applications of Sequencing Technologies NGS sequencing has proven exceptionally useful in scientific research. Although the sequence reads are short the mass of them together provides a large amount of data that can be put to good use.
Figure 30.27 Exome enrichment for Illumina systems. (a) Denaturing of the dsDNA library. (b) Hybridization of biotinylated capture probes to the target regions. (c) Enrichment of the target regions using streptavidin beads. (d) Elution of the target fragments from the capture beads. Source: adapted after Illumina, Inc.
Genomic Sequencing and Resequencing Sequencing the first human genome was an exceptionally expensive and lengthy project – estimated cost $3 billion. NGS technologies now allow resequencing of a human genome for under $1000 and in under a week. With the availability of cheap mass sequencing the genomes of organisms that would not normally be sequenced can be constructed although several issues exist using the short read data. If no reference genome is available one has to be assembled de novo. This is a challenging task especially for short read technologies where the read length is often shorter than the length of repeated sequences that are widely spread through eukaryotic genomes. Several programs using a range of techniques such as de Bruijn or String figures have been published to deal with the task of constructing larger contiguous sequences, contigs, from short reads. One way to try and extend the capacity of short reads over the length of short repeats is by using paired and mate-pair sequencing strategies.
30 DNA Sequencing
813
Despite the best efforts of biologists and computer scientists the repeat nature of eukaryotic genomes has led to several newly genome assemblies that are unfortunately made up of fragmented contigs and would benefit from the promise of the third generation of sequencing technologies presented below. In cases such as humans where a good reference genome exists it is possible to produce a consensus by aligning reads to the reference. This is widely used in human genome resequencing where variation from the reference is being studied. Single base and small deletions can be detected by modern alignment programs whereas larger structural variants, inversions, deletions, or chromosomal rearrangements may be detected using paired or matepair information. In some cases it is not required to sequence the whole genome of an individual but only a reduced portion such as only the coding sequences for the genes, the exome. This approach has been heavily exploited in disease screening in humans where it was at one stage still impractical to sequence the whole genome. With the advent of higher throughput from the sequencers many large projects have moved to whole genome sequencing with the advantages of more structural variant and non-coding variant information. However, exome sequencing still provides a good way to sequence the gene coding information from several individuals. Tag Counting – RNASeq and ChipSeq The production of millions if not billions of small sequence reads lends itself very well to various semi-quantitative and qualitative techniques that use different ways to isolate nucleic acids with the specific aim of investigating a particular biological question. The field has expanded widely, encompassing 70 techniques (Illumina has published a summary of available methods – ForAllYouSeqMethods.pdf, available online). Two of the earliest and most widely used techniques, RNASeq and ChipSeq, are outlined below. RNASeq (Mortazavi, A. (2008) et al. Mapping and quantifying mammalian transcriptomes by RNA-seq. Nat. Methods, 5, 621–628) allows the investigation of gene expression by measuring RNA levels in a cell or tissue. Total RNA is isolated from the tissue and enriched for coding sequences by either capturing poly-A tail containing molecules or by depletion of ribosomal RNA with probes that perform specific sequestering. The RNA is then converted into a single stranded cDNA library using random hexamer priming, following which the second strand of the cDNA is synthesized, creating a library of cDNA molecules that can be further processed in library preparation. Once the library is prepared it is sequenced as normal on the sequencer. The short read tags are then aligned to a genome and the alignment positions mapped back to the relevant gene annotation. The number of aligned counts per gene, transcript, or exon is taken as a measure of gene expression, with higher tag counts indicating higher gene expression. Paired end sequencing or aligning with an aligner capable of splitting the short read alignments up so that they span exon junctions are used when investigating alternative transcript usage in gene expression. ChipSeq (chromatin-immuno-precipitation sequencing) (Robertson, G. et al. (2007) Genome-wide profiles of STAT1 DNA association using chromatin immunoprecipitation and massively parallel sequencing. Nat. Methods, 4, 651–657) is used to investigate the binding sequences of DNA binding proteins such as transcription factors. DNA and bound proteins (chromatin) are reversibly crosslinked using formaldehyde. The chromatin is sheared and enriched by immunoprecipitation of the DNA binding the protein using an antibody specific for the protein in question. After enrichment the crosslinking is reversed and the DNA is processed into a sequencing library and sequenced. The short reads are aligned against a genome and the number of tags aligning above background and the shape of the peak of aligned reads is used to ascertain the binding position of the protein under investigation. In the case of transcription factors the question is often: Which gene is downstream of the binding site?
30.2.2 Single Molecule Sequencing The above established sequencing methods require an amplification, either by classic cloning in the “older” processes or by clonal amplification in the newer massively parallel sequencing techniques. The availability of new more sensitive detectors and methods based on other
Figure 30.28 Ion torrent: (a) Single reaction chamber of a sequencing chip with a single bead bound DNA template, the sensor, and electronics. With the addition of dNTPs, protons (H+) are set free that change the pH of the chamber. These changes are measured by the chamber’s sensor. (b) Variation of bases with time. The peak heights of the figure represent the number of bases detected. The bases incorporated are identified by the step in the synthesis and the dNTP added. Source: adapted after Ion Torrent Systems, Inc. (now Thermo Fisher Scientific GENEART GmbH).
814
Part IV: Nucleic Acid Analytics
Figure 30.29 PacBio RS single molecule real time DNA sequencing. (a) Single ZMW with a polymerase bound to the lower surface of the chamber. The illumination from underneath only affects the lower region of the chamber. (b) Diagram of the sequencing process (incorporation, fluorescence emission, cleavage of the fluorophore). Source: Eid, J. et al. (2009) Science, 323, 133–138. With permission, Copyright 2009, American Association for the Advancement of Science.
principles brings the possibility of measuring single molecules within reach. This has brought on the concept of third-generation sequencing techniques focused on single molecules. Single Molecule Real Time (SMRT) DNA Sequencing (Pacific Biosciences) This technique sees a single DNA molecule being sequenced in a very small chamber called a zero mode waveguide (ZMW) in a sequencing cell. There are thousands of ZMWs in the chip and thereby thousands of molecules are sequenced in parallel and also in real time. The special thing about the ZMW is that it is so narrow that light is trapped at its base when emitted from sets of fluorescently labeled nucleotides due to nucleotide incorporation: 1. The starting material is fragmented and end repaired. A hairpin adaptor is ligated on each end resulting in a closed circular molecule. The hairpin structure contains a sequencing primer complementary sequence. The final products are size selected and cleaned. 2. The sequencer primer is hybridized onto the hairpin structure. After which Phi-29 polymerase is added and an initiation complex formed that can be loaded onto the chip and sequenced. 3. Once immobilized in the ZMW the sequence of the DNA template is measured by the fluorescence of incorporated molecules. Each base has a different fluorescent label and when incorporated leads to a large burst of light emitted that is trapped for some time in the ZMW, giving a clearly measurable signal. This clear incorporated signal is much stronger than that of unincorporated labeled nucleotides that move in and out of the volume at the bottom of the ZMW quickly. The terminal phosphate carrying the base specific fluorophore is removed with the addition of the next labeled nucleotide and the subsequent cleaving of the phosphate. This results in a real time film of bases being incorporated, light being emitted, next base being incorporated, and the next burst of light being emitted (Figure 30.29).
Figure 30.30 Native hemolysin pore incorporated in a membrane. Source: Zwolak, M. and Ventra, M.D. (2008) Rev. Modern Phys., 80, 141–165. With permission, Copyright 2008, American Physical Society.
Nanopore Sequencing A nanopore is a very small pore that will allow ions to flow through when a voltage is applied to it in a conducting fluid. The pattern of current generated by the flow of ions is characteristic based on the pore shape and changes also in a characteristic fashion when DNA is threaded through the pore. By threading DNA through a pore the sequence of nucleotides is called by the characteristic changes observed due to different nucleotides as they pass through (Figure 30.30). This technique promises several potential advantages over other previous techniques in that it can measure a DNA molecule of any length without being restricted to a set of run cycles and that it measures a single molecule without need for amplification. The difficulties with the technique lie in the small changes in current, which are difficult to measure accurately enough to reliably call the base sequence. Despite the practical difficulties in measuring DNA flowing through nanopores some companies have attempted to bring a nanopore sequencer to market. Oxford Nanopore Technologies is the nearest to commercially launching a product at the time of writing. Using protein nanopores placed in an artificial non-conducting chip surface, DNA is threaded through the pore by a progressive engine attached to one end of the DNA during library preparation, while the other end receives a hairpin sequence. As the DNA passes through the pore at its narrowest point the sequence causes a change in the current flowing and this is measured by the chip’s detectors under each pore. When one strand of the DNA strand has been sequenced the hairpin passes through the pore and the second strand is also sequenced, allowing the information from both strands to be used in base calling.
30 DNA Sequencing
Further Reading Ansorge, W., Voss H., and Zimmermann, J. (1996) DNA Sequencing Strategies, John Wiley & Sons, Inc., New York. Bentley, D.R. et al. (2008) Accurate whole human genome sequencing using reversible terminator chemistry. Nature, 456, 53–59. Blazej, R.G. et al. (2007) Inline injection microdevice for attomole-scale Sanger DNA sequencing. Anal. Chem., 79, 4499–4506. Branton, D. et al. (2008) Nanopore sequencing. Nat. Biotechnol., 26, 1146–1153. Brenner, S. (2000) Gene expression analysis by massively parallel signature sequencing (MPSS) on microbead arrays. Nat. Biotechnol., 18, 630–634. Clarke, J. et al. (2009) Continuous base identification for single-molecule nanopore DNA sequencing. Nat. Nanotechnol., 4, 265–270. Craig, D.W., Pearson, J.V., Szelinger, S. et al. (2008) Identification of genetic variants using bar-coded multiplex sequencing. Nat. Methods, 5, 887–893. Deamer, D. (2010) Nanopore analysis of nucleic acids bound to exonucleases and polymerases. Annu. Rev. Biophys., 39, 79–90 Drmanac, R. (2010) Human genome sequencing using unchained base reads on self-assembling DNA nanoarrays, Science, 327, 78–81. Fuller, C. et al. (2010) The challenges of sequencing by synthesis. Nat. Biotechnol., 27, 1013–1023. Hodges, E. et al. (2007) Genome-wide in situ exon capture for selective resequencing. Nat. Genet., 39, 1522–1527. Horner, D.S., Pavesi, G., Castrignano, T. et al. (2009) Bioinformatics approaches for genomics and post genomics applications of next-generation sequencing. Briefings Bioinformatics, 11, 181–197. Korlach, J. et al. (2008) Selective aluminum passivation for targeted immobilization of single DNA polymerase molecules in zero-mode waveguide nanostructures. Proc. Natl. Acad. Sci. U.S.A., 105, 1176–1181. Levene, M.J. et al. (2003) Zero-mode waveguides for single-molecule analysis at high concentrations. Science, 299, 682–686. Mardis, E.R. (2003) Next-generation DNA sequencing methods. Annu. Rev. Genomics Human Genet., 9, 387–402. Margulies, M. et al. (2005) Genome sequencing in microfabricated high-density picolitre reactors. Nature, 437, 376–380. Maxam, A. and Gilbert, W. (1977) A new method for sequencing DNA. Proc. Natl. Acad. Sci. U.S.A., 74, 560–564. Niedringhaus, T.P. et al. (2011) Landscape of next-generation sequencing technologies. Anal. Chem., 83, 4327–4341. Pop, M. and Salzberg, S.L. (2008) Bioinformatics challenges of new sequencing technology. Trends Genet., 24, 142–149. Ramanathan, A., Huff, E.J., Lamers, C.C. et al. (2004) An integrative approach for the optical sequencing of single DNA molecules. Anal. Biochem., 330, 227–241. Richardson, P. (2010) Special issue: next generation DNA sequencing. Genes, 385–387. Rothberg, J.M., Leamon, J.H. (2008) The development and impact of 454 sequencing. Nat. Biotechnol., 26, 1117–1124. Rothberg, J.M. et al. (2008) An integrated semiconductor device enabling non-optical genome sequencing. Nature, 475, 348–352. Sanger, F., Nicklen, S., Coulson, A.R. (1977) DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. U.S.A., 74, 5463–5467. Schadt, E.E. et al. (2010) A window into third-generation sequencing. Hum. Mol. Genet., 19, R227–R240. Schuster, S.C. et al. (2008) Method of the year, next-generation DNA sequencing. Functional genomics and medical applications. Nat. Methods, 5, 11–21. Shendure, J. and Ji, H. (2008) Next-generation DNA sequencing. Nat. Biotechnol., 26, 1135–1145. Stoddart, D. et al. (2009) Single-nucleotide discrimination in immobilized DNA oligonucleotides with a biological nanopore. Proc. Natl. Acad. Sci. U.S.A., 106, 7702–7707. Stoddart, D. et al. (2010) Nucleobase recognition in ssDNA at the central constriction of the alphahemolysin pore. Nano Lett., 10, 3633–3637. Sultan, M. et al. (2008) A global view of gene activity and alternative splicing by deep sequencing of the human transcriptome. Science, 321, 956–960. Tabor, S. and Richardson, C.C. (1990) DNA sequence analysis with a modified bacteriophage T7 DNA polymerase. J. Biol. Chem., 265, 8322–8328. Trepagnier, E.H. et al. (2007) Controlling DNA capture and propagation through artificial nanopores. Nano Lett., 7, 2824–2830. Venter, J.C., Adams, M.D., Meyers, E.W. et al. (2001) The sequence of the human genome. Science, 291, 1304–1351. Wang, H. and Branton, D. (2001) Nanopores with a spark for single-molecule detection. Nat. Biotechnol., 19, 622–623.
815
Analysis of Epigenetic Modifications Reinhard Dammann
31
Justus Liebig-University of Gießen, Institute of Genetics, Department of Biology and Chemistry, Heinrich-Buff-Ring 58-62, 35392 Gießen, Germany
With the complete sequencing of the human genome the number of genes that are involved in the complex interaction of the cellular development is now assessable. Basically, each genome consists of four bases: adenine, thymine, cytosine, and guanine, which, however, can be modified covalently. Since these modifications are inherited after DNA replication, the coded information of the genome changes considerably. The most important DNA modifications are methylation of cytosine at the C5-postion to 5-methylcytosine (5mC) and the methylation of adenine at the N6-position to N6-methyladenine (N6mA) (Figure 31.1). Methylation of adenine is mainly found in prokaryotes (Dam-methylation, (GN6mATC) and serves as a protective mechanism of its own DNA against sequence specific restriction enzymes. Cytosine methylation is frequently found in bacteria (Dcm-methylation, C5mCWGG) and is also detected as a modification in plants, invertebrates, and vertebrates (i.e., CpG-methylation, 5m CG). In mammals cytosines are mainly methylated, when the base is followed by guanine, which is designated as dinucleotide 5mCpG. This methylation is achieved in vivo by DNA methyltransferases (DNMTs). Interestingly, an additional 5mCpA- and 5hmC-methylation has been found in human stem cells. In addition, in the brain, 5-hydroxymethylcytosine (5hmC) has been detected and results from the oxidation of 5mC by TET enzymes. TET further oxidizes 5hm C to 5-formylcytosine (5fC) and 5-carboxylcytosine (5caC), which can be converted into C by base excision repair. The function of 5hm C is not yet completely understood and may either represent an intermediate modification in the DNA demethylation pathway or a specific epigenetic mark. In human somatic cells 5m C base makes up only 1% of all DNA bases, but 70–80% of the CpG are methylated. In the human genome the dinucleotide CpG is underrepresented, but is often found in GC-rich sequences, so-called CpG islands. Nearly 60% of all human genes harbor a CpG island in their promoter region. Normally, these CpG-island promoters are unmethylated. Methylation of CpG in the promoter region has an important influence on the regulation of gene expression and leads to epigenetic inactivation of the affected gene. Epigenetic control plays an important role in inheritance of gene activity; since DNA methylation modifies the information of the genome without changing the primary sequence pattern of the DNA. Cytosine methylation directly influences the gene activity by the binding of regulatory proteins (e.g., methyl binding domain proteins) to the methylated sequence and indirectly influences gene expression by inactivation of the chromatin structure and through altered histone modifications. Thus, the fifth base 5m C acts as a reversible epigenetic switch and is essentially involved in the inheritance of the gene activity. Since hypermethylation of regulatory sequences results in inactivation of gene expression, these epigenetic changes are considered to be an important mechanism in the inactivation of tumor suppressor genes. DNAmethylation not only plays a fundamental role in carcinogenesis but also in cellular development and aging. Furthermore, DNA-methylation determines allele specific expression of paternal and maternal inherited genes, a mechanism that is termed imprinting. DNA methylation is also involved in the processes of dosage compensation by X-chromosome-inactivation. Moreover chromatin structure and modifications of nucleosomes are also important for the epigenetic gene Bioanalytics: Analytical Methods and Concepts in Biochemistry and Molecular Biology, First Edition. Edited by Friedrich Lottspeich and Joachim Engels. 2018 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2018 by Wiley-VCH Verlag GmbH & Co. KGaA.
Figure 31.1 5-Methylcytosine, 5-hydroxymethylcytosine, and N6-methyladenine.
818
Part IV: Nucleic Acid Analytics
regulation. While active genes display an open chromatin structure (euchromatin) with acetylated histones, silenced genes show a closed chromatin (heterochromatin) with deacetylated histone. Additionally, nucleosomes are altered by methylation, phosphorylation, and other modifications of histones. This chapter describes several methods that can be used to analyze epigenetic modifications (DNA methylation and other chromatin changes).
31.1 Overview of the Methods to Detect DNA-Modifications There are six main techniques for analyzing DNA-methylation:
Chemical modifications of the unmethylated bases with: 1. bisulfite.
Protein specific analyses of DNA-sequences with:
Real Time PCR, (quantitative) PCR, Section 29.2.5
2. methylation sensitive restriction enzymes; 3. 5m C-binding domain (MBD) proteins; 4. 5m C, 5hm C, 5f C or 5ca C -specific antibodies. Analysis of the configuration of the bases of the complete DNA with: 5. DNA hydrolysis; 6. nearest neighbor-analysis (Table 31.1).
The different reactivities of modified cytosine and cytosine for bisulfite induced deamination to uracil is used to determine the methylation status of the DNA. With protein specific methods the different activity of restriction enzymes or binding of mC-binding proteins (MBD or antibodies) is used to analyze the methylation status of the DNA. Certain methylation sensitive restriction nucleases do not cut their recognition site if the DNA is methylated, whereas others only cut methylated sites in the DNA or are insensitive against methylation. These enzymes allow analysis of the methylation status of the respective restriction cutting sites. 5m C-Binding domain proteins (MBDs) and mC-antibodies (5m C or 5hm C) are utilized to precipitate the modified DNA and to quantify its methylation status by real time PCR. Moreover, antibodies that recognize 5f C or 5ca C are available. A further possibility is in analyzing the composition of the DNA bases. The genomic DNA is completely hydrolyzed and different modifications of a
Table 31.1 Overview of important methods for analysis of the DNA-modifications. Method Bisulfite modification
Concept 5m
Chemical resistance of C and C to deamination to uracil by bisulfite (5f C, 5ca C, and C are deaminated to U) 5hm
Methylation sensitive restriction enzymes
Different accessibility of methylated DNA for restriction enzymes
5m
Precipitation of DNA or chromatin with MBD proteins
C-Binding domain (MBD) protein-specific analyses (i.e., MIRA) Antibody specific for modified DNA (i.e., MeDIP)
Precipitation of DNA with 5m C-, C-, 5f C-, or 5ca C-antibodies
Identification
Scope
– – – –
Sequencing restriction analysis (COBRA) methylation specific PCR (MSP) TET assisted bisulfite sequencing (TAB-Seq) – Southern blotting – qPCR
All modified Cs in a DNA fragment can be analyzed as well in the upper as in the lower strand
– – – – –
The methylation status of a specific DNA fragment or region can be analyzed
5hm
DNA hydrolysis
Complete analysis of the different base modifications
Nearest neighbor-analysis
Analysis of the different DNA modifications in connection to the 3´ -base
– – – – – – – – –
Pulldown and sequencing qPCR microarray immunofluorescence Immunoprecipitation and sequencing qPCR microarray immunofluorescence HPLC chromatography mass spectrometry HPLC chromatography mass spectrometry
Only DNA-modifications within restriction site can be analyzed
The methylation status of a specific DNA fragment or region can be analyzed
Different modifications of the genomic DNA can be analyzed The amount of distinct modifications can be analyzed
31
Analysis of Epigenetic Modifications
819
single base are analyzed. With the nearest neighbor-analysis dinucleotides are labeled and subsequently their composition is dissected. However, DNA hydrolysis and the nearest neighbor-analysis are not able to reveal the exact sequence context of the modified bases. These methods are explained next in more detail.
31.2 Methylation Analysis with the Bisulfite Method The easiest and most effective way to analyze DNA methylation is by the bisulfite technique. This method has a very high resolution and it is possible to analyze the methylation status of the whole DNA-population at a specific sequence or to dissect the methylation pattern of single DNA fragments. This technique was developed in 1974, but it was only in 1992 that it became popular after further development by Frommer and coworkers. Meanwhile it has come into broad usage because of its high resolution and reliability. The principle of this method is the reaction of bisulfite (HSO3 ) with DNA, which converts cytosine (5f C, 5ca C, and C) into uracil. The C6 position of an accessible cytosine is sulfonated by a high bisulfite concentration (3.0 M) and acidic conditions (pH 5.0). In this process the amino group at the C4 position is hydrolyzed and uracil is generated (Figure 31.2). The particularity of this reaction is that methylated cytosines (5m C and 5hm C) are not converted and remain as cytosines. Thus, unmethylated cytosines (deaminated to U) and methylcytosines (remain C) can be distinguished. (Figure 31.3). However, this method cannot distinguish between 5m C and 5hm C. In a PCR reaction the bisulfite-treated DNA is amplified and analyzed by using primers, which are complementary to the deaminated DNA sequence. This PCR results in the substitution of uracil by thymine (Figure 31.2). Through the PCR amplification the bisulfite method is highly sensitive and only a small amount of genomic DNA (50 ng) is needed. It is also possible to analyze the DNA methylation in less than 100 cells. Before the bisulfite treatment it can be of advantage if the cells or the DNA are embedded in agarose so that the loss of DNA can be minimized. A problematic issue of the bisulfite treatment is the incomplete conversion of Cs into Ts. The partial denaturation of the DNA during the bisulfite treatment may lead to incomplete deamination of unmethylated Cs to Ts and this could lead to misinterpretation of methylated Cs. To overcome this problem, different modifications of the bisulfite method have been established. One possibility is to digest the DNA in short fragments with a restriction enzyme before its denaturation. (However, no restriction site should be in the investigated DNA region.) During the bisulfite treatment, by a repetitive denaturing in a thermocycler, the deamination of the DNA can also be improved. However, by too intensive treatment, the DNA could be degraded into small fragments. It is relatively simple to verify the complete bisulfite conversion of the DNA by PCR amplification and sequencing. The presence of methylated C in a non-CpG context is mostly likely an artifact of an incomplete bisulfite reaction and this should be verified with an alternative method. In practice the bisulfite technique has proved to be a very efficient and reliable procedure to analyze DNA methylation.
Figure 31.2 Sodium bisulfite catalyzes the deamination of unmethylated cytosine or to uracil.
31.2.1 Amplification and Sequencing of Bisulfite-Treated DNA The bisulfite method allows the gene specific analysis of methylation levels of a cell population and of the methylation pattern of single DNA molecules. This depends on whether PCR products are directly sequenced (i.e., pyrosequencing) or if they are first subcloned and then sequenced. Since the two DNA strands of the DNA are no longer complementary after bisulfite conversion, it is possible to investigate strand specific methylation with separate primer pairs (Figure 31.3). The PCR amplification leads to a conversion of U (unmethylated C) into T, and from methylated C into C. On the complementary strand G (opposite to an originally unmethylated C) is then converted into A. It is very easy to mimic the bisulfite conversion in silico with a word processing program to generate the deaminated sequence necessary to design primers for the amplification of the bisulfite-converted DNA. The following aspects should be considered when designing the primer: 1. To exclude the amplification of not bisulfite-modified DNA, primer pairs should include some deaminated C (T in forward-primer and respective A in the reverse primer).
Sequencing by Synthesis, Classic Pyrosequencing, Section 30.2.1
820
Part IV: Nucleic Acid Analytics
Figure 31.3 Principle of the methylation analysis with the bisulfite method. DNA is denatured and treated with bisulfite. With this method the methylated cytosines (5m C and 5hm C) are conserved while the unmethylated cytosines (C, 5f C and 5ca C) are deaminated to uracil (U) and appear as thymine (T) after PCR amplification. Note that after the bisulfite treatment the DNA strands are not complementary and can be amplified with different primer pairs (A, B or C, D).
2. The primer should not include CpGs in their original DNA sequence, to avoid a specific amplification of methylated or unmethylated DNA (see also methylation specific PCR). If this is unavoidable, it is possible to insert Y (pyrimidine: C or T) instead of C and in the complementary strand an R (purine: G or A) instead of G. 3. Since primers do not contain Cs (in the complementary strand no Gs) their annealing temperature is often low and therefore primers with a length of 25–30 nt should be used. 4. For pyrosequencing a biotinylated primer and a sequence primer are necessary. 5. The PCR product should not be longer than 500 bp since longer DNA fragments will be amplified at low rate. This is caused by the fact that the bisulfite treatment degrades DNA and long intact DNA molecules are not available for the amplification. 6. The methylation pattern of individually complementary DNA molecules can be analyzed. Methylation in the context of a double strand DNA can be analyzed by the ligation of a hairpin linker prior to the bisulfite treatment. Semi-nested or Nested PCR, Section 29.3.1 Chemical Cleavage According to Maxam Gilbert, Section 30.1.3
Normally, 50 ng bisulfite-treated DNA is fully sufficient for a PCR reaction. If only a small amount of DNA is available, the sensitivity of the detection can be increased by a semi-nested or nested PCR. The DNA methylation is identified either by direct sequencing of PCR products or by sequencing of many individual DNA molecules after cloning in a vector system. One advantage of the bisulfite method is that the DNA methylation can be detected by conventional sequencing (didesoxy or Maxam–Gilbert method) or pyrosequencing (Figure 31.4). Methylated C as well as unmethylated C can be detected with the DNA sequencing method: m C occurs as C and unmethylated C as T (Figure 31.4). Hence, the methylation of all CpG in a DNA fragment can be analyzed. From the pyrosequencing additional quantitative conclusions can be drawn about the level of methylation of the PCR products and the presence of incomplete conversions of unmethylated cytosines (see bisulfite control in Figure 31.4). Alternatively, sequencing from short DNA fragments can be accomplished by mass spectrometry. During the PCR reaction a preferential amplification (bias) of unmethylated or methylated DNA may occur. In samples representing a certain mixture of methylated DNA and unmethylated DNA, this PCR bias can be investigated. To generate methylated standards DNA is methylated by a CpG-methylase (e.g., SssI-methylase) before bisulfite conversion and if necessary by cloning of PCR products in Escherichia coli.
31.2.2 Restriction Analysis after Bisulfite PCR In Vitro Restriction Analysis, Section 27.1.4
For further analysis of bisulfite-modified DNA different sensitive detection methods have been developed. One of them is the restriction analysis of PCR products of bisulfite-treated DNA. This method was termed combined bisulfite restriction analysis (CoBRA). The principle is that methylated C remains C after bisulfite treatment in a CpG sequence while unmethylated C is converted into T. If the C is in a palindromic sequence (i.e., 5´ -TCGA) the methylation can be investigated with a restriction enzyme. The restriction enzyme TaqI cuts the recognition sequence TCGA, so that the “methylated” PCR products are digested (Figures 31.5 and 31.6). On the other hand, if this restriction site is missing and the product is not cut, the PCR product originates from an
31
Analysis of Epigenetic Modifications
821
Figure 31.4 Examples of bisulfite methylation analysis after conventional sequencing (a) or pyrosequencing (b) of PCR products of the RASSF1Apromoter. (a) After bisulfite treatment and PCR amplification all unmethylated cytosines (C) are replaced by thymines (T). Methylated Cs in CpG context are resistant against this conversion. (b) Pyrograms of three sequencing reactions with the sequence YGTTYGGTTYGYGTTTGTTA and different levels of methylation of the PCR products. Double height of the signal indicates the incorporation of two nucleotides.
Figure 31.5 Principle of the restriction analysis after bisulfite PCR. The DNA is denatured, treated with bisulfite, and amplified by PCR. While unmethylated cytosines are modified to thymines (T) the methylated cytosines (m C) remain C. For example, in the “methylated” DNA the restriction cutting site for TaqI (5´ -TCGA) and in the “unmethylated” DNA the recognition site for TasI (5´ -AATT) emerges.
822
Part IV: Nucleic Acid Analytics
Figure 31.6 Example of a restriction analysis of the RASSF1A-promoter after bisulfite-PCR. The 205 bp long “unmethylated” PCR product is not cut by Taql. The “methylated” PCR product is digested in 171 bp (partial methylated), 90 bp, and 81 bp fragments. (The 34 bp long fragment is not visualized.) A 100 bp marker (M) serves as length standard for the 2% agarose gel.
Table 31.2 Enzymes for the restriction analysis of bisulfite modified DNA of PCR products. Restriction enzyme
Recognition sequence
For methylated DNA (CpG): TaqI
T/CGA
BstUI
CG/CG
MaeII, HpyCH4IV or TaiI
A/CGT
BsiWI
C/GTACG
PvuI
C/GATCG
ClaI
AT/CGAT
MluI
A/CGCGT
For unmethylated DNA (TpG):
unmethylated DNA (Figures 31.5 and 31.6). For COBRA all enzymes with CG in the recognition sequence can be used as diagnostic restriction enzyme (Table 31.2) and also those with only one C at the 3´ -end (i.e., EcoRI: GAATTC). However, the most common enzymes are the four base pair cutter Taql, BstUI, and MaeII, since their recognition sites are more abundant (Table 31.2). Interestingly, this assay can also be used to verify the complete conversion of C into T in the analyzed DNA. Thereby, a new restriction site is created through the bisulfite conversion. For example, a cutting site for Taql (TCGA) is only created from the original sequence 5´ -CCGA when the unmethylated 5´ -C was modified to T, but the second C is not converted in a methylated CpG context (Figure 31.5). The same principle can also be used to investigate an unmethylated C in a CpG sequence. If a C at the 3´ -end of a putative recognition site is modified to T, a new restriction cutting site is only created in the deaminated “unmethylated” DNA (Figure 31.5). For example, if the sequence AATCG is modified to AATTG, a restriction site for the enzyme TasI (AATT) will be found in a PCR product amplified from unmethylated DNA (Figure 31.5). One main limitation of COBRA is that only the analysis of the DNA methylation at restriction enzyme recognition sites is possible and therefore not all CpG in a DNA molecule can be investigated.
31.2.3 Methylation Specific PCR
TasI, Tsp509I
/AATT
AseI, VspI
AT/TAAT
SspI
AAT/ATT
Figure 31.7 Principle of the methylation specific PCR. DNA is denatured and treated with bisulfite. In this process methylated cytosines (m C) are kept while unmethylated Cs are deaminated to uracils (Us). The “methylated” DNA is amplified with methylation specific primer (MF and MR) and the “unmethylated” DNA with unmethylation specific primer (UF and UR).
In 1996 Herman and coworkers developed methylation specific PCR (MSP) to increase the sensitivity in detecting methylated DNA after bisulfite treatment. MSP analysis is very sensitive and can detect up to 0.1% methylated (or unmethylated) DNA sequences per sample. The MSP method uses different primer pairs for the amplification of methylated and unmethylated DNA after the bisulfite modification (Figure 31.7). These primers are located at specific CpGs and
31
Analysis of Epigenetic Modifications
their amplification rate reveals the methylation status of these Cs. For the amplification of methylated DNA a methylation specific primer pair with Cs in the forward primer and Gs in the reverse primer at the investigated CpG sites is utilized (Figure 31.7). Therefore, these primers only bind and amplify the previously methylated bisulfite modified DNA. In contrast, for the amplification of the unmethylated bisulfite-treated DNA an unmethylated specific primer pair is used, where the C in the forward primer is replaced by T (in the reverse primer G is replaced by A). These primers only bind and amplify the previously unmethylated bisulfite modified DNA. After gel electrophoresis the methylation status is directly detected by the amount of methylated and unmethylated PCR products (Figure 31.8). To increase the specificity of the amplification rate of “methylated” or “unmethylated” bisulfite-treated DNA by MSP, several aspects should be considered during the primer design. Moreover it is important to utilize DNA controls with known methylation status (i.e. methylated, unmethylated, and unconverted negative controls). The following aspects should be considered for the primer design:
The methylation specific forward primer should harbor a C (reverse-primer a G) at the 3´ -end. The unmethylation specific forward primer should have a T at the 3´ -end (reverse-primer an A).
The primer should have three to four CpG or TpG to ensure a specific amplification of the methylated or unmethylated DNA, respectively.
To increase the specificity of the primer pairs for bisulfite modified DNA the primer should harbor several Ts for “deaminated Cs” in the forward-primer and As in the reverse-primer.
The advantage of the MSP method is based on its high sensitivity and the simplicity of the assay. MSP has been combined with the real time-detection (real time MSP). The MethyLight method uses during the PCR methylation- (or unmethylation-) specific TaqMan probes. Therefore, this technique is utilized to quantify the methylation levels of the analyzed CpGs.
31.3 DNA Analysis with Methylation Specific Restriction Enzymes Some restriction endonucleases do not cut the DNA when their recognition sites are methylated, while other restriction enzymes are insensitive to such DNA methylation. For a third group of enzymes methylation of their recognition site is necessary for the cutting. These restriction endonucleases are used to identify methylated Cs or methylated As in their recognition sites. Often the enzymes HpaIl und Mspl are used for the analysis of the methylation status of cytosines at the CpG dinucleotides (Figure 31.9). Both enzymes recognize the sequence 5’CCGG, but only HpaIl is able to cut this sequence, when the second cytosine is unmethylated. In contrast the methylation insensitive isoschizomer Mspl is utilized to cut methylated and unmethylated DNA. Mspl cuts the methylated CmCGG-sequence as well as the unmethylated CCGG-sequence. The different methylation specific restriction fragments of HpaIl und Mspl are analyzed by Southern blot and PCR (Figure 31.9). For Southern blot analysis approximately 10 μg of genomic DNA is necessary. This technique allows a quantitative estimation of the methylated rate at the specific cutting site. Southern blot and hybridization can be conducted by a standard protocol.
823
Figure 31.8 Example of a methylation specific PCR of the RASSF1A-promoter. After bisulfite treatment a previously unmethylated DNA is amplified with the unmethylation specific primer pair (u) and a 105 bp PCR product is detectable after gel electrophoresis. On the other hand, a 93 bp PCR product is obtained by amplification of previously methylated DNA with the methylation specific primer pair (m). In a partial methylated bisulfite-treated DNA, PCR products for both primer pairs are detectable. A 100 bp marker (M) served as length standard for the 2% agarose gel.
824
Part IV: Nucleic Acid Analytics
Figure 31.9 DNA-methylation analysis with methylation specific restriction enzymes. The methylation sensitive enzyme Hpall cuts only unmethylated DNA. The methylation specific inhibition is analyzed by Southern blot or PCR and controlled with the insensitive enzyme Mspl, which cuts both the methylated and the unmethylated DNA. For the Southern blot analysis the DNA is digested with an additional insensitive restriction enzyme (R).
Isoschizomers are distinct restriction endonucleases with identical recognition sites, which generate similar or different cleavage products.
For the PCR analysis after a methylation specific restriction, little genomic DNA (50 ng) is needed and this allows detection of low levels of methylated DNA. However, this technique is prone to restriction artifacts (e.g., uncomplete digestion, see below) and therefore should be well controlled. For this analysis two primers flanking the restriction site should be designed. PCR products are analyzed by gel electrophoresis and compared to specific controls. Only in the methylated sample is a fragment detected after a restriction digest with the methylation sensitive enzyme HpaIl and PCR (Figure 31.9). As control, no fragment should be obtained after digestion with the insensitive Mspl and PCR amplification. As further control, the DNA can be digested with a restriction enzyme that cuts outside of the analyzed fragment. After this restriction a PCR product should be detected. For the methylation analysis of m CpG a number of methylation sensitive restriction enzymes can be utilized. Several of these enzymes are listed in Table 31.3. With these enzymes not only the methylation of known cutting sites can be investigated, but also novel methylated DNA Table 31.3 Methylation sensitive enzymes and insensitive isoschizomers for restriction analysis. Sensitive enzyme (insensitive isoschizomer)
Methylated recognition sequence (isoschizomer)
m
CpG-methylation
HpaII (MspI)
/CmCGG (C/mCGG)
BstUI
m
NotI
GmC/GGCmCGC
AscI
GG/mCGmCGCC
SmaI (XmaI)
CCmC/GGG (C/CmCGGG)
CG/mCG
Dcm-methylation – CmCWGG EcoRII (BstNI)
/CmCWGG (CmC/WGG)
SfoI (NarI)
GGC/GCmCWGG (GG/CGCmCWGG) G/GTACmCWGG (GG/TACmCWGG)
Acc65I (KpnI) m
Dam-methylation – G ATC MboI (Sau3AI)
/GmATC (/GmATC)
DpnII (Sau3AI)
/GmATC (/GmATC)
BclI
T/GmATCA
AlwI
GGmATC(4/5)
31
Analysis of Epigenetic Modifications
regions can be isolated that are resistant to digestion by methylation sensitive restriction enzymes. Methylation-sensitive arbitrarily primed PCR, differential methylation hybridization, and restriction land-mark genome scanning are examples of methods used to identify potentially methylated DNA regions in the genome (see Further Reading). For the methylation analysis with restriction enzymes a certain caution is necessary: The incomplete digestion of unmethylated DNA with a methylation sensitive enzyme can be mistaken as a partial methylated restriction site. Since the restriction of the genomic DNA can be inhibited by contamination of the sample with cell membranes, carbohydrates, or lipopolysaccharides or by utilizing wrong conditions during the reaction (e.g., salt concentration or pH), the purity of the genomic DNA is essential. In Escherichia coli cytosine is only methylated within the sequence 5´ -CmCWGG (W = A or T), which is called Dcm-methylation. The status of methylation of this sequence can be analyzed with the isoschizomer EcoRII I-BstNI (Table 31.3). While EcoRII does not cut the methylated sequence (CmCWGG), the isoschizomer BstNI cuts both the methylated and the unmethylated sequence. Some enzymes are sensitive to Dcm methylation; their consensus sequences are listed in Table 31.3. In prokaryotes methylation of adenine can be found in the sequence 5´ -GmATC, which is termed Dam-methylation. To analyze this adenine methylation the isoschizomers Mbol/Sau3A can be used (Table 31.3). Both enzymes recognize the sequence GATC. Mbol is sensitive for m A and the methylated sequence is not cut. In contrast Sau3A is insensitive for this methylation and cuts the sequence GmATC. The methylated GmATC-sequences can also be detected with Dpnl. Interestingly, the enzyme Dpnl only cuts the DNA when adenines on both strands of its recognition site are methylated. It should be considered that because of Dcm- and Dam-methylation the cloning of certain DNA fragments from Escherichia coli is sometimes problematical when certain restriction sites are methylated – especially when the methylation motif is encoded through the flanking sequence. The restriction enzymes Clal (ATCGAT) and Xbal (TCTAGA) do not cut the DNA when an adenine has been modified by an overlapping Dam methylation (ATCGmATC respective TCTAGmATC). The enzyme Stul (AGGCCT) can be inhibited by an overlapping Dcm-methylation (AGGCmCTGG). Thus it is important to consider flanking bases when a specific recognition site is not cut by the appropriated restriction enzyme. This problem can be avoided by utilizing a methylation insensitive isoschizomer or an Escherichia coli strain that is negative for Dcm- or Dam-methylation.
31.4 Methylation Analysis by Methylcytosine-Binding Proteins This method utilized specific proteins that exhibit high affinity binding to methylated DNA. Different proteins have been isolated from mammalian cells, which can bind methylated cytosine and are involved in the inactivation of gene expression and changes of chromatin state. These proteins are termed methyl-CpG-binding proteins (MeCPs) and possess a methyl binding domain (MBD). Several different proteins (i.e., MeCP2, MBD1, MBD2, and MBD3) have been characterized and isolated. These MBD proteins are used for different methods to analyze the methylation status of specific genomic regions or to identify novel differentially methylated regions (Figure 31.10). For example, tagged MBD proteins are immobilized to nickel agarose beads and used for enrichment of methylated DNA. Genomic DNA is fragmented by sonication or restriction digestion and purified on a column. At a specific salt concentration the methylated DNA is bound to the column, and the unmethylated DNA is eluted. Afterwards the methylated DNA is eluted with high salt concentration and analyzed. Subsequently the methylated DNA can be quantified by real time-PCR, hybridized on microarrays, or analyzed by deep sequencing. The methylated-CpG island recovery assay (MIRA) uses the ability of MBD3L binding to MBD2 and thereby increases the affinity of MBD2 to methylated DNA. In MIRA recombinant GST-tagged MBD2b protein and His-marked MBD3L1 protein is expressed in bacteria and is purified with glutathione-Sepharose 4B or the respective Ni-NTA agarose beads. The genomic DNA is isolated and cut with the enzyme MseI. MseI recognizes the TTAA cutting site but cuts rarely in CpG islands; 1 μg purified GST-MBD2b and 1 μg His-MBD3L are pre-incubated together with 500 ng
825
826
Part IV: Nucleic Acid Analytics
Figure 31.10 Analysis of the DNA-methylation with methyl-binding proteins. The genomic DNA is fragmented by ultrasound or restriction digestion in small fragments and bound specifically to methylcytosine binding proteins (MBD) or to methylcytosine antibodies, precipitated, and purified. The methylated DNA can be quantified by PCR, sequenced, or analyzed on microarrays.
unmethylated DNA (e.g., bacterial DNA) and then incubated for some hours with approximately 500 ng of fragmented genomic DNA. During this step the methylated DNA is bound to MBD2b/ MBD3L. The methylated DNA is precipitated with immobilized glutathione paramagnetic particles and the beads are washed. Subsequently, the methylated DNA is eluted from the beads and analyzed. For example, after linker ligation the enriched DNA can be amplified, labeled, and hybridized on microarrays. As an example application, tumor specific DNA methylation can be revealed by fluorescence marking of DNA from tumor tissue with Cy5 (red) and compared to DNA from normal tissue, which was marked with Cy3 (green) (Figure 31.11). With the chromatin immunoprecipitation (ChIP) method endogenous MBD proteins are linked with formaldehyde to genomic DNA in vivo. Subsequently, the protein–DNA complexes are purified and precipitated with anti-MBD-antibody (see also Section 31.7). Again the methylated DNA can be analyzed by PCR, next generation sequencing (NGS), or microarray (ChIP on chip). These methods allow the identification of methylated DNA-sequences in a certain genome.
31.5 Methylation Analysis by Methylcytosine-Specific Antibodies
Figure 31.11 Analysis of differential methylation levels of normal and tumor tissue. Genomic DNA is fragmented and purified with MBD proteins (MIRA) or methylcytosine specific antibodies (MeDIP). The methylated DNA is marked by fluorescence and analyzed on microarrays.
A further method for detecting and quantifying modified DNA is based on the development of antibodies binding specifically to 5m C, 5hm C, 5f C, or 5ca C (Figure 31.10). These antibodies interact with single strand modified DNA. For example, methylated DNA can be precipitated and analyzed. The sensitivity of the anti-5-methylcytosine antibody is very high. The monoclonal mouse 5m C-antibody interacts specifically with methylated DNA in 10 ng genomic DNA that contains only 3% 5m C. With these antibodies methylated DNA can be precipitated and analyzed by real time-PCR, deep sequencing, or microarray-technology (Figures 31.10 and 31.11). This method is also called the methylated DNA immune-precipitation (MeDIP) method. During the MeDIP procedure 4 μg of genomic DNA is cut into 300–1000 bp fragments by sonication and denatured by heating. An aliquot of the sheared DNA can be used as an input control. Subsequently, the DNA is incubated with 10 μl monoclonal mouse 5m C antibody at 4 °C for several hours. The methylated DNA is precipitated with anti-mouse IgG beads,
31
Analysis of Epigenetic Modifications
827
purified, and eluted through a proteinase K digestion. The enriched methylated DNA can now be investigated by real-time PCR, sequencing, or microarray analysis. The 5m C, 5hm C, 5f C, or 5ca C antibodies together with immunofluorescence can be utilized to investigate chromosomal segments with a high occurrence of modified cytosines.
31.6 Methylation Analysis by DNA Hydrolysis and Nearest Neighbor-Assays The following methods allow us to analyze the frequency of modified bases in the DNA and for the nearest neighbor-assays their context as dinucleotides can be detected (e.g., 5m CpN or 5hm CpN). However, with these methods, it is not possible to locate the modified base in the genomic sequence. Since the DNA of contaminated organisms can influence the analysis results, the investigated cells should be free of foreign DNA from viruses, mycoplasms, or other endoparasites. With the DNA hydrolysis method the DNA is completely hydrolyzed. Subsequently, the base composition is fractionized and the modified bases are quantified. Since products of a chemical hydrolysis are rather complex, the enzymatic hydrolysis is the preferred method. Spleen phosphodiesterase or Micrococcus-nuclease produces 3´ -phosphorylated mono-nucleosides. Pancreatic DNase I or snake venom phosphodiesterase produces 5´ -phosphorylated mono-nucleosides. Afterwards, 3´ - or 5´ -phosphates are eliminated with an alkaline phosphatase and the hydrolysis products are identified by different techniques such as high performance liquid chromatography (HPLC), mass spectrometry, or capillary electrophoresis (CE). With HPLC it is possible to detect in 2.5 μg DNA from 0.04 up to 0.005% 5m C. With the nearest neighbor analysis the frequency of methylated bases in context of the 3´ neighbor can be dissected. For this purpose the purified genomic DNA is marked with one of four [α-32P]dNTP by nick translation at accidental strand breaks that can be generated by DNase I (Figure 31.12). Then the DNA is digested with Micrococcus nuclease and calf thymus phosphodiesterase (exonuclease) to 3´ -dNMP, with the radioactive 32 P of the 5´ -position of the marked nucleotide located now at the 3´ -position of the 5´ -base. By adsorption chromatography or HPLC the 3´ -marked dNMP are separated and then compared to different modified standards. For example, one can analyze how often 5m C or N6m A is found with respect to one out of four different 5´ -bases. Since this method was rather complex, it has been refined. DNA is digested with a restriction enzyme and then the cutting site is marked with a specific [α-32P]-dNTP and Klenow. To analyze methylated CpG, the DNA can be digested with Mbol (/GATC) and the cutting site can be labeled with [α-32P]-dGTP and Klenow (Figure 31.12). With nucleases the DNA is digested to 3´ -marked-dNMP and the modified base can be quantified with chromatography. The intensity of labeled 5m dCp, dCp, dTp, dGp, and dAp indicates the quantity of 5m dCpG, dCpG, dTpG, dGpG, and dApG, respectively, at MboI cutting sites. Alternatively, the DNA can be digested with Fokl (GGATGN9–13) and marked with one of the four [α-32P]dNTPs.
Figure 31.12 Principle of the nearest neighbor-analysis. (a) α-32P-dGTP is inserted at the 3´ -end of the methylated adenine and analyzed by chromatography. (b) The DNA is digested with Mbol, the cutting site is labeled with α-32PdGTP and Klenow. Subsequently methylated Cs are quantified.
828
Part IV: Nucleic Acid Analytics
Chromatin Immunoprcipitation (ChIP), Chapter 32.4.7
The advantage of Fokl is that DNA modifications can be investigated independently from their recognition site but in the context all four 3´ -downstream bases (NpA, NpG, NpT, and NpG). With the enzyme MvoI (CC/WGG) only the methylation of CpA and CpT can be analyzed. In principle, several enzymes could be used; however, it should be considered that the selected restriction enzyme is not sensitive to DNA modifications. The nearest neighboranalysis is a preferred technique to identify new DNA modifications; however, with this method these modifications cannot be localized within the genome.
31.7 Analysis of Epigenetic Modifications of Chromatin Chromatin modifications and the specific binding of proteins (e.g., transcription factors) to DNA are preferably investigated with chromatin immunoprecipitation (ChIP) (Figure 31.13). This method is based on the crosslinking of nucleic acids with proteins or by crosslinking from proteins among each other with formaldehyde. The crosslinked protein–nucleic acid complexes are precipitated with a specific antibody against the protein or the chromatin modification. The precipitated complexes are washed stringently to elute unspecific bound chromatin. By heating the crosslinks are made reversible and the proteins are digested. The purified DNA can be investigated with PCR (real-time PCR), deep sequencing (ChIP Seq) or microarray (ChIP-onchip). One important requirement for this method is that the antibody for the analyzed modification or protein works also in the crosslinked chromatin condition. For ChIP analysis, DNA binding proteins are crosslinked in vivo by 1% formaldehyde. During this procedure proteins are linked covalently among each other as well as with the genomic DNA. This crosslinking can be carried out in cell culture for 10 min at 37 °C. Afterwards the cells are washed and harvested. The cells are lysed in a SDS buffer and the chromatin is sheared by sonication into fragments of approximately 500 bp. Subsequently, the chromatin is diluted in a binding buffer to 200 up to 300 μg μl 1 protein and an aliquot is retained as an input-control. To reduce the unspecific binding of antibodies the chromatin sample is pre-incubated with protein A agarose. Then the chromatin is incubated at 4 °C overnight with a specific antibody against the chromatin modification or factor (Figure 31.13). As a negative control in parallel an unspecific mouse IgG antibody can be used or as positive control a histone H3 antibody is suitable. The bound chromatin is precipitated by addition of protein A agarose for 1 h at 4 °C and centrifuged. Subsequently, all samples are intensely washed. Finally, crosslinking of samples and the input-control are reversed by the addition of 5 M NaCl and incubation at 65 °C overnight. Proteins are digested by proteinase K and DNA is purified. The enriched DNA can now be analyzed by real-time PCR, ChIP-on-chip (chromatin immunoprecipitation on microarray) or ChIP-Seq (chromatin immune-precipitation with ultradeep sequencing).
31.8 Chromosome Interaction Analyses Figure 31.13 Analysis of DNA-associated protein modifications or factors with chromatin immunoprecipitation (ChIP). Chromatin is crosslinked with formaldehyde and fragmented by sonication in approximately 500 bp. The chromatin is incubated with specific antibodies, precipitated, and purified. The purified DNA can be quantified by PCR, sequenced (ChIP-Seq), or analyzed on microarrays (ChIP on Chip).
During the interphase chromosomes are uncoiled and organized in different topologically associating domains (TADs) and chromosomal territories. In this configuration chromatin interacts intrachromosomally as well as interchromosomally. For example, there are intrachromosomal interactions between enhancers and promoters. This chromosomal topology can be investigated with the chromosome conformation capture technique (CCC or 3C analysis) (Figure 31.14). During the 3C analysis chromatin is crosslinked, cut, and ligated. During this treatment a preferential ligation of associated and crosslinked DNA fragments occurs and these specific interactions can be analyzed by quantitative PCR. For the 3C method the chromatin is incubated with formaldehyde to covalently crosslink protein–protein and protein–DNA interactions. This treatment can be performed in cell culture with formaldehyde. The cells are lysed and the crosslinked chromatin is cut with a restriction enzyme. Subsequently, the restriction enzyme is deactivated and the chromatin is ligated by T4 DNA ligase. During this step all DNA fragments that are fixed by crosslinking are preferentially
31
Analysis of Epigenetic Modifications
829
ligated (Figure 31.14). Afterwards crosslinks are reversed by an incubation at 65 °C overnight and the DNA can be purified and precipitated. The amount of ligation products can be analyzed by quantitative real-time PCR. The more often ligation products are detected the higher is the probability that DNA fragments interact in vivo. With the help of circular chromosome conformation capture (4C- or circular 3C-analysis) new chromosomal interactions can be detected (Figure 31.14). In this analysis the ligation products resulting from the described 3Canalysis are cut by a further restriction enzyme and transformed in a circular DNA (circular DNA) with a second ligation. These circular ligation products can be amplified by inverse PCR and the unknown DNA sequence can be identified by a sequencing or microarray technique. For this analysis anchor primers for a fixed so called anchor point are utilized. With the inverse PCR the sequence of unknown circular DNA can be detected. In this respect, a PCR from the known DNA fragment (bait-DNA) into the unknown ligation product is performed.
Hi-C is another modification of 3C that can identify chromatin interactions in a genome-wide manner. Again cells are fixed with formaldehyde and DNA is cut with a restriction enzyme. Subsequently, 5´ overhangs are filled in with biotinylated nucleotides and Klenow. Then a blunt-end ligation is performed under very dilute conditions. This results in a library of ligation products that represent interacting genomic regions. This library is sheared, and the junctions are pulled-down with streptavidin beads. Finally, interacting regions can be identified by paired-end sequencing. Another technique combines ChIP and chromosome configuration capture with high throughput sequencing. This technique, called ChIA-PET for chromatin interaction analysis by paired-end tag sequencing, allows the identification of genome wide chromosome interactions that are associated with a specific chromatin factor or modification.
31.9 Outlook The DNA modifications and chromatin play important roles in the regulation of gene expression (epigenetics). Since the sequence of the human genome has been elucidated, the DNA sequence is known. The next challenge is to decode the epigenetic modifications and configuration of the chromatin. Tissue- and illness-specific epigenetic patterns could be utilized as biomarkers for early diagnostics of diseases and their molecular classifications.
Further Reading Beck, S. and Rakyan, V.K. (2008) The methylome: approaches for global DNA methylation profiling. Trends Genet., 24, 231–237. Collas, P. (ed.) (2009) Chromatin Immunoprecipitation Assays, Methods and Protocols Series: Methods in Molecular Biology, Volume 567, Springer. Dammann, R., Li, C., Voon, J.H., Chin, E.L., Bates, S., and Pfeifer, G.P. (2000) Epigenetic inactivation of a RAS association domain family protein from the lung tumour suppressor locus 3p21.3. Nat. Genet., 25, 315–319. Dekker, J., Rippe, K., Dekker, M., and Kleckner, N. (2002) Capturing chromosome conformation. Science, 295, 1306–1311. Esteller, M. (ed.) (2005) DNA Methylation: Approaches, Methods and Applications, CRC Press, Boca Raton. Fullwood, M.J., Liu, M.H., Pan, Y.F., Liu, J., Xu, H., Mohamed, Y.B., Orlov, Y.L., Velkov, S., Ho, A., Mei, P.H., Chew, E.G., Huang, P.Y., Welboren, W.J., Han, Y., Ooi, H.S., Ariyaratne, P.N., Vega, V.B., Luo, Y., Tan, P.Y., Choy, P.Y., Wansa, K.D., Zhao, B., Lim, K.S., Leow, S.C., Yow, J.S., Joseph, R., Li, H., Desai, K.V., Thomsen, J.S., Lee, Y.K., Karuturi, R.K., Herve, T., Bourque, G., Stunnenberg, H.G., Ruan, X., Cacheux-Rataboul, V., Sung, W.K., Liu, E.T., Wei, C.L., Cheung, E., and Ruan, Y. (2009) An oestrogen-receptor-alpha-bound human chromatin interactome. Nature, 462, 58–64. Herman, J.G., Graff, J.R., Myohanen, S., Nelkin, B.D., and Baylin, S.B. (1996) Methylation-specific PCR: a novel PCR assay for methylation status of CpG islands. Proc. Natl. Acad. Sci. U.S.A., 93, 9821–9826. Lieberman-Aiden, E., van Berkum, N.L., Williams, L., Imakaev, M., Ragoczy, T., Telling, A., Amit, I., Lajoie, B.R., Sabo, P.J., Dorschner, M.O., Sandstrom, R., Bernstein, B., Bender, M.A., Groudine, M.,
Figure 31.14 Analysis of chromosomal interactions by the chromosome conformation capture technique. The chromosomal interactions are fixed by formaldehyde crosslinking. Chromatin is cut by a restriction digestion and the DNA fragments are ligated. The ligation products are purified and analyzed by quantitative PCR (real-time PCR) or by other techniques (e.g., inverse PCR).
830
Part IV: Nucleic Acid Analytics Gnirke, A., Stamatoyannopoulos, J., Mirny, L.A., Lander, E.S., and Dekker, J. (2009) Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science, 326, 289–293. Lister, R., Pelizzola, M., Dowen, R.H., Hawkins, R.D., Hon, G., Tonti-Filippini, J., Nery, J.R., Lee, L., Ye, Z., Ngo, Q.M., Edsall, L., Antosiewicz-Bourget, J., Stewart, R., Ruotti, V., Millar, A.H., Thomson, J.A., Ren, B., and Ecker, J.R. (2009) Human DNA methylomes at base resolution show widespread epigenomic differences. Nature, 462, 315–322. Xiong, Z. and Laird, E.W. (1997) COBRA: a sensitive and quantitative DNA methylation assay. Nucleic Acids Res., 25, 2532–2534. Zhao, Z., Tavoosidana, G., Sjölinder, M., Göndör, A., Mariano, P., Wang, S., Kanduri, C., Lezcano, M., Sandhu, K.S., Singh, U., Pant, V., Tiwari, V., Kurukuti, S., and Ohlsson, R. (2006) Circular chromosome conformation capture (4C) uncovers extensive networks of epigenetically regulated intra- and interchromosomal interactions. Nat. Genet., 38, 1341–1347.
Protein–Nucleic Acid Interactions Rolf Wagner Gustav-Stresemann-Sraße 15, 41352 Korschenbroich, Germany
Protein–nucleic acids interactions are fundamental for all events in living organisms that serve the conservation and propagation of genetic information. All steps during the flow of genetic information, such as replication, transcription, translation, as well as events during chromatin remodeling, repair, maturation, or transport are characterized through extensive contacts between nucleic acids and diverse classes of proteins. Notable examples of such proteins are polymerases, transcription factors, helicases, topoisomerases, ligases, or ribosomes and telomerases, the latter representing itself complexes between RNA and proteins. Hence, for modern molecular biology or biochemistry it is of central importance to unravel the molecular mechanisms underlying the recognition between proteins and nucleic acids. Although the basic molecular details in the recognition between proteins and DNA or RNA share many common principles there are several subtle differences caused by the fundamental different higher-order structures and functions of both nucleic acid classes leading to specific peculiarities in their interaction mechanisms. A separate chapter is therefore devoted to describe methods for the analysis of RNA–protein complexes. It should be emphasized, however, that many of the methods described are similarly suitable for the analysis of RNA–protein complexes as well as DNA-protein complexes. This is especially true for the physical methods described in this chapter.
32.1 DNA–Protein Interactions 32.1.1 Basic Features for DNA–Protein Recognition: Double-Helical Structures DNA predominantly exists as a double-stranded helical structure, which over large ranges consists of the B-form structure as postulated by Watson and Crick. This helical structure is characterized by two base-paired polynucleotide strands, which are intertwined plectonemically by two strands of opposite polarity (strands cannot be pulled apart without unwinding). In this structure, the negative charges of the sugar-phosphate backbone point outwards at an optimal distance. The helix is furthermore characterized by a major groove and a minor groove, which are wound in right-handed turns around the helix axis. The paired aromatic bases (A:T and G:C) are stacked parallel on top of each other perpendicular to the helix axis (tilt). Neighboring base pairs are twisted relative to each other by 36° in right-hand turns (twist), which results in a full helical turn after about ten base pairs. Donor and acceptor positions of the nucleotide bases are involved in base pairing within the helix and thus shielded by the sugar-phosphate backbone from functional side groups of potential outside proteins. Apart from the interactions of polymerases and some single-strand binding proteins the recognition of most DNA binding proteins occurs without breaking the double-stranded base pairing structure. Hence, the interaction between DNA and Bioanalytics: Analytical Methods and Concepts in Biochemistry and Molecular Biology, First Edition. Edited by Friedrich Lottspeich and Joachim Engels. 2018 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2018 by Wiley-VCH Verlag GmbH & Co. KGaA.
32
832
Part IV: Nucleic Acid Analytics
Figure 32.1 Structure of helical B-DNA. (a) Arrangement of the sugar-phosphate backbone and the major and minor grooves. (b) Chemical recognition motifs within the grooves. A: H-bond acceptors, D: H-bond donors, and M: methyl group.
proteins does not involve specific Watson–Crick-type base pairing, which is otherwise extremely important for nucleic acid interactions during biological processes. However, the grooves of the DNA double helix provide very specific surfaces for the recognition of protein structures in which each specific base pair exhibits an individual pattern of H-bond donors, -acceptors, or methyl groups for the interaction with amino acid side chains of proteins. The helical grooves of the DNA play a predominant role in the interaction between DNA and proteins (Figure 32.1). There are of course special proteins that recognize less abundant DNA structures, such as single-stranded DNA or alternative helix structures, like Z-DNA, for instance, which represents a left-handed structure. Interestingly, these proteins often share structural similarities with RNAbinding proteins.
32.1.2 DNA Curvature
Figure 32.2 Parameters describing helical DNA conformations. (a) DNA-helix parameters; (b) schematic models explaining DNA curvature.
In total, the structure of DNA does not only represent DNA in the B-form. In fact, the exact helical geometry of a given DNA results from the sequence of the different base pairs. Depending on the individual sequence the DNA does not uniformly follow a B-form structure but local differences in the helical conformation occur as a result of deviations in rotational (twist, tilt, or roll angles) or translational parameters (shift, rise, slide) of the base pairs. Often these structural deviations cause a curved path (curvature) of the otherwise straight DNA conformation. Such changes in the DNA contour often provide additional recognition signals for the interaction of specific proteins. Characteristically, curvature arises at consecutive A:T base-pairs clustered in helical phase. Yet, DNA curvature can also result from GGCC sequence repetitions. There is, however, a difference in the direction of the curvature induced by A:T clusters or GGCC repeats. At A:T clusters the minor groove points to the inside of the curvature whereas GGCC repeats result in a curvature with the minor groove pointing to the outside of the curvature. The resulting contour angles are quite remarkable and for a single A:T cluster angles between 12° and 23° have been noted. Several models have been put forward to explain the occurrence of curvature. The most descriptive are probably the ApA-wedge- and the B-junction models (Figure 32.2). In a simplified way, the ApA-wedge model predicts that each A:T dinucleotide causes a change in the tilt and roll angles. This results in a widening of the stacked bases similar to a wedge. Several such alterations in helical phase (ten base-pair distance corresponding to one helical turn) lead to a continuous DNA curvature. In the B-junction model the curvature is explained by the fact that only in the B-form DNA are the stacked bases perpendicular to the helix axis while in other helical conformations, such as the A-form, the plane of the bases relative to the helix axis is changed by the tilt angle. DNA sequences consisting of A:T clusters have a tendency to exist in the DNA A-form. A kink of the helix axis occurs at the junction between the A- and B- conformations because the stacking properties of the aromatic bases force all base pairs to remain packed in a parallel manner. Notably, DNA curvature must not always be static to enable the interaction of curvature-dependent proteins. Often, anisotropic flexibility of the DNA (preferential bending in one direction) suffices to cause a shortened persistence length, facilitating the deformation in one but not the other
32 Protein–Nucleic Acid Interactions
direction. A specific interaction is supported if an adequate adaptation of the DNA structure to the protein surface is possible. This type of induced conformational change is termed DNA bending. DNA curvature, and the concomitant efficiency to bind various proteins, depends on several external parameters including temperature (normally DNA curvature melts above 50 °C) or the presence of bivalent ions, such as Mg2+ and Ba2+, which generally enhance curvature, while antibiotics like distamycin, which binds into the minor groove of A:T-rich sequences, cause a reduction in curvature. Moreover, superhelicity has a profound effect on DNA curvature. How can the curvature of DNA be detected or even the position and intensity of DNA curvature be determined and how could one show that binding of a protein might alter the curvature of a given DNA? The simplest approach applicable in almost every laboratory is gel electrophoresis. As outlined below, DNA curvature reduces the gel electrophoretic mobility. In other words, the speed of migration of a curved DNA fragment is slower compared to a non-curved DNA fragment of the same length. To determine the degree of curvature of a particular DNA one has to compare the gel electrophoretic mobilities of the curved (μobs) and the non-curved (μact) DNA fragments. This can easily be done for DNA fragments between 100 and 500 bp with native polyacrylamide gels between 8% or 10%. To measure the difference in mobility the gel electrophoresis has to be performed at low (50 °C) temperature at which the curvature but not the base pairs of the double strands melts. The ratio of the two mobilities is termed the k-factor (k = μact/μobs). A k-factor larger than 1 (k > 1) indicates that the DNA is curved, whereby the magnitude of the k-factor correlates with the angle of the curvature. Moreover, the magnitude of the k-factor depends on the position of the center of curvature within the DNA fragment, which means that reduction of the mobility caused by curvature correlates with the end-to-end distance of the DNA fragment. In fact, the end-toend distance of a curved DNA is a function of the angle of the curvature and the position of the center of the curvature with respect to the fragment ends. Hence, for the end-to-end distance of a curved DNA fragment the position of the center of curvature matters. For the same angle the distance is smaller if the center of curvature is closer to the middle of the fragment than to its ends. This means that DNA fragments of equal size with the same curvature exhibit the lowest gel electrophoretic mobility if their center of curvature is localized in the middle of the fragment. If one determines the mobility of a DNA fragment with a given curvature in the middle of the fragment (μM) and compares this mobility with the same curvature localized at the fragment ends (μE) it is possible to derive the curvature angle α from the empirical relationship μM/μE = cos(α/2). This gel electrophoretic measurement not only allows determination of the degree of curvature it also discloses the center of curvature within a certain DNA fragment. Special plasmids have been constructed to clone curved DNA fragments or curved protein binding sites at different positions within a series of DNA fragments of exactly equal length (circular permutation assay). The different DNA fragments are generated by hydrolysis with a set of restriction enzymes for which cleavage sites have been positioned as direct repeats flanking a central cloning site for the uptake of the desired DNA. The combined restriction digest results in a set of DNA fragments of identical size with the inserted DNA at a different distance relative to the fragment ends. A plot of the gel electrophoretic mobility of the different fragments against the distance of the fragment ends in bp yields the center of curvature as the extrapolated position of minimal mobility. The curvature angle can additionally be derived from μM/μE = cos α/2 (Figure 32.3).
32.1.3 DNA Topology Single-stranded DNA forms are extremely rare; hence, most existing DNA conformations can be described by different double-helical forms and sections of static or dynamic curvature. However, biological DNA are often covalently closed circles and/or very large with the ends fixed and not free to move. Such structures give rise to different topological isomers, which are characterized by additional parameters. The topology of DNA molecules is very important for biological processes such as replication or transcription, which generally involve DNA–protein interactions. Some fundamental facts to understand the effect of topology on DNA–protein interaction are listed below.
833
Figure 32.3 Scheme of a permutation analysis to determine the center of curvature within a given DNA. (a) Arrangement of DNA fragments of identical length resulting from restriction hydrolysis with enzymes A–G. Restriction sites of the enzymes A–G for the integration of a DNA-binding region or a curved DNA flank the cloning site (grey box) as direct repeats. Hydrolyses with the different enzymes result in fragments of equal length in which the region for the integration of the DNA in question exhibits different distances to the fragment ends. (b) Schematic depiction of a retardation gel with the different DNA fragments. (c) Diagram showing the relative mobilities of the DNA fragments as a function of the position of the cloning site relative to the fragment start. The center of the curvature is derived by extrapolation of the position with minimal mobility versus the base position.
834
Part IV: Nucleic Acid Analytics The spatial description of molecules that exist as closed circles or which are fixed at their ends require an additional dimension defining the topology of the system. Such molecules can exist as different topoisomers. In the case of circular DNA molecules different superhelical structures give rise to different topoisomers. Superhelical structures are divided into positive (left-handed screw, DNA overwound) and negative (right-handed screw, DNA underwound) superhelical windings. Circular DNA with neither positive nor negative superhelical windings is termed relaxed.
For a more detailed description of DNA topology and related phenomena the reader is referred to specialized literature. The parameters relevant for topological molecules are defined by a simple equation: LK T W W R
Figure 32.4 Schematic illustration of the parameters LK (a), TW (b), and WR (c) describing superhelical DNA structures.
Figure 32.5 Coupling between transcription and superhelical DNA according to the twin supercoiled domain model.
(32.1)
where LK is the linking number, TW is the twisting number, and WR is the writhing number. The linking number LK describes how often a DNA strand is intertwined; LK represents the topological constant, which can only be changed if a DNA strand is broken; LK is necessarily an integer. The twist TW gives the number of rotations of the antiparallel DNA strands around the helix axis. In B-DNA the twist TW is for instance 10.5 bp per turn. The writhing number WR reflects the three-dimensional contour of the helix axis and describes the number of superhelical over- or underwindings. For relaxed circular DNA without any superhelical windings WR = 0. For such a molecule the linking number and twisting number are identical (LK = TW, which follows from LK = TW + WR). In the case of right-handed superhelical windings the writhing number is negative (WR < 0), while for left-handed supercoils it is positive (WR > 0) (Figure 32.4). How can these parameters be influenced? As outlined above, the topological constant LK can only be changed by breaking covalent bonds (the responsible enzymes in the cell are called topoisomerases). Both WR and TW are prone to changes by a number of biological relevant processes, which are related to protein binding. Examples are changes in twist (overwinding or melting of the doublestrand structure). Proteins that change the twist upon DNA binding either enhance or reduce the superhelicity of the DNA. In turn, enhancement or reduction of the superhelicity may cause a change in the binding affinity of such proteins. Processes like the intercalation of aromatic amino acids between DNA base pairs change twist automatically. In addition, the intercalation of dyes (e.g., ethidium bromide) or antibiotics, which bind into the grooves of DNA, has a similar effect. The superhelicity also changes if base pars are disrupted and the DNA melts in response to protein binding because unwinding the DNA strands reduces TW. This effect is valid for all polymerases, which cause DNA-melting over a defined range! Therefore, transcription has a direct influence on DNA superhelicity and vice versa. Note that polymerases, owing to their size and steric constraints, are unable to rotate with the necessary speed (∼300 rpm). As a consequence, regions of positive and negative superhelicity flank the section, where RNA polymerase moves (twin supercoiled domain model). Within the cell such regions of superhelical over- or underwound DNA are normally relaxed by cellular enzymes (topoisomerases) (Figure 32.5).
32 Protein–Nucleic Acid Interactions
835
32.2 DNA-Binding Motifs Comparative structural analyses of known DNA-binding proteins have led to the classification of characteristic amino acid sequence motifs for the recognition and binding of DNA. The most prominent DNA-binding motifs can be divided into five major classes (Figure 32.6): helix-turnhelix structures, leucine zipper structures, zinc-finger domains, helix-loop-helix domains, and β-sheet structures. Helix-turn-helix structures (HTHs) consist of a section of roughly 20 amino acids in length in which two α-helices are linked by a short β-loop (turn) of approximately four amino acids with an invariant glycine at the second position. Both α-helices are oriented almost perpendicular to each other. The helix closer to the C-terminus is defined as the recognition helix. The recognition helix, which fits exactly into the DNA major groove, is responsible for the recognition. HTH-binding proteins are ubiquitous in prokaryotes and eukaryotes. In prokaryotes HTH-proteins generally recognize palindromic DNA sequences and for that reason normally exist as symmetrical dimers or even-numbered oligomers. Members of eukaryotic HTH-proteins, for instance the homeodomain protein family, bind non-symmetrical DNA sequences as monomers or heterodimers. Some contain additional N-terminal sequences that facilitate binding through the interaction with the DNA minor groove. A variant of HTHproteins is those containing winged-HTH domains. In winged-HTH-proteins the recognition motif is extended by a third α-helix with a neighboring β-sheet. This secondary structural element makes additional contacts with the DNA backbone. Zinc-finger proteins exist in many variations and are mainly found in eukaryotes. They all are characterized by a tetrahedral coordination of one or two Zn-ions by conserved cysteines or histidines, which stabilize modular domains of the protein. In the classic case of Zn-finger transcription factors two antiparallel β-sheets, which are linked by a loop with an α-helix, are coordinated by a Zn-ion between two cysteines and two histidines. The DNA contact is maintained by the α-helix, which recognizes a stretch of three base pairs through the major groove. Zinc-finger proteins often consist of multiple such motifs arranged in a consecutive way, such that they are wound helically around the DNA during binding. A special situation is found in Gal4, a yeast transcription factor. Here, two neighboring Zn-ions are bound coordinatively by six cysteines, with each two cysteines sharing one Zn-binding (shared ligands). The two Zn-ions stabilize the position of two α-helices, which also interact with the DNA major groove. Leucine zipper proteins are designated according to their mechanism of dimerization. They exist as homo- or heterodimers and have almost exclusively been described in eukaryotes. They are composed of an α-helical recognition helix linked to a C-terminal dimerization helix. Dimerization is maintained through hydrophobic interactions between two amphipathic dimerization helices, which form a coiled-coil structure. This structure is characterized by the interaction of each two hydrophobic amino acid residues (generally leucines) separated by two α-helical repeats (heptad repeat) oriented almost on the same site of the helix. The leucine side chains are arranged like the teeth of a zipper. The DNA interaction is maintained by the two separate N-terminal domains that contain positively charged side chains (basic region). These recognition helices are formed like a fork, which fits in opposite directions of the DNA major groove. Helix-loop-helix proteins (HLHs) are related to zipper proteins. They consist of a shorter DNA binding helix and a longer dimerization α-helix, which is linked by an unstructured loop to a four-helix-bundle. HLH proteins form homo- or heterodimers similarly to leucine zipper proteins. Each one of the α-helixes from the two dimers binds into the DNA major groove. The binding specificity and affinity can thus be modulated by different protein partners. β-Folds of proteins use their particular secondary structure as the principal element for DNA binding. A pair of anti-parallel β-strands adapts itself into the DNA minor groove. Solved high-resolution structures (e.g., the TATA binding protein (TBP)) reveals that the conserved β-leaf structure of two pseudo-identical domains form a saddle-like structure, which fits into the minor groove of the DNA recognition sequence. Aromatic side chains of two conserved phenylalanines at the end of each β-sheet are intercalated between two DNA base pairs. This interaction creates a DNA kink, such that the DNA points away from the binding protein.
Figure 32.6 Schematic depiction of different DNA binding motifs.
836
Part IV: Nucleic Acid Analytics
32.3 Special Analytical Methods Several very powerful methods, starting from very simple to those having a high to very high technical cost, are presented below. As a detailed introduction to the technical and theoretical requirements exceeds the intension of this chapter, different methods will only be introduced briefly and their applications exemplified.
32.3.1 Filter Binding One of the earliest methods for the analysis of protein–nucleic acid interactions is the filter binding technique, which relies on the principle that proteins bind to nitrocellulose membranes while, for example, nucleic acids, if not too large or complex, migrate through the membrane during a filter process. In a typical binding experiment a mixture of protein and the putative target nucleic acid (preferably radiolabeled) is filtered through a nitrocellulose membrane. The non-bound nucleic acid is subsequently washed from the filter. Existing protein–nucleic acid complexes are retarded on the filter by the protein. The use of radiolabeled nucleic acid allows the amount of complex to be determined by counting the radioactivity on the filter. If adequately performed a differential determination of the filtrate is also possible. Although filter binding is not a real equilibrium binding method it yields relatively exact quantitative data. Filter binding therefore serves to determine apparent binding constants and, moreover, is suitable for slower kinetic measurements. Owing to the low technical requirements and the fact that the method is simple, fast, and generally applicable, filter binding still belongs to the frequently used methods. Note, however, that the mechanism of interaction between a given protein and the nitrocellulose membrane is not completely known. It has been observed that certain proteins are not bound or lose their binding properties by induced conformational changes.
32.3.2 Gel Electrophoresis EMSA, electrophoretic mobility shift analysis, or short mobility shift or gel retardation describes a method used to separate protein–nucleic acid complexes from the free nucleic acid