Chromosome Architecture: Methods and Protocols (Methods in Molecular Biology, 2476) 107162220X, 9781071622209

Table of contents :
Preface
Contents
Contributors
Chapter 1: A Next Generation of Advances in Chromosome Architecture
1 Introduction
References
Chapter 2: Single-Molecule Narrow-Field Microscopy of Protein-DNA Binding Dynamics in Glucose Signal Transduction of Live Yeas...
1 Introduction
2 Materials
2.1 Fluorescently Labeled Yeast Strains
2.2 Sample Preparation
2.3 Narrow-Field Microscope
2.4 Computational Analysis
3 Methods
3.1 Growing Yeast Strains
3.2 Plasma Cleaning Coverslips
3.3 Preparing Cell Samples
3.4 Agarose Pad Preparation
3.5 Obtaining Single-Molecule Data
3.6 Tracking Single Molecules
3.7 Analyzing Single-Molecule Trajectories
3.8 Categorizing Tracks by Cell Compartment
4 Notes
References
Chapter 3: Convolutional Neural Networks for Classifying Chromatin Morphology in Live-Cell Imaging
1 Introduction
1.1 Chromatin Morphology Changes Throughout Cell´s Life Cycle
1.2 Machine Learning Strategies for Cell-State Classification
1.3 Exploratory Visualization of CNN Feature Embedding
2 Materials
3 Methods
3.1 Annotating Training Images Using Napari
3.2 Training a CNN for Chromatin Morphology Classification
3.2.1 Training the CNN on a Train Set
3.2.2 Evaluating the CNN on a Validation Set
3.3 Using the CNN for Chromatin Morphology Classification on Unseen Data
3.3.1 Using the Trained CNN for Inference on a Test Set
3.3.2 Visualizing the CNN Feature Embedding
4 Notes
References
Chapter 4: Fluorescence Recovery After Photobleaching (FRAP) to Study Dynamics of the Structural Maintenance of Chromosome (SM...
1 Introduction
2 Materials
2.1 Growth Media
2.2 Slides and Microscope
3 Methods
3.1 Preparation of Bacterial Cultures for Microscopy
3.2 Preparation of Microscopy Slide
3.3 Microscopy
3.4 Image Analysis
4 Notes
5 Conclusions
References
Chapter 5: Atomic Force Microscopy of DNA and DNA-Protein Interactions
1 Introduction
2 Materials
2.1 General Materials
2.2 AFM Cantilevers
2.3 Buffer Solutions
2.4 DNA Substrates
3 Methods
3.1 Preparation of Mica Substrate
3.2 Methods for DNA Adsorption on a Mica Substrate for AFM Imaging in Fluid
3.2.1 DNA Adsorption Using Divalent Cations
3.2.2 DNA Adsorption Using PLL
3.2.3 DNA Adsorption Using PLL-b-PEG
3.3 Pre-Imaging Setup for High-Resolution AFM in Fluid
3.4 Optimizing AFM Imaging in PFT for High-Resolution AFM Imaging on DNA
3.5 Methods for DNA-Protein Imaging by AFM in Fluid
4 Notes
References
Chapter 6: Live-Cell Visualization of DNA Transfer and Pilus Dynamics During Bacterial Conjugation
1 Introduction
2 Materials
2.1 Medium for Cell Culture and Dye Reagents
2.2 Agarose´s Pad and Microfluidics Plates
2.3 Image Analysis Suite
3 Methods
3.1 Cysteine Substitution Mutants for Maleimide Bioconjugation
3.2 Labeling of the Conjugative Pilus
3.3 Visualization of Plasmid DNA Transfer
3.4 Live-Cell imaging on Agarose-Mounted Slides
3.5 Live-Cell Imaging Within Microfluidic Chambers
3.6 Imaging Acquisition
3.7 Image Analysis
4 Notes
References
Chapter 7: Transverse Magnetic Tweezers Allowing Coincident Epi-Fluorescence Microscopy on Horizontally Extended DNA
1 Introduction
2 Materials
2.1 Buffers and Special Reagents
2.2 Preparation of Microspheres Labeled with Anti-Digoxigenin (PAG-AD-MS)
2.3 Preparation of Oxygen Scavenger System
2.4 Preparation of Microscopy Substrates
2.5 Preparation of Horizontal DNA Tethers
2.6 Combined Magnetic Tweezers and Epi-Fluorescence Microscope
3 Methods
3.1 Preparation of Microspheres Labeled with Anti-Digoxigenin
3.2 Preparation of Oxygen Scavenger System
3.3 Preparation of Microscopy Substrates
3.4 Preparation of Horizontal DNA Tethers
3.5 Manipulation of DNA-Tethered Superparamagnetic Microspheres
4 Notes
References
Chapter 8: Atomistic Molecular Dynamics Simulations of DNA in Complex 3D Arrangements for Comparison with Lower Resolution Str...
1 Introduction
1.1 Overview of Simulation Protocols
1.2 Computer Hardware, Software, and MD Force Fields
1.3 Analysis Tools of DNA Simulations for Comparison with Experiments
2 Materials
2.1 MD Simulation, Analysis, and Visualization Software
2.2 AMBERTOOLS Modules
2.3 Source Files
3 Methods
3.1 Molecular Dynamics Simulations
3.2 Conformational Analysis
4 Notes
References
Chapter 9: Replication Labeling Methods for Super-Resolution Imaging of Chromosome Territories and Chromatin Domains
1 Introduction
2 Materials
2.1 Cell Culture and Labeling
2.2 Sample Staining, Fixation, and Microscopy
3 Methods
3.1 F-ara-EdU Labeling of Individual Chromosome Territories
3.2 Replication Domains and Origins for Fixed/Live Samples
3.3 Differential Labeling of Sister Chromatids for Fixed Samples (Fig. 5)
4 Notes
References
Chapter 10: DNA-Protein Interactions Studied Directly Using iSCAT Imaging of GNP-Tagged Proteins
1 Introduction
2 Materials
2.1 Interferometric Scattering Microscope
2.2 Digoxigenin-Labeled Oligonucleotides
2.3 5x ABC Buffer
2.4 ABT Buffer
2.5 mPEG Solution
2.6 1x TE Buffer
2.7 Anti-digoxigenin Solution
2.8 Flow Cells
3 Methods
3.1 Slide and Coverslip Preparation
3.2 Plasma Cleaning
3.3 Slide and Coverslip Silanization
3.4 Flow Cell Construction
3.5 Calibration Curve Data Collection
3.6 Annealing Oligonucleotides
3.7 Experiment Setup
3.8 Image Acquisition
3.9 Image Processing
3.10 Calibration Curve
3.11 Data Analysis
4 Notes
References
Chapter 11: Escherichia coli Chromosome Copy Number Measurement Using Flow Cytometry Analysis
1 Introduction
2 Materials
3 Methods
3.1 Standard Strains
3.2 dnaAts Mutants
3.3 Staining Cells for Flow Cytometry
4 Notes
References
Chapter 12: Dynamics of Bacterial Chromosomes by Locus Tracking in Fluorescence Microscopy
1 Introduction to Live-Cell Locus Tracking at High Frame Rate
2 Materials
2.1 Bacteria Strains and Reagents
2.2 Microscopy
3 Methods
3.1 Sample Preparation
3.2 Video Acquisition
3.3 Image Analysis
3.4 Considerations on Precision and System-Specific Aspects of the Localization Data
3.5 Choice of the Protein Fusion to Label Foci
3.6 Considerations on the Correlation of Loci Motility to Intensity
3.7 Cellular Density Estimation
3.8 Control of Cell Lysis
3.9 Refraction Index Estimation
4 Notes
References
Chapter 13: Measuring Nuclear Mechanics with Atomic Force Microscopy
1 Introduction
2 Materials
2.1 Atomic Force Microscope
2.2 Cantilevers
2.3 Coverslip cleaning
3 Methods
3.1 Cell Seeding: Full-Adhesion Measurements
3.2 Cell Seeding: Initial Adhesion Measurements
3.3 Nuclei Seeding: Isolated Nuclei
3.4 Cantilever Calibration
3.5 Mechanical Measurements of the Nucleus
3.6 Data Analysis: Calculation of the Young´s Modulus E
4 Notes
References
Chapter 14: Copy Number Analysis of the Yeast Histone Deacetylase Complex Component Cti6 Directly in Living Cells
1 Introduction
2 Materials
2.1 Yeast Strains
2.2 Super-Resolution Microscopy
3 Methods
3.1 Fluorescent Labeling of the Cti6 Protein
3.2 Conformation PCR
3.3 Preparing Cells for Microscopy
3.4 Preparing Microscopy Slides Using 125 μL Gene Frame
3.5 Data Acquisition
3.6 Copy Number Analysis
4 Notes
References
Chapter 15: Super-Resolution Microscopy and Tracking of DNA-Binding Proteins in Bacterial Cells
1 Introduction
2 Materials
2.1 Lambda Red Recombination
2.2 Cell Culture, Labeling, and Slide Preparation
2.3 Microscope
2.4 Data Analysis
3 Methods
3.1 Lambda Red Integration to Generate an Endogenous Fluorescent Fusion
3.2 Cell Culture
3.3 Microscopy
3.4 Reconstructing a Super-Resolution Map of Localizations
3.5 Short Exposure Times to Quantify Diffusion
3.6 Long Exposures to Quantify Binding Kinetics
3.7 Clustering Analysis
4 Notes
References
Chapter 16: Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells Using Single-Molecule Localization Microsco...
1 Introduction to the Role of Live-Cell Single-Molecule Localization Microscopy
2 Materials
2.1 SMLM Microscope Configuration
2.1.1 Sample Illumination
2.1.2 Sample Detection
2.2 Labeling Mammalian Cells
2.2.1 Fluorophore Choice
2.2.2 Designing Expression Vectors
3 Methods
3.1 Sample Preparation
3.2 Microscope Setup
3.3 Experimental Design for Single-Molecule Imaging and Tracking
3.3.1 Short-Exposure Tracking of Single-Protein Diffusion and Binding to Chromatin
3.3.2 Long-Exposure Tracking of Single Chromatin-Bound Proteins for Probing Chromatin Diffusion
3.3.3 Time-Lapse Experiments for Determining Residence Time
3.3.4 Structural Imaging
3.4 Data Analysis
3.4.1 Quantifying Numbers of Molecules
3.4.2 Extracting Clustering Parameters
3.4.3 Extracting Biophysical Parameters from SPT
3.4.4 Residence Time
3.4.5 Using Photobleaching Rates to Analyze Single-Molecule FRET Trajectories
3.4.6 Analyzing Images with High Density of Localizations
4 Notes
References
Chapter 17: The End Restraint Method for Mechanically Perturbing Nucleic Acids In Silico
1 Introduction
1.1 DNA Mechanics In Vivo
1.2 Experimental Methods
1.3 Simulation Approaches
1.4 Advantages of Our Approach
2 Materials
2.1 Molecular Dynamics Software
2.2 Visualization Software
2.3 Analysis Utilities
2.4 Bespoke Scripts and Input Files
3 Methods
3.1 Initial System Preparation
3.2 Applying Torsion
3.3 Equilibration
3.4 Running the Umbrella-Sampling Simulations
3.5 Analysis Strategies
4 Notes
References
Chapter 18: Visualizing the Replisome, Chromosome Breaks, and Replication Restart in Bacillus subtilis
1 Introduction
2 Materials
2.1 Growth of B. subtilis.
2.2 General Microscopy
3 Methods
3.1 Growth Conditions for Visualizing the Nucleoid, Origins, and Replisome in B. subtilis
3.2 Growth Conditions for DNA Damage Assays
3.3 Growth Conditions for Replication Restart Assays
3.4 Preparation of Multi-Spot Microscope Slides
3.5 Preparation of Microscope Slide (Gene Frame)
4 Notes
References
Chapter 19: High-Throughput Imaging of Bacillus subtilis
1 Introduction
2 Materials
2.1 Media and Equipment for the Growth of B. Subtilis
2.2 Components for Imaging
3 Methods
3.1 Exponential Growth of Cells
3.2 Prepare 96-Well Pedestals
3.3 Mounting B. subtilis Strains on the Agarose Pedestals
3.4 High-Throughput Imaging of B. subtilis Strains Using Wide-Field Fluorescence Microscopy
3.5 Option 1: Setting Software for Semiautomated Image Acquisition
3.6 Option 2: Setting Software for Automated Image Acquisition
3.7 Characterization of Mutant Phenotypes by Quantitative Image Analysis
4 Notes
References
Chapter 20: Measuring Nuclear Organization of Proteins with STORM Imaging and Cluster Analysis
1 Introduction
2 Materials
2.1 Etch Solution
2.2 Imaging Buffer
2.3 Coverslip Preparation
3 Methods
3.1 Cell Preparation
3.2 Immunofluorescence
3.3 Imaging Sample Preparation
3.4 STORM Image Acquisition
3.5 STORM Image Processing
3.6 Cluster Analysis of STORM Data
4 Notes
References
Chapter 21: Single-Molecular Quantification of Flowering Control Proteins Within Nuclear Condensates in Live Whole Arabidopsis...
1 Introduction
2 Materials
3 Methods
3.1 Preparation of Live Arabidopsis Roots for Microscopy
3.2 Preparation of a Standard Yeast Sample for MICROSCOPY
3.3 Preparation of an AiryScan Alignment Sample
3.4 Optional: Slimfield Imaging of the Standard Sample
3.5 AiryScan Imaging of the Standard and Root Samples
3.6 Single-Molecule Brightness Using Literature Estimates
3.7 Optional: Verification of Single-Molecule Calibration Using Slimfield Imaging
3.8 Applying the Calibration to the Analysis of AiryScan Plant Root Images
4 Notes
4.1 Preparation of Live Roots for Investigation
4.2 Preparation of the Standard Yeast Sample
4.3 Preparation of an AiryScan Alignment Sample
4.4 Slimfield Imaging
4.5 AiryScan imaging
4.6 Single-Molecule Calibration, Verification, and Implementation
References
Index

Citation preview

Methods in Molecular Biology 2476

Mark C. Leake Editor

Chromosome Architecture Methods and Protocols Second Edition

METHODS

IN

MOLECULAR BIOLOGY

Series Editor John M. Walker School of Life and Medical Sciences University of Hertfordshire Hatfield, Hertfordshire, UK

For further volumes: http://www.springer.com/series/7651

For over 35 years, biological scientists have come to rely on the research protocols and methodologies in the critically acclaimed Methods in Molecular Biology series. The series was the first to introduce the step-by-step protocols approach that has become the standard in all biomedical protocol publishing. Each protocol is provided in readily-reproducible step-bystep fashion, opening with an introductory overview, a list of the materials and reagents needed to complete the experiment, and followed by a detailed procedure that is supported with a helpful notes section offering tips and tricks of the trade as well as troubleshooting advice. These hallmark features were introduced by series editor Dr. John Walker and constitute the key ingredient in each and every volume of the Methods in Molecular Biology series. Tested and trusted, comprehensive and reliable, all protocols from the series are indexed in PubMed.

Chromosome Architecture Methods and Protocols Second Edition

Edited by

Mark C. Leake Departments of Physics and Biology, University of York, York, UK

Editor Mark C. Leake Departments of Physics and Biology University of York York, UK

ISSN 1064-3745 ISSN 1940-6029 (electronic) Methods in Molecular Biology ISBN 978-1-0716-2220-9 ISBN 978-1-0716-2221-6 (eBook) https://doi.org/10.1007/978-1-0716-2221-6 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Humana imprint is published by the registered company Springer Science+Business Media, LLC part of Springer Nature. The registered company address is: 1 New York Plaza, New York, NY 10004, U.S.A.

Preface Following the success of the first edition of Chromosome Architecture: Methods and Protocols as part of the Methods in Molecular Biology series, this new edition comprises updated chapters from the original authors as well as new content from new contributors in this fast-moving field. This issue, as with the original, will consist of a collection of cutting-edge laboratory protocols, techniques, and applications in use today by some of the leading international experts in the broad field of “chromosome architecture.” A key difference in emphasis, compared with the previous collections of review articles published in this area, is the emphasis on the development and application of complex techniques and protocols, which increase the physiological relevance of chromosome architecture investigation compared to the methods utilized previously; these developments are manifest both through application of far more complex bottom-up assays in vitro and through striving to maintain the native physiological context through investigation of living, functional cells. The length scale of precision of experimental protocols in this area has improved dramatically over the recent years, and many cutting-edge methods now utilize state-of-the-art single-molecule approaches, both for imaging the DNA content of chromosome and proteins that bind to DNA and for using methods that can controllably manipulate single-DNA molecules. This issue will also include more complex, physiologically representative methods to investigate chromosome architecture through the use of advanced computational methods and mathematical analysis. University of York, York, UK

Mark C. Leake

v

Contents Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

v ix

1 A Next Generation of Advances in Chromosome Architecture . . . . . . . . . . . . . . . . Mark C. Leake 2 Single-Molecule Narrow-Field Microscopy of Protein-DNA Binding Dynamics in Glucose Signal Transduction of Live Yeast Cells . . . . . . . . . Adam J. M. Wollman and Mark C. Leake 3 Convolutional Neural Networks for Classifying Chromatin Morphology in Live-Cell Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kristina Ulicna, Laure T. L. Ho, Christopher J. Soelistyo, Nathan J. Day, and Alan R. Lowe 4 Fluorescence Recovery After Photobleaching (FRAP) to Study Dynamics of the Structural Maintenance of Chromosome (SMC) Complex in Live Escherichia coli Bacteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anjana Badrinarayanan and Mark C. Leake 5 Atomic Force Microscopy of DNA and DNA-Protein Interactions . . . . . . . . . . . . Philip J. Haynes, Kavit H. S. Main, Bernice Akpinar, and Alice L. B. Pyne 6 Live-Cell Visualization of DNA Transfer and Pilus Dynamics During Bacterial Conjugation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kelly Goldlust, Agathe Couturier, Laurent Terradot, and Christian Lesterlin 7 Transverse Magnetic Tweezers Allowing Coincident Epi-Fluorescence Microscopy on Horizontally Extended DNA . . . . . . . . . . . . . . . Stephen J. Cross, Claire E. Brown, and Christoph G. Baumann 8 Atomistic Molecular Dynamics Simulations of DNA in Complex 3D Arrangements for Comparison with Lower Resolution Structural Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . George Watson, Victor Velasco-Berrelleza, and Agnes Noy 9 Replication Labeling Methods for Super-Resolution Imaging of Chromosome Territories and Chromatin Domains . . . . . . . . . . . . . . . . . . . . . . . Ezequiel Miron, Joseph Windo, Fena Ochs, and Lothar Schermelleh 10 DNA-Protein Interactions Studied Directly Using iSCAT Imaging of GNP-Tagged Proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Robert J. Charman and Neil M. Kad 11 Escherichia coli Chromosome Copy Number Measurement Using Flow Cytometry Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Michelle Hawkins, John Atkinson, and Peter McGlynn 12 Dynamics of Bacterial Chromosomes by Locus Tracking in Fluorescence Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Leonardo Mancini, Estelle Crozat, Avelino Javer, Marco Cosentino Lagomarsino, and Pietro Cicuta

1

vii

5

17

31 43

63

75

95

111

129

145

155

viii

13

Contents

Measuring Nuclear Mechanics with Atomic Force Microscopy . . . . . . . . . . . . . . . ´ lia dos Santos, Florian Rehfeldt, and Christopher P. Toseland A Copy Number Analysis of the Yeast Histone Deacetylase Complex Component Cti6 Directly in Living Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sviatlana Shashkova, Thomas Nystro¨m, and Mark C. Leake Super-Resolution Microscopy and Tracking of DNA-Binding Proteins in Bacterial Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chloe´ J. Cassaro and Stephan Uphoff Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells Using Single-Molecule Localization Microscopy . . . . . . . . . Maike Steindel, Igor Orsine de Almeida, Stanley Strawbridge, Valentyna Chernova, David Holcman, Aleks Ponjavic, and Srinjan Basu The End Restraint Method for Mechanically Perturbing Nucleic Acids In Silico. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jack W. Shepherd and Mark C. Leake Visualizing the Replisome, Chromosome Breaks, and Replication Restart in Bacillus subtilis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hannah Gaimster, Charles Winterhalter, Alan Koh, and Heath Murray High-Throughput Imaging of Bacillus subtilis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Paula Montero Llopis, Ryan Stephansky, and Xindan Wang Measuring Nuclear Organization of Proteins with STORM Imaging and Cluster Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ´ lia dos Santos, Rosemarie E. Gough, Lin Wang, A and Christopher P. Toseland Single-Molecular Quantification of Flowering Control Proteins Within Nuclear Condensates in Live Whole Arabidopsis Root . . . . . . . . . . . . . . . . Alex L. Payne-Dwyer and Mark C. Leake

249

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

329

14

15

16

17

18

19 20

21

171

183

191

209

263

277

293

311

Contributors BERNICE AKPINAR • London Centre for Nanotechnology, University College London, London, UK; Molecular Sciences Research Hub, Department of Chemistry, Imperial College London, London, UK JOHN ATKINSON • School of Medical Sciences, Institute of Medical Sciences, University of Aberdeen, Foresterhill, Aberdeen, UK ANJANA BADRINARAYANAN • National Centre for Biological Sciences, Bengaluru, Karnataka, India SRINJAN BASU • Wellcome-MRC Cambridge Stem Cell Institute, University of Cambridge, Cambridge, UK CHRISTOPH G. BAUMANN • Department of Biology, University of York, York, UK CLAIRE E. BROWN • Department of Biology, University of York, York, UK CHLOE´ J. CASSARO • Department of Biochemistry, University of Oxford, Oxford, UK ROBERT J. CHARMAN • School of Biosciences, University of Kent, Canterbury, UK VALENTYNA CHERNOVA • Wellcome-MRC Cambridge Stem Cell Institute, University of Cambridge, Cambridge, UK PIETRO CICUTA • Department of Physics, Cavendish Laboratory, University of Cambridge, Cambridge, UK AGATHE COUTURIER • Microbiologie Mole´culaire et Biochimie Structurale (MMSB), Universite´ Lyon 1, CNRS, INSERM, UMR5086, Lyon Cedex 07, France STEPHEN J. CROSS • Department of Biology, University of York, York, UK ESTELLE CROZAT • Centre de Biologie Inte´grative de Toulouse, Laboratoire de Microbiologie et de Ge´ne´tique Mole´culaires, Universite´ de Toulouse, CNRS, UPS, Toulouse, France; Department of Physics, Cavendish Laboratory, University of Cambridge, Cambridge, UK NATHAN J. DAY • Institute of Structural and Molecular Biology, University College London, London, UK; Institute for the Physics of Living Systems, University College London, London, UK IGOR ORSINE DE ALMEIDA • Wellcome-MRC Cambridge Stem Cell Institute, University of Cambridge, Cambridge, UK ´ LIA DOS SANTOS • Department of Oncology and Metabolism, University of Sheffield, A Sheffield, UK HANNAH GAIMSTER • Centre for Bacterial Cell Biology, Biosciences Institute, Newcastle University, Newcastle upon Tyne, UK KELLY GOLDLUST • Microbiologie Mole´culaire et Biochimie Structurale (MMSB), Universite´ Lyon 1, CNRS, INSERM, UMR5086, Lyon Cedex 07, France ROSEMARIE E. GOUGH • Department of Oncology and Metabolism, University of Sheffield, Sheffield, UK MICHELLE HAWKINS • Department of Biology, University of York, York, UK PHILIP J. HAYNES • London Centre for Nanotechnology, University College London, London, UK; Molecular Science Research Hub, Department of Chemistry, Imperial College London, London, UK; Department of Physics and Astronomy, University College London, London, UK LAURE T. L. HO • Institute of Structural and Molecular Biology, University College London, London, UK; MRC Laboratory of Molecular and Cellular Biology, London, UK

ix

x

Contributors

DAVID HOLCMAN • Group of Computational Biology and Applied Mathematics, Institute of Biology, Ecole Normale Supe´rieure, Paris, France AVELINO JAVER • Institute of Biomedical Engineering, Department of Engineering Science, University of Oxford, Oxford, UK; Department of Physics, Cavendish Laboratory, University of Cambridge, Cambridge, UK NEIL M. KAD • School of Biosciences, University of Kent, Canterbury, UK ALAN KOH • Centre for Bacterial Cell Biology, Biosciences Institute, Newcastle University, Newcastle upon Tyne, UK MARCO COSENTINO LAGOMARSINO • IFOM, FIRC Institute of Molecular Oncology, Milan, Italy; Physics Department, University of Milan, and INFN, Milan, Italy MARK C. LEAKE • Departments of Physics and Biology, University of York, York, UK CHRISTIAN LESTERLIN • Microbiologie Mole´culaire et Biochimie Structurale (MMSB), Universite´ Lyon 1, CNRS, INSERM, UMR5086, Lyon Cedex 07, France ALAN R. LOWE • Institute of Structural and Molecular Biology, University College London, London, UK; London Centre for Nanotechnology, University College London, London, UK; Institute for the Physics of Living Systems, University College London, London, UK; The Alan Turing Institute, London, UK KAVIT H. S. MAIN • London Centre for Nanotechnology University College London, London, UK; UCL Cancer Institute, University College London, London, UK LEONARDO MANCINI • Department of Physics, Cavendish Laboratory, University of Cambridge, Cambridge, UK PETER MCGLYNN • Department of Biology, University of York, York, UK EZEQUIEL MIRON • Department of Biochemistry, University of Oxford, Oxford, UK PAULA MONTERO LLOPIS • MicRoN Core, Harvard Medical School, Boston, MA, USA HEATH MURRAY • Centre for Bacterial Cell Biology, Biosciences Institute, Newcastle University, Newcastle upon Tyne, UK AGNES NOY • Department of Physics, University of York, Heslington, York, UK THOMAS NYSTRO¨M • Department of Microbiology and Immunology, Institute of Biomedicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden FENA OCHS • Department of Biochemistry, University of Oxford, Oxford, UK ALEX L. PAYNE-DWYER • Department of Physics, University of York, York, UK; Department of Biology, University of York, York, UK ALEKS PONJAVIC • School of Physics and Astronomy and School of Food Science and Nutrition, University of Leeds, Leeds, UK ALICE L. B. PYNE • London Centre for Nanotechnology, University College London, London, UK; Department of Materials Science and Engineering, University of Sheffield, Sheffield, UK FLORIAN REHFELDT • Experimental Physics 1, University of Bayreuth, Bayreuth, Germany LOTHAR SCHERMELLEH • Department of Biochemistry, University of Oxford, Oxford, UK SVIATLANA SHASHKOVA • Department of Microbiology and Immunology, Institute of Biomedicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden; Department of Physics, University of York, York, UK JACK W. SHEPHERD • Department of Physics, University of York, York, UK CHRISTOPHER J. SOELISTYO • Institute of Structural and Molecular Biology, University College London, London, UK; Institute for the Physics of Living Systems, University College London, London, UK MAIKE STEINDEL • Wellcome-MRC Cambridge Stem Cell Institute, University of Cambridge, Cambridge, UK

Contributors

xi

RYAN STEPHANSKY • MicRoN Core, Harvard Medical School, Boston, MA, USA STANLEY STRAWBRIDGE • Wellcome-MRC Cambridge Stem Cell Institute, University of Cambridge, Cambridge, UK LAURENT TERRADOT • Microbiologie Mole´culaire et Biochimie Structurale (MMSB), Universite´ Lyon 1, CNRS, INSERM, UMR5086, Lyon Cedex 07, France CHRISTOPHER P. TOSELAND • Department of Oncology and Metabolism, University of Sheffield, Sheffield, UK KRISTINA ULICNA • Institute of Structural and Molecular Biology, University College London, London, UK; London Centre for Nanotechnology, University College London, London, UK; Institute for the Physics of Living Systems, University College London, London, UK STEPHAN UPHOFF • Department of Biochemistry, University of Oxford, Oxford, UK VICTOR VELASCO-BERRELLEZA • Department of Physics, University of York, Heslington, York, UK LIN WANG • Central Laser Facility, Research Complex at Harwell, Science and Technology Facilities Council, Rutherford Appleton Laboratory, Harwell, Oxford, UK XINDAN WANG • Department of Biology, Indiana University, Bloomington, IN, USA GEORGE WATSON • Department of Physics, University of York, Heslington, York, UK JOSEPH WINDO • Department of Biochemistry, University of Oxford, Oxford, UK CHARLES WINTERHALTER • Centre for Bacterial Cell Biology, Biosciences Institute, Newcastle University, Newcastle upon Tyne, UK ADAM J. M. WOLLMAN • Newcastle University Biosciences Institute, Newcastle University, Newcastle, UK

Chapter 1 A Next Generation of Advances in Chromosome Architecture Mark C. Leake Abstract New insight into the architecture of chromosomes, their molecular composition, structure and spatial location, and time-resolved features, has grown enormously through developments of a range of pioneering interdisciplinary approaches that lie at the interface of the life and physical sciences. These involve several state-of-the-art “physics of life” tools that are both experimental and theoretical, used in conjunction with molecular biology methods which enable investigation of chromosome structure and function in vitro, in vivo, and even in silico. In particular, a move towards far greater quantitation has enabled transformative leaps in our understanding. These have involved valuable improvements to the spatial and temporal resolution of quantitative measurements, such as in vivo super-resolved light microscopy and singlemolecule biophysics methods, which facilitate probing of dynamic chromosome processes hitherto impossible. Similarly, there have been important advances in the theoretical biophysics approaches which have enabled advances in predictive modeling to generate new understanding of the modes of operation of chromosomes across all domains of life. Here, I discuss these advances, and review the current state of our knowledge of chromosome architecture and speculation where future advances may lead. Key words Single-molecule biophysics, Super-resolution, DNA, Nucleus

1

Introduction This updated edition comprises a collection of truly cutting-edge laboratory protocols, techniques, and applications in use today by some of the leading international experts in the broad field of “chromosome architecture.” A key difference in emphasis, compared with previous collections of articles published in this area, is the emphasis on the development and application of complex techniques and protocols which increase the physiological relevance of chromosome architecture investigation compared to methods utilized previously—these developments are manifest both through application of far more complex bottom-up assays in vitro as well as in striving to maintain the native physiological context through investigation of living, functional cells [1]. In particular, experimental methods have used advances in optical microscopy [2],

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_1, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

1

2

Mark C. Leake

especially the use of fluorescence microscopy methods, to probe functional, living cells [3–12]. The length scale of precision of experimental protocols in this area has improved dramatically over recent years and many cutting-edge methods now utilize state-ofthe-art single-molecule approaches [13–15], both for imaging the DNA content of chromosome and proteins that bind to DNA and for using methods that can controllably manipulate single-DNA molecules and can image its structure to a precision better than the standard optical resolution limit [16, 17]. This volume also includes more complex, physiologically representative methods to investigate chromosome architecture through the use of advanced computational methods and mathematical analysis. What is clear is that the combination of pioneering molecular biology, biochemistry, and genetics methods with emerging, exciting tools from biophysics, bioengineering, computer science, and biomathematics is transforming our knowledge of functional chromosome architecture. Improvements in these fields are likely to add yet more insight over the next few years into the complex interactions between multiple key molecular players inside chromosomes.

Acknowledgments Support from the EPSRC (EP/T002166/1) and the Leverhulme Trust (RPG-2017-340) is acknowledged. References 1. Wollman AJM, Miller H, Zhou Z et al (2015) Probing DNA interactions with proteins using a single-molecule toolbox: inside the cell, in a test tube and in a computer. Biochem Soc Trans 43:139–145 2. Wollman AJM, Nudd R, Hedlund EG et al (2015) From Animaculum to single molecules: 300 years of the light microscope. Open Biol 5: 150019 3. Lenn T, Leake MC, Mullineaux CW (2008) Are Escherichia coli OXPHOS complexes concentrated in specialized zones within the plasma membrane? Biochem Soc Trans 36: 1032–1036 4. Plank M, Wadhams GH, Leake MC (2009) Millisecond timescale slim-field imaging and automated quantification of single fluorescent protein molecules for use in probing complex biological processes. Integr Biol (Camb) 1: 602–612 5. Chiu S-W, Leake MC (2011) Functioning nanomachines seen in real-time in living bacteria using single-molecule and super-resolution

fluorescence imaging. Int J Mol Sci 12: 2518–2542 6. Robson A, Burrage K, Leake MC (2013) Inferring diffusion in single live cells at the singlemolecule level. Philos Trans R Soc Lond Ser B Biol Sci 368:20120029 7. Bryan SJ, Burroughs NJ, Shevela D et al (2014) Localisation and interactions of the Vipp1 protein in cyanobacteria. Mol Microbiol 94:1179–1195 8. Llorente-Garcia I, Lenn T, Erhardt H et al (2014) Single-molecule in vivo imaging of bacterial respiratory complexes indicates delocalized oxidative phosphorylation. Biochim Biophys Acta 1837:811–824 9. Reyes-Lamothe R, Sherratt DJ, Leake MC (2010) Stoichiometry and architecture of active DNA replication machinery in Escherichia coli. Science 328:498–501 10. Badrinarayanan A, Reyes-Lamothe R, Uphoff S et al (2012) In vivo architecture and action of bacterial structural maintenance of chromosome proteins. Science 338:528–531

Next Generation Chromosome Architecture Advances 11. Wollman A, Leake MC (2015) Millisecond single-molecule localization microscopy combined with convolution analysis and automated image segmentation to determine protein concentrations in complexly structured, functional cells, one cell at a time. Faraday Discuss 184: 401–424 12. Lenn T, Leake MC (2015) Single-molecule studies of the dynamics and interactions of bacterial OXPHOS complexes. Biochim Biophys Acta 1857:224–231 13. Leake MC (2013) The physics of life: one molecule at a time. Philos Trans R Soc Lond Ser B Biol Sci 368(1611):20120248

3

14. https://pubmed.ncbi.nlm.nih.gov/2462 6744/ 15. https://pubmed.ncbi.nlm.nih.gov/28869217/ 16. Miller H, Zhaokun Z, Wollman AJM et al (2015) Superresolution imaging of single DNA molecules using stochastic photoblinking of minor groove and intercalating dyes. Methods 88:81–88 17. Lenn T, Leake MC (2012) Experimental approaches for addressing fundamental biological questions in living, functioning cells with single molecule precision. Open Biol 2: 120090

Chapter 2 Single-Molecule Narrow-Field Microscopy of Protein-DNA Binding Dynamics in Glucose Signal Transduction of Live Yeast Cells Adam J. M. Wollman and Mark C. Leake Abstract Single-molecule narrow-field microscopy is a versatile tool to investigate a diverse range of protein dynamics in live cells and has been extensively used in bacteria. Here, we describe how these methods can be extended to larger eukaryotic, yeast cells, which contain subcellular compartments. We describe how to obtain singlemolecule microscopy data but also how to analyze these data to track and obtain the stoichiometry of molecular complexes diffusing in the cell. We chose glucose-mediated signal transduction of live yeast cells as the system to demonstrate these single-molecule techniques as transcriptional regulation is fundamentally a single-molecule problem—a single repressor protein binding a single binding site in the genome can dramatically alter behavior at the whole cell and population levels. Key words Single-molecule biophysics, Signal-transduction, Yeast

1

Introduction Bulk biochemical methods can only measure mean ensemble properties while single-molecule techniques allow the heterogeneity in molecular biology to be explored which often leads to a new understanding of the biological system involved [1]. The use of fluorescent protein fusions to act as reporters can provide significant insight into a wide range of biological processes and molecular machines, for enabling insight into stoichiometry and architecture as well as details of molecular mobility inside living, functional cells with their native physiological context intact [2–7]. Singlemolecule narrow-field microscopy, and its similar counterpart slim field microscopy, is a versatile tool to investigate a diverse range of protein dynamics in live cells which can be used in conjunction with fluorescent protein fusion strains to generate enormous insight into biological processes at the single-molecule level. In bacteria, it has been used to investigate the components of the

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_2, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

5

6

Adam J. M. Wollman and Mark C. Leake

replisome [8], structural maintenance of chromosomes [9], DNA repair mechanisms [10, 11], and carbon-fixing, carboxysome in photosynthesis [12]. In narrow-field microscopy, the normal fluorescence excitation field is reduced to originally encompass only a single cell but, with advances in laser and camera technology, it can now encompass multiple cells. This produces a Gaussian excitation field (30–3000 μm2) with 100–1000 times the laser excitation intensity of standard epifluorescence microscopy. This intense illumination causes fluorophores to emit many more photons, generating much greater signal intensity relative to normal camera-imaging noise and, hence, facilitates millisecond timescale imaging of single fluorescently labeled proteins. The millisecond timescale is fast enough to keep up with the diffusional motion present in the cytoplasm of cells and can also sample the fast molecular transitions that occur, particularly during signal transduction. Currentgeneration solid-state lasers are now easily capable of hundreds of milliwatts output, allowing a larger excitation field. This is complemented by the improved sensitivity of back-illuminated sCMOS cameras, which have faster frame rates and larger sensors. Single fluorescent proteins or complexes of proteins can be considered point sources of light and so appear as spatially extended spots in a fluorescence image due to diffraction by the microscope optics [13]. Narrow-field microscopy data consists of a time series of images of spots which require a significant amount of in silico analysis. Spots must be identified by software, the intensity of these spots quantified to calculate their stoichiometry, and their position tracked over time to produce a trajectory. We have applied narrow-field microscopy to glucose signal transduction in budding yeast, Saccharomyces cerevisiae. All cells dynamically sense their environment through signal transduction mechanisms. The majority of these mechanisms rely on gene regulation through cascades of protein-protein interactions which transmit signals from sensory elements to responsive elements within each cell. The Mig1 protein is an essential transcription factor in this mechanism in yeast. Mig1 is a Cys2-His2 zinc finger DNA-binding protein [14] which binds several glucose-repressed promoters [15–18]. In the presence of extracellular glucose it is poorly phosphorylated and predominantly located in the nucleus [19, 20] where it recruits a repression complex to the DNA [21]. If extracellular glucose concentration levels are depleted, Mig1 is phosphorylated by the sucrose non-fermenting protein (Snf1) [22–24], resulting in a redistribution of mean localization of Mig1 into the cytoplasm [19, 25, 26]. Thus, Mig1 concentration levels in the cell nucleus and cytoplasm serve as a readout of glucose signal transduction in budding yeast [27]. Mig1 has been labeled with the green fluorescent protein, GFP, and in the same strain, a ribosome component, Nrd1, almost completely localized to the

Single-Molecule Narrow-Field

7

nucleus, has been labeled with the mCherry fluorescent protein [20]. We have used narrow-field microscopy to track single Mig1GFP complexes as they diffuse in the nucleus and cytoplasm in the presence and absence of extracellular glucose. We discovered that Mig1 forms clusters responsible for gene regulation [28] and now a new picture of transcription factor condensates and clusters is emerging [29–31]. Our methods have also extended into 3D imaging [32] and to use Mig1 stoichiometry to “barcode” the gene regulatory state of the cell [33]. Here we describe in detail how to obtain single-molecule data of fluorescently labeled transcription factors in live yeast but also methods used to analyze the data obtained. We describe ADEMS code [34], the custom-parallelized MATLAB software we have created to track fluorescent molecules and quantify their stoichiometry, which we have also recently adapted into Python, with significant speed increases, in a new package called PySTACHIO [35]. We also show how fluorescence images of Mig1-GFP and Nrd1-mcherry can be segmented to identify the boundary of the cell and nucleus, respectively, and thus how trajectories can be categorized by their different subcellular compartments.

2

Materials

2.1 Fluorescently Labeled Yeast Strains

1. MATa MIG1-GFPHIS3 NRD1-mCherry-hphNT1METLYS S. cerevisiae strain in the BY4741 background [20]. 2. Yeast strains stored at with 20% glycerol.

80  C in YNB media supplemented

3. Yeast nitrogen base (YNB) media (Formedium) pH 6. 4. Yeast nitrogen base (YNB) media (Formedium) pH 6 supplemented with 4% glucose. 5. 2 Yeast nitrogen base (YNB) media (Formedium) pH 6 supplemented with 8% glucose. 2.2 Sample Preparation

1. Standard microscope slides (Fisher). 2. Coverslips (Menzel-Gl€aser). 3. Gene frames (17 mm  28 mm) and spreaders (Thermo Scientific). 4. 2% Agarose solution. 5. Plasma cleaner (Harrick Plasma PDC-32G). 6. Desktop centrifuge (Sigma 1-14).

2.3 Narrow-Field Microscope

For narrow-field microscopy, ~6 W/cm2 excitation light must be delivered to the sample centered at 488 nm for millisecond imaging of GFP. This must be combined with a high-speed (kHz) camera

8

Adam J. M. Wollman and Mark C. Leake

capable of detecting single fluorescent proteins. Our microscope is constructed from the following: 1. A Zeiss microscope body with a 100 TIRF 1.49 numerical aperture (NA) Olympus oil immersion objective lens and an xyz nano-positioning stage (Nanodrive, Mad City Labs). 2. 50 mW Obis 488 nm and 561 nm lasers for fluorescence excitation. 3. A dual-pass GFP/mCherry dichroic with 25 nm transmission windows centered on 525 nm and 625 nm was used underneath the objective lens turret. 4. A high-speed camera (Prime 95B, Teledyne Photometrics, USA) was used to image at typically 5 ms/frame with the magnification set at ~50 nm per pixel. 5. Our camera integrates dynamic I/O allowing direct digital modulation of the lasers. This allowed camera acquisition to be synchronized with laser one time, removing blank frames at the start of the acquisition. 6. The camera sensor was split between a GFP and mCherry channel using a bespoke color splitter consisting of a dichroic centered at pass wavelength 560 nm and emission filters with 25 nm bandwidths centered at 525 nm and 594 nm. 7. The microscope was controlled using Micromanager [36]. 2.4 Computational Analysis

3

1. We use MATLAB 2020b, which has the advantage that many functions are built in such as curve fitting, and code can be trivially parallelized. We have also recently implemented the same methods in Python through our new PySTACHIO package [35].

Methods

3.1 Growing Yeast Strains

1. Prepare 5 mL YNB supplemented with 4% glucose in 15 mL Falcon tube. 2. Scrape a small quantity of frozen yeast culture from the cryovial using a pipette tip without defrosting the vial. 3. Swirl pipette tip in prepared media and leave the culture to grow overnight at 30  C, shaking at 200 rpm.

3.2 Plasma Cleaning Coverslips

Glass coverslips must be plasma cleaned before use to remove any material left on the glass from manufacture. We have found this material to be fluorescent under the intense excitation light of narrow-field microscopy. This produces a fluorescent background in an image or false-positive spots.

Single-Molecule Narrow-Field

9

1. Place coverslips into the plasma cleaner. 2. Seal with the valve door and turn on the vacuum pump, pressing the edges of the door to insure a good seal. 3. Turn the radio-frequency (RF) generator to high. If the vacuum pressure is low enough, a pink-violet glow will emanate from the chamber, but if the pressure is not yet low enough, continue to press on the seal until more air is evacuated by the pump. 4. Once plasma is generated, slightly open the valve to allow a small amount of air into the chamber. The plasma will change color to violet indicating oxygen plasma. 5. After coverslips have been exposed to oxygen plasma for ~1 min, turn off the RF and vacuum pump and slowly open the valve. 6. Once atmospheric pressure is restored, remove the coverslips and store in a clean petri dish. 3.3 Preparing Cell Samples

1. For glucose conditions, the overnight culture can be used as the cell sample. 2. For absence of glucose conditions, prepare 1 mL of YNB. 3. Pellet 1 mL of cell culture by spinning in a desktop centrifuge for 2 min at 3000 rpm and discard the supernatant. 4. Resuspend in 500 μL YNB and pellet again. 5. Discard the supernatant and resuspend in 100 μL YNB or more depending on the cell density.

3.4 Agarose Pad Preparation

Cells were imaged on agarose pads suffused with media. This ensures that cells remain healthy during imaging while also immobilizing them. Figure 1 illustrates agarose pad assembly.

Fig. 1 Schematic of agarose pad assembly

10

Adam J. M. Wollman and Mark C. Leake

1. Prepare 500 μL 2YNB, supplemented with 8% glucose, to image cells in high-glucose conditions. 2. Remove the larger of the two clear plastic covers from the gene frame and apply the frame to a glass slide to create a rectangular well. 3. Melt 2% agarose solution in a microwave. 4. Pipette 500 μL of hot agarose into the 2YNB and quickly mix before removing 500 μL and pipetting into the well on the slide. 5. Quickly apply a plastic spreader (included with Gene Frames) to the well to remove excess agarose and leave a thin even layer in the well. 6. Slide the spreader off the pad carefully and remove the second plastic cover from the gene frame. 7. Pipette 5 μL of cell sample onto the pad in ~10 droplets and leave to dry for ~5 min before covering with a plasma-cleaned coverslip. 3.5 Obtaining SingleMolecule Data

1. Place the sample on the microscope and locate a cell or a cluster of cells by imaging in bright field. 2. Focus on the mid body of the cell by adjusting the focus to the point of minimum contrast. 3. Acquire 10 bright-field frames at 50 ms exposure time with no camera gain. All images are saved with raw pixel values as stacked Tag Image Format (TIF) files using the open microscopy environment (OME) standard. 4. Turn off the bright field and set gain to maximum. 5. Acquire 100 frames at 5 ms exposure time with the 561 nm laser at 15 mW power to obtain images of the mCherry signal; this will be used to define the nucleus. 6. Acquire 1000 frames at 5 ms exposure time with the 488 nm laser at 30 mW power to obtain a time series of GFP fluorescence images (see Note 1). 7. Repeat for ~30 cells or however many are required for a robust statistical sample (see Note 2).

3.6 Tracking Single Molecules

Bespoke MATLAB code has been written to track bright spots in fluorescence image time series. The steps performed by the code are outlined here. 1. Load TIF file containing single-molecule tracks, in this case the 1000 frame 488 nm exposure, into MATLAB as an mnp array, m pixels by n pixels by p frames. 2. Apply a top hat transformation to each frame to even the background and threshold the resulting image using Otsu’s method.

Single-Molecule Narrow-Field

11

Fig. 2 Schematic of iterative Gaussian masking to determine spot centroid. Left: A fluorescence image of Mig1-GFP in a yeast cell with a spot identifiable. Right: Schematic illustration of iterative Gaussian masking to find the spot centroid. The spot pixels are shown as a 3D surface and convolved with a 2D Gaussian centered on the centroid estimate. The resulting image is used to find a new centroid estimate and the process iterated until convergence

3. Dilate the resulting binary image with a disk-shaped structural element and then erode with the same element to remove bright noise pixels. 4. Perform an ultimate erosion to leave only nonzero pixels at potential spot locations (see Note 3). 5. At each of these potential spot locations, define a square 16-pixel region of interest (ROI) and a 5-pixel radius circular ROI centered on the intensity centroid of the potential spot. 6. Convolve the circular ROI with a 2D Gaussian function with 3-pixel width centered on the current intensity centroid and use this to determine a new intensity centroid. 7. Repeat step 6 until the centroid position converges on the final sub-pixel spot center coordinates. Figure 2 illustrates one iteration of a Mig1-GFP spot. 8. Define the spot’s total intensity as the sum of the pixel values inside the circular ROI corrected for background by subtracting the mean pixel value of the remaining pixels in the square ROI. 9. Define the spot’s signal-to-noise ratio (SNR) as the spot’s total intensity divided by the standard deviation of the remaining pixels in the square ROI and discard spots with SNR 2 and assign a new trajectory number or continue an existing one if a spot is already part of a trajectory. Thus, a complete set of trajectories for the time series is acquired. 3.7 Analyzing Single-Molecule Trajectories

Once trajectories are obtained, they are analyzed to determine the number of fluorophores present in each spot. The intensity of a single fluorophore is first found from the distribution of all spot intensities over the whole time series. The most common intensity value is that of a single fluorophore as all traces bleach to this value. The number of fluorophores present in each spot can be determined by dividing the initial value by the single fluorophore value. The initial intensity is determined by fitting an exponential decay function to each trace. 1. Generate histogram or kernel density estimation (KDE) of all the intensity values of all the spots found in trajectories. Figure 3, left, shows the KDE of mig1-GFP intensity values obtained in a single cell. 2. As every complex of fluorophores is bleaching, the most common intensity value in the intensity distribution is the characteristic single fluorophore intensity and the peak value in the distribution. Figure 3, right, shows spot intensity as a function of time with the single GFP value marked with a line (see Note 4).

Fig. 3 Left spot intensity as a function of time with the single GFP intensity value marked as a dotted line. Right distribution of spot intensities with the single GFP value marked

Single-Molecule Narrow-Field

13

3. Either fit an exponential to all the spot intensity values as a function of time. Use the time constant from the global fit, to fit exponentials to each trajectory, provided that it is >3 points within the first 200 frames and its initial intensity is >2 the single fluorophore intensity. 4. Or approximate the fit using a simple linear regression to the first 4 points of the trajectory. This approximation works well, both during the exponential phase of a photobleach (which is approximately linear initially) and during stepwise photobleaching. 5. The number of fluorophores present in each spot is the initial intensity in each trajectory’s fit divided by the characteristic single fluorophore intensity (see Note 5). 3.8 Categorizing Tracks by Cell Compartment

Trajectories are analyzed in terms of their location in the cell. The cell and nuclear boundaries are determined by segmenting GFP and mCherry fluorescence frame-averaged images. This allows trajectories to be defined as nuclear, cytoplasmic, and trans-nuclear. Figure 4 shows example frame averages and segmentation and categorized trajectories overlaid on a bright-field image of a yeast cell. 1. Generate frame averages over the first five frames of the Nrd1mCherry and Mig1-GFP acquisitions. These will be used to determine the nucleus and cell boundary. 2. Threshold these images above the full width half maximum (FWHM) of the background peak in the pixel intensity distribution to obtain a binary image. 3. Erode this image with a 4-pixel disk-shaped structural element to remove any bright single pixels and create a smoother edge around the object. These masks define the nucleus and cell pixels. 4. Divide the tracks into those which are always in the nucleus, always in the cytoplasm, or at different times—trans-nuclear tracks.

4

Notes 1. During fluorescence acquisitions, the camera must be synchronized with the laser or begin acquiring frames just before the laser illuminates the sample. This allows all of the emitted light from the sample to be captured by the camera which is crucial for analysis. 2. When moving through the sample during imaging, care must be taken to move in a set direction through the sample. This

14

Adam J. M. Wollman and Mark C. Leake

Fig. 4 Top: Fluorescence image of Mig1-GFP. Middle: Fluorescence image of Nrd1-mCherry. Bottom: Bright-field image with cell and nucleus outlines (yellow and cyan, respectively) and nuclear, cytoplasmic, and trans-nuclear tracks (red, green, and blue, respectively) overlaid

ensures that every cell that is imaged has not had any prior exposure to the laser and thus no photobleaching has occurred. 3. Steps 2–4 in Subheading 3.6 identify candidate spots in fluorescence images. This process is not strictly necessary as spots are evaluated using iterative Gaussian masking which could be performed at every pixel location in an image. This would be very computationally intensive but by identifying candidates, the number of possible spot locations is reduced along with the processing time. 4. The intensity of a single fluorophore can be independently verified in an in vitro assay by imaging antibody-immobilized fluorophores on a coverslip. This intensity will vary from that observed in vivo due to the different conditions inside a cell, particularly the differing pH.

Single-Molecule Narrow-Field

15

5. Any fluorophores which are within the diffraction-limited spot width of each other (~250 nm for GFP) will appear to be a single spot in a narrow-field image. This distance is large compared to the size of proteins such as Mig1 of a few nanometers and so it is possible that some spots in a frame are not molecular complexes but separate molecules within 250 nm of each other. These events are short lived for diffusing molecules and so the stoichiometry of complexes is obtained from the fits to the intensity vs. time traces.

Acknowledgments We thank Sviatlana Shashkova and Stefan Hohmann (University of Gothenburg, Sweden) for donation of yeast cell strains and assistance with yeast cell culturing. MCL was assisted by a Royal Society URF and research funds from the Biological Physical Sciences Institute (BPSI) of the University of York, UK. References 1. Wollman AJMM, Miller H, Zhaokun Z et al (2015) Probing DNA interactions with proteins using a single-molecule toolbox: inside the cell, in a test tube and in a computer. Biochem Soc Trans 43:139–145 2. Lenn T, Leake MC, Mullineaux CW (2008) Are Escherichia coli OXPHOS complexes concentrated in specialized zones within the plasma membrane? Biochem Soc Trans 36: 1032–1036 3. Plank M, Wadhams GH, Leake MC (2009) Millisecond timescale slimfield imaging and automated quantification of single fluorescent protein molecules for use in probing complex biological processes. Integr Biol (Camb) 1: 602–612 4. Chiu S-W, Leake MC (2011) Functioning nanomachines seen in real-time in living bacteria using single-molecule and super-resolution fluorescence imaging. Int J Mol Sci 12: 2518–2542 5. Robson A, Burrage K, Leake MC (2013) Inferring diffusion in single live cells at the singlemolecule level. Philos Trans R Soc Lond Ser B Biol Sci 368:20120029 6. Bryan SJ, Burroughs NJ, Shevela D et al (2014) Localisation and interactions of the Vipp1 protein in cyanobacteria. Mol Microbiol 94:1179–1195 7. Llorente-Garcia I, Lenn T, Erhardt H et al (2014) Single-molecule in vivo imaging of bacterial respiratory complexes indicates

delocalized oxidative phosphorylation. Biochim Biophys Acta 1837:811–824 8. Reyes-Lamothe R, Sherratt DJ, Leake MC (2010) Stoichiometry and architecture of active DNA replication machinery in Escherichia coli. Science 328:498–501 9. Badrinarayanan A, Reyes-Lamothe R, Uphoff S et al (2012) In vivo architecture and action of bacterial structural maintenance of chromosome proteins. Science 338:528–531 10. Stracy M, Wollman AJM, Kaja E et al (2019) Single-molecule imaging of DNA gyrase activity in living Escherichia coli. Nucleic Acids Res 47:210–220 11. Syeda AH, Wollman AJM, Hargreaves AL et al (2019) Single-molecule live cell imaging of Rep reveals the dynamic interplay between an accessory replicative helicase and the replisome. Nucleic Acids Res 47:6287–6298 12. Sun Y, Wollman AJM, Huang F et al (2019) Single-organelle quantification reveals stoichiometric and structural variability of carboxysomes dependent on the environment. Plant Cell 31:1648–1664 13. Wollman AJM, Nudd R, Hedlund EG et al (2015) From Animaculum to single molecules: 300 years of the light microscope. Open Biol 5: 150,019 14. Lundin M, Nehlin JO, Ronne H (1994) Importance of a flanking AT-rich region in target site recognition by the GC box-binding

16

Adam J. M. Wollman and Mark C. Leake

zinc finger protein MIG1. Mol Cell Biol 14: 1979–1985 15. Nehlin JO, Carlberg M, Ronne H (1991) Control of yeast GAL genes by MIG1 repressor: a transcriptional cascade in the glucose response. EMBO J 10:3373–3377 16. Klein CJL, Olsson L, Nielsen J (1998) Glucose control in Saccharomyces cerevisiae: the role of MIG1 in metabolic functions. http://www. ncbi.nlm.nih.gov/pubmed/9467897 17. Ghillebert R, Swinnen E, Wen J et al (2011) The AMPK/SNF1/SnRK1 fuel gauge and energy regulator: structure, function and regulation. FEBS J 278:3978–3990 18. Broach JR (2012) Nutritional control of growth and development in yeast. Genetics 192:73–105 19. De VMJ, Waddle JA, Johnston M (1997) Regulated nuclear translocation of the Mig1 glucose repressor. Mol Biol Cell 8:1603–1618 20. Bendrioua L, Smedh M, Almquist J et al (2014) Yeast AMP-activated protein kinase monitors glucose concentration changes and absolute glucose levels. J Biol Chem 289: 12,863–12,875 21. Treitel MA, Carlson M (1995) Repression by SSN6-TUP1 is directed by MIG1, a repressor/ activator protein. Proc Natl Acad Sci U S A 92: 3132–3136 22. Smith FC, Davies SP, Wilson WA et al (1999) The SNF1 kinase complex from Saccharomyces cerevisiae phosphorylates the transcriptional repressor protein Mig1p in vitro at four sites within or near regulatory domain 1. FEBS Lett 453:219–223 23. Ostling J, Carlberg M, Ronne H (1996) Functional domains in the Mig1 repressor. Mol Cell Biol 16:753–761 24. Ostling J, Ronne H (1998) Negative control of the Mig1p repressor by Snf1p-dependent phosphorylation in the absence of glucose. Eur J Biochem 252:162–168 25. Frolova E (1999) Binding of the glucosedependent Mig1p repressor to the GAL1 and GAL4 promoters in vivo: regulation by glucose and chromatin structure. Nucleic Acids Res 27: 1350–1358 26. DeVit MJ, Johnston M (1999) The nuclear exportin Msn5 is required for nuclear export

of the Mig1 glucose repressor of Saccharomyces cerevisiae. Curr Biol 9:1231–1241 27. Wollman AJMJM, Leake MCC (2015) Millisecond single-molecule localization microscopy combined with convolution analysis and automated image segmentation to determine protein concentrations in complexly structured, functional cells, one cell at a time. Faraday Discuss 184:401–424 28. Wollman AJ, Shashkova S, Hedlund EG et al (2017) Transcription factor clusters regulate genes in eukaryotic cells. elife 6:e27451 29. Cho WK, Spille JH, Hecht M et al (2018) Mediator and RNA polymerase II clusters associate in transcription-dependent condensates. Science (80-) 361:412–415 30. Chong S, Dugast-Darzacq C, Liu Z et al (2018) Imaging dynamic and selective low-complexity domain interactions that control gene transcription. Science (80-) 361: eaar2555 31. Sabari BR, Dall’Agnese A, Boija A et al (2018) Coactivator condensation at super-enhancers links phase separation and gene control. Science (80-) 361:eaar3958 32. Wollman AJM, Hedlund EG, Shashkova S et al (2020) Towards mapping the 3D genome through high speed single-molecule tracking of functional transcription factors in single living cells. Methods 170:82–89 33. Shashkova S, Nystro¨m T, Leake MC et al (2021) Correlative single-molecule fluorescence barcoding of gene regulation in Saccharomyces cerevisiae. Methods 193:62–67 34. Miller H, Zhou Z, Wollman AJM et al (2015) Superresolution imaging of single DNA molecules using stochastic photoblinking of minor groove and intercalating dyes. Methods 88: 81–88 35. Shepherd JW, Higgins EJ, Wollman AJM et al (2021) PySTACHIO: Python Single-molecule TrAcking stoiCHiometry Intensity and simulatiOn, a flexible, extensible, beginner-friendly and optimized program for analysis of singlemolecule microscopy data. Comput Struct Biotechnol J 19:2021.03.18.435952 36. Edelstein A, Amodaj N, Hoover K, et al (2010) Computer control of microscopes using manager, /pmc/articles/PMC3065365/

Chapter 3 Convolutional Neural Networks for Classifying Chromatin Morphology in Live-Cell Imaging Kristina Ulicna, Laure T. L. Ho, Christopher J. Soelistyo, Nathan J. Day, and Alan R. Lowe Abstract Chromatin is highly structured, and changes in its organization are essential in many cellular processes, including cell division. Recently, advances in machine learning have enabled researchers to automatically classify chromatin morphology in fluorescence microscopy images. In this protocol, we develop userfriendly tools to perform this task. We provide an open-source annotation tool, and a cloud-based computational framework to train and utilize a convolutional neural network to automatically classify chromatin morphology. Using cloud compute enables users without significant resources or computational experience to use a machine learning approach to analyze their own microscopy data. Key words Machine learning, Live-cell imaging, Cell cycle, Image analysis, Computational biology

1

Introduction Chromatin, the higher order structure of nuclear DNA and proteins, constitutes the majority of the nucleus of eukaryotic cells. Changes in chromatin structure lay the basis of many essential nuclear processes, including transcription, meiosis, mitosis, and apoptosis [1].

1.1 Chromatin Morphology Changes Throughout Cell’s Life Cycle

The cell cycle is the coordination of highly controlled molecular processes which result in the replication of the cell’s genetic material and its equal partitioning into two daughter cells upon division (mitosis) [2]. The cell cycle stage leading to mitosis is referred to as interphase, the longest phase between subsequent cell divisions defined by cell growth and DNA replication event. The interphase

Kristina Ulicna, Laure T. L. Ho and Christopher J. Soelistyo and Nathan J. Day contributed equally. Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_3, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

17

18

Kristina Ulicna et al.

is then followed by mitosis, during which the cell’s DNA is compacted, partitioned, and distributed into the newly formed cells. Although mitosis is a continuous process, it can be subdivided into discrete phases such as pro(meta)phase, characterized by chromosome condensation and nuclear envelope breakdown; metaphase, during which duplicated chromosomes align to the metaphase plate and their kinetochores attach to centrosome microtubules; and finally ana(telo)phase, described by sister chromatid separation into newly formed daughter cells, where new nuclear envelope is formed and cytokinesis begins [2]. As a quality control mechanism, the cell can decide to undergo apoptosis, which represents a highly regulated cell signaling process of biochemical and/or mechanical nature leading to programmed cell death. The progression through all cell cycle stages is an orchestrated series of events with visually distinct morphological changes to the cell nucleus (Fig. 1). To visualize the nuclear dynamics, various fluorescent dyes have historically been used, including DAPI [3] and Hoechst 33342 [4]. These permeate the cell membrane and directly bind to nucleic acids, which limits their application to fixed samples or to relatively short imaging periods (several hours to days), respectively. However, for longer term live-cell microscopy over several cell divisions across multiple generations, it is preferable to engineer cell lines to stably express fluorescent reporter(s) of nuclear architecture. This can be achieved by hijacking the primary function of core histone proteins, namely H2A, H2B, H3, and H4 [5], which bind to DNA to form nucleosome building blocks of chromatin, a higher order of DNA structure [6]. Therefore, overexpression of fusion histone proteins with an attached fluorescent tag, such as GFP or mCherry proteins, offers real-time capturing of the nuclear distribution, condensation level, or apoptosis-triggered degradation of chromatin in live eukaryotic cells. In experiments which require imaging of live-cell samples containing hundreds to thousands of single cells over long durations, manual annotation of each individual cell’s stage at any given time point represents a laborious, time-consuming, and error-prone task due to the high volume of single-cell observations. Because of the characteristic appearance of the chromatin throughout the cell cycle, automated labelling of microscopy image patches cropped around cell nuclei is possible with computer vision approaches, which also significantly speeds up the analysis of high microscopy volumes. 1.2 Machine Learning Strategies for Cell-State Classification

In recent years, several machine learning-based methods have been developed to classify cells according to their morphology. For example, a support vector machine (SVM) was implemented to classify fluorescence images of HeLa cells into one of the eight states (interphase, prophase, prometaphase, metaphase, early

CNN Annotator of Chromatin Morphology

interphase

A)

prometaphase

P

M

metaphase

anaphase

19

interphase

B)

A

I

WT

I

+FP

mitosis NEB

NEF H1

G2phase

cell cycle

Sphase

H2A H2B

G1phase

H3 H4 FP

checkpoint

DNA nucleosome

apoptosis

histone octamer

chromatin (DNA + protein)

interphase

Fig. 1 Chromatin morphology during cell cycle. (a) Cell cycle progression with phase-specific changes to the H2B-GFP-labeled chromatin condensation level. Single-cell time-lapse image patches capturing the process from nuclear envelope breakdown (NEB) to nuclear envelope formation (NEF) are illustrated (top). During interphase, the G1, S, and G2 phases are separated by cell cycle checkpoints that regulate progression through the cycle; alternatively, the cell may undergo apoptosis, or programmed cell death (bottom). Images taken every 4 min. Scale bar ¼ 5 μm. (b) Chromatin labeling with a fluorescent histone reporter. A vector with gene encoding for a chosen histone protein (H2B) fused to a visualization tag, such as green fluorescent protein (GFP), can be stably transfected into a cell line. Overexpression of H2B-FP under strong promoter control encourages the incorporation of the tagged histone protein over the wild-type version into the nascent histone octamer, which the nuclear DNA wraps around to form the resulting higher chromatin structure. The GFP tag is associated with the nucleic acid into nucleosome structures, which allows for easy chromatin visualization in live cells using fluorescence microscopy modality

anaphase, late anaphase, telophase, or apoptosis) [7]. Fluorescence images, exhibiting the cells’ chromatin morphology, were embedded in a multidimensional feature space. These quantitative features included the circularity and perimeter length of the cell. The SVM then deduced the optimal hyperplane(s) within the feature space separating data points of different categories. Another algorithm commonly used is a Random Forest classifier, a variant of the Decision Tree algorithm that—like the SVM— classifies data points based on their associated features. Ilastik is a

20

Kristina Ulicna et al.

bioimage analysis tool that uses the Random Forest approach to classify objects or individual pixels according to a set of user-defined categories (e.g., “nucleus,” “cytoplasm”) and manual labels [8]. For individual pixels, the features obtained are the output of image filters, and they include pixel intensity and pixel edge-ness, among others. Convolutional neural networks (CNNs) are deep learning algorithms that have also been used to classify cells based on their morphology. A major advantage of CNNs is that, unlike other approaches, they do not require feature engineering—the network learns to extract image features that fulfill the task. Over the last decade, this network architecture has demonstrated significant utility in classifying images across many different image domains [9, 10]. As an example from the microscopy domain, CNNs have been used to classify actin-labeled fluorescence images of human breast cells as belonging to either “normal” (MCF-10A) or “cancerous” (MCF-7 or MDA-MB-231) cell lines [11]. The high performance of the model—more accurate than that of human experts—demonstrates the potential of CNNs to correctly classify cells according to actin filament structure. As we will demonstrate, this capability can be applied to chromatin structure classification. We have previously used CNNs to classify chromatin morphology in MDCK cell lines [10, 12]. In this protocol, we provide the tools and a guide on how the user can create a dataset of single-cell image patches from an example dataset, and train a convolutional neural network to automatically assign labels to single cells throughout the entire cell cycle progression (Fig. 2). We also describe how to evaluate the network performance, infer new classifications on previously unseen examples in a fully automated manner, and visualize the encoded image representations in reduced dimensionality embeddings. 1.3 Exploratory Visualization of CNN Feature Embedding

Single-cell data is often high-dimensional and requires tools to enable exploratory visualization. To tackle dimensionality problems encountered when analyzing high-dimensional data as we are doing here, dimensionality reduction techniques have been developed and are now applied in various areas [13, 14]. These approaches reduce the high number of dimensions by transforming the data into lower dimensional space while attempting to retain intrinsic features of the original data [15]. This low-dimensional embedding is usually 2D or 3D, which greatly facilitates visualization and interpretation of the high-dimensional data. Techniques broadly use two approaches, either by trying to preserve the overall data distance structure, e.g., in the widely used principal component analysis (PCA) [16], or by prioritizing local neighbor distances over global distances, e.g., in t-Distributed Stochastic Neighbor Embedding (t-SNE) [17] and Uniform Manifold Approximation and Projection (UMAP) [18].

CNN Annotator of Chromatin Morphology

21

A) Acquisition

Annotation

Training

Validation

Inference

D_CNN_Inference_and_Embedding.ipynb A_CNN_Annotation_ with_Napari.ipynb Image cultured cells via time-lapse fluorescence microscopy

Annotate new training images to provide ground truth labels

C_CNN_Training_ and_Validation.ipynb Set CNN model hyperparameters & train the neural network

Measure model performance by standard metrics (confusion matrix)

Yes

Use the model to visualize clustering

B_CNN_Data_Preview_Images.ipynb

No

B) Multi-well plate with seeded cells

Use the model to classify new images

Is the CNN performance suitable?

Fluorescence image sequence of H2B-GFP labelled MDCK cells

64x64 px image patch

Convolutional neural network

“Ground truth” label

“Anaphase”

Automated fluorescence microscopy

~80% training set ~20% testing set time

MaxPool SoftMax

Fully connected

Conv 2D Batch Norm ReLU

Fig. 2 Overview of the protocol pipeline. (a) Schematic representation of the protocol pipeline involved in acquiring data, training a convolutional neural network for classification, and its utilization in inference of chromatin morphology in live-cell microscopy data. Corresponding Jupyter notebooks for each part of this protocol are color-coded accordingly. (b) Practical implementation of the protocol. First, time-lapse microscopy data of live proliferating cells in culture are acquired. Next, users annotate a sample of single-cell images to generate training (~80%) and testing (~20%) sets of images, with the associated ground-truth labels. Finally, these image-label pairs are used to train a convolutional neural network (CNN) as illustrated. The CNN consists of several convolutional layers that extract image features, followed by densely connected layers that transform the features into a categorical label. Conv 2D, two-dimensional convolution operation; batch norm, batch normalization; ReLU, rectified linear unit. Scale bar ¼ 50 μm

PCA is a common linear method where high-dimensional data are projected onto the axes that can explain the greatest underlying variance [16]. It also allows to quantify the individual contribution of input variables, thus making this method one of the most interpretable and commonly used approaches. However, due to the complex nonlinear nature of microscopy imaging data, it is often necessary to use nonlinear techniques [15, 19] such as t-SNE or UMAP to properly interpret our CNN’s latent representations of the training images. Although t-SNE and UMAP yield similar low-dimensional embeddings, UMAP provides an added feature to reuse a computed transform function on new unseen data in order to project it into the same low-dimensional space [18]. UMAP has also been proposed to have quicker performance than t-SNE and other techniques when scaling to considerably larger datasets [14, 18]. In this protocol, we use UMAP to visually validate whether the CNN has

22

Kristina Ulicna et al.

successfully learned relevant image features to correctly classify the different cell cycle stages. However, these approaches can also be extended to visualize orthogonal reporters. For example, using reporters for both chromatin morphology and nuclear envelope, one could cluster cells by chromatin morphology but visualize the structure of the nuclear envelope associated with each mitotic stage.

2

Materials 1. This protocol relies on a source of high-quality fluorescence microscopy data. These are typically time-lapse image sequences in two or three spatial dimensions where at least one channel is a fluorescent marker for chromatin (see Note 1). Each image should be stored as a multidimensional TIFF file in the standard format (see Note 2). 2. A computer connected to the internet, capable of running a web browser. 3. An installation of Python (see Note 3) of version 3.7 or higher on your machine (see Note 4). 4. A Google account to train models in the cloud using the provided Google Colab environment (see Note 5). 5. Follow the installation instructions on the GitHub repository to create a clean virtual environment with all needed Python libraries and our cnn-annotator package: https://github.com/ lowe-lab-ucl/cnn-annotator (see Note 6).

3

Methods

3.1 Annotating Training Images Using Napari

1. At the command line, make sure that you are in your local cnnannotator directory and type jupyter notebook; you will then be redirected to a Jupyter notebook environment in your default browser. Navigate to the /notebooks/ directory where the repository structure is outlined. 2. Open A_CNN_Annotation_with_Napari.ipynb by clicking on the file name. A separate tab will open. 3. In the notebook, modify the file path to your image data (default: /data/MDCK_H2B_GFP_movie.tif). 4. If you are using your own classification labels, create those labels in the notebook using the CellState structure (see Note 7). 5. Run the notebook and a separate Napari instance with a GUI window will pop up (see Note 8).

CNN Annotator of Chromatin Morphology

A)

Napari interface for single-cell annotation

B)

23

Training image examples annotated in Napari ”Pro”

”Int”

”Met”

”Pro”

”Apo”

”Ana”

”Apo”

”Int”

”Met”

”Ana”

Int - interphase, Pro – prometaphase, Met – metaphase, Ana– anaphase, Apo - apoptosis

Fig. 3 Napari widget for single-cell annotation. (a) A screenshot of the graphical user interface (GUI) provided by Napari, with the custom-designed annotation widget provided as part of this protocol. The widget allows users to annotate their images with categorical labels representing different chromatin morphologies, and to export these single-cell image patches and annotations for further analysis and training of the CNN using cloud computation. (b) Representative single-cell image patches and annotated label pairs serving as ground-truth dataset extracted using the annotation widget

6. Inside Napari, select a label corresponding to a particular chromatin morphology and annotate the image data by placing a small marker at the center of each cell with the appropriate morphology (Fig. 3) (see Note 9). 7. Aim to annotate several examples of each state that you wish to classify (see Note 10). 8. Export the data using the “export” button. This writes a single zip file containing all of the image patches as well as the locations in T(Z)YX and ground-truth labels. This zip file can be used as a source of training data for the CNN (see Note 11). 9. You can close the Napari instance when done. Repeat steps 5– 8 to create additional annotation zip files for training, validation, and future inference (see Note 12). 10. (Optional) If you wish to assess the class balance and visualize your annotations, return to the Jupyter notebook interface, then open, and run B_CNN_Data_Preview_Images.ipynb. 3.2 Training a CNN for Chromatin Morphology Classification 3.2.1 Training the CNN on a Train Set

1. From the GitHub repository, click on the “Open in Colab” notebook for training (see Note 13). 2. Follow the instructions in the notebook, and upload the training and validation zip files to the Colab /content/train/ and /content/validation/ subfolders. 3. Make sure that you set aside some data for validation (see Note 14). 4. Set the hyperparameters required for training (see Note 15). 5. Run the Colab notebook to train the model.

24

Kristina Ulicna et al.

A)

B) Ground Truth Labels

False Pos

C)

In t False Pos

o Pr False Pos

t Me

Recall =

False Neg

A

Precision =

False Neg

False Neg

True Pos

False Neg

na False Pos

F1-Score = 2 *

=

*

o Ap In t

⁄

(

)

P ro

CNN Predicted Labels

t Me

An

a

Ap

o

F1 Score = 0.95

Fig. 4 Evaluation of CNN performance with multi-class categorization. (a) Interactive TensorBoard interface for visualizing the training of the CNN in real time. Metrics include the training loss, overall accuracy (shown), and a confusion matrix of training accuracy for each categorical class. (b) A confusion matrix for CNN network performance displaying high accuracy on the diagonal of the matrix, indicating agreement between the ground-truth labels (rows) and the CNN prediction labels (columns). Calculation of “anaphase” classspecific F1 score reflective of network’s recall (sensitivity) and precision (specificity), according to the equations shown in (c) 3.2.2 Evaluating the CNN on a Validation Set

1. Monitor the training performance using TensorBoard (see Note 16), which will include the calculation of a confusion matrix (Fig. 4) (see Note 17). 2. If the network performs satisfactorily on the separate validation data, it means it was sufficiently trained and can now be used to classify examples in new datasets (see Notes 18 and 19). 3. Save out and download the trained model (default name: model.h5) for future inference (see Note 20).

3.3 Using the CNN for Chromatin Morphology Classification on Unseen Data

Once the CNN has been successfully trained, the goal is to use it to explore unseen data. The inference network takes an unseen image of a cell nucleus and returns the prediction of its classification label.

3.3.1 Using the Trained CNN for Inference on a Test Set

1. From the GitHub repository, click on the “Open in Colab” notebook for inference (see Note 13). 2. Upload your trained model (default name: Colab default /content/ folder.

model.h5)

to the

CNN Annotator of Chromatin Morphology

A)

”Pro”

”Met”

”Int”

”Pro”

B)

”Apo”

”Ana”

”Apo”

Anaphase

”Int”

”Met”

25 ”Ana”

individual observations

Metaphase

representative patch

Apoptosis

Pro

separation between classes Int

Prometaphase

Interphase Fig. 5 2D representation of the CNN feature embedding. (a) Representative predictions from the trained network for the given images. (b) A two-dimensional image-based representation of the CNN output, created using the dimensionality reduction method UMAP. The image represents the manifold generated using 20,000 example images, with individual data points representing individual observations overlaid. Clear clustering of image patches with discrete chromatin morphologies (and their categorical labels) can be observed

3. Upload your test zip file(s) to the Colab subfolder.

/content/test/

4. Run the notebook to perform predictions on real data (Fig. 5) (see Note 21).

26

Kristina Ulicna et al.

5. Calculate the confusion matrix using the notebook (see Note 17). 6. Calculate additional performance metrics, including the F1 score, precision, and recall (see Note 22). 3.3.2 Visualizing the CNN Feature Embedding

4

1. (Optional) Perform UMAP dimensionality reduction along with a manifold projection of single-cell image patches to visualize the network’s classification in 2D (Fig. 5) (see Note 23).

Notes 1. Although we use H2B-GFP as a fluorescent marker proxy for chromatin organization, other fluorescent reporters, such as DAPI, can also be used. 2. We use TIFF stacks of image data for the annotation stage. These are usually derived from time-lapse microscopy acquisitions and are ordered as T(Z)YXC. Each channel, for example GFP, is a separate stack of images. The bright-field transmission data can also be added as an additional input to improve the accuracy of the classification. Although this protocol is primarily designed for two-dimensional datasets, the same approach can be utilized for volumetric imaging data. We provide some example image data in the GitHub repo. 3. We recommend installing a complete scientific distribution of Python, such as Anaconda. Follow the installation instructions for your platform here: https://www.anaconda.com/ products/individual. 4. You can check which version of Python you have installed by typing python -version at the command line. We recommend version 3.7 or greater. 5. You will need a Google email address and need to be signed in to run notebooks in the Google Colab environment. Sign up for an account here: https://accounts.google.com/signup. 6. Please note that we are constantly adding new features and updating the live repository. For the latest version of the protocol and more detailed instructions, please remember to check https://github.com/lowe-lab-ucl/cnn-annotator. You will also find a requirements.txt on the repo with which you can install the necessary packages using other means and then install the cnn-annotator package from GitHub. 7. We recommend using the following labels for annotating the data: 0—Interphase, 1—Prometaphase, etc. However, it is possible to generate your own labels depending on the goal of the classification task. In the notebook, we have defined five

CNN Annotator of Chromatin Morphology

27

states and given each a unique numerical value. However, it is possible to define your own states (for example, “5—Telophase”) by simply adding another entry to this structure, with a unique numerical value. 8. We use the multidimensional image viewer Napari [20] for data visualization. Read more about Napari here: https://napari. org/. 9. Annotations that are too close to the edge of the field of view will be exported but will be padded with zero values, and excluded from training by default. However, you are given the option of using or excluding these examples during training of the model. It is useful to provide examples when it is likely that your real dataset will also include examples that are close to the edge of the field of view. 10. You should aim to provide at least 500 examples of each class to properly train the network. This dataset should be free from errors and is referred to as the “ground-truth dataset.” 11. Annotation files will be named using a common format and saved to the local drive. Each file is a zip archive containing the raw image patches and the associated metadata. You can find several examples within the GitHub repo (default save path: / data/annotation_MM-DD-YYYY--hh-mm-ss.zip). 12. Each annotation file will be given a unique time and date stamp, allowing multiple datasets to be annotated. We recommend providing data from multiple different experiments to properly sample the real data distribution. This will prevent overfitting during training and present a generalizable model as the result. 13. The “Open in Colab” links can be found on the GitHub repository page: https://github.com/lowe-lab-ucl/cnnannotator. 14. In addition to the training dataset, it is important to reserve some example images for evaluating the resultant network. These images are “held out” and not used during training. This would typically be 10–20% of your training set, and can be achieved by saving additional annotated images in Subheading 3.2. These annotations should be set aside and not used when training the network. This enables a fair evaluation of the network’s performance. 15. The following hyperparameters can be modified before training your model: BATCH_SIZE—Number of training images the model is trained on in one iteration, after which the model computes its loss before continuing to train.

28

Kristina Ulicna et al. BUFFER_SIZE—Number of elements in a buffer from which each training example will be picked at random for shuffling purposes; it is recommended to have a buffer size equal or larger to the full dataset size for optimal shuffling. TRAINING_EPOCHS—One training epoch is a full pass over the entire training set, i.e., the network will have seen every image once; more specifically one epoch consists of (dataset_size / BATCH_SIZE) training iterations. BOUNDARY_AUGMENTATION—If set to “True”: when building the training dataset, random simulations of cropped images that usually occur at the edge of imaging volumes will be added to the training dataset. This improves the generality of the model. INPUT_SHAPE—Dimensions of an input training image to the network, typically in the format (width, height, channels) for 2D datasets.

16. On TensorBoard, you should see a decrease in the loss and an increase in the accuracy as the network trains. This is an indication that the network’s performance on the validation set is improving. Click on the “SCALARS” tab to see the accuracy and loss during training. Click on the “IMAGES” tab to see the confusion matrix during training. 17. A confusion matrix is an NN table (where N is the number of classes) that is usually used to evaluate the performance of a classification network on a set of testing or validation data. The number in row A and column B of the table is equal to the number of testing examples of ground-truth class A that have been classified by the network as belonging to class B. Therefore, the numbers shown in the main diagonal of the table (from top-left to bottom-right) indicate those examples that have been classified correctly by the network. 18. Evaluating the network’s performance on a set of previously unseen data is a way of ensuring that the network is able to generalize, and is not simply “remembering” the examples in the training dataset. Hence, it is a way of ensuring that the network is able to conduct the task that it has been trained to do. In general, the acceptable level of accuracy is dependent on the task. For labeling image sequences, we routinely achieve accuracies >0.95 and often >0.99. 19. When using a trained network for inference on unseen data, we assume that the new data are independent and identically distributed (i.i.d.) to the training data. 20. The trained model is stored in the HDF5 file format and can be reused once trained. We recommend keeping a local copy of the trained model once you are satisfied with the performance.

CNN Annotator of Chromatin Morphology

29

21. The trained network can now be used to assign classification labels to cells in unseen sequences. This is very useful when trying to automatically extract division events from time-lapse sequences or for clustering image sets with orthogonal fluorescence markers. 22. The “recall” of a certain class C is the proportion of groundtruth examples of C that have been correctly classified. Thus, a low recall indicates that there are many ground-truth examples of C that have been incorrectly classified as belonging to another class. The “precision” of class C is the proportion of examples classified by the network as belonging to C that are actually ground-truth examples of C. Thus, a low precision indicates that there are many examples classified by the network as belonging to C, but that are actually ground-truth examples of other classes. The F1 score of class C is the harmonic mean of precision and recall, and acts as a metric for evaluating the performance of the network on that specific class. These three class-specific metrics are important because the overall “accuracy” of a network when applied to a testing or validation set describes the general performance of the network across all classes, and does not make clear the difference in performance between classes. By calculating these metrics, we may assess how well the network performs for any specific class. 23. By analyzing the UMAP manifold projection, we can see how image patches are localized in the feature embedding. Distinct clusters of images should correspond to the various defined classes, implying that the network has appropriately learned underlying image features to properly classify the patches. If images of the same class do not cluster as you would expect, one of the reasons could be that the n_neighbors parameter was not properly chosen. For more information about some important UMAP parameters, check out this documentation: https://umap-learn.readthedocs.io/en/latest/parameters. html#.

Acknowledgments This work was supported by BBSRC grant BB/S009329/1 to ARL. KU, LTLH, and CJS were supported by BBSRC London Interdisciplinary Doctoral Programme studentships. NJD was supported by an MRC DTP studentship (MR/N013867/1). We gratefully acknowledge Giulia Vallardi for providing example data and Guillaume Charras for discussions.

30

Kristina Ulicna et al.

References 1. Kaplan N, Moore IK, Fondufe-Mittendorf Y et al (2009) The DNA-encoded nucleosome organization of a eukaryotic genome. Nature 458:362–366 2. Mitchison TJ, Salmon ED (2001) Mitosis: a history of division. Nat Cell Biol 3:E17–E21 3. Kapuscinski J (1995) DAPI: a DNA-Specific Fluorescent Probe. Biotech Histochem 70: 220–233 4. Durand RE (1982) Use of Hoechst 33342 for cell selection from multicell systems. J Histochem Cytochem 30:117–122 5. Kornberg RD, Thomas JO (1974) Chromatin structure: oligomers of the histones. Science 184:865–868 6. Kornberg RD, Lorch Y (1999) Twenty-five years of the nucleosome, fundamental particle of the eukaryote chromosome. Cell 98: 285–294 7. Held M, Schmitz MHA, Fischer B, et al (2010) CellCognition: time-resolved phenotype annotation in high-throughput live cell imaging. Nat Methods 7:747–754 8. Berg S, Kutra D, Kroeger T et al (2019) Ilastik: interactive machine learning for (bio)image analysis. Nat Methods 16:1226–1232 9. Krizhevsky A, Sutskever I, Hinton GE (2017) ImageNet classification with deep convolutional neural networks. Commun ACM 60: 84–901 10. Bove A, Gradeci D, Fujita Y et al (2017) Local cellular neighborhood controls proliferation in cell competition. MBoC 28:3215–3228 11. Oei RW, Hou G, Liu F et al (2019) Convolutional neural network for cell classification

using microscope images of intracellular actin networks. PLoS One 14:e0213626 12. Ulicna K, Vallardi G, Charras G, et al (2021) Automated deep lineage tree analysis using a Bayesian single cell tracking approach. Front Comput Sci 3 https://doi.org/10.3389/ fcomp.2021.734559 13. Koch FC, Sutton GJ, Voineagu I, et al (2021) Supervised application of internal validation measures to benchmark dimensionality reduction methods in scRNA-seq data. Brief Bioinformatics 22:6 https://doi.org/10.1101/ 2020.10.29.361451 14. Becht E, McInnes L, Healy J et al (2019) Dimensionality reduction for visualizing single-cell data using UMAP. Nat Biotechnol 37:38–44 15. Van Der Maaten L, Postma E, Van den Herik J (2009) Dimensionality reduction: a comparative review. J Mach Learn Res 10:13 16. Hotelling H (1933) Analysis of a complex of statistical variables into principal components. J Educ Pyschol 24:417–441 17. Van Der Maaten L, Hinton G (2008) Visualizing data using t-SNE. J Mach Learn Res 9: 2579–2605 18. McInnes L, Healy J, and Melville J (2018) UMAP: uniform manifold approximation and projection for dimension reduction. arXiv preprint arXiv:1802.03426v2 19. Yin H (2007) Nonlinear dimensionality reduction and data visualization: a review. Int J Automat Comput 4:294–303 20. Sofroniew N, Lambert T, Evans K, et al (2021) napari/napari: 0.4.3rc0, Zenodo. doi: https:// doi.org/10.5281/zenodo.4435160

Chapter 4 Fluorescence Recovery After Photobleaching (FRAP) to Study Dynamics of the Structural Maintenance of Chromosome (SMC) Complex in Live Escherichia coli Bacteria Anjana Badrinarayanan and Mark C. Leake Abstract MukBEF, a structural maintenance of chromosome (SMC) complex, is an important molecular machine for chromosome organization and segregation in Escherichia coli. Fluorescently tagged MukBEF forms distinct spots (or “foci”) composed of molecular assemblies in the cell, where it is thought to carry out most of its chromosome-associated activities. Here, we outline the technique of fluorescence recovery after photobleaching (FRAP) as a method to study the properties of YFP-tagged MukB in fluorescent foci. This method can provide important insight into the dynamics of MukB on DNA and be used to study its biochemical properties in vivo. Key words Chromosome organization, MukBEF, E. coli, Fluorescence microscopy, FRAP

1

Introduction The bacterial chromosome is compacted nearly a 1000-fold into a cell where it is faithfully replicated, transcribed, and segregated. Not only is it highly compacted, but it is also spatially organized with chromosomal regions occupying specific positions inside the cell [1, 2]. The highly conserved structural maintenance of chromosome (SMC) complex, MukBEF, plays a central role in E. coli to maintain chromosome organization and ensure faithful chromosome segregation [3, 4]. The MukBEF complex consists of three proteins: The SMC-like MukB and two accessory proteins MukE and MukF. Deletion of any of these components results in a Muk
phenotype that includes temperature sensitivity, production of anucleate cells, and loss of wild-type chromosome organization. Functional fluorescent fusions of MukB, E, or F all form foci in

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_4, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

31

32

Anjana Badrinarayanan and Mark C. Leake

Fig. 1 Fluorescent fusions of MukBEF form foci in cells. YFP-tagged MukB, MukE, and MukF form foci in cells. Representative cells are shown in this figure. Fluorescent focus is highlighted with *

cells (Fig. 1), with two foci on average around the origin of replication [3, 5]. Recent studies using an array of microscopy-based approaches and genetic tools have provided insight into the properties of Muk foci and have supported the idea that foci are the centers of activity of the MukBEF complex [3–7]. The techniques used in these studies are widely applicable to understanding the functions of other proteins/protein complexes in vivo. In general, advances in live-cell imaging in combination with the use of green fluorescent protein (GFP) and its variants have facilitated the ability to study the composition and dynamics of protein complexes in a cellular context [8–19]. In this chapter, we describe the techniques of fluorescence recovery after photobleaching (FRAP) [20] and fluorescence loss in photobleaching (FLIP) to study the dynamics of YFP-tagged MukB, E, or F in foci [6]. During FRAP, a subset of fluorescent molecules (typically, molecules in one of the two Muk foci) are irreversibly photobleached using a laser beam with high intensity of illumination. After the brief pulse of bleaching, images are recorded for subsequent time frames at lower laser intensities to observe the recovery of fluorescence at the bleached spot either by diffusion of the non-bleached molecules or by active exchange of bleached molecules in the focus with unbleached ones. Since MukBEF forms two fluorescent foci, we can also record the loss in fluorescence of the unbleached focus (FLIP) during the photobleaching experiment. In an ideal scenario, the rate of recovery after photobleaching should be comparable to the loss in intensity of the unbleached focus. This chapter briefly describes the method to grow cells and prepare slides, similar to that described previously [21], and outlines a typical FRAP experiment as well as a simple method of data analysis. As stated earlier, the method can be modified to study other protein complexes as well. For these experiments, E. coli cells are grown under conditions that result in nonoverlapping replication cycles, so each cell has two Muk foci on average.

FRAP as a Tool to Study SMC Dynamics In Vivo

2

33

Materials Instructions for the construction of YFP-tagged MukBEF components are beyond the scope of this chapter. However, a note is included on strain construction that might be useful (see Note 1).

2.1

Growth Media

1. Luria Broth: 10 g Yeast extract, 5 g NaCl, and 5 g tryptone in 1 L of water. pH is adjusted to 7 and LB is sterilized by autoclaving. 2. 10XM9 salt solution: 63 g Na2HPO4, 30 g KH2PO4, 5 g NaCl, and 10 g NH4Cl in 1 L of water. Sterilize by autoclaving. 3. 1XM9-glycerol: 1XM9 salts, 0.5 mg/mL of thiamine, 0.1% 1 M MgSO4, 0.1% 100 mM CaCl2, and 0.2% of glycerol as carbon source in water. Make 100 mL of this solution. 4. 2XM9-glycerol: Same as 1XM9-glycerol but in half the quantity of water. Make 50 mL of this solution. 5. 2% Agarose: Invitrogen ultrapure agarose can be used for this. For 50 mL, add 1 g of agarose to 50 mL of water. 6. 1% Agarose + M9-glycerol (this mixture is used for microscope slide preparation): Mix melted 2% agarose with 2XM9-glycerol in a 1:1 ratio. I usually mix 500 μL of each solution by pipetting in an Eppendorf and immediately use this to prepare the microscopy slide.

2.2 Slides and Microscope

1. Microscope slides: VistaVision microscope slides (VWR). 2. Coverslips: Micro cover glasses, thickness 1.5, 24  50 mm (VWR). 3. Gene frame seals (Thermo Scientific Catalog number AB0578). 4. Microscope: UltraView PerkinElmer Spinning Disc Confocal microscope with FRAP module, 100 1.35NA oil immersion objective, electron-multiplying charge-coupled device (ImagEM, Hamamatsu Photonics), and UltraView PK Bleaching Device for photobleaching. Assuming that you will be imaging YFP-tagged MukBEF, the microscope should have laser lines for 514 nm (see Note 2 for alternative laser lines). 5. Immersion oil: Immersol W oil NA 1.339 (Zeiss). 6. Software: Volocity imaging software (PerkinElmer) for image acquisition and ImageJ for image analysis.

34

3

Anjana Badrinarayanan and Mark C. Leake

Methods

3.1 Preparation of Bacterial Cultures for Microscopy

1. Streak bacterial cells from a frozen stock on LB agar plates with appropriate antibiotic at 37  C. As far as possible, use fresh cells no more than 2 weeks old. All cultures are grown by shaking to provide sufficient aeration. Most cells can grow at 37  C (see Note 3 about growing ΔmukBEF cells). The steps listed below are for a strain carrying a YFP-tagged version of MukB. The same procedure can be followed for other tagged components of the complex. 2. Pick a single colony from the plate prepared in step 1 and resuspend it in 5 mL of LB. Allow the culture to grow until stationary phase (5–6 h). 3. Make a 1 in 5000 dilution of the above into 5 mL of 1XM9glycerol and allow this culture to grow overnight. 4. The following day, subculture the cells in fresh 1XM9-glycerol (~1 in 1000 dilution) and allow the cells to grow till OD 0.1–0.2 (measured using a spectrophotometer). This should take 2–3 h (see Note 4 for details about generation time of E. coli cells grown in M9-glycerol). 5. Spin down 500 μL of culture from step 4 at 8000 rpm for 1 min. Remove the supernatant and resuspend the pellet in 50 μL of 1XM9-glycerol. Cells are now ready to be spotted on the microscope slide and imaged.

3.2 Preparation of Microscopy Slide

The procedure described here has been previously outlined in detail in another volume of this series [21]. A condensed version of this protocol is provided below. 1. The gene frame is first stuck on a clean glass slide by removing its clear plastic cover. Make sure to stick the frame smoothly on all sides without leaving wrinkles (see Note 5 on why gene frames are used). 2. Take 500 μL of 1% agarose + M9-glycerol and immediately transfer it to the center of the gene frame prepared in the above step (see Note 6). 3. Place a coverslip on top of this, press it down to remove excess agarose, and flatten the solution evenly in the frame. Let this stand for a few minutes, until the agarose has dried and solidified. 4. Once the agarose has solidified, slide the coverslip off and let the agarose dry for a couple of extra minutes. 5. Take 5 μL of culture prepared earlier (step 5, Subheading 3.1) and spot it on the agarose. Try to evenly distribute it across the slide by applying multiple spots and tilting the slide to allow

FRAP as a Tool to Study SMC Dynamics In Vivo

35

spreading. Allow the slide to dry for a couple of more minutes. It is essential to do so as excess water will hamper step 6 of this section. 6. Remove the top plastic cover of the gene frame. On the sticky side of the frame carefully place a coverslip. Make sure that the coverslip is placed evenly and avoid the formation of air pockets. Once the coverslip has made contact with all four sides of the frame, you can press it down gently to even out its adhesion. 3.3

Microscopy

1. Turn on the lasers, microscope, and computer. Then turn on Volocity (the acquisition software). 2. Add a drop of immersion oil to the objective and place the slide on top of the lens. 3. Cells should be focused using bright field or DIC. Avoid focusing using fluorescence to prevent photobleaching. An ideal field of view for imaging should have cells evenly distributed and in focus. A typical field can have up to 50 cells. 4. Open the settings for YFP (514 nm laser) on Volocity and reduce laser power to 4–6% (see Note 7). Under camera settings, set the frame rate to 300 ms for image capture. Focus and take a picture. 5. In the picture, you will be able to see typically two distinct MukB-YFP foci per cell. The aim of the experiment is to bleach one of the two foci and record fluorescence recovery after bleaching (see Note 8 on the use of cephalexin to elongate cells). 6. Open the FRAP module to set up bleaching conditions (see Note 9 on FRAP calibration). Pulse bleach is ideally done with 6–15% laser intensity for 15 ms. The number of cycles of bleaching is limited to 1. A region of interest (ROI) is drawn around the focus to be bleached. This is usually a diffractionlimited region of ~300 nm (see Note 10 on the size of ROI). 7. Using the PhotoKinesis menu, choose up to six ROIs (one ROI per cell) in one field of view. ROIs can be chosen by drawing a region around a MukB-YFP focus. Then set the conditions for acquisition. Typically, take 2–3 pre-bleach images and after pulse-bleaching (step 6), record recovery of fluorescence every 15 s for 3 min or every 30 s for 5 min. Again, image capture should be done at lower laser intensity (4–6%) at a 300 ms capture rate. The entire module is automated. Once the settings have been applied and acquisition has started, images will be acquired in the sequence desired: two pre-bleached images, followed by pulse-photobleaching of the ROIs selected, followed by image capture with low laser

36

Anjana Badrinarayanan and Mark C. Leake

intensities for 3 or 5 min. Movies are saved as stack files that can be opened in ImageJ. 8. Repeat the above procedure after moving to a new field of view that is distinct from the field previously imaged (see Note 11). 3.4

Image Analysis

1. Open images (saved as a stack) in ImageJ. It is important to remove background fluorescence prior to extracting information on focus intensity. This is done using the background subtraction module in ImageJ. Apply subtraction to the entire stack. 2. For FRAP measurements draw a region of interest around the spot that was bleached in the experiments in Subheading 3.3. Also draw a second ROI around the entire cell to calculate total cellular fluorescence intensity for the cell undergoing pulsebleaching. Use ImageJ’s “Measure Intensity” tool to extract mean and total intensity values for each ROI through the entire stack. 3. For FLIP measurements, the same procedure (step 2) should be repeated for an ROI drawn around the unbleached focus in the cell undergoing pulse-bleaching. 4. FRAP, FLIP, and total cellular fluorescence intensities for a cell in a movie can now be copied and pasted into Excel. To compare recovery across cells, intensity of ROIs should be normalized to highest pre-bleach intensities. 5. Before calculating recovery times, it is important to correct for photobleaching due to fluorescence excitation during imaging. This is done by normalizing total cellular intensity at each time point to the total cellular intensity soon after photobleaching. 6. Now the intensity of a bleached focus at a given time point can be calculated using the following equation, which corrects measured intensity values for any photobleaching which may have occurred: I ðt Þ ¼ ðIb ðt Þ=Ib max Þ=ðIc ðt Þ=Ic max Þ Where: Ib(t) ¼ intensity of ROI at time t (post-bleach). Ibmax ¼ maximum intensity of ROI (pre-bleach). Ic(t) ¼ intensity of whole cell at time t. Icmax ¼ intensity of whole cell soon after bleach. 7. By plotting the I(t) values for a bleached or unbleached focus over the time of imaging, you can get an estimate of FRAP or FLIP, respectively (see Note 12 on expected outcomes and controls) (Fig. 2).

FRAP as a Tool to Study SMC Dynamics In Vivo

37

Fig. 2 Using FRAP to study dynamics of MukB-YFP in foci. Above: Representative time lapse of a cell with MukB-YFP foci during a FRAP experiment is shown. The region of interest (ROI) that is pulse-bleached is highlighted with a circle, pulse-bleaching is indicated with *, and recovery after bleaching is indicated by the arrow. Below: Quantification of FRAP experiment is shown. Two pre-bleach images were taken prior to pulse-bleaching of fluorescence in the ROI. Images were taken every 30 s after bleaching to record fluorescence recovery after bleaching. Normalized intensity is plotted for the bleached focus (FRAP) and for the control, unbleached focus (FLIP) in the same cell

4

Notes 1. It is ideal to construct fluorescent fusions of proteins in the chromosome at the endogenous locus of the gene. One efficient way of strain construction in E. coli is using the λ-red recombination system [22]. MukBEF genes are arranged in an operon (in the order mukF-mukE-mukB). While C-terminal fusions to MukB and MukE are fully functional, MukF needs to be tagged in its N-terminus for function to be maintained. A short linker of about 8–10 amino acids (glycine, serine, and alanine rich) is typically inserted between MukB, E, or F and the fluorescent protein (monomeric form of YFP, mYPet, has been used to image MukBEF in previous experiments [6]). 2. Typically pulse-bleaching should be carried out using the same wavelength as used for imaging. In the case of YFP, this is the peak absorption wavelength of 514 nm. In the event that the YFP laser is not powerful enough for pulse-bleaching, a 488 nm laser can be used for this step of the experiment.

38

Anjana Badrinarayanan and Mark C. Leake

3. Wild-type E. coli cells can be grown at 37  C in rich or minimal media. However, ΔmukB, E, or F strains or strains with mutants of MukB are temperature sensitive and ideally grow at room temperature (~22  C). When doing an experiment that involves ΔmukB (or mutant MukB) and wild-type cells, you should grow both cultures at 22  C so that the conditions are comparable during imaging as well. 4. The generation time for E. coli in M9-glycerol is ~100 min. Cells grown in these conditions have nonoverlapping replication cycles and are simpler to study processes such as chromosome organization, replication, and segregation using microscopy. When growing cells in M9-glycerol, it is important to ensure that cells do not go into late stationary phase (O.D.600 ~ 1) as the recovery time (lag phase) to return to exponential growth will be prolonged. 5. Agarose pads can dry out or desiccate when kept for a long period during imaging (especially at high temperatures such as 37  C). In order to prevent this, gene frames are used. 6. As stated earlier, M9-glycerol provides ideal growth conditions for microscopy-based experiments in E. coli. Another important advantage of using M9-glycerol over LB is the lower autofluorescence in M9. Background fluorescence can pose a problem during imaging and in particular during analysis of fluorescence intensity in the cell. It is always advisable to use media with low levels of autofluorescence for this reason. Autofluorescence can be further reduced using low-fluorescence agarose, for example the NuSieve GTG Agarose from Lonza Biosciences. 7. Ideal laser intensity settings will vary between microscope setups. It is recommended to use low laser intensities during imaging in order to reduce photobleaching or phototoxicity effects. The same applies for laser intensity used for pulsebleaching. It is recommended to use an intensity that is high enough to completely bleach fluorescence in the region of interest, but not too high that other regions of the cell are bleached as well. 8. Since E. coli cells are small in size, FRAP experiments can sometimes cause bleaching across the entire cell, which can complicate the analysis of fluorescence recovery. One way of circumventing this problem is to treat E. coli cells with the cell division inhibitor cephalexin (100 mg/mL) for 2–3 generations prior to imaging. This will result in the production of elongated cells with multiple, segregated chromosomes. If cephalexin is used, it should also be added to the agarose pad to prevent cells from dividing during imaging.

FRAP as a Tool to Study SMC Dynamics In Vivo

39

9. Before you start a FRAP experiment it is important to ascertain that the pulse-bleach is centered on the region of interest chosen in the cell. This can be done using the “FRAP Calibration Wizard” in Volocity. For this, you will need a slide with GFP fluorescence (I use a fluorescence marker to make this). 10. Since bacterial cells are small, try to use a small ROI for pulsebleaching to avoid bleaching a large area of the cell. 11. While imaging, it is important to ensure that cells are still actively growing on the agarose pad. I recommend FRAP imaging of cells on an agarose pad for no longer than 2 h. More traditional time-lapse movies can be carried out for longer (as long as the cells continue to grow). 12. There are, broadly, two typical outcomes of a FRAP experiment: a. There is no recovery after photobleaching and the slight increase in fluorescence intensity in the ROI after pulsebleaching is due to diffusion of free fluorescent molecules into the area. b. There is active recovery of fluorescence as assessed by a significant increase in intensity in the ROI after pulsebleaching. You should be able to see the return of a MukB focus in this case. In order to test for the physiological relevance of this recovery, you can use MukB mutants that should not show recovery after photobleaching [6].

5

Conclusions Here, we have described an easy-to-implement protocol for performing FRAP in live bacteria, and shown how simple analysis can render important quantitative details about the molecular turnover of a functional molecular machine MukBEF. We note that recent other examples of FRAP on live E. coli have included details of the molecular turnover of the DNA replication machinery (the replisome), indicating that almost all of the replisome protein components turn over appreciably apart from the DnaB replicative helicase that seems to serve as a molecular anchor for the whole replication machinery [23]. Also, FRAP has recently been used to quantify molecular dynamics of proteins incorporated into molecular assemblies called aggresomes [24] that are composed of liquidliquid phase-separated condensates that act to increase the fitness of the bacterial cell in response to its ability to cope with a range of hard stresses including antibiotics [25]. FRAP is clearly a robust and powerful tool, and its future use may become even more valuable if developments are implemented to correlate multiple single-molecule biophysics methods [26], for example to extend its utility to 3D multispectral tracking [27], fluorescence polarization data to infer correlated molecular orientation changes [28], as well as correlating mobility signals with changes to the cell physiology, such as molecular crowding [29].

40

Anjana Badrinarayanan and Mark C. Leake

Acknowledgments M.L. acknowledges support from the Engineering and Physical Science Research Council (EPSRC EP/T002166/1, EP/N031431/1), the Biotechnology and Biological Sciences Research Council (BB/P000746/1, BB/R001235/1), and the Royal Society (IEC\NSFC\191406). References 1. Le TB, Laub MT (2014) New approaches to understanding the spatial organization of bacterial genomes. Curr Opin Microbiol 22C:15–21 2. Wang X, Llopis PM, Rudner DZ (2013) Organization and segregation of bacterial chromosomes. Nat Rev Genet 14:191–203 3. Danilova O, Reyes-Lamothe R, Pinskaya M et al (2007) MukB colocalizes with the oriC region and is required for organization of the two Escherichia coli chromosome arms into separate cell halves. Mol Microbiol 65: 1485–1492 4. Nolivos S, Sherratt D (2014) The bacterial chromosome: architecture and action of bacterial SMC and SMC-like complexes. FEMS Microbiol Rev 38:380–392 5. Badrinarayanan A, Lesterlin C, Reyes-Lamothe R et al (2012) The Escherichia coli SMC complex, MukBEF, shapes nucleoid organization independently of DNA replication. J Bacteriol 194:4669–4676 6. Badrinarayanan A, Reyes-Lamothe R, Uphoff S et al (2012) In vivo architecture and action of bacterial structural maintenance of chromosome proteins. Science (New York, NY) 338: 528–531 7. Nicolas E, Upton AL, Uphoff S et al (2014) The SMC complex MukBEF recruits topoisomerase IV to the origin of replication region in live Escherichia coli. MBio 5: e01001–e01013 8. Shaner NC, Steinbach PA, Tsien RY (2005) A guide to choosing fluorescent proteins. Nat Methods 2:905–909 9. Schermelleh L, Heintzmann R, Leonhardt H (2010) A guide to super-resolution fluorescence microscopy. J Cell Biol 190:165–175 10. Wollman AJM, Miller H, Zhou Z et al (2015) Probing DNA interactions with proteins using a single-molecule toolbox: inside the cell, in a test tube and in a computer. Biochem Soc Trans 43:139–145 11. Plank M, Wadhams GH, Leake MC (2009) Millisecond timescale slim-field imaging and

automated quantification of single fluorescent protein molecules for use in probing complex biological processes. Integr Biol 1:602–612 12. Robson A, Burrage K, Leake MC (2013) Inferring diffusion in single live cells at the singlemolecule level. Philos Trans Roy Soc Lond B Biol Sci 368:20120029 13. Llorente-Garcia I, Lenn T, Erhardt H et al (2014) Single-molecule in vivo imaging of bacterial respiratory complexes indicates delocalized oxidative phosphorylation. Biochim Biophys Acta 1837:811–824 14. Lenn T, Leake MC (2012) Experimental approaches for addressing fundamental biological questions in living, functioning cells with single molecule precision. Open Biol 2: 120090 15. Chiu S-W, Leake MC (2011) Functioning nanomachines seen in real-time in living bacteria using single-molecule and super-resolution fluorescence imaging. Int J Mol Sci 12: 2518–2542 16. Leake MC (2010) Shining the spotlight on functional molecular complexes: the new science of single-molecule cell biology. Commun Integr Biol 3:415–418 17. Bryan SJ, Burroughs NJ, Shevela D et al (2014) Localisation and interactions of the Vipp1 protein in cyanobacteria. Mol Microbiol 94(5):1179–1195 18. Chiu S-W, Roberts MAJ, Leake MC et al (2013) Positioning of chemosensory proteins and FtsZ through the Rhodobacter sphaeroides cell cycle. Mol Microbiol 90:322–337 19. Lenn T, Leake MC, Mullineaux CW (2008) Are Escherichia coli OXPHOS complexes concentrated in specialized zones within the plasma membrane? Biochem Soc Trans 36: 1032–1036 20. Sprague BL, McNally JG (2005) FRAP analysis of binding: proper and fitting. Trends Cell Biol 15:84–91 21. Reyes-Lamothe R (2012) Use of fluorescently tagged SSB proteins in in vivo localization experiments. Methods Mol Biol 922:245–253

FRAP as a Tool to Study SMC Dynamics In Vivo 22. Datsenko KA, Wanner BL (2000) One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci U S A 97:6640–6645 23. Beattie T, Hapadia N, Nicolas E, Uphoff S, Wollman AJM, Leake MC, Reyes-Lamothe R (2017) Frequent exchange of the DNA polymerase during bacterial chromosome replication. elife 6:e21763 24. Pu Y, Li Y, Jin X, Tian T, Ma Q, Zhao Z, Lin S-L, Chen Z, Li B, Yao G, Leake MC, Lo C-L, Bai F (2019) ATP-dependent dynamic protein aggregation regulates bacterial dormancy depth critical for antibiotic tolerance. Mol Cell 7:143–156 25. Jin X, Lee J-E, Schaefer C, Luo X, Wollman AJM, Tian T, Zhang X, Chen X, Li Y, McLeish TCB, Leake MC, Bai F (2021) Membraneless organelles formed by liquid-liquid phase separation increase bacterial fitness. Sci Adv. https://doi.org/10.1101/2021.06.24. 449778 26. Leake MC (2021) Correlative approaches in single-molecule biophysics: a review of the

41

progress in methods and applications. Methods. https://doi.org/10.1101/2021.06.24. 449778 27. Wollman AJM, Hedlund EG, Shashkova S (2020) Leake MC towards mapping the 3D genome through high speed single-molecule tracking of functional transcription factors in single living cells. Methods 170:82–89 28. Shepherd JW, Payne-Dwyer AL, Lee J-E, Syeda A, Leake MC (2021) Combining single-molecule super-resolved localization microscopy with fluorescence polarization imaging to study cellular processes. J Phys Photonics 3:034010 29. Shepherd JS, Lecinski S, Wragg J, Shashkova S, MacDonald C, Leake MC (2020) Molecular crowding in single eukaryotic cells: using cell environment biosensing and single-molecule optical microscopy to probe dependence on extracellular ionic strength, local glucose conditions, and sensor copy number. Methods: S1046-2023(20)30236-X. https://doi.org/ 10.1016/j.ymeth.2020.10.015

Chapter 5 Atomic Force Microscopy of DNA and DNA-Protein Interactions Philip J. Haynes, Kavit H. S. Main, Bernice Akpinar, and Alice L. B. Pyne Abstract Atomic force microscopy (AFM) is a microscopy technique that uses a sharp probe to trace a sample surface at nanometer resolution. For biological applications, one of its key advantages is its ability to visualize the substructure of single molecules and molecular complexes in an aqueous environment. Here, we describe the application of AFM to determine the secondary and tertiary structure of surface-bound DNA, and its interactions with proteins. Key words Atomic force microscopy, AFM, DNA, Supercoiling, Double helix, DNA-protein binding

1

Introduction Atomic force microscopy (AFM, see Fig. 1) is a unique tool to obtain structural information of single biomolecules at ~1 nm spatial resolution. AFM allows for the characterization of molecules adsorbed on a planar substrate in aqueous solution, i.e., without the need for chemical fixation, staining, or vitrifying (Fig. 1). AFM can be performed in air or in fluid. In-air AFM is more straightforward in operation and can provide static snapshots of reactions that involve DNA in solution (i.e., taking place while the DNA was free in solution and not bound to the planar substrate) by drying the sample before imaging. However, AFM in liquid has yielded the highest spatial resolution on DNA [1–3] and can also be used to observe biomolecular dynamics [4–6]. AFM has been extensively used to visualize DNA supercoiling [7–11], noncanonical DNA secondary structures [11, 12], DNA origami assemblies [13, 14], interactions with therapeutic agents [15, 16], and DNA-protein complexes [17–22], with the additional advantage that binding events and conformational changes can be monitored in real time [5, 6, 21]. AFM can discern the helical pitch of DNA [23–27] and

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_5, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

43

44

Philip J. Haynes et al.

Fig. 1 Schematic of AFM in aqueous solution. A sharp tip is scanned line by line across the sample surface to build up an image of the surface topography. The topography at each scanned point is a function of the tip-sample interaction which is monitored by measuring the bending of the cantilever to which the tip is attached. The bending of the cantilever is usually detected via a laser beam deflected on a position-sensitive detector (4-quadrant photodiode). The sample is mounted on a (usually piezoelectric) scanner for threedimensional positioning with sub-nanometer accuracy. The sample, the tip, and the cantilever are immersed in liquid. Inset: magnified for clarity

RNA [28], and under appropriate imaging conditions resolve the two strands of the DNA double helix (Fig. 2a) [1–3]. For high-resolution AFM imaging, the sample must be adsorbed onto an atomically flat substrate, such that observed topographic features can be attributed to the biological sample and not to the substrate. Muscovite mica is often used as a substrate for AFM imaging, since its structure consists of many weakly interacting planes. These planes can be cleaved using sticky tape, resulting in an atomically flat surface. The disadvantage of mica as a substrate for DNA imaging is that both mica and DNA are negatively charged at neutral pH in aqueous solution, impeding the adsorption of DNA to the mica. Various methods that modify the surface charge of mica have been developed to facilitate DNA adsorption. These include functionalizing the mica with aminopropyltriethoxy silane (APTES) or aminopropyl silatrane (APS) to create a positively charged surface [29]; using monovalent and divalent cations to bridge the charge repulsion [30]; adsorbing a positively charged lipid bilayer to the mica to facilitate electrostatically driven adsorption [24]; and functionalizing with polymers (e.g., poly-L-lysine) to create a positively charged monolayer on the surface [9]. Here we cover the divalent cation method and the use of poly-L-lysine-based mica functionalization which facilitate high-resolution imaging of DNA. This allows for structural insights into variations in DNA structure at both local and global levels, at a resolution where both the minor and major grooves can be resolved (Fig. 2a).

Atomic Force Microscopy of DNA and DNA-Protein Interactions

45

Fig. 2 High-resolution topographic images of DNA acquired by PeakForce Tapping mode (Subheading 3.4). The divalent cation method (Subheading 3.2.1) is used to adsorb (a) DNA plasmids and (b) 339 base-pair DNA minicircles. In (a), the two strands of the DNA double helix are captured. Inset: A higher resolution image digitally straightened and overlaid with a cartoon representation of the B-DNA crystal structure. Color scales: 2.5 nm (main), 1.2 nm (inset). In (b), defects and disruptions in the canonical B-form DNA are observed (red triangles), as a step change in the angle of the helix. Color scale (scale bar in a): 2.5 nm (ref. [11], with permission). (c) A DNA plasmid adsorbed onto PLL1000–2000-functionalized mica (Subheading 3.2.2) where the chains of poly-L-lysine making up the underlying substrate are resolved. Color scale: 8 nm (adapted from ref. [31], with permission)

In typical AFM experiments, the sample preparation is a compromise between the need to immobilize the DNA for high spatial resolution and, for real-time imaging of binding events and conformational changes, the need to allow DNA sufficient freedom for structural rearrangements. In addition, when using salts to facilitate DNA adsorption on a substrate, there is—in particular for the commonly used NiCl2—the possibility of salt accumulation on the substrate, which may compromise the resolution and interpretation of the resulting AFM images [1–3]. The poly-L-lysine preparation has the advantage of not requiring particular salts in the solution. However, due to its gross positive charge and adhesiveness, it may bring in other contaminants from the solution. This is further exacerbated when interrogating DNA-protein interactions, as nonspecific protein binding to the underlying substrate is problematic at biologically relevant protein concentrations. At these concentrations, nonspecifically bound protein obscures the specifically adsorbed DNA-bound protein complexes (Subheading 3.5). This nonspecific binding is further amplified when the DNA-protein affinity is considered low and higher concentrations of protein are required to observe binding. To curtail this, the surface can be PEGylated using diblock copolymers that contain poly-L-lysine (surface-binding domain) and polyethylene glycol (surface-passivating domain) [31]. The addition of poly(L-lysine)b-poly(ethylene glycol) (PLL-b-PEG) achieves selective DNA adsorption while minimizing nonspecific protein adsorption, where PEG suppresses protein binding by steric repulsion.

46

Philip J. Haynes et al.

In addition to sample preparation, a critical element for highresolution AFM imaging is force sensitivity and control [1–3]. Generally, the AFM probe needs to exert a force on the sample to be able to record the surface topography. However, if this force is too large, it can cause sample deformation or contamination of the probe, both of which can reduce the resolution of the resulting image. To minimize the force exerted on the sample, a wide range of operational modes have been developed. The best highresolution AFM images in the literature appear highly similar. This illustrates that there is more than one route to high-resolution imaging given that all required parameters are optimized; however, these modes vary in the ease by which the imaging is achieved. In particular, early DNA double helix imaging was carried out using phase and frequency modulation techniques, which typically achieve high sensitivity using stiffer cantilevers [1, 2]. Here, we describe a widely implemented method for in-liquid AFM imaging of DNA that has been successfully employed to visualize the DNA double helix: rapid force-distance imaging mode (also called Force Modulation or PeakForce™ Tapping mode, see Fig. 3). Rapid force-distance AFM minimizes the tip-sample interaction by measuring the force applied by the tip to the sample at each point during imaging/scanning. It does so by taking repeated force curves across the sample whereby the tip is

Fig. 3 PeakForce Tapping is a widely adopted AFM mode which allows routine imaging in fluid. (a) The cantilever tip is driven to oscillate sinusoidally (b) at frequencies much lower than its resonant frequency, resulting in intermittent contact with the surface and low lateral tip-sample interactions. The dashed line indicates the position of the cantilever tip and the diamond indicates the feedback point, set via the Sync Distance New function. (c) The interaction force is minimized and controlled by a continuous feedback loop. Illustrations showing PeakForce Tapping force curves as a function of time and z-position, showing the tip-sample approach (teal) and withdraw (magenta)

Atomic Force Microscopy of DNA and DNA-Protein Interactions

47

approached to and retracted from the surface. As the tip interacts with the surface the applied force is measured with respect to the baseline away from the surface, for each force curve (Fig. 3b). By measuring the height at which the force reaches a predefined setpoint, we can determine the sample topography at each interaction point, thereby modulating the vertical tip-sample distance. The surface preparations described below are not limited to rapid force-distance imaging (PeakForce Tapping) and the AFM system of choice. Our description presumes some knowledge about elementary AFM operation, such as can be found in instrument manuals.

2

Materials Prepare and store all reagents at room temperature, unless indicated otherwise.

2.1 General Materials

1. 15–50 mL Falcon™ tubes. 2. Borosilicate glass bottle and volumetric flask (both 500 mL). 3. Eppendorf™ tubes, 0.5 and 2 mL, DNA, and protein LoBind. 4. 6 mm Mica substrates (Agar Scientific). 5. Magnetic stainless steel disks (15  0.8 mm). 6. Adhesive backed PTFE (Bytac® surface protection laminate). 7. Scalpel or punch. 8. Scotch™ tape. 9. Araldite® 2-part epoxy resin. 10. Stainless steel and plastic-tip tweezers. 11. Petri dishes (35  10 mm).

2.2

AFM Cantilevers

An appropriate cantilever should be chosen from the list below. The cantilever manufacturer, spring constant (N/m), resonant frequency in fluid (kHZ, see Note 1), and nominal tip radius (nm) are provided. 1. FastScan-D (Bruker, 0.25 N/m, 110 kHz, 5 nm). 2. PEAKFORCE-HIRS-F-B (Bruker, 0.12 N/m, 30 kHz, 1 nm). 3. BioLever mini (Olympus, 0.1 N/m, 25 kHz, 10 nm). 4. MSNL-E (Bruker, 0.05 N/m, 7 kHz, 2 nm).

2.3

Buffer Solutions

1. Prepare the following buffers for adsorption and imaging: (a) Nickel adsorption and imaging buffer: 3 mM NiCl2, 20 mM HEPES pH 7.4, store on bench, ambient temperature. For 1 L, weigh out 4.76 g HEPES (molecular

48

Philip J. Haynes et al.

weight: 238.30 g/mol) and 0.95 g NiCl2 (molecular weight: 237.69 g/mol). (b) Poly-L-lysine imaging buffer: 20 mM HEPES pH 7.4 with or without 120 mM NaCl; see Note 2. Used in methods both with and without copolymerization with polyethylene glycol (PEG). 2. Gather a volumetric flask and borosilicate glass bottle (both 500 mL), and clean them by soaking and shaking in detergent (e.g., PCC-54™) and rinsing with ultrapure water (Milli-Q®, resistivity >18.2 MΩ) until no bubbles remain (see Note 3). 3. Weigh out the necessary reagents and dissolve in 200 mL ultrapure water in the glass bottle. Use a magnetic stirrer to aid the dissolution process if required. The spatula and stirrer should be cleaned prior to use by rinsing with ethanol and ultrapure water and air-dried. 4. Pour into the volumetric flask and fill up just below 500 mL with ultrapure water. 5. Pour back into the bottle. Check the pH of the solution using a well-calibrated and cleaned pH meter (three-point calibrated at pH 4, 7, and 10). 6. Adjust to the required pH using small amounts of concentrated (~1 M) HCl or NaOH and a clean magnetic stirrer. 7. Pour into the volumetric flask and fill to the 500 mL mark with ultrapure water. 8. Filter the buffer by passage through a 0.2 μm syringe and store long-term in a glass bottle or short-term in a 15–50 mL Falcon™ tube. 9. Ensure that the buffer is clean (see Note 4) and store on the bench at room temperature. 2.4

DNA Substrates

1. 339 Base-pair DNA minicircles (store at 4  C) in 20 mM HEPES pH 7.4, at a concentration of ~10 ng/μL (see Note 5). 2. 496 Base-pair linear DNA (store at 4  C) in 10 mM HEPES pH 7.8 at concentrations between 1.5 and 20 ng/μL (see Note 5).

3

Methods Mica and DNA are both negatively charged at a neutral pH in aqueous solution. There are a number of methods used to facilitate the adsorption of DNA to the mica surface [9, 24, 29–31]. The following sections describe three methods to facilitate DNA and DNA-protein adsorption on mica, which are appropriate for imaging DNA in liquid (see Note 6).

Atomic Force Microscopy of DNA and DNA-Protein Interactions

49

Mica substrates can be attached to a steel disk to be mounted on magnetic sample holders as common in AFM instruments (see Note 7). 3.1 Preparation of Mica Substrate

1. Cut the adhesive PTFE into circles of the same size as the steel sample disks (15 mm), using either a punch or scalpel (see Note 8). 2. Peel off the backing of the PTFE cut-out and adhere to the steel disk. 3. Mix the Araldite® 2-part epoxy resin 50:50 on a disposable surface, e.g., weighing boat. Using a pipette tip transfer a small amount of the mixed epoxy to the center of the Teflon (use tip as a capillary to help bring up the glue). 4. Cleave the 6 mm mica disk on one side with scotch tape. With the cleaved mica disk facing down, immediately place on top of the epoxy droplet and press flat. 5. Leave the glue to dry and cure overnight. 6. Once cured, cleave the mica using Scotch tape to reveal an atomically flat clean substrate (see Note 9).

3.2 Methods for DNA Adsorption on a Mica Substrate for AFM Imaging in Fluid 3.2.1 DNA Adsorption Using Divalent Cations

Divalent cations (in this case Ni2+) can be used to overcome the electrostatic repulsion between DNA and mica, thus facilitating DNA adhesion to the mica, which can also be tuned via the cationic concentration in the solution as outlined below. 1. Immediately before DNA adsorption, cleave a 6 mm mica disk that has been prepared as described in Subheading 3.1. 2. Cover the freshly cleaved mica with 20 μL of nickel adsorption buffer (see Note 10). 3. Add 4 μL DNA (1 ng/μL, see Note 11) and distribute evenly in the meniscus by gently purging. 4. Adsorb for 30 min. Then gently exchange the buffer to the nickel imaging buffer four times to remove any unbound DNA. 5. Add sufficient nickel imaging buffer to form a droplet covering the sample (dependent on the AFM system, see Note 12). 6. Mount the sample on AFM.

3.2.2 DNA Adsorption Using PLL

DNA adsorption on mica can also be assisted by surface modification of mica. Poly-L-lysine creates a cationic monolayer on the mica surface due to its protonated amino groups. DNA can then bind to the positively charged groups. The procedure for this is outlined below.

50

Philip J. Haynes et al.

1. Immediately before exposure to poly-L-lysine, cleave a mica disk that has been prepared as described in Subheading 3.1. 2. Cover the mica with 10 μL poly-L-lysine solution (0.01% w/v, 150,000–300,000 MW). 3. Incubate for 1 min. 4. Hold the disk at an angle and thoroughly rinse under a stream of ultrapure water. Or, equivalently, wash 5 with ultrapure water. 5. Blot the disk and add 20 μL of a poly-L-lysine imaging buffer. 6. Add 10 μL DNA (1 ng/μL, see Note 11) and distribute evenly in the meniscus by gently purging. 7. Absorb for 10 min (see Note 2). Then wash four times with the poly-L-lysine imaging buffer to remove any unbound DNA (see Note 13). 8. Add sufficient poly-L-lysine imaging buffer on the sample (dependent on the AFM system, see Note 12). 9. Mount the sample on AFM. Figure 2 shows DNA plasmids adsorbed on a mica substrate by both the divalent cation (Fig. 2a) and poly-L-lysine (Fig. 2c) methods. Both methods yield stable DNA adsorption on the substrate for imaging by AFM. 3.2.3 DNA Adsorption Using PLL-b-PEG

The poly(L-lysine)-b-poly(ethylene glycol) (PLL-b-PEG) block copolymer, PLL10-b-PEG113 (subscript refers to the number of monomer repeats), can be purchased as a lyophilized powder and dissolved in ultrapure water to 1 mg/mL on the day of imaging (see Note 14). It is supplemented 1:1 with poly-L-lysine solution (PLL1000–2000, 0.01% w/v, 150,000–300,000 MW) to aid DNA adsorption. The significant decrease in nonspecific background protein with the addition of PLL-b-PEG is shown in Fig. 5b, c. The substrate preparation is outlined below. 1. Prepare a  20 μL 1:1 PLL10-b-PEG113/PLL1000–2000 (PLL-b-PEG/PLL) mixture in an Eppendorf. 2. Cleave a mica disk that has been prepared as described in Subheading 3.1. 3. Immediately after cleaving, add 20 μL of the PLL-b-PEG/PLL mixture and leave to incubate in a humidified petri dish for 45 min (see Note 15). 4. Wash 5 with ultrapure water and a further 5 times with polyL-lysine imaging buffer (see Note 2). 5. Add 20 μL of 496 bp linear DNA (1.5 ng/μL, 7.8 nM, see Notes 11 and 16), with or without protein, to the disk and

Atomic Force Microscopy of DNA and DNA-Protein Interactions

51

gently mix. If adsorbing DNA with protein, the DNA and protein must be preincubated at room/ambient temperature before deposition. 6. Incubate the sample on the disk for 30 min and, with the imaging buffer, wash 5 and make up to 30 μL (see Note 12). 7. Mount the sample on AFM, and approach and image as described in Subheading 3.3. For protein-exchange assays where you add protein to a surface with DNA already adsorbed, carry out a buffer exchange for protein in the imaging buffer (see Note 17). Reapproach and resume imaging after a 5-min incubation. 3.3 Pre-Imaging Setup for HighResolution AFM in Fluid

1. Prior to imaging, soak the chosen cantilever in a petri dish containing isopropanol:ethanol (1:1) for several hours and dry by blotting. 2. Ensure that cantilevers are totally dry before plasma cleaning in air for 30 s at 10% power (Zepto, Deiner Electronics). 3. Mount the plasma-cleaned cantilever in the AFM and align the laser. Leave the AFM to equilibrate in buffer solution, using a clean freshly cleaved mica disk during sample preparation (see Note 18). 4. Exchange the blank mica disk for the sample mica disk. Several variables need to be optimized to record highresolution images by AFM. These variables include sample preparation, cantilever characteristics, and AFM operation. The sample preparations above should yield DNA that is sufficiently bound to the mica substrate to facilitate high-resolution imaging. 5. Prepare a DNA sample as described in Subheadings 3.1 and 3.2.1 or Subheading 3.2.2, and place the sample in the AFM. 6. Select an appropriate cantilever for imaging DNA and place in the fluid cell. (a) Cantilevers with spring constants 0.3 N/m are preferable for achieving the highest resolution when using the AFM imaging modes described here—allowing imaging of DNA at forces 18 MΩ and Total Organic Carbon 0.9; this isolates signals which arise from singlemolecule binders. 7. Calculate the lifetime by calculating the difference between the frames at which minima and maxima were reached and then multiply this by the frame rate in milliseconds (Fig. 6). 8. Create a histogram of all binding lifetimes; from this create a cumulative frequency plot. 9. Create a cumulative decay function by subtracting each point of the cumulative frequency plot from the total number of binders. 10. Fit the cumulative decay function to an exponential function (see below); to determine whether the fit is single or multiexponential, convert the binding frequency histogram to a logarithmic scale—if the resulting line is straight, it is a single exponential; if it is formed from two lines it is a double exponential; and if it is formed from three it is a triple exponential, etc.: f ðx Þ ¼ amp  exp ð
k  x Þ where k is the release rate constant, and x is the binned lifetime. 11. The resulting k describes the release rate constant, the number of bind molecules releasing per second—take the reciprocal of k to convert to lifetime (Fig. 7). 12. In double and triple exponentials, the amp for each describes the proportionate size of each population.

4

Notes 1. Laser output power should be matched to the requirements of the camera being used—since iSCAT is carried out in a shotnoise limited regime, in a stable well-constructed iSCAT

140

Robert J. Charman and Neil M. Kad

a.

b.

c.

d.

Lifeme (ms) = number of pixels/1000

Fig. 6 Example of protein binding and unbinding. Binding (a) and unbinding (b) of proteins to target DNA produce distinctive PSFs after processing. A kymograph of this region (c) and the subsequent intensity plot (d) show the key signature from which lifetimes can be accurately calculated. Both binding and unbinding show a smooth fade in and fade out, with the frame in which the protein bound/unbound being the frame in which maximum contrast is reached

system. The signal-to-noise (SNR) as a result of shot-noise is given by N SNR ¼ pffiffiffiffiffi N

iSCAT Imaging of DNA-Protein Interactions

141

Fig. 7 Example double-exponential fit of UvrA binding to fluorescein-labeled DNA. The double-exponential fit (a) of a cumulative decay function produced from UvrA lifetimes on fluorescein-labeled DNA reveals the presence (b) of two populations, which correspond (c) to a short (0.104 s) and long lifetime (1.830 s)

where N is the average number of events, i.e., equivalent to the number of photons reaching a pixel per exposure, meaning shot noise is reduced to its minimum when the camera is running with pixels near their full-well capacity. The required illumination intensity can be calculated using the full-well capacity of the camera being used (33 ke
for the Flir Grasshopper USB3 GS3-U323S6M-C), the total illumination area (20  20 μm), and the pixel size (0.0325  0.0325 μm). To calculate the required illumination intensity, the number of events per pixel needs to be tuned to the full-well capacity of the camera in use. Firstly, the number of pixels being illuminated needs to be calculated: Total pixels ¼

Illumination area ðμmÞ Pixel size ðμmÞ

With an illumination area of 20  20 μm, and a pixel size of 0.0325  0.0325 μm, a total of 378,698 pixels are illuminated. To saturate all pixels 33 ke
is needed per pixel; however quantum efficiency of the camera also needs to be considered (76% for the Flir Grasshopper USB3 GS3-U3-23S6M-C). This means per exposure time (1 ms for 1000 fps) 1.64  1010 photons need to reach the camera:   Total pixels ðphotonsÞ  Full
well capacity ðe
Þ Required photons ms
1 ¼ Quantum efficiency From this the required output power can be calculated from   Total photons ms
1  h Output power ðmW Þ ¼ λ ðnmÞ

142

Robert J. Charman and Neil M. Kad

where h is the Planck constant, and λ is the wavelength of the laser used. Please note that the use of an intensity-reducing beam splitter (e.g., 50:50 beam splitter) needs to be accounted for. 2. Longer wavelength lasers are recommended for unlabeled scattering; however, use of a 532 nm laser is effective when using gold nanoparticle labels as this wavelength closely matches their plasmon resonance frequencies. 3. ABT buffer and mPEG solution should be remade every 2 weeks. 4. Holes are drilled using a Dremel electric hand drill with diamond-coated dental drill tips. 5. Gaskets are made from sheets of double-side sticky tape (with a thickness of 180 μm); these are cut automatically using a digital cutting machine. They are cut into rectangles (with dimensions 30 mm  20 mm), with a central region (with dimensions 20 mm  10 mm) removed, leaving a border thickness of 5 mm. The thickness of the gasket controls the final volume of the flow cell. This can be modified to accommodate additional inlet holes, as long as this is kept consistent across experiments. 6. Plasma cleaning, mPEG, and ABT are all used to prevent nonspecific binding of labeled proteins to the surface (SEA-Block is a commercially available alternative to ABT which is just as efficient). 7. Leaving the oligonucleotide mixture to cool slowly by leaving it in the heat block after heating leads to more efficient annealing. 8. YOYO-1 can be used to check for the presence of DNA on the surface of the flow cell (either on a separate fluorescent microscope or with a separate fluorescence channel implemented within the iSCAT system). 9. Ratiometric Imaging and Lifetime Detection MATLAB programs are freely available from https://github.com/Kad-Lab. 10. iSCAT PSFs are estimated as previously described [6]. In brief, a radial series of 360 intensity profiles are taken (each with a length of 35 pixels), with one degree of separation between each, extending out from the centre of the iSCAT PSF. These profiles are then averaged to create an accurate estimate of the true iSCAT PSF, isolating it form local inhomogeneities. This profile can then be reprojected as a 2D PSF to be used for experimental image convolution.

iSCAT Imaging of DNA-Protein Interactions

143

References 1. Andrecka J et al (2015) Structural dynamics of myosin 5 during processive motion revealed by interferometric scattering microscopy. eLife:4. https://doi.org/10.7554/eLife.05413 2. Kukura P et al (2009) High-speed nanoscopic tracking of the position and orientation of a single virus. Nat Methods 6(12):923–927. https://doi.org/10.1038/nmeth.1395 3. Young G et al (2018) Quantitative mass imaging of single biological macromolecules. Science 360(6387):423–427. https://doi.org/10. 1126/science.aar5839 4. de Wit G et al (2018) Revealing compartmentalized diffusion in living cells with interferometric scattering microscopy. Biophys J 114(12):

2945–2950. https://doi.org/10.1016/j.bpj. 2018.05.007 5. Cole D et al (2017) Label-free single-molecule imaging with numerical-aperture-shaped interferometric scattering microscopy. ACS Photonics 4(2):211–216. https://doi.org/10.1021/ acsphotonics.6b00912 6. Taylor RW, Mahmoodabadi RG, Rauschenberger V, Giessl A, Schambony A, Sandoghdar V (2019) Interferometric scattering microscopy reveals microsecond nanoscopic protein motion on a live cell membrane. Nature Photonics 13 (7):480–7. https://doi.org/10.1038/s41566019-0414-6

Chapter 11 Escherichia coli Chromosome Copy Number Measurement Using Flow Cytometry Analysis Michelle Hawkins, John Atkinson, and Peter McGlynn Abstract Flow cytometry is a high-throughput technique that analyzes individual particles as they pass through a laser beam. These particles can be individual cells and by detecting cell-scattered light their number and relative size can be measured as they pass through the beam. Labeling of molecules, usually via a fluorescent reporter, allows the amount of these molecules per cell to be quantified. DNA content can be estimated using this approach and here we describe how flow cytometry can be used to assess the DNA content of Escherichia coli cells. Key words Flow cytometry, Genome copy number, Genome content, Escherichia coli, Replication, Chromosome, DnaA

1

Introduction The start sites of DNA replication are called origins and E. coli has a single circular chromosome with one origin called oriC. Origin activity is predominantly regulated by the DnaA initiator protein [1, 2]. DnaA binds to 9-bp repeat elements near oriC and mediates unwinding of the origin region so that helicase loading can occur [3]. Once oriC fires, two replication forks travel in opposite directions until they reach the terminus region and complete genome replication (Fig. 1a). In optimal conditions the average genome replication time of cells in an exponentially growing population (~40 min) can be longer than the doubling time of E. coli (~20 min) [4]. This is possible because oriC can reinitiate before previous sets of replication forks have finished replicating the genome [5]. The resulting concurrent rounds of replication mean that multiple sets of replisomes can be active in a single cell (Fig. 1a). The number of active replisomes in a cell is a useful readout for replication and fork progression defects [6], while the ability of cells to complete ongoing rounds of replication can also

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_11, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

145

146

Michelle Hawkins et al.

A

ter 2

Active replisomes at time 0

8

4

B

Chromosome copies after 2 hour “run-out”

Number of cells

C Flow cytometry output

12 4

8

12 4

8

12 4

8

Chromosome copy number

Fig. 1 “Run-out” experiment principle and output. (a) E. coli chromosome schematics showing concurrent rounds of DNA replication and how this alters active replisome numbers. Black circles represent ori C. Arrows indicate replisome direction. (b) After 2 h of “run-out” conditions where origin initiation and cell division are inhibited, all the forks will have finished replicating their template and the final number of chromosome copies will equal the number of active replication forks at the time of antibiotic addition. (c) Representative chromosome copy number profiles with fluorescence measurement as a proxy for the total DNA content

report on the efficiency of fork progression [7]. Here we describe how to quantify active replisome number by inhibiting replication initiation and cell division before using flow cytometry to assess total genome content. In this method origin initiation in an exponentially growing population is inhibited using rifampicin. Rifampicin inhibits DNA-dependent RNA synthesis by binding to RNA polymerase and blocking elongation [8]. Inhibition of transcription prevents any further initiation from oriC but active replication forks are not disrupted. Since we are interested in measuring genome content, cell division must be prevented in order to avoid a reduction of the number of chromosomes per cell. Cephalexin is a β-lactam antibiotic that disrupts cell division by binding to peptidoglycans in the cell wall. This inhibits cross-linking of membrane peptidoglycans and downstream septation events that lead to cell division [9]. DNA replication proteins are unaffected by both antibiotics

Escherichia coli Genome Content Measurement

147

and therefore the replication machinery continues to progress and complete chromosome duplication. A similar method was originally employed by Steen and colleagues to study initiation timing [10] and this technique is generally known as a “run-out” experiment. In this protocol cells are first grown in the absence of antibiotics. At a given time point, rifampicin and cephalexin are added to enable all active replication forks to complete genome replication without any subsequent cell division. Cells are then stained with a dye that fluoresces only when bound to DNA and analyzed using flow cytometry to measure the DNA content in individual cells. DNA content can be used to estimate the total number of chromosomes present within cells. This estimation requires analysis of cells grown under conditions known to favor the presence of only a single pair of replication forks at any one time. The majority of these cells will therefore have a fluorescence signal equivalent to two chromosomes. The estimated final number of genome copies enables the number of active replication forks in a cell upon addition of rifampicin and cephalexin to be inferred (Fig. 1b). Here we describe how to perform “run-out” experiments for standard E. coli strains Subheading 3.1 and temperature-sensitive DnaA mutants Subheading 3.2. The method detailed in Subheading 3.2 is for strains where the mutant DnaA is active at 30  C but nonfunctional at 42  C due to instability at higher temperatures [11]. Temperature-based control of origin initiation enables synchronization of cells by extended growth in nonpermissive conditions where initiation is prevented. Once all cells have completed ongoing rounds of replication a brief return to permissive conditions is employed to allow DnaA to function and initiate a new round of replication from oriC, after which the cells are returned to a higher temperature to inhibit further rounds of initiation. This approach allows the majority of cells to initiate one round of chromosome replication within a short (10 min) time window. Tracking the DNA content as a function of time using flow cytometry then allows progression of the newly initiated replication forks to be followed. Approximate rates of replication fork movement between different strains can thus be compared as the DNA content increases from one to two chromosomes (Fig. 2). Flow cytometry relies on particles flowing in a narrow stream through a laser. Measurements of scattered light are recorded as “events.” These events can be confused by debris in the cell suspension and by neighboring particles being detected as a single event, termed doublets. To avoid debris being recorded as an event it is important to filter all reagents used in preparation of your sample suspension. Doublets can be avoided by using a homogenous sample preparation and can be excluded during and post-analysis. Most flow cytometry takes advantage of a fluorescent indicator dye that binds to a cell or molecule of interest. When the laser excites these fluorophores they emit light and this fluorescence

Michelle Hawkins et al.

0

10

Time (minutes)

148

20

30

40

50

60 1

2

Chromosome copy number

Fig. 2 Monitoring progression of genome duplication using an E. coli dnaA temperature-sensitive mutant. Strain HB159 [7] was grown at 42  C for 2 h and then shifted to 30  C at time 0. At this time point the majority of cells contained a single chromosome equivalent since cell division could proceed after completion of ongoing rounds of replication but replication could not be reinitiated during the 42  C incubation period. After 10 min at 30  C the culture was then shifted back to 42  C. The DNA content of the majority of cells increased as a function of time, reflecting the speed with which the chromosome was duplicated

serves as a proxy for the relative amount of the cell or molecule being interrogated. The wide variety of fluorescent dyes and laser wavelengths available make cytometry a versatile method that can be used for a range of prokaryotic and eukaryotic species [12]. Here we are interested in total bacterial DNA content as a measure of genome copy number after “run-out.” We stain our fixed cells with Sytox® Green but other suitable dyes are available. Sytox® Green only permeates dead cells and binds to all nucleic acid. RNase A is used to remove RNA in the sample and avoid unwanted nucleic acid binding. When Sytox® Green is bound to DNA it has absorption and emission maxima of 502 and 523 nm, respectively, and so

Escherichia coli Genome Content Measurement

149

can be excited by a 488 nm argon-ion laser or another 450–490 nm source [13]. We use a Beckman Coulter CyAn™ ADP Analyzer for our analysis but most flow cytometers are equipped to excite and detect at these wavelengths. Here we describe how to prepare E. coli cells for flow cytometry with Sytox® Green as the fluorophore Subheading 3.3. Protocol alterations are required to stain eukaryotes with Sytox® Green and may also be required for other bacterial species. A full description of optimizing flow cytometry parameters for analysis is beyond the scope of this method since they vary depending on the cytometer setup and software. The key considerations are to account for any debris and cell autofluorescence by running buffer and unstained cell control samples. Detectable events in these samples are background readings and should be minimal in order to proceed. The sensor voltage parameters should be optimized so that fluorescence levels corresponding to 1n and at least 8n genome copies can be displayed simultaneously on the x-axis. Some mutant strains can contain cells with 16 genome copies. We count at least 10,000 events but routinely collect data for samples of 100,000. For an excellent overview of flow cytometry mechanics and outputs please refer to Givan (2010) [14].

2

Materials 1. LB medium: 10 g/L Tryptone, 5 g/L yeast extract, 5 g/L NaCl. pH to 7.0 with NaOH. 2. Minimal medium (MM): 2.64 g/L KH2PO4, 4.34 g/L Na2HPO4·12H2O, 1 g/L (NH4)2SO4, 0.02% MgSO4·7H2O, 0.001% Ca(NO3)2·4H2O, 0.00005% FeSO4·7H2O. Prepare 150 mL batches of 56 salts (5.28 g/L KH2PO4, 8.68 g/L Na2HPO4·12H2O, 2 g/L (NH4)2SO4, 2 mL 10% MgSO4·7H2O, 1 mL 1% Ca(NO3)2·4H2O, 50 μL 1% FeSO4·7H2O. Make FeSO4 just prior to use. Autoclave and add an equal volume of sterile distilled water. A carbon source and any necessary supplements can then be added. For our MG1655 strain background we add 4.8 mL 20% glucose (0.32% final concentration) and 300 μL of 0.1% thiamine/ vitamin B1 (0.0001% final concentration) before use. 3. 1  PBS: 137 mM NaCl, 2.7 mM KCl, 4.3 mM Na2HPO4, 1.47 mM KH2PO4 adjusted to a pH of 7.4. The solution should be filtered twice with 0.22 μm filters to avoid debris that could interfere with flow cytometry. 4. Rifampicin: 10 mg/mL in dimethyl formamide wrapped in foil and stored at 20  C. 5. Cephalexin: 10 mg/mL in dH2O wrapped in foil and stored at 20  C.

150

Michelle Hawkins et al.

6. Sytox® Green (Invitrogen): Dilute from 5 mM stock to 50 μM working stock in DMF. Store wrapped in foil at 20  C (see Note 1). 7. RNase A: Make up 1 mg/mL in dH2O working stock and store at 20  C (see Note 1). 8. Sytox mix: 1 μM Sytox® Green, 50 μg/mL RNase A, made up in 1  PBS (see Note 1). 9. Flow cytometer: Analyzer with a 450–490 nm source for use with Sytox® Green and the appropriate flow cytometry tubes. We use a Beckman Coulter CyAn™ ADP Analyzer.

3 3.1

Methods Standard Strains

All of our E. coli growth is carried out at 37  C in an air incubator shaking at 170 RPM. 1. Inoculate 10 mL of LB with a single colony of your strain of interest. Pre-warm 10 mL of LB per strain in a 37  C incubator to prepare for step 2. 2. After the culture has grown to an OD650 of 0.4, dilute the culture to an OD650 of 0.01 in 10 mL of LB pre-warmed to 37  C (see Note 2). 3. Grow the culture to mid-exponential phase (OD650 of 0.4–0.6) (see Note 3). 4. Take a 100 μL sample and transfer to a 1.5 mL tube. Spin the sample down for 1 min in a bench-top centrifuge at 17,000  g. Remove the supernatant and resuspend the pellet in 100 μL of 1  PBS. Add 400 μL of 100% methanol, touch vortex, and store at 20  C (see Note 4). 5. Add rifampicin to the cell culture to a final concentration of 100 μg/mL and cephalexin to a final concentration of 15 μg/ mL. Touch vortex. For antibiotic concentrations adjust the current volume by subtracting the amount of any culture removed to measure absorbance. Antibiotic addition is time 0. 6. Grow the culture for 2 h in the presence of rifampicin and cephalexin and then take a sample as in step 4 (see Note 5). 7. For a genome copy number control grow wild-type E. coli in 5 mL minimal media containing 0.32% glycerol. Grow to stationary phase (overnight growth) and then carry out steps 5 and 6. These cells provide a 1n + 2n genome copy number control reflecting cells in the population prior to antibiotic addition that contained either no replication forks or two forks (Fig. 1c, left panel).

Escherichia coli Genome Content Measurement

3.2

dnaAts Mutants

151

Steps 3–9 should be carried out as quickly as possible. 1. Inoculate 5 mL of LB with a single colony of your dnaAts strain. We have tested several different dnaAts alleles and have found that dnaA46 [7] provides the highest fraction of cells that initiate replication within the permissive temperature time window (see step 6 below). Grow at 30  C to an OD650 of 0.2–0.3. 2. Transfer tubes to a 42  C shaking (180 RPM) water bath and grow for 2 h. At this point move LB to an 18  C and a 55  C incubator to prepare for steps 4 and 7 (5 mL and 10 mL per strain, respectively). 3. Take a 100 μL sample and transfer to a 1.5 mL tube. 4. Add 5 mL of LB pre-chilled to 18  C directly to the culture and move to the 30  C water bath. 5. Process the sample taken in step 3. Spin the sample down for 1 min in a bench-top centrifuge at 17,000  g. Remove the supernatant and resuspend the pellet in 100 μL of 1  PBS. Add 400 μL of 100% methanol, touch vortex, and store at 20  C. 6. After 10-min incubation at 30  C, take a 100 μL sample and transfer to a 1.5 mL tube. 7. Add 10 mL of LB pre-warmed to 55  C directly to the culture and transfer to the 42  C water bath. 8. Process the sample taken in step 6 as in step 5. 9. Take 400 μL samples (as in step 5) at 10-min intervals for 70 min (80 min after the shift to 30  C in step 4). An optional sample can be taken 2 h after step 4 (see Note 5).

3.3 Staining Cells for Flow Cytometry

1. Spin the fixed samples for 1 min in a bench-top centrifuge at 17,000  g. 2. Remove the supernatant and resuspend the pellet in 500 μL of 1  PBS. 3. Spin the samples for 1 min in a bench-top centrifuge at 17,000  g. 4. Remove the supernatant and resuspend pellets in 1 mL of Sytox mix (see Note 6). 5. Transfer the samples to flow cytometry tubes. All stained cells should be kept out of the light using dark tubes or foil and stored at 4  C (see Note 7).

152

4

Michelle Hawkins et al.

Notes 1. Make up the working stocks (50 μM Sytox® Green in DMF, 1 mg/mL RNase A in dH2O) the night before they are required for the Sytox mix. The working stocks can be kept for 1 week at 20  C. 2. Depending on your strain background, it may not be necessary to grow the culture to exponential phase twice. In such cases proceed to step 3 after step 1. 3. Growth to exponential phase takes 2–3 h for wild-type E. coli strains. 4. This sample is from an exponential phase population. Since the population will contain cells at every stage of genome replication it will produce a continuous distribution of genome content. This is not informative in most cases and probably only needs carrying out once per strain to check for gross replication defects such as abnormal overall DNA content. 5. Once cells have been fixed with methanol they can be stored at 20  C indefinitely before processing and staining. 6. At this stage the pellet is invisible. Depending on the specifications of the flow cytometer used for analysis, you may need to dilute samples to avoid exceeding the flow rate at which data collection is accurate. An appropriate dilution should be optimized empirically but for guidance a recommended suspension density is 5  105 to 5  106 cells/mL [14]. In our experience diluting cells twofold in Sytox mix gives good results with a Beckman Coulter CyAn™ ADP Analyzer. We dilute half the stained cells so that we can also run undiluted samples if necessary later. 7. Stained cells kept in the dark will still eventually bleach. We recommend storing stained E. coli for no longer than two months. Touch vortex samples before flow cytometry analysis to avoid cell clumps.

Acknowledgments We thank Karen Hogg and the Imaging and Cytometry Laboratory in the Technology Facility at the University of York for technical assistance and support. This work was supported by BBSRC grant BB/I001859/1 and the University of York.

Escherichia coli Genome Content Measurement

153

References 1. Jameson KH, Wilkinson AJ (2017) Control of initiation of DNA replication in Bacillus subtilis and Escherichia coli. Genes (Basel), 8 2. Hansen FG, Atlung T (2018) The DnaA tale. Front Microbiol 9:319 3. Mott ML, Berger JM (2007) DNA replication initiation: mechanisms and regulation in bacteria. Nat Rev Microbiol 5:343–354 4. Helmstetter CE, Cooper S (1968) DNA synthesis during the division cycle of rapidly growing Escherichia coli B/r. J Mol Biol 31: 507–518 5. O’Donnell M, Langston L, Stillman B (2013) Principles and concepts of DNA replication in bacteria, archaea, and eukarya. Cold Spring Harb Perspect Biol:5 6. Myka, K.K., Hawkins, M., Syeda, A.H., Gupta, M.K., Meharg, C., Dillingham, M.S., Savery, N.J., Lloyd, R.G. and McGlynn, P. (2017) Inhibiting translation elongation can aid genome duplication in Escherichia coli. Nucleic Acids Res, 45, 2571-2584 7. Atkinson J, Gupta MK, Rudolph CJ, Bell H, Lloyd RG, McGlynn P (2011) Localization of an accessory helicase at the replisome is critical in sustaining efficient genome duplication. Nucleic Acids Res 39:949–957

8. Campbell EA, Korzheva N, Mustaev A, Murakami K, Nair S, Goldfarb A, Darst SA (2001) Structural mechanism for rifampicin inhibition of bacterial RNA polymerase. Cell 104:901–912 9. Bush K (2012) Antimicrobial agents targeting bacterial cell walls and cell membranes. Rev Sci Tech 31:43–56 10. Skarstad K, Boye E, Steen HB (1986) Timing of initiation of chromosome replication in individual Escherichia coli cells. EMBO J 5: 1711–1717 11. Hansen, F.G., Koefoed, S. and Atlung, T. (1992) Cloning and nucleotide sequence determination of twelve mutant dnaA genes of Escherichia coli. Mol Gen Genet, 234, 14-21 12. Robinson JP (2018) Overview of flow cytometry and microbiology. Curr Protoc Cytom 84: e37 13. Roth BL, Poot M, Yue ST, Millard PJ (1997) Bacterial viability and antibiotic susceptibility testing with SYTOX green nucleic acid stain. Appl Environ Microbiol 63:2421–2431 14. Givan AL (2011) Flow cytometry: an introduction. Methods Mol Biol 699:1–29

Chapter 12 Dynamics of Bacterial Chromosomes by Locus Tracking in Fluorescence Microscopy Leonardo Mancini, Estelle Crozat, Avelino Javer, Marco Cosentino Lagomarsino, and Pietro Cicuta Abstract In the last two decades, it has been shown that bacterial chromosomes have remarkable spatial organization at various scales, and they display well-defined movements during the cell cycle, for example to reliably segregate daughter chromosomes. More recently, various labs have begun investigating also the short time dynamics (displacements during time intervals of 0.1 s–100 s), which should be related to the molecular structure. Probing these dynamics is analogous to “microrheology” approaches that have been applied successfully to study mechanical response of complex fluids. These studies of chromosome fluctuation dynamics have revealed differences of fluctuation amplitude across the chromosome, and different characters of motion depending on the time window of interest. Different fluctuation amplitudes have also been observed for the same chromosomal loci under antibiotic treatments, with magnitudes that are correlated to changes in intracellular density and thus crowding. We describe how to carry out tracking experiments of single loci and how to analyze locus motility. We point out the importance of considering in the analysis the number of GFP molecules per fluorescent locus, as well as the nature of the protein they are fused to, and also how to measure intracellular density. Key words Chromatin, Bacterial nucleoid, Loci and foci, Fluorescence imaging, Mean squared displacement, Polymer dynamics

1

Introduction to Live-Cell Locus Tracking at High Frame Rate In recent years, it has started to be recognized that the physical structure of the bacterial nucleoid plays an important role in fundamental biological processes, e.g., segregation and transcription (Fig. 1). This chapter focuses on methods to characterize chromosomal dynamics at short timescales (intervals between 0.1 s and 100 s), where the motion should be sensitive to the state of the nucleoprotein complex and its environment, without being dominated by large-scale motions related to the genome segregation and cell growth. The movement of fluorescent chromosomal labels in live Escherichia coli can be obtained by custom-designed single

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_12, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

155

156

Leonardo Mancini et al.

Fig. 1 (a) Schematic illustration of supercoiled chromosome sections, giving a branched spatial structure; (b) illustration of the structure and condensation inducing the role of nucleoid-associated proteins (NAP). Adapted from Ref. [3]

particle tracking programs and we describe one here. The collection of a large number of trajectories is crucial, making it possible to characterize not just average chromosomal dynamics, but also binning for time in the cell cycle, or subcellular localization, or number of foci, or a combination of those, to tease apart some of the diverse factors that affect chromosome mobility [1, 2]. The bacterial chromosome spontaneously condenses, together with its associated RNA and proteins, into the cell’s central region to form a structure called “nucleoid” [4, 5], which is a highly organized and dynamic structure [6–9]; see Fig. 1. It is now recognized that the physical properties and organization of the nucleoid are intimately related to important cellular functions, particularly chromosomal segregation [7, 10–13] and gene expression [14, 15]. Bacteria are particularly interesting organisms for a variety of reasons, including a “physics” perspective. Indeed, they are in some ways simpler than eukaryotic cells. They do not have organelles or internal membranes, their volume is three orders of magnitude smaller than a typical eukaryote cell, and they have a very compact genome [16], and yet have the extraordinary complexity of life. E. coli is by far the best characterized bacterium. Its genetics and molecular biology are well understood, it is easy to maintain, and a large collection of strains and tools for manipulation is available. Therefore E. coli is very often the first choice in studies of any general aspect in bacterial biology. The development of systems that allow tagging specific chromosomal loci, using fluorescent proteins through genetic engineering, has allowed the in vivo exploration of the dynamics of specific chromosomal regions [17, 18]. Most of the studies of chromosomal markers have focused on the segregation dynamics (some examples in Refs. [9, 18–21]). Remarkably, only a handful of

Dynamics of Bacterial Chromosomes by Locus Tracking in Fluorescence Microscopy

157

experimental studies have addressed the problem of characterizing the chromosomal structure from a physics viewpoint [6, 22–27]. In this chapter we present an approach inspired by microrheology, where, through the analysis of the movement of individual probes, it is possible to extract details on the local environment [28– 32]. Particularly important in this respect is the pioneering work of Weber et al. [25–27, 33], where it is shown how the chromosomal markers exhibit different dynamics than what is expected from free diffusion or standard polymer physics models [33]. They attributed this behavior to the viscoelastic properties of the cytoplasm, and in a subsequent work, they found that the amplitude of the markers’ vibrations depends on a parameter as central to cell functioning as ATP [26]. This suggests that the dynamics of chromosome and cytoplasm may be intimately linked to the physiological state of the cell. Indeed, system-level perturbations such as those imposed by certain antibiotic treatments are capable of influencing the behavior of the markers [34, 35]. Macromolecular crowding of the cell cytoplasm is one of the mechanisms linking antibiotic action and changes in chromosomal dynamics [35]. In this chapter, we describe the series of tools we developed to analyze the movement of the chromosomal markers. This includes the creation of a program to track each individual fluorescent focus with subpixel resolution, as well as an analysis of the different types of errors that could arise from it, and a series of strategies to correct them. As examples, we present the key results obtained in our lab, on the short time dynamics in a collection of 27 strains where each strain has a fluorescent marker in a different chromosomal position. The effect of three different growth media is also considered. We also show that the choice of the fluorescently labeled protein is critical, as one of the most widely used ParB(P1)-GFP seems to enhance the cohesion of replicated chromosomes, in the Ter domain of E. coli, while this effect is not detectable elsewhere on the chromosome [36]. The markers all share a similar subdiffusive behavior, but their mobility depends on (a) the chromosomal localization, (b) the subcellular position, and (c) the fluorescence intensity (which we have tested to be a GFP locus size effect as opposed to an error of localization) [1]. Furthermore, the use of a matP mutant, a protein binding specifically to the Ter region of the chromosome, shows that such a method can help to characterize the role of proteins in vivo; in this case, we showed that MatP, while keeping replicated Ter regions together, does not impede chromosomal dynamics outside of this event. The high number of tracks allowed us to decipher between unreplicated Ter region and replicated, but not segregated, ones, which is not possible with a simple analysis of fluorescence [36]. Lastly, we describe a protocol for the estimation of intracellular density in bulk, which can be used as a proxy for molecular crowding [35]. Again, we will use as an

158

Leonardo Mancini et al.

example results obtained in media and strains used in our lab, but the protocol can be generalized to other strains, species, and conditions.

2

Materials

2.1 Bacteria Strains and Reagents

We initially studied a collection of 27 E. coli strains with the GFP-ParB(P1)/parS fluorescent label system (kindly provided by the Espeli and Boccard laboratories [20]). This collection of strains has a P1 parS inserted at 27 different positions around the chromosome; loci are assigned a name according to the macrodomain they belong to. The expression of the ParB-GFP fusion protein is driven by the pALA2705 plasmid; no IPTG induction is required to produce the ParB-GFP levels necessary to visualize loci in the cells [21]. As a control, IPTG can be added half an hour before visualization. Another set of E. coli strains was used in a following study [36], with parS sites from pMT1 inserted at nearly the same loci as the parS(P1) (Espeli and Boccard laboratories). In this case, a small amount of IPTG (30 μM) was necessary to image the foci; ParB was expressed from pFHC2973 [21]. All the chemical reagents were obtained from Sigma-Aldrich unless stated otherwise. Estimation of bacterial intracellular density can be performed in virtually any medium that supports E. coli growth and at any growth stage.

2.2

The E. coli cells on agar pads were imaged on a Nikon Eclipse Ti-E inverted microscope using a 60
oil-immersion objective with NA 1.45. The images were further magnified with a 2.5
TV adapter before detection on an Andor iXon EM-CCD camera. The camera used in EM gain mode, capable of detecting single fluorophores, yields a high signal-to-noise ratio which in turn gives a high loci localization precision. Field-of-view scanning and image acquisition were automated using custom-written software.

3

Microscopy

Methods

3.1 Sample Preparation

Strains were grown overnight in LB at 37  C with ampicillin 100 μg/mL, or in M9 supplemented with complementary salts (MgSO4 2 mM, CaCl2 100 μM, tryptophan 4 μg/mL, and thymidine 5 μg/mL), and 0.4% glucose. Cultures were diluted 200:1 into M9 minimal salts (Difco) and, if required, a different carbon source: 0.4% glycerol, 150-min doubling time; 0.4% glucose (Surechem Products), 115-min doubling time; or 0.4% glucose and 0.5% casamino acids (Difco), 75-min doubling time. For observations on agar pads, 10 μL of this culture was deposited on a pad swollen of the same medium used for the exponential

Dynamics of Bacterial Chromosomes by Locus Tracking in Fluorescence Microscopy

159

growth and 1.5% of agarose. The pad was then sealed between two coverslips with silicone grease, preventing bacteria motility but allowing bacteria growth. The sample was kept on the microscope at 30  C during the data acquisition for typically 40 min, waiting for 20 min before taking the first video. 3.2

Video Acquisition

During a typical experiment, up to 30 fields of view were chosen manually, and then sequentially scanned automatically. Focus was maintained while scanning using the Nikon “perfect focus” hardware autofocus system. Movies were taken for 45 s at a frame rate of 9.6 fps with an exposure time of 104 ms producing approximately 425 frames. Agar experiments were performed at 30.0  C.

3.3

Image Analysis

A variety of different algorithms have been developed to track individual particles. Due to the variety of the studied systems, there is not a single gold standard. However, independently of each individual implementation, it is possible to distinguish three required steps in all algorithms [37], each of them with its own challenges and potential sources of error: 1. Localization or identification of particles in individual frames: In the case of diffraction-limited spots, the most important constraint is the signal-to-noise ratio, SNR, where the signal is the spot intensity, and the noise is the intensity fluctuations over the background. It was shown before that at SNR 9 is considered an outlier, and will be excluded from the mean and standard deviation calculation. This step makes use of the MATLAB function (spatialMovAveBG) developed by Jaqaman et al. [38]. A5. Identify the statistically meaningful local maxima. Considering 3
3 pixel neighborhoods, for each pixel of the original image, we select the ones for which the central pixel is a maximum. In order to determine if these local maxima have statistical significance, they are tested against the background mean and standard deviation (obtained at step A4) using the cumulative density function of a normal distribution. The local background intensity and noise are used as the μ and σ parameters of the distribution. Only maxima in the 90% tail of the distribution are considered as valid particles. This part of the procedure follows closely Jaqaman et al. [38] and makes use of the function (locmax2d) from that work. B. Subpixel resolution detection of the position. Using the original unprocessed images, the regions around the candidate particles are fitted to a 2D Gaussian (Fig. 3). B1. Using the raw images, a region of 5
5 pixels centered in the local maximum obtained above is selected. Each region is fitted to the function [37]     2  2 2 Z ðx, y Þ ¼ I 0 exp  ðx  μx Þ þ y  μy =2s þ B,

162

Leonardo Mancini et al.

Fig. 3 Subpixel resolution detection of the position. (a) Raw image (scale bar 1 μm). (b) 3D representation of the zoomed region. In order to obtain subpixel resolution, each peak is fitted to a 2D gaussian. (c) Typical resulting trajectory for 430 frames

where s ¼ 0:21 λ=ðNA  pÞ, λ is the wavelength, NA is the microscope numerical aperture, and p is the pixel size. In our experiments, λ ¼ 510 nm, NA 1.4, and p ¼ 106 nm. B2. The difference between the local maxima position and the correction given by μy and μx is computed. The data is rejected as an artifact if μ2x + μ2y > 2. C. Linking of the trajectories. In this step the positions detected along the different time frames are assembled to reconstruct the trajectories. C1. The closest neighbors between two consecutive frames are assigned to the same trajectory, except if their separation is larger than two pixels. C2. Particles not assigned to any trajectory in the previous frame are considered as new trajectories. C3. If the end of a trajectory is separated by less than two pixels and less than three frames from the beginning of another trajectory, the two trajectories are joined. The frames between the two original trajectories, where no particle was identified, are considered as “not a number” (NaN). Displacements that require to include NaN data are excluded from the analysis.

Dynamics of Bacterial Chromosomes by Locus Tracking in Fluorescence Microscopy

3.4 Considerations on Precision and System-Specific Aspects of the Localization Data

163

The algorithm described above allowed us robust and unsupervised analysis of thousands of movies, proving hundreds of thousands of locus tracks. Of the many possible sources of error, it is worth focusing here just on the fundamental limit in the precision of localization that comes from the number of photons collected during an exposure. This was calculated in [40] as
ε2 ¼ s 2 =N þ a 2 =12=N þ 4 √π s 3 b 2 = a N 2 , where ε2 is the static error, s is the standard deviation of the pointspread function, a is the pixel size, N is the number of photons collected, and b is the background noise. For a given exposure time, N is proportional to the intensity registered by the camera. The first term arises from the fluctuations in collected photons at a given time due to the Poissonian nature of photon emissions, the second term is a consequence of the finite size of the pixels, and the third term considers the effect of background noise. This equation indicates that as the intensity decreases the static error drastically increases. With the ParB/parS system there is typically a large distribution of intensities, and also under repeated imaging there is photobleaching. So the localization of some loci (increasing fraction during multiple acquisitions) will be subject to non-negligible “static error.” For accurate statistics, it is needed to take this variable error into account, and/or consider only loci above a threshold intensity for which the static error is negligible. This can easily be tested: we carried out tests by changing the intensity of the excitation illumination, ideally on the same field of view, and verifying the lower illumination threshold at which localization error becomes noticeable (Fig. 4). We also compared motility of loci in living cells with loci in fixed cells, and we generated artificial datasets as in silico “null models” [1, 2].

Fig. 4 The static error level is lower than the MSD(0.1 s) at every intensity range. The predicted levels of errors (gray circle) are smaller than the intensity-binned ensemble average MSD at lag times of 0.1 s (blue upward triangles), but only at larger lag times (10s, green downward triangles) the error is expected to be negligible (at least 10 times smaller than the MSD). Data from either (a) Ori1 or (b) Ter3 locus

164

Leonardo Mancini et al.

Fig. 5 Error subtraction increases the MSD slope and unmasks the effect of dynamic error. The graphs show the result (blue circles) of subtracting the predicted static error (gray triangles) from the raw MSD (gray squares). The MSDs are calculated using only data with narrow-intensity window on data from Ori1 (a) and Ter3 (b) loci. In Ori1, after the error subtraction, the apparently linear curve shows a downturn at short lag times. The effect is likely to be the dynamic error previously hidden by the static error. In the Ter3 locus, that has lower motility, the effect of static error is still present suggesting that this locus could be close to the analysis limit, although the presence of a different mobility regime cannot be discarded

The static error and the real displacement are independent; therefore, it is possible to subtract the error from the observed MSD (“MSDraw”). As reported in ref. [1], the loci in the Ter domain have a lower mobility; therefore, similar levels of static error will have a larger effect on the MSD (Fig. 5a). MSDs from more mobile loci typically present a quite linear behavior in the log-log plots (Fig. 5b). However, once the static error is subtracted, the resulting MSD has a downturn at short lag times for Ori1, most likely a consequence of the dynamic error. The dynamic and static errors have opposite contributions. Since the exposure time and the lag time in our data are equal (movies are under continuous illumination), the dynamic error is not negligible, and can be masked under the effect of the static error. 3.5 Choice of the Protein Fusion to Label Foci

In our first experiments [1, 2], we used ParB(P1)-GFP to image the foci. Since then, it became clear that this version of ParB increased post-replicative cohesion of tagged loci (Fig. 6) [36]. We therefore tested another version of ParB, pMT1, and found that the postreplicative cohesion was greatly reduced in the Ter region of the chromosome. Especially, the large drop in mobility observed on the highest intensity loci when using ParB(P1) was not observed with ParB(pMT1) and the number of segregated foci was closer to the one observed with other methods [41]. This does not undermine previous results, since over a large range of loci intensity the motility is quite independent on locus size, and ParB-P1 has been used successfully to label other parts of the chromosome [21, 41, 42]. It is possible that its tendency toward cohesion can itself be a useful physical probe of chromosome structure, albeit of a more

Dynamics of Bacterial Chromosomes by Locus Tracking in Fluorescence Microscopy

165

Fig. 6 Different behaviors of fluorescently labeled ParB proteins. The figure shows the medians of MSD of ParB-P1 (P1) and ParB-pMT1 (T1) foci at 5 s as a function of intensity of the foci, for two genetic loci: Ori2 (O2), near the origin of replication, and Ter3 (T3) or Ter4 (T4), near the terminus of replication. While the mobility of O2 foci decreases slowly with the increase of intensity, a sharp fall is observed at Ter when it is labeled with ParB-P1 protein, illustrating a probable aggregation of the protein. Error bars show the standard error of the mean. Adapted from Crozat et al. (2020) [36]

perturbative nature. Still, switching to the pMT1 version of the tag is prudent if one wishes to have a tag with the minimal perturbation of the chromosome. 3.6 Considerations on the Correlation of Loci Motility to Intensity

In using ParB(P1)-GFP as in [1, 2] we showed a quite dramatic “slowing down” of loci motility with increasing locus fluorescence. If data is scarce, one may want to use all loci of all intensities, rather than “gate” for a narrow window. One pragmatic approach to that is presented in [34] and is basically a normalization of the motility to that of a “reference” intensity. Note that in the presence of photobleaching, as also discussed in [34], one should recalculate the non-photobleached intensity before applying such a normalization procedure. These are all complications that arise from a labeling that does not have defined number of units.

3.7 Cellular Density Estimation

In diluted samples, the optical density (traditionally measured at 600 nm for bacteria) is a better proxy of the biomass than of the cell number. Indeed, the latter is susceptible to changes in cell size as larger cells are more likely to scatter light and will therefore yield higher optical densities. While bacteria like E. coli change size substantially during their growth and across different conditions, they tend to maintain a constant density inside the cell such that their biomass per OD600 unit remains constant. Some conditions however, such as in the presence of certain antibiotics or osmotic shocks, are capable of perturbing density homeostasis. These

166

Leonardo Mancini et al.

Fig. 7 Estimation of intracellular density in bulk cultures and controls. (a) The OD600 scales with biomass. (b) The repeated wash in water or low-salt buffers for estimating biomass can cause cell lysis and must be controlled for. (c) RI changes linearly with BSA concentration. The 1.79 slope is from [46]. (d) Progressive addition of BSA zeroes the OD of the cells. The BSA concentration that zeroes the OD can then be mapped onto (c) to estimate the refractive index of the cells

changes to the intracellular density can be measured by plotting the OD600 against cell mass per unit volume of culture (Fig. 7a). To obtain OD600 measurements, aliquots of culture are diluted to yield a measured optical density within the linear range of the spectrophotometer (commonly between OD600 0.05 and 0.4) [43]. The biomass of cells can instead be measured in the form of dry mass [44]. Relatively large cultures (>200 mL) are harvested by centrifugation at 8000
g for 10 min and resuspended in 200 mL water.

Dynamics of Bacterial Chromosomes by Locus Tracking in Fluorescence Microscopy

167

Cells are washed again, resuspended in 2 mL water, and placed on pre-weighted aluminum boats. These are then left to dry overnight at 70  C and weighted again. 3.8 Control of Cell Lysis

Depending on the physiological state of the cells, the washing treatment that aims at eliminating salts and other extracellular contents of the culture may be harsh enough to cause cell lysis [45], which would affect biomass estimations. To check whether repeated washes are changing cell numbers, it is sufficient to compare the OD600 of the samples after each wash (Fig. 7b).

3.9 Refraction Index Estimation

Another assumption that is made when taking optical density readings of cells is that the refractive index (RI) of both cells and medium remains constant. Should the difference between the two RIs increase, also the OD would, while cells that have a RI equal to the one of the medium have zero optical density. It is possible to take advantage of this dependency to estimate the RI of cells and thus ensure that it is not changing between different time points or conditions. The addition of bovine serum albumin (BSA) to the medium changes the RI in a known way (Fig. 7c) such that to each BSA concentration corresponds a RI. Progressively adding BSA to the medium decreases the optical density of the cell suspension. The concentration that zeroes the OD of the cells can be mapped onto the [BSA] to RI curve to infer the RI of the cells (Fig. 7d).

4

Notes 1. Broad foci intensity distributions that extend over one or two decades of total fluorescence are typical in the analyzed samples. These distributions are observed under constant illumination even in individual experiments and they seem to be a consequence of variation of GFP-ParB expression in individual cells. Therefore, we expect to observe drastic differences in the level of static error between individual foci. A common procedure to determine the ε2 is to apply the tracking algorithm to movies of immobile particles [47]. The recorded motion is then a measurement of the tracking precision of the experimental setup. Here, to avoid the effect of photobleaching we used fluorescent beads dried over a coverslip. The processed signal is altered by inserting different neutral density filters into the fluorescent excitation path. As expected, the observed static error is drastically affected with the recorded intensity. While for the beads the effect can be entirely ascribed to changes in precision of the image analysis procedure, the focus intensity has an important effect on its dynamics; see Fig. 4. This effect is independent of tracking errors and rather a consequence of

168

Leonardo Mancini et al.

“physics” of size, or binding to DNA of the GFP-ParB complex on the chromosome. Further work will be required on this aspect and relating it to polymer physics models. 2. Each locus track is analyzed to extract an individual MSD, as a function of lag time τ, and all tracks from a given locus are averaged using both sliding and non-sliding means. Time- and ensemble-averaged MSD(τ) data of each locus is fitted with the function MSDðτÞ ¼ 4Dapp τα , where the fitted parameters are 4Dapp and α which characterize the amplitude and degree of subdiffusivity of the motion. 3. Each movie contains a set of a minimum of 10 moving loci, typically about 35. The typical tracks were composed of about 450 frames separated by 0.104-s intervals. To test the effect of finite track length, measurements were compared with simulations of fractional Brownian motion generated to obtain the same number of tracks and frames per track with similar Dapp considering the two cases α ¼ 0.4 and α ¼ 0.5. Only loci whose trajectories are followed for at least 280 frames are used for averages (more than 80% of the tracks are 430 frames long). For each experimental run, 15–30 movies are recorded. Each set of movies (biological replicate) typically generates between 500 and 2000 individual loci tracks.

Acknowledgments We are very grateful to K. Dorfman, V.G. Benza, B. Sclavi, A. Spakowitz, O. Espeli, P.A. Wiggins, N. Kleckner, L. Mirny, and G. Fraser for helpful discussions; Z. Long, E. Nugent, M. Grisi, K. Siriwatwetchakul, J. Kotar, and C. Saggioro for their help with the experimental setups and bacterial strains; and O. Espeli and F. Boccard for the gift of bacterial strains developed in their laboratory. This work was supported by the International Human Frontier Science Program Organization, grant RGY0070/ 2014, and Consejo Nacional de Ciencia y Tecnologia (CONACYT) and UKRI grant EP/T002778/1. References 1. Javer A, Long Z, Nugent E et al (2013) Shorttime movement of E. coli chromosomal loci depends on coordinate and subcellular localization. Nat Commun 4:1–8. https://doi.org/ 10.1038/ncomms3003 2. Javer A, Kuwada NJ, Long Z et al (2014) Persistent super-diffusive motion of Escherichia

coli chromosomal loci. Nat Commun 5. https://doi.org/10.1038/ncomms4854 3. Wang X, Llopis PM, Rudner DZ (2013) Organization and segregation of bacterial chromosomes. Nat Rev Genet 14:191–203 4. Valkenburg JAC, Woldringh CL (1984) Phase separation between nucleoid and cytoplasm in

Dynamics of Bacterial Chromosomes by Locus Tracking in Fluorescence Microscopy Escherichia coli as defined by immersive refractometry. J Bacteriol 160:1151–1157. https:// doi.org/10.1128/jb.160.3.1151-1157.1984 5. Mondal J, Bratton BP, Li Y et al (2011) Entropy-based mechanism of ribosomenucleoid segregation in E. coli Cells. Biophys J 100:2605–2613. https://doi.org/10.1016/j. bpj.2011.04.030 6. Wiggins PA, Cheveralls KC, Martin JS et al (2010) Strong intranucleoid interactions organize the Escherichia coli chromosome into a nucleoid filament. Proc Natl Acad Sci U S A 107:4991–4995. https://doi.org/10.1073/ pnas.0912062107 7. Fisher JK, Bourniquel A, Witz G et al (2013) Four-dimensional imaging of E. coli nucleoid organization and dynamics in living cells. Cell 153:882–895. https://doi.org/10.1016/j. cell.2013.04.006 8. Hadizadeh Yazdi N, Guet CC, Johnson RC, Marko JF (2012) Variation of the folding and dynamics of the Escherichia coli chromosome with growth conditions. Mol Microbiol 86: 1318–1333. https://doi.org/10.1111/mmi. 12071 9. Youngren B, Nielsen HJ, Jun S, Austin S (2014) The multifork Escherichia coli chromosome is a self-duplicating and self-segregating thermodynamic ring polymer. Genes Dev 28: 71–84. https://doi.org/10.1101/gad. 231050.113 10. Jun S, Wright A (2010) Entropy as the driver of chromosome segregation. Nat Rev Microbiol 8:600–607 11. Jun S, Mulder B (2006) Entropy-driven spatial organization of highly confined polymers: Lessons for the bacterial chromosome. Proc Natl Acad Sci U S A 103:12388–12393. https:// doi.org/10.1073/pnas.0605305103 12. Lim HC, Surovtsev IV, Beltran BG et al (2014) Evidence for a DNA-relay mechanism in ParABS-mediated chromosome segregation. elife 2014. https://doi.org/10.7554/eLife. 02758 13. Jung Y, Jeon C, Kim J et al (2012) Ring polymers as model bacterial chromosomes: confinement, chain topology, single chain statistics, and how they interact. Soft Matter 8: 2095–2102. https://doi.org/10.1039/ c1sm05706e 14. Bryant JA, Sellars LE, Busby SJW, Lee DJ (2014) Chromosome position effects on gene expression in Escherichia coli K-12. Nucleic Acids Res 42:11383–11392. https://doi.org/ 10.1093/nar/gku828 15. Cagliero C, Grand RS, Jones MB et al (2013) Genome conformation capture reveals that the

169

Escherichia coli chromosome is organized by replication and transcription. Nucleic Acids Res 41:6058–6071. https://doi.org/10. 1093/nar/gkt325 16. Alberts B, Johnson A, Lewis J et al (2002) Molecular biology of the cell, 4th edn. Garland Science, New York 17. Reyes-Lamothe R, Wang X, Sherratt D (2008) Escherichia coli and its chromosome. Trends Microbiol 16:238–245 18. Wang X, Liu X, Possoz C, Sherratt DJ (2006) The two Escherichia coli chromosome arms locate to separate cell halves. Genes Dev 20: 1727–1731. https://doi.org/10.1101/gad. 388406 19. Joshi MC, Bourniquel A, Fisher J et al (2011) Escherichia coli sister chromosome separation includes an abrupt global transition with concomitant release of late-splitting intersister snaps. Proc Natl Acad Sci U S A 108: 2765–2770. https://doi.org/10.1073/pnas. 1019593108 20. Espeli O, Mercier R, Boccard F (2008) DNA dynamics vary according to macrodomain topography in the E. coli chromosome. Mol Microbiol 68:1418–1427. https://doi.org/ 10.1111/j.1365-2958.2008.06239.x 21. Nielsen HJ, Li Y, Youngren B et al (2006) Progressive segregation of the Escherichia coli chromosome. Mol Microbiol 61:383–393. https://doi.org/10.1111/j.1365-2958.2006. 05245.x 22. Kuwada NJ, Cheveralls KC, Traxler B, Wiggins PA (2013) Mapping the driving forces of chromosome structure and segregation in Escherichia coli. Nucleic Acids Res 41:7370–7377. https://doi.org/10.1093/nar/gkt468 23. Pelletier J, Halvorsen K, Ha B-Y et al (2012) Physical manipulation of the Escherichia coli chromosome reveals its soft nature. Proc Natl Acad Sci U S A 109(40):E2649–E2656. h t t p s : // d o i . o r g / 1 0 . 1 0 7 3 / p n a s . 1208689109/-/DCSupplemental 24. Hong SH, Toro E, Mortensen KI et al (2013) Caulobacter chromosome in vivo configuration matches model predictions for a supercoiled polymer in a cell-like confinement. Proc Natl Acad Sci U S A 110:1674–1679. https:// doi.org/10.1073/pnas.1220824110 25. Weber SC, Theriot JA, Spakowitz AJ (2010) Subdiffusive motion of a polymer composed of subdiffusive monomers. Phys Rev E Stat Nonlinear Soft Matter Phys 82:011913. https:// doi.org/10.1103/PhysRevE.82.011913 26. Weber SC, Spakowitz AJ, Theriot JA (2012) Nonthermal ATP-dependent fluctuations contribute to the in vivo motion of chromosomal

170

Leonardo Mancini et al.

loci. Proc Natl Acad Sci U S A 109: 7338–7343. https://doi.org/10.1073/pnas. 1119505109 27. Weber SC, Thompson MA, Moerner WE et al (2012) Analytical tools to distinguish the effects of localization error, confinement, and medium elasticity on the velocity autocorrelation function. Biophys J 102:2443–2450. https://doi.org/10.1016/j.bpj.2012.03.062 28. Cicuta P, Donald AM (2007) Microrheology: a review of the method and applications. Soft Matter 3:1449–1455. https://doi.org/10. 1039/b706004c 29. Levi V, Gratton E (2007) Exploring dynamics in living cells by tracking single particles. Cell Biochem Biophys 48:1–15 30. Saxton MJ (1997) Single-particle tracking: the distribution of diffusion coefficients. Biophys J 72:1744–1753. https://doi.org/10.1016/ S0006-3495(97)78820-9 31. Meijering E, Dzyubachyk O, Smal I (2012) Methods for cell and particle tracking. In: Methods in enzymology. Academic Press Inc., pp 183–200 32. Crocker JC, Grier DG (1996) Methods of digital video microscopy for colloidal studies. J Colloid Interface Sci 179:298–310. https:// doi.org/10.1006/jcis.1996.0217 33. Weber SC, Spakowitz AJ, Theriot JA (2010) Bacterial chromosomal loci move subdiffusively through a viscoelastic cytoplasm. Phys Rev Lett 104:238102. https://doi.org/10. 1103/PhysRevLett.104.238102 34. Wlodarski M, Raciti B, Kotar J et al (2017) Both genome and cytosol dynamics change in E. coli challenged with sublethal rifampicin. Phys Biol 14. https://doi.org/10.1088/ 1478-3975/aa5b71 35. Wlodarski M, Mancini L, Raciti B et al (2020) Cytosolic crowding drives the dynamics of both genome and cytosol in Escherichia coli challenged with sub-lethal antibiotic treatments. iScience 23(10):101560. https://doi. org/10.1016/j.isci.2020.101560 36. Crozat E, Tardin C, Salhi M et al (2020) Postreplicative pairing of sister ter regions in Escherichia coli involves multiple activities of MatP. Nat Commun 11:1–12. https://doi.org/10. 1038/s41467-020-17606-6 37. Cheezum MK, Walker WF, Guilford WH (2001) Quantitative comparison of algorithms

for tracking single fluorescent particles. Biophys J 81:2378–2388. https://doi.org/10. 1016/S0006-3495(01)75884-5 38. Jaqaman K, Loerke D, Mettlen M et al (2008) Robust single-particle tracking in live-cell timelapse sequences. Nat Methods 5:695–702. https://doi.org/10.1038/nmeth.1237 39. Gonzalez RC, Woods RE, Prentice Hall P Digital Image Processing Third Edition Pearson International Edition prepared by Pearson Education 40. Thompson RE, Larson DR, Webb WW (2002) Precise nanometer localization analysis for individual fluorescent probes. Biophys J 82: 2775–2783. https://doi.org/10.1016/ S0006-3495(02)75618-X 41. Stouf M, Meile JC, Cornet F (2013) FtsK actively segregates sister chromosomes in Escherichia coli. Proc Natl Acad Sci U S A 110:11157–11162. https://doi.org/10. 1073/pnas.1304080110 42. Nielsen HJ, Ottesen JR, Youngren B et al (2006) The Escherichia coli chromosome is organized with the left and right chromosome arms in separate cell halves. Mol Microbiol 62: 331–338. https://doi.org/10.1111/j. 1365-2958.2006.05346.x 43. Stevenson K, McVey AF, Clark IBN et al (2016) General calibration of microbial growth in microplate readers. Sci Rep 6:1–7. https:// doi.org/10.1038/srep38828 44. Basan M, Zhu M, Dai X et al (2015) Inflating bacterial cells by increased protein synthesis. Mol Syst Biol 11:836. https://doi.org/10. 15252/msb.20156178 45. Tomasek K, Bergmiller T, Guet CC (2018) Lack of cations in flow cytometry buffers affect fluorescence signals by reducing membrane stability and viability of Escherichia coli strains. J Biotechnol 268:40–52. https://doi.org/10. 1016/j.jbiotec.2018.01.008 46. Crespi A, Lobino M, Matthews JCF et al (2012) Measuring protein concentration with entangled photons. Appl Phys Lett 100: 233704. https://doi.org/10.1063/1. 4724105 47. Savin T, Doyle PS (2005) Static and dynamic errors in particle tracking microrheology. Biophys J 88:623–638. https://doi.org/10. 1529/biophysj.104.042457

Chapter 13 Measuring Nuclear Mechanics with Atomic Force Microscopy A´lia dos Santos, Florian Rehfeldt, and Christopher P. Toseland Abstract Atomic force microscopy is an ideal tool to map topography and mechanical properties of materials on the micro- and nanoscale. Here, we describe its application to measure and analyze the mechanics, in particular the effective Young’s elastic modulus E* of the mammalian nucleus in live cells. We present three approaches which enable the mechanics to be probed under varying conditions. This includes fully adhered cells, initially adhered cells which lack an established cytoskeleton, and purified nuclei to study their isolated response. Key words Atomic force microscopy, Young’s modulus, Nucleus, Mechanics, Mechanobiology

1

Introduction The nucleus of eukaryotic cells houses the genetic material. The physical properties of the nucleus are defined by multiple factors which include the lamina, cytoskeleton, chromatin structure, and organization, and they are therefore linked to the regulation of cellular events [1]. While the importance of cellular mechanics is nowadays widely acknowledged and extensively studied, in particular in the context of mechanically guided differentiation of human mesenchymal stem cells [2–4], only recently the nucleus is beginning to receive the same level of attention [5–7]. The mechanical properties of the nucleus are perturbed in many diseases such as cancer and premature-aging syndromes [1]. Moreover, it was recently revealed that DNA damage triggers large-scale mechanical alterations to the nucleus which may promote DNA repair. Such a mechanism may lead to protection which drives therapy resistance [6]. It is therefore important to measure nuclear mechanics to understand regulation and how changes occur. Current methods to measure mechanics of soft biological samples include optical

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_13, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

171

172

A´lia dos Santos et al.

Fig. 1 Sketch of AFM measurement. A flexible cantilever probe indents the nucleus of a cell. The deflection of the cantilever is measured through the reflection of a laser beam onto a four-quadrant photodiode. Analysis of tip indentation versus applied force allows us to measure the effective Young’s modulus E* of the material, which is a measure of elasticity

tweezers, magnetic tweezers, and micropipettes as briefly summarized in [8]. Here, we describe the use of atomic force microscopy (AFM) to probe the nucleus of mammalian cells. AFM is an ideal tool as it can probe the topography as well as the mechanical properties of nuclei on the micro- and nanoscale [3, 9]. In particular, we highlight approaches to directly probe mechanical changes in the nucleus by using (i) initially adhered cells that lack an established cytoskeleton and contractile membrane cortex, which would otherwise mask nuclear mechanics, and (ii) purified nuclei from mammalian cells. The AFM allows probing of the mechanical properties of soft materials by measuring the deflection of a cantilever, as the cantilever tip indents the sample at a certain applied force (Fig. 1). Cantilever tips come in different geometries as described in detail later and are therefore ideally suited to probe subcellular structures.

2

Materials

2.1 Atomic Force Microscope

The measurements described here were performed with an MFP 3D BIO (Asylum Research, Oxford Instruments, Santa Barbara, USA) on top of an inverted fluorescence microscope (IX71, Nikon Instruments, Japan). The combined instrument is mounted on a Halcyonics vibration isolation table (Accurion GmbH, Go¨ttingen, Germany) inside an acoustic enclosure (Asylum Research) to minimize noise.

Measuring Nuclear Mechanics with Atomic Force Microscopy

2.2

Cantilevers

2.3 Coverslip cleaning

173

Measurements are performed using the longer triangular cantilever of a TR 400 PB chip that has a pyramidal tip with an opening angle of 35
(Olympus). 1. Place glass coverslips (see Note 1) in an appropriate coverslip rack and plasma clean for 15–20 min. 2. Remove the rack from plasma generator and transfer it into a glass container. 3. Fill the container with absolute ethanol and place in a water bath sonicator for 5 min (see Note 2). 4. Discard solution and wash coverslips twice in absolute ethanol. Dry for 1–2 h at 70
C. Store coverslips submerged in sterile Milli-Q water (see Note 3). 5. Place clean coverslips in 6-well plates and UV-sterilize them for 15–30 min, before seeding cells.

3

Methods

3.1 Cell Seeding: Full-Adhesion Measurements

1. Seed approximately 50,000 cells per coverslip the day before performing AFM measurements (see Notes 4 and 5).

3.2 Cell Seeding: Initial Adhesion Measurements

1. Seed approximately 100,000–120,000 cells per coverslip 30–45 min before performing AFM measurements (see Notes 4, 6, and 7).

2. Before measurements, cells should be washed once with cell culture media, to remove any dead cells, and transfer coverslip to AFM coverslip holder. Example image is shown in Fig. 2.

2. Before measurements, wash once with cell culture media, to remove any dead or nonattached cells and transfer coverslip to AFM coverslip holder. Example image is shown in Fig. 2. 3.3 Nuclei Seeding: Isolated Nuclei

1. (Optional) Treat coverslips with 0.1 mg/mL poly-D-lysine for 30 min to improve attachment of nuclei. 2. Seed approximately 100,000–120,000 nuclei per coverslip 30 min before performing AFM measurements. 3. Wash with phosphate-buffered saline (PBS) to remove debris and nonattached nuclei (see Note 8). Example image is shown in Fig. 2.

3.4 Cantilever Calibration

1. Place a cleaned and dry coverslip on the coverslip holder and then transfer it to the inverted microscope. 2. Mount the cantilever onto the cantilever holder using forceps (see Note 9).

174

A´lia dos Santos et al.

Fig. 2 Mechanical measurements in the nucleus of fully adhered and initially adhered cells and in isolated nuclei. (a) Bright-field images of HeLa cells after 24-h adhesion (full adhesion), after 1-h adhesion (initial adhesion), and HeLa-isolated nuclei. Scale bar represents 5 μm. (b) E* values measured for the three different conditions—nuclei measured in fully adhered cells display significantly higher stiffness. Mean  SE is shown and P values were calculated using a two-tailed T-test, assuming equal variance; *P < 0.05

3. Place cantilever holder on the AFM head and place it on the inverted microscope. 4. Adjust laser spot on the cantilever end to ensure that sum signal is maximal. 5. Adjust photodetector so that noncontact deflection is as close to 0 V as possible (see Note 10). 6. Perform a single noncontact (or free space) “force curve” sufficiently away from the surface using the full range of the piezo (Fig. 3a) to calibrate for virtual deflection and generate deflection offset values for the correction of force-distance curves. 7. Engage cantilever in contact to the surface of the coverslip as the next calibration step needs to be performed on a hard surface (see Note 11). 8. Once the cantilever is on the surface (middle range of piezo voltage), take a contact force-distance curve covering roughly 20% of the positive deflection range (in our case 2 V out of 10 V from photodiode) (Fig. 3b). This procedure should be performed several times (e.g., 30), and the mean value for the analyzed curves is used as deflection sensitivity (i.e., photodiode voltage per piezo travel length, see Fig. 3c) (see Note 12).

Measuring Nuclear Mechanics with Atomic Force Microscopy

175

Fig. 3 Cantilever calibration. (a) Virtual deflection and deflection offset of cantilever. (b) Deflection-distance curve over a set deflection of approximately 700 nm equaling a voltage of 2.0 V from the photodiode for calibration on glass. (c) Calculation of deflection sensitivity through averaging and fitting of data generated by deflection-distance curves in (b). (d) Thermal spectrum of the cantilever for spring constant calibration using the thermal noise method

9. Withdraw the cantilever and ensure that it is far from the surface. Set deflection to 0 V and capture thermal data (take at least 100 samples) to calibrate the spring constant of the cantilever (see Note 13) by the thermal fluctuation method [10]. Fit the resonance peak in the power spectrum to calculate the spring constant of the cantilever (see Fig. 3d). 10. Increase the distance between cantilever and coverslip significantly to avoid damage and remove AFM head. 11. Add PBS to a coverslip and mount on coverslip holder for calibration of the cantilever in liquid (see Notes 14 and 15). 12. Readjust laser to a maximum sum and deflection to 0 V (see Note 16). 13. Repeat steps 7–11 for deflection sensitivity and thermal calibration in liquid. 14. For subsequent mechanical measurements, use the deflection sensitivity values generated during the calibration in liquid and the spring constant value obtained during the air calibration (see Note 13).

176

A´lia dos Santos et al.

3.5 Mechanical Measurements of the Nucleus

1. After calibration in liquid, remove the AFM head and place coverslips with cells or isolated nuclei on the coverslip holder. Add medium or buffer to the coverslips and replace AFM head at the center of the holder (see Notes 17 and 18). 2. Take a bright-field image of the selected cell/nucleus for future reference. 3. Place cantilever above the center of the nucleus, ensuring that there are no interfering cells in the immediate surroundings. 4. Engage cantilever and lower it toward the nucleus until piezo Z voltage is around the mean value of the range. 5. Run a force-distance curve with an approach-retract cycle of 5–10 μm, at an approach velocity of not more than 5 μm/s and a maximum indentation force of 1.5 nN (see Notes 19–22). Figure 4a shows an example of a force-distance curve taken above a nucleus of an initially adhered cell. It shows a flat curve offset, and contact point and detachment point for approach and retraction curves, respectively. Figure 4b, c shows examples of unsuitable curves for mechanical measurements; Fig. 4b shows anomalies possibly caused by slippage of the cell during measurement and Fig. 4c possibly caused by something in the cell surrounding interfering with the cantilever. 6. Measure the next cell (see Notes 23 and 24).

Fig. 4 Examples of mechanical measurements in live-cell nuclei. (a) Force-distance curve of the nucleus showing approach (red) and retraction (blue) curves. Contact point and detachment points are indicated and a flat offset in the curves is visible. (b) Examples of anomalies in force-distance are indicated with arrows. These can be caused due to slippage or movement of the cell/nucleus during measurements. (c) Offset of the force-distance curve is not flat (indicated by box) and therefore not suitable for curve fitting and E* calculation

Measuring Nuclear Mechanics with Atomic Force Microscopy

3.6 Data Analysis: Calculation of the Young’s Modulus E

177

1. Import curves into IGOR or other suitable analysis software. 2. Correct the offset of the approach force-distance curve by using a linear fit (see Note 25). 3. Calculate sample indentation from the point of contact between the cantilever and the sample. To do this, subtract the deflection from the piezoelectric z value (which describes the distance the cantilever moved into the sample). 4. Calculate the force using Hooke’s law (F ¼ k D, where F is the force, k is the spring constant, and D is the deflection). 5. Fit force-indentation data to a Hertz model (see, e.g., [3]), or other appropriate theoretical model, to calculate the effective Young’s modulus E (see Note 26). In our case we fitted the following function: πF 1  ν2 E¼ δ2 tan α




where ν is the Poisson’s ratio (assumed to be 0.45 [9, 11]), α the opening angle of the pyramidal tip, and F the applied force.

4

Notes 1. No. 1.5 glass coverslips are recommended for AFM measurements. 2. At this point coverslips may be treated with 1–4%(v/v) 3-aminopropyltriethoxysilane (APTES). This will introduce amino groups on the glass surface and can be used for further reaction with glutaraldehyde for improved cell immobilization. 3. As an alternative to plasma cleaning, coverslips can also be cleaned by incubating them in etch solution (5:1:1 ratio of H2O:H2O2 (50 wt. %):NH4OH (28–30%) for 2 h, in a 70
C water bath, as described in [12]). 4. Cells can be cultured following normal procedures before seeding on coverslips. We recommend that cell lines are passaged at least three times and are growing stably before AFM measurements. 5. The number of cells can be adjusted depending on the cell line, treatment, or time of seeding before measurement. There should be enough cells so that searching is not too time consuming but they should be well spaced so that surrounding cells do not interfere with the cantilever during measurements. Moreover, enough time should elapse to enable full adhesion. 6. Time of adhesion varies between cell lines. 30–45 min is optimized for HeLa cells. Other cell lines, or cells under different

178

A´lia dos Santos et al.

treatments, may require different adhesion times. For example, bone marrow-mesenchymal stem cells (BM-MSCs) attach faster and may be fully adhered by 60 min. Cells should be adhered enough that they do not move away when they come into contact with the cantilever; however, in bright-field images, the nucleus should occupy the majority of the cell volume (Fig. 2a). To optimize adhesion times, cells can also be fixed at different time points and stained with phalloidin—there should be minimal actin fiber formation and only at the cell edges. 7. For some cell lines, coverslip coating with poly-D-lysine (0.1 mg mL1) might improve cell attachment at initial adhesion stages. 8. Poly-D-lysine coating will improve attachment of nuclei but it may also increase the amount of debris and membranes attached to coverslip. To reduce this, nuclear isolation steps can be optimized to ensure a cleaner prep [5, 13]. 9. For the measurements on live cells and isolated nuclei, cantilevers are typically chosen with a spring constant in the range of 10–100 pN/nm. Different tip geometries can be used. Keep in mind that there is a trade-off between sharper tips that probe the nucleus more locally but exert higher local stresses (force per area) and tips that are more blunt (or spherical) that average over larger volumes. For the measurements shown in Fig. 2 we used cantilevers with pyramidal tips with an opening angle of 35
(longer cantilever of OMCL-TR400PB, Olympus). 10. At this point, due to electrostatic forces, deflection values may be unstable. Introducing an ozone generator next to the AFM head for a few seconds might help to reduce this effect. 11. If engage was not successful, retract cantilever and check the value of photodetector—it should be 0 V. If the cantilever fails to reach the surface when taking a force-distance curve, check Z voltage value. 12. Take note of all the values generated during calibration as some of these will be needed at a later stage. 13. In general the spring constant calibration in air is much more accurate than in aqueous environments due to less viscous damping. In the case that due to electrostatic attraction no meaningful calibration in air is possible, you need to use the value for the spring constant obtained from liquid. In general, the two values will differ by up to 20%. 14. For calibration in liquid, use PBS or the buffer that will be used during measurements. At this point, the cantilever might also be incubated in buffer with 3% (m/V) BSA (molecular biology grade is recommended) for 30–60 min. This step is optional

Measuring Nuclear Mechanics with Atomic Force Microscopy

179

but could be useful to reduce adhesion of the cantilever when using stickier samples—for example, isolated nuclei or cells treated with toxic drugs. 15. If the cantilever is dry (e.g., has not been incubated with BSA) add a small drop of buffer with a pipette to the cantilever; this will reduce the likelihood of the cantilever breaking when entering the solution. 16. Sample might need a few minutes to equilibrate if deflection is too variable. Avoid using cold buffer to minimize temperature changes in the AFM. 17. Cells can be imaged in regular cell culture media, with supplements such as FBS, and isolated nuclei can be imaged using PBS or other suitable buffers. Be aware of the fact that buffer solutions that are not CO2 independent potentially change their pH due to the lower partial pressure of CO2 in ambient conditions. 18. New coverslips with cells/nuclei should be left to equilibrate for 5–10 min before starting measurements. 19. These parameters can vary depending on the cell type and conditions of measurements. We recommend 10–20 curves on the same point for averaging during analysis and calculation of Young’s modulus. 20. We recommend an indentation force of 1.5 nN for measurements, which when using a cantilever with a spring constant of 30pN nm1 will produce a detectable deflection to the cantilever (50 nm). 21. If the cell moves under the cantilever, discard measurement. If this happens for too many measurements, coating the coverslip with poly-D-Lysine could improve measurements. 22. If the cantilever is sticking to the cell membrane during retraction and not detaching from the cell surface, preincubation with BSA will also reduce this effect. However, treatments to the cells may enhance the cantilever-cell interactions and therefore render the measurements nonviable. It would be better to try different cantilevers—material/shapes. 23. Depending on the quality of the data, we recommend 20–30 cells to be imaged per condition. 24. We recommend that measurements are not performed over long periods of time on the same coverslip (30–60 min). This will avoid increased cellular stress and is especially important for initially adhered cells, as they will continue adhering during measurements. This time frame will be dependent on the cell type and how fast cells adhere.

180

A´lia dos Santos et al.

25. This should be calculated in a “noncontact” region above the cell and the deflection signal in this region should be flat. 26. The theoretical model applied will depend on the tip geometry. The Hertz model gives good and reproducible values for coneshaped tips (and therefore in the first approximation also for pyramids), but relies on several assumptions and has some limitations. For example, the sample is assumed to be much thicker than the probed indentation and should be homogeneous and isotropic. As a result, sample indentation should not generally be larger than 10% of the total height of the sample. The Hertz model also assumes that the contact between the sample and the cantilever is free of adhesion, which is a valid assumption if the approach curve is used for calculating E. While nuclei and cells in general are heterogeneous and not isotropic it is best to refer to the fitted Young’s modulus as the “effective Young’s modulus.”

Acknowledgments We thank the UKRI-MRC (MR/M020606/1) and the Royal Society (IE170270) for funding. References 1. Dos Santos A, Toseland CP (2021) Regulation of nuclear mechanics and the Impact on DNA damage. Int J Mol Sci 22(6). https://doi.org/ 10.3390/ijms22063178 2. Discher DE, Janmey P, Wang Y-l (2005) Tissue cells feel and respond to the stiffness of their substrate. Science 310(5751):1139–1143. https://doi.org/10.1126/science.1116995 3. Engler AJ, Rehfeldt F, Sen S, Discher DE (2007) Microtissue elasticity: measurements by atomic force microscopy and its influence on cell differentiation. In: Methods in cell biology, vol 83. Academic Press, pp 521–545. https://doi.org/10.1016/S0091-679X(07) 83022-6 4. Zemel A, Rehfeldt F, Brown AEX, Discher DE, Safran SA (2010) Optimal matrix rigidity for stress-fibre polarization in stem cells. Nat Phys 6:468. https://doi.org/10.1038/nphys1613 ´ , Hari-Gupta Y, 5. Lherbette M, dos Santos A Fili N, Toseland CP, Schaap IAT (2017) Atomic force microscopy micro-rheology reveals large structural inhomogeneities in single cell-nuclei. Sci Rep 7(1):8116. https://doi. org/10.1038/s41598-017-08517-6 6. dos Santos A, Cook AW, Gough RE, Schilling M, Olszok NA, Brown I, Wang L,

Aaron J, Martin-Fernandez ML, Rehfeldt F, Toseland CP (2021) DNA damage alters nuclear mechanics through chromatin reorganization. Nucleic Acids Res 49(1):340–353. https://doi.org/10.1093/nar/gkaa1202 7. Swift J, Ivanovska IL, Buxboim A, Harada T, Dingal PCDP, Pinter J, Pajerowski JD, Spinler KR, Shin J-W, Tewari M, Rehfeldt F, Speicher DW, Discher DE (2013) Nuclear Lamin-A scales with tissue stiffness and enhances matrix-directed differentiation. Science 341(6149). https://doi.org/10.1126/sci ence.1240104 8. Rehfeldt F, Schmidt CF (2017) Physical probing of cells. J Phys D Appl Phys 50(46):463001 9. Rehfeldt F, Brown AEX, Raab M, Cai S, Zajac AL, Zemel A, Discher DE (2012) Hyaluronic acid matrices show matrix stiffness in 2D and 3D dictates cytoskeletal order and myosin-II phosphorylation within stem cells. Integr Biol 4(4):422–430. https://doi.org/10.1039/ C2IB00150K 10. Hutter JL, Bechhoefer J (1993) Calibration of atomic-force microscope tips. Rev Sci Instrum 64(7):1868–1873. https://doi.org/10.1063/ 1.1143970

Measuring Nuclear Mechanics with Atomic Force Microscopy 11. Kaliman S, Jayachandran C, Rehfeldt F, Smith A-S (2014) Novel growth regime of MDCK II model tissues on soft substrates. Biophys J 106(7):L25–L28. https://doi.org/10.1016/ j.bpj.2013.12.056 ´ , Cook AW, 12. Hari-Gupta Y, Fili N, dos Santos A Gough RE, Reed HCW, Wang L, Aaron J, Venit T, Wait E, Grosse-Berkenbusch A, Gebhardt JCM, Percipalle P, Chew T-L, MartinFernandez M, Toseland CP (2020) Nuclear myosin VI regulates the spatial organization

181

of mammalian transcription initiation. bioRxiv. https://doi.org/10.1101/2020.04.21. 053124 ´ , Cook A, 13. Fili N, Hari-Gupta Y, Dos Santos A Poland S, Ameer-Beg SM, Parsons M, Toseland CP (2017) NDP52 activates nuclear myosin VI to enhance RNA polymerase II transcription. Nat Commun 8(1):1871. https://doi.org/10.1038/s41467-01702050-w

Chapter 14 Copy Number Analysis of the Yeast Histone Deacetylase Complex Component Cti6 Directly in Living Cells Sviatlana Shashkova, Thomas Nystro¨m, and Mark C. Leake Abstract Proteins are one of the key components of cellular life that play a crucial role in most biological processes. Therefore, quantification of protein copy numbers is essential for revealing and better understanding of cellular behavior and functions. Here we describe a single-molecule fluorescence-based method of protein copy number quantification directly in living cells. This enables quick and reliable estimations and comparison of the protein of interest abundance without implementing large-scale studies. Key words Histone deacetylase, Single-molecule, Copy number, Living cells

1

Introduction Proteins are one of the main functional components of cells. Protein abundance is the key in the regulation of protein-protein and protein-nucleic acid interactions as well as enzymatic reactions which comprise most biological processes. This makes quantification of protein copy numbers essential for revealing and better understanding of cellular behavior and functions. Initially, methods for protein number estimation relied on gene expression measurements, such as microarray hybridization [1]. However, protein homeostasis resembles a balance between protein production, including mRNA processing and translation, and degradation. Therefore, reliable quantification can be obtained only by counting protein numbers directly. Multiple large-scale high-throughput studies on eukaryotic proteome, mainly that of the budding yeast Saccharomyces cerevisiae, have been performed using various methods. This includes quantitative mass spectrometry, western blotting, fluorescence-based methods such as widefield and confocal microscopy approaches, as well as flow cytometry [2–8].

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_14, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

183

184

Sviatlana Shashkova et al.

We report an easy manual method based on a single-molecule fluorescence data, which allows for a quick estimation of protein numbers in living eukaryotic cells. We have applied such an approach to the Cti6 protein of the chromatin remodeling system in the budding yeast S. cerevisiae. Cti6 is a component of the large histone deacetylase complex Rdp3L and participates in multiple cellular processes including iron tolerance and metabolism maintenance [9]. Cti6 plays an important role in gene expression regulation via its binding to the SAGA activator complex [10]. Moreover, Cti6 has been reported to convert one of the largest repression machineries, the Ssn6-Tup1 global co-repressor complex, into an activator [9, 11]. Here we report in detail all necessary steps to estimate molecular copy numbers of a protein of interest in living yeast S. cerevisiae cells based on the fluorescent signal. We describe endogenous genomically integrated labeling of the Cti6 protein with a monomeric form of GFP, mGFP, which contains the A206K point mutation [12]. We used a super-resolution Slimfield microscopy setup to visualize Cti6-mGFP directly in living cells. We then applied previously reported MATLAB code [13, 14] to track fluorescent complexes and estimate the intensity of a single fluorescent molecule. We then show how these data can be used to estimate Cti6 copy numbers per cell.

2 2.1

Materials Yeast Strains

1. The BY4741 wild-type yeast S. cerevisiae cells, pmGFPS plasmid [14] (HIS3, GFPmut3 S65G, S72A, A206K). 2. 40% (w/v) Glucose solution. 3. Yeast nitrogen base (YNB) media (with complete amino acid supplement and -his): 1.7 g/L Yeast nitrogen base without amino acids and (NH4)2SO4, 5 g/L (NH4)2SO4, complete amino acid supplement, or -his amino acid supplement as indicated by the manufacturer, pH 5.8–6.0, supplemented with 4% w/v glucose (added after autoclaving). To prepare YNB plates, add 20 g/L agar before autoclaving. 4. Yeast peptone dextrose (YPD) liquid medium (10 g/L yeast extract, 20 g/L bacto-peptone) supplemented with 4% w/v glucose (added after autoclaving): To prepare YPD plates, add 20 g/L agar before autoclaving. 5. Phusion high-fidelity DNA polymerase 2 U/μL kit, dNTP mix. 6. Primers: For mGFP::his3 fragment amplification, forward (Fw) and reverse (Rv) primers contain ~16–20 bp of the 50 and 30 ends, respectively, of the mGFP::his3 fragment from pmGFPS plasmid plus ~50 bp sequences up- (Fv) and

Cti6 Copy Numbers

185

Fig. 1 Primer design for (a) amplification of a mGFP::his3 fragment for further genomic integration to label the Cti6 protein and (b) verification of a correct insertion of the fluorophore sequence

downstream (Rv) of the Cti6 STOP codon (Fig. 1a). For the conformation PCR, forward (Fconf) and reverse (Rconf) primers are sequences of ~16–20 bp located at the 30 end of the CTI6 gene and 50 end of the mGFP::his3 (Fig. 1b). 7. 0.8% Agarose gel with Midori Green or any other nucleic acid stain in 0.5 TAE buffer. 8. 1 kb DNA ladder, 6 loading buffer. 9. 1 M LiAc water solution. 10. 100 mM LiAc water solution. 11. 50% w/v PEG-4000 solution in water. 12. 10 mg/mL Single-strand DNA (we use salmon sperm DNA). 13. 0.02% SDS solution in water. 14. Standard epifluorescence microscope: We used Zeiss Axio Observer.Z1 inverted microscope with Apotome and Axiocam 506 345 camera, and a Plan-Apochromat 100/1.40 Oil DIC M27 objective. 2.2 Super-Resolution Microscopy

1. We utilized the super-resolution fluorescence Slimfield microscope, with ~6 W/cm2 excitation intensity directed onto a sample, 50 mW Obis 488 nm laser, 100 1.49 TIRF 1.49 NA oil-immersion objective, and EMCCD camera (Prime 95B scientific CMOS; Teledyne Photometrics) with 50 nm/ pixel magnification [14, 15]. 2. 125 μL Gene Frames. 3. Plasma-cleaned BK7 glass microscope coverslip (22  50 mm).

186

Sviatlana Shashkova et al.

4. 2 YNB complete liquid medium supplemented with 8% w/v glucose. 5. 2% Agarose solution in Milli-Q water. For the computational analysis, we used a combination of MATLAB, ImageJ FiJi, and Excel software.

3

Methods

3.1 Fluorescent Labeling of the Cti6 Protein

1. Streak the BY4741 wild-type cells from a frozen stock, using a sterile pipette tip on a freshly prepared YPD plate, and incubate at 30  C for at least 24 h. 2. Set a culture in a 14 mL tube by inoculating 3 mL of YPD with cells grown in a YPD plate. Incubate at 30  C, 180 rpm, overnight. 3. Amplify the mGFP::his3 fragment from the pmGFPS plasmid [14] by using Fw and Rv primers, dNTP mix, and Phusion high-fidelity DNA polymerase 2 U/μL kit as recommended by the manufacturer, with the following PCR program: Initiation

98.0°C

– 30 sec

Denaturation

98.0°C

– 10 sec

Annealing

60.6°C

– 20 sec

Elongation

72.0°C

– 60 sec

Final elongation 72.0°C

30 cycles

– 5 min

Keep at 4  C. Verify the size of the fragment (here ~1920 bp) on 0.8% agarose gel by mixing 5 μL of a PCR reaction mix with 1 μL of 6 loading dye and applying onto a gel (100 V, 25 min, in 0.5 TAE buffer). Use 1 kb DNA ladder as a reference for the size of the amplified fragment. Store at 20  C until further use. 4. Inoculate 50 mL of YPD medium with the overnight culture to the final OD600 ~0.2. Grow until mid-logarithmic phase, OD600 ~0.7–1.0. 5. Transform the culture with 70 μL of the PCR mix containing mGFP::his3 fragment by standard LiAc protocol [16]. In brief, mix 50 μl of cells washed with water and 100 mM LiAc, with 240 μL of 50% w/v PEG-4000, 36 μL of 1 M LiAc, 5 μL of 10 mg/mL salmon sperm DNA, and 70 μL of he PCR mix containing mGFP::his3 fragment. Incubate the entire mix at

Cti6 Copy Numbers

187

30  C for 30 min, and then heat-shock at 42  C for 25 min. Wash (see Note 1) and plate the transformation mix onto YNB-his plates. Incubate at 30  C for 3 days. 6. Confirm the colonies for the green signal under the microscope. We used Zeiss Axio Observer.Z1 inverted microscope with Apotome and Axiocam 506 345 camera, and a PlanApochromat 100/1.40 Oil DIC M27 objective. 7. Verify correct integration of the mGFP::his3 fragment by conformation PCR (in detail below). 8. Streak correct clones on a new YNB-his plate (see Note 2), and incubate for 24 h at 30  C. Store at 4  C. For longer storing, collect grown cells, mix in a cryo tube with 1 mL of YP + 20% glycerol solution, and put into 80  C freezer. 3.2 Conformation PCR

1. Extract genomic DNA: take a bit of a transformant into 20 μL of 0.02% SDS, and heat at 98  C for 3 min. Spin down at 3824  g, for 10 s. Mix 10 μL of the supernatant with 10 μL of Milli-Q water. 2. Perform PCR using Phusion high-fidelity DNA polymerase 2 U/μL kit as recommended by the manufacturer, with the PCR program as above but using 58.9  C as an annealing temperature. Verify the size of the fragment (here ~175 bp) on 0.8% agarose gel (100 V, 25 min, in 0.5 TAE buffer).

3.3 Preparing Cells for Microscopy

1. Set overnight cultures of BY4741 wild-type and Cti6-mGFP strains in two 14 mL tubes by inoculating 3 mL of YNB complete with cells grown on plates. Incubate at 30  C, 180 rpm. 2. In the morning, subculture the cells into 2 3 mL of fresh YNB complete medium to the final OD600 ~ 0.2. Grow for 4 h at 30  C, 180 rpm. 3. Spin cells down at 3000 rpm, for 1 min, and remove the supernatant. Resuspend cells in 0.5 mL of fresh YNB complete medium.

3.4 Preparing Microscopy Slides Using 125 μL Gene Frame

1. Remove the larger plastic cover and apply the frame onto a standard microscopy glass slide. 2. Prepare 0.5 mL of 2 YNB complete medium supplemented with 8% w/v glucose in a 1.5 mL Eppendorf tube. 3. Add 0.5 mL of melted 2% agarose water solution, quickly mix by pipetting up and down, take 0.5 mL of obtained mixture, and add onto the slide within the Gene Frame. 4. Quickly apply a plastic cover slip from the Gene Frame package to remove the access and evenly distribute the solution within the well. Let it sit for a few minutes. 5. Carefully remove the plastic coverslip by sliding it off.

188

Sviatlana Shashkova et al.

6. Remove the smaller plastic cover from the frame. 7. Add ca 5 μL of the prepared cell in droplets across the entire pad. Leave it to dry for 3–5 min. 8. Cover the pad with a plasma-cleaned BK7 glass microscope coverslip (22  50 mm), and avoid air bubbles. 3.5

Data Acquisition

1. Place the sample under the Slimfield microscope and locate a cell; adjust the focus to the middle of the cell. 2. Acquire two bright-field images with 10-ms exposure time. 3. Turn off the bright field and acquire 1000 images at 5-ms exposure time with the 488 nm laser at 20 mW power to obtain images of Cti6-mGFP. 4. Repeat for at least 30 cells. 5. Repeat the whole procedure for the nonfluorescent parental strain BY4741; this will be used for autofluorescence correction during the copy number estimation.

3.6 Copy Number Analysis

1. Load the TIF file containing fluorescence data acquisition into MATLAB as n x m pixels by p frames and perform tracking of individual fluorescent spots as described previously [17]. 2. We estimate the fluorescence intensity of an individual mGFP molecule (Isingle) based on the intensity distribution of the Cti6-mGFP spots identified in the last 2/3 of the acquisition frames (see Note 3). At this point, single photobleaching events become more apparent and detectable; therefore, the most common value (the peak) represents the intensity of a single fluorophore (Fig. 2). 3. Load TIF file containing fluorescence data acquisition into ImageJ Fiji software and open the brightest frame.

Fig. 2 Distribution of all Cti6-mGFP spot intensities (a) and spots identified toward the end of photobleaching (b). Peak values are indicated

Cti6 Copy Numbers

189

Fig. 3 Jitter plot of Cti6-mGFP copy numbers per cell across a population of 32 individual cells. Bar indicates the median

4. Choose the “oval” selection shape on the ImageJ tool bar and draw a circle around the cell. Keep the same size of the selection for all measurements. 5. To obtain numeric values of total fluorescence intensity of the cell press “Ctrl” + “M” (Analyze ! Measure). Record Raw Integrated Density number (RawIntDen) in Excel. Repeat for all files containing fluorescence data of BY4741 Cti6-mGFP cells. 6. Repeat steps 1–5 for the BY4741 wild-type fluorescence files. Calculate the average autofluorescence. 7. Subtract the average autofluorescence value from each RawIntDen value recorded from the BY4741 Cti6-mGFP cells. 8. To estimate the copy numbers of Cti6-mGFP per cell, divide obtained values by Isingle. Figure 3 shows distribution of Cti6 numbers per cell across a population of 32 cells. We estimated the mean number of Cti6 molecules as 2368  1029.

4

Notes 1. To increase the transformation efficiency, prior to plating, the cells can be washed and incubated in YPD medium supplemented with 4% w/v glucose for 1 h at 30  C, 180 rpm. 2. As the fragment is genomically integrated, it is very unlikely that it will be lost during cellular division. Therefore, once the correct clones are confirmed, it is not mandatory to have a selective medium. 3. Isingle can also be estimated by imaging purified mGFP immobilized on a coverslip.

190

Sviatlana Shashkova et al.

Acknowledgments This work was supported by the Royal Society Newton International Fellowship Alumni (AL\191025, AL\201039), Knut and Alice Wallenberg Foundation (KAW 2017-0091, KAW 2015.0272), Swedish Research Council (VR 2019-03937), and the Leverhulme Trust (grant RPG-2019-156). References 1. Spellman PT, Sherlock G, Zhang MQ et al (1998) Comprehensive identification of cell cycle-regulated genes of the yeast Saccharomyces cerevisiae by microarray hybridization. Mol Biol Cell 9:3273–3297. https://doi.org/10. 1091/mbc.9.12.3273 2. Ho B, Baryshnikova A, Brown GW (2018) Unification of protein abundance datasets yields a quantitative Saccharomyces cerevisiae proteome. Cell Syst 6:192–205. e3. https:// doi.org/10.1016/j.cels.2017.12.004 3. Tkach JM, Yimit A, Lee AY et al (2012) Dissecting DNA damage response pathways by analysing protein localization and abundance changes during DNA replication stress. Nat Cell Biol 14:966–976. https://doi.org/10. 1038/ncb2549 4. Chong YT, Koh JLY, Friesen H et al (2015) Yeast proteome dynamics from single cell imaging and automated analysis. Cell 161: 1413–1424. https://doi.org/10.1016/j.cell. 2015.04.051 5. Davidson GS, Joe RM, Roy S et al (2011) The proteomics of quiescent and nonquiescent cell differentiation in yeast stationary-phase cultures. Mol Biol Cell 22:988–998. https://doi. org/10.1091/mbc.E10-06-0499 6. De Godoy LMF, Olsen JV, Cox J et al (2008) Comprehensive mass-spectrometry-based proteome quantification of haploid versus diploid yeast. Nature 455:1251–1254. https://doi. org/10.1038/nature07341 7. De´nervaud N, Becker J, Delgado-Gonzalo R et al (2013) A chemostat array enables the spatio-temporal analysis of the yeast proteome. Proc Natl Acad Sci U S A 110:15842–15847. https://doi.org/10.1073/pnas.1308265110 8. Ghaemmaghami S, Huh WK, Bower K et al (2003) Global analysis of protein expression in yeast. Nature 425:737–741. https://doi. org/10.1038/nature02046 9. Puig S, Lau M, Thiele DJ (2004) Cti6 is an Rpd3-Sin3 histone deacetylase-associated protein required for growth under iron-limiting conditions in Saccharomyces cerevisiae. J Biol

Chem 279:30298–30306. doi:https://doi. org/10.1074/jbc.M313463200 10. Papamichos-Chronakis M, Petrakis T, Ktistaki E et al (2002) Cti6, a PHD Domain protein, bridges the Cyc8-Tup1 corepressor and the SAGA coactivator to overcome repression at GAL1. Mol Cell 9:1297–1305. https://doi. org/10.1016/S1097-2765(02)00545-2 11. Hill SM, Hao X, Gro¨nvall J et al (2016) Asymmetric inheritance of aggregated proteins and age reset in yeast are regulated by Vac17dependent vacuolar functions. Cell Rep 16: 826–838. https://doi.org/10.1016/j.celrep. 2016.06.016 12. Zacharias DA, Violin JD, Newton AC, Tsien RY (2002) Partitioning of lipid-modified monomeric GFPs into membrane microdomains of live cells. Science 296:913–916. https://doi.org/10.1126/science.1068539 13. Wollman AJM, Leake MC (2016) Single molecule narrow-field microscopy of protein-DNA binding dynamics in glucose signal transduction of live yeast cells. Methods Mol Biol 1431: 5–15. https://doi.org/10.1007/978-1-49393631-1_2 14. Wollman AJM, Shashkova S, Hedlund EG et al (2017) Transcription factor clusters regulate genes in eukaryotic cells. elife 6:e27451. https://doi.org/10.7554/eLife.27451 15. Laidlaw KME, Bisinski DD, Shashkova S et al (2021) A glucose-starvation response governs endocytic trafficking and eisosomal retention of surface cargoes in budding yeast. J Cell Sci 134. https://doi.org/10.1242/jcs.257733 16. Gietz RD, Schiestl RH (2007) Frozen competent yeast cells that can be transformed with high efficiency using the LiAc/SS carrier DNA/PEG method. Nat Protoc 2:1–4. https://doi.org/10.1038/nprot.2007.17 17. Wollman AJM, Leake MC (2016) Singlemolecule narrow-field microscopy of protein-DNA binding dynamics in glucose signal transduction of live yeast cells. Chromosom Archit Methods Protoc:5–15. https://doi.org/10. 1007/978-1-4939-3631-1_2

Chapter 15 Super-Resolution Microscopy and Tracking of DNA-Binding Proteins in Bacterial Cells Chloe´ J. Cassaro and Stephan Uphoff Abstract The ability to detect individual fluorescent molecules inside living cells has enabled a range of powerful microscopy techniques that resolve biological processes on the molecular scale. These methods have also transformed the study of bacterial cell biology, which was previously obstructed by the limited spatial resolution of conventional microscopy. In the case of DNA-binding proteins, super-resolution microscopy can visualize the detailed spatial organization of DNA replication, transcription, and repair processes by reconstructing a map of single-molecule localizations. Furthermore, DNA-binding activities can be observed directly by tracking protein movement in real time. This allows identifying subpopulations of DNA-bound and diffusing proteins, and can be used to measure DNA-binding times in vivo. This chapter provides a detailed protocol for super-resolution microscopy and tracking of DNA-binding proteins in Escherichia coli cells. The protocol covers the genetic engineering and fluorescent labeling of strains and describes data acquisition and analysis procedures, such as super-resolution image reconstruction, mapping single-molecule tracks, computing diffusion coefficients to identify molecular subpopulations with different mobility, and analysis of DNA-binding kinetics. While the focus is on the study of bacterial chromosome biology, these approaches are generally applicable to other molecular processes and cell types. Key words Super-resolution fluorescence microscopy, Single-molecule imaging, Single-particle tracking, DNA-binding proteins, DNA repair, Lambda Red recombination, Escherichia coli

1

Introduction Super-resolution fluorescence microscopy [1] has matured as a technique and is now widely applied in fundamental research. Photoactivated localization microscopy (PALM) [2] and stochastic optical reconstruction microscopy (STORM) [3] reach image resolution below the diffraction limit of light by localizing individual spatially isolated fluorophores. For use in living cells, these fluorophores can be genetically encoded, as photoactivatable fluorescent proteins that are switched from a nonfluorescent state to a fluorescent state such that only a sparse subset of fluorophores is visible at any time. Alternatively, genetically encoded protein tags, such as

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_15, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

191

192

Chloe´ J. Cassaro and Stephan Uphoff

the HaloTag, can be covalently labeled with synthetic cellpermeable dyes [4]. For example, the commonly used fluorophore tetramethylrhodamine (TMR) switches spontaneously and reversibly between bright and dark states inside cells and has served as a basis from which brighter variants and photoactivatable synthetic fluorophores have been developed [5]. Automated image analysis detects fluorescent spots and determines their centroid positions. A super-resolution image can then be reconstructed from the list of molecule localizations that have been recorded sequentially over a series of images. The study of microorganisms particularly benefits from the ~10-fold increase in image resolution, allowing the visual examination of subcellular molecular structures, such as cell wall components, cell division machinery, and chromosomes [6–21]. Here, we focus on the application of PALM and single-molecule tracking to study DNA-binding proteins in Escherichia coli. Conventional fluorescence microscopy obscures the measurement of most of these proteins because they bind DNA transiently and are distributed throughout the bacterial nucleoid. By imaging single molecules, it is possible to observe unsynchronized reaction events, transient states, and small molecular subpopulations without population averaging. Beyond recording static structures, the ability to determine precise localizations of single fluorescent molecules has enabled tracking proteins in live cells [22]. With the super-resolution microscopy concept, this approach can now be applied for arbitrary densities of labeled molecules [23]: each fluorophore activation gives a glimpse into the function of a single protein. Reaction events, such as protein degradation or the binding of a DNA repair enzyme to a DNA damage site, are marked by a change in the diffusion characteristics [11, 24]. The apparent diffusion coefficient can be computed from a protein trajectory and provides information about the frequency of DNA-binding events and the fraction of time spent bound to DNA or in a transition state during the search for DNA target sites [25–28]. Progress has also been made in the application of live-cell super-resolution microscopy to study DNA-binding proteins in eukaryotic cells [29–32]. This chapter provides a detailed protocol covering the sample preparation and labeling for PALM imaging and data analysis procedures. Two alternative labeling strategies are presented, with fluorescent protein fusion, or HaloTag fusion. A detailed comparison of these different labeling strategies can be found here [33]. Similar general principles also apply to other super-resolution microscopy modalities such as STORM. First, the protocol shows the construction of E. coli strains carrying an endogenous photoactivatable fluorescent protein or HaloTag fusion using lambda Red recombination [34]. As opposed to exogenous plasmid expression systems, this approach maintains native expression levels and

Super-Resolution Microscopy of DNA-Binding Proteins

193

ensures complete replacement of the native gene with the fluorescent version. The following steps in the protocol include the preparation of cell cultures for microscopy, labeling for the HaloTag fusion protein, PALM data acquisition, and data processing to obtain single-molecule localizations and tracks. Once localizations and tracks have been recorded, there are many options for further analysis. Here, the most common and general approaches are presented, such as reconstruction of super-resolution images, mapping single-molecule tracks, computing diffusion coefficients to identify molecular subpopulations with different mobility, and analysis of DNA-binding kinetics. The protocol is illustrated using data of DNA polymerase I (Pol1), a DNA-binding protein with key functions in DNA replication and DNA repair in E. coli. Photoactivated single-molecule tracking has been applied to directly visualize binding events of single Pol1 enzymes at DNA repair sites following DNA alkylation damage [11], and to characterize its mobility during the search for target sites in unperturbed cells [25].

2

Materials

2.1 Lambda Red Recombination

1. E. coli AB1157 background strain (other E. coli K12 strains such as MC4100 or MG1655 can also be used). 2. Plasmid pKD46 [34], encoding the lambda Red integration proteins under an arabinose-inducible promoter. pKD46 carries an ampicillin resistance marker and the temperaturesensitive origin of replication repA101ts (to be grown at 30
C). 3. Plasmid encoding the photoactivatable fluorescent protein PAmCherry [35] or HaloTag [4] with an N-terminal flexible linker and an antibiotic resistance marker (e.g., see [11, 36] or [24, 33]). 4. PCR primers to amplify the insertion fragment from the template plasmid. The protocol provides guidance on primer design. 5. PCR kit with a high-fidelity polymerase. 6. Dpn1 enzyme. 7. LB medium and LB agar plates. 8. Antibiotics as required for pKD46 plasmid and selection of the lambda Red insertion (e.g., ampicillin, kanamycin). 9. 10% Arabinose solution: Freshly dissolved in dH2O and sterilized using a 0.2 μm filter. 10. PCR primers to verify the insertion.

194

Chloe´ J. Cassaro and Stephan Uphoff

2.2 Cell Culture, Labeling, and Slide Preparation

1. LB medium and LB agar plates. 2. Supplemented M9 medium: 100 mL 5 M9 salts, 1 mL 1 M MgSO4, 500 μL 100 mM CaCl2, 10 mL 50 MEM amino acids, 5 mL 100 μg/mL L-proline, 50 μL 0.5% thiamine, 5 mL 20% glucose, dH2O up to 500 mL. Sterilize with 0.2 μm filter. 3. Microscope coverslips (no. 1.5 thickness) burnt in a furnace at 500
C for 1 h or in a plasma etcher to remove fluorescent background contamination. 4. Low-fluorescence molecular biology-grade agarose (e.g., BioRad). 5. Promega HaloTag Ligand TMR.

2.3

Microscope

Detailed descriptions of how to design and assemble a custom total internal reflection fluorescence (TIRF) microscope for singlemolecule imaging can be found in [2, 15, 37, 38]. Suitable commercial instruments are also available from different manufacturers. Typical components that are ideal for single-molecule imaging are the following: 1. 100 NA 1.4 oil-immersion objective. 2. Electron-multiplying CCD camera. 3. 405 nm laser with at least 20 mW output power—for PAmCherry. 4. 561 nm laser with at least 50 mW output power. 5. TIRF excitation module. 6. Transmitted light illumination.

2.4

Data Analysis

Automated data processing can be performed in MATLAB (MathWorks). Optional toolboxes with useful functions include the following: 1. Image Processing Toolbox, containing functions to read and write image files, as well as filtering, registration, and segmentation tools. 2. Statistics Toolbox, containing tools for plotting data histograms and performing statistical tests. 3. Optimization Toolbox, containing curve fitting functions (e.g., lsqcurvefit).

3

Methods In the following protocols, two alternative labeling strategies using either a photoactivatable fluorescent protein or the HaloTag are described. Most steps are similar for both strategies; any differences are indicated.

Super-Resolution Microscopy of DNA-Binding Proteins

3.1 Lambda Red Integration to Generate an Endogenous Fluorescent Fusion

195

This protocol follows the method developed by Datsenko and Wanner [34] that employs the phage lambda Red recombinase to generate an endogenous C-terminal PAmCherry fusion or HaloTag fusion with a protein of interest in E. coli. The gene encoding PAmCherry with a flexible N-terminal linker or HaloTag is PCR amplified from a template plasmid together with an antibiotic resistance gene that allows for selection of the chromosomal integration. 1. Transform the target E. coli strain with plasmid pKD46. This plasmid encodes the lambda Red integration factors and an ampicillin resistance marker. The transformed strain needs to be grown with ampicillin at 30
C to maintain the temperaturesensitive plasmid. 2. Find the gene sequence of the protein you wish to tag using the online E. coli genome browser: http://microbes.ucsc.edu/cgibin/hgGateway?db¼eschColi_K12. 3. Design PCR primers flanking the PAmCherry or HaloTag and antibiotic resistance genes on the template plasmid and add overhangs with homology to the chromosomal insertion site. For the forward PCR primer, use 40–50 nt of homology at the 30 terminal end of the gene just upstream of the stop codon and add a primer sequence at the 50 terminus of the fusion linker on the template plasmid. For the reverse PCR primer, use 40–50 nt of homology immediately downstream of the stop codon of the gene and add a primer sequence downstream of the antibiotic resistance gene on the template plasmid. 4. Run a 50 μL PCR to amplify the lambda Red insertion fragment from the template plasmid using a high-fidelity DNA polymerase. 5. Add 1 μL of Dpn1 enzyme to the PCR product and incubate for at least 2 h at 37
C. This digests the template plasmid. 6. Run the PCR product on a 1% agarose gel (e.g., 100 V for 1 h). Cut out the corresponding DNA band and extract the DNA (e.g. using the Qiagen gel extraction kit). Elute the DNA in a small volume of dH2O (e.g., 30 μL) to obtain a concentrated solution (typically 30–200 ng/μl). 7. Grow a 5 mL culture of the target strain carrying plasmid pKD46 in LB medium with 50 μg/mL ampicillin at 30
C overnight. 8. Dilute 500 μL of the culture in 50 mL LB medium containing 50 μg/mL ampicillin and 0.2% arabinose. Grow the culture at 30
C until it reaches an OD600 of 0.6. 9. Place the culture on ice for 10 min and swirl to chill it abruptly.

196

Chloe´ J. Cassaro and Stephan Uphoff

10. Spin down the culture in a chilled 50 mL tube for 10 min in a centrifuge at 2900 x g, cooled to 4
C. 11. Repeat several cycles of centrifugation and resuspension of the culture in decreasing volumes of ice-cold dH2O: 50 mL, 35 mL, 2 mL, and 500 μL. Keep the culture on ice between steps and centrifuge at 4
C. 12. Add 1–6 μL of DNA (typically 30–200 ng/μL concentration) to 60 μL of cell suspension and incubate on ice for 15 min. 13. Electroporate the cells with DNA in a pre-chilled cuvette. Following electroporation, immediately add 1 mL of LB medium (at room temperature) and mix gently. 14. Recover the cells at 37
C for 1 h and plate on LB agar containing the appropriate antibiotic to select for insertions. Incubate plates at 37
C overnight to promote the loss of the temperature-sensitive plasmid pKD46. 15. Plates typically show up to several tens of cell colonies. Pick several colonies and streak again to isolate single colonies. Replica plate on LB agar with 50 μg/mL ampicillin to confirm the loss of plasmid pKD46. 16. Verify correct insertion of the fusion construct by colony PCR and sequencing using primers flanking the insertion site. 17. Use P1 phage transduction [39] to move the fusion allele to a strain which has not been transformed with plasmid pKD46. 18. Functionality of the fusion protein should be assessed by comparing the growth rate of the generated strain to the wild-type strain and using other appropriate assays (e.g., sensitivity to DNA damage for a strain carrying a fusion of a DNA repair protein). See Note 1 for more information. 3.2

Cell Culture

The following protocol is for the culture of E. coli AB1157 strains exhibiting wild-type phenotypes. Mutant strains may require different growth conditions. 1. Two days before a microscopy session: Streak cells from a frozen glycerol stock onto a LB agar plate containing the appropriate antibiotic to select for the strain expressing the fusion protein of interest. 2. The next day, inoculate a LB culture from a single cell colony and grow for 3–5 h at 37
C. 3. Add 5 μL of LB culture to 4 mL of supplemented M9 medium and grow overnight at 37
C. 4. The next morning, dilute 80 μL of the culture in 4 mL of supplemented M9 medium and grow for 2 h at 37
C so that cells are imaged during early exponential growth phase (~OD 0.05–0.1).

Super-Resolution Microscopy of DNA-Binding Proteins

197

5. Pellet cells in a benchtop centrifuge at 3300  g for 3 min. For the PAmCherry fusion, proceed to step 6 and for the HaloTag fusion (detailed protocol here: [33]), follow steps 7–13. 6. Thoroughly resuspend the cell pellet in 10 μL of supernatant medium. For the PAmCherry fusion, go to step 13 directly. 7. For the HaloTag fusion, resuspend the cell pellet in 100 μL of M9 medium and add 2.5 μM of TMR dye. Mix thoroughly and incubate at 25
C for 30 min. 8. Pellet the cells and resuspend in 1 mL of M9 medium to wash any free dye. 9. Move the cells to a fresh tube, spin down, and resuspend. Repeat this 3 times to remove free dye. 10. Add 1 mL of M9 medium to resuspend, and let the cells recover for 30 min at 37
C. 11. Spin down 1 mL of cells, resuspend in 1 mL of M9 medium, and spin down again. 12. Thoroughly resuspend the cell pellet in 10 μL of supernatant medium. 13. Microscopy should be performed immediately after preparation of the concentrated cell suspension. 3.3

Microscopy

1. Prepare 5 mL of a 1% low-fluorescence agarose solution in supplemented M9 medium. 2. Spread 1 mL of melted agarose solution on a microscope coverslip and place a burnt coverslip onto the solidifying gel to create a flat pad. 3. Remove the burnt coverslip from the gel pad and drop 1 μL of concentrated cell suspension onto it. Place a new burnt coverslip on top to sandwich the cells. 4. Place the sample on the microscope and bring cells into focus using transmitted light illumination. For the PAmCherry fusion, follow steps 5–8 and for the HaloTag fusion, follow steps 9–13. 5. For the PAmCherry fusion, activate the electron-multiplying gain of the EMCCD camera and display the live camera data. Only background noise should be visible. Typical frame acquisition rates for single-molecule imaging in live cells are between 10 and 100 frames/s. 6. Switch on the 561 nm laser for excitation of PAmCherry and the 405 nm laser for photoactivation. Wait for individual fluorescent spots to appear inside cells and adjust the TIRF illumination angle to maximize the brightness of the spots.

198

Chloe´ J. Cassaro and Stephan Uphoff

7. Record a short movie, examine the saved images, and optimize the acquisition settings (frame rates and laser intensities) as explained in Note 2. 8. For data acquisition, first expose cells to 561 nm excitation for a few seconds to bleach the autofluorescence background, and then start recording a movie. Switch on the 405 nm laser for PAmCherry photoactivation. For tracking experiments, activated molecules should be visible for a few frames until photobleaching. Gradually increase the 405 nm intensity over the course of the movie while the pool of pre-activated molecules depletes. Keep the density of activated molecules sparse at less than one spot per cell on average in each frame. Record a movie of several thousand frames until most molecules have been activated (Fig. 1). For the PAmCherry fusion, go to step 13 at the end of this step. 9. For the HaloTag fusion, switch on the 561 nm laser for excitation of the HaloTag-TMR. Activate the electron-multiplying gain of the EMCCD camera and display the live camera data. 10. A pre-bleaching step is required for a few seconds before imaging, depending on the density of fluorescent spots in cells. Wait for individual fluorescent spots to be sufficiently sparse and adjust the TIRF illumination angle to maximize the brightness of the spots.

Fig. 1 Example images of a PALM recording. The transmitted light image shows live E. coli cells immobilized on an agarose pad. Cells are expressing a fusion of DNA polymerase 1 (Pol1) with PAmCherry. Example frames of a PALM movie show isolated fluorescent spots of single Pol1-PAmCherry molecules. Different molecules become photoactivated in different frames such that their precise positions can be recorded over time. Scale bars: 1 μm

Super-Resolution Microscopy of DNA-Binding Proteins

199

11. Record a short movie, examine the saved images, and optimize the acquisition settings (frame rate and laser intensities) as explained in Note 2. 12. For data acquisition, first expose cells to 561 nm excitation for a few seconds to bleach the autofluorescence background and the HaloTag-TMR dyes, and then start recording a movie. Record a movie of several thousand frames until most molecules have bleached or a desired amount of data has been collected (Fig. 1). 13. Several movies of different cells can be recorded for approximately 45 min before the agarose gel starts to dry out. 3.4 Reconstructing a Super-Resolution Map of Localizations

1. Use a localization algorithm to identify fluorescent spots and determine their centroid with high precision. Several algorithms have been developed for this purpose, e.g., [40– 43]. See Note 3 for guidance. 2. Use a segmentation algorithm such as MicrobeTracker or SuperSegger in MATLAB [44, 45] to determine the outline of the cells, so that only fluorescent spots within cells are included in the analysis. 3. Display the resulting localizations as a scatterplot (Fig. 2a, b). 4. Generate a super-resolution image by binning localizations into a two-dimensional color-coded histogram (Fig. 2c). Alternatively, each localization can be rendered as a Gaussian spot with a width corresponding to the estimated localization error. The super-resolution image is then reconstructed as the superposition of all Gaussian spots (Fig. 2d). 5. The super-resolution image can be used to analyze structures of protein assemblies in cells [9, 14, 15, 32], presence of protein foci [10, 12], or partitioning of proteins between different cellular compartments [17, 20, 27, 46].

3.5 Short Exposure Times to Quantify Diffusion

In live cells, protein movement can be measured by recording a PALM movie at high frame rates and short exposure times (e.g., 15 ms/frame) and using an automated tracking algorithm. 1. Determine molecule tracks by linking localizations that appear nearby in consecutive frames. Simple algorithms use a fixed tracking window within which localizations get connected [47], while other methods apply global tracking analysis [48]. See Note 4 for more information on the following analysis procedure. 2. Compute the mean-squared displacement (MSD) from the (x, y) localizations for each track with at least N ¼ 5 localizations: MSD ¼ 1/(N – 1) ∑i¼1N–1 (xi+1 – xi)2 + (yi+1 – yi)2.

200

Chloe´ J. Cassaro and Stephan Uphoff

Fig. 2 Super-resolution image reconstruction. (a) Localizations of Pol1-PAmCherry from a PALM recording of 7.500 frames (example frames in Fig. 1) are displayed as a scatterplot. (b) Localizations mapped onto the

Super-Resolution Microscopy of DNA-Binding Proteins

201

3. Compute the apparent diffusion coefficient (D*) per track using the mean-squared displacement: D* ¼ MSD/(4 Δt). Here, Δt is the time interval between frames. 4. Plot a histogram to examine the distribution of diffusion coefficients. Molecule subpopulations with different mobility (e.g., mobile and bound molecules) can be identified as distinct species in the histogram (Fig. 3). 5. In the case of a bound and a mobile population of molecules, the fraction of molecules in the bound state can be quantified using a threshold on the apparent diffusion coefficient (Fig. 3b, c). 6. An alternative approach to quantify subpopulations with different mobility is to fit the distributions using an analytical equation [7, 13, 17], and reference molecules belonging to each potential subpopulation can be used to resolve the different diffusive states possible for the protein in the cell (see Note 5). 3.6 Long Exposures to Quantify Binding Kinetics

The time a single photoactivated molecule such as PAmCherry can be observed for is limited by irreversible photobleaching. This duration is typically only a few frames, which complicates quantification of binding kinetics from tracking data recorded at high temporal resolution. HaloTag with TMR dyes presents the advantage of reduced photobleaching, which allows imaging for a longer period [33]. A simple solution is to reduce laser intensities—which extends the photobleaching lifetime—while equally increasing exposure times to circumvent the reduced emission intensities. At long exposure times, mobile molecules will appear as blurred spots, while bound molecules result in diffraction-limited spots. A spotfinding algorithm can be used to extract diffraction-limited spots only, and the average binding time corresponds to the average number of frames a diffraction-limited spot is visible for. It is important to correct for the residual effect of photobleaching on the apparent binding time by measuring photobleaching rates under identical illumination conditions using fixed cells or a sample for which binding times are very long compared to the average bleaching time. Crucially, photobleaching should not be faster than the average binding time because small errors in the estimation of the photobleaching rate will cause very large errors in the estimation of the binding time constant. More information on this type of analysis can be found in reference [11].

ä Fig. 2 (continued) transmitted light image of cells. (c) Histogram visualization: Localizations were binned into a two-dimensional grid of subpixels (38 nm x 38 nm, i.e., 4 x 4 subpixels per original pixel). The number of localizations per bin is represented according to the colors in the scale bar. (d) Gaussian kernel visualization: The image is reconstructed by summing normalized Gaussian kernels with 40 nm standard deviation (equivalent to the localization precision) centered on the localizations. The boxed regions are shown magnified. Scale bars: 1 μm

202

Chloe´ J. Cassaro and Stephan Uphoff

Fig. 3 Photoactivated single-molecule tracking analysis. Tracks of Pol1-PAmCherry molecules were generated from the localization data in Fig. 2 and shown on a transmitted light image of cells. The histograms show the distribution of apparent diffusion coefficients. Live cells were treated with the DNA-damaging agent methyl methanesulfonate (100 mM MMS) for 20 min before imaging. Base-excision repair of MMS damage creates gapped DNA substrates to which Pol1 binds for DNA repair synthesis. Single-molecule tracking of Pol1 allows identifying such events by the low apparent diffusion coefficient. (a) Data for all tracks with at least 5 localizations. (b) Detecting individual bound molecules by plotting only tracks with an apparent diffusion coefficient below 0.15 μm2/s. (c) All observed tracks, color coded by their classification as bound or mobile, in red and blue, respectively. Scale bars: 1 μm

Super-Resolution Microscopy of DNA-Binding Proteins

3.7 Clustering Analysis

4

203

Clustering algorithms identify foci of localizations that might represent a spatial organization of proteins performing their function in cells. Methods to quantify clustering include k-means, nearestneighbor clustering, and point-correlation analysis. These algorithms report the size and number of localizations per cluster. K-means assigns all observed localizations to a fixed number of clusters. This number has to be defined prior to the analysis. Nearest-neighbor clustering groups localizations that are within a user-defined distance threshold [10]. Point-correlation analysis examines the spatial correlation between localizations, and compares the observed distribution to a random homogenous distribution of points [49, 50].

Notes 1. Protein functionality can be affected by the fusion to a fluorescent protein. Therefore, care should be taken to closely compare growth characteristics and specific phenotypes of the strain carrying the fluorescent fusion protein to the wild-type strain. For nonessential proteins, functionality of the fusion can also be evaluated by comparison with a strain in which the gene is deleted [51]. Additional control experiments can be performed that abolish specific activities of the protein (e.g., by introducing point mutations, drug treatments, or deletion of protein interaction partners). In addition, control experiments are required to ensure that the labels used (fluorescent proteins or protein tags) do not perturb the functionality, normal diffusive behaviors, and states of the proteins studied. For example, tracking labels which are not fused to a protein can be used to confirm their lack of interaction with other particles within the cell [28]. Furthermore, many fluorescent proteins are prone to dimerize or multimerize which can cause aggregation artifacts of fusion proteins [52, 53]. Localization clusters may therefore represent protein aggregates instead of a genuine spatial organization of the native protein. Moreover, photoactivatable fluorescent proteins typically blink; that is, they randomly convert between bright and dark states once they have been photoactivated by 405 nm light [54, 55]. Each fusion protein may therefore be observed several times, which can result in apparent clustering of tracks and complicates counting molecule numbers. Conventional fluorescence imaging with a monomeric fluorescent protein fusion can be performed to evaluate the genuine presence of clusters [52]. 2. Before data acquisition it is important to identify appropriate imaging conditions. To this end, record a short movie, examine the images, and change the microscope settings as required.

204

Chloe´ J. Cassaro and Stephan Uphoff

For example: (A) Increase the 561 nm laser power or use a longer camera exposure time if the fluorescent spots are too dim against the background noise. (B) Decrease the 561 nm laser power or use faster camera exposures if the spots are very clear but appear to bleach very quickly. Estimation of diffusion coefficients requires observing the same molecule for multiple frames. (C) Shorten the camera exposure time if the fluorescent spots appear blurry due to fast movement of the fusion protein during each frame. Appropriate laser intensities and exposure times will vary between instruments due to differences in the alignment and quality of optical components. 3. Microscopy images are subject to several factors that complicate the precise localization of single fluorophores. These factors include image pixelation, camera read noise, electron-multiplying noise, background noise, and photon shot noise. Background noise can be reduced by TIRF excitation, while photon shot noise is influenced by the fluorophore brightness. Once all experimental factors have been optimized, the localization algorithm is key to maximize the localization precision [56]. Most localization algorithms use either leastsquares [40, 57] or maximum likelihood parameter estimation [41, 42]. A detailed assessment of localization software can be found here: [43]. 4. In addition to minimizing localization error, a key metric for the performance of localization algorithms is the recall, i.e., the fraction of successfully detected spots. The recall fraction decreases with increasing density of fluorescent spots per image. However, high densities are often desired because the time required to record a PALM movie that represents the majority of labeled molecules in the sample is limited by the number of accurate localizations per frame. Moreover, the total density of all localizations dictates the resolution of the reconstructed image according to Shannon’s sampling theorem. Several algorithms have been developed that substantially improve the localization recall in images of densely overlapping fluorescent spots [58, 59]. Because of photobleaching, the number of localizations per track is usually small, which causes large statistical uncertainty in the estimation of the diffusion coefficient of a single molecule. For this reason, only tracks with a certain minimum number of steps should be included in the analysis (e.g., at least 5 localizations or 4 steps). In addition, it is useful to apply a tracking algorithm that uses a memory parameter to link molecule tracks when a localization is missed due to blinking or low signal-to-noise ratio [25]. The term apparent diffusion coefficient is used because this measurement of the diffusion coefficient includes several biases: confinement of molecules within the cell volume or cellular compartments leads to an underestimation of the diffusion coefficient.

Super-Resolution Microscopy of DNA-Binding Proteins

205

Furthermore, molecular movement during the exposure time reduces the apparent diffusion coefficient because the localizations represent time-averaged positions. The localization error, on the other hand, effectively adds a random step to each true position, which causes an overestimation of the diffusion coefficient. This error can be corrected by measuring the average localization error σloc using a sample with immobile molecules and applying the equation D* ¼ MSD/(4 Δt) – σ loc2/Δt. Frameworks have been developed to extract kinetic parameters for transitions between different diffusion states while accounting for confinement and tracking window boundaries [26]. Another approach uses time-averaged diffusion analysis, also taking into account these experimental biases to estimate diffusion states, interconversion diffusion coefficients, and population fractions [28]. Finally, single-molecule tracking analysis procedures can be verified using simulations of Brownian diffusion within cells [11, 26, 28]. 5. Molecular diffusion is normally governed by the size of a particle and viscosity of the medium. However, the mobility of DNA-binding proteins in bacteria is largely independent of their size, but instead determined by the duration and frequency of transient interactions with the chromosomal DNA [25]. During their search for target sites, DNA-binding proteins spend the majority of time bound nonspecifically to DNA. This should be taken into account when interpreting the observed mobilities. Finally, to derive biological meaning from the diffusion distributions of molecules within a cell, biological controls can be used. Control experiments performed using mutants of the protein of interest in a single diffusive state (DNA binding, freely diffusing, etc.) can help define the different subpopulations present within a cell. Alternatively, complete removal of chromosomal DNA can also be used as a control for its influence on protein mobility [25].

Acknowledgments Rodrigo Reyes-Lamothe, David J. Sherratt, and Achillefs N. Kapanidis helped with the original development of this protocol. Katarzyna Ginda and David J. Sherratt are thanked for their comments on the manuscript. Research in the Uphoff lab is funded by a Wellcome Trust and Royal Society Sir Henry Dale Fellowship (206159/Z/17/Z), a Wellcome-Beit Prize (206159/Z/17/B), a Lister Institute Research Fellowship Prize, and a tutorial fellowship at New College Oxford. C.J.C. is supported by funding from the Biotechnology and Biological Sciences Research Council (BBSRC) [grant number BB/M011224/1].

206

Chloe´ J. Cassaro and Stephan Uphoff

References 1. Hell SW (2007) Far-field optical nanoscopy. Science 316:1153–1158 2. Betzig E, Patterson GH, Sougrat R et al (2006) Imaging intracellular fluorescent proteins at nanometer resolution. Science (New York, NY) 313:1642–1645 3. Rust MJ, Bates M, Zhuang X Sub-diffractionlimit imaging by stochastic optical reconstruction microscopy (STORM). Nat Methods 3: 793–795 4. Los GV, Encell LP, McDougall MG, Hartzell DD, Karassina N, Zimprich C, Wood MG, Learish R, Ohana RF, Urh M, Simpson D, Mendez J, Zimmerman K, Otto P, Vidugiris G, Zhu J, Darzins A, Klaubert DH, Bulleit RF, Wood KV (2008) HaloTag: a novel protein labeling technology for cell imaging and protein analysis. ACS Chem Biol 3(6): 3 7 3 – 3 8 2 . h t t p s : // d o i . o r g / 1 0 . 1 0 2 1 / cb800025k 5. Grimm JB, English BP, Choi H, Muthusamy AK, Mehl BP, Dong P, Brown TA, LippincottSchwartz J, Liu Z, Lionnet T, Lavis LD (2016) Bright photoactivatable fluorophores for single-molecule imaging. Nat Methods 13(12):985–988. https://doi.org/10.1038/ nmeth.4034 6. Biteen JS, Thompson MA, Tselentis NK et al (2008) Super-resolution imaging in live Caulobacter crescentus cells using photoswitchable EYFP. Nat Methods 5:947–949 7. English BP, Hauryliuk V, Sanamrad A et al (2011) Single-molecule investigations of the stringent response machinery in living bacterial cells. Proc Natl Acad Sci U S A 108: E365–E373 8. Garner EC, Bernard R, Wang W et al (2011) Coupled, circumferential motions of the cell wall synthesis machinery and MreB filaments in B. subtilis. Science 333:222–225 9. Greenfield D, McEvoy AL, Shroff H et al (2009) Self-organization of the Escherichia coli chemotaxis network imaged with superresolution light microscopy. PLoS Biol 7: e1000137 10. Badrinarayanan A, Reyes-Lamothe R, Uphoff S et al (2012) In vivo architecture and action of bacterial structural maintenance of chromosome proteins. Science 338:528–531 11. Uphoff S, Reyes-Lamothe R, de Leon FG et al (2013) Single-molecule DNA repair in live bacteria. Proc Natl Acad Sci U S A 110: 8063–8068 12. Endesfelder U, Finan K, Holden SJ, Endesfelder U, Finan K, Holden SJ et al

(2013) Multiscale spatial organization of RNA polymerase in Escherichia coli. Biophys J 105:172–181 13. Bakshi S, Dalrymple RM, Li W et al (2013) Partitioning of RNA polymerase activity in live Escherichia coli from analysis of singlemolecule diffusive trajectories. Biophys J 105: 2676–2686 14. Fiche J-B, Cattoni DI, Diekmann N et al (2013) Recruitment, assembly, and molecular architecture of the SpoIIIE DNA pump revealed by super-resolution microscopy. PLoS Biol 11:e1001557 15. Holden SJ, Pengo T, Meibom KL et al (2014) High throughput 3D super-resolution microscopy reveals Caulobacter crescentus in vivo Z-ring organization. Proc Natl Acad Sci U S A 111:4566–4571 16. Diepold A, Kudryashev M, Delalez NJ et al (2015) Composition, formation, and regulation of the cytosolic c-ring, a dynamic component of the type III secretion injectisome. PLoS Biol 13:e1002039 17. Stracy M, Lesterlin C, de Leon FG et al (2015) Live-cell super-resolution microscopy reveals the organization of RNA polymerase in the bacterial nucleoid. Proc Natl Acad Sci U S A 112(32):E4390–E4399. 201507592 18. Zhang Z, English BP, Grimm JB, Kazane SA, Hu W, Tsai A, Inouye C, You C, Piehler J, Schultz PG, Lavis LD, Revyakin A, Tjian R (2016) Rapid dynamics of general transcription factor TFIIB binding during preinitiation complex assembly revealed by single-molecule analysis. Genes Dev 30(18):2106–2118. https:// doi.org/10.1101/gad.285395.116 19. Li Y, Schroeder JW, Simmons LA, Biteen JS (2018) Visualizing bacterial DNA replication and repair with molecular resolution. Curr Opin Microbiol 43:38–45. https://doi.org/ 10.1016/j.mib.2017.11.009 20. Bayas CA, Wang J, Lee MK, Schrader JM, Shapiro L, Moerner WE (2018) Spatial organization and dynamics of RNase E and ribosomes in Caulobacter crescentus. Proc Natl Acad Sci U S A 115(16):E3712–E3721. https://doi. org/10.1073/pnas.1721648115 21. Stracy M, Jaciuk M, Uphoff S, Kapanidis AN, Nowotny M, Sherratt DJ, Zawadzki P (2016) Single-molecule imaging of UvrA and UvrB recruitment to DNA lesions in living Escherichia coli. Nat Commun 7:12568. https://doi. org/10.1038/ncomms12568

Super-Resolution Microscopy of DNA-Binding Proteins 22. Saxton MJ, Jacobson K (1997) Single-particle tracking: applications to membrane dynamics. Annu Rev Biophys Biomol Struct 26:373–399 23. Manley S, Gillette JM, Patterson GH et al (2008) High-density mapping of singlemolecule trajectories with photoactivated localization microscopy. Nat Methods 5: 155–157 24. Jones EC, Uphoff S (2021) Single-molecule imaging of LexA degradation in Escherichia coli elucidates regulatory mechanisms and heterogeneity of the SOS response. Nat Microbiol. 6:981–990. https://doi.org/10.1038/ s41564-021-00930-y 25. Stracy M, Schweizer J, Sherratt DJ, Kapanidis AN, Uphoff S, Lesterlin C (2021) Transient non-specific DNA binding dominates the target search of bacterial DNA-binding proteins. Mol Cell. https://doi.org/10.1016/j.molcel. 2021.01.039 26. Vink JNA, Brouns SJJ, Hohlbein J (2020) Extracting transition rates in particle tracking using analytical diffusion distribution analysis. Biophys J 119(10):1970–1983. https://doi. org/10.1016/j.bpj.2020.09.033 27. Vink JNA, Martens KJA, Vlot M, McKenzie RE, Almendros C, Estrada Bonilla B, Brocken DJW, Hohlbein J, Brouns SJJ (2020) Direct visualization of native CRISPR target search in live bacteria reveals cascade DNA surveillance mechanism. Mol Cell 77(1):39–50. e10. https://doi.org/10.1016/j.molcel.2019. 10.021 28. Rocha J, Corbitt J, Yan T, Richardson C, Gahlmann A (2019) Resolving cytosolic diffusive states in bacteria by single-molecule tracking. Biophys J 116(10):1970–1983. https://doi. org/10.1016/j.bpj.2019.03.039 29. Gebhardt JCM, Suter DM, Roy R et al (2013) Single-molecule imaging of transcription factor binding to DNA in live mammalian cells. Nat Methods 10:421–426 30. Etheridge TJ, Boulineau RL, Herbert A et al (2014) Quantification of DNA-associated proteins inside eukaryotic cells using singlemolecule localization microscopy. Nucleic Acids Res 42:e146–e146 31. Mehta GD, Ball DA, Eriksson PR, Chereji RV, Clark DJ, McNally JG, Karpova TS (2018) Single-molecule analysis reveals linked cycles of RSC chromatin remodeling and Ace1p transcription factor binding in yeast. Mol Cell 72(5):875–887. e879. https://doi.org/10. 1016/j.molcel.2018.09.009 32. Gormal RS, Padmanabhan P, Kasula R, Bademosi AT, Coakley S, Giacomotto J, Blum A, Joensuu M, Wallis TP, Lo HP, Budnar S, Rae J,

207

Ferguson C, Bastiani M, Thomas WG, Pardon E, Steyaert J, Yap AS, Goodhill GJ, Hilliard MA, Parton RG, Meunier FA (2020) Modular transient nanoclustering of activated beta2-adrenergic receptors revealed by singlemolecule tracking of conformation-specific nanobodies. Proc Natl Acad Sci U S A 117(48):30476–30487. https://doi.org/10. 1073/pnas.2007443117 33. Banaz N, Makela J, Uphoff S (2019) Choosing the right label for single-molecule tracking in live bacteria: side-by-side comparison of photoactivatable fluorescent protein and Halo tag dyes. J Phys D: Appl Phys 52(6):064002. https://doi.org/10.1088/1361-6463/ aaf255 34. Datsenko KA, Wanner BL (2000) One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci U S A 97:6640–6645 35. Subach FV, Patterson GH, Manley S et al (2009) Photoactivatable mCherry for highresolution two-color fluorescence microscopy. Nature Meth 6:153–159 36. Reyes-Lamothe R, Possoz C, Danilova O et al (2008) Independent positioning and action of Escherichia coli replisomes in live cells. Cell 133:90–102 37. Roy R, Hohng S, Ha T (2008) A practical guide to single-molecule FRET. Nat Methods 5:507–516 38. Uphoff S, Sherratt DJ, Kapanidis AN (2014) Visualizing protein-DNA interactions in live bacterial cells using photoactivated singlemolecule tracking. J Visual Exp: JoVE 39. Thomason LC, Costantino N, Court DL (2007) E. coli genome manipulation by P1 transduction. In: Current protocols in molecular biology. John Wiley & Sons, New York 40. Holden SJ, Uphoff S, Hohlbein J et al (2010) Defining the limits of single-molecule FRET resolution in TIRF microscopy. Biophys J 99: 3102–3111 41. Mortensen KI, Churchman LS, Spudich JA et al (2010) Optimized localization analysis for single-molecule tracking and superresolution microscopy. Nat Methods 7: 377–381 42. Smith CS, Joseph N, Rieger B et al (2010) Fast, single-molecule localization that achieves theoretically minimum uncertainty. Nat Methods 7:373–375 43. Sage D, Pham TA, Babcock H, Lukes T, Pengo T, Chao J, Velmurugan R, Herbert A, Agrawal A, Colabrese S, Wheeler A, Archetti A, Rieger B, Ober R, Hagen GM, Sibarita JB, Ries J, Henriques R, Unser M, Holden S

208

Chloe´ J. Cassaro and Stephan Uphoff

(2019) Super-resolution fight club: assessment of 2D and 3D single-molecule localization microscopy software. Nat Methods 16(5): 387–395. https://doi.org/10.1038/s41592019-0364-4 44. Sliusarenko O, Heinritz J, Emonet T, JacobsWagner C (2011) High-throughput, subpixel precision analysis of bacterial morphogenesis and intracellular spatio-temporal dynamics. Mol Microbiol 80(3):612–627. https://doi. org/10.1111/j.1365-2958.2011.07579.x 45. Stylianidou S, Brennan C, Nissen SB, Kuwada NJ, Wiggins PA (2016) SuperSegger: robust image segmentation, analysis and lineage tracking of bacterial cells. Mol Microbiol 102(4): 690–700. https://doi.org/10.1111/mmi. 13486 46. Bakshi S, Siryaporn A, Goulian M et al (2012) Super-resolution imaging of ribosomes and RNA polymerase in live Escherichia coli cells. Mol Microbiol 85:21–38 47. Crocker JC, Grier DG (1996) Methods of digital video microscopy for colloidal studies. J Colloid Interface Sci 179:298–310 48. Persson F, Linde´n M, Unoson C et al (2013) Extracting intracellular diffusive states and transition rates from single-molecule tracking data. Nat Methods 10:265–269 49. Veatch SL, Machta BB, Shelby SA et al (2012) Correlation functions quantify superresolution images and estimate apparent clustering due to over-counting. PLoS One 7: e31457 50. Sengupta P, Jovanovic-Talisman T, LippincottSchwartz J (2013) Quantifying spatial organization in point-localization super-resolution images using pair correlation analysis. Nat Protocols 8:345–354

51. Baba T, Ara T, Hasegawa M et al (2006) Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: the Keio collection. Mol Syst Biol 2. https://doi.org/10. 1038/msb4100050 52. Landgraf D, Okumus B, Chien P et al (2012) Segregation of molecules at cell division reveals native protein localization. Nat Methods 9(5): 480–482 53. Wang S, Moffitt JR, Dempsey GT et al (2014) Characterization and development of photoactivatable fluorescent proteins for single-molecule–based super-resolution imaging. Proc Natl Acad Sci U S A 111:8452–8457 54. Lee S-H, Shin JY, Lee A et al (2012) Counting single photoactivatable fluorescent molecules by photoactivated localization microscopy (PALM). Proc Natl Acad Sci U S A 109: 17436–17441 55. Durisic N, Laparra-Cuervo L, Sandoval´ lvarez A ´ et al (2014) Single-molecule evaluaA tion of fluorescent protein photoactivation efficiency using an in vivo nanotemplate. Nat Methods 11:156–162 56. Small A, Stahlheber S (2014) Fluorophore localization algorithms for super-resolution microscopy. Nat Methods 11:267–279 57. Thompson RE, Larson DR, Webb WW (2002) Precise Nanometer Localization analysis for individual fluorescent probes. Biophys J 82: 2775–2783 58. Holden SJ, Uphoff S, Kapanidis AN (2011) DAOSTORM: an algorithm for high- density super-resolution microscopy. Nat Methods 8: 279–280 59. Zhu L, Zhang W, Elnatan D et al (2012) Faster STORM using compressed sensing. Nat Methods 9:721–723

Chapter 16 Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells Using Single-Molecule Localization Microscopy Maike Steindel, Igor Orsine de Almeida, Stanley Strawbridge, Valentyna Chernova, David Holcman, Aleks Ponjavic, and Srinjan Basu Abstract Single-molecule localization microscopy (SMLM) allows the super-resolved imaging of proteins within mammalian nuclei at spatial resolutions comparable to that of a nucleosome itself (~20 nm). The technique is therefore well suited to the study of chromatin structure. Fixed-cell SMLM has already allowed temporal “snapshots” of how proteins are arranged on chromatin within mammalian nuclei. In this chapter, we focus on how recent developments, for example in selective plane illumination, 3D SMLM, and protein labeling, have led to a range of live-cell SMLM studies. We describe how to carry out single-particle tracking (SPT) of single proteins and, by analyzing their diffusion parameters, how to determine whether proteins interact with chromatin, diffuse freely, or do both. We can study the numbers of proteins that interact with chromatin and also determine their residence time on chromatin. We can determine whether these proteins form functional clusters within the nucleus as well as whether they form specific nuclear structures. Key words Chromatin, Super-resolution microscopy, PALM, STORM, SPT, SPIM, Fluorescence imaging, Mean squared displacement, Jump distance, Residence time, Diffusion coefficient, Anomalous exponent

1

Introduction to the Role of Live-Cell Single-Molecule Localization Microscopy Understanding protein/chromatin dynamics is critical to understanding how chromosome architecture is regulated. Advances in super-resolution imaging have led to the imaging of live cells well below the diffraction limit of light (~250 nm), allowing us to now dissect previously inaccessible dynamics that are critical to chromosome function. The suitable imaging approach generally depends on the spatial scale at which chromosome function is being studied (Fig. 1a). The mammalian nucleus is ~10–30 μm in size within which chromosomes occupy discrete regions of ~3 μm called “chromosome

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_16, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

209

210

Maike Steindel et al.

A

3D confocal + Airyscan 3D SIM STED Fixed cell SMLM Live cell SMLM

B

C 4 µm

SMLM reconstruction in 3D, Diffraction-limited microscopy

SMLM reconstruction in 2D

e.g. DHPSF

Fig. 1 (a) Length scales at which live-cell single-molecule localization microscopy (SMLM) is currently used to study chromatin structure. This is compared to other common microscopy approaches. (b) Schematic of single-molecule localization microscopy (SMLM) adapted from Horrocks et al. [4]. Conventional imaging is diffraction limited. By turning one molecule at a time to the “on” state, it is possible to precisely localize them at higher resolution (~10 nm) and reconstruct a super-resolution image. (c) Schematic comparison between the output of diffraction-limited, 2D SMLM, and 3D SMLM methods

territories” [1], although intermingling between adjacent chromosomes does occur [2]. Most of the DNA within chromosomes is packaged into nucleosomes of around 20 nm but these nucleosomes are not uniformly compacted into chromosomes, instead forming structures at various orders of magnitude. Recent advances

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . .

211

in chromatin electron microscopy tomography, super-resolution fluorescence microscopy, and proximity-based sequencing approaches (e.g., chromosome conformation capture approaches) are finally allowing us to probe 3D genome architecture at unprecedented resolution at these different scales [3–14]. These approaches confirm physical “megadomains” observed through traditional imaging approaches of active and inactive genes at the 5–20 Mb scale, thought to be less densely packed euchromatin and more condensed heterochromatin. However, they have also revealed smaller sub-compartments (topologically associating domains or TADs) ranging in size from 40 kb to 3 Mb, and even the formation of smaller compartments or sub-TADs. At the same time, high-resolution imaging approaches have revealed the formation of nuclear condensates at these scales that are thought to enrich for specific proteins [15]. Some of these nuclear condensates (including heterochromatin regions) form membraneless compartments through liquid-liquid phase separation [16–19]. How these compartments assemble/disassemble, how they influence chromatin dynamics, and how proteins are included/excluded from them are topics of considerable interest, with studies suggesting that proteins travel differently through heterochromatic vs. euchromatic regions [20]. Understanding the dynamics of proteins or condensates that act at these smaller levels of compaction has driven the development of several higher resolution imaging approaches. The main three live-cell imaging approaches developed for studying the dynamics of chromatin and chromatin-binding proteins at high resolution are the following: 1. 3D confocal microscopy with deconvolution (e.g., Airyscan). 2. Microscopy approaches that pattern the illumination light such as stimulated emission depletion (STED) [21], reversible saturable optical fluorescence transition (RESOLFT) microscopy [22, 23], structured illumination microscopy (SIM) [24], and MINFLUX [25, 26]. 3. Single-molecule localization microscopy (SMLM) approaches that detect and localize single molecules including stochastic optical reconstruction microscopy (STORM) [27], photoactivated localization microscopy (PALM) [28], and fluorescence photoactivation localization microscopy (fPALM) [29]. SMLM approaches separate molecules temporally such that their point-spread functions (PSFs) do not overlap spatially during the image acquisition (Fig. 1b, c). This is achieved either through under-labeling of proteins in the sample with fluorophores or by taking advantage of fluorophores that can modulate their fluorescence between an emissive “on” state and a non-emissive “off” state. The resolution is then no longer

212

Maike Steindel et al.

determined by the diffraction limit [30] but instead by the uncertainty with which a single molecule can be super-localized and detected, the labeling density and the size of the probe used to label the molecule. Live-cell SMLM approaches can address dynamics that occur at the level of nucleosomes as well as larger structures that require a resolution of ~10–20 nm. It has allowed single-particle tracking (SPT) of proteins on chromatin. Such studies have been used to extract information on the diffusion coefficients of proteins, their chromatin-binding kinetics, as well as the search time of a protein for its specific site on DNA [31–37]. In addition, live-cell SMLM has allowed us to study the spatial distribution of proteins within the nucleus such as transcription factors and chromatin regulators [3, 20, 38, 39] and to track the movement or compaction of chromatin through the tracking of chromatin-bound proteins and single-molecule Fo¨rster resonance energy transfer (FRET) [32, 40– 46]. It has even been used to image more complex structures such as heterochromatin fibers [47]. In this chapter, we describe how live-cell SMLM can be used to acquire these parameters for your protein of choice.

2

Materials

2.1 SMLM Microscope Configuration

A typical SMLM microscope requires collimated excitation lasers (and activation lasers for photoswitchable proteins/dyes) for the relevant wavelengths that are aligned and focused at the back aperture of a high N.A. objective lens mounted, commonly, on an inverted microscope frame. An example setup for imaging of either the photoconvertible protein mEos or the photoactivatable dye PA-JF549 (see Subheading 2.2.1) is shown below (Fig. 2a) and includes the following components: l

Optical table, vibration isolated (e.g., Thorlabs/Newport).

l

High-numerical-aperture-objective lens (e.g., Nikon Apo TIRF 60X N.A. ¼ 1.49 Oil).

l

Inverted optical microscope (e.g., Nikon Eclipse Ti2).

l

Motorized sample stage (e.g., Prior HLD117).

l

Focus drift compensation (e.g., Nikon Perfect Focus System).

l

Excitation laser, 488 nm: to image green form of mEos before photoactivation (e.g., Toptica, iBeam Smart 488).

l

Excitation laser, 561 nm: to image PA-JF549 or red form of mEos after photoactivation (e.g., Cobolt, Jive 200).

l

Activation laser, 405 nm: to photoconvert mEos or PA-JF549 (e.g., Oxxius, LaserBoxx 405).

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . .

213

Fig. 2 Typical microscope setup for single-particle tracking. (a) Beam path and lasers common to a microscope setup with activation and excitation lasers as well as a depiction of how oblique angle illumination and phase mask-based 3D detection approaches are set up. (b) Comparison of setups for wide-field, partial illumination, and selective plane illumination microscopy (SPIM). Excitation beam (yellow arrow) and collection objectives are indicated. (c) Schematic of common 3D detection approaches. (Left) Astigmatism, (middle) double-helix point-spread function (DH-PSF), and (right) multifocus microscopy approaches involve the use of a cylindrical lens, phase mask, or multifocal grating, respectively, to either shape the beam or split the signal into different regions of the camera (in the case of multifocus microscopy)

214

Maike Steindel et al. l

Laser power meter (e.g., Thorlabs PM100D).

l

Mechanical shutters for pulsing of lasers (e.g., Thorlabs SH05/ Prior HF200HT).

l

An electron-multiplying charge-coupled device (EMCCD) camera for detection of individual fluorophores with high quantum efficiency (e.g., Photometrics Evolve/Delta 512, Andor iXon3 897) or scientific complementary metal–oxide–semiconductor (sCMOS) camera (e.g., Photometrics Prime 95B, Hamamatsu Orca Flash 4.0).

l

High-efficiency optical filters and dichroics (e.g., Semrock, USA): – Multi-band dichroic filter (e.g., Semrock, Di01-R405/488/ 561/635). – 488 long-pass filter (e.g., Semrock, BLP01-488R) – 561 long-pass filter (e.g., Semrock, BLP01-561R) and bandpass filter (Semrock, FF01-587/35).

2.1.1 Sample Illumination

l

Wavelength-optimized phase masks can be added (see Subheading 2.1.2) for 3D data acquisition such as a double-helix phase mask (DoubleHelix, USA) or cylindrical lens for astigmatism.

l

Software for image acquisition such as Micro-manager (https:// www.micro-manager.org) and image processing software such as ImageJ/Fiji [48, 49].

The sample may be illuminated using wide-field illumination. This is appropriate in mammalian cells for 3D detection, where it is necessary to excite molecules in all the axial planes of interest simultaneously, but the approach is inefficient due to excitation of out-of-focus fluorescence. A common technique employed to increase the signal-to-background ratio of a fluorophore is to confine the excitation geometry by reducing the volume of the nucleus that is illuminated by the excitation source and therefore reducing the background signal generated by cellular autofluorescence and laser scattering [50]. These can be implemented in software via smart filtering of existing datasets [51], or more typically experimentally using one of several selective plane illumination microscopy (SPIM) approaches (Fig. 2b). Common ways to illuminate the sample usually involve partial illumination of the nucleus: 1. Oblique plane microscopy (OPM) images samples on a standard inverted microscope setup at an inclined plane (typically at 30 ), using a refocusing system, which enables illumination with an inclined thin light sheet (800 nm). High-NA-objective lenses can be combined with oblique plane microscopy for SPT (eSPIM and obSTORM) [52, 53]. Use of a bespoke glasstipped objective lens further improves the performance

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . .

215

[54]. This enables light-sheet illumination without any additional components, and is compatible with multiwell plate imaging and galvanometric scanning for high-speed 3D imaging. 2. Highly inclined and laminated optical sheet or HILO microscopy [55] only requires the use of a high N.A. objective and is therefore the simplest technique to implement for SPT of molecules within a mammalian nucleus. Whereas in total internal reflection fluorescence (TIRF) microscopy the beam is reflected off the top surface of the imaging coverslip by the incident light being directed above the critical angle, HILO brings in the light beam at a subcritical angle such that it refracts through the glass but at an oblique angle that creates a light sheet illuminating only part of the nucleus. TIRF itself is rarely used as it only illuminates the bottom surface of the cell (~100 nm) and so usually misses most of the nucleus, only really allowing the study of nuclear membrane-associated proteins. The limitation of HILO microscopy is the thickness of the sheet (6 μm), which is larger than the typical z range (500 nm). 3. Reflected light-sheet microscopy (RLSM) [31] and singleobjective SPIM (soSPIM) [56] are approaches based on the reflection of a light sheet across the cell at the detection plane. The first requires an excitation objective from above and detection from below whereas the second excites and detects from the single inverted objective lens. Both require some skill to assemble. For example, the RLSM requires the positioning of an objective lens and a disposable tipless atomic force microscopy (AFM) cantilever (e.g., fHYDRA2R-100 N-TL-10, Nanoscience) next to the cell being imaged. Note that the requirement of the extra objective lens can be removed by inverting the cantilever, enabling RLSM in multiwell plates [57]. The cantilever is coated with a 1 nm titanium layer followed by a 40 nm aluminum layer by thermal evaporation so it can act as a mirror to reflect a light sheet coming from above across the cell. The light-sheet illumination is thinner (~1 μm) and so this leads to lower background fluorescence than HILO while offering improved light collection efficiency when compared to OPM. At a low density, the signal-to-background ratio improvement is only 1.5-fold but it can be as much as fivefold at high densities [31]. This approach is therefore particularly appropriate if there is a high density of fluorophores in the focal plane of detection either because the molecules are not photoactivatable or because large numbers of molecules must be activated quickly (for example, to visualize structures).

216

Maike Steindel et al.

4. Prism-coupled light-sheet illumination [47]: This is where the illumination is from below through a prism and it follows the same principle as RLSM in that the light sheet is in the detection plane. 5. Lattice light-sheet illumination [58] is more difficult to experimentally implement but is useful if a thinner sheet or more even illumination is required over larger distances, e.g., when imaging several cells at the same time or when imaging thicker samples such as tissues or embryos. The excitation lattice is propagation invariant because the transverse profile of the beam remains unaltered during free-space propagation, leading to properties such as a tight (600 nm).

218

Maike Steindel et al.

3. No. 1.5 glass coverslip-bottomed dishes, e.g., from Mat-tek or Nunc Lab-Tek II dishes. 4. Fixation buffers: Methanol, ethanol, formaldehyde, glutaraldehyde, phosphate-buffered saline (PBS). 2.2.1 Fluorophore Choice

Since SMLM requires fitting of distinct molecules/puncta to a PSF, overlapping of puncta is circumvented by a reduction in the number of detectable emitters in a given frame. A simple method to achieve this is under-labeling, where a sufficiently low probe concentration is used to ensure sparse signal distribution within the nucleus (along with selective plane illumination volumes, e.g., RLSM, soSPIM, or lattice light-sheet illumination; see Subheading 2.1.2). It should be noted that the signal must be sparser for some PSF shaping methods that occupy more pixels such as the DH or tetrapod PSF approaches. While this under-labeling approach is sufficient in experiments where low statistical power can be compensated through an increase in sample size, some experiments require large numbers of molecules detected per cell, such as experiments involving molecule counts per cell or subnuclear localization. In these cases, fluorophores must be chosen that can exist in distinct fluorescent states to allow signal emission to be manipulated either chemically by altering environmental conditions such as the oxidation of reduced groups within the dye molecule (e.g., hydrocyanine dyes) or through illumination strategies (e.g., UV irradiation). Fluorophores are often classified into three categories: l

Photoactivatable dyes undergo conformational changes from a nonfluorescent to a fluorescent conformation.

l

Photoswitchable dyes reversibly transition between the “on” and the “off” states to effectively blink.

l

Photoconvertible dyes transition between different absorption and emission maxima, so that in a given detection channel, the molecule effectively transitions from an “on” to an “off” state, but undergoes the reverse transition in a different detection channel.

While this categorization is not mutually exclusive and can be combined (e.g., IrisFP, which can switch from a nonfluorescent to a green fluorescent state but also from a nonfluorescent to a red fluorescent state [74]), these different modes of transition are trade-offs that need to be applied in accordance to the experiment. The major types of fluorophore to choose from are as follows: 1. Fluorescent proteins (FPs) can be genetically encoded to the target gene so that it will be synthesized in fusion to the target protein. Provided that the target protein concentration is sufficiently low, FPs with high photostability and fluorescence quantum yield such as YPet or TagRFP can be used [31, 75, 76]. For higher protein concentrations, FPs have been

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . .

219

modified so that they can be activated into a fluorescent “on” state using an activation laser (often 405 nm) [28, 77–80], where the power of the activation laser determines the number of molecules in the “on” state. For example, PA-GFP can be photoactivated by 405 nm UV irradiation to achieve a 200-fold increase in photon yield when imaged using 488 nm excitation. Similarly, UV can be used to photoconvert EosFP-derived proteins from green to red fluorescence [81–83] or to modulate the fluorescence of photoswitchable FPs such as Dronpa and Dendra/Dendra2 [84, 85]. When choosing FPs to use, it is important to take into account that some FPs may not fold and mature as well as others and so have a smaller percentage of their molecules actually able to switch to the fluorescent “on” state. When studying whether proteins cluster together, it is also important to take into account that some of these proteins have dimerization tendencies. For these experiments, one should take advantage of FPs such as mEos3.2-derived FPs that have been modified to reduce the tendency to dimerize [82, 83]. 2. HaloTag®/SNAP-/eDHFR-tagged proteins are enzymes developed to specifically label proteins with organic dyes that are introduced to cells [37, 86–89]. Common strategies for preventing overlap of single dye-labeled proteins are to either under-label proteins in the cell with dyes such as JF549 or use photoactivatable dyes that are sensitive to UV such as PA-JF549 [86, 90, 91], as described above for FPs. Dyes generally emit more photons than FPs and so can be used to track molecules at higher spatial resolution or for longer than is possible with FPs. This is ideal for determining residence times or for capturing longer trajectories. Long trajectories are essential when trying to capture switching of states during a trajectory. Blinking dyes have also been developed for the purposes of structural imaging [92]. The major disadvantage of dye labeling when compared to FPs is that it can be difficult to label all the proteins in the sample and so the percentage of labeled proteins may need to be quantified and optimized before interpreting experiments such as counting studies. For example, factors such as the cell permeability of these dyes can be variable between cell types (dyes that are not permeable may need to be electroporated or injected into the cell to increase labeling efficiency). In addition, some dyes show nonspecific binding to lipid membranes and this can again be dependent on the cell type. To reduce the likelihood of detecting non-specifically bound dye molecules, dyes have been developed that become fluorescent only when chemically conjugated to the enzyme [88]. However, these dyes may not be photoactivatable or as photostable as other dyes and so would be used only if non-specific binding is a problem.

220

Maike Steindel et al.

3. Dyes can also be used to directly label proteins such as nanobodies before they are injected/transfected/electroporated into the cell [93, 94] but this approach is less common given that the protein will then need to be expressed, purified, and labeled in vitro. 4. Unnatural amino acids with side groups that can react with dyes introduced to cells [95]: This approach, although very promising, has yet to be widely adopted by the community. The suitability of a dye for a given experiment can also be described in terms of the following photophysical properties [96, 97]: (a) Photostability or photobleaching quantum yield: If a fluorophore is more photostable, it can be imaged for longer before irreversible photobleaching. These long track lengths are ideal for determining the residence time of proteins bound to chromatin. Dyes generally have greater photostability than fluorescent proteins. Strategies have also been developed to use Fo¨rster resonance energy transfer (FRET) to increase the photostability of photoactivatable/photoconvertible fluorophores, which can be used to detect protein-protein interactions [46]. (b) Fluorescence brightness (a function of a fluorophore/s extinction coefficient and its quantum yield): For example, if a fluorophore has a high fluorescence quantum yield, more photons are emitted for a given number of photons absorbed. Higher photon counts give rise to greater localization precision and therefore a better characterization of diffusion. (c) Switching cycles/dye molecule: When used in photoswitching experiments, the number of switching cycles can vary from fluorophore to fluorophore but is generally higher for dyes than for fluorescent proteins. A high number of switching cycles allows the structure to be imaged repetitively over a longer time frame (proteins developed for RESOLFT often have the most switching cycles). However, the blinking of photoswitchable dyes leads to a single target being detected more than once and so these dyes may be less suited for counting experiments (see Subheading 3.4.1). In contrast, FPs/dyes that can be activated fewer times, such as PA-JF549 and mEos2/3, may be more appropriate for counting experiments. (d) “On”/“off” ratio (or duty cycle) during switching cycle: Fluorophores with a lower “on”/“off” duty cycle may be useful when studying denser structures that have a higher likelihood of fluorophores overlapping during their blinking cycles whereas a greater “on”/“off” duty cycle allows the structure to be imaged quicker.

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . . 2.2.2 Designing Expression Vectors

221

There are several common ways to express your tagged protein and the preferred method is dependent on the question being asked: 1. Transient transfection of a tagged protein expressed from a promoter of choice is the easiest but least favored approach. It does not label every cell and shows cell-to-cell variation in the percentage of fluorescent molecules labeled. Analysis of data like this can be difficult as expression-level differences can often affect binding of proteins to chromatin and to other chromatin-binding proteins. However, it is still possible to interpret changes in binding that arise from mutations as long as the expression level is not too high. Transient transfection is usually used to verify the expression and localization of a construct prior to making a stable cell line. 2. Stable expression of the tagged protein is usually important when characterizing the binding dynamics of a protein since association rates are concentration dependent. It means that there is a consistent expression level between cells. Stable expression can be achieved using the same expression vector as above but transfection is followed by selection using an antibiotic-resistant gene (also on the vector) and expansion of a colony with a given expression level arising from a single cell. However, over time these expression constructs are sometimes silenced. In these cases, PiggyBac vectors that insert genes at specific transposon sites are popular and exhibit less silencing [98]. 3. Knock-in cell line generation is the preferred approach as then proteins are labeled at close to the endogenous levels of the protein. However, this approach is relatively time consuming and involves recombining the fluorophore sequence into the relevant genomic site usually through RNA-guided genome editing, based on clustered, regularly interspaced, short palindromic repeat (CRISPR) technology [99]. If carrying out counting/clustering experiments, a homozygous knock-in cell line can mean that there are no untagged proteins to worry about. If proteins associate with other proteins before binding to DNA, maintaining expression levels similar to the endogenous protein in this way allows a better mimic of the endogenous interactions being studied. It is worth noting that the chosen tag can still affect protein degradation and so its actual expression level should be checked by Western blotting. Other considerations when designing an expression construct are the following: 1. Promoter choice: The expression level of a promoter and its likelihood of being silenced are cell line dependent [100] and so this should be considered depending on the specific cell line being used.

222

Maike Steindel et al.

2. Antibiotic resistance gene: This allows the cells to be selected. There are a range of antibiotic resistance genes and some take longer than others to select. For example, puromycin is often quicker than geneticin selection. 3. Does tagging affect protein function? It is critical to check the function of the protein after tagging through a functional study. Sometimes tagging of the N- but not the C-terminus can affect the function of the protein or vice versa. Also, adding amino acid linkers between the protein and its tag can sometimes rescue protein function.

3

Methods

3.1 Sample Preparation

1. Cells should be grown on coverslip-bottomed dishes in phenol red-free medium (see Notes 1 and 2). Keep away from shortwavelength light (typically 350–450 nm) as much as possible if working with UV-sensitive fluorophores to prevent prior photoactivation. 2. When imaging proteins tagged with fluorescent proteins, proceed straight to imaging. HaloTag®/SNAP-tagged proteins must, however, be labeled in the cells using the following protocol: (a) Replace the medium of cells with dye for 15 min (30 min for SNAP dyes or if labeling is inefficient). Incubate cells in normal culture incubator conditions. (b) Wash cells three times with the medium to remove dye and then incubate in normal culture incubator conditions for another 30 min (see Notes 3 and 4). (c) Finally replace the medium for imaging. The concentration of the dye depends on the type of experiment. For single-molecule tracking, low concentrations of dye should be added to prevent overlap of molecules (typically ~5 nM). For imaging of structures using blinking/photoactivatable dyes where the intention is to label all the protein, higher concentrations are required (typically ~10 μM). The exact dye concentration depends on the cellular protein concentration and so should be determined by finding the concentration at which there is no further increase in the number of labeled molecules.

3.2 Microscope Setup

There are changes in the microscope setup depending on the fluorophore of choice, so here we describe the imaging of mEos3 or HaloTag-tagged proteins labeled using PA-JF549/JF549 to give an idea of the steps involved.

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . .

223

1. Measure the laser power densities at the sample plane and use neutral density filters to make sure that they are as follows: (a) 561 nm laser ~1 kW cm2 (b) 488 nm laser ~1 kW cm2 (c) 405 nm laser ~10–100 W cm2. 2. Check optical setup components, e.g., filters, phase masks (for 3D), and beam size (smaller if higher power density is required, e.g., for short exposure or stroboscopic imaging). 3. Prepare a calibration sample of TetraSpeckTM beads or fluorospheres [101] (placed on a poly-L-lysine-coated coverslip so they attach to the surface) and then use this to find the coverslip. Lock the piezo focus to ensure finer z control. If two cameras are being used for detection, these beads can be used to align cameras so these beads overlap. For 3D approaches, a z stack of these beads should be carried out at this stage using the piezo controller to allow subsequent estimation of the z position from the actual data. 4. Prepare a fixed cell sample for setting up the imaging (see Note 5): l l

Option 1: Methanol:ethanol (1:1), 6 min at 20  C. Option 2: 4% Formaldehyde in PBS, 20 min at room temperature. Wash with PBS three times before imaging.

5. Focus on the center of the nucleus using the white light. Then adjust the optimal HILO angle or set up the light sheet (for example, for RLSM) to achieve the highest excitation at the specific detection plane. For mEos3 imaging, this can be optimized using the 488 nm laser as the non-photoactivated molecule emits in the green part of the spectrum. For dyes such as PA-JF549 that are dark before photoactivation, this can be optimized by activating some of the PA-JF549 molecules by a pulse of 405 nm laser followed by imaging using the 561 nm laser. 6. Set up image acquisition parameters (e.g., using MicroManager): laser shutters (e.g., 405/488/561 lasers), EM gain (usually 250 for Photometrics Delta), camera ID (if more than one camera), exposure time, continuous vs. pulsed imaging. 7. The power of the activation laser (in this case 405 nm) must be optimized to ensure the activation of isolated single molecules. Bright puncta should show single-step photobleaching intensity traces over time under continuous 561 nm excitation to confirm that they are single molecules. Activation laser powers should be lowered if there are spatial overlap between emitters.

224

Maike Steindel et al.

8. The power of the excitation laser (in this case 561 nm) must be optimized for the specific question being asked. We use fixed cells here to prevent the analysis of molecules that may move out of the detection plane. There is a trade-off between precision (a measure of uncertainty of the spatial position of a single molecule) and bleaching time (the time taken for a fluorophore to irreversibly transition to a nonfluorescent state) and the type of experiment determines the suitable laser power. For measuring diffusion parameters such as the jump distance, higher 561 nm power is used to obtain higher precision at the expense of smaller track lengths. When trying to determine residence time or to better estimate biophysical parameters such as the anomalous exponent (see Subheading 3.4.3), longer track lengths are useful. Note that the addition of noncytotoxic triplet-state quenchers such as 5 mM Trolox can be used to reduce photobleaching of many dyes [46, 102, 103]. 9. When trying to image a structure, a fixed cell sample can be used to estimate the emitter density and therefore the number of molecules that should be detected to be able to effectively visualize the structure. Molecules that form simple structures such as spherical hubs may take less time and so allow faster imaging than molecules that form more complicated spatial patterns, e.g., fibers. In addition, the active emitter concentration, that is, the ratio of molecules in the “on” versus the “off” state, is critical to prevent overlap of the PSF of individual molecules. 10. Finally start acquisition using automated scripts if necessary. 3.3 Experimental Design for SingleMolecule Imaging and Tracking

There are several types of experiments that can be carried out based on the type of experimental question (Fig. 3–5). For ease of discussion, we will predominantly focus on the concepts behind these experiments using the mEos or HaloTag tags and the 3D-DHPSF single-molecule imaging approach [33]. However, these principles apply to any other fluorophore being imaged in two or three dimensions. SPT can be used in live cells to determine the diffusion parameters, residence times, or spatial distribution of single proteins, as well as the diffusion parameters of chromatin in 2D or 3D.

3.3.1 Short-Exposure Tracking of Single-Protein Diffusion and Binding to Chromatin

The exposure time at which a protein should be tracked depends on the specific question being asked and the diffusion rate of the protein itself. Most chromatin-binding proteins exhibit multiple modes of diffusion—they (1) diffuse freely in the nucleus but also (2) bind nonspecifically and specifically to DNA and/or nucleosomes (Fig. 5a) [31–35, 37, 93, 104]. To detect freely diffusing proteins (often with a diffusion coefficient of 1–20 μm2 s1), it is necessary to image at fast exposures (1–20 ms). Note that most cameras can acquire data at 10–20 μm2s1), motion blurring may occur even at very short exposure times. As the camera exposure may not easily go below 10 ms, stroboscopic imaging can be carried out, where an acoustooptical modulator is used to reduce the illumination time below 1 ms [105], which can also be achieved via TTL (transistortransistor logic) control of laser diodes. At the fast exposures required to capture freely diffusing molecules, chromatin-bound molecules appear predominantly immobile with their effective diffusion coefficient often being within the precision limit. If the aim of the experiment is to determine the number of molecules that are freely diffusing vs. chromatin bound, an exposure time must be chosen such that the freely diffusing molecule is clearly faster than the chromatin-bound protein at the precision limit. For example, at a precision of 30 nm, chromatinbound molecules will have an effective diffusion coefficient

226

Maike Steindel et al.

Fig. 4 Extracting quantitative spatial information from SMLM. (a) Counting experiments (that account for fluorophore blinking): To determine the number of molecules, localizations are identified and tracked over time and then the number of tracked localizations is converted to the number of molecules after correcting for the blinking rate of the fluorophore. (b) Spatial clustering of molecules: Algorithms developed to identify clustered molecules involve the use of a density metric (in this example, we show the Voronoi approach which uses Voronoi cell area/volume as a metric for density). Molecules are assigned a density metric (left) that is then compared to the density metric calculated for a simulated random distribution of the same number of molecules occupying the same area (middle top). Molecules are then assigned as clustered if they have a higher density metric than that observed for randomly distributed molecules. Finally, neighboring molecules that have been assigned as clustered are connected to generate clusters (top right) that can then be used to extract biological parameters (bottom right) such as the cluster size (area/volume), the number of clusters/cell, and the number of molecules/cluster (note that clustering algorithms must take into account blinking as in (a)). (c) Table comparing spatial clustering algorithms

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . .

227

Fig. 4 (continued)

(squared displacement/time) of 0.03*0.03/0.02 ¼ 0.045 μm2 s1 at 20-ms exposure and an effective diffusion coefficient 0.03*0.03/ 0.005 ¼ 0.18 μm2 s1 at 5-ms exposure. Although, in both cases, it is possible to distinguish chromatin-bound proteins from freely diffusing proteins with a diffusion coefficient >0.18 μm2 s1, there is a trade-off when choosing an exposure time for the imaging. Specifically, the exposure time needs to allow for the molecule to move enough to track it over several frames while avoiding motion blur, i.e., the molecule needs to move just enough to track motion above the precision limit but not too much within the chosen exposure window to avoid motion blur. As a result, slowly diffusing proteins are easier to distinguish from chromatinbound proteins at 20-ms rather than at 5-ms exposures due to the lower effective diffusion coefficient of chromatin-bound proteins, and fast-diffusing proteins show more motion blurring and so become harder to detect at 20-ms than at 5-ms exposure. It should be noted that 3D imaging and tracking, which have been made possible through advances in approaches like 3D-DHPSF imaging (see Subheading 2.1.2), have significantly improved our ability to generate longer, continuous tracks required for tMSD analysis (Fig. 5b), specifically for freely diffusing molecules, and thereby to accurately determine the biophysical

228

Maike Steindel et al.

Fig. 5 Extracting dynamic information from SMLM. (a) (Top) Representative images of freely diffusing and chromatin-bound PA-JF549-HaloTag-tagged proteins collected by 3D DH PSF imaging at 20-ms and 500-ms exposure. At 500-ms exposure, motion blurring of freely diffusing proteins allows localization of only chromatin-bound proteins. (Bottom left) 20-ms trajectories allow detection of both freely diffusing and chromatin-bound proteins. (Bottom right) 500-ms trajectories allow us to study whether chromatin movement is slow/fast diffusing. (b) For 20-ms exposure imaging, MSD vs. time plots identify proteins that are freely diffusing vs. chromatin bound. (c) For 500-ms exposure imaging, chromatin-bound proteins exhibit slowvs. fast-diffusing vs. directed chromatin movement. Through trajectory segmentation or fitting of diffusion coefficient distributions, it is possible to extract (d) the percentage of molecules in each of these states, or (e) the diffusion coefficient of these states through Gaussian fitting. (Note that, for 500-ms exposure imaging, the anomalous exponent can also be extracted from these plots to determine how tethered the chromatin is.) (f) Pulse-chase imaging at 500-ms exposure involves varying the lag time (tl) between exposures and is used to determine the residence time. (g) By fitting at different time lapses to an exponential distribution defined by keff and then (h) plotting the lag time tl against the keff .tl, one can extract the residence time koff and the photobleaching rate kb

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . .

229

parameters (see Subheading 3.4.3) of freely diffusing and chromatin-bound molecules by allowing us to record and account for axial movement [32, 33]. Therefore, 3D approaches are preferred when extracting biophysical parameters of freely diffusing molecules. In addition to fixed-cell experiments, useful controls include proteins such as HaloTag-tagged histone H2B and HaloTag-NLS (HaloTag with a nuclear localization sequence) that should be bound and unbound, respectively. Chromatin-binding mutants are ideal if possible for the protein of interest as they better reflect the size and shape of the chromatin-binding protein being imaged and therefore would be more likely to show similar biophysical properties. 3.3.2 Long-Exposure Tracking of Single Chromatin-Bound Proteins for Probing Chromatin Diffusion

Longer exposure times (~500 ms) make use of “motion blurring” of faster diffusing molecules, which allows exclusive characterization of the immobilized/bound molecules in the nucleus (Fig. 5a) [31, 105, 106]. This experimental setup enables the quantitation of changes in the mobility/diffusive state of native chromatin by tracking the movement of labeled single chromatin-bound protein molecules [32, 107]. The long exposure can result in considerable cell movement being recorded. To account for this, it is recommended to include a nuclear envelope marker, such as Lamin B (LMNB1), or a fiducial (e.g., fluorescent beads or gold nanoparticles) marker in your experiments. Without fiducial controls, analysis should be limited to movement that occurs faster than that measured for cell movement. Labeling of global chromatin-binding proteins, such as H2B [107], provides information about global chromatin movement. However, it is also vital to characterize the dynamics/diffusion characteristics of a specific endogenous genomic locus. This allows us to probe how sequence-specific chromatin regulators influence the movement of genomic loci or how movement at specific loci differs between cell types [32, 43]. Live-cell tracking of specific genomic loci can be achieved by the generation of (1) knock-in cell lines with arrays of DNA-binding sites and expression of fluorophore-tagged proteins that bind to these loci, e.g., bacterial Tet and Lac repressor proteins that bind to unique sequences not commonly found in mammalian cells [40–42], or (2) cell lines expressing fluorophore-tagged TALEN (transcription activator-like effector nucleases) or dCas9 (CRISPR-associated protein 9) proteins targeted to a genomic site of interest [32, 43, 44]. All of these approaches require cloning and cell line generation. Imaging success is largely dependent on a large number of fluorophores binding to the locus of interest to allow visualization above background (with photon yield dependent on the brightness of the

230

Maike Steindel et al.

dye, etc.). Although initially limiting single-locus labeling to repetitive sequences, various approaches have now been developed to overcome this limitation, such as the use of a CARGO plasmid to deliver 36 different gRNAs targeting a 2 kb window close to the target locus in a transgenic cell line with inducible dCas9-eGFP expression [32, 43]. In addition, in order to relate the motion of individual loci to other genomic/gene-regulatory elements, simultaneous imaging of two genomic loci is required. Both types of approaches, Tet/Lac systems and TALEN/dCas9/dCas13 systems, have now been successfully applied to study dual-color chromatin organization in live cells [42, 45]. For these experiments, it is important to validate correct subnuclear labeling of the targeted locus using orthogonal techniques such as DNA-FISH. Additionally, care should be taken to target proteins up/downstream of the sequence of interest so as to avoid protein binding affecting the function of the genomic sequence. Although labeled protein binding might affect the diffusive properties of the locus of interest, if the collected data is a relative comparison between different, identically treated/processed samples, we can assume that potential changes to the locus dynamics will be consistent across samples and so relative change between treatments is informative. 3.3.3 Time-Lapse Experiments for Determining Residence Time

Motion blurring through the use of longer exposure times, as explained above, can further be harnessed to determine the residence time of chromatin-bound proteins (the time the molecule remains bound) (Fig. 5f). Residence time determination from SPT is complicated by the problem of photobleaching. If photobleaching occurs at a slower timescale than the residence time of the protein, continuous imaging can be carried out with a fixed cell control to measure and correct for the photobleaching rate [34, 37]. However, if photobleaching occurs at a quicker timescale than the residence time of the protein, the residence time cannot be determined by continuous imaging and requires a time-lapse experiment [31, 32] (Fig. 5f). By reducing the duty cycle of the 561 nm laser after an activation pulse of 405 nm, it is possible to perform time-lapse imaging. If increasing the time between pulses results in the same track length, then the residence time is greater than the pulse length. If the track length decreases, then the residence time can be determined. Therefore, the experiment involves several imaging experiments in which the time between pulses is increased. It should be noted that measuring residence times for proteins bound for very long periods of time (such as histone H2B) is difficult as the cell moves and so it can be difficult to connect trajectories if the time between pulses is too large.

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . .

231

3.3.4 Structural Imaging

Accurate live-cell imaging of a structure depends on its size and complexity. By imaging the structure in fixed cells, the number of localizations required to accurately reconstruct a super-resolved image of the structure of interest can be determined. In superresolution imaging, temporal resolution is often traded off for increased spatial resolution, so for dynamic biological structures it is important to consider these temporal constraints. To visualize how the structures change over time, a sliding window analysis can be applied [87, 108]. So for example, if it takes 10 s to resolve the structure but imaging is carried out at 50-ms time resolution, the structure generated from every 10-s sliding window of localization data (t to t + 10 s, t + 0.05 to t + 10.05, . . .) can be used to generate a movie at 50-ms time resolution where each frame is a sliding window. However, it is often the case that the structure moves slower than this in which case to prevent bleaching of the fluorophores, a time-lapse experiment can be designed, for example, such that a 10-s pulse of continuous imaging is carried out every 5 min.

3.4

Data requires several preprocessing steps prior to analysis:

Data Analysis

1. Visual inspection of data using software like ImageJ/Fiji is usually required to ensure that localizations are not overlapping (if there is overlap, ideally one should reduce the photoactivation rate if possible; otherwise see Subheading 3.4.6). 2. Fluorescent puncta localization programs (see review [109] to choose the appropriate method based on the density of localizations and 2D/3D approach being used): Such programs typically apply a band-pass filter (since noise occurs at high frequency and background at low frequency) to accurately identify the initial fluorescence localizations, followed by fitting of the raw data above a user-defined threshold, that (1) have a high signal-to-noise ratio (typically >6 for mEos3 or PA-JF549 dyes) and (2) have a PSF width predicted by theory (to avoid fitting of noise or overlapping molecules) (Fig. 3a). Fitting can be through either Gaussian or non-Gaussian PSF models with non-Gaussian approaches such as spline fitting often being preferred for 3D approaches such as biplane and double helix (e.g., SMAP-2018). For 3D analysis, these programs also typically require a calibration file to estimate the z-position (generated from a z stack of a TetraSpeckTM bead; see Subheading 3.1). Blurring that occurs when molecules move quickly during an exposure can sometimes be accounted for [36]. 3. Tracking algorithms (e.g., GitHub—wb104/trajectory-analy sis: Python code to analyze trajectories that allow inputs from PeakFit or EasyDHPSF datasets) (Fig. 3b): A key parameter here is the “-maxJumpDistance,” the distance between dots in

232

Maike Steindel et al.

adjacent frames. This should be high for freely diffusing proteins (e.g., 800 nm for low-density 5–20-ms exposure images), low for motion-blurred chromatin-bound proteins (e.g., 500 nm for 500-ms exposure images), and even lower for fixed-cell images (e.g., 50 nm). 3.4.1 Quantifying Numbers of Molecules

Much progress has been made recently in the quantification of SMLM approaches (Fig. 4a). Here we describe three main considerations when analyzing this kind of experiment: 1. Are all the proteins tagged? If a knock-in cell line has not been generated and the untagged endogenous protein is still present, the ratio of tagged to untagged protein must be accounted for. 2. Are all the tagged proteins fluorescent? Even if all the proteins are tagged (e.g., in a knock-in cell line), dye labeling in the case of HaloTag®/SNAP tag is often protein dependent and for endogenous fluorophores, their maturation is rarely 100% (e.g., GFP is ~80% and mEos2 is ~50%) [110–112]. The percentage of expressed fluorophores that are fluorescent can be calculated by comparing the concentration predicted by fluorescence against the actual protein concentration calculated from the absorbance at 280 nm (either using the purified tagged protein or by quantitative Western blot from a known number of cells). 3. Does the molecule undergo fluorescence intermittency or “blinking”? Many photoactivatable fluorophores go to a recoverable/reversible dark state rather than via the photobleaching pathway, which can result in systematic overcounting (Fig. 4a). For example, mEos2 molecules blink an average of two times before bleaching [113]. Recent work has shown that these blinking events can be separated temporally and spatially [114], that is, they occur within a specific time frame after the first localization. By applying a filter to take into account the precision of the detected molecule (count as one molecule if tracked in successive frames) and by calculating the number of expected blinking events for a specific fluorophore, it is possible to determine the number of molecules [113, 115]. To summarize:

Protein count ¼

3.4.2 Extracting Clustering Parameters

Number of molecules counted ðBlinks=fluorophoreÞ  ðFraction of fluorophores fluorescent Þ

Approaches that help us understand how molecules are distributed throughout the nucleus are critical for us to fully harness the spatial information generated by SMLM data. Of particular interest is the formation of clusters of molecules found at intranuclear

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . .

233

compartments such as the nuclear condensates or “hubs” of transcription factors that control transcription or the heterochromatin compartments arising from liquid-liquid phase separation within the nucleus [15, 20, 116]. This section explores SMLM methods which have been applied to the identification of clusters of chromatin-binding proteins, specifically how they can be used to identify the number of clusters present, the number of molecules within each cluster, the relative nuclear position of the clusters, and the existence of any supranuclear arrangements (e.g., inclusion of or exclusion from nucleoli) (Fig. 4b). Two common approaches are (1) spatial statistical methods and (2) “density-based spatial clustering of applications with noise” (DBSCAN). More recent methods that have been applied are (3) Bayesian analysis and (4) Voronoi tessellation (reviewed in [117, 118]) (Fig. 4b, c). Details of these approaches are as follows: 1. Spatial descriptive statistics (pair correlation analysis [39, 119, 120], Ripley’s K-, H-, and L-functions [121], and the nearestneighbor distance distribution [122]) allow for the derivation of global descriptors such as the clustering degree or average cluster radius, but do not directly identify individual clusters themselves. While useful, these methods fail to inform on the cluster heterogeneity, cluster shape, or positioning of a cluster within a nucleus. Extending this method to 3D datasets often necessitates processing 2D projections [123] which can introduce artifacts [117]. 2. DBSCAN, in contrast, can identify individual clusters. This can be performed on both 2D data and 3D point clouds [124, 125]. A main caveat is that the algorithm depends upon user-specified parameters for density threshold and length scale. Subjective parameter selection can introduce systematic bias, while sample-specific parameter tuning can be computationally expensive for 3D datasets. In addition, DBSCAN cannot account for localization uncertainties, which are important for 3D datasets where the axial (z-axis) resolution may be different to the lateral (x/y-axis) resolution. A grid-based version of DBSCAN partially addresses this limitation but has until very recently been restricted to 2D datasets and is further confounded by small or overlapping clusters. This is especially sensitive to high point cloud densities and noise levels [126, 127]. Simulation-aided DBSCAN (SAD) has been developed to mitigate these issues by training assignments on simulated clusters of varying sizes [128, 129]. 3. Bayesian analysis is a model-based approach that uses a priori information on the molecular distribution to replace the userspecified parameters of DBSCAN and it explicitly accounts for localization uncertainties. It is capable of operating on 2D and 3D datasets [130, 131] but has limitations: (i) computationally

234

Maike Steindel et al.

intensive, requiring multiple hours to evaluate a single 3D sample; (ii) sensitivity to imaging artifacts and user-defined probabilities; and (iii) being limited to detecting largely uniform, circular, or spherical Gaussian clusters as prescribed by the a priori model. 4. SR-Tessler (2D; [132]) and ClusterViSu (2D and 3D; [133, 134]) are algorithms based on Voronoi tessellation [135] (Fig. 4b). Here, localizations are partitioned into Voronoi cells with an associated density, defined as the reciprocal of the Voronoi Cell area (2D), or volume (3D). Clustered points are identified based upon a density threshold generated by a Monte Carlo approach. Advantages of this method include the ability to identify individual clusters by the generation and thresholding of a density map created by interpolating the Voronoi cell density over the domain. A promising outlook for cluster identification includes graph theoretic and machine learning approaches [117, 136–139] but to date these have not been implemented on SMLM data of chromatin-binding proteins. Despite their successes, none of these approaches have fully removed the dependency on prior knowledge or user input. This makes it difficult to automate and batch process data analysis. Nevertheless, any one of these clustering algorithms can be incredibly useful, when its limitations are respected. It should also be noted that improvements in the determination of clusters occur through not only refinements in algorithms but also empirical quantification. For example, control experiments with known number of molecules per cluster are ideal [115] but not always experimentally feasible. Finally, it should be noted that data preprocessing to remove multi-blinking [96] and multi-labeling [140] of a single protein that could give rise to false-positive pseudo-clusters is crucial to make meaningful biological inferences from the outputs of any of these clustering approaches [118]. Varying the labeling density is also a robust approach for accounting for real clustering vs. blinking of fluorophores [141]. It is also worth noting that clustering analysis can be affected by whether or not the entire structure is in the plane, so it may be important (for 2D detection approaches) to only count molecules within the plane by analyzing the shape and intensity of the localized spot [51] or by using 3D detection methods described previously. 3.4.3 Extracting Biophysical Parameters from SPT

Biophysical parameters can be extracted from SPTs to determine a range of biological parameters (Fig. 5): (1) percentage of chromatin-bound vs. freely diffusing proteins (Fig. 5b, d) [31, 33, 34]; (2) how fast proteins diffuse in different regions of the nucleus [20]; and (3) how fast chromatin moves (Fig. 5c, e) [32, 43, 142–144].

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . .

235

Extracting biophysical parameters from SPTs requires one to first predefine a model for motion. Possible options include pure or confined diffusion, diffusion with a space-dependent drift, switching between different states characterized by different motions, and many other possibilities (drift is a directional movement that can arise technically from the microscope itself or biologically from energy-dependent directed motion). The parameters to extract are given in the associated stochastic equations. Diffusion [145, 146]: The case of assuming pure diffusion is simple because it is characterized by a single parameter: the diffusion constant D, which is independent of the spatial location of the trajectories. Various estimators have therefore been constructed based on fitting the distribution of the mean-squared displacement (MSD), |X(t + Δt)  X(t)|2. The classical law of Brownian motion states that the average of the MSD can be described as follows: < jX ðt þ Δt Þ  X ðt Þj2 >¼ 4D ðΔt Þ ðfor 2D trajectoriesÞ < jX ðt þ Δt Þ  X ðt Þj2 >¼ 6D ðΔt Þ for 3D trajectoriesÞ For practical reasons, this is often measured by averaging over long trajectories. However, computing the MSD along long trajectories can hide spatial heterogeneity because by definition, long trajectories span large distances where the medium organization may change (the effective diffusion coefficient can change with the density of obstacles and thus depends on the location) (see later, Analysis of samples with heterogeneity in the underlying medium). Anomalous diffusion: At an intermediate timescale, we sometimes observe additional behaviors such as correlated motion: for example, this is the case for the motion of a chromatin locus. Indeed, because the motion of the neighboring chromatin regions matters, chromatin movement is characterized by anomalous diffusion which can be described for small time steps as < jX ðt þ Δt Þ  X ðt Þj2 >¼ A ðΔt Þα The parameter A does not carry much physical information but the anomalous exponent α can be used to understand the behavior of the molecule: 0 < α < 1 indicates sub-diffusion or constrained diffusion, where a particle motion is restricted between obstacles. This can create two scales: a small one where the moving particle is restricted and a larger one which corresponds to hopping between regions of confinement. α > 1 is observed for processes involving super-diffusion, a combination of Brownian motion with a persistent directed motion. Drift is also an interesting parameter to be estimated for trajectories exhibiting directed motion, but this requires redundant trajectories (see below).

236

Maike Steindel et al.

More refined classifications are also possible: α < 1/2 characterizes a locus positioned on chromatin with cross-linkers (betapolymer model [147]); α ¼ 1/2 for a Rouse chromatin locus; and α ¼ 2/3 for a locus located on a polymer embedded in a fluid. In addition, when a chromatin locus can “feel” the rotation of the nucleus, the anomalous exponent of a chromatin locus can increase, leading to α > 1/2 [148]. Switching between different states of diffusion [149]: Another classical paradigm to interpret trajectories is the switching motion (1) between proteins that are chromatin bound versus those that are freely diffusing and (2) between slow- and fastdiffusing chromatin (monitored using long exposure imaging of chromatin-bound proteins). In these cases, four parameters are needed for the most elementary case: two diffusion coefficients and two switching rates. In case of more states, including states characterized by drift, a similar procedure can be used [150]. To extract these parameters, many trajectories are required to increase the likelihood of transitions within a trajectory. Trajectory segmentation pipelines can then be implemented to separate each state and to estimate diffusion coefficients or switching rates [32, 151, 152]. Effect of localization error on recovering diffusion [149]: A major difficulty in accurately extracting the parameter for diffusion or anomalous diffusion processes is localization error. This additional noise is often modeled as a Gaussian distribution with variance σ and is determined empirically from fixed-cell datasets acquired at the same exposure and laser powers (see Note 5). The measured MSD (MSDmeasured) therefore includes both this noise σ and the diffusion coefficient evaluated [150] at time resolution Δt, and is defined by MSDmeasured ¼ D ðΔt Þ þ σ 2 =2Δt: Δt When making biological interpretations involving changes in MSDmeasured, values below σ 2/2Δt should not be considered because they are within the localization error. Note that for anomalous diffusion, the relationship becomes MSDmeasured ¼ A ðΔt Þα þ σ 2 ðΔt Þ=2: Analysis of samples with heterogeneity in the underlying medium: To account for cases in which medium heterogeneity is expected, the cell should be divided by a grid into small bins and analysis be subsequently performed on these smaller bins within the cell. Note that the choice of the time step Δt is critical: if too large, local obstacles are missed, and if too small, the Brownian motion model could be invalid (possibly dominated by localization error). The output of this analysis is a diffusion coefficient map D(x) [153]. If large numbers of short trajectories have been collected, they can be used to estimate the local properties of the underlying medium from this type of data. Using a stochastic model, it is

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . .

237

possible to recover the local drift and diffusion coefficient [146, 153]. Due to the averaging of many trajectories, the impact of localization error is negligible [150]. This binning approach also allows one to identify larger scale features such as local accumulation of proteins, regions where molecules can aggregate or be exchanged. Jump distance analysis: For single-fluorophore tracking where the fluorophore has fast-changing diffusion parameters or rarely lasts long enough for MSD analysis, switching analysis may be more reliable [154]. For switching analysis, jump distances within intervals [r, r + dr] traveled by single particles in Δt are counted to estimate the probability that a particle starting at r1 ¼ 0 will be encountered within a shell of radius r and width dr at time Δt. The integrated probability distribution function




Zr 2

P r , Δt ¼ 2


 r2 p r 2 dr 2 ¼ 1  e 4DΔt

0

can be fitted to experimental data. When there are m species present, the integrated distribution is the sum of m terms: m
 X P r 2 , Δt ¼ j ¼1

2 fj  r e 4D j Δt dr 2 4D j Δt

where fj is the fraction of particles in mode j and Dj is the associated diffusion coefficient. The relative likelihood for each model is determined using the Bayesian Information Criterion (BIC) [155] as previously described for the case of comparing different diffusion models for single-particle tracking data [156]. In short, using this approach we calculated the BIC for each model from BIC ¼ ln ½nðp þ 1Þ þ nð ln ½2πRSSn þ 1Þ, where n is the number of data points, p is the number of free parameters for the fit, and RSS is the residual sum of squares of the fit. The fit with the lowest BIC is used. As with the diffusion analysis above, it is important when using switching data to take into account the jump distance that corresponds to the localization error of the imaging data as this is the lower limit that can be measured. In addition, when 2D imaging is carried out, it is important to account for extreme cases where molecules that diffuse quickly (>5 μm2 s1) are less likely to have as many jumps recorded as jumps are more likely to occur out of the detection plane [46]. More recent approaches used Bayesian modeling to determine not only how many types of diffusion occur for molecules but also how often the molecules make transition from one state to another by assigning a jump distance to each diffusion mode [151].

238

Maike Steindel et al.

3.4.4 Residence Time

Short residence times: If the residence time is smaller than the bleaching time, the residence time can be determined from the number of frames a molecule is observed in its “on” state before it turns “off” (ton). Specifically, exponential fitting of the ton distribution is carried out using the following equation to extract the apparent off rate, kmeasured: f ðt on Þ ¼ A e kmeasured :t on As the residence time increases relative to the bleaching time, this measured koff parameter (kmeasured) is affected more and more by the bleaching rate (kb) as described by this equation: kmeasured ¼ koff þ kb For small residence times, fixed-cell experiments can be used to calculate the photobleaching rate kb and this can be subtracted from the measured residence time kmeasured to extract the actual residence time koff. To determine whether more than one residence time occurs, multi-exponential fitting is carried out (for example to extract koff, 1, koff, 2, etc.) and BIC analysis is used to determine the appropriate number of exponential distributions (see Subheading 3.4.3). Long residence times: For longer residence times, one must instead analyze pulse-chase experiments (see Subheading 3.3.3) (Fig. 5f–h). If we pulse the excitation at different lag times Tlag (Fig. 5f) and fit exponential distributions to these experiments (Fig. 5g), we can then extract the contribution of bleaching and the residence time (Fig. 5h)—since the bleaching rate is related to the number of exposures rather than the total time Ttotal, the equation can be written as follows: kmeasured ¼ koff þ kb :T total =T lag If we presume two off rates (koff,1 and koff,2), the data can then be fit to the following equation (Fig. 5g), where B is the fraction of molecules with koff, 1:    ! f ðt Þ ¼ A Be

 kb

T total T lag þkoff ,

1

þ ð1  B Þe

 kb

T total T lag þk off , 2

For dyes that bleach slowly, this kind of analysis is often ignored and it is assumed that the residence time is shorter than the bleaching time. However, it may still be useful to carry out this analysis if only to prove that there are no longer residence times unaccounted for. If the residence time is long, 3D imaging should be carried out instead of 2D imaging since chromatin-bound molecules may otherwise diffuse out of the focal plane. It is also important in these cases to account for cell movement when tracking localizations to reduce the likelihood of connecting localizations from adjacent cell nuclei.

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . .

239

3.4.5 Using Photobleaching Rates to Analyze Single-Molecule FRET Trajectories

FRET can be used to monitor changes in protein proximity and thus changes in protein-protein interactions or in chromatin compaction [46]. Specifically, FRET decreases the signal intensity and increases the photostability of photoactivatable fluorophores such as mEos3.2 and PA-JF549. Therefore, by extracting intensity and track length information from individual trajectories, it is possible to detect changes in FRET or in specific protein-protein interactions. In addition, binning approaches described in Subheading 3.4.3 can be used to track signal intensity/track length and thus to determine regions of the nucleus with increased FRET.

3.4.6 Analyzing Images with High Density of Localizations

The data analysis for images of structures depends on the density of the fluorophores being imaged. Clearly in live cells, by imaging more molecules in a given time frame, the time resolution of imaging a structure can be improved but fitting overlapping molecules can reduce the overall spatial resolution. In order to deal with this problem, approaches have been developed to deal with highdensity images (see comparisons in [109]). These approaches include the following: (1) Multi-emitter fitting, fitting of multiple PSFs to overlapping molecules [157–160]—packages include DAOSTORM, CSpline, and ThunderSTORM. (2) Iterative deconvolution, presuming that the emitters in the high-density image are far enough apart in any given image to deconvolute them [161]—packages include DeconSTORM. (3) Compressed sensing, a sparse-signal recovery technique that deals with the problem of always getting enough photoactivations to reconstruct a structure [162, 163]—packages include FALCON and ADCG. (4) 3B analysis produces a map of the statistical likelihood of an emitter being in a given position through Bayesian-based fitting of an emitter that blinks and bleaches [164, 165]. This approach allows more emitter overlap but is limited by its data processing time, which has recently been improved using Haar wavelet kernel analysis [166]. (5) Machine learning approaches are also being developed to improve the fitting of overlapping emitters—packages include DeepSTORM3D [167].

4

Notes 1. Adding 2 M glycine to clean slides before adding the cells is useful when adding dyes for Halo/SNAP-tagged proteins that would otherwise stick to the coverslip.

240

Maike Steindel et al.

2. One should grow cells in phenol red-free media for a few days as we have noticed that changes in media prior to an experiment can result in higher background and also some initial stress of the cells. 3. This protocol can be very dye dependent. Some dyes aggregate, so must be vortexed before adding. Some dyes take longer to escape the cells and so require an increased number of washing steps, sometimes requiring an overnight wash step. Of course, this increases the time since the label was first added and this can be a problem if protein turnover is fast. In these cases, dyes that are only fluorescent when bound to the protein are preferable. 4. When the dye of interest is not cell permeable, electroporation of the cells can increase the efficiency of dyes entering the cell but it requires replating of the cells to allow for recovery (1–4 h depending on the cell line). This can potentially be overcome via direct nanopipette injection of dyes [168]. It is worth bearing in mind that fixation often introduces artifacts with not all proteins being fixed [169] but also that even fixed cells show some movement of proteins bound to chromatin presumably as the chromatin still moves a little even in fixed conditions. To fix the chromatin further, you may want to add 0.2% glutaraldehyde [170] to the formaldehyde fixation mix followed by a 7-min wash with 0.1% sodium borohydride.

Acknowledgments The manuscript was written by MS, IA, SES, VC, AP, DH, and SB. The figures shown were made by MS, IA, and VC. We would like to thank the Trinity College Stem-Cell Medicine Senior Postdoctoral Researcher Fund and the Wellcome Trust/MRC Cambridge Stem Cell Intitute for funding SB. We thank the Wellcome Trust 4-Year PhD Programme in Stem Cell Biology and Medicine for funding MS. SS is funded by a Joint Research Grant from the Isaac Newton Trust, Wellcome Trust ISSF, University of Cambridge, and the CSCI Wellcome Trust seed fund. Finally, we would like to thank Yi Lei Tan, Edward J.R. Taylor, Prof Ernest D. Laue, and Dr. Steven F. Lee for contributions to an earlier version of this book chapter. We would like to thank Prof Ernest D. Laue and Prof Sir David Klenerman for generous use of the microscopes on which these images were collected.

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . .

241

References 1. Cremer T, Kurz A, Zirbel R, Dietzel S, Rinke B et al (1993) Role of chromosome territories in the functional compartmentalization of the cell nucleus. Cold Spring Harb Symp Quant Biol 58:777–792 2. Branco MR, Pombo A (2006) Intermingling of chromosome territories in interphase suggests role in translocations and transcriptiondependent associations. PLoS Biol 4:e138 3. Stevens TJ, Lando D, Basu S, Atkinson LP, Cao Y et al (2017) 3D structures of individual mammalian genomes studied by single-cell Hi-C. Nature 544:59–64 4. Lando D, Basu S, Stevens TJ, Riddell A, Wohlfahrt KJ et al (2018) Combining fluorescence imaging with Hi-C to study 3D genome architecture of the same single cell. Nat Protoc 13:1034–1061 5. Ma X, Ezer D, Adryan B, Stevens TJ (2018) Canonical and single-cell Hi-C reveal distinct chromatin interaction sub-networks of mammalian transcription factors. Genome Biol 19: 174 6. Lakadamyali M, Cosma MP (2020) Visualizing the genome in high resolution challenges our textbook understanding. Nat Methods 17:371–379 7. Bintu B, Mateo LJ, Su JH, SinnottArmstrong NA, Parker M et al (2018) Super-resolution chromatin tracing reveals domains and cooperative interactions in single cells. Science 362 8. Trzaskoma P, Ruszczycki B, Lee B, Pels KK, Krawczyk K et al (2020) Ultrastructural visualization of 3D chromatin folding using volume electron microscopy and DNA in situ hybridization. Nat Commun 11:2120 9. Dekker J, Rippe K, Dekker M, Kleckner N (2002) Capturing chromosome conformation. Science 295:1306–1311 10. Lieberman-Aiden E, van Berkum NL, Williams L, Imakaev M, Ragoczy T et al (2009) Comprehensive mapping of longrange interactions reveals folding principles of the human genome. Science 326:289–293 11. Rao SS, Huntley MH, Durand NC, Stamenova EK, Bochkov ID et al (2014) A 3D map of the human genome at kilobase resolution reveals principles of chromatin looping. Cell 159:1665–1680 12. Chang LH, Noordermeer D (2020) Of dots and stripes: the morse code of micro-C reveals the ultrastructure of transcriptional and architectural mammalian 3D genome organization. Mol Cell 78:376–378

13. Hsieh TH, Weiner A, Lajoie B, Dekker J, Friedman N, Rando OJ (2015) Mapping nucleosome resolution chromosome folding in yeast by micro-C. Cell 162:108–119 14. Boettiger A, Murphy S (2020) Advances in chromatin imaging at Kilobase-scale resolution. Trends Genet 36:273–287 15. Mir M, Bickmore W, Furlong EEM, Narlikar G (2019) Chromatin topology, condensates and gene regulation: shifting paradigms or just a phase? Development 146 16. Strom AR, Emelyanov AV, Mir M, Fyodorov DV, Darzacq X, Karpen GH (2017) Phase separation drives heterochromatin domain formation. Nature 547:241–245 17. Larson AG, Elnatan D, Keenen MM, Trnka MJ, Johnston JB et al (2017) Liquid droplet formation by HP1alpha suggests a role for phase separation in heterochromatin. Nature 547:236–240 18. Sanulli S, Trnka MJ, Dharmarajan V, Tibble RW, Pascal BD et al (2019) HP1 reshapes nucleosome core to promote phase separation of heterochromatin. Nature 575:390–394 19. Han X, Yu D, Gu R, Jia Y, Wang Q et al (2020) Roles of the BRD4 short isoform in phase separation and active gene transcription. Nat Struct Mol Biol 27:333–341 20. Liu Z, Legant WR, Chen BC, Li L, Grimm JB et al (2014) 3D imaging of Sox2 enhancer clusters in embryonic stem cells. elife 3: e04236 21. Klar TA, Jakobs S, Dyba M, Egner A, Hell SW (2000) Fluorescence microscopy with diffraction resolution barrier broken by stimulated emission. Proc Natl Acad Sci U S A 97: 8206–8210 22. Dreier J, Castello M, Coceano G, Caceres R, Plastino J et al (2019) Smart scanning for low-illumination and fast RESOLFT nanoscopy in vivo. Nat Commun 10:556 23. Hofmann M, Eggeling C, Jakobs S, Hell SW (2005) Breaking the diffraction barrier in fluorescence microscopy at low light intensities by using reversibly photoswitchable proteins. Proc Natl Acad Sci U S A 102: 17565–17569 24. Gustafsson MG (2000) Surpassing the lateral resolution limit by a factor of two using structured illumination microscopy. J Microsc 198:82–87 25. Balzarotti F, Eilers Y, Gwosch KC, Gynna AH, Westphal V et al (2017) Nanometer resolution imaging and tracking of fluorescent

242

Maike Steindel et al.

molecules with minimal photon fluxes. Science 355:606–612 26. Gwosch KC, Pape JK, Balzarotti F, Hoess P, Ellenberg J et al (2020) MINFLUX nanoscopy delivers 3D multicolor nanometer resolution in cells. Nat Methods 17:217–224 27. Rust MJ, Bates M, Zhuang X (2006) Subdiffraction-limit imaging by stochastic optical reconstruction microscopy (STORM). Nat Methods 3:793–795 28. Betzig E, Patterson GH, Sougrat R, Lindwasser OW, Olenych S et al (2006) Imaging intracellular fluorescent proteins at nanometer resolution. Science 313:1642–1645 29. Hess ST, Girirajan TP, Mason MD (2006) Ultra-high resolution imaging by fluorescence photoactivation localization microscopy. Biophys J 91:4258–4272 30. Abbe E (1873) Beitr€age zur Theorie des Mikroskops und der mikroskopischen Wahrnehmung. Arch Mikrosk Anat 9:413–468 31. Gebhardt JC, Suter DM, Roy R, Zhao ZW, Chapman AR et al (2013) Single-molecule imaging of transcription factor binding to DNA in live mammalian cells. Nat Methods 10:421–426 32. Basu S, Shukron O, Ponjavic A, Parruto P, Boucher W et al (2020) Live-cell 3D singlemolecule tracking reveals how NuRD modulates enhancer dynamics. bioRxiv. https://doi. org/10.1101/2020.04.03.003178 33. Carr AR, Ponjavic A, Basu S, McColl J, Santos AM et al (2017) Three-dimensional superresolution in eukaryotic cells using the double-helix point spread function. Biophys J 112:1444–1454 34. Wiese M, Bannister AJ, Basu S, Boucher W, Wohlfahrt K et al (2019) Citrullination of HP1gamma chromodomain affects association with chromatin. Epigenetics Chromatin 12:21 35. Zhang W, Aubert A, Gomez de Segura JM, Karuppasamy M, Basu S et al (2016) The nucleosome remodeling and deacetylase complex NuRD is built from preformed catalytically active sub-modules. J Mol Biol 428: 2931–2942 36. Izeddin I, Recamier V, Bosanac L, Cisse II, Boudarene L et al (2014) Single-molecule tracking in live cells reveals distinct targetsearch strategies of transcription factors in the nucleus. elife 3 37. Chen J, Zhang Z, Li L, Chen BC, Revyakin A et al (2014) Single-molecule dynamics of enhanceosome assembly in embryonic stem cells. Cell 156:1274–1285

38. Zhao ZW, Roy R, Gebhardt JC, Suter DM, Chapman AR, Xie XS (2014) Spatial organization of RNA polymerase II inside a mammalian cell nucleus revealed by reflected lightsheet superresolution microscopy. Proc Natl Acad Sci U S A 111:681–686 39. Cisse II, Izeddin I, Causse SZ, Boudarene L, Senecal A et al (2013) Real-time dynamics of RNA polymerase II clustering in live human cells. Science 341:664–667 40. Germier T, Audibert S, Kocanova S, Lane D, Bystricky K (2018) Real-time imaging of specific genomic loci in eukaryotic cells using the ANCHOR DNA labelling system. Methods 142:16–23 41. Germier T, Kocanova S, Walther N, Bancaud A, Shaban HA et al (2017) Realtime imaging of a single gene reveals transcription-initiated local confinement. Biophys J 113:1383–1394 42. Alexander JM, Guan J, Li B, Maliskova L, Song M et al (2019) Live-cell imaging reveals enhancer-dependent Sox2 transcription in the absence of enhancer proximity. elife 8 43. Gu B, Swigut T, Spencley A, Bauer MR, Chung M et al (2018) Transcription-coupled changes in nuclear mobility of mammalian cis-regulatory elements. Science 359: 1050–1055 44. Eykelenboom JK, Tanaka TU (2020) Zooming in on chromosome dynamics. Cell Cycle 19:1422–1432 45. Dickerson D, Gierlin´ski M, Singh V, Kitamura E, Ball G et al (2016) High resolution imaging reveals heterogeneity in chromatin states between cells that is not inherited through cell division. BMC Cell Biol 17:33 46. Basu S, Needham LM, Lando D, Taylor EJR, Wohlfahrt KJ et al (2018) FRET-enhanced photostability allows improved singlemolecule tracking of proteins and protein complexes in live mammalian cells. Nat Commun 9:2520 47. Hu YS, Zhu Q, Elkins K, Tse K, Li Y et al (2013) Light-sheet Bayesian microscopy enables deep-cell super-resolution imaging of heterochromatin in live human embryonic stem cells. Opt Nanoscopy 2 48. Edelstein A, Amodaj N, Hoover K, Vale R, Stuurman N (2010) Computer control of microscopes using micromanager. Curr Protoc Mol Biol Chapter 14(Unit14):20 49. Schneider CA, Rasband WS, Eliceiri KW (2012) NIH Image to ImageJ: 25 years of image analysis. Nat Methods 9:671–675

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . . 50. Lee SA, Ponjavic A, Siv C, Lee SF, Biteen JS (2016) Nanoscopic cellular imaging: confinement broadens understanding. ACS Nano 10: 8143–8153 51. Palayret M, Armes H, Basu S, Watson AT, Herbert A et al (2015) Virtual-‘light-sheet’ single-molecule localisation microscopy enables quantitative optical sectioning for super-resolution imaging. PLoS One 10: e0125438 52. Yang B, Chen X, Wang Y, Feng S, Pessino V et al (2019) Epi-illumination SPIM for volumetric imaging with high spatial-temporal resolution. Nat Methods 16:501–504 53. Kim J, Wojcik M, Wang Y, Moon S, Zin EA et al (2019) Oblique-plane single-molecule localization microscopy for tissues and small intact animals. Nat Methods 16:853–857 54. Sapoznik E, Chang BJ, Huh J, Ju RJ, Azarova EV et al (2020) A versatile oblique plane microscope for large-scale and highresolution imaging of subcellular dynamics. elife 9 55. Tokunaga M, Imamoto N, Sakata-Sogawa K (2008) Highly inclined thin illumination enables clear single-molecule imaging in cells. Nat Methods 5:159–161 56. Galland R, Grenci G, Aravind A, Viasnoff V, Studer V, Sibarita JB (2015) 3D high- and super-resolution imaging using singleobjective SPIM. Nat Methods 12:641–644 57. Ponjavic A, Ye Y, Laue E, Lee SF, Klenerman D (2018) Sensitive light-sheet microscopy in multiwell plates using an AFM cantilever. Biomed Opt Express 9:5863–5880 58. Chen BC, Legant WR, Wang K, Shao L, Milkie DE et al (2014) Lattice light-sheet microscopy: imaging molecules to embryos at high spatiotemporal resolution. Science 346:1257998 59. Chang BJ, Kittisopikul M, Dean KM, Roudot P, Welf ES, Fiolka R (2019) Universal light-sheet generation with field synthesis. Nat Methods 16:235–238 60. Kwakwa K, Savell A, Davies T, Munro I, Parrinello S et al (2016) easySTORM: a robust, lower-cost approach to localisation and TIRF microscopy. J Biophotonics 9:948–957 61. Deschamps J, Rowald A, Ries J (2016) Efficient homogeneous illumination and optical sectioning for quantitative single-molecule localization microscopy. Opt Express 24: 28080–28090 62. Michalet X, Colyer RA, Scalia G, Ingargiola A, Lin R et al (2013) Development of new photon-counting detectors for singlemolecule fluorescence microscopy. Philos

243

Trans R Soc Lond Ser B Biol Sci 368: 20120035 63. Huang B, Wang W, Bates M, Zhuang X (2008) Three-dimensional super-resolution imaging by stochastic optical reconstruction microscopy. Science 319:810–813 64. Pavani SR, Thompson MA, Biteen JS, Lord SJ, Liu N et al (2009) Three-dimensional, single-molecule fluorescence imaging beyond the diffraction limit by using a double-helix point spread function. Proc Natl Acad Sci U S A 106:2995–2999 65. Shechtman Y, Weiss LE, Backer AS, Lee MY, Moerner WE (2016) Multicolour localization microscopy by point-spread-function engineering. Nat Photonics 10:590–594 66. Juette MF, Gould TJ, Lessard MD, Mlodzianoski MJ, Nagpure BS et al (2008) Threedimensional sub-100 nm resolution fluorescence microscopy of thick samples. Nat Methods 5:527–529 67. Abrahamsson S, Chen J, Hajj B, Stallinga S, Katsov AY et al (2013) Fast multicolor 3D imaging using aberration-corrected multifocus microscopy. Nat Methods 10:60–63 68. Sims RR, Abdul Rehman S, Lenz MO, Benaissa SI, Bruggeman E et al (2020) Single molecule light field microscopy. Optica 7: 1065–1072 69. Shtengel G, Galbraith JA, Galbraith CG, Lippincott-Schwartz J, Gillette JM et al (2009) Interferometric fluorescent superresolution microscopy resolves 3D cellular ultrastructure. Proc Natl Acad Sci U S A 106:3125–3130 70. Shtengel G, Wang Y, Zhang Z, Goh WI, Hess HF, Kanchanawong P (2014) Imaging cellular ultrastructure by PALM, iPALM, and correlative iPALM-EM. Methods Cell Biol 123: 273–294 71. Deschamps J, Mund M, Ries J (2014) 3D superresolution microscopy by supercritical angle detection. Opt Express 22: 29081–29091 72. Jouchet P, Cabriel C, Bourg N, Bardou M, Pou¨s C et al (2021) Nanometric axial localization of single fluorescent molecules with modulated excitation. Nat Photonics 15: 297–304 73. Cabriel C, Bourg N, Jouchet P, Dupuis G, Leterrier C et al (2019) Combining 3D single molecule localization strategies for reproducible bioimaging. Nat Commun 10:1980 74. Wu B, Piatkevich KD, Lionnet T, Singer RH, Verkhusha VV (2011) Modern fluorescent proteins and imaging technologies to study

244

Maike Steindel et al.

gene expression, nuclear localization, and dynamics. Curr Opin Cell Biol 23:310–317 75. Shaner NC, Lin MZ, McKeown MR, Steinbach PA, Hazelwood KL et al (2008) Improving the photostability of bright monomeric orange and red fluorescent proteins. Nat Methods 5:545–551 76. Nguyen AW, Daugherty PS (2005) Evolutionary optimization of fluorescent proteins for intracellular FRET. Nat Biotechnol 23: 355–360 77. Lippincott-Schwartz J, Altan-Bonnet N, Patterson GH (2003) Photobleaching and photoactivation: following protein dynamics in living cells. Nat Cell Biol:S7–S14 78. Lippincott-Schwartz J, Patterson GH (2009) Photoactivatable fluorescent proteins for diffraction-limited and super-resolution imaging. Trends Cell Biol 19:555–565 79. Patterson GH, Lippincott-Schwartz J (2002) A photoactivatable GFP for selective photolabeling of proteins and cells. Science 297: 1873–1877 80. Subach FV, Patterson GH, Manley S, Gillette JM, Lippincott-Schwartz J, Verkhusha VV (2009) Photoactivatable mCherry for highresolution two-color fluorescence microscopy. Nat Methods 6:153–159 81. McKinney SA, Murphy CS, Hazelwood KL, Davidson MW, Looger LL (2009) A bright and photostable photoconvertible fluorescent protein. Nat Methods 6:131–133 82. Marathe P, Mahadeva Swamy HS, Nair D, Bhattacharyya D (2020) mEosBrite are bright variants of mEos3.2 developed by semirational protein engineering. J Fluoresc 30: 703–715 83. Zhang MS, Chang H, Zhang YD, Yu JW, Wu LJ et al (2012) Rational design of true monomeric and bright photoactivatable fluorescent proteins. Nat Methods 9:727–U297 84. Verkhusha VV, Chudakov DM, Gurskaya NG, Lukyanov S, Lukyanov KA (2004) Common pathway for the red chromophore formation in fluorescent proteins and chromoproteins. Chem Biol 11:845–854 85. Habuchi S, Ando R, Dedecker P, Verheijen W, Mizuno H et al (2005) Reversible singlemolecule photoswitching in the GFP-like fluorescent protein Dronpa. Proc Natl Acad Sci U S A 102:9511–9516 86. Grimm JB, English BP, Chen J, Slaughter JP, Zhang Z et al (2015) A general method to improve fluorophores for live-cell and singlemolecule microscopy. Nat Methods 12: 244–250. 3 p following 50

87. Wombacher R, Heidbreder M, van de Linde S, Sheetz MP, Heilemann M et al (2010) Live-cell super-resolution imaging with trimethoprim conjugates. Nat Methods 7:717–719 88. Wang L, Tran M, D’Este E, Roberti J, Koch B et al (2020) A general strategy to develop cell permeable and fluorogenic probes for multicolour nanoscopy. Nat Chem 12:165–172 89. Niekamp S, Stuurman N, Vale RD (2020) A 6-nm ultra-photostable DNA FluoroCube for fluorescence imaging. Nat Methods 17: 437–441 90. Grimm JB, Brown TA, English BP, Lionnet T, Lavis LD (2017) Synthesis of Janelia Fluor HaloTag and SNAP-Tag ligands and their use in cellular imaging experiments. Methods Mol Biol 1663:179–188 91. Grimm JB, English BP, Choi H, Muthusamy AK, Mehl BP et al (2016) Bright photoactivatable fluorophores for single-molecule imaging. Nat Methods 13:985–988 92. Zheng Q, Ayala AX, Chung I, Weigel AV, Ranjan A et al (2019) Rational design of fluorogenic and spontaneously blinking labels for super-resolution imaging. ACS Cent Sci 5: 1602–1613 93. Speil J, Baumgart E, Siebrasse JP, Veith R, Vinkemeier U, Kubitscheck U (2011) Activated STAT1 transcription factors conduct distinct saltatory movements in the cell nucleus. Biophys J 101:2592–2600 94. Traenkle B, Rothbauer U (2017) Under the microscope: single-domain antibodies for live-cell imaging and super-resolution microscopy. Front Immunol 8:1030 95. Nikic I, Plass T, Schraidt O, Szymanski J, Briggs JA et al (2014) Minimal tags for rapid dual-color live-cell labeling and superresolution microscopy. Angew Chem Int Ed Engl 53:2245–2249 96. Dempsey GT, Vaughan JC, Chen KH, Bates M, Zhuang X (2011) Evaluation of fluorophores for optimal performance in localization-based super-resolution imaging. Nat Methods 8:1027–1036 97. Wang S, Moffitt JR, Dempsey GT, Xie XS, Zhuang X (2014) Characterization and development of photoactivatable fluorescent proteins for single-molecule-based superresolution imaging. Proc Natl Acad Sci U S A 111:8452–8457 98. Di Matteo M, Matrai J, Belay E, Firdissa T, Vandendriessche T, Chuah MK (2012) PiggyBac toolbox. Methods Mol Biol 859: 241–254

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . . 99. Sander JD, Joung JK (2014) CRISPR-Cas systems for editing, regulating and targeting genomes. Nat Biotechnol 32:347–355 100. Mao G, Marotta F, Yu J, Zhou L, Yu Y et al (2008) DNA context and promoter activity affect gene expression in lentiviral vectors. Acta Biomed 79:192–196 101. Balinovic A, Albrecht D, Endesfelder U (2019) Spectrally red-shifted fluorescent fiducial markers for optimal drift correction in localization microscopy. J Phys D Appl Phys 52:204002 102. Cordes T, Vogelsang J, Tinnefeld P (2009) On the mechanism of Trolox as antiblinking and antibleaching reagent. J Am Chem Soc 131:5018–5019 103. Rasnik I, McKinney SA, Ha T (2006) Nonblinking and long-lasting single-molecule fluorescence imaging. Nat Methods 3: 891–893 104. Mazza D, Abernathy A, Golob N, Morisaki T, McNally JG (2012) A benchmark for chromatin binding measurements in live cells. Nucleic Acids Res 40:e119 105. Elf J, Li GW, Xie XS (2007) Probing transcription factor dynamics at the singlemolecule level in a living cell. Science 316: 1191–1194 106. Etheridge TJ, Boulineau RL, Herbert A, Watson AT, Daigaku Y et al (2014) Quantification of DNA-associated proteins inside eukaryotic cells using single-molecule localization microscopy. Nucleic Acids Res 42:e146 107. Gomez-Garcia PA, Portillo-Ledesma S, Neguembor MV, Pesaresi M, Oweis W et al (2021) Mesoscale modeling and singlenucleosome tracking reveal remodeling of clutch folding and dynamics in stem cell differentiation. Cell Rep 34:108614 108. Endesfelder U, van de Linde S, Wolter S, Sauer M, Heilemann M (2010) Subdiffraction-resolution fluorescence microscopy of myosin-actin motility. ChemPhysChem 11:836–840 109. Sage D, Pham TA, Babcock H, Lukes T, Pengo T et al (2019) Super-resolution fight club: assessment of 2D and 3D singlemolecule localization microscopy software. Nat Methods 16:387–395 110. Durisic N, Laparra-Cuervo L, SandovalAlvarez A, Borbely JS, Lakadamyali M (2014) Single-molecule evaluation of fluorescent protein photoactivation efficiency using an in vivo nanotemplate. Nat Methods 11: 156–162

245

111. Ulbrich MH, Isacoff EY (2007) Subunit counting in membrane-bound proteins. Nat Methods 4:319–321 112. Annibale P, Scarselli M, Kodiyan A, Radenovic A (2010) Photoactivatable fluorescent protein mEos2 displays repeated photoactivation after a long-lived dark state in the red photoconverted form. J Phys Chem Lett 1: 1506–1510 113. Lando D, Endesfelder U, Berger H, Subramanian L, Dunne PD et al (2012) Quantitative single-molecule microscopy reveals that CENP-A(Cnp1) deposition occurs during G2 in fission yeast. Open Biol 2:120078 114. Annibale P, Vanni S, Scarselli M, Rothlisberger U, Radenovic A (2011) Identification of clustering artifacts in photoactivated localization microscopy. Nat Methods 8:527–528 115. Ricci MA, Manzo C, Garcia-Parajo MF, Lakadamyali M, Cosma MP (2015) Chromatin fibers are formed by heterogeneous groups of nucleosomes in vivo. Cell 160:1145–1158 116. Strom AR, Brangwynne CP (2019) The liquid nucleome—phase transitions in the nucleus at a glance. J Cell Sci 132 117. Khater IM, Nabi IR, Hamarneh G (2020) A review of super-resolution single-molecule localization microscopy cluster analysis and quantification methods. Patterns (N Y) 1: 100038 118. Nieves DJ, Owen DM (2020) Analysis methods for interrogating spatial organisation of single molecule localisation microscopy data. Int J Biochem Cell Biol 123:105749 119. Vanleeuwen JMJ, Groeneveld J, Deboer J (1959) New method for the calculation of the pair correlation function. 1. Physica 25: 792–808 120. Sengupta P, Jovanovic-Talisman T, Lippincott-Schwartz J (2013) Quantifying spatial organization in point-localization superresolution images using pair correlation analysis. Nat Protoc 8:345–354 121. Ripley BD (1977) Modeling spatial patterns. J Roy Stat Soc B Met 39:172–212 122. Cover TM, Hart PE (1967) Nearest neighbor pattern classification. IEEE T Inform Theory 13:21 123. Malkusch S, Muranyi W, Muller B, Krausslich HG, Heilemann M (2013) Single-molecule coordinate-based analysis of the morphology of HIV-1 assembly sites with near-molecular spatial resolution. Histochem Cell Biol 139: 173–179

246

Maike Steindel et al.

124. Gan JH, Tao YF (2015) DBSCAN revisited: mis-claim, un-fixability, and approximation. In: Sigmod’15: Proceedings of the 2015 ACM Sigmod International Conference on Management of Data, pp 519–530 125. Ester M, Kriegel H-P, Sander J, Xu X (1996) A density-based algorithm for discovering clusters in large spatial databases with noise. In: Proceedings of the second international conference on knowledge discovery and data mining. AAAI Press, Portland, Oregon, pp 226–231 126. Mazouchi A, Milstein JN (2016) Fast Optimized Cluster Algorithm for Localizations (FOCAL): a spatial cluster analysis for superresolved microscopy. Bioinformatics 32: 747–754 127. Nino DF, Djayakarsana D, Milstein JN (2020) FOCAL3D: A 3-dimensional clustering package for single-molecule localization microscopy. PLoS Comput Biol 16 128. Nan X, Collisson EA, Lewis S, Huang J, Tamguney TM et al (2013) Single-molecule superresolution imaging allows quantitative analysis of RAF multimer formation and signaling. Proc Natl Acad Sci U S A 110: 18519–18524 129. Endesfelder U, Finan K, Holden SJ, Cook PR, Kapanidis AN, Heilemann M (2013) Multiscale spatial organization of RNA polymerase in Escherichia coli. Biophys J 105: 172–181 130. Rubin-Delanchy P, Burn GL, Griffie J, Williamson DJ, Heard NA et al (2015) Bayesian cluster identification in single-molecule localization microscopy data. Nat Methods 12: 1072–1076 131. Griffie J, Shlomovich L, Williamson DJ, Shannon M, Aaron J et al (2017) 3D Bayesian cluster analysis of super-resolution data reveals LAT recruitment to the T cell synapse. Sci Rep-UK 7 132. Levet F, Hosy E, Kechkar A, Butler C, Beghin A et al (2015) SR-Tesseler: a method to segment and quantify localization-based superresolution microscopy data. Nat Methods 12:1065–1071 133. Andronov L, Orlov I, Lutz Y, Vonesch JL, Klaholz BP (2016) ClusterViSu, a method for clustering of protein complexes by Voronoi tessellation in super-resolution microscopy. Sci Rep 6:24084 134. Andronov L, Michalon J, Ouararhni K, Orlov I, Hamiche A et al (2018) 3DClusterViSu: 3D clustering analysis of superresolution microscopy data by 3D Voronoi tessellations. Bioinformatics 34:3004–3012

135. Okabe A (2000) Spatial tessellations: concepts and applications of Voronoi diagrams. Wiley, Chichester; New York. xii, 671 p. 136. Williamson DJ, Burn GL, Simoncelli S, Griffie J, Peters R et al (2020) Machine learning for cluster analysis of localization microscopy data. Nat Commun 11:1493 137. Khater IM, Meng F, Nabi IR, Hamarneh G (2019) Identification of caveolin-1 domain signatures via machine learning and graphlet analysis of single-molecule super-resolution data. Bioinformatics 35:3468–3475 138. Khater IM, Meng F, Wong TH, Nabi IR, Hamarneh G (2018) Super resolution network analysis defines the molecular architecture of caveolae and caveolin-1 scaffolds. Sci Rep 8:9009 139. Wong TH, Khater IM, Joshi B, Shahsavari M, Hamarneh G, Nabi IR (2021) Single molecule network analysis identifies structural changes to caveolae and scaffolds due to mutation of the caveolin-1 scaffolding domain. Sci Rep 11:7810 140. Jimenez A, Friedl K, Leterrier C (2020) About samples, giving examples: optimized single molecule localization microscopy. Methods 174:100–114 141. Baumgart F, Arnold AM, Leskovar K, Staszek K, Folser M et al (2016) Varying label density allows artifact-free analysis of membrane-protein nanoclusters. Nat Methods 13:661–664 142. Backlund MP, Joyner R, Moerner WE (2015) Chromosomal locus tracking with proper accounting of static and dynamic errors. Phys Rev E Stat Nonlinear Soft Matter Phys 91:062716 143. Amitai A, Seeber A, Gasser SM, Holcman D (2017) Visualization of chromatin decompaction and break site extrusion as predicted by statistical polymer modeling of single-locus trajectories. Cell Rep 18:1200–1214 144. Hauer MH, Seeber A, Singh V, Thierry R, Sack R et al (2017) Histone degradation in response to DNA damage enhances chromatin dynamics and recombination rates. Nat Struct Mol Biol 24:99–107 145. Saxton MJ, Jacobson K (1997) Single-particle tracking: applications to membrane dynamics. Annu Rev Bioph Biom 26:373–399 146. Holcman D, Hoze N, Schuss Z (2015) Analysis and interpretation of superresolution single-particle trajectories. Biophys J 109: 1761–1771 147. Shukron O, Seeber A, Amitai A, Holcman D (2019) Advances using single-particle

Studying the Dynamics of Chromatin-Binding Proteins in Mammalian Cells. . . trajectories to reconstruct chromatin organization and dynamics. Trends Genet 35: 685–705 148. Amitai A, Holcman D (2017) Polymer physics of nuclear organization and function. Phys Rep 678:1–83 149. Reingruber J, Holcman D (2011) Transcription factor search for a DNA promoter in a three-state model. Phys Rev E Stat Nonlinear Soft Matter Phys 84:020901 150. Hoze N, Holcman D (2015) Recovering a stochastic process from super-resolution noisy ensembles of single-particle trajectories. Phys Rev E Stat Nonlinear Soft Matter Phys 92:052109 151. Persson F, Linden M, Unoson C, Elf J (2013) Extracting intracellular diffusive states and transition rates from single-molecule tracking data. Nat Methods 10:265–269 152. Monnier N, Barry Z, Park HY, Su KC, Katz Z et al (2015) Inferring transient particle transport dynamics in live cells. Nat Methods 12: 838–840 153. Hoze N, Holcman D (2017) Statistical methods for large ensembles of super-resolution stochastic single particle trajectories in cell biology. Annu Rev Stat Appl 4:189–223 154. Weimann L, Ganzinger KA, McColl J, Irvine KL, Davis SJ et al (2013) A quantitative comparison of single-dye tracking analysis tools using Monte Carlo simulations. PLoS One 8 155. Schwarz G (1978) Estimating the dimension of a model. Ann Stat 6:461–464 156. Lagerholm BC, Andrade DM, Clausen MP, Eggeling C (2017) Convergence of lateral dynamic measurements in the plasma membrane of live cells from single particle tracking and STED-FCS. J Phys D Appl Phys 50: 063001 157. Huang F, Schwartz SL, Byars JM, Lidke KA (2011) Simultaneous multiple-emitter fitting for single molecule super-resolution imaging. Biomed Opt Express 2:1377–1393 158. Holden SJ, Uphoff S, Kapanidis AN (2011) DAOSTORM: an algorithm for high- density super-resolution microscopy. Nat Methods 8: 279–280 159. Babcock H, Sigal YM, Zhuang X (2012) A high-density 3D localization algorithm for stochastic optical reconstruction microscopy. Opt Nanoscopy 1

247

160. Ovesny M, Krizek P, Borkovec J, Svindrych Z, Hagen GM (2014) ThunderSTORM: a comprehensive ImageJ plug-in for PALM and STORM data analysis and super-resolution imaging. Bioinformatics 30:2389–2390 161. Mukamel EA, Babcock H, Zhuang X (2012) Statistical deconvolution for superresolution fluorescence microscopy. Biophys J 102: 2391–2400 162. Min J, Vonesch C, Kirshner H, Carlini L, Olivier N et al (2014) FALCON: fast and unbiased reconstruction of high-density superresolution microscopy data. Sci Rep 4:4577 163. Babcock HP, Moffitt JR, Cao Y, Zhuang X (2013) Fast compressed sensing analysis for super-resolution imaging using L1-homotopy. Opt Express 21: 28583–28596 164. Cox S, Rosten E, Monypenny J, JovanovicTalisman T, Burnette DT et al (2011) Bayesian localization microscopy reveals nanoscale podosome dynamics. Nat Methods 9: 195–200 165. Hu YS, Nan X, Sengupta P, LippincottSchwartz J, Cang H (2013) Accelerating 3B single-molecule super-resolution microscopy with cloud computing. Nat Methods 10: 96–97 166. Marsh RJ, Pfisterer K, Bennett P, Hirvonen LM, Gautel M et al (2018) Artifact-free highdensity localization microscopy analysis. Nat Methods 15:689–692 167. Nehme E, Freedman D, Gordon R, Ferdman B, Weiss LE et al (2020) DeepSTORM3D: dense 3D localization microscopy and PSF design by deep learning. Nat Methods 17:734–740 168. Li B, Ponjavic A, Chen WH, Hopkins L, Hughes C et al (2021) Single-molecule light-sheet microscopy with local nanopipette delivery. Anal Chem 93:4092–4099 169. Festuccia N, Owens N, Papadopoulou T, Gonzalez I, Tachtsidi A et al (2019) Transcription factor activity and nucleosome organization in mitosis. Genome Res 29:250–260 170. Stanly TA, Fritzsche M, Banerji S, Garcia E, Bernardino de la Serna J et al (2016) Critical importance of appropriate fixation conditions for faithful imaging of receptor microclusters. Biol Open 5:1343–1350

Chapter 17 The End Restraint Method for Mechanically Perturbing Nucleic Acids In Silico Jack W. Shepherd and Mark C. Leake Abstract Far from being a passive information store, the genome is a mechanically dynamic and diverse system in which torsion and tension fluctuate and combine to determine structure and help regulate gene expression. Much of this mechanical perturbation is due to molecular machines such as topoisomerases which must stretch and twist DNA as part of various functions including DNA repair and replication. While the broadscale mechanical response of nucleic acids to tension and torsion is well characterized, detail at the single base pair level is beyond the limits of even super-resolution imaging. Here, we present a straightforward, flexible, and extensible umbrella-sampling protocol to twist and stretch nucleic acids in silico using the popular biomolecular simulation package Amber—though the principles we describe are applicable also to other packages such as GROMACS. We discuss how to set up the simulation system, decide force fields and solvation models, and equilibrate. We then introduce the torsionally constrained stretching protocol, and finally we present some analysis techniques we have used to characterize structural motif formation. Rather than defining forces or fictional pseudoatoms, we instead define a fixed translation of specified atoms between each umbrella-sampling step, which allows comparison with experiment without needing to estimate applied forces by simply using the fractional end-to-end displacement as a comparison metric. We hope that this easy-to-implement solution will be valuable for interrogating optical and magnetic tweezers data on nucleic acids at base pair resolution. Key words Molecular dynamics, DNA topology, DNA stretching, Supercoiling, AMBER, Mechanical perturbation

1

Introduction

1.1 DNA Mechanics In Vivo

In the living cell, DNA is a dynamic and constantly modulated system which interacts with itself as well as with other agents such as molecular machines and transcription factors. Regulating this are two primary mechanisms—sequence recognition and changes in structural conformation. In the former case, specific nucleotide sequences are recognized by transcription proteins as the start of a gene which is then read until the so-called stop codon is reached. However, whether this is possible is regulated by the overall

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_17, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

249

250

Jack W. Shepherd and Mark C. Leake

configuration of the genome; genes can be switched on and off either by physicochemical means such as methylation or by burying the gene in the middle of the nucleus such that it cannot be read. Indeed, these dual-regulation mechanisms may explain the longstanding mystery of why multiple copies of the same gene exist in multiple parts of the genome. Conversely, it has been shown that under mechanical stress, some transcription factor-binding sites form a stable conformation that may be used to promote transcription factor recognition even in mechanically perturbed conditions. It is therefore vital to understand the effect of torsion and tension on the structure of DNA, be it in the live cell [1], in vitro, or computationally [2], to elucidate genome expression. Moreover, as synthetic DNA and protein-based molecular machines are developed, knowing how DNA functions as a material rather than as a biological agent will become of some importance. 1.2 Experimental Methods

DNA’s mechanical properties have been of interest ever since its structure was definitively determined by Rosalind Franklin and colleagues. Initially, these took the form of fiber diffraction experiments in which purified DNA was semidehydrated, and imaged using X-ray diffraction, which led to several structures of DNA and DNA-ligand systems being elucidated [3]. However, the constraints imposed by the sample preparation and the static nature of the experimental technique meant that possible experiments were limited to static structure determination rather than time-resolved imaging of ongoing biomolecular processes. In the 1980s however, two types of “tweezers” were developed using single-molecule biophysics techniques [4], which revolutionized the study of DNA as a material and led to greater insights in the physics of life at the single-molecule level [5]. The first type, known as magnetic tweezers, consists of a permanent or electromagnet on the outside of a sample. DNA is then tethered to a magnetic microbead at one end and at the other to the sample surface, or to another bead which has been immobilized on the coverslip. The magnet can then be moved, or the magnetic field rotated using electromagnetic coils [6], to exert a force on the bead, or twisted to exert torque. While this goes on, the DNA may be imaged using intercalating fluorescent dyes [7], or the position of the bead (and hence absolute force applied) can be tracked using bright-field imaging and centroid detection. The second tweezers is known as the optical or laser tweezers and uses typically a latex or polystyrene bead which is trapped using a near-infrared laser. The laser is focused by the objective lens, and at the focus the radiation pressure of the light diffracting through the bead is sufficient to exert a Hookean force toward the center of the focus. Again, DNA can be tethered to the bead and at the other end to the coverslip or to a second bead which may also be trapped

The End Restraint Method for Mechanically Perturbing Nucleic Acids In Silico

251

in a so-called dumbbell assay. The DNA may be imaged with fluorescence microscopy, or its structure inferred from forceextension behavior, or both. These tweezers have led to many important discoveries including the force-extension behavior of DNA [8], of supercoiled DNA [9], and of DNA/protein systems [10]. DNA has been hyperstretched [11], and overwound to form Pauling-like conformations [12]. When using fluorescence microscopy, color-changing dyes may be used to measure the proportion of DNA which is singlestranded (melted) or remaining in the familiar double-helix structure [13, 14]. Fluorescence imaging can enable insight into DNA architecture and dynamics by imaging proteins that bind to DNA using live-cell microscopy, such as in studying DNA replication [15, 16], remodeling [17], and DNA topology manipulation [18], but can be used in vitro in multispectral imaging, for example to DNA repair proteins translocating along DNA that has been tagged using a different color dye [19], or it can be combined with other quantitative techniques such as fluorescence polarization which can infer the orientations of the base pairs in which the dye is intercalated [20–22]. These approaches, along with highprecision structural imaging methods such as atomic force microscopy (AFM) [23], have demonstrated that DNA is a complex and dynamic system where multiple competing motifs can coexist depending on the stress and torsion applied, and through inference the likely structures have been determined. While these approaches are extremely powerful and have elucidated some of DNA’s mechanical properties, there are some levels of detail which are unreachable through optical microscopy means. Thanks to the diffraction limit, individual base pairs cannot be imaged directly. Instead, we turn to fluorescent reporter molecules which are excited by a given wavelength of laser light. We can then localize these molecules individually as long as they are sufficiently separated in time or space. The localization precision possible through this method is generally of the order 20–30 nm [7], well below the diffraction limit but still representing almost 100 base pairs, while the dye itself is generally bound between 4 and 6 base pairs [24]. There is therefore a clear mismatch between the length scale that the molecule is actually reporting on and the length scale it is taken to be reporting on. Moreover, either binding a dye in groove or intercalating it between base pairs has significant structural implications; in the case of a commonly used intercalating dye YOYO-1 for example the contour length can increase by up to 50% with associated underwinding and torsional stiffening of the double helix [24, 25], while the specific mechanical stabilization or destabilization effects of an intercalator are unknown. It is therefore something of an open question as to how close to the unlabeled structure a fluorescently labeled tract of DNA will be.

252

Jack W. Shepherd and Mark C. Leake

1.3 Simulation Approaches

Fortunately, as computers have grown more powerful and generalpurpose graphics processing units (GPGPUs) have developed, larger time- and length-scale simulations have become possible at both an atomistic level (using for example Amber [26]) and a coarse-grained level (for example using oxDNA [27]). Various strategies have been devised to mechanically perturb DNA systems using each regime which we summarize here. First, it is possible to define a so-called repulsion plane in a simulation cell. This is a fictional barrier which repels a given subset of simulated atoms using a desired potential, generally either linear or harmonic. This can be used to stretch nucleic acids by placing the plane in the center of the contour and repelling the two ends. Alternatively, fictional “pseudoatoms” can be defined. These do not participate in the overall simulation but are placed away from the nucleic acid fragment toward the edge of the simulation cell and are used as anchors, held fixed in place while specified atoms are harmonically attracted toward them with a potential well. Both of these approaches have been successfully used to show DNA-melting behavior [28], but neither is well suited to torsional constraint. The repulsion plane can be modified however using an additional coordinate, usually defined as the twist angle between the first and last base pairs [29]. An additional force-field parameter can then be specified which restrains this angle to be in a specified range, essentially exerting torsional control, though this has the disadvantage of being difficult to implement. Other simulations have been run using an infinite length of DNA—that is, a fragment which joins itself at the edge of the periodic simulation box. Here, the DNA is torsionally constrained thanks to it being bound to itself, and base pairs were deleted to retain the same supercoiling density in fewer base pairs, in essence increasing the overtwisting [30]. This produced Pauling-like DNA (P-DNA) in which the base pairs were exposed and the backbone tightly wound around itself with a helical pitch of just a few base pairs. However, the utility of this approach beyond simulations of those types is limited. Also using the DNA-bound-to-itself method of torsional constraint are minicircle-based simulations [31]. These are circular DNA constructs, the supercoiling density of which can be modified prior to simulation. Using these, Pyne, Noy, and colleagues demonstrated the rich landscape of supercoiled minicircle DNA using both atomic-level simulations and AFM [32], demonstrating the power of combined theory and experiment studies. Minicircles, however, are not suitable for a single-strand stretching experiment.

1.4 Advantages of Our Approach

Here we present our protocol for stretching DNA at a fixed supercoiling density by making use of atomic restraints on both terminal base pairs of a nucleic acid fragment and discrete movements of one end of the fragment to produce a stretch [33] (see a flowchart

The End Restraint Method for Mechanically Perturbing Nucleic Acids In Silico

253

Fig. 1 (a) Flowchart of the end-restraints method for nucleic acid perturbation. The steps inside the dashed box must be repeated to build up a large overall stretch in many small increments. (b) Each stretch even seeds the next, and by seeding step n + 1 before step n is finished, the simulation can be efficiently parallelized across umbrella-sampling windows

representation of this in Fig. 1). The advantages of this, we believe, are as follows: first, it is an exceptionally conceptually and practically straightforward method which is readily implemented in any simulation suite; second, because the end-to-end length of the nucleic acid is fixed, comparison with optical or magnetic tweezer assays is rendered more straightforward than force-matching approaches; third, because the method makes use of umbrella sampling, it is possible to parallelize across sampling windows to reduce overall runtime (see Fig. 1); fourth, it is trivial to perform these simulations with any desired supercoiling density, up to any desired overstretch; fifth, it is possible to run any desired energy minimization routine at the start of each umbrella-sampling window, so for especially complex systems a full simulated annealing stage could be implemented if needed; sixth, by lowering the strength of the restraint potential on the terminal base pairs some movement of the ends is allowed— the end-to-end length can then be used as the coordinate for the weighted histogram method of free energy calculation (WHAM) and the change in free energy due to conformation change can be extracted; and seventh, this approach is trivially compatible with ligand binding. To our knowledge, our protocol is unique in this list of advantages. We note however that there are some disadvantages of our approach. Most notably, each perturbation is a discrete single step, after which the structure is energetically minimized. We do not therefore access the transition between sub-conformations; we simply see them in their formed states. Because we artificially keep the terminal base pairs in a canonical conformation, care must be taken to use a sufficiently long fragment so that edge effects are not a

254

Jack W. Shepherd and Mark C. Leake

concern at the center, where analysis will then be focused. We also do not generally access the forces applied due to these restraints (it is possible to extract these values but in our experience they are invariably too noisy to be useful) so comparison with experiment at that level is largely impossible. However, for linear stretching simulations with torsional constraint these drawbacks are surmountable.

2

Materials

2.1 Molecular Dynamics Software

Here we focus on atomistic molecular dynamics using Amber and its associated force fields. For these structurally perturbed simulations it is necessary to choose a modern well-tested force field and we in general use either the OL15 or the bsc1 nucleic acid force field [34]. For solvation we use the Generalized Born model when working with implicit solvation and a TIP3P [35] model for explicit solvation, with appropriate ion parameters. Systems are prepared with the Amber utilities nab and leap.

2.2 Visualization Software

For most atomistic operations we recommend Visual Molecular Dynamics (VMD) [36], a freely available biomolecule visualization suite which can be run using input scripts or through the visual interface. However, for presenting coarse-grained structures (e.g., with a whole nucleotide represented by a single flat cuboid) two alternatives are Ovito [37] and Chimera [38], both free and scriptable as well as operating in GUI mode.

2.3

Analysis Utilities

The free AmberTools suite is usually sufficient for the analysis that is needed for these linear stretching simulations as writhe and plectoneme formation is prohibited by the simulation setup. Using AmberTools’ cpptraj [39] we can find canonical and noncanonical hydrogen bonds, while with sander we can evaluate the total interaction energy between nucleotides. More difficult is conformational analysis as the resulting highly mechanically perturbed structures lead to anomalous results (for example, if a base pair melts the assumed vector linking the nucleotides ceases to be meaningful). However, for unmelted structures it is possible to use Curves+ [40]. Other physical properties such as the persistence length can be estimated using the SerraNA [41] utility which uses the molecular flexibility and correlations between fluctuations in different base pairs to estimate mechanical parameters. However, we note that the underlying assumption in SerraNA is that the ends of the DNA are free to fluctuate so the output for an end-constrained simulation will be only a rough estimate.

2.4 Bespoke Scripts and Input Files

Various scripts will need to be written to perform the functions described below. Our own pipeline is composed of bash, awk, and Python 3 scripts chained together in one bash script file which also

The End Restraint Method for Mechanically Perturbing Nucleic Acids In Silico

255

calls the necessary tleap and cpptraj instances during the simulation. In general we recommend developing these scripts in pure Python which will be most portable and maintainable. Input files to Amber will need to be written on a case-by-case basis though we give some examples below.

3

Methods

3.1 Initial System Preparation

1. Create a linear piece of DNA using the fd_helix and putpdb functions in nab. This by default lays the helical axis along the simulation z-axis which will be useful for implicit solvent simulations. For ease we will call the resulting pdb structure nab_structure.pdb. 2. If applying no additional supercoiling, prepare input files for Amber’s pmemd.cuda. To do this, it is best to write a reusable tleap script. In the case of an implicit solvent simulation this can be as simple as the following: a = loadpdb(“nab_structure.pdb”) saveamberparm a nab_structure.prmtop nab_structure.inpcrd

Save this as tleap.in and run as follows: $ tleap –f tleap.in. This will create two files—nab_structure.prmtop and nab_structure.inpcrd—which can be given to pmemd.cuda for simulation. If using TIP3P water, the input file is only slightly more complex. For an octahedral simulation cell with a size 12
12
50 Å: a = loadpdb(“nab_structure.pdb”) solvatebox a TIP3PBOX {12 12 50} addions a Na+ 0 saveamberparm a nab_structure.prmtop nab_structure.inpcrd

In order to avoid having to resolvate the structure at any point during the stretching simulation, it is imperative to select a box large enough that the stretched DNA will still fit inside it comfortably, with ~12 Å clearance between the box edge and any atom in the DNA. Here note also that we have included a line which adds Na + ions so that the overall system is electrically neutral. This is essentially proceeding at 0 M salt concentration—hardly biologically relevant. See Notes for adding extra ions.

256

Jack W. Shepherd and Mark C. Leake

3.2

Applying Torsion

Application of torsion is of course nothing more than rotating the terminal base pair around the helical axis to a desired supercoiling density. This is most easily achieved immediately after preparation of the nab_structure.pdb file when the helical axis lies along z. Read in the pdb file to your scripting language of choice and simply apply the 3D rotation matrix to rotate by some step around z. For large rotations, it will be necessary to perform energy minimization between smaller rotation events to build up a large supercoiling density. Having done this, you will need to prepare the pmemd.cuda input files as described in 3.1 above.

3.3

Equilibration

Equilibration is reasonably straightforward and proceeds according to several well-known protocols. For the purposes of the input file examples below we will assume that the simulation duplex is 20 base pairs long, therefore containing 40 nucleotides in total. 1. If using explicit solvent, begin by performing steepest descent and conjugate gradient on the water and ions only, keeping the DNA fixed in place. Example input file: &ctrl imin=1, ncyc=1000, maxcyc=2000, ntb=1, igb=0, cut=12, ntr=1 / DNA restraints 50.0 RES 1 40 END END

2. Perform steepest descent and conjugate gradient minimization on the middle of the NA fragment, keeping end base pairs only fixed. We find that this step is crucial—in general we use up to hundreds of thousands of minimization steps to ensure that the structure is as close as possible to its ground state. Example input file: &ctrl imin=1, ncyc=10000, maxcyc=20000, ntb=1, igb=0, cut=12 / DNA end restraints 50.0 RES 1,20,21,40 END END

The End Restraint Method for Mechanically Perturbing Nucleic Acids In Silico

257

3. If using explicit solvent, perform a 100 ps simulation in the NPT ensemble, heating the system from 0 to 300 K linearly through the simulation. However, if using implicit solvent this step is unnecessary for simple simulations though it would likely be useful for complex nucleic acid/ligand systems. 3.4 Running the Umbrella-Sampling Simulations

Finally, the time has come to perform the umbrella-sampling simulations themselves. We proceed according to the flowchart in Fig. 1. 1. First, simulate the unstretched conformation making sure to keep the end restraints in place. This can be done with the user’s preferred parameters, though we recommend keeping the time step small (2 fs) and using the maximum velocity keyword (maxvel) to limit the maximum atomic velocity to ca. 20. Physically the need for this arises thanks to the hard restraints on the end base pairs—if a water atom gets too close to this hard wall, it will be repelled at a very high force which may lead to the simulation blowing up energetically. The need for this is lessened by using softer restraints on the end base pairs, though for consistency if nothing else it is worthwhile deciding on a maximum velocity and sticking with it. Ensure to use the positional restraints on the terminal base pairs. 2. While the molecular dynamics goes on, monitor the various physical parameters that indicate that it has reached thermodynamic equilibrium—for example density and total energy. At the beginning of the simulation the system will still be equilibrating so there will be a number of frames that need to be discarded or at least ignored for the purposes of analysis. 3. When thermodynamic equilibrium appears to have been reached, you may then take a snapshot from the system by copying the current restart (.rst) file to a new directory. This will be the starting conformation for the simulation of the first stretching event. 4. How to proceed here depends on whether you are using explicit or implicit solvation. In the table below we summarize how to approach each. Implicit solvation Here, the easiest way to proceed is to use cpptraj to create a new pdb file, for example: $ cpptraj –p dna.prmtop –y dna.rst – x dna.pdb l This pdb file will still have the helical axis along z l

Explicit solvation If explicit solvation is being used, two methods may be employed. Most straightforwardly, a similar .pdb file may be created using cpptraj as described for the implicit case l

(continued)

258

Jack W. Shepherd and Mark C. Leake

Implicit solvation l It is therefore trivial to read in the pdb file, and subtract 1 from the z component of position for the terminal nucleotides by using the residue number l This can be done in a variety of ways but the most straightforward may be using awk: {if($5 ¼¼ 1 || $5 ¼¼ 40){ printf(“ATOM %6i %4 s %3 s %5i % 8.3f%8.3f%8.3f 1.00 0.00\n”, $2, $3,\ $4, $5, $6, $7, $8–1); else{ print $0 } } l If saved as, e.g., stretch.awk, this script can then be run as follows:

$ awk --f stretch. awk < dna. pdb > dna_stretched.pdb l This dna_stretched.pdb can then be used as the starting pdb for minimization and simulation as described above.

Explicit solvation l awk may then also be used to move the terminal base pair by checking the residue number l However, in our experience the pdb files often fail to encode the box information properly l It is therefore necessary to redefine the box using tleap when it is loaded in to create the topology and coordinate files using the setbox command l Alternatively, the .rst file can be modified directly if it is saved as a plain-text netCDF file l The easiest approach to this is to read the file using Python, and then move coordinates for the correct residues, though the netCDF does not specify residue numbers— these can be taken from the topology file l For explicit solvent simulations, especially if using a truncated octahedral box rather than a cuboid, it is not guaranteed for the DNA helical axis to lie along z however l It is therefore in general necessary to specify a translation axis for the terminal base pair, which may be worked out from atomic coordinates of the nucleic acid

5. Having stretched the nucleic acid by 1 Å, this is used as the starting structure for minimization and simulation as described above. 3.5 Analysis Strategies

1. Finally, the overall simulation can be analyzed either in individual umbrella-sampling windows or by joining the windows into one simulation—making sure to discard the start of each window which may not be in equilibrium. 2. Use Curves+ and SerraNA to estimate physical properties, though note that in particularly perturbed states these estimates will likely be unreliable. 3. In general there are two useful properties for observing motif formation: hydrogen bonding and stacking energy. 4. To calculate the hydrogen bonding, we generally use cpptraj with a 120 angle cutoff and a 3.5 Å distance cutoff.

The End Restraint Method for Mechanically Perturbing Nucleic Acids In Silico

259

5. This produces a large table of hydrogen-bonding interactions, and using Python or another scripting language, the canonical and noncanonical hydrogen bonds can be extracted. 6. For the stacking energy, the AmberTools utility sander is used. However, sander’s esander routine does not decompose energies by residue-residue interaction so the user must do this manually. 7. The easiest way—though time consuming to do—is simply to iterate through each possible pair of nucleotides, strip out all others, and run esander. This comes at some computational cost, however. For canonical base stacking between only nearest neighbors, it is relatively cheap—N operations where N is the number of nucleotides in the system. However, for noncanonical stacking, in principle this would take (N  1)(N  2)/ 2 operations, quickly becoming prohibitively expensive (for 40 nucleotides this is 741 separate calculations). However, we find that with a linear fragment of DNA being stretched we can in practice limit the operations to next-nearest neighbors, including cross-strand neighbors, i.e., the three nearest neighbors on the opposite strand plus the two next-nearest neighbors for a total of 5 per nucleotide, though these are degenerate—each calculation will be used for two nucleotides. The number of calculations is therefore 5 N/2, a significant saving. 8. Alternatively, it is possible to find the pairwise energies by using other postprocessing analysis software—GROMACS does have this functionality but it comes with a necessary file conversion step. 9. Either way, by calculating these energies, multiple files will be generated. From these, for each nucleotide, it is possible to extract the nonbonded energy (principally van der Waals interactions) and sum all contributions for each nucleotide. The total canonical or noncanonical stacking is thus known. 10. If a high noncanonical stacking energy or number of hydrogen bonds is seen to persist over many frames it is indicative of a semi-stable structure being formed. The easiest way to see the trends in these physical quantities is by plotting a heatmap in which the x-axis is base pair, the y-axis is the time axis, and the color corresponds to the physical quantity under examination. 11. All apparent structures should then be interrogated with visualization software and the precise details obtained from the hydrogen bond and energy analysis.

260

Jack W. Shepherd and Mark C. Leake

12. Finally, if soft restraints are used on the terminal base pairs, the free energy may be estimated using WHAM. In order to do this, the integration coordinate must first be determined from the trajectory files. In general we have had good success using the end-to-end length of the DNA as the coordinate, found trivially from the atomic xyz positions.

4

Notes Ionic Strength 1. If using implicit solvation, set the salt concentration using the saltcon keyword in the pmemd.cuda input file. This is measured in moles/L so saltcon ¼ 0.2 would generate Debye-Huckel screening parameters to approximate 200 mM salt.

2. If using explicit solvation, the user must solvate the system and note the volume of the bounding box. From this one can work out the required number of salt ions needed and add them manually after adding enough ions to electrically neutralize the system. 3. Imagining that the number of ions needed was 20, we could proceed as follows: a = loadpdb(“dna.pdb”) solvatebox a TIP3PBOX {15 15 50} addions a Na+ 0 addions a Na+ 20 addions a Cl- 20 saveamberparm a dna.prmtop dna.inpcrd

4. Of course this process is also scriptable, and includes calling tleap once to generate the solvated system, then saving as a pdb file, extracting the volume from the automatically generated leap.log logfile, and then loading the solvated system back into tleap to add the ions before saving the topology and coordinate files for simulation.

Acknowledgments Thank you for helpful discussions with Agnes Noy, Matt Probert, and Robert Greenall at the University of York. The work was supported by the Leverhulme Trust (RPG-2019-156, RPG-2017-340) and EPSRC (EP/N027639/1).

The End Restraint Method for Mechanically Perturbing Nucleic Acids In Silico

261

References 1. Lenn T, Leake MC (2016) Single-molecule studies of the dynamics and interactions of bacterial OXPHOS complexes. Biochim Biophys Acta Bioenerg 1857(3):224–231. https://doi.org/10.1016/j.bbabio.2015. 10.008 2. Wollman AJM, Miller H, Zhou Z, Leake MC (2015) Probing DNA interactions with proteins using a single-molecule toolbox: inside the cell, in a test tube and in a computer. Biochem Soc Trans 43(2):139–145. https://doi. org/10.1042/BST20140253 3. Stubbs G (1999) Developments in fiber diffraction. Curr Opin Struct Biol 9(5): 615–619. https://doi.org/10.1016/S0959440X(99)00014-7 4. Miller H, Zhou Z, Shepherd J, Wollman AJM, Leake MC (2018) Single-molecule techniques in biophysics: a review of the progress in methods and applications. Rep Prog Phys 81(2): 24601. https://doi.org/10.1088/13616633/aa8a02 5. Leake MC (2013) The physics of life: one molecule at a time. Philos Trans R Soc B Biol Sci 368(1611):20120248. https://doi.org/10. 1098/rstb.2012.0248 6. Zhou Z, Miller H, Wollman AJM, Leake MC (2015) Developing a new biophysical tool to combine magneto-optical tweezers with superresolution fluorescence microscopy. PhotoDermatology 2(3):758–772. https://doi.org/ 10.3390/photonics2030758 7. Miller H, Zhou Z, Wollman AJM, Leake MC (2015) Superresolution imaging of single DNA molecules using stochastic photoblinking of minor groove and intercalating dyes. Methods 88:81–88. https://doi.org/10. 1016/j.ymeth.2015.01.010 8. Smith SB, Cui Y, Bustamante C (1996) Overstretching B-DNA: the elastic response of individual double-stranded and single-stranded DNA molecules. Science (80-) 271(5250): 795–799. https://doi.org/10.1126/science. 271.5250.795 9. Strick TR, Allemand JF, Bensimon D, Croquette V (1998) Behavior of supercoiled DNA. Biophys J 74(4):2016–2028. https:// doi.org/10.1016/S0006-3495(98)77908-1 10. Lipfert J, Kerssemakers JWJ, Jager T, Dekker NH (2010) Magnetic torque tweezers: measuring torsional stiffness in DNA and RecADNA filaments. Nat Methods 7(12): 977–980. https://doi.org/10.1038/nmeth. 1520

11. Schakenraad K et al (2017) Hyperstretching DNA. Nat Commun 8(1):1–7. https://doi. org/10.1038/s41467-017-02396-1 12. Allemand JF, Bensimon D, Lavery R, Croquette V (1998) Stretched and overwound DNA forms a Pauling-like structure with exposed bases. Proc Natl Acad Sci U S A 95(24):14152–14157. https://doi.org/10. 1073/pnas.95.24.14152 13. King GA, Gross P, Bockelmann U, Modesti M, Wuite GJL, Peterman EJG (2013) Revealing the competition between peeled ssDNA, melting bubbles, and S-DNA during DNA overstretching using fluorescence microscopy. Proc Natl Acad Sci U S A 110(10): 3859–3864. https://doi.org/10.1073/pnas. 1213676110 14. Miller HL et al (2020) Biophysical characterisation of DNA origami nanostructures reveals inaccessibility to intercalation binding sites. Nanotechnology 31(23):235605. https://doi. org/10.1088/1361-6528/ab7a2b 15. Reyes-Lamothe R, Sherratt DJ, Leake MC (2010) Stoichiometry and architecture of active DNA replication machinery in Escherichia coli. Science (80-) 328(5977):498–501. https://doi.org/10.1126/science.1185757 16. Syeda AH et al (2019) Single-molecule live cell imaging of Rep reveals the dynamic interplay between an accessory replicative helicase and the replisome. Nucleic Acids Res 47(12): 6287–6298. https://doi.org/10.1093/nar/ gkz298 17. Badrinarayanan A, Reyes-Lamothe R, Uphoff S, Leake MC, Sherratt DJ (2012) In vivo architecture and action of bacterial structural maintenance of chromosome proteins. Science (80-) 338(6106):528–531. https:// doi.org/10.1126/science.1227126 18. Stracy M et al (2019) Single-molecule imaging of DNA gyrase activity in living Escherichia coli. Nucleic Acids Res 47(1):210–220. https:// doi.org/10.1093/nar/gky1143 19. Heller I et al (2013) STED nanoscopy combined with optical tweezers reveals protein dynamics on densely covered DNA. Nat Methods 10(9):910–916. https://doi.org/10. 1038/nmeth.2599 20. Backer AS, Biebricher AS, King GA, Wuite GJL, Heller I, Peterman EJG (2019) Singlemolecule polarization microscopy of DNA intercalators sheds light on the structure of S-DNA. Sci Adv. https://doi.org/10.1126/ sciadv.aav1083

262

Jack W. Shepherd and Mark C. Leake

21. Backer AS et al (2021) Elucidating the role of topological constraint on the structure of overstretched DNA using fluorescence olarization microscopy. J Phys Chem B 125(30): 8351–8361. https://doi.org/10.1021/ACS. JPCB.1C02708 22. Shepherd JW, Payne-Dwyer AL, Lee J-E, Syeda A, Leake MC (2021) Combining single-molecule super-resolved localization microscopy with fluorescence polarization imaging to study cellular processes. J Phys Photon 3(3):34010. https://doi.org/10.1088/ 2515-7647/ac015d 23. Yoshua SB, Watson GD, Howard JAL, VelascoBerrelleza V, Leake MC, Noy A (2021) Integration host factor bends and bridges DNA in a multiplicity of binding modes with varying specificity. Nucleic Acids Res. https://doi. org/10.1093/nar/gkab641 24. Gu¨nther K, Mertig M, Seidel R (2010) Mechanical and structural properties of YOYO-1 complexed DNA. Nucleic Acids Res 38(19):6526–6532. https://doi.org/10. 1093/nar/gkq434 25. Kundukad B, Yan J, Doyle PS (2014) Effect of YOYO-1 on the mechanical properties of DNA. Soft Matter 10(48):9721–9728. https://doi.org/10.1039/c4sm02025a 26. Case DA et al. (2018) AMBER17. 2017. San Fr. Univ. Calif. 27. Ouldridge TE, Louis AA, Doye JPK (2011) Structural, mechanical, and thermodynamic properties of a coarse-grained DNA model. J Chem Phys 134(8). https://doi.org/10. 1063/1.3552946 28. Marin-Gonzalez A, Vilhena JG, Perez R, Moreno-Herrero F (2017) Understanding the mechanical response of double-stranded DNA and RNA under constant stretching forces using all-atom molecular dynamics. Proc Natl Acad Sci U S A 114(27):7049–7054. https:// doi.org/10.1073/pnas.1705642114 29. Olson WK, Zhurkin VB (2000) Modeling DNA deformations. Curr Opin Struct Biol 10(3):286–297. https://doi.org/10.1016/ S0959-440X(00)00086-5 30. Randall GL, Zechiedrich L, Pettitt BM (2009) In the absence of writhe, DNA relieves torsional stress with localized, sequencedependent structural failure to preserve B-form. Nucleic Acids Res 37(16): 5568–5577. https://doi.org/10.1093/nar/ gkp556 31. Mitchell JS, Harris SA (2013) Thermodynamics of writhe in DNA minicircles from

molecular dynamics simulations. Phys Rev Lett 110(14):148105. https://doi.org/10. 1103/PhysRevLett.110.148105 32. Pyne ALB et al (2021) Base-pair resolution analysis of the effect of supercoiling on DNA flexibility and major groove recognition by triplex-forming oligonucleotides. Nat Commun 12(1):1–12. https://doi.org/10.1038/ s41467-021-21243-y 33. Shepherd JW, Greenall RJ, Probert MIJ, Noy A, Leake MC (2020) The emergence of sequence-dependent structural motifs in stretched, torsionally constrained DNA. Nucleic Acids Res. https://doi.org/10.1093/ nar/gkz1227 34. Galindo-Murillo R et al (2016) Assessing the current state of Amber force field modifications for DNA. J Chem Theory Comput 12(8): 4114–4127. https://doi.org/10.1021/acs. jctc.6b00186 35. Price DJ, Brooks CL (2004) A modified TIP3P water potential for simulation with Ewald summation. J Chem Phys 121(20):10096–10103. https://doi.org/10.1063/1.1808117 36. Humphrey W, Dalke A, Schulten K (1996) VMD: Visual molecular dynamics. J Mol Graph 14(1):33–38. https://doi.org/10. 1016/0263-7855(96)00018-5 37. Stukowski A (2010) Visualization and analysis of atomistic simulation data with OVITO-the open visualization tool. Model Simul Mater Sci Eng 18(1):15012. https://doi.org/10.1088/ 0965-0393/18/1/015012 38. Pettersen EF et al (2004) UCSF Chimera: a visualization system for exploratory research and analysis. J Comput Chem 25(13): 1605–1612. https://doi.org/10.1002/jcc. 20084 39. Roe DR, Cheatham TE (2013) PTRAJ and CPPTRAJ: software for processing and analysis of molecular dynamics trajectory data. J Chem Theory Comput 9(7):3084–3095. https:// doi.org/10.1021/ct400341p 40. Lavery R, Moakher M, Maddocks JH, Petkeviciute D, Zakrzewska K (2009) Conformational analysis of nucleic acids revisited: Curves+. Nucleic Acids Res 37(17): 5917–5929. https://doi.org/10.1093/nar/ gkp608 41. Velasco-Berrelleza V, Matthew B, Shepherd JW, Leake MC, Golestanian R, Noy A (2020) SerraNA: a program to determine nucleic acids elasticity from simulation data. Phys Chem Chem Phys 22(34):19254–19266. https://doi. org/10.1039/D0CP02713H

Chapter 18 Visualizing the Replisome, Chromosome Breaks, and Replication Restart in Bacillus subtilis Hannah Gaimster, Charles Winterhalter, Alan Koh, and Heath Murray Abstract Research over the last two decades has revealed that bacterial genomes are highly organized and that bacteria have sophisticated mechanisms in place to ensure their correct replication and segregation into progeny cells. Here we discuss techniques that can be used with live bacterial cells to analyze DNA replisome dynamics, double-strand chromosome breaks, and restart of repaired replication forks. Key words DNA replication, Replisome, DNA damage, Double-strand DNA breaks, SOS response, Replication restart

1

Introduction Faithful transmission of genetic material is required for the viability of all cells. DNA replication must be initiated only once per cell cycle to ensure precise coordination of genome duplication with cell growth and division. In most bacteria chromosomes are replicated bidirectionally from a single genetic locus termed oriC (replication origin of the chromosome; Fig. 1a). The master bacterial initiation protein DnaA binds to specific sites within oriC and assembles into an ATP-dependent filament that unwinds the DNA duplex. In the model organism Bacillus subtilis, initiation of chromosome replication requires two more essential proteins, DnaD and DnaB, both of which have DNA-remodeling activities [1] and bind to the origin prior to helicase loading [2]. DnaD interacts with both DnaA and DNA and allows for subsequent recruitment of DnaB. In turn, DnaB recruits the hexameric helicase DnaC and its dedicated chaperone DnaI [3]. Together this machinery loads one replicative helicase around each single DNA strand of the open complex. Helicase activity further enlarges the open complex, providing space for loading the DNA replication elongation machinery, which in B. subtilis is composed of two DNA

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_18, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

263

264

Hannah Gaimster et al. A: Chromosomal replication initiation DnaC DnaI DnaA

DnaA

DnaA

oriC

DnaG DnaN

DnaB

DnaD

+

B: DNA damage response

C: Replication Restart

DNA damage Replisome PriA Single strand nick

SsbA

DSB DnaD DnaB Activation of SOS response

DNA repair genes upregulated

SOS response mediators: RecA and LexA

DnaC DnaI

Cell division inhibition: YneA

DnaC-DnaI

Fig. 1 An overview of the processes of chromosomal replication initiation, DNA damage response, and replication restart in B. subtilis. Proteins and genomic regions, which are used later in this chapter, are indicated in bold and color

polymerases (PolC and DnaE), the clamp loader complex (τ3δδ0 ), and the processivity clamp (β2). Each “replisome” proceeds along one chromosome arm to synthesize a new copy of the genome, converging roughly equidistant from oriC [4].

Visualizing the Replisome, Chromosome Breaks, and Replication Restart. . .

265

Replisomes encounter many obstacles as they proceed along the genome, some of which can arrest the replication machinery, leading to its dissociation from the DNA template. Following repair of the replication fork, helicase must be reloaded in a DNA structure-specific (but sequence independent) manner by the ubiquitous restart protein PriA. Here PriA must ensure that only a single helicase is loaded on the lagging strand. Bacterial cells, with their relatively small genomes and rapid generation times, provide excellent model systems with which to understand the fundamental principles of DNA replication and replisome restart, which are often conserved throughout the kingdom of life. Researchers have developed tools for visualizing and analyzing these processes within single living bacterial cells. Below we describe methods to investigate DNA replication and to detect both double-stranded DNA breaks (DSBs) and replisome restart in the bacterium B. subtilis, a fast-growing and genetically tractable model organism. A reporter used frequently to study DNA and protein localization in living cells is the green fluorescent protein (GFP), along with its derivatives that have distinct excitation and emission properties to allow multicolor imaging (cyan ¼ CFP, yellow ¼ YFP, red ¼ RFP). These reporters can be genetically fused to a protein of interest to assess its position within the cell. Additionally, because neither fixation nor permeabilization are required for this approach, the signal can be detected in living cells over time to capture their dynamic behavior. To visualize the entire bacterial chromosome, fluorescent proteins (FPs) can be fused to the highly abundant nucleoid-associated histone-like protein HBsu (Fig. 2) [5]. HBsu is a homodimeric DNA-binding protein which belongs to the HU family of nucleoidassociated proteins (NAPs). HBsu engages DNA nonspecifically and HBsu-GFP is observed to bind throughout the nucleoid [6]. Figure 2 shows that in a wild-type strain growing exponentially, chromosomes are well segregated and present in each cell. To visualize the position of the replisome within a cell, several of the DNA replication proteins that compose this machinery can be fused to FPs. For example, Reyes-Lamothe et al. successfully tagged a wide range of replication proteins with the monomeric yellow fluorescent protein mYPet and used single-molecule imaging approaches to determine their stoichiometry within the replisome [7]. In our lab, we often visualize the replisome by tagging the sliding clamp DnaN, which accumulates into “clamp zones” of ~100 copies on the lagging strand behind active replication forks [8]. Figure 2 shows that in a wild-type strain, exponentially growing B. subtilis cells display 1-2 DnaN-GFP foci per cell. To visualize the physical duplication of the chromosome during DNA replication, individual genetic loci can be targeted by FP-labeled site-specific DNA-binding proteins. The most

266

Hannah Gaimster et al.

Green: GFP

Merge

GFP-DnaN

Hbsu-GFP

Red: membrane dye

Fig. 2 Visualizing the bacterial chromosome (Hbsu-GFP) and the replisome (GFP-DnaN) in B. subtilis. Strains were grown at 30  C until mid-exponential phase. Cell membranes were stained with Nile red dye to distinguish individual cells. Bulk chromosome morphology was visualized using the nucleoid-associated DNA-binding protein HBsu-GFP and the replisome was visualized using GFP-DnaN

commonly used systems in B. subtilis are based on the repressors TetR and LacI binding tetO (19bp) and lacO (21bp) sites, respectively (Fig. 3). Here arrays of operator sites are integrated at a specific locus on the genome, which then become bound by their cognate fluorescently tagged repressor. Figure 3 illustrates how this system can be used to observe the replication of one chromosome arm, by labeling both a site near the chromosome origin (oriC) (TetR-YFP binding a tetO array; green dot) and a second site located ~150 kilobases downstream (Lacl-CFP binding a lacO array; red dot). The time between duplication of the two loci can be used to estimate the rate of replisome movement along the chromosome. Furthermore, by counting the number of origins within each cell, it is possible to determine the frequency of DNA replication initiation. Chromosomes must be fully replicated for the genetic information to be equally segregated into daughter cells prior to cytokinesis. One of the chromosome lesions capable of impeding complete DNA replication is a DSB. These breaks are an especially hazardous form of DNA damage as unrepaired or erroneously repaired DSBs can ultimately cause genome rearrangements, loss of genetic information, mutations, or cell death [9]. It is thought that one of the most common causes of DSBs is when the replisome encounters a nick in the DNA template [10] (Fig. 1b).

Visualizing the Replisome, Chromosome Breaks, and Replication Restart. . . (i) HindIII

Random NNNNNNNNNN

AATTGTGAGCGGATAACAATT 5'

N10

lacO

N10

lacO

N10

N 10

lacO

(v) N10

lacO

267

3'

XbaI

or 5'

HindIII

N10

tetO

N 10

tetO

N10

tetO

N10

tetO

N10

ge ne

Cloning of tetO array and tetR-gfp with their respective flanking genes into the plasmid for stable integration into the chromosome by double crossover

ne ge

B

C tetR-gfp integration plasmid

(vi)

A

te tO

gfp R-

ne ge

ar ra y

te t

tetO array integration plasmid

XbaI

In a cell having recently undergone cell division, there is a single chromosome containing one origin and one distal region.

TCCCTATCAGTGATAGAGA

(ii)

3'

gene D B

After initiation of DNA replication occurs at the origin, there are two origins and one distal region.

C

D A

B. subtilis chromosome

(vii) TetR-GFP

(iii)

tet

O

a rr ay

tetR-gfp

Visualization of TetR-GFP binding to tetO array by epifluorescence microscopy

(iv) 345˚ Distal marker

B. subtilis chromosome

359˚ Origin marker

360˚/0˚ Origin

As replication proceeds along the chromosome arm, the distal region is duplicated.

(viii)

The cell division machinery localizes to mid-cell to drive cytokinesis, leading to the formation of two identical daughter cells.

Terminus 180˚

Fig. 3 Cartoon representation showing the construction of reporter systems and an example of their use. (i) Organization of the lacO/tetO operator sequence (up to 240 operator repeats) arranged into arrays that contain 10 base pairs of randomized nucleotide sequence between each operator sequence. (ii) Plasmids containing either the tetO array or the tetR-gfp are stably integrated into the chromosome by double crossover. (iii) Successful construction allows visualization of the tetO-proximal locus bound by TetR-GFP using epifluorescence microscopy with an example of how this approach can be used (iv). Strains were grown at 30  C in minimal growth media to avoid overlapping rounds of DNA replication; (v) shows a newborn cell that recently underwent cell division with the origin marked with a green dot and the distal region with a red dot. (vi) A cell has initiated DNA replication and segregated the duplicated origin regions. (vii) DNA replication and segregation proceed normally and the distal region is copied. (vii) Successful formation of two identical daughter cells

During vegetative growth in B. subtilis, DSBs must be repaired via homologous recombination (HR) to allow completion of chromosome replication [11]. HR uses an identical DNA template to prime DNA synthesis and repair the lesion. Briefly, the process

268

Hannah Gaimster et al.

involves the following steps. First, recognition and processing of the DSB generate a 30 single-stranded DNA (ssDNA) tail [12]. Second, the recombinase RecA is loaded onto the ssDNA and uses the substrate to identify a homologous DNA molecule. Third, RecA promotes strand invasion via pairing of the 30 ssDNA tail to form a displacement loop [13]. Fourth, PriA-dependent helicase loading is required to promote replisome assembly and DNA synthesis [14]. Fifth, the newly synthesized strand is ligated to a parental strand to remove any remaining nicks. Sixth, crossover junctions are resolved by endonucleases to complete the repair of the DSB. A well-studied and important DNA damage repair pathway that allows bacteria to sense DSBs and react accordingly is the widely conserved SOS response (Fig. 1b). In B. subtilis, the highly conserved RecA and LexA proteins are central to the regulation of the SOS transcriptional response [15]. RecA filaments interacting with ssDNA promote the auto cleavage of LexA, a DNA-binding repressor that prevents transcription of 63 genes across 26 operons in the SOS regulon [16]. Expression of these SOS genes upregulates a wide variety of DNA repair pathways (such as excision repair, translesion DNA synthesis, and HR) and cell cycle checkpoints [17]. FPs can be used to detect DSBs in individual living cells, both directly and indirectly. Indirect detection of DSBs can be achieved by fusion of gfp to the yneA promoter, which is negatively regulated by LexA. YneA is upregulated during the SOS response and acts to inhibit cell division, allowing time to complete chromosome segregation following DSB repair [18]. Gozzi et al. constructed a fluorescent transcriptional reporter, PyneA-gfp, to monitor SOS induction in B. subtilis during biofilm formation. A fluorescent signal was observed throughout cells in aged biofilms, indicating that yneA expression levels were high at this time [19]. This reporter can also be used to assess SOS induction in planktonic B. subtilis cultures (Subheading 3.2, Fig. 4). When cells are exposed to DNA-damaging agents such as the reactive oxygen species (H2O2) or antibiotics (zeocin), PyneA expression levels increase indicating DSB formation. This approach, however, does not provide information on how many DSBs are present or how they are localized in the cell. In E. coli, an elegant approach has been developed to directly localize DSBs using the fluorescently tagged Gam protein (from bacteriophage Mu), an ortholog of eukaryotic and bacterial Ku [20, 21]. Biochemically, Gam is a highly specific double-stranded DNA end-binding protein [22, 23] and thus can be used to specifically detect chromosomal DSBs to help determine their locations in the genome. In B. subtilis, addition of a DNA-damaging agent such as zeocin results in the appearance of Gam-GFP foci (Subheading 3.2, Fig. 4). The regularly spaced distribution of Gam-GFP

Visualizing the Replisome, Chromosome Breaks, and Replication Restart. . .

Green: GFP

Merge

Zeocin

+ IPTG

Pspac gam-GFP

Uninduced

Zeocin

H2O2

PyneA GFP

-

Phase contrast

269

Fig. 4 Visualizing DSBs in B. subtilis. Strains were grown at 30  C until mid-exponential phase with supplements added as indicated. The presence of DSBs when treated with DNA-damaging agents was visualized either indirectly using PyneA GFP or directly using Pspac gam-GFP

Zeocin

+ IPTG

-

Hannah Gaimster et al.

PhsA+1T mNeonGreen-priA

270

Fig. 5 Visualizing restart of repaired replication forks in B. subtilis. Strains were grown at 37  C up to early-log phase, induced with IPTG, and further incubated until mid-exponential phase. The presence of green fluorescent foci is used as a proxy for replication fork collapse, with their numbers increasing when cells are treated with a DNA-damaging agent

observed following zeocin treatment is reminiscent of replisome localization, providing support for DNA replication-dependent models of DNA breakage [20]. In B. subtilis, replisome restart relies on the conserved replication protein PriA, aided by the ssDNA-binding protein SsbA [24]. While PriA-mediated restart is believed to be linked to the structure of repaired replication forks, its precise mode of action to load helicase around the lagging strand is unclear (Fig. 1c). A GFP-PriA fusion can be used to investigate replisome restart. When expressed in exponentially growing cells, GFP-PriA generally localizes as 1–2 foci per cell. In contrast, when DSBs are generated via treatment with the antibiotic zeocin, the number of foci per cell significantly increases (Subheading 3.3, Fig. 5). Below we describe specific techniques to image live cells of B. subtilis, either during unperturbed DNA replication or under stress conditions caused by DSBs. It is important to note that each of the DNA-labeling strategies outlined above have potential shortcomings. First, FP-tags on endogenous proteins always affect their activity to some degree, so although cell growth, chromosome organization, and segregation may appear normal in an otherwise wild-type strain background, combining multiple FP-tagged proteins in one strain or combining FP-tagged proteins with other mutations may result in synthetic phenotypes [25]. Since no single labeling technique is without caveats, it is advisable to always utilize complementary approaches wherever possible.

Visualizing the Replisome, Chromosome Breaks, and Replication Restart. . .

2

271

Materials All solutions are prepared using ultrapure water and analytical grade reagents. All solutions and reagents are prepared and stored at room temperature unless specifically stated. All components are sterilized either by autoclaving at 15 psi for 30 min or for heatlabile solutions by filtering (0.45 μm).

2.1 Growth of B. subtilis.

1. Cultures are grown either in rich medium (Penassay Broth (PAB) or Luria-Bertani (LB)) or chemically defined derivatives from Spizizen minimal medium (SMM: NH4SO4 (0.2% w/v), K2HPO4 (1.4% w/v), KH2PO4 (0.6% w/v), sodium citrate dehydrate (0.1% w/v), MgSO4.7H2O (0.02% w/v)). Use of complex media such as PAB and LB will produce high autofluorescence while SMM-based media have lower autofluorescence levels. There is also a positive correlation between DNA replication initiation and nutrient-mediated growth rates. Therefore, in complex media such as PAB and LB, cells undergo faster growth rate as compared to minimal media supplemented with the various types of carbon sources. Media such as SMM which support slower growth rate will result in cells having fewer origins/chromosomes per cell, therefore making analysis less complicated. These factors need to be taken into consideration and carefully balanced when deciding what media to use and it is likely that optimization will be required for the specific strain and fluorophore being used. 2. SMM should be supplemented with the following on the day the experiment is performed: Fe-NH4-citrate (0.001 mg/mL final, 2 mg/mL stock), MgSO4.7H2O (6 mM, 1 M stock), CaCl2 (0.1 mM final, 0.1 M stock), MnSO4.4H2O (0.13 mM final, 65 mM stock), glutamate (0.1% w/v final, 5% w/v stock), and tryptophan (0.02 mg/mL final, 2 mg/mL stock). Primary carbon sources added are either succinate (2.0% w/v final, 40% w/v stock), glycerol (0.5% w/v final, 50% w/v stock), or glucose (2% w/v final, 40% w/v stock). Casein hydrolysate (200 μg/mL final, 20 mg/mL stock) can be added for faster growth and getting to higher cell density. The carbon source will influence the numbers of origins/ chromosomes per cell with glucose causing the highest origin numbers followed by glycerol and then succinate. Ideally, casein hydrolysate should only be used for overnight cultures as it will affect the number of origins/chromosomes per cell observed in the different carbon sources. 3. Inducers are added as needed: tetracycline (1 μg/mL final, 1 mg/mL stock), anhydrotetracycline (500 ng/mL, 100 μg/ mL stock), or IPTG (0.1 or 1 mM final, 1 M stock).

272

Hannah Gaimster et al.

4. Test tubes or 125 mL conical flasks. 5. Shaking incubator with required temperature setting. 2.2 General Microscopy

1. Prepare ~1.5% agarose gel in the media that was used for the cell culture. Minimize the use of complex media such as PAB, LB, and CH, which produce high autofluorescence while minimal medium has a lower autofluorescence level. Always prepare fresh agarose solution on the day of the experiment, ensure that the agarose is completely dissolved to prevent agarose crystallization, and store the solution at ~60  C to prevent it from solidifying. 2. Multi-spot microscope slide (Fig. 6) or a Gene Frame (Life Technologies, cat no: AB-0578; 17 mm  28 mm) mounted on a clear microscope slide (Fig. 6). 3. Glass coverslip. 4. Surgical scalpel blades. 5. Prepare 70% ethanol. 6. Epifluorescence microscope with appropriate filter sets. 7. METAMORPH software for image capture. 8. ImageJ software for data analysis, can be downloaded at the following URL: http://imagej.nih.gov/ij/.

(I)

top view (1 to 2 strains) 6 mm

3.5 mm

4 mm

side view (width) cover slip

1.5% agarose pad gene Frame (ii) top view (1 to 4 strains)

side view (width) cover slip

air pocket cells side view (length)

cover slip

Fig. 6 Images showing multi-spot microscope slides. Schematic diagrams showing the layout of a gene frame microscope slide for either (i) two or (ii) four different strains

Visualizing the Replisome, Chromosome Breaks, and Replication Restart. . .

3

273

Methods

3.1 Growth Conditions for Visualizing the Nucleoid, Origins, and Replisome in B. subtilis

1. To visualize cells during exponential growth, 2 mL of starter cultures were grown overnight at either 30  C or 37  C in SMM-based medium.

3.2 Growth Conditions for DNA Damage Assays

1. To visualize cells during exponential growth, 2 mL of starter cultures were grown overnight at 30  C in PAB or LB media.

2. Dilute culture 1:100 into freshly prepared SMM supplemented with succinate (2.0%), glycerol (0.5%), or glucose (2.0%). 3. Allow cultures to achieve at least three mass doublings (A600 0.3–0.5) before observation by microscopy (see Subheadings 3.4 and 3.5).

2. Dilute culture 1:100 into freshly prepared PAB or LB media (+ 1 mM IPTG for Pspac Gam-GFP reporter). 3. Allow cultures to grow for 2–3 h. 4. Add DNA-damaging agent (1 mM H2O2 or 100 μg/mL zeocin) and leave for 1 h. 5. Observe cells by microscopy (see Subheadings 3.4 and 3.5).

3.3 Growth Conditions for Replication Restart Assays

1. To visualize cells during exponential growth, 5 mL of starter cultures were grown overnight at 37  C in SMM-based medium supplemented with casein hydrolysate and glycerol as the primary carbon source. 2. Dilute cells 1:100 into freshly prepared SMM-based medium supplemented with glycerol only in a 125 mL conical flask. 3. Allow cultures to grow for 3–4 h. 4. Add 0.1 mM IPTG and further incubate for 60 min. 5. Observe cells by microscopy (see Subheadings 3.4 and 3.5). Zeocin, if required, should be mixed directly with the agarose pad at 100 μg/mL and cells imaged within 30 min.

3.4 Preparation of Multi-Spot Microscope Slides

1. Clean the multi-spot microscope slide with 70% ethanol and ensure that it is dry and free of dust. 2. Pipette 750 μL of molten agarose onto the middle of the slide. Ensure that there is no air bubble in the agarose as this might interfere with how the agarose pad will harden. Ensure that the agarose pad is evenly spread onto the microscope slide to achieve the best imaging for the field of cells. To achieve this, it is recommended that the gene frame be used. 3. Quickly place a standard microscope slide on top of the molten agarose and squeeze any excess agarose out by gently pressing down on the slide. Avoid exposing the agarose pad to the external environment for too long as this will cause excessive drying.

274

Hannah Gaimster et al.

4. Carefully remove the standard microscope slide on top of the multi-spot microscope slide to reveal the agarose. Air-dry to allow the media to absorb into the pad. This normally takes about 5 min. Avoid exposing the agarose pad to the external environment for too long as this will cause excessive drying. 5. Pipette ~1 μL of cell culture onto the agarose pad. Agitate the culture before removing samples for imaging to reduce cells from clumping together. 6. Air-dry to allow the media to absorb into the pad. This normally takes about 5 min. Avoid exposing the agarose pad to the external environment for too long as this will cause excessive drying. 7. Place a clean coverslip on top of the cells. To prevent air bubbles from accumulating under the coverslip, use the edge of the pen and move it firmly from one end of the coverslip to the other. This will force any air pockets between the coverslip and agarose pad out. 8. The microscope sample is now ready for imaging. 3.5 Preparation of Microscope Slide (Gene Frame)

1. Clean the microscope slide with 70% ethanol and ensure that it is dry and free of dust. 2. Carefully remove the plastic cover from the gene frame (i.e., the side with the solid plastic covering the whole gene frame) which will reveal the sticky pad. 3. Attach the gene frame with the sticky pad onto the middle of a standard microscope slide. Use the edge of a pen and move along the frame ensuring that there are no air bubbles between the frame and the slide (note that the upper side of the sticky gene frame will remain covered with a plastic mask). 4. Pipette 500 μL of molten agarose onto the middle of the gene frame. Ensure that the agarose exceeds the volume of the gene frame and that there are no air bubbles in the agarose. 5. Quickly place a standard microscope slide on top of the molten agarose. Squeeze excess agarose out by gently pressing down on the slide. 6. To harden the agarose, place the gene frame horizontally into a petri dish that contains a small ball of wet tissue (to prevent desiccation). Cover the petri dish and store it in the refrigerator at 4  C for at least 30 min. Avoid exposing the agarose pad to the external environment for too long as this will cause excessive drying. 7. To prepare the slides for microscopy, warm the microscope slides by shifting the petri dishes into the desired temperature (depending on experimental requirements) for at least 30 min.

Visualizing the Replisome, Chromosome Breaks, and Replication Restart. . .

275

Reduce exposing the sample and the microscope slide to rapid temperature changes. Try to do all the remaining steps at the desired experimental temperature. 8. Carefully remove the clear microscope slide on top of the gene frame to reveal the agarose. 9. To provide a source of oxygen create air pockets using a scalpel to cut out agar strips (Fig. 6). 10. Pipette ~1 μL of cell culture onto the agarose pad. Agitate the culture before removing samples for imaging to reduce cells from clumping together. 11. Air-dry to allow the media to absorb into the pad. This normally takes about 5 min. Avoid exposing the agarose pad to the external environment for too long as this will cause excessive drying. 12. Remove the mask from the gene frame and firmly attach a clean coverslip, ensuring that no gaps are present. To prevent air bubbles from accumulating under the coverslip, use the edge of the pen and move it firmly from one end of the coverslip to the other. This will force any air pockets between the coverslip and agarose pad out. 13. The microscope sample is now ready for imaging.

4

Notes 1. B. subtilis requires thorough aeration for optimum growth and development; therefore, ensure vigorous shaking of the culture. Normally a >1:20 ratio (culture volume:container volume) is desired. 2. Maintain and grow overnight culture of strains that harbor the large ~150 tetO array with tetracycline (1 μg/mL) (only if the strain is resistant to tetracycline) or anhydrotetracycline (500 ng/mL) and large ~150 lacO array with IPTG (1 mM) to inhibit TetR and LacI binding to its operator sites, respectively. 3. Use fresh colonies streaked onto nutrient agar plates for each experiment. 4. Allow strains that exhibit slower growth rates a longer overnight incubation period to allow the overnight culture to achieve a high density. 5. Prewarm media before use to prevent temperature shock to the bacteria.

276

Hannah Gaimster et al.

6. Care should be taken to prevent B. subtilis from entering the sporulation developmental pathway as this could interfere with the positioning of the chromosome and thus the origin. References 1. Zhang W et al. (2005) The Bacillus subtilis DnaD and DnaB proteins exhibit different DNA remodelling activities. J Mol Biol 351(1):66–75 2. Bailey S, Eliason WK, Steitz TA (2007) Structure of hexameric DnaB helicase and its complex with a domain of DnaG primase. Science 318(5849):459–463 3. Smits WK, Goranov AI, Grossman AD (2010) Ordered association of helicase loader proteins with the Bacillus subtilis origin of replication in vivo. Mol Microbiol 75(2):452–461 4. Mott ML, Berger JM (2007) DNA replication initiation: mechanisms and regulation in bacteria. Nat Rev Microbiol 5:343 5. Ko¨hler P, Marahiel MA (1997) Association of the histone-like protein HBsu with the nucleoid of Bacillus subtilis. J Bacteriol 179(6):2060 6. Marbouty M et al. (2015) Condensin- and replication-mediated bacterial chromosome folding and origin condensation revealed by Hi-C and super-resolution imaging. Mol Cell 59(4):588–602 7. Reyes-Lamothe R, Sherratt DJ, Leake MC (2010) Stoichiometry and architecture of active DNA replication machinery in Escherichia coli. Science 328(5977):498–501 8. Su’etsugu M, Errington J (2011) The replicase sliding clamp dynamically accumulates behind progressing replication forks in Bacillus subtilis cells. Mol Cell 41(6):720–732 9. Badrinarayanan A, Le TBK, Laub MT (2015) Rapid pairing and resegregation of distant homologous loci enables double-strand break repair in bacteria. J Cell Biol 210(3):385–400 10. Kuzminov A (2001) Single-strand interruptions in replicating chromosomes cause double-strand breaks. Proc Natl Acad Sci 98(15):8241 11. Lenhart JS et al. (2012) DNA Repair and Genome Maintenance in 12. Yeeles JTP, Dillingham MS (2010) The processing of double-stranded DNA breaks for recombinational repair by helicase–nuclease complexes. 9(3):276–285 13. Lovett CM, Roberts JW (1985) Purification of a RecA protein analogue from Bacillus subtilis. J Biol Chem 260(6):3305–3313

14. Marsin S et al. (2001) Early steps of Bacillus subtilis primosome assembly. J Biol Chem 276(49):45818–45825 15. Cheo DL, Bayles KW, Yasbin RE (1991) Cloning and characterization of DNA damageinducible promoter regions from Bacillus subtilis. J Bacteriol 173(5):1696–1703 16. Groban ES et al. (2005) Binding of the Bacillus subtilis LexA protein to the SOS operator. Nucleic Acids Res 33(19):6287–6295 17. Goranov AI et al. (2006) Characterization of the global transcriptional responses to different types of DNA damage and disruption of replication in Bacillus subtilis. J Bacteriol 188(15): 5595–5605 18. Kawai Y, Moriya S, Ogasawara N (2003) Identification of a protein, YneA, responsible for cell division suppression during the SOS response in Bacillus subtilis. Mol Microbiol 47(4): 1113–1122 19. Gozzi K et al. (2017) Bacillus subtilis utilizes the DNA damage response to manage multicellular development. NPJ Biofilms Microbio 3:8–8 20. Shee C et al. (2013) Engineered proteins detect spontaneous DNA breakage in human and bacterial cells. eLife 2:e01222 21. d’Adda di Fagagna F et al. (2003) A DNA damage checkpoint response in telomereinitiated senescence. Nature 426(6963): 194–198 22. Williams JG, Radding CM (1981) Partial purification and properties of an exonuclease inhibitor induced by bacteriophage Mu-1. J Virol 39(2):548 23. Abraham ZH, Symonds N (1990) Purification of overexpressed gam gene protein from bacteriophage Mu by denaturation-renaturation techniques and a study of its DNA-binding properties. Biochem J 269(3):679–684 24. Michel B, Sinha AK, Leach DRF (2018) Replication Fork breakage and restart in Escherichia coli. Microbiol Mol Biol Rev: MMBR 82(3): e00013–e00018 25. Gruber S, Errington J (2009) Recruitment of condensin to replication origin regions by ParB/SpoOJ promotes chromosome segregation in B. subtilis. Cell 137(4):685–696

Chapter 19 High-Throughput Imaging of Bacillus subtilis Paula Montero Llopis, Ryan Stephansky, and Xindan Wang Abstract Bacillus subtilis is a widely used model bacterium to study cellular processes and development. The availability of an arrayed mutant library gave us the opportunity to cytologically analyze every mutant and screen for new genes involved in cell shape determination, cell division, and chromosome segregation. Here we describe a high-throughput method to image arrayed B. subtilis mutant libraries using wide-field fluorescence microscopy. We provide a detailed description of growing the arrayed strain collection, preparing slides containing agarose pedestals, setting up the microscopy procedure, acquiring images, and analyzing the images. Key words Bacillus subtilis, High-throughput imaging, Automated imaging, Arrayed mutant library, Fluorescence microscopy

1

Introduction Bacillus subtilis is widely used as a model organism to study processes important for bacteria growth and development. This bacterium is extremely genetically tractable due to its natural competence, which is a physiological state that allows bacteria to uptake extracellular DNA and recombine to their own genome [1]. Natural competence makes it easy to obtain genetically modified strains, and enables high-throughput transformations to add desired traits to existing arrayed mutant collections [2]. While genetic screens such as Tn-Seq [3] are powerful tools to identify essential genes in certain strain backgrounds or growth conditions, they rely on strong growth defects. Cytological screens like the one described here make it possible to discover new factors that are important for cellular processes that may be missed in other screens because of the lack of strong growth defects. Additionally, cytological screens can be used to determine the spatial localization of fluorescent protein fusions in single cells [4]. Furthermore, cytological screening of arrayed mutant libraries can be streamlined by systematic quantification of fluorescence images. Thanks to the

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_19, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

277

278

Paula Montero Llopis et al.

Frozen Arrayed Strain Collection

Innoculate 96-deep well plates O/N 30oC 900rpm

Dilute and exponential growth 37oC 900rpm

Spot on agarose pedestals

Fig. 1 Overview of the experimental procedure

inherent quantitative nature of fluorescence images, the cytological characteristics of each mutant can be determined by quantifying different aspects of the cells, including cell length, width and area, nucleoid length, width and area, and number of nucleoids per cell. Thus, these quantifications enable the characterization of every mutant and their classification according to the phenotype they display [5–7]. For example, when disrupted, genes involved in cell division generate long cells and mini-cells [8]. When a gene of unknown function displays such phenotypes, this gene would most likely be involved in some aspect of cell division. Therefore, systematic analysis and classification are key in assigning a function to an uncharacterized gene or inferring the pathway the gene is involved in. In this chapter, we describe a detailed method to perform high-throughput imaging and analyses of B. subtilis from an arrayed know-out strain collection in which the parental strain expresses HbsU-mGFPmut3 [9] fusion protein and cytoplasmic mCherry [10] to visualize and quantify chromosome and cell dimensions, respectively (Fig. 1).

2

Materials The arrayed strain collection imaged here is derived from the B. subtilis knockout strain collection constructed by Koo et al. [2]. We first extracted the genomic DNA from this strain collection using Promega Wizard SV 96 DNA purification system (Cat# A2370). Then we performed high-throughput transformation of the genomic DNA into a strain expressing HbsU-mGFPmut3 fusion protein and cytoplasmic mCherry. This chapter describes the procedure to image this new arrayed mutant library. We note that this chapter builds on our previous chapter in which we describe tools and protocols to grow and image B. subtilis [11]. Here we include quantitative high-throughput tools to characterize B. subtilis mutants.

High-Throughput Imaging of Bacillus subtilis

279

Media components are prepared in Milli-Q water (ddH2O) unless otherwise stated. Where indicated, liquid media are sterilized by autoclaving for 30 minutes (min) in liquid cycle, or by filtering through a 0.22 μm syringe-driven filter or a bottle-top filter. Glassware is sterilized by autoclaving for 30 min in a dry cycle. Sterile media components and glassware are stored at room temperature unless otherwise specified. 2.1 Media and Equipment for the Growth of B. Subtilis

1. Defined rich casein hydrolysate medium (CH medium) [12] component CHI+II: 10 mg/mL casein hydrolysate (Neogen #7229A), 4.7 mg/mL L-glutamate sodium salt monohydrate, 3.2 mg/mL L-asparagine monohydrate, 2.5 mg/mL L-alanine, 2.72 mg/mL potassium phosphate monobasic anhydrous (KH2PO4), 2.68 mg/mL ammonium chloride (NH4Cl), 1.1 mg/mL sodium sulfate (Na2SO4), 1 mg/mL ammonium nitrate (NH4NO3), and 0.01 mg/mL ferric chloride 6-hydrate (FeCl3·6H2O); autoclaved. 2. CH medium component CHIII: 0.66 mg/mL calcium chloride dihydrate. (CaCl2·2H2O), and 1.21 mg/mL magnesium sulfate anhydrous (MgSO4); autoclaved. 3. CH medium component CHIV: 1.67 mg/mL manganese sulfate monohydrate (MnSO4·H2O); autoclaved. 4. CH medium component CHV: 2 mg/mL L-Tryptophan; filtered and distributed in 10 mL aliquots in 15 mL conical tubes. Store at 4
C. 5. Round-bottom 96-deep-well plates (Fisher 40002-014). 6. 96-pin replicator tool. 7. Handheld Vacuum Aspirator with 8-Channel Adapter (Argos Technologies HandE-Vac 13050-32). 8. Temperature-controlled plate shaker (Multitron Infors HT). 9. Breathable cover for 96-deep-well plates (Denville Scientific AeraSeal B1212-7S). 10. Sparkle cleaner (VWR 500032-904).

2.2 Components for Imaging

1. Custom-sized glass slides (127  85 mm, 1 mm thickness; ordered from Hausser Scientific). 2. Custom-sized cover glass (115  75 mm, 0.17 mm thickness No. 1.5; ordered from VWR and prepared by Thermo custom order 115  75–1.5-VWR cover glass). 3. Two of custom-made stainless steel molds for 96 pedestals (see Fig. 2 for dimensions). 4. Molecular grade agarose (VWR 12001-870). 5. Microwave oven. 6. Kimwipes.

280

Paula Montero Llopis et al. 128 mm 8 mm

9mm

7 mm

10 mm 2 mm

9mm

86 mm

2 mm

1 mm Thickness

Side view

Agarose inside wells/holes Stacked molds Glass slide

Side view

Agarose pedestals

Single mold Glass slide

Fig. 2 Dimensions of the custom-made stainless steel molds. Top, dimensions of the molds built in a machine shop. These numbers are needed for the JOBS macro to define the custom plate. Bottom, schematics of the side view of the slide/mold/agarose assembly before (upper panel) and after (lower panel) exposing the pedestals

7. 70% Ethanol in a squirt bottle. 8. Fully motorized Nikon Ti inverted microscope equipped with a Perfect Focus System (PFS), a Plan Apo 100/1.45 N.A. Ph3 oil-immersion objective (or a Plan Apo 60/1.4 Ph3 objective lens), an Andor Zyla 4.2 plus sCMOS camera, Nikon Ti-S-ER motorized stage with a multi-well stage insert (TI-SH-W; Nikon), filter cubes for GFP (Chroma 49002), Texas Red (Chroma 49008), and a Lumencor Spectra X light engine. The microscope is operated using Nikon Elements software equipped with a JOBS module. Similar microscopes and software can be used.

High-Throughput Imaging of Bacillus subtilis

3

281

Methods

3.1 Exponential Growth of Cells

1. From frozen B. subtilis strains stored in 96-well format, inoculate directly into 96-deep-well plates containing 300 μL of CH (see Note 1). Wipe the foil that covers the frozen stock with 70% ethanol before carefully peeling it from one side. Sterilize the pin tool by submerging the pins in a shallow container containing 100% ethanol and flaming them (see Note 2). Once the pins are cooled down, firmly press the pinning tool onto the frozen 96-well stock and immediately inoculate into the 96-deep-well plates containing 300 μL growth media (Fig. 1; see Note 3). Seal the plate with a breathable cover. 2. Incubate the cultures overnight at 30
C, shaking at 900 rpm. 3. In the next morning dilute the cultures in 300 μL of fresh CH growth medium in 96-deep-well plates to an ~OD600 of 0.02. Seal the plate with a breathable cover. 4. Put the plates in a plate shaker and grow for 5–7 generations (in our case 30–40 min/generation) at 900 rpm at 37
C.

3.2 Prepare 96-Well Pedestals

1. While the cells are growing, prepare 2% agarose solution in growth media to make the pedestals: dissolve 2 g of molecular grade agarose in 100 mL of growth medium. Microwave until homogenous. Keep at 65
C (see Note 4). 2. Prepare the agarose pads (Fig. 3, see Note 5). Clean two custom-sized glass slides per plate by wiping them with a Kimwipe dampened with ethanol from a squirt bottle. Avoid using too much ethanol which will produce strikes on the glass. Next, wipe the slides with a Kimwipe dampened with Sparkle cleaner to remove any water-soluble debris. Layer the glass slides and the stainless steel molds as shown in Fig. 3. Clean an area of the bench with water and then 70% ethanol. Place a glass slide on the bench and the two overlaid stainless steel molds on top. Pour the 2% agarose mixture to the center of the assembly (see Note 6), slowly moving toward the edges. Immediately after, place a second slide over the agarose-covered assembly at a shallow angle and gently lower it (Fig. 3). Gently place a weight on top to remove excess agarose. Leave on the bench for 15–20 min to allow the agarose to solidify (see Note 7). Using a razor blade, cut the excess solidified agarose around the assembly and turn it over. The glass slide that was against the bench is now on top (see Notes 8 and 9). 3. Carefully remove excess agarose from all 4 sides of the assembly with the razor blade. 4. Expose the surface of the agarose by sliding the top glass slide longitudinally. Avoid pulling, as this will warp and distort the

282

Paula Montero Llopis et al.

Components

Prepare slide and molds to cast the pedestals

Pour agarose

Put weight down and wait ~15-20 min

Use a razor blade to remove excess agarose

Expose the agarose surface

Clean excess agarose with a kimwipe

Finished pedestals

Apply the droplets to each row on the coverslip

Continue applying droplets of all strains from the multi-well the plate

Apply top slide

Use a razor blade inserted between the molds to expose the pedestals

Prepare the coverslip on the Depress the multi-channel cleaned, discarded mold pipet plunger to expell a small droplet only

Invert the pedestals and lower them onto the droplets on the coverslip

Finished mounted pedestals

Fig. 3 Preparation of the 96 agarose pedestal assembly. During imaging, the glass slide is fastened on the stage by lab tape (see Fig. 4). The gap between the cover glass and the mold, which is set by the height of the removed mold, allows the agarose pads to be exposed to air in the humidified temperature-controlled incubator encasing the microscope. For B. subtilis, adequate oxygen is important for the health of the cells

agarose. Remove the thin layer of excess agarose from the edges of the mold using a Kimwipe. Slide a razor blade at one of the corners of the assembly, in between the two stainless steel molds. Pry the two stainless steel molds apart by twisting the razor blade. This will expose the agarose pedestals. Wipe the thin layer of agarose from the edges of the mold and around the pedestals (see Note 10).

High-Throughput Imaging of Bacillus subtilis

Place the mounted pedestals on a pipette tip box

Invert the multi-well stage insert over the pedestal assembly

Fasten the pedestal assembly to the stage with lab tape

Apply immersion oil to the top lens of the objective lens

Place the stage insert on the microscope stage

Screw the stage insert. Move the objective lens to the A1 position

283

Apply immersion oil to the coverglass

Start imaging

Fig. 4 Mounting of the agarose pedestal assembly on the microscope for high-throughput imaging. The agarose pedestal assembly is fastened to the stage insert and immersion oil is applied to the surface of the cover glass

5. Clean the discarded slide, mold with soap and water as soon as they are removed, and then rinse them with ddH2O and 70% ethanol. Reuse them for the following steps or for a new batch. 6. Prepare a custom-sized cover glass to mount the strains on. Place a clean and dry stainless steel mold from step 5 on a clean bench. Place the custom cover glass over the mold, which will guide where to drop the cells on the glass slide (see below). Make a mark at the top left corner of the cover glass with a sharpie to designate where the A1 well/pedestal will be (Fig. 3; see Note 11). 3.3 Mounting B. subtilis Strains on the Agarose Pedestals

1. Once the agarose pedestals are exposed, spin down the 96-deep-well plate containing exponentially growing B. subtilis at 1000 g for 2 min. 2. Using a handheld multi-well vacuum tool, aspirate most of the supernatant, leaving ~20–50 μL of the growth medium (see Notes 12 and 33). 3. Using a 12-well multichannel pipette set to 15 μL, gently pipette up and down to resuspend the cell pellets in each well, avoiding air bubbles. Take 15 μL of each cell suspension from one row of the deep-well plate (e.g., A1–12; see Note 13). Gently depress the plunger of the multichannel pipette to expose a small droplet of cell suspension in each pipette tip,

284

Paula Montero Llopis et al.

without letting them fall (Fig. 3; see Note 14). Touch the surface of the cover glass to spot each droplet (see Note 15). 4. Repeat step 3 until the cell suspension from each well is spotted on the cover glass. 5. Pick the exposed agarose pedestals, invert them so that they are facing down, and carefully lower them onto the cover glass with the arrayed droplets of cells (see Note 16). 6. Image the arrayed strains immediately (see below; see Note 17). 3.4 High-Throughput Imaging of B. subtilis Strains Using WideField Fluorescence Microscopy

1. Mount the 96-well pedestal assembly onto a multi-well stage insert. To do so, place the glass-slide side of the pedestal assembly on an empty pipette tip box, so that the cover glass side of the assembly is facing up (Fig. 4). Invert the multi-well stage insert and place it over the pedestal assembly. Fasten the pedestal assembly to the stage insert by sticking all four sides with lab tape (see Note 18). 2. Apply immersion oil to the surface of the cover glass using an eye drop applicator or a small brush (see Note 19). 3. Apply immersion oil to the top lens of a Plan Apo 100/1.4 oil objective lens (see Note 20). 4. Before mounting the stage insert onto the microscope stage, ensure that the nosepiece is in the lowest position (see Note 21). 5. Attach the stage insert to the microscope stage by fastening the screws with a #2 Allen key. Be careful not to overtighten the screws as they can break. 6. Optimize acquisition settings to produce high signal-to-noise ratio images (see Note 22). Proceed with semiautomated Subheading 3.5 or automated Subheading 3.6 image acquisition depending on the capability of your acquisition software.

3.5 Option 1: Setting Software for Semiautomated Image Acquisition

1. Position the pedestal assembly so that the Plan Apo 100/1.4 Ph3 objective lens is below the pedestal at A-1 position (Fig. 4). Slowly bring the nosepiece up until the immersion oil on the lens is in contact with the surface of the cover glass. Turn on the Perfect Focus System (PFS) and continue carefully bringing the nosepiece up using the fine focus knob until the PFS is engaged. Alternatively, look down the eyepiece using transmitted light and carefully focus on the surface of the agarose (see Notes 23–25). 2. Once the cells are in focus, adjust the optics in the transmitted light path for Koehler illumination to ensure bright and even illumination across the field of view (FOV). This is critical to produce high-contrast and -resolution phase-contrast images.

High-Throughput Imaging of Bacillus subtilis

285

3. Set up the experiment in Nikon Elements ND acquisition. In the λ tab, enter the optical configurations to be imaged (e.g., GFP, mCherry). In the XY tab, generate a relative XY array of points or FOVs to be imaged. To do so, Check “Relative XY” in the XY tab, click on custom, and in the pop-up window select the well plate tab. Enter the number of FOV to be imaged (rows and columns; e.g., 3  2) and enter the distance between points, which should be greater than the size of the FOV with a particular camera and total magnification (see Note 26). Enter a distance that would not create overlapping points. Note that the distance is in mm. Click on apply to create the points. The relative positions will appear in the XY tab. 4. Select a stage speed of Manage devices, highlight Nikon Ti, and click on Configure Devices. In the pop-up window select the appropriate speed. 5. Use the XYZ navigator tab to move from pedestal to pedestal automatically (right click on the workspace and select it under Acquisition controls). Make sure that the distance is set to 9 mm, which is the distance between the centers of adjacent pedestals. Use the arrows to navigate from pedestal to pedestal. Click on the arrow once; the stage will move from the center of one pedestal to the next. 6. Adjust the focus if necessary and press run to automatically acquire multipoint, multicolor fluorescence images (see Note 27). 7. Continue until images for all 96 pedestals have been acquired (see Note 28). 3.6 Option 2: Setting Software for Automated Image Acquisition

If the acquisition software is equipped with a JOBS module, a custom macro can be implemented to perform the automated acquisition of all 96 pedestals. This will enhance the highthroughput capability of the system (see Note 29). 1. Perform steps 1 and 2 in Subheading 3.5. 2. Set the stage speed to 1 mm/s. 3. Create a custom JOBS to automatically acquire 6 FOV per agarose pedestal, using the PFS and two-pass software-based autofocusing routines to maintain focus (example in Fig. 5). The steps are described below. 4. In the Plate Definition window, click “Custom Plate” to create a custom plate with the mold dimensions displayed in Fig. 2. 5. Align the custom pedestal plate in the Align Plate window, to ensure that the distance between pedestals is correct (see Note 30).

286

Paula Montero Llopis et al.

Fig. 5 Example of custom JOBS macro to fully automate acquisition of the pedestal assembly

6. Select the pedestals to image by highlighting the wells in the custom plate displayed in the WellSelection window. 7. Define the AutoFocus settings by selecting the “Manual” mode with a 20 μm range with “Two passes.” Set the first coarse pass to 3 μm and the second fine pass to 1 μm. Use a transmitted optical configuration (Phase) to avoid photobleaching and phototoxicity. In case of AutoFocus error, uncheck “Move to original Z on failure.”

High-Throughput Imaging of Bacillus subtilis

287

8. Add a Well Loop in which the PFS turns off, the stage moves to the center of each pedestal, the PFS is re-engaged, and 6 FOV or positions are generated in a random position within each pedestal to avoid bias. 9. Within the Well Loop, set up a Point Loop in which the macro will perform an autofocusing procedure using the PFS (Manual Mode, 5 μm range, 1 μm coarse step, and 0.3 μm fine step) in each point generated for each pedestal. 10. Define the acquisition parameters that contain the optical configurations for each well, such as the fluorescence and transmitted light channels (e.g., GFP, mCherry, Phase). 11. Select the directory to save the images. 12. Run the custom macro. 3.7 Characterization of Mutant Phenotypes by Quantitative Image Analysis

There are many open-source software available to quantify transmitted light or fluorescence images of bacterial cells [13–19]. We used the Oufti application to analyze the images of our arrayed fluorescent strain collection [14] (www.oufti.org). Oufti allows subpixel detection of cells and nucleoids using either phase contrast or fluorescence signal. Additionally, this software enables batch processing and quantification of the detected cells or structures. 1. Launch the Oufti application to open the main GUI. 2. Click on the “Phase” button to load the mCherry fluorescence images (which will provide information on the cell dimensions; see Note 31). 3. Click on “Load parameters” on the bottom right corner of the GUI window (see Note 32). 4. Optimize the segmentation parameters to ensure that the application detects the cells correctly (see Note 33). Once optimized, save the parameter set for batch analysis. 5. Click on “Signal 1” button to load the HbsU-mGFPmut3 fluorescence images. These will provide information on the dimensions of the nucleoids. 6. Select ObjectDetection from the top menu to detect the nucleoids within the cell outlines. 7. Optimize the segmentation parameters and save them for batch analysis. 8. Once the analysis for all strains is obtained, use the built-in tools to obtain statistics on cell dimension (length, width, standard deviation of length, standard deviation of width) and nucleoid characteristics (length, width, standard deviation of length, standard deviation of width, number of nucleoids per cell; see Fig. 6). Classification of these cellular characteristics will facilitate assigning functions to uncharacterized genes [5– 7] and generating hypotheses for further validation.

288

Paula Montero Llopis et al.

A

WT

Cell Dimension Length Width Area Volume

HbsUmGFPmut3 mCherry

Nucleoid Dimension Length Width Area Volume Other #Nucleoid/Cell Position of nucleoid in the cell Fraction of Cell volume occupied by DNA

Acquisition B

Separate Channels

Detection

Quantification

Classification

WT

ΔdivIVA

ΔripX

Δfpb

ΔbglP

ΔyoyJ

Fig. 6 Analysis pipeline and sample images from the screening. (a) Overview of the analysis pipeline to classify each phenotype. (b) Fluorescence images of phenotypes displayed by individual mutants. Green, HbsUmGFPmut3; red, mCherry. Scale bar, 5 μm

4

Notes 1. CH medium has lower background fluorescence compared to LB medium, making it more suitable for wide-field fluorescence imaging. 2. Ensure that the Bunsen burner is away from the ethanol container and that no ethanol is dripping from the pin before flaming to avoid fire hazards.

High-Throughput Imaging of Bacillus subtilis

289

3. Inoculate cultures at the end of the day to prevent overgrowth in the next morning. 4. To prevent agarose from boiling over, heat the agarose solution in the microwave at 5-s pulses and gently swirl the bottle to mix thoroughly between pulses, until the agarose is fully melted. Use caution to prevent burns. 5. Prepare the agarose pedestals ~20 min before harvesting the cells to ensure that they are not overdried. 6. Use caution when pouring the molten agarose over the pedestals, as it can cause burns. An excess of agarose is needed to prevent air bubbles. 7. The agarose makes everything slippery, and the glass and molds will shift. Carefully align the layers so that the holes in the two molds are superimposed. Hold the assembly for ~1 min to stabilize it. 8. Some agarose may get in between the bench and the slide, sticking the assembly to the bench. Hold the assembly by the long sides, avoiding pressing down, and pull toward you. 9. Using this side of the assembly ensures that the pedestals are even and flat. 10. The bottom slide and the bottom mold are not removed. They provide stability to the pedestal array. The molds and glass slides can be reused as long as they are kept clean. 11. Instead of being spotted on the agarose pedestals, the cells will be spotted on the surface of the cover glass. The mold at the bottom of the cover glass serves as a guide of where to spot the cells. 12. Avoid sucking the pellets up and losing the cells. Change the pipet tips to avoid cross contamination between strains. Start a new box of pipet tips and use the same column as the one in the plate when removing the supernatant. This reduces mistakes. 13. Only 1–2 μL of the 15 μL cell suspension will be spotted on the cover glass. Taking a larger volume than needed avoids air bubbles. 14. When using the multichannel pipette, firmly attach the tips to ensure optimal performance. If one of the droplets does not come out, continue anyway. Then use a regular pipet to spot the missing ones. 15. Be careful not to slide the cover glass over the mold when spotting the cell suspensions. This would lose reference of the positions. If this happens, start over with a new cover glass. 16. This prevents air bubbles and overdrying of agarose while spotting the cells. The cover glass will be fastened to the pedestals by surface tension.

290

Paula Montero Llopis et al.

17. To image several plates, stagger the timing of the experiments so that the next plate is ready ~1 h after the previous one. 18. Avoid placing the tape over the cover glass, as it will put pressure on the pedestals and squeeze them. 19. Do not apply excess oil as it will drip over the microscope components. Apply the oil on the surface of the cover glass over the pedestals, sufficient so that the top lens will be covered at all times with oil as it moves from each pedestal to the next. 20. Critical! Avoid directly touching the top lens with the applicator so as not to damage the objective lens. 21. This prevents damaging the objective lens. 22. To achieve a good signal-to-noise ratio (SNR) in a fluorescence image, adjust the light intensity and/or the exposure time so that the range of intensities within an image covers the dynamic range of the camera used. This depends on the bit depth (e.g., 12 bit or 16 bit; number of possible intensity levels in an image) and the associated camera gain. The human eye cannot distinguish enough intensities in a 12-bit image, and therefore it is critical to evaluate the histogram of intensities to determine whether the range of intensities is appropriate. Fluorescence background introduces noise and therefore reduces SNR. Finally, ensure that the exposure time and power density of the light source do not photobleach the fluorophore, as this can result in photodamage and irreproducibility. 23. Here we use a fully motorized, inverted Nikon Ti microscope equipped with a Perfect Focus System to maintain focus. The specific microscope is not required for this project, and any that has similar capability would likely work. 24. Caution is needed when focusing to prevent damage to the objective lens. 25. Although a phase-contrast objective lens reduces fluorescence signal, it produces high-contrast images of bacterial cells, improving their detection using transmitted light. Ensure that the appropriate phase-contrast annulus is set up in the condenser turret for the phase-contrast objective lens used (e.g., Nikon Plan Apo 100/1.4 Ph3 requires the Ph3 annulus to be positioned in the light path above the condenser lens). 26. To estimate the size of the FOV, multiply the size of the photodiode array (pixel) in the camera chip (e.g., 2048  2048), by the physical size of the photodiode (e.g., 6.5 μm) and divide by the total magnification (objective magnification and any additional optovar magnification, e.g., 60  1). In the example above, 2048  6.5/60 ¼ 221.9 μm. The distance between positions in each pedestal should be at least 0.25 μm to prevent imaging overlapping FOVs.

High-Throughput Imaging of Bacillus subtilis

291

27. Move from left to right, down one row, right to left. This will minimize the acquisition time. 28. The acquisition time in a semiautomated acquisition is ~35–40 min. Make different dilutions in the morning to ensure that the next plate will be ready to mount after the current plate is imaged. 4–6 plates can be imaged a day. 29. Making the pedestals even is key in the success of the automated acquisition. Set the speed to 1 mm/s. Apply enough oil on the surface of the cover glass. Avoid air bubbles between the cover glass and the pedestals. The PFS with longer working distance range performs better at maintaining focus. 30. The alignment routine only needs to be done once per imaging session as the dimensions and the position of the pedestals within the stage insert are constant. 31. Oufti detects objects from images loaded in the “Phase” channel in the main GUI. However, the detection can be done using transmitted light or fluorescence images. Ensure that the Invertimage value is set to 1 in the segmentation parameters. 32. Oufti contains several parameter sets that can be loaded. These are presets for different bacterial cell types. Optimization of the parameters will be required. See the tutorials provided in www. oufti.org for details. 33. Image quality is key in successful segmentation and detection in any analysis software. Images that contain too many cells close to each other will be challenging for segmentation. If cell density needs to be adjusted, modify the volume of the remaining medium at Subheading 3.3, step 2. At the analysis step, revision of the cell outlines is important for successful detection of the cells or nucleoids.

Acknowledgments This work was supported by the National Institutes of Health Grants R01GM141242 to Xindan Wang, and R01GM086466 and R01GM127399 to David Z. Rudner. References 1. Dubnau D, Blokesch M (2019) Mechanisms of DNA uptake by naturally competent bacteria. Annu Rev Genet 53:217–237. https://doi. org/10.1146/annurev-genet112618-043641 2. Koo BM, Kritikos G, Farelli JD, Todor H, Tong K, Kimsey H, Wapinski I, Galardini M,

Cabal A, Peters JM, Hachmann AB, Rudner DZ, Allen KN, Typas A, Gross CA (2017) Construction and analysis of two genomescale deletion libraries for Bacillus subtilis. Cell Syst 4(3):291–305. e297. https://doi. org/10.1016/j.cels.2016.12.013

292

Paula Montero Llopis et al.

3. van Opijnen T, Bodi KL, Camilli A (2009) Tn-seq: high-throughput parallel sequencing for fitness and genetic interaction studies in microorganisms. Nat Methods 6(10): 767–772. https://doi.org/10.1038/nmeth. 1377 4. Werner JN, Gitai Z (2010) High-throughput screening of bacterial protein localization. Methods Enzymol 471:185–204. https://doi. org/10.1016/S0076-6879(10)71011-9 5. Boutros M, Heigwer F, Laufer C (2015) Microscopy-based high-content screening. Cell 163(6):1314–1325. https://doi.org/10. 1016/j.cell.2015.11.007 6. Pennisi E (2016) IMAGING. ‘Cell painting’ highlights responses to drugs and toxins. Science 352(6288):877–878. https://doi.org/ 10.1126/science.352.6288.877 7. Rohban MH, Singh S, Wu X, Berthet JB, Bray MA, Shrestha Y, Varelas X, Boehm JS, Carpenter AE (2017) Systematic morphological profiling of human gene and allele function via cell painting. elife 6. https://doi.org/10. 7554/eLife.24060 8. Rothfield L, Justice S, Garcia-Lara J (1999) Bacterial cell division. Annu Rev Genet 33: 423–448. https://doi.org/10.1146/annurev. genet.33.1.423 9. Wang X, Montero Llopis P, Rudner DZ (2014) Bacillus subtilis chromosome organization oscillates between two distinct patterns. Proc Natl Acad Sci U S A. https://doi.org/10. 1073/pnas.1407461111 10. Graham TG, Wang X, Song D, Etson CM, van Oijen AM, Rudner DZ, Loparo JJ (2014) ParB spreading requires DNA bridging. Genes Dev 28(11):1228–1238. https://doi.org/10. 1101/gad.242206.114 11. Wang X, Montero Llopis P (2016) Visualizing Bacillus subtilis during vegetative growth and spore formation. Methods Mol Biol 1431: 275–287. https://doi.org/10.1007/978-14939-3631-1_19

12. Harwood CR, Cutting SM (1990) Molecular biological methods for bacillus. Wiley 13. Sliusarenko O, Heinritz J, Emonet T, JacobsWagner C (2011) High-throughput, subpixel precision analysis of bacterial morphogenesis and intracellular spatio-temporal dynamics. Mol Microbiol 80(3):612–627. https://doi. org/10.1111/j.1365-2958.2011.07579.x 14. Paintdakhi A, Parry B, Campos M, Irnov I, Elf J, Surovtsev I, Jacobs-Wagner C (2016) Oufti: an integrated software package for high-accuracy, high-throughput quantitative microscopy analysis. Mol Microbiol 99(4): 767–777. https://doi.org/10.1111/mmi. 13264 15. Ducret A, Quardokus EM, Brun YV (2016) MicrobeJ, a tool for high throughput bacterial cell detection and quantitative analysis. Nat Microbiol 1(7):16077. https://doi.org/10. 1038/nmicrobiol.2016.77 16. Guberman JM, Fay A, Dworkin J, Wingreen NS, Gitai Z (2008) PSICIC: noise and asymmetry in bacterial division revealed by computational image analysis at sub-pixel resolution. PLoS Comput Biol 4(11):e1000233. https:// doi.org/10.1371/journal.pcbi.1000233 17. Smit JH, Li Y, Warszawik EM, Herrmann A, Cordes T (2019) ColiCoords: a python package for the analysis of bacterial fluorescence microscopy data. PLoS One 14(6):e0217524. https://doi.org/10.1371/journal.pone. 0217524 18. Hartmann R, van Teeseling MCF, Thanbichler M, Drescher K (2020) BacStalk: a comprehensive and interactive image analysis software tool for bacterial cell biology. Mol Microbiol 114(1):140–150. https://doi.org/ 10.1111/mmi.14501 19. Jeckel H, Drescher K (2020) Advances and opportunities in image analysis of bacterial cells and communities. FEMS Microbiol Rev. https://doi.org/10.1093/femsre/fuaa062

Chapter 20 Measuring Nuclear Organization of Proteins with STORM Imaging and Cluster Analysis A´lia dos Santos, Rosemarie E. Gough, Lin Wang, and Christopher P. Toseland Abstract Super-resolution microscopy enables the high-precision localization of proteins. Therefore, it is possible to investigate the spatial organization of proteins within the nucleus to understand how their organization relates to regulation and function. Here, we present methodology for single-molecule localization microscopy and cluster analysis where we cover sample preparation, image acquisition, and data analysis. Key words STORM, Single molecule, Nucleus, Nuclear organization, Cluster analysis

1

Introduction The spatial organization of proteins within the nucleus is linked to their regulation and function. This is best exemplified within transcription regulation where the spatial organization has been debated and studied for over two decades [1–4]. The formation of transcription centers has been suggested to increase the local concentration of enzymes and render these nuclear processes more efficient [5]. Enzymatic clustering occurs in many cellular processes, with examples in replication [6] and DNA repair [7]. It is therefore vital to be able to implement an imaging modality with the ability to define the number, size, and composition of these nuclear clusters. STochastic Optical Reconstruction Microscopy (STORM) is a single-molecule localization technique able to address this need while readily achieving a resolution of 20–30 nm [8]. STORM utilizes standard organic fluorescent dyes commonly conjugated to commercial antibodies or covalent Halo or SNAP ligands. High-precision single-molecule localization is achieved by sequentially switching on a small number of sparsely distributed

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_20, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

293

294

A´lia dos Santos et al.

fluorophores. This occurs through a process called photoswitching. The typical fluorescent dyes (e.g., Alexa-Fluor 647) are not capable of photoswitching under standard microscopy conditions. The process is induced under high excitation power density (>5 kW/ cm2) in the presence of a thiol-containing buffer (e.g., 2-mercaptoethanol). Upon excitation, the fluorophore reacts with the thiol group to enter a long-lived dark state but will spontaneously reactivate, which generates the blinks and enables localization. The individual fluorophores are localized with nanometer precision and the localizations are accumulated across thousands of frames. An image is finally rendered from the list of localizations. A detailed review on this methodology can be found here [9]. Importantly, single-molecule localization enables molecular counting and density to be explored for our protein of interest. It is therefore possible to analyze how individual molecules are organized in relation to each other to establish if they form clusters which, in turn, can be related to protein function.

2

Materials

2.1

Etch Solution

Prepare the etch solution by mixing at a ratio of 5:1:1: Sterilefiltered H2O, ammonium hydroxide (NH4OH, ACS reagent, 28.0–30.0% NH3 basis), and hydrogen peroxide (H2O2, 50 wt. 50% in H2O, stabilized). To make enough solution to clean 50–100 coverslips use 500 mL filter-sterilized water, 100 mL H2O2, and 100 mL NH4OH and then filter the solution through a 0.2 μM PES membrane (see Note 1).

2.2

Imaging Buffer

Buffer A: 10 mM Tris (pH 8.0), 50 mM NaCl. Buffer B: 50 mM Tris (pH 8.0), 10 mM NaCl, 10%(w/v) glucose. GLOX solution: 14 mg Glucose oxidase, 50 μL catalase (17 mg/ mL), 200 μL buffer A (see Note 2).

2.3 Coverslip Preparation

When preparing the coverslips, always wear a lab coat, gloves, and glasses to avoid exposing skin to the corrosive chemicals. The procedure should be carried out in a chemical fume hood. 1. For STORM imaging, round 25 mm No. 1.5 coverslips should be used (see Note 3). These are compatible with the Attofluor cell chamber (Invitrogen #A-7816). Clean coverslips in the etch solution (see Subheading 2.1). 2. Place the coverslips in a rack holder using tweezers and transfer them into a glass beaker. 3. Immerse the racks in the etch solution and cover the beaker with tinfoil (see Note 4).

Measuring Nuclear Organization of Proteins with STORM Imaging and Cluster. . .

295

4. Place the beaker with coverslips inside a larger glass beaker halffilled with distilled H2O and place on a heating block set to 70
C for 2 h (see Note 5). 5. Use tweezers to transfer coverslip racks into a clean beaker filled with Milli-Q-autoclaved H2O. Wash the coverslips by moving the rack up and down. Repeat this process three times with fresh H2O each time. 6. Immerse coverslip racks into 100% MeOH or EtOH. 7. In a cell culture hood, place the rack onto a clean tissue and using tweezers remove excess MeOH/EtOH by shaking the coverslip or by blowing with filtered air (see Note 6). 8. Finally, store the clean and dry coverslips in a clean staining rack and store in a sealed box to prevent dust settling on the coverslips. 9. Before use, place coverslips in 6-well tissue culture plates and UV-sterilize coverslips for 15–30 min before seeding cells.

3 3.1

Methods Cell Preparation

1. Seed cells at desired density onto the pre-cleaned coverslips in a 6-well plate (see Note 6). 2. Allow the cells to attach and perform cell treatments as required (see Note 8).

3.2 Immunofluorescence

1. Wash cells twice with phosphate buffer saline (PBS). 2. Fix cells in 4% (w/v) paraformaldehyde (PFA) in PBS. 3. Wash cells three times with PBS (5 min per wash). 4. Quench residual PFA and cell autofluorescence by incubating cells with 50 mM NH4Cl in PBS for 15 min. 5. Wash coverslips three times with PBS (5 min per wash). If the immunofluorescence (IF) protocol needs to be done in Trisbuffered saline (TBS), wash cells an additional two times in TBS (see Note 9). 6. Permeabilize and simultaneously block cells for 30 min with blocking buffer (3% (w/v) BSA, 0.1% triton (v/v) X-100 in PBS or TBS), and leave at room temperature either on the bench or rocking. 7. Without washing coverslips, incubate the permeabilized cells for 1 h (see Note 10) with the desired concentration of primary antibody in blocking buffer (see Note 11). To incubate, place 200 μL of diluted antibody onto parafilm and then invert coverslips (cells facing down) onto the antibody solution (see Note 12).

296

A´lia dos Santos et al.

8. Re-invert coverslips back into the plate and wash three times (10 min each wash) with washing buffer—0.2% (w/v) BSA and 0.05% (v/v) Triton X-100 in PBS or TBS (see Note 13). 9. Incubate coverslips, as described above, for 1h with the appropriate conjugated secondary antibody in blocking buffer (see Note 14). 10. Wash coverslips three times (10 min per wash) with washing buffer. 11. Wash coverslips with PBS, or TBS, and then 1 with PBS (see Note 15). 12. Perform a second post-IF fixation with 4% (w/v) PFA in PBS for 10 min (see Note 16). 13. Wash coverslips three times with PBS (10min). 14. Replace with PBS supplemented with 0.02% sodium azide (NaN3) (2 mL/well) (see Note 17). 15. Parafilm plates and cover with silver foil. Store at 4
C. 3.3 Imaging Sample Preparation

1. Prepare the imaging buffer by combining 7 μL GLOX, 7 μL 2-mercaptoethanol, and 690 μL buffer B. Store in the dark at 40C (see Note 18). 2. Wash the coverslip three times in PBS and one time in filtered Milli-Q H2O. Mount in the Attofluor cell chamber. 3. Add imaging buffer to the chamber (see Note 19).

3.4 STORM Image Acquisition

The following STORM image acquisition and processing methods are produced by assuming that users implement 2D multicolor STORM imaging using a ZEISS Elyra PS1 microscope. Two proteins of interest in the cell nucleus are labeled with organic dyes Alexa-Fluor 488 (AF488) and Alexa-Fluor 647 (AF647), respectively. 1. Switch on the microscope and computer. Open the ZEISS ZEN software (black version). Click the Start System button to boot up, and wait for the user interface to appear. 2. Go to the Acquisition tab, in the Setup Manager group, and switch on the 405 nm, 488 nm, and 642 nm lasers in the Laser section. Then select the alpha Plan-Apochromat 100/1.46 Oil DIC objective by clicking Objective icon in the optical path figure in the Imaging Setup section (see Note 20). 3. Mount the cell sample holder, on the microscope stage. 4. Find the cells on the focal plane. The cells may be found by direct visual observation through the microscope’s eyepiece using either bright-field (BF) or fluorescence imaging (see Note 21). For the BF imaging, under the Locate tab, click open the Transmitted Light Lamp icon and click the ON button in the Microscope Control section. For fluorescence

Measuring Nuclear Organization of Proteins with STORM Imaging and Cluster. . .

297

imaging, click open the Excite Reflected Light icon, click the ON button, then click the Transmission Light icon, and choose FSet77 HE, a triple-band beam splitter, to allow multicolor fluorescence imaging. Position the cells in the center of the field of view. 5. Implement laser wide-field (WF) fluorescence imaging to find regions of interest (ROIs) (see Note 22). Under the Acquisition tab, in the Setup Manager group, in the Imaging Setup section, activate laser WF imaging mode by clicking the Laser WF button. Automatically, the TV1: EM CCD Andor PALM button is activated, indicating that the EMCCD camera (Andor iXon 897) has been assigned. 6. Select the appropriate optical filter set, such as BP 420-480 + BP 495-550 + LP 650 in this case, by clicking the Transmission Light icon, and choose the desired filters. 7. Next, in the Acquisition Parameter group, set up two-color imaging by firstly adding the second track (color) (see Note 23). This can be done by clicking the + button in the Channels section. Then, select the appropriate lasers for each track by ticking the boxes in the Lasers option. In this case, tick the 642 laser box when the Track 1 banner is activated, and then tick the 488 laser box when the Track 2 banner is activated. The initial laser power may be set to 0.5–2 (%) of the total power. At this point, the display colors in each track can also be designated by selecting the desired colors from the drop-down lists in the color options in the Track banners. 8. The final step is to assign Exposure Time [ms] and EMCCD Gain in the camera settings. The suitable initial values may be 100 (ms) and 50, respectively. 9. Now click the Continuous button under the Acquisition tab; real-time images should appear in the image display area (see Note 24). Move the cell sample around by translating the microscope stage, until a ROI is found. Take the WF image of the ROI by clicking the SNAP button next to the Continuous button, and save the image by clicking the floppy disc icon to the appropriate directory. 10. Acquire raw STORM images. Firstly, set up a time series for raw STORM image acquisition. Under the Acquisition tab, tick the Time Series box. Accordingly, the Time Series section is activated in the Multidimensional Acquisition group. Assign an appropriate frame number, e.g., 10k–50k (see Note 25), in the Cycles option. 11. To optimize two-color imaging, a sectioned sequential imaging strategy is required. In the Imaging Setup section of the Setup Manager group, select the Sequence in the Switch track every option. As a result, the Sequence Size option is activated in the Acquisition Mode section of the Acquisition

298

A´lia dos Santos et al.

Parameter group. Assign an appropriate number, e.g., 300 (frames), as the section size for both tracks (see Note 26). During data acquisition, the image series alternate in every assigned number of frames between the two tracks. Furthermore, it is necessary to use the microscope’s internal trigger for the fast switching. In the Acquisition Parameter group, activate this trigger by choosing Internal in the Trigger in option in the Acquisition Mode section. 12. Reduce the frame size to increase imaging speed. In the Acquisition Parameter group, assign appropriate pixel numbers, e.g., 256  256 (25.6  25.6 μm) or 128  128 (12.8  12.8 μm), in the Frame Size option in the Acquisition Mode section. To match the chosen frame size, reduce the laser illumination area by choosing either TIRF_HP or TIRF_μHP, respectively, after clicking TIRF icon in the optical path diagram in the Imaging Setup section of the Setup Manager group. 13. Assign camera parameters and prevent axial drifts. In the Acquisition Parameter group, in the Channels section, set the Exposure Time [ms] and EMCCD Gain to be 30 (ms) and 250, respectively. To prevent axial drifts, a focus locking strategy has to be implemented. In the Focus Devices and Strategy section, choose Definite Focus in the Autofocus Mode option. Then tick Autofocus every n Timepoint box, and assign an appropriate frame number, e.g., 300 (frames), as the time gap between two consecutive focal position compensation actions in the axial direction. 14. Pre-bleach the cells. To observe single-molecule photoswitching (blinks), it is necessary to shine high-intensity laser beams to the cells in order to push the majority of the dye molecules into a long-lived dark state. This procedure should start from track 1, i.e., AF647. Tick track 1 box and untick track 2 box in the Channels section, and then click the Continuous button to monitor the cell images in real time. Gradually increase the 642 nm laser power up to 100 (%) until single-molecule blinks appear on a dark background. Once the blink quasiequilibrium is set, reduce the laser power to an appropriate value, e.g., between 40 (%) and 80 (%) (see Notes 27 and 28). After that, implement this procedure in track 2 in a similar manner. 15. Apply highly inclined and laminated optical (HILO) illumination (Fig. 1a) to reduce the background. Since the nuclear proteins are packed in a confined space, it is beneficial to limit the imaging depth to about 1 μm using the HILO illumination (Fig. 1a). In the Channels section, click and activate the HILO button in both tracks in the Illumination mode option. The corresponding TIRF Angle (˚) option changes accordingly.

Measuring Nuclear Organization of Proteins with STORM Imaging and Cluster. . .

299

Fig. 1 STORM imaging and cluster analysis of proteins in the nucleus. (a) Schematic showing HILO illumination of samples for STORM imaging. Only a small sheet of the cell is illuminated—this reduces the levels of out-offocus background fluorescence and increases signal-to-noise ratio. (b) Wide-field (left panel) and STORM (right panel) images of two proteins—pSer5-RNA polymerase II labeled with Alexa-Fluor 488 (green) and NDP52 labeled with Alexa-Fluor 64 (red). Scale bar: 5μm. (c) ClusDoc rendering of localizations generated by ZEN analysis of STORM data shown in (b). pSer5-RNA polymerase II is now shown in red and NDP52 in green. ROI selected for cluster analysis is also shown with the solid black line. (d) Linearized Ripley K function across the selected ROI. L(r)-r function shows the spatial distribution of molecules across a radius r. Positive values indicate that the protein is clustered within the ROI, with a maximum of clustering at r ¼ 100 nm (blue line). A random distribution would have L(r)-r values of zero across r, while negative values would be indicative of a segregated distribution. (e) DBSCAN analysis establishes the number of neighbors within a certain radius x— usually the mean value for localization precision. If enough neighbors are found within this user-defined radius, the group of localizations is defined as a cluster. (f) Following DBSCAN analysis, cluster maps are generated for both channels. Clusters are represented in green or red, for channels 1 and 2, respectively. Non-clustered molecules are displayed in gray

300

A´lia dos Santos et al.

16. Acquire raw STORM data. Add the 405 nm activation illumination in both tracks by ticking the 405 laser box in the Lasers option in the Channels section. The initial laser power may be set to 0.1 (%) of the total power (see Note 29). Then start data acquisition by clicking the Start Experiment button under the Acquisition tag (see Note 30). 17. When performing two-color STORM, it is necessary to acquire two-color fluorescent bead images for channel alignment purposes. The channel alignment procedure eliminates residual chromatic aberrations and other systematic aberrations in multicolor imaging. Mount the Tool for calibration Multi Spec slide, containing 140 nm fluorescent beads, provided by ZEISS on the microscope stage. Take two-color WF images of the beads (see Note 31) as described previously. Save the images for later use. 3.5 STORM Image Processing

The localization analysis identifies single-molecule spots as pointspread functions (PSFs) in each raw image, then fits the PSFs with a 2D Gaussian function, and then calculates the coordinates of their centers of mass. The coordinates and other related parameters are recorded in a table of results. The table is used to generate a rendered super-resolution image. 1. Open raw images in the same ZEN software by clicking the open folder icon and applying the appropriate directory. Under the Processing tab, in the Method section, expand the PALM drop-down list and choose the PALM option. Move to the Method Parameters section, and select the opened images by clicking Select button. Choose Account for overlap in the Setting drop-down list. This function allows simultaneous Gaussian fitting to partially superimposed PSFs, enabling dense molecule localizations within a crowded environment. In the Peak Finder option, assign Peak Mask Size and Peak intensity to noise values to be 9 (pixels) and 6, respectively. Choose x,y 2D Gauss Fit in the Fit Model drop-down list in the Localizer option (see Note 32). 2. Check if the desired single-molecule spots in each frame have been identified in the preview images that are shown next to the raw images. This can be done by choosing various frames in the Time option under the Dimensions tab shown below the image display area. The Peak Mask Size and Peak intensity to noise values may be altered until the identifications of single-molecule spots are satisfactory. Then click Apply button to implement localization analysis. This may take a while depending on the size of the raw images. Save the processed image in an appropriate directory.

Measuring Nuclear Organization of Proteins with STORM Imaging and Cluster. . .

301

3. The axial drifts in the raw images have been corrected in real time using the Definite Focus function during image acquisition; however the lateral drifts caused by the vibrations and thermal expansions can only be corrected in the image postprocessing. It is normally sufficient to use feature detection and cross-correlation method to correct the lateral drifts. Under the PAL-Drift tab shown below the image display area, choose Model-Based in the Correction Type drop-down list, then choose Automatic in the Segments drop-down list, and then assign 8 in the max option. Click the Apply button in this section. 4. The fluorophores remaining on a fluorescent “on” state over a few consecutive frames are treated as multiple fluorophores in the initial localization analysis. However, this is detrimental to the credibility of the reconstructed cellular structures, and can fundamentally fail quantitative single-molecule analysis. The grouping function is designed to find these fluorophores and combine them as a single localization. Empirically, under the PAL-Grouping tab shown below the image display area, assign 5 (frames), 10 (frames), and 2 (pixels) in the Max On Time, Off Gap, and Capture Radius options, respectively, for channel HR1, i.e., AF647. The corresponding values for channel HR2, i.e., AF 488, are 5 (frames), 8 (frames), and 2 (pixels) (see Note 33). 5. It is useful to examine the statistical results of PSF size and localization precision, and then filter out any irregular localizations. Under the PAL-Statistics tab shown below the image display area, choose Histogram in the Plot type drop-down list, and then choose PSF half width in the Histogram Source drop-down list. Examine the PSF size histograms, and estimate the mean value of the PSF sizes (precisely speaking they are the half sizes of PSFs) that is normally in the range of 120-150 nm. Then move to the PAL-Filter tab; assign a range centered at the mean value, e.g., minimum 50 (nm) and maximum 250 (nm), in the PSF Half Width option; and then tick the box to activate this option. The PSF size histogram changes accordingly. In the PAL-Statistics tab, choose Precision in the Histogram Source drop-down list. Examine the localization precision histograms, and check if they comply with a normal (Gaussian) distribution. If there are long tails on either ends of the histograms, filter out the excessive localization data points using the Localisation Precision option under the PAL-Filter tab, e.g., assign minimum 5 (nm) and maximum 50 (nm) precision. Lastly tick the box to activate this option. 6. It is necessary to eliminate residual chromatic aberrations and other systematic aberrations in the images. Firstly, generate a calibration table from the laser WF images of fluorescent beads.

302

A´lia dos Santos et al.

Open the bead calibration images, under the Processing tab, in the Method section, and click and choose Channel Alignment option. Then, in the Method Parameters section, select the opened images by clicking the Select button, then tick Fit, choose Affine from the Alignment mode drop-down list, and click the Apply button. The aberration-corrected image will be generated. Click the Save button in the Method Parameters section. A dialogue box will appear; save the calibration data in bin file format. Next, apply the calibration data to the STORM localization analysis results. Go back to the STORM localization analysis window, and select the image file being analyzed by clicking the Select button in the Method Parameters section. Untick Fit, then click Load button, and apply the appropriate directory where the calibration data, i.e., the bin file, is stored. Click Apply button to implement channel alignment and an aberration-free STORM image will appear. Save the results. 7. Right click the localization table shown in the image display area, and choose Save table in the drop-down list. The real product of STORM imaging is the table containing all the localization information rather than an image. This table may be used for further analysis, such as clustering analysis, which is not available in the ZEN software. 8. The most common way to present a STORM image is to render each localization to a Gaussian function taking into account its localization precision. Under the PAL-Rendering tab, assign half the size of the mean value of localization precision, e.g., 10 (nm/pixel), in the Pixel Resolution XY option, and then choose Gauss in the Display Mode drop-down list. Save the results. An example rendered image and corresponding WF image is shown in Fig. 1b for the potential transcription regulator NDP52 [10] and pSer-RNA Polymerase II. 3.6 Cluster Analysis of STORM Data

The cluster analysis described here uses ClusDoC [11] but other software are available for this analysis. 1. Download ClusDoC [11] package from https://github.com/ PRNicovich/ClusDoC. 2. Open ClusDoC file location in MATLAB and open the graphical user interface (GUI) for the software by typing “ClusDoC” into Command line window. 3. Load x, y coordinate table for STORM localizations in ClusDoC (see Notes 34 and 35). This will render an image of the STORM acquisition in ClusDoC with all detected particles. 4. Select a folder to save the analysis (see Note 36).

Measuring Nuclear Organization of Proteins with STORM Imaging and Cluster. . .

303

Fig. 2 ClusDoC analysis of STORM data. (a) Colocalization heat maps for channels 1 and 2, generated from STORM data shown in Fig. 1b. Channel 1 shows the density of colocalization for NDP52 and channel 2 for pSer5-RNA polymerase II. Molecules with colocalization scores of 1 (red) are perfectly colocalized and molecules with values of 1 (blue) are separated from each other. (b) Histograms generated from ClusDoC analysis showing relative frequency of colocalization scores for all molecules in the ROI

5. Draw a region of interest (ROI). Figure 1c shows the whole nucleus as an ROI for ClusDoC analysis. 6. Run global clustering analysis (“Ripley K for all”) (Fig. 1d–f) (see Note 37). 7. Run Clus-DoC analysis (“ClusDoC for all”) (see Notes 38 and 39). This will run both colocalization and cluster (densitybased spatial clustering of applications with noise—DBSCAN) analysis. User-defined parameters can be introduced in the new window. a. For degree of colocalization analysis, a coordinate-based method is employed to determine the correlation between individual molecules in each channel [12]. A histogram of DoC stores for each channel is produced along with a colorcoded colocalization cluster map that reflects the DoC value for each molecule (Fig. 2). The colocalization threshold is usually set at 0.4, above which molecules will be considered colocalized. The L(r)-r radius can be obtained from the Ripley K analysis and set to the radius that corresponds to the maximum value of the L(r)-r function. b. For DBSCAN analysis (Fig. 1e, f), user-defined values can be introduced separately for the two channels—Ɛ represents the search radius and can be set to the localization precision of the STORM data, in nanometers. MinPts (minimum number of neighbors within the defined radius) for a cluster

304

A´lia dos Santos et al.

can also be user defined—we recommend a minimum of 5. As before, L(r)-r radius can also be set as the maximum radius of the Ripley K function. Cluster contour can also be defined (see Notes 40, 41, and 42). 8. Click “Export Results Tables” (see Note 43).

4

Notes 1. The etch solution can be stored at room temperature and used up to three times. 2. Vortex to dissolve glucose oxidase and centrifuge at 6500  g. Keep the supernatant and store at 4
C for up to 2 weeks. 3. The coverslip thickness has to be No. 1.5, i.e., 170 μm; otherwise the image quality may be compromised. High-precision coverslips (thickness error within 5 μm) are preferred. 4. Coverslips should be completely submerged under approximately an excess of 5 cm of the etch solution. 5. Use a thermometer to measure the temperature and do not let the temperature exceed 70
C as the etch solution will begin to boil and coverslips will be displaced from the racks. To avoid coverslips coming out of the racks we use a smaller empty beaker and balance on top of the coverslip racks. 6. Filtered air must be used to avoid residue being deposited on the clean glass. 7. When using HeLa cells, we recommend a seeding density of 320K but this will vary depending on the cell type and treatments used. 8. Do not label cells with DNA dyes excited at 405 nm because this will interfere with imaging but DNA dyes can be used for STORM imaging of DNA [13]. Transfected proteins, such as Halo- or SNAP-tagged fusions, can be labeled with appropriate dyes before fixation. In this instance, we recommend using the Janelia-Fluor dyes, in particular JF549. 9. If immunofluorescence is performed using anti-phospho antibodies all subsequent steps should be performed in TBS. 10. Antibody concentrations should be optimized using STORM. Higher concentrations of primary antibody are frequently required to label samples at sufficient density for singlemolecule imaging. 11. Incubation times are dependent upon the antibody used. 12. Antibody incubation and any subsequent steps should be kept in the dark to avoid any photobleaching of dyes. It can also be helpful to add a dampened piece of tissue between coverslips to make sure that they do not dry out while incubating.

Measuring Nuclear Organization of Proteins with STORM Imaging and Cluster. . .

305

13. For washes, BSA is used in the buffer to reduce nonspecific interactions. 14. For STORM imaging, we recommend Alexa-Fluor dyes due to their photophysical properties, brightness, and stability [14]. We have found using Alexa-Fluor 647 and Alexa-Fluor 488 [4, 15, 16] to be a good combination for two-color experiments. It is important to select fluorophores with a wide spectral separation to avoid background/cross-excitation. 15. Here we leave coverslips in PBS regardless of whether the experiment has been carried out in TBS. 16. This double fixation is used to ensure that both the primary and secondary antibodies are fixed in place and avoids motion during imaging. 17. NaN3 prevents microbial growth on coverslips, allowing storage for longer periods of time (up to 4 weeks). 18. The GLOX solution acts as an oxygen-scavenging system to enhance the photostability of the organic fluorophores throughout the long acquisition times of experiments. 2-Mercaptoethanol, or equivalent thiol solution, is required for photoswitching—where molecules reside in a dark (off) state for long periods of time, before spontaneously becoming activated, thus enabling localization. 19. The solution will enable imaging for at least 2 h. After this time, or if blinking is markedly reduced during STORM acquisition, replenish imaging chamber with fresh imaging buffer. 20. Wait for 15 min before commencing an experiment, allowing the EMCCD camera to cool down to 63
C, the electronics and mechanics to stabilize, and the lasers to warm up. 21. If it is still difficult to identify cells in BF, it might be beneficial to switch to fluorescence imaging mode. 22. As the prerequisite for STORM imaging, WF fluorescence imaging provides not only a fast route to identify ROIs, but also the opportunity to take a WF image as a comparison to the reconstructed STORM image. From the latter, useful information, such as when to terminate the raw STORM image acquisition and how much artifact may have been introduced in the STORM image, can be obtained. 23. Assign the dye with longer wavelength, i.e., AF647, to track 1 and the one with shorter wavelength, i.e., AF488, to track 2. This is to ensure that the pre-bleach of AF647 dye molecules will not bleach AF488 prematurely during the pre-bleach procedure in STORM imaging at a later stage. 24. The default switching mode between tracks is “Frame” that causes a noticeable delay in real-time imaging. To resolve this,

306

A´lia dos Santos et al.

the “Frame Fast” mode can be applied. Under the Acquisition tab, in the Setup Manager group, in the Imaging Setup section, choose Frame Fast in the Switch track every option. At this point, an info window pops out showing the following message: “Several settings will be overwritten. Setting of the first track will be used. Continue?” Click the Yes button and the camera’s Exposure Time [ms] and EMCCD Gain setting for track 2 will be overwritten by those assigned to track 1. Furthermore, in the Acquisition Parameter group, in the Acquisition Mode section, activate the microscope’s internal trigger for the EMCCD camera by choosing Internal in the Trigger in option. 25. When imaging an unfamiliar sample, the frame number may be set to a high value initially, e.g., 50k; then take raw STORM images with the “Online Processing” function on. Using the prerequisite WF image as the reference, observe the reconstructed STORM image in the real-time display, and manually terminate the raw image acquisition at an appropriate time point. After several runs, an optimal frame number may be established empirically. 26. If the Sequence Size is too small, the dye molecules recovered from the dark state dominate each individual section, causing low signal-to-noise ratio in the raw images. On the other hand, if the Sequence Size is too big, large and/or abrupt lateral drifts may occur between the two consecutive sections in both tracks, hindering the accuracy of drift corrections. 27. The density of blinks from dye molecules is directly affected by the imaging buffer. As a rule of thumb, if the density of blinks in each frame is too low (or high), try to decrease (or increase) the thiol buffer concentration, and/or decrease (or increase) the buffer pH value. Empirically, a dozen blinks sparsely appearing in the 25.6  25.6 μm field may be considered as a suitable state in the raw STORM image acquisition. 28. The corresponding power densities of the 642 nm and 488 nm lasers at the focal plane are 7–14 kW/cm2 and 7–12 kW/cm2, respectively, when the laser illumination is set to be the TIRF_HP mode. 29. The corresponding power density of the 405 nm laser at the focal plane is 1 W/cm2. The power of the 405 nm laser may be increased manually when the number of blinks decreases during the image acquisition. If the photophysical properties of the dyes are well known, the increase of the 405 nm laser power may be implemented automatically. In the Online Processing Options, tick the Control activation power box, and then assign a percentage, e.g., 1(%) to the End Power option. Therefore, the power of the 405 nm laser gradually increases from zero to 1% during the raw STORM image acquisition.

Measuring Nuclear Organization of Proteins with STORM Imaging and Cluster. . .

307

30. The “Online Processing” function may be deployed during raw STORM image acquisition. In the Acquisition Parameter group, in the Online Processing Options section, tick the Online Processing PALM box, then tick the Fit Gauss 2D box, and leave the default values, i.e., 9 (pixels) and 6, respectively, in the Peak Mask Size and Peak intensity to noise options. The real-time reconstruction of STORM image appears next to the raw image display during the image acquisition. 31. Select an area where dozens of fluorescent beads are sparsely distributed in the 25.6  25.6 μm field. Assign appropriate Exposure Time [ms] and EMCCD Gain values in the camera settings, e.g., 2 (%) and 50, respectively. 32. When the Account for overlap is selected, the Maximum cluster size option becomes available. Assign an appropriate value, e.g., 10, depending on the computing power of the PC. The value defines the maximum number of molecules to be fitted using 2D Gaussian function in a cluster (an area where multiple molecules are densely packed). In the PSF Half Width option, assign an appropriate value, e.g., 177.9 (nm, the default value for the chosen objective). More precisely, the experimental PSF half width can be derived from the fluorescent bead calibration data and applied here. Under the Processing tab, in the Method section, select Experimental PSF, then click the Select button, and then Apply button; the averaged PSF image appears in the image display area. Click the info tab on the left-hand side; the experimental 2D FWHM values are summarized in the Notes area. 33. The blinking behavior from dye molecules is dependent on many factors, such as laser illumination intensity, chemical components in the buffer, pH value of the buffer, and temperatures. The fluorescent “on/off” time cannot be inferred from the blink statistics either. To apply the grouping function in the most reliable way, the fluorescent “on/off” time is determined by measuring control samples. For example, some sparse dye molecules may be coated on a glass surface; then the blinks can be induced under the same experimental condition as the cell samples. The fluorescent “on/off” time may be measured from multiple single-molecule spot sequences. The other way to seek reasonable fluorescent “on/off” time is through trial and error. It is sensible to use two frames in the Max On Time option initially, and 0 frame in the Off Gap option; then increase these two values gradually and examine the changes in the localization precision histogram. When reaching a point that the smallest precision is achieved, the fluorescent “on/off” time may be determined at this point.

308

A´lia dos Santos et al.

34. This is the .txt file generated during the PALM analysis step. 35. As we previously described, for STORM imaging, the longer wavelength (e.g., Alexa-Fluor 647) is assigned to track 1. When the localization table is uploaded to ClusDoC, the software defaults channel 1 to green and channel 2 when rendering an image. This means that red and green channels will be switched in comparison to the STORM image (Fig 1b, c). 36. ClusDoC will automatically save plots, cluster/density maps, and tables in this folder. 37. Ripley K analysis will give a general overview of how the particles are distributed within the ROI. The software generates plots of L(r)-r versus r, which represent the linearized form of the Ripley K function. A randomly distributed population will have values of L(r)-r of zero, while clustered molecules will have positive values and segregated ones will have negative values (Fig. 1d). 38. ClusDoC analysis is divided into two steps. First the software calculates colocalization scores for the two channels and then a DBSCAN step detects and characterizes individual clusters within the ROI. 39. The “ClusDoC for all” step only applies when analyzing two-channel STORM data. For cluster analysis of singlechannel data “DBSCAN for all” is used instead. 40. Smooth radius for cluster contour will affect circularity calculations for clusters and might also affect other parameters such as cluster area. We recommend a value of 7 for smooth radius; however, this is sample dependent and should be optimized. 41. Although one of the advantages of ClusDoC is that processing times are short, occasionally, MATLAB errors occur. This is common when performing the analysis using computers with low processing capabilities or for ROIs that have particularly high localization numbers or particularly large clusters. Settings such as “array size” and “memory use” in MATLAB can be increased to help processing. Alternatively, dividing the ROI into smaller sub-ROIs can also help and data from these can be combined at a later stage. 42. All the analysis is automatically saved as .csv files, .mat files, and graphic representations (Fig. 1f) and includes information on cluster size, area, average number of molecules per cluster, and percentage of molecules in cluster. 43. Results tables are exported in .txt format to the same file location as the input STORM localization table. These files contain detailed colocalization scores, as well as cluster area and number of points for each cluster.

Measuring Nuclear Organization of Proteins with STORM Imaging and Cluster. . .

309

Acknowledgments We thank the UKRI-MRC (MR/M020606/1) and UKRI-STFC (19130001) for funding. References 1. Cho WK, Jayanth N, English BP, Inoue T, Andrews JO, Conway W, Grimm JB, Spille JH, Lavis LD, Lionnet T, Cisse II (2016) RNA Polymerase II cluster dynamics predict mRNA output in living cells. Elife 5. https:// doi.org/10.7554/eLife.13617 2. Jackson DA, Hassan AB, Errington RJ, Cook PR (1993) Visualization of focal sites of transcription within human nuclei. EMBO J 12(3): 1059–1065 3. Papantonis A, Cook PR (2013) Transcription factories: genome organization and gene regulation. Chem Rev 113(11):8683–8705. https://doi.org/10.1021/cr300513p ´ , Cook AW, 4. Hari-Gupta Y, Fili N, dos Santos A Gough RE, Reed HCW, Wang L, Aaron J, Venit T, Wait E, Grosse-Berkenbusch A, Gebhardt JCM, Percipalle P, Chew T-L, MartinFernandez M, Toseland CP (2020) Nuclear myosin VI regulates the spatial organization of mammalian transcription initiation. bioRxiv. https://doi.org/10.1101/2020.04.21. 053124 5. Mao YS, Zhang B, Spector DL (2011) Biogenesis and function of nuclear bodies. Trends Genet 27(8):295–306. https://doi.org/10. 1016/j.tig.2011.05.006 6. Kennedy BK, Barbie DA, Classon M, Dyson N, Harlow E (2000) Nuclear organization of DNA replication in primary mammalian cells. Genes Dev 14(22):2855–2868. https://doi. org/10.1101/gad.842600 7. Misteli T, Soutoglou E (2009) The emerging role of nuclear architecture in DNA repair and genome maintenance. Nat Rev Mol Cell Biol 10(4):243–254. https://doi.org/10.1038/ nrm2651 8. Rust MJ, Bates M, Zhuang X (2006) Subdiffraction-limit imaging by stochastic optical reconstruction microscopy (STORM). Nat Methods 3(10):793–795. https://doi.org/ 10.1038/nmeth929 9. Bates M, Jones SA, Zhuang X (2013) Stochastic optical reconstruction microscopy (STORM): a method for superresolution fluorescence imaging. Cold Spring Harb Protoc 6:

498–520. https://doi.org/10.1101/pdb. top075143 10. Fili N, Hari-Gupta Y, Dos Santos A, Cook A, Poland S, Ameer-Beg SM, Parsons M, Toseland CP (2017) NDP52 activates nuclear myosin VI to enhance RNA polymerase II transcription. Nat Commun 8(1):1871. https://doi.org/10.1038/s41467-01702050-w 11. Pageon SV, Nicovich PR, Mollazade M, Tabarin T, Gaus K (2016) Clus-DoC: a combined cluster detection and colocalization analysis for single-molecule localization microscopy data. Mol Biol Cell 27(22): 3627–3636. https://doi.org/10.1091/mbc. E16-07-0478 12. Malkusch S, Endesfelder U, Mondry J, Gelleri M, Verveer PJ, Heilemann M (2012) Coordinate-based colocalization analysis of single-molecule localization microscopy data. Histochem Cell Biol 137(1):1–10. https:// doi.org/10.1007/s00418-011-0880-5 13. Dos Santos A, Cook AW, Gough RE, Schilling M, Olszok NA, Brown I, Wang L, Aaron J, Martin-Fernandez ML, Rehfeldt F, Toseland CP (2021) DNA damage alters nuclear mechanics through chromatin reorganization. Nucleic Acids Res 49(1):340–353. https://doi.org/10.1093/nar/gkaa1202 14. Toseland CP (2013) Fluorescent labeling and modification of proteins. J Chem Biol 6(3): 85–95. https://doi.org/10.1007/s12154013-0094-5 ´ , Fili N, Hari-Gupta Y, Gough RE, 15. dos Santos A Wang L, Martin-Fernandez M, Aaron J, Waite E, Chew T-L, Toseland CP (2020) Binding partners regulate unfolding of myosin VI to activate the molecular motor. bioRxiv. https:// doi.org/10.1101/2020.05.10.079236 16. Emmanouilidis I, Fili N, Cook AW, HariGupta Y, Dos Santos A, Wang L, MartinFernandez ML, Ellis PJI, Toseland CP (2021) A targeted and tuneable DNA damage tool using CRISPR/Cas9. Biomolecules 11(2). https://doi.org/10.3390/biom11020288

Chapter 21 Single-Molecular Quantification of Flowering Control Proteins Within Nuclear Condensates in Live Whole Arabidopsis Root Alex L. Payne-Dwyer and Mark C. Leake Abstract Here we describe the coupled standardization of two complementary fluorescence imaging techniques and apply it to liquid-liquid phase-separated condensates formed from an EGFP fluorescent reporter of flowering control locus A (FCA), a protein that associates with chromosomal DNA in plants during epigenetic regulation of the flowering process. First, we use home-built single-molecule Slimfield microscopy to establish a fluorescent protein standard. This sample comprises live yeast cells expressing Mig1 protein, a metabolic regulator which localizes to the nucleus under conditions of high glucose, fused to the same type of EGFP label as for the FCA fusion construct. Then we employ commercial confocal AiryScan microscopy to study the same standard. Finally, we demonstrate how to quantify FCA-EGFP nuclear condensates in intact root tips at rapid timescales and apply this calibration. This method is a valuable approach to obtaining single-molecule precise stoichiometry and copy number estimates of protein condensates that are integrated into the chromosome architecture of plants, using confocal instrumentation that lacks de facto single-molecule detection sensitivity. Key words Biomolecular condensates, Liquid-liquid phase separation, Single-molecule counting, Microscope calibration, AiryScan, Slimfield

1

Introduction Flowering at the correct time represents a vital evolutionary pressure on a vast majority of commercial crop plants. The timing of each flowering event is largely predicated on exposure to cold winter temperatures, as monitored in Arabidopsis thaliana by the regulatory state of flowering locus C (FLC) [1]. Prior to cold exposure, the FLC gene is repressed by an autonomous pathway, in which 30 processing of antisense transcripts is linked with chromatin-silencing markers and transcriptional changes [2]. Flowering control locus A (FCA) is one of the RNA-processing proteins involved that directly interact with the polyadenylation machinery

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6_21, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

311

312

Alex L. Payne-Dwyer and Mark C. Leake

at FLC [3]. In addition to prion-like domains, FCA contains two RNA recognition motifs which indicate functional self-assembly [4]. Indeed, FCA has been discovered to form nuclear condensates in living cells via liquid-liquid phase separation [5]. Phase-separated condensates are strongly evidenced in facilitating gene regulation in a diverse range of cases across living kingdoms [6–8], often as a direct result of their physical properties such as locally increased concentration and buffering of functional molecules [9]. The number of active molecules within each condensate is therefore of interest. An investigator might reasonably then task themselves to detect these nuclear condensates in the living organism and to quantify the local reservoir of functional protein, in this case FCA, within each. However, the nuclear condensates cannot be isolated without perturbation [10], nor are typical commercial microscopes capable of single-molecule quantification at the millisecond timescales on which smaller nuclear bodies diffuse [5, 11]. Conversely, singlemolecule imaging techniques are limited when applied to multicellular tissues due to scattering and autofluorescence, particularly in plants [12]. While liquid-phase-separated condensates have been largely investigated using in vitro analogs, there remains a keen demand for more accessible experiments in vivo. To overcome these issues, we describe here the coupled standardization of these two complementary fluorescence imaging techniques. First, we use dedicated single-molecule Slimfield microscopy [13, 14] to establish a known amount of labeled protein in a standard sample; second, we employ confocal AiryScan microscopy to study the same standard; and finally, we interrogate nuclear condensates in whole, wild-type root tips at rapid timescales and apply the calibration. This correspondence between instruments is well defined since fluorescent protein fusions obtain a consistent brightness under standard imaging and biochemical conditions [15]. The concept is simple: we establish the singlemolecule sensitivity of one instrument and lend it to a second microscope which allows superior throughput, ease of use, and imaging contrast (Fig. 1). In our example, the standard sample consists of live yeast cells bearing protein fused to the same enhanced green fluorescent protein (EGFP) label [16]. This protein is the metabolic regulator Mig1 which localizes to the nucleus under conditions of high glucose [16–19]. This approach presents the same chromophore in a similar biochemical environment to the system of interest: FCA-EGFP in the plant nucleus. Finally, we describe the calibration of the AiryScan root and yeast data using the Slimfield standard, to convert the local fluorescence intensity into the number of labeled proteins per condensate.

Quantifying Protein in Biomolecular Condensates in Live Arabidopsis

313

Fig. 1 Flowchart describing the process of lending the single-molecule sensitivity of Slimfield microscopy to the AiryScan confocal microscope suitable for plant root imaging. The process relies on a combination of standardized imaging, a careful choice of standard sample, and reproducible analyses

2

Materials Recommended: A microscope capable of efficient and unequivocal detection of single molecules of EGFP; here we use a custom Slimfield microscope described previously (see 4.4, steps 1 and 2) with a high-performance oil-immersion objective (Nikon 100). Again, a 488 nm wavelength laser line is required for EGFP excitation. Required: A high-performance confocal microscope (Zeiss LSM880 + AiryScan) with a high magnification immersion objective (Zeiss 63) and acquisition and processing software (Zeiss, Zen Black). This instrument must be capable of direct imaging of the desired fusion in plant nuclei in live roots. A 488nm wavelength laser line is required for EGFP excitation. For growing plants: l

Arabidopsis thaliana seeds: – Of a specified accession with stable transgenic fusion: pFCA:: FCA-eGFP [5] in Landsberg erecta [20] – Of the control strain lacking the fusion

l

5% Hypochlorite bleach in UHQ H2O.

l

1 L Autoclaved Murashige and Skoog media (4.3 g MS0 powder + 0.5 g MES buffer + 1L UHQ H2O, adjusted to pH 5.7 with NaOH).

l

1 L Autoclaved Murashige and Skoog media as above + 1% w/w Difco Bacto agar.

314

Alex L. Payne-Dwyer and Mark C. Leake l

100 mm Square petri dishes (plates).

l

300 mL 2% w/w Agarose in UHQ H2O.

l

3M Micropore tape.

l

Aluminum foil.

l

Sterile razor blades and tweezers.

l

Laminar flow cabinet.

l

Cold room or refrigerator at 4  C.

l

l

Growth chamber at 22  C (programmable light cycle, humidity, and temperature control). Racks for holding plates vertically. For growing standard yeast sample:

l

Saccharomyces cerevisiae (e.g., on frozen stab at 80  C): – Of a specified background (BY4741) with Mig1-EGFP fusion [16] – Of the control strain lacking the fusion

l

Standard petri dishes

l

Loose-top vials.

l

Autoclaved yeast nutrient broth (YNB) liquid media + 0.2% w/w glucose.

l

Autoclaved YNB liquid media + 4% w/w glucose.

l

Autoclaved yeast extract–peptone–dextrose (YEP) media + 4% w/w glucose + 1% w/w agarose.

l

Shaking incubator at 30  C.

l

Plate incubator at 30  C.

For aligning the AiryScan detector and for preparing yeast and plant slides: l

Fluorescent beads matched to AiryScan excitation line: TetraSpeck™ Microspheres, 0.1 μm diameter, Invitrogen, 10 μL.

l

Prolong Diamond mountant, 20 μL.

l

Gene Frames 125 μL, Thermo Fisher.

l

Standard microscope slides, 26  76 mm, cleaned.

l

l

Oblong coverslips, #1.5 (170 um), 22  50 mm, cleaned (see Note 4.3, step 1). An air-fed, benchtop, radio-frequency plasma cleaning unit and vacuum pump (Harrick Plasma), or 10 mL conc. HCl and approx. 20mL absolute ethanol. For processing:

l

A multicore workstation or PC capable of high-performance video processing.

Quantifying Protein in Biomolecular Condensates in Live Arabidopsis

3

315

l

ImageJ (open-source Fiji distribution).

l

PYSTACHIO (Python based) or ADEMScode (MATLAB based): Particle tracking software suitable for low signal-tonoise input [20, 21].

l

CoPro (MATLAB-based) image processing software [22].

Methods

3.1 Preparation of Live Arabidopsis Roots for Microscopy

The aim of this section is to grow seedlings vertically and when ready, roots are easily mounted on a slide for AiryScan imaging. 1. For each strain, add 20–30 seeds to separate, labeled screw-cap vials (see Note 4.1, step 1). Add 5% v/v bleach and shake for 5 min. Rinse three times in UHQ water. 2. Pour the MS agar medium into six square petri dishes. Seal and leave to set (see Note 4.1, step 2). 3. When the plates have set, sow the seeds: three plates for each strain. Seal the plates with Micropore tape (see Note 4.1, step 3). 4. Stratify the seeds by storing the plates in darkness (cover with foil) at 4  C for 2–7 days. 5. Grow the plants by setting the plates upright in the growth chamber and leaving them for 5–7 days (see Note 4.1, step 4). 6. On the day of imaging, slides should be prepared less than 1h before microscopy (see Note 4.1, step 5). Prepare an agar pad by adhering the Gene Frame to a clean slide. Melt the agarose and mix 500 μL hot agarose with 500 μL of imaging media (for plants, MS liquid media without agar). Fill the well immediately with 500 μL mixture and skim off the excess using a slide to leave a flat agar surface. Wait for 5 min for this to set. 7. Select roots that are 1–3 cm long, without mold or damage. Gently pull out each seedling by the leaves using tweezers. Pipette 10 μL of liquid MS media onto the agar pad. Cut off the final 8–10 mm of root tip and place this in the center of the pad. Seal with a large coverslip, taking care to avoid trapping air. Use within 1 h (see Note 4.1, step 5).

3.2 Preparation of a Standard Yeast Sample for MICROSCOPY

The aim of this section is to grow yeast under controlled conditions including glucose supply, such that the Mig1 protein segregates to the nucleus. The yeast should present as a flat, sparse monolayer suitable for both Slimfield and AiryScan imaging. 1. Thaw each yeast stab and streak under sterile atmosphere onto separate YEP agar plates for each strain. Incubate for 48 h at 30  C (see Note 4.2, step 1).

316

Alex L. Payne-Dwyer and Mark C. Leake

2. Seal the plates with Parafilm and store at 4  C (for up to one month). 3. On the day before imaging, set up the clean laminar flow hood. Add 7 mL liquid YNB media at 4% glucose to each of the five 15 mL vials for each strain. To the first, add isolated colonies from the corresponding plate. 4. Then vortex to mix, transfer 1 mL into the next vial, and repeat to get successive eightfold dilutions: 1, 1/8, 1/64, 1/1024, and 1/4096 (see Note 4.2, step 2). Leave these in the incubator overnight (12–18 h) (see Note 4.2, step 3). 5. On the day of imaging, measure the optical densities (OD600) and select the culture closest to OD 0.4–0.5. Dilute again 5 into fresh YNB liquid media at 4% glucose and leave for 1 h in the 30  C incubator. 6. Construct an agarose pad as above (see Subheading 3.1, step 6) but using YNB liquid media at 4% glucose instead of MS media. Spot 2 μL of culture onto a Gene Frame agarose pad and seal with the coverslip. 3.3 Preparation of an AiryScan Alignment Sample

The aim of this section is to prepare a bright, stable control sample with which the AiryScan excitation laser, pinhole, and secondary detector can be mutually aligned. 1. Sonicate the fluorescent beads for 15 min. 2. Place a clean coverslip (see Note 4.3, step 1) on a clean benchtop and pipette 5 μL undiluted bead stock in the center. Spread the drop out with the pipette tip and leave to dry for 5 min. 3. Place a clean slide on the benchtop and pipette 10 μL of Prolong Diamond onto the center of the slide. 4. Add the coverslip with the bead-coated side facing downwards. Leave with the coverslip facing down in the dark for 16 h (overnight) to allow the mountant to cure with the beads in place. Store at 4  C (see Note 4.3, step 2).

3.4 Optional: Slimfield Imaging of the Standard Sample

This optional section aims to generate Slimfield photobleaching sequences (see Note 4.4, step 1) of yeast nuclei in a high-glucose condition, for the purposes of evaluating (i) the initial intensity of each yeast nucleus, (ii) the characteristic brightness of a single EGFP molecule in vivo, and thus (iii) the number of molecules in a yeast nucleus. Since literature values of the number of Mig1EGFP per cell are available (see Note 4.6, step 3), this section is not strictly required for a basic AiryScan calibration but is recommended for improved accuracy.

Quantifying Protein in Biomolecular Condensates in Live Arabidopsis

317

1. Align the Slimfield microscope [20] (see Notes 4.4, steps 1–3) with 488 nm excitation using the fluorescent bead sample. 2. Mount the Mig1-EGFP yeast slide. Find the cells in bright field and focus on the cell midbody. 3. Record a bright-field image for later segmentation and inspection. 4. Set the fluorescence acquisition parameters (see Note 4.4, step 4) such that (i) the sample is not initially saturating the camera (Fig. 2a, top center), and (ii) when the sample is undergoing photobleaching and almost spent, individual EGFP molecules can be seen blinking across individual frames. Images should be saved as 16-bit grayscale with camera metadata in OME TIFF format. 5. Set the number of frames to acquire sufficiently large to capture this photoblinking regime. 6. Record the standardized excitation and emission settings, including laser power and filter types. 7. Change to bright field and find a new, pristine yeast cell. 8. Acquire in fluorescence mode with the same exposure and power settings, ensuring that the sample is not exposed to the laser until the first frame of the recording takes place (see Note 4.4, step 5). 9. Repeat with the same settings to obtain >50 cells from each of at least 3 independent cultures. 10. Repeat for the unlabeled strain. 3.5 AiryScan Imaging of the Standard and Root Samples

This section details the process of aligning the AiryScan instrument (see Note 4.5, step 1), followed by imaging roots and then the standard yeast sample. 1. Mount the alignment bead sample with the 63 oil-immersion objective. Set the AiryScan “RS” mode and turn on the laser at 1% power. On the Maintenance tab, select “Adjust in Continuous mode” to allow the pinhole to move in response to the AiryScan detector. Refocus onto the beads in “Continuous” mode until the signal is centered on the AiryScan detector. Select “Store position” and deselect “Adjust in Continuous mode.” The AiryScan is now ready to use for a dim sample. 2. Mount the FCA-EGFP root sample. Use “Locate” mode to center the root tip under bright-field illumination. Change to Acquisition tab and use “overview”-style settings at low zoom to ensure that the nuclei are in focus (Fig. 2b). 3. Save an acquisition so that these overview settings can be reproduced with the “Reuse” function in Zen Black. Images should be saved as 16-bit grayscale with metadata in .CZI format.

318

Alex L. Payne-Dwyer and Mark C. Leake

Fig. 2 Representative acquisitions for the Mig1-EGFP yeast culture and FCA-EGFP root. (a) Yeast cells in highglucose condition expressing Mig1-GFP that localizes to the nucleus, with segmented nuclear and cellular boundaries where visible; (first row) Slimfield microscopy and (second row) LSM880 confocal after AiryScan processing; (left) transmitted light, (center) Mig1-EGFP fusion strain, and (right) parental control showing autofluorescence. Scale bar 2 μm. (b, c) LSM880 confocal images of an FCA-EGFP-labeled Arabidopsis root; (b) in zoomed-out overview: (top) transmitted light and (bottom) fluorescence channels; scale bar 10 μm. (c) A single epidermal nucleus, in single slice and maximum intensity projection along z, before and after AiryScan processing. These frames show spots corresponding to biomolecular condensates of FCA-EGFP. The condensates reside in the nucleoplasm surrounding the nucleolus (central dark area). Scale bar 5 μm

4. Choose the cell type of interest. These must occur at a consistent layer (epidermal or cortex) and distance from the root tip (meristem, dividing zone, or elongation zone). FCA condensates are typically brighter/higher contrast in cortex nuclei in the dividing region.

Quantifying Protein in Biomolecular Condensates in Live Arabidopsis

319

5. Explore the AiryScan mode settings to determine a suitable standard imaging configuration (see Notes 4.5, steps 2 and 3) which can capture FCA condensates of all sizes and intensities. 6. Fix and record these settings by saving the acquisitions so that they can be recovered by means of the “Reuse” function. 7. Acquire and save z-stacks of the root nuclei of the desired type using the standard settings. Repeat to obtain >50 cells from 3 independent cultures. 8. Repeat for the unlabeled control strain. 9. Mount the yeast sample. Acquire and save z-stacks of the standard yeast cells using the standard settings (Fig. 2a, second row). Repeat to obtain >50 cells from 3 independent cultures. 10. Repeat for the unlabeled control strain. 11. Navigate to the Processing tab. Perform batch-wise “AiryScan Processing” with a suitable choice of processing “Strength” to generate super-resolved volumes without artifacts (see Note 4.5, step 4). 3.6 Single-Molecule Brightness Using Literature Estimates

The purpose of this section is to analyze the newly acquired images to extract the characteristic brightness of a single molecule [20] in the AiryScan data, here termed Isingle(A). 1. Collate the AiryScan processing output and, if available, Slimfield image sequences on the workstation. 2. Load an AiryScan z-stack of the standard yeast data into ImageJ using the “BioFormats” plug-in. 3. Perform a maximum intensity projection to collapse all the z-slices (see Note 4.6, step 1). Save as TIFF. 4. Segment the yeast nuclei to generate binary pixel masks, for example using local thresholding in ImageJ, and then iterating to optimize the heuristics (see Note 4.6, step 2). CoPro software [22] requires input as a “_segmentation.mat” MATLAB array file. To generate this, use ImageJ to save as binary TIFF, then use MATLAB to convert this to a binary array, and save as the .mat file. 5. Measure the total fluorescence intensity of the nucleus in pixel gray value units using CoPro software with the dummy input Isingle ¼ 1. Use a dark pixel value (typically bias of 100) as the “background,” and the segmentation masks. Use the default option for a simulated point-spread function. 6. The output of CoPro is a .mat file detailing the brightness of each yeast nucleus in terms of counts. This gives the raw integrated count (RIC) for each nucleus. It also gives the size of the segmented area.

320

Alex L. Payne-Dwyer and Mark C. Leake

Fig. 3 Violin plots of fluorescent integrated counts (area corrected) of yeast nuclei under AiryScan microscopy as obtained from CoPro with dummy input of Isingle ¼ 1 for control and Mig1-EGFP (n ¼ 44, 50 nuclei, respectively). The individual nuclei are shown as circles, with boxplot overlaid. The difference between population mean counts (black crosses) is proportional to both the mean number of Mig1-EGFP molecules in each nucleus, N, and the characteristic Isingle(A), the number of counts associated with each EGFP in every AiryScan standard acquisition

7. Repeat for all the other labeled AiryScan yeast sequences; also repeat for the unlabeled parent strain z-stack, where the RIC represents only the total autofluorescence intensity (Fig. 3). 8. For each of the 2 yeast strains, take the mean of the AiryScan RIC over all the corresponding nuclei: RIC ðlabeled Þ ¼

n 1X RIC ðlabeled, nucleus i Þ n i¼1

9. Find the mean copy number of Mig1-EGFP molecules per nucleus, N. This can be verified independently using CoPro on the yeast Slimfield sequences if available (see Section 3.7 below), though estimates are already known for Mig1-EGFP from published work (see Notes 4.6, steps 3 and 4). 10. Take the difference of the two average RIC values, with the control weighted by the mean cell areas in each strain determined by CoPro. Then divide by N to estimate the Isingle value for AiryScan:

Quantifying Protein in Biomolecular Condensates in Live Arabidopsis

Isingle ðA Þ ¼

321

Area ðlabeled Þ RIC ðlabeled Þ  RIC ðcontrol Þ  Area ðcontrol Þ

N

The result should be representative of any EGFP molecule under the same AiryScan imaging conditions, including condensed FCA-EGFP molecules in the plant nuclei. 3.7 Optional: Verification of SingleMolecule Calibration Using Slimfield Imaging

The core use of Slimfield microscopy is to estimate the singlemolecule brightness and the absolute number of localized molecules directly from image sequences. This section is concerned with analyzing the Slimfield imaging of the yeast nuclei to refine the Isingle calibration above. 1. Use PySTACHIO [23] or ADEMScode [21] software to track the Slimfield images of yeast with Mig1-EGFP. 2. The integrated pixel value brightness of one EGFP molecule in Slimfield data, Isingle(S), can be estimated quickly by calculating the modal spot brightness after 1/3 of the photobleaching has occurred. This is an inbuilt feature of PySTACHIO and is governed by the “InVivoIsingle” script in ADEMScode. Alternatively, identification of consistent stepwise photobleaching is a more robust method of finding Isingle(S) for Slimfield [24]. 3. Segment individual nuclei in the Slimfield data using the inbuilt tools in PySTACHIO or ADEMScode for yeast nuclei. Generate masks for each field of view that will be included in the next step, CoPro. 4. Run CoPro in MATLAB on the Slimfield data, with input including the Isingle(S), a dark pixel value (typically bias of 100) as the “background,” and the segmentation masks. Use the default option for a simulated point-spread function. 5. The output of CoPro is a .mat file detailing the brightness of each yeast nucleus in terms of the number of equivalent EGFP molecules. 6. Repeat steps 3–5 for the parental control strain. 7. Take the average and SEM of the copy numbers in the EGFPlabeled and control cases. Take the difference of these two numbers. This result is N, the number of actual molecules of Mig1-EGFP per nucleus. This empirical result can be used in the calculation in Subheading 3.6 instead of the literature value, to refine Isingle(A), the brightness counts corresponding to a single EGFP in the AiryScan data. 8. Ensure to propagate the SEM uncertainty on each value (by addition in quadrature) from the CoPro output, through to the estimation of the number of molecules per nucleus, N, and finally Isingle(A), based on the number of nuclei, n, included in the calculation. For n ¼ 50, the relative error in Isingle(A) is expected to fall around 10%.

322

Alex L. Payne-Dwyer and Mark C. Leake

3.8 Applying the Calibration to the Analysis of AiryScan Plant Root Images

In this section the calibration is applied to the AiryScan data for FCA-EGFP-bearing plant nuclei. The trajectory analysis divides the integrated counts of each trajectory’s initial time point by the Isingle(A) brightness; this refinement yields stoichiometry not simply per cell or per nucleus, but of FCA-EGFP molecules per condensate focus. 1. Track the roots using PySTACHIO or ADEMScode software to detect foci and link them into trajectories. 2. Determine the cumulative uncertainty in the stoichiometry rescaling by adding in quadrature: (i) the fractional error in Isingle(A) to (ii) the baseline Slimfield sensitivity of 0.7 EGFP molecules (see Note 4.6, step 3). 3. Prepare the trajectory analysis script (“TrackAnalyser” in ADEMScode). Set the input Isingle to the identified value of Isingle(A), and the kernel density width, params.stoichKDW, to the total rescaling uncertainty. 4. The stoichiometry distribution will be displayed as a kernel density estimate (see Fig. 4) and saved to the output.mat file. 5. The Isingle(A) value can be applied to other datasets for which the settings are maintained—that is to say, the number of counts per EGFP per frame remains consistent with the standard yeast sample (see Notes 4.6, steps 5 and 6).

Fig. 4 Stoichiometry of FCA-EGFP assemblies within analyzed AiryScan volume stacks in live root tips (n ¼ 2 replicates, 22 nuclei, 163 foci). Scaled from integrated intensity counts by Isingle(A) ¼ 134  15 PGV. Kernel density estimate with kernel of 0.71 molecules

Quantifying Protein in Biomolecular Condensates in Live Arabidopsis

4

323

Notes

4.1 Preparation of Live Roots for Investigation

1. Dry seeds are very susceptible to electrostatic dispersal, so avoid pouring/handling them and use a long metal spatula instead. 2. If roots are stressed, such as by dehydration, they may present aberrant phenotypes including high autofluorescence. Ensure that the depth of the agar on the plates is at least 8 mm (around 70 mL) and that the plates are closed when the agar is cooling. 3. It is helpful to draw a horizontal line across the top half of the underside of the plate as a guide for sewing. A pipettor is too aggressive; use a 1000 μL pipette tip and a gloved finger to provide fine control, so the wet seeds are dispensed individually. The tape retains water while allowing gas exchange; do not use impermeable tape. 4. The Arabidopsis growth chamber should be set to a fixed temperature, humidity, light intensity, and light duration. Here, 22 oC and 50% relative humidity with a 16-h day/8h night cycle are recommended. Vernalization in a cold chamber may be added after the growth period but is not required for this calibration protocol. 5. The cut roots are susceptible to osmotic shock, so MS media is preferred over UHQ water for imaging. If prepared correctly each root will last for about 1 h before degrading; the signature that must be avoided is if the nuclei begin to disperse and debris accumulates inside the cells. Once this occurs discard the roots and cut another set for a new slide. This protocol is intended for short-term endpoint imaging. For time-lapse imaging, whole seedlings must be mounted and other influences such as root growth must be accommodated [25, 26].

4.2 Preparation of the Standard Yeast Sample

1. Yeast inocula are susceptible to contamination by fast-growing bacteria, so all culturing is best performed in sterile atmosphere. 2. The serial dilutions must match the incubation time. Based on an optimal yeast doubling time of about 1.5 h, then an overnight gap of 12–18 h is covered by dilutions around 2(-12/1.5) ¼ 1/256 and 2(-18/1.5) ¼ 1/4096. Additional variability arises since the agar-grown cultures do not homogenize easily under vortexing and the amount inoculated from the plate into each vial is not easily controlled. Multiple dilutions are necessary to provide contingency. 3. Pull the lids up after setting the vials in the shaking incubator to make sure that they are not airtight; otherwise the culture will become anaerobic.

324

Alex L. Payne-Dwyer and Mark C. Leake

4.3 Preparation of an AiryScan Alignment Sample

1. Coverslips must be free of fluorescent contaminants, or the yeast sample will suffer from extraneous background. To clean slides and coverslips in tandem, either use a low-pressure air plasma cleaner (Harrick) for 10 min or clean the coverslips for 30 min in 1% hydrochloric acid (7 mL conc. HCl in 250 mL UHQ H2O on a shaker). Rinse thoroughly in UHQ H2O. Incubate in absolute ethanol on a shaker for 30 min. Leave glass to dry. Store the slips for up to a month in a sealed container such as a Falcon tube or slide box. 2. Once the bead sample has cured, store it in the fridge for up to one month. Do not freeze it. Vectashield or glycerol can be used as an alternative mountant, but the beads will dislodge more easily.

4.4 Slimfield Imaging

1. Details of our bespoke Slimfield setup comprising off-the-shelf optics and optomechanics are published extensively elsewhere [27–35]. Here, only single-color imaging in GFP channel is required, though the major determinants of performance are (i) the correct matching of emission filters to the fluorophore and (ii) the quality of the scientific camera (here a Photometrics Prime 95B running at the highest gain setting). Suitable commercial alternatives for microbiological standard samples include TIRF-capable microscopes such as the Oxford NanoImager (ONI). 2. Slimfield reveals the quantitative characteristics of individual tracked bodies in greater detail than simply the number of foci and their spatial (co)distribution within the cell. Information on the diffusive mobility of bodies and that of their internalized molecules can be obtained, as well as the estimates of stoichiometry and protein copy that are discussed here. Methods to explore the mobility of loci in Arabidopsis using complementary techniques are described elsewhere [36]. 3. While the excitation beam must be collimated (as in a TIRF microscope), the yeast nuclei must be imaged at normal beam incidence since they do not rest on the coverslip. The excitation beam must be wide enough that the illuminated area containing at least one whole nucleus is appreciably uniform. The emission output of one molecule is then approximately the same, regardless of its location in the nucleus. 4. For all fluorescence-based intensity measurements, there is a trade-off between trajectory length and frame rate. If the laser power is too high, the sample will photobleach very quickly and the tracks of individual molecules will be short. If the power is too low, then the exposure time must be extended (frame rate reduced) to retain single-molecule detection in each frame. For our implementation, the region of interest is 190  190 pixels

Quantifying Protein in Biomolecular Condensates in Live Arabidopsis

325

¼ 10  10 μm at 50 nm/pixel; the typical exposure time is 5 ms (189 frames per second); the laser beam is 14 μm wide (Gaussian 1/e2 diameter) and the peak intensity is 5 kW/cm2. 5. The sample can be protected from premature photobleaching either by digital control of the laser or by control of the excitation shutter. The signal will therefore not be photobleached prior to the acquisition and all molecules of Mig1-EGFP will be visible in the first few frames. 4.5 AiryScan imaging

Zeiss AiryScan instruments [37] use a second photodetector to measure the real confocal point-spread function during the acquisition of each confocal image. This information is then used in postprocessing steps to undo the blur from optical diffraction. A carefully chosen deconvolution step will refine the spatial resolution as low as 150 nm. Protocols are available for generating confocal z-stacks of Arabidopsis nuclei [36]. However, the requirements for fixing the settings are much stricter here. Once the settings are standardized, they cannot be adapted later to suit unseen samples. Do not change any settings or optics on either the excitation or the emission paths, just the slide and the stage and focal positions. These filters, power, and exposures must remain fixed, or the subsequent calibration will become invalid; the protocol here would need to be repeated. There is a clear benefit of the “Reuse” function on a wellmaintained microscope with multiple users. The laser power and pixel dwell time must be adjusted to give best contrast for smaller, diffraction-limited FCA bodies (Fig. 2c). The region of interest, z-slices, and z-spacing must be set so that an entire nucleus is captured to avoid undercounting. A suitable starting point for FCA-EGFP is recommended as follows: AiryScan “RS” acquisition mode, and confocal region of interest: 9  9 μm ¼ 100  100 pixels at pixel size 90 nm. This reduces to 7.6  7.6 μm per slice after AiryScan processing. Laser power: 4% of source max ~0.8 mW. Exposure: 56 ms/slice  15 z-slices at 0.6 μm per step ¼ 0.8 s per volume 9 μm deep. The optimal pixel size is less than the diffraction limit since mild spatial oversampling is preferred for super-resolved tracking of foci. The photobleaching decay time of EGFP under these conditions is about 1 min of exposure ~1000 slices, or ~50 cycles. The Wiener deconvolution used in the AiryScan processing is a nonlinear high-pass filter, so it is vulnerable to introducing severe spotlike artifacts at low input SNR. For dim samples, the filter strength needs to be kept low to avoid extra false positives when tracking. Filter strengths of 5.0 are recommended for z-stacks and just 2.3 for still images, instead of the “Auto” Zen Black settings.

326

Alex L. Payne-Dwyer and Mark C. Leake

4.6 Single-Molecule Calibration, Verification, and Implementation

1. The maximum intensity projection is only necessary to collapse the XYZT sequences to representative XYT images, so that all (local maxima) can be tracked, since the tracking software is designed for 2D image sequences over time. Segmentation is also much simpler in 2D. However, this step necessarily makes some assumptions about the sparsity of the fluorophore distribution. There is a limitation that if the number density of condensates is very high then some local maxima will be lost in the projection and not detected, or their assignment to tracks made more ambiguous. Since the maximal intensity projection also skews the pixel values higher as a function of the number of slices, the number of slices should also be kept constant for the calibration to hold. Alternatively, one could track all the individual frames across z and t, though these would be noisy and computationally challenging and risk overcounting the molecules, were any to drift in z position during the scan. 2. CoPro [22] uses a traditional image processing pipeline, designed for supervised segmentation in the ADEMScode suite that is dependent on manual input and/or optimizing heuristics. These heuristics include the number and order of morphological operations, such as closing and eroding, which must be determined by trial and error. ImageJ is also capable of performing segmentation in this way. A benefit of ImageJ and MATLAB is that the interfaces allow intuitive visual assessment of each iterative step or result. However, there is a wealth of alternative, less heavily supervised methods for segmenting 2D images of yeast cells, such as convolutional neural networks which could scale more readily to larger image sets [27]. 3. The standard Slimfield estimate is 806  353 molecules (mean  SD.) of Mig1-EGFP per cell at high-glucose condition, of which 226  155 reside in the nucleus [16]. The resulting Isingle(A) value is 134  13 counts/molecule EGFP (n ¼ 50 cells, 3 replicates). Our abbreviated calibration method relies on this literature estimate and thus assumes that the mean total copy of Mig1 protein is consistent and reproducible for each population of yeast cells [18]. Ideally one would perform the AiryScan and Slimfield standard acquisitions on the day with the same standard slide(s), and then repeat on two more days to generate three biological replicates. However, the variance in the original Slimfield measurement of the stepwise photobleaching of single molecules (0.7 per EGFP) tends to dominate over the uncertainty in the AiryScan measurements. 4. We do not make a correction for any maturation effect for EGFP, but a separate protocol is available to characterize this if appropriate [20]. Under the assumptions of fast maturation, robustness to environmental conditions, and simple three-state

Quantifying Protein in Biomolecular Condensates in Live Arabidopsis

327

fluorescence behavior, this protocol may not be limited to EGFP, but may be useful for exploring calibration of other constitutively fluorescent proteins. 5. The method assumes that the brightness of the fusion protein is indistinguishably similar in the two samples. For deeper imaging within the plant, the scattering loss will increase. Since the yeast is imaged at the surface, it necessarily cannot reflect the reduction in effective molecular brightness as a function of depth. This means that the calibration strictly provides a lower bound estimate on the stoichiometry in the plant nuclei, which becomes less accurate with depth. 6. Applying the calibration provides for single-molecule resolution between peaks in the stoichiometry but cannot change the fundamental sensitivity cutoff (noise floor) of the AiryScan instrument. As such, individual, isolated molecules are likely to remain undetected by the AiryScan instrument. This is visible as a steep shoulder at the lower end of the AiryScanderived stoichiometry distribution (Fig. 4).

Acknowledgments Thanks to Geng-Jen Jang and Caroline Dean (John Innes Centre, Norwich) for donating Arabidopsis seeds for the FCA-EGFP fusion construct, and the Bioscience Technology Facility (BTF) at the University of York for supporting confocal microscopy resources. This work was funded by EPSRC (EP/T002166/1). References 1. Berry S, Dean C (2015) Environmental perception and epigenetic memory: mechanistic insight through FLC. Plant J 83(1):133–148 2. Wu Z, Fang X, Zhu D et al (2020) Autonomous pathway: flowering locus c repression through an antisense-mediated chromatinsilencing mechanism. Plant Physiol 182(1): 27–37 3. Xu C, Wu Z, Duan HC et al (2021) R-loop resolution promotes co-transcriptional chromatin silencing. Nat Commun 12(1):1790 4. Macknight R, Bancroft I, Page T et al (1997) FCA, a gene controlling flowering time in Arabidopsis, encodes a protein containing RNA-binding domains. Cell 89(5):737–745 5. Fang X, Wang L, Ishikawa R et al (2019) Arabidopsis FLL2 promotes liquid–liquid phase separation of polyadenylation complexes. Nature 569(7755):265–269 6. McSwiggen DT, Mir M, Darzacq X et al (2019) Evaluating phase separation in live

cells: diagnosis, caveats, and functional consequences. Genes Dev 33(23–24):1619–1634 7. Sabari BR, Dall’Agnese A, Young RA (2020) Biomolecular Condensates in the Nucleus. Trends Biochem Sci 45(11):961–977 8. Emenecker RJ, Holehouse AS, Strader LC (2020) Emerging Roles for Phase Separation in Plants. Dev Cell 55(1):69–83 9. Holehouse AS, Pappu RV (2018) Functional Implications of Intracellular Phase Transitions. Biochemistry 57(17):2415–2423 10. Pliss A, Levchenko SM, Liu L et al (2019) Cycles of protein condensation and discharge in nuclear organelles studied by fluorescence lifetime imaging. Nat Commun 10(1):455 11. Bayguinov PO, Oakley DM, Shih CC et al (2018) Modern Laser Scanning Confocal Microscopy. Curr Protoc Cytom 85(1):e39 12. Berthet B, Maizel A (2016) Light sheet microscopy and live imaging of plants. J Microsc 263(2):158–164

328

Alex L. Payne-Dwyer and Mark C. Leake

13. Lenn T, Leake MC (2012) Experimental approaches for addressing fundamental biological questions in living, functioning cells with single molecule precision. Open Biol 2(6): 120090 14. Plank M, Wadhams GH, Leake MC (2009) Millisecond timescale slimfield imaging and automated quantification of single fluorescent protein molecules for use in probing complex biological processes. Integr Biol 1(10): 602–612 15. Chen Y, Mu¨ller JD, Ruan Q et al (2002) Molecular brightness characterization of EGFP in vivo by fluorescence fluctuation spectroscopy. Biophys J 82(1):133–144 16. Wollman AJM, Shashkova S, Hedlund EG et al (2017) Transcription factor clusters regulate genes in eukaryotic cells. eLife 25(6):e27451 17. Shashkova S, Wollman AJM, Leake MC et al (2017) The yeast Mig1 transcriptional repressor is dephosphorylated by glucose-dependent and -independent mechanisms. FEMS Microbiol Lett. 364(14):fnx133 18. Shashkova S, Nystro¨m T, Leake MC et al (2021) Correlative single-molecule fluorescence barcoding of gene regulation in Saccharomyces cerevisiae. Methods 193:62–67 19. Wollman AJM, Hedlund EG, Shashkova S et al (2020) Towards mapping the 3D genome through high speed single-molecule tracking of functional transcription factors in single living cells. Methods 170:82–89 20. Zapata L, Ding J, Willing EM et al (2016) Chromosome-level assembly of Arabidopsis thaliana Ler reveals the extent of translocation and inversion polymorphisms. Proc Natl Acad Sci U S A 113(28):E4052–E4060 21. Miller H, Zhou Z, Wollman AJM et al (2015) Superresolution imaging of single DNA molecules using stochastic photoblinking of minor groove and intercalating dyes. Methods 88: 81–88 22. Wollman AJM, Leake MC (2015) Millisecond single-molecule localization microscopy combined with convolution analysis and automated image segmentation to determine protein concentrations in complexly structured, functional cells, one cell at a time. Faraday Discuss 184: 401–424 23. Shepherd JW, Higgins EJ, Wollman AJM et al (2021) PySTACHIO: Python single-molecule TrAcking stoiCHiometry Intensity and simulatiOn, a flexible, extensible, beginner-friendly and optimized program for analysis of singlemolecule microscopy data. Comput Struct Biotechnol J 19:4049–4058 24. Leake MC, Chandler JH, Wadhams GH et al (2006) Stoichiometry and turnover in single,

functioning membrane protein complexes. Nature 443(7109):355–358 25. Rahni R, Birnbaum KD (2019) Week-long imaging of cell divisions in the Arabidopsis root meristem. Plant Methods 15(30) 26. Stoeva D, Go¨schl C, Corliss B et al (2017) Long-term confocal imaging of Arabidopsis thaliana roots for simultaneous quantification of root growth and fluorescent signals. Plant Genomics: Methods Mol Biol 1610:169–183 27. Jin X, Lee J-E, Schaefer C et al (2021) Membraneless organelles formed by liquid-liquid phase separation increase bacterial fitness. Sci Adv 7(43):eabh2929 28. Sun Y, Wollman AJM, Huang F et al (2019) Single-organelle quantification reveals stoichiometric and structural variability of carboxysomes dependent on the environment. Plant Cell 31(7):1648–1664 29. Cosgrove J, Novkovic M, Albrecht S et al (2020) B cell zone reticular cell microenvironments shape CXCL13 gradient formation. Nat Commun 11(1):3677 30. Miller H, Cosgrove J, Wollman AJM et al (2018) High-speed single-molecule tracking of CXCL13 in the B-follicle. Front Immunol 9:1073 31. Stracy M, Wollman AJM, Kaja E et al (2019) Single-molecule imaging of DNA gyrase activity in living Escherichia coli. Nucleic Acids Res 47(1):210–220 32. Badrinarayanan A, Reyes-Lamothe R, Uphoff S et al (2012) In vivo architecture and action of bacterial structural maintenance of chromosome proteins. Science 338(6106):528–531 33. Dresser L, Hunter P, Yendybayeva F et al (2021) Amyloid-β oligomerization monitored by single-molecule stepwise photobleaching. Methods 193:80–95 34. Syeda AH, Wollman AJM, Hargreaves AL et al (2019) Single-molecule live cell imaging of Rep reveals the dynamic interplay between an accessory replicative helicase and the replisome. Nucleic Acids Res 47(12):6287–6298 35. Reyes-Lamothe R, Sherratt DJ, Leake MC (2010) Stoichiometry and architecture of active DNA replication machinery in Escherichia coli. Science 328(5977):498–501 36. Meschichi A, Rosa S (2021) Visualizing and measuring single locus dynamics in Arabidopsis thaliana. In: Sanchez-Serrano J.J., Salinas J. (eds) Humana, New York, NY. Arabidopsis Protocols: Methods Mol Biol 2200:213–224 37. Wu X, Hammer JA (2021) ZEISS AiryScan: optimizing usage for fast, gentle, superresolution imaging. In: Brzostowski J., Sohn H. (eds) Humana, New York, NY. Confocal Microscopy: Methods Mol Biol 2304:111–130

INDEX A AFM imaging .......................................................... 44, 46, 49–51, 53, 58, 99 Arabidopsis thaliana ............................................ 313, 315 Atomic force microscopy (AFM) ............................43–47, 49–51, 53–60, 96, 97, 101, 102, 171–180, 217, 253, 254

B Bacillus subtilis ..................................................... 265–294 Bacterial chromosome..................................... 31, 63, 155–168, 194, 265, 267, 268 nucleoid .................................. 64, 155, 156, 194, 288

C Chromatin states ....................................... 18, 212, 227, 230, 238 territories ................................................111–127, 211 Chromosome organization......................31, 38, 111, 156, 269, 272 Cluster analysis ..................................................... 295–311

D Diffusion coefficients ............................................ 194, 195, 203, 204, 206, 214, 227, 230, 237–239 DNA architecture and dynamics ...................................... 253 binding protein ............................................. 6, 64, 68, 89, 129, 193–207, 267, 268, 270, 272 damage............................................60, 112, 171, 194, 195, 198, 204, 266, 268, 270, 273 micromanipulation repair ........................................ 6, 112, 171, 194, 195, 198, 204, 253, 267, 269, 270, 295 replication ............................................. 17, 39, 68, 71, 96, 111, 112, 145–148, 152, 195, 253, 265–269, 272, 295 replication domains ........................................ 112, 126 segregation ......................................71, 117, 269, 272 supercoiling ................................ 43, 80, 96, 101, 254

DnaA....................................................145, 147, 148, 265 Double helix ...................................................... 44–46, 53, 54, 95–97, 101, 215–217, 233, 253

E Epigenetics ........................................................... 111, 115 Escherichia coli ............................................ 31–40, 64, 66, 145–150, 152, 155–158, 165, 194, 195, 197, 198, 200, 270

F Flow cytometry ............................................145–152, 183 Fluorescence microscopy confocal............................................................ 33, 115, 183, 213, 315 epifluorescence .................................................... 75–92 high-throughput ............................................ 279–293 localization microscopy........................................... 193 narrowfield............................................................. 5–15 single-molecule....................................................2, 194 STORM ................................................................... 193 super-resolution ...................................................... 193 Fluorescence recovery after photobleaching (FRAP)...........................................................31–39 Fluorescent proteins.............................................. 5–8, 19, 32, 37, 64, 68, 70, 156, 160, 193–196, 205, 220, 222, 224, 267, 279, 314, 329

G Genome content................................................... 146, 148, 152 copy number................................................... 148, 150 organization........................................... 112, 156, 171

H Histones...........................................................18, 19, 115, 183–190, 231, 233

I Image segmentation................................. 13, 65, 70, 289, 293, 319, 321, 323, 328

Mark C. Leake (ed.), Chromosome Architecture: Methods and Protocols, Methods in Molecular Biology, vol. 2476, https://doi.org/10.1007/978-1-0716-2221-6, © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2022

329

CHROMOSOME ARCHITECTURE: METHODS AND PROTOCOLS

330 Index L

S

Label free imaging interferometric scattering (iSCAT) ............... 129–142 Lambda Red Recombination .............................. 194, 195 Live cell imaging .............................................32, 68, 114, 118, 119, 121, 213, 233

Saccharomyces cerevisiae.......................6, 7, 183, 184, 316 Signal transduction ..................................................... 5–15 Simulations atomistic ..........................................99, 101, 254, 255 molecular dynamics.................................. 95–107, 255 Single-molecule biophysics ......................................................... 39, 252 imaging ............................................... 2, 79, 194, 196, 199, 204, 212, 219, 224, 226–233, 252, 267, 305, 314, 315, 322 tracking ...........................................39, 194, 195, 204, 206, 224, 226–233 Single particle tracking .......................155, 214, 215, 239 Sporulation .................................................................... 277

M Machine learning............................................18, 236, 242 Magnetic tweezers transverse ............................................... 76, 81, 83, 85 Mean squared displacement ....................... 200, 203, 237 MukBEF .......................................................31–34, 37, 39

N Nuclear mechanics ............................................... 171–180 Nucleosome.................................... 18, 19, 212, 214, 226

T

Polymer dynamics ......................................................... 157

Time-lapse microscopy ...................................... 21, 25, 68 Total internal reflection fluorescence (TIRF)............................................. 8, 77–79, 130, 185, 196, 199, 200, 206, 214, 217, 300, 326

R

V

Residence times .......................................... 221, 222, 226, 230, 232, 238, 241

Vegetative growth ......................................................... 269

P