Supramolecular chemistry in water 9783527344673, 3527344675, 9783527814916, 9783527814923

Table of contents :
Content: Water runs deep --
Water-compatible host systems --
Artificial peptide and protein receptors --
Recognition, transformation, detection of nucleotides and aqueous nucleotide-based materials --
Carbohydrate receptors --
Ion receptors --
Coordination compounds --
Aquesous supramolecular polymers and hydrogels --
Foldamers --
Vesicles and micelles --Monolayer-protected gold nanoparticles for molecular sensing and catalysis --
Optical probes and sensors --
probes for medical imaging --
Supramolecular catalysis in water.

Citation preview

­Supramolecular Chemistry in Water Edited by Stefan Kubik

Editor Prof. Dr. Stefan Kubik

Technische Universität Kaiserslautern Fachbereich Chemie ‐ Organische Chemie Erwin‐Schrödinger‐Straße 67663 Kaiserslautern Germany Cover Image: Courtesy of Stefan Kubik, TU Kaiserslautern

All books published by Wiley‐VCH are carefully produced. Nevertheless, authors, editors, and publisher do not warrant the information contained in these books, including this book, to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate. Library of Congress Card No.: applied for British Library Cataloguing‐in‐Publication Data

A catalogue record for this book is available from the British Library. Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at . © 2019 Wiley‐VCH Verlag GmbH & Co. KGaA, Boschstr. 12, 69469 Weinheim, Germany All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form – by photoprinting, microfilm, or any other means – nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Print ISBN: 978‐3‐527‐34467‐3 ePDF ISBN: 978‐3‐527‐81491‐6 ePub ISBN: 978‐3‐527‐81493‐0 oBook ISBN: 978‐3‐527‐81492‐3 Typesetting  SPi Global, Chennai, India Printing and Binding

Printed on acid‐free paper 10 9 8 7 6 5 4 3 2 1

­I dedicate this book to Julius Rebek Jr. who sparked my fascination and that of many others with supramolecular chemistry.

vii

Contents Preface  xv 1

Water Runs Deep  1 Nicholas E. Ernst and Bruce C. Gibb

1.1 The Control of Water  1 1.2 The Shape of Water  2 1.3 The Matrix of Life as a Solvent  4 1.4 Solvation Thermodynamics  6 1.5 The Three Effects  9 1.5.1 The Hydrophobic Effect  11 1.5.2 The Hofmeister Effect  19 1.5.3 The Reverse Hofmeister Effect  23 1.6 Conclusions and Future Work  24 ­ Acknowledgments  25 ­References  25 2

Water‐Compatible Host Systems  35 Frank Biedermann

2.1 General Overview  35 2.2 Acyclic Systems  36 2.2.1 Acyclic Molecular Recognition Units  36 2.2.2 Molecular Tweezers  38 2.2.3 Foldamers  39 2.2.4 Compartmentalized Structures Formed by Surfactant‐Like Molecules  40 2.3 Macrocyclic Receptors that Bind Charged Guests  42 2.3.1 Crown Ethers, Cryptands, and Spherands  42 2.3.2 Bambus[n]urils  44 2.3.3 Calix[n]arenes  45 2.3.4 Pillar[n]arenes  48 2.4 Macrocyclic Receptors that (also) Bind Non‐charged Organic Guests  50 2.4.1 Cyclodextrins  50 2.4.2 Cucurbit[n]urils  54 2.4.3 Deep Cavitands  58

viii

Contents

2.4.4 Molecular Tubes  62 2.5 Practitioner’s Guidelines for Choosing a Water‐Compatible Host  64 2.5.1 Guest Binding Affinity and Selectivity  64 2.5.2 Availability/Scalability  65 2.5.3 Functionality  65 2.5.4 Solubility  66 2.5.5 Biocompatibility/Toxicity  67 ­References  67 3

Artificial Peptide and Protein Receptors  79 Joydev Hatai and Carsten Schmuck

3.1 Introduction  79 3.2 Peptide Recognition  79 3.2.1 Calixarenes  80 3.2.2 Guanidiniocarbonyl Pyrroles  80 3.2.3 Cucurbiturils  82 3.2.4 Metal Complexes  84 3.2.5 Phosphonates  86 3.2.6 Thiourea‐Containing Copolymers  87 3.3 Protein Recognition  88 3.3.1 Molecular Tweezer: Huntingtin Protein (htt)  89 3.3.2 Foldamer: Human Carbonic Anhydrase  89 3.3.3 Tetravalent Peptide: β‐Tryptase  90 3.3.4 Semisynthetic Fusicoccin Derivative: 14‐3‐3/Gab2 Protein  91 3.3.5 Ruthenium Complex: Cytochrome C 92 3.3.6 Nitrilotriacetic Acid–Peptide Conjugate: His‐Tag Calmodulin  93 3.3.7 Cucurbit[7]uril: Native Insulin and Human Growth Hormone  95 3.3.8 Phosphonated Calix[6]arene: Cytochrome C 96 3.3.9 p‐Sulfonatocalixarene: Human Insulin Α 96 3.3.10 Multivalent Calixarene: Platelet‐Derived Growth Factor  97 3.4 Sensor Arrays for Proteins  99 3.4.1 Tripodal Peptide‐Containing Receptors: Proteins and Glycoproteins  99 3.4.2 Substituted Porphyrins: Proteins and Metalloproteins  100 3.4.3 Poly(p‐phenyleneethynylene)s: Proteins  101 3.4.4 Chemiluminescent Nanomaterials: Proteins and Cells  103 3.5 Combinatorial Fluorescent Molecular Sensors for Proteins  104 3.5.1 Probe for MMP, GST, and PDGF Protein Families  104 3.5.2 Probe for Amyloid Beta Proteins  107 3.6 Conclusions and Future Directions  108 ­ References  109 4

Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials  115 Isabel Pont, Cristina Galiana‐Rosello, Alberto Lopera, Jorge González‐García, and Enrique García‐España

4.1 Introduction  115 4.2 Nucleotide Structures  118

Contents

4.3 4.3.1 4.3.2 4.3.3 4.3.4 4.4 4.4.1 4.4.2 4.4.3 4.5

Nucleotide Receptors  119 Receptors Without Aromatic Units  119 Receptors with Aromatic Units  123 Metal Complexes as Nucleotide Receptors  131 Catalytic Aspects  134 Nucleotide Sensing  140 General Aspects  140 UV–vis Sensing  140 Fluorescence Sensing  142 Soft Materials Incorporating Nucleotides, Nucleosides, and Nucleobases  147 4.6 Biomedical Applications  150 4.7 Challenges and Future Perspectives  151 ­Acknowledgment  152 ­References  153 5

Carbohydrate Receptors  161 Anthony P. Davis

5.1 Introduction  161 5.2 Organic Molecular Receptors  163 5.2.1 Acyclic Receptors  164 5.2.2 Macrocyclic Receptors  167 5.2.3 Macropolycyclic Cage Receptors  171 5.3 Metal Complexes as Carbohydrate Receptors  178 5.4 Boron‐Based Receptors  180 5.5 Conclusions  184 ­References  186 6

Ion Receptors  193 Luca Leoni, Antonella Dalla Cort, Frank Biedermann, and Stefan Kubik

6.1 Introduction  193 6.1.1 Potential Applications for Ion Receptors  194 6.1.2 Binding Modes of Ion Receptors  194 6.2 Cation Receptors  197 6.2.1 Neutral Receptors  197 6.2.1.1 Crown Ethers and Cryptands  197 6.2.1.2 Cyclodextrins 198 6.2.1.3 Cucurbiturils 199 6.2.1.4 Cavitands 201 6.2.2 Negatively Charged Receptors  202 6.2.2.1 Cyclophanes 202 6.2.2.2 Cryptophanes 204 6.2.2.3 Calixarenes 204 6.2.2.4 Pillararenes 205 6.2.2.5 Molecular Tweezers  206 6.2.2.6 Acyclic Cucurbiturils  208 6.2.3 Metal‐Containing Receptors  209

ix

x

Contents

6.2.3.1 Metallacycles 209 6.2.3.2 Coordination Cages  210 6.3 Anion Receptors  211 6.3.1 Metal‐Containing Receptors  211 6.3.1.1 Coordination Cages  212 6.3.1.2 Tetraazamacrocycle‐Based Receptors  214 6.3.1.3 Diethylenetriamine‐ and Bis(2‐pyridylmethyl)amine‐Based Receptors  215 6.3.1.4 Tris(2‐aminoethyl)amine and Tris(2‐pyridylmethyl)amine‐Based Receptors  218 6.3.1.5 Miscellaneous 220 6.3.2 Positively Charged Receptors  221 6.3.2.1 Receptors with Quaternary Ammonium Groups  221 6.3.2.2 Amine‐Based Receptors  223 6.3.2.3 Guanidine‐Based Receptors  225 6.3.2.4 Imidazolium‐Based Receptors  227 6.3.3 Negatively Charged Receptors  228 6.3.4 Neutral Receptors  231 6.4 Zwitterion Receptors  236 6.5 Conclusion and Future Challenges  238 ­ References  239 7

Coordination Compounds  249 Anna J. McConnell and Marc Lehr

7.1 Introduction  249 7.2 Organometallic Compounds  249 7.2.1 Macrocycles  251 7.2.2 Cages  252 7.3 Metallomacrocycles  253 7.4 Metallosupramolecular Helicates  255 7.4.1 Transition Metal Helicates  255 7.4.2 Lanthanide Helicates  257 7.5 Metallosupramolecular Bowls and Tubes  260 7.6 Metallosupramolecular Cages  262 7.6.1 Design Considerations  263 7.6.2 Thermodynamics of Guest Binding  263 7.6.3 Cage and Guest Dynamics upon Encapsulation  265 7.6.4 Chiral Recognition  266 7.6.5 Encapsulation of Biorelevant Molecules  266 7.6.6 Stabilization of Encapsulated Species  269 7.6.7 Controlling Reactivity  269 7.6.8 Catalysis  270 7.7 Metal–Organic Frameworks  272 7.8 Challenges and Future Directions  273 ­References  274

Contents

8

Aqueous Supramolecular Polymers and Hydrogels  285 Daniel Spitzer and Pol Besenius

8.1 Introduction  285 8.2 Hydrogen‐Bonded Supramolecular Systems  287 8.3 Host–Guest Induced Supramolecular Polymers and Hydrogels  292 8.4 Metal–Ligand Coordinated Systems  296 8.5 π‐Conjugated Systems  301 8.6 Low Molecular Weight Hydrogelator Systems  307 8.7 Peptide‐Based Molecular Amphiphiles and Their Supramolecular Systems  314 8.8 Bioinspired Systems  321 8.9 Challenges and Future Directions  326 ­References  326 9 Foldamers  337 Morgane Pasco, Christel Dolain, and Gilles Guichard

9.1 Introduction  337 9.2 Discrete Protein‐Like Architectures by Lateral Assemblies of Helical Foldamers  338 9.2.1 Bioinspired Helix Assemblies: Top‐Down Approaches  340 9.2.2 Bioinspired Helix Assemblies: Bottom‐Up Approaches  344 9.3 Helix Duplexes in Aqueous Solution  350 9.4 Assemblies of Extended Chains  355 9.5 Elongated Nanostructures by Self‐Assembly  357 9.6 Applications  359 9.6.1 Host–Guest Interactions With and Within Helix Bundles  359 9.6.2 Self‐Assembling Foldamers Targeting Heparin  362 9.6.3 Catalysis with Self‐Assembled Foldamers  363 9.6.4 Foldamer‐Mediated Protein Oligomerization  364 9.6.5 Nanopores by Insertion of Foldamers into Phospholipid Membranes  366 9.7 Challenges and Future Directions  366 ­Acknowledgments  367 ­References  367 10

Vesicles and Micelles  375 Wilke C. de Vries and Bart Jan Ravoo

10.1 Introduction  375 10.2 Building Blocks and Structure of Vesicles and Micelles  376 10.2.1 Conventional Building Blocks and Packing Parameter  376 10.2.2 Driving Forces and Dynamics  379 10.2.3 Nonconventional Building Blocks  382 10.3 Stimulus‐Responsive Vesicles and Micelles  387 10.3.1 Endogenous Stimuli: Redox and pH  387 10.3.1.1 Redox  387

xi

xii

Contents

10.3.1.2 pH  389 10.3.2 Exogenous Stimuli: Light and Temperature  391 10.3.2.1 Light  391 10.3.2.2 Temperature  392 10.4 Vesicles and Micelles as Template Structures for Nanomaterials  393 10.4.1 Condensation and Polymerization Reactions Using Template Structures  393 10.4.2 Stabilization of Vesicle and Micelle Structures by Cross‐Linking  394 10.4.3 Polymer Shells Enclosing Vesicle Templates  395 10.5 Molecular Recognition of Vesicles and Micelles in Biomimetic Systems and Nanomaterials  397 10.5.1 Macrocyclic Amphiphiles  397 10.5.2 Carbohydrate and Peptide‐Based Recognition  399 10.5.3 DNA‐Based Recognition  402 10.6 Challenges and Future Directions  404 ­References  405 11

Monolayer‐Protected Gold Nanoparticles for Molecular Sensing and Catalysis  413 Fabrizio Mancin, Leonard J. Prins, Federico Rastrelli, and Paolo Scrimin

11.1 Introduction  413 11.2 Analytical Techniques  414 11.2.1 Nuclear Magnetic Resonance Spectroscopy  414 11.2.2 Electron Paramagnetic Resonance Spectroscopy  416 11.2.3 Fluorescence Spectroscopy  417 11.2.4 Isothermal Titration Calorimetry  417 11.2.5 Surface‐Enhanced Raman Scattering  418 11.3 Molecular Recognition and Chemosensing of Small Molecules  418 11.3.1 Multivalent Binding Interactions at the Monolayer Surface  419 11.3.2 Binding Pockets in the Monolayer  420 11.3.3 Gold Nanoparticle‐Based Chemosensors  426 11.3.3.1 Indicator Displacement Assays  426 11.3.3.2 NMR Chemosensing  428 11.4 Catalysis by Nanozymes  430 11.5 Controlling Molecular Recognition Processes at the Monolayer  435 11.5.1 Regulatory Mechanisms  435 11.5.2 Adaptive Multivalent Surfaces  438 11.6 Challenges and Future Directions  442 ­References  442 12

Optical Probes and Sensors  449 Pavel Anzenbacher, Jr and Lorenzo M. Mosca

12.1 12.2 12.3 12.3.1

Introduction and Lexicon  449 Brief Fundamentals of Molecular Photoprocesses  451 Some Comments on the Design of Probes and Sensors  455 General Aspects  455

Contents

12.3.2 Fighting with Water  457 12.4 Probes and Sensors for Electroneutral Species  459 12.4.1 Carbohydrates  459 12.5 Probes and Sensor for Cations  462 12.5.1 Alkali and Alkali‐Earth Cations  462 12.5.2 First‐Row Transition Metal Ions  464 12.5.3 Heavy Metal Ions, Particularly Cadmium and Mercury  467 12.6 Probes and Sensors for Anions  469 12.6.1 Fluoride  469 12.6.2 Cyanide  472 12.6.3 Inorganic and Organic Phosphates  473 12.6.4 Carboxylates  482 12.6.5 Other Anions of Interest  487 12.6.6 Sensors for Multiple Anions  487 12.7 Sensing of Biomacromolecules  489 12.8 Challenges and Future Directions  491 ­References  492 13

Probes for Medical Imaging  501 Felicia M. Roland and Bradley D. Smith

13.1 Medical Imaging  501 13.2 Structure and Supramolecular Properties of Molecular Probes  503 13.2.1 Structure  503 13.2.2 Linkers  503 13.2.3 Reporter Groups  504 13.2.4 Design Aspects  504 13.3 Targeting Groups for Receptors  506 13.3.1 Drug‐Like Molecules  506 13.3.2 Vitamins  507 13.3.3 Peptides  508 13.3.4 Antibodies  508 13.3.5 Aptamers  510 13.4 Signal Enhancement Strategies  511 13.4.1 Intracellular Accumulation  511 13.4.2 Signal Activation by Enzymes  512 13.5 Targeting Cell Surface Biomolecules  513 13.5.1 Anionic Phospholipids  513 13.5.2 Glycans  514 13.5.3 Antigens  515 13.6 Clinical Development  516 13.6.1 Government Approval  516 13.6.2 Multimodal Approaches  518 13.6.3 Theranostic Approaches  519 13.7 Future Role of Supramolecular Chemistry  520 ­Acknowledgments  521 ­References  521

xiii

xiv

Contents

14

Supramolecular Catalysis in Water  525 Piet W. N. M. van Leeuwen and Matthieu Raynal

14.1 Introduction  525 14.2 Classification of Supramolecular Catalysts Operating in Water  527 14.2.1 Mass Transfer Promotion Through Substrate Sequestration (S1)  529 14.2.2 Catalysis by Confinement (S2)  529 14.2.3 Directed Substrate Reactivity (S3)  531 14.2.4 Construction and Modulation of the Catalytic Structure (S4)  532 14.3 Synthetic Hosts for Catalysis in Water  533 14.3.1 Cyclodextrins (CD)  536 14.3.2 Cucurbit[n]urils (CBn)  536 14.3.3 Hosts with Aromatic Walls  537 14.3.4 Velcrands  538 14.3.5 Octa‐acid  538 14.3.6 Metallocages  538 14.3.7 Hyperbranched Polymers  539 14.3.8 Dendrimers  539 14.3.9 Micelles  540 14.3.10 Vesicles 541 14.4 Supramolecular Catalysts for the Aqueous Biphasic Hydroformylation Reaction  542 14.5 Supramolecular Catalysts for Other Organometallic Reactions in Water  547 14.6 Future Directions  550 ­References  551 Index  567

xv

Preface Water is the element of life. Without water, life on this planet could have neither developed nor would it have survived. That water plays such a crucial role in this context is partly due to its unique properties as a solvent. Water is polar, has a high boiling point considering the low molecular weight of water molecules, and dissolves many salts and polar organic molecules, while apolar molecules tend to phase‐separate. These properties are intimately linked with the propensity of water molecules to form an infinite and dynamic network in the condensed phase in which the individual components are held together by relatively strong cooperative hydrogen bonds. This network can rearrange to incorporate ions or polar neutral molecules. The associated hydration of these solutes suppresses interactions between them, whereas interactions between apolar molecules, whose hydration is thermody­ namically unfavorable, are reinforced. Consequently, apolar solvents do not mix with water, amphiphiles aggregate to form micelles or vesicles, and polymers (e.g. proteins) fold such that apolar regions are buried inside the resulting struc­ ture, whereas polar ones are positioned along the solvent‐accessible outside. The underlying recognition, aggregation, and folding processes encompass intra‐ or intermolecular molecular interactions under thermodynamic control, position­ ing them at the heart of supramolecular chemistry. Because aspects of water structure cause these recognition phenomena to be governed by effects that are absent in other solvents, supramolecular chemistry in water is special and often challenging. Nevertheless, supramolecular chemistry has not shied away from water. For example, studies on the host–guest chemistry of cyclodextrins, cyclophanes, polyazacryptands or related receptors, micelles or liposomes, or transport phe­ nomena were predominantly carried out in aqueous environments. The choice of solvent in these investigations was mostly based on practical considerations, however, such as the properties and solubilities of the binding partners. Only in recent years a clear trend emerged to deliberately develop supramolecular sys­ tems that work in water also with a view on potential practical uses. Examples are biomedical applications in which supramolecular receptors are used to mediate and control biochemical processes, analytical applications aiming at detecting harmful environmental contaminants, or the control of chemical transforma­ tions by suitable supramolecular catalysts. At the same time, the understanding of the peculiarities of water and the associated effects on recognition events have

xvi

Preface

progressed to such a level that the development of supramolecular systems work­ ing in water has become increasingly more successful. This book should illus­ trate these developments. Chapter 1 of the book shows how and to what state our knowledge about the structure of water in the condensed phase and the effects of water on recogni­ tion phenomena have developed in recent years. This chapter thus lays the foundation for understanding the behaviors of the systems presented in the rest of the book. Chapter 2 then introduces structures and properties of many important host systems that can be used in water, most of which will reappear in later chapters. The following four chapters focus on several types of sub­ strates that can be bound with appropriate receptors in water, namely, peptides and proteins (Chapter 3), nucleotides (Chapter 4), carbohydrates (Chapter 5), and ions (Chapter 6). The focus will then shift to supramolecular systems that can be generated by self‐assembly in water. Chapter 7 describes coordination compounds, Chapter  8 polymers and gels, Chapter  9 foldamers, Chapter  10 vesicles and micelles, and Chapter 11 surface‐modified gold nanoparticles. All of these chapters elude to concepts, structural aspects of the respective sys­ tems, analytical techniques to characterize them, and also potential appli­ cations wherever appropriate. Applications are also the central topics of the final three chapters of the book, with Chapter 12 summarizing supramolecular strategies to sense analytes in water by means of optical methods, Chapter 13 presenting optical probes for medicinal imaging, and Chapter 14 focusing on supramolecular catalysts. The broad range of subjects treated, from fundamental aspects to applied ones, should provide extensive insight into a timely and exciting research field at the same time illustrating some directions into which supramolecular chemistry is currently heading. It is thus hoped that the book will be a useful compendium for scientists working in the area and that it might also motivate interested read­ ers to join the community and contribute with original and new ideas. This book would not have been possible without the help from a group of highly competent and reliable authors. I am extraordinarily grateful to all of them for their efforts and excellent contributions. I also warmly thank Elke Maase and Shirly Samuel from Wiley for their help and support. Kaiserslautern, Germany October 2018

Stefan Kubik

1

1 Water Runs Deep Nicholas E. Ernst and Bruce C. Gibb Tulane University, Department of Chemistry, 2015 Percival Stern Hall, 6400 Freret Street, New Orleans, LA, 70118, USA

1.1 ­The Control of Water There is current consensus among archeology scholars that there are five cradles of civilization, geographical locations where the first civilizations emerged thousands of years ago. These five areas – the Fertile Crescent, the Indus River, the Yellow River, the Central Andes, and Mesoamerica – each independently gave rise to a new level of human existence [1–3]. What do these sites have in common? They all shared a combination of favorable geography and the opportunity to control water. Because of its central role in life on Earth, the idea of controlling water manifests in many aspects of human endeavor. In agriculture, the key is irrigation – how to bring water to the fertile land and crops. “Controlling” the oceans and rivers requires harbors, moorings, and boats to open up trading routes or the possibility of fishing, while controlling sewage has been key to the success of cities. Controlling water also allowed energy production in the form of watermills and, in more modern times, dams, tidal barrages, and wave energy convertors. Unfortunately, the control of water can also be used to justify war. Conflicts between nations, states, or groups, for example, the violence in the war in Sudanese Darfur, can be partially attributed to water [4, 5]. For many different reasons, the control of water at the molecular and nanoscale levels is also key. Many scientific disciplines work toward being able to understand and control water purification for waste treatment and desalination [6, 7], water flow for microfluidics and ultra‐hydrophobic materials [8], and solvation to in turn control dissolved matter. At this level, the control of water is free of famine and conflict. However, it is plagued by beguiling complexity; the solvent of life, this “matrix of the world and of all its creatures” (Paracelsus, 1493–1541 ce) is a truly complex substance possessing endless emergent phenomena [9, 10]. This means that it is easy for pathological science [11] to make wild claims about the memory properties of water [12, 13] or the existence of polywater [14, 15]. Although, at the time, the latter stoked a public fear reminiscent of Vonnegut’s Cat’s Cradle [16], polywater has passed into history. Alas the same cannot be Supramolecular Chemistry in Water, First Edition. Edited by Stefan Kubik. © 2019 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2019 by Wiley-VCH Verlag GmbH & Co. KGaA.

2

1  Water Runs Deep

said for memory water and infinitely diluted homeopathic remedies. And lest it be thought that there are no new controversies around water, consider also the relatively recent idea that water is capable of storing charge [17–19]. As the complexity of water means that it is exceedingly difficult to truly understand how water interacts with itself and how it interacts with solutes [20], the primary aim of this chapter is to outline what we do and do not know about supramolecular interactions in aqueous solution. Correspondingly, a secondary aim is to minimize situations wherein an observation might be explained away by the waving of hands. For as has been noted previously, “When confronted with unexpected experimental results water structure has commonly been used deus ex machina for explaining the observations.” [21] We can and should do better than this.

1.2 ­The Shape of Water At the heart of water and aqueous solutions lies a contentious question that goes back millennia – what is the structure of liquid water? The Greek philosopher Empedocles (490–430 bce), like many thinkers around the world, considered water to be one of the four classical elements: earth, water, air, and fire. More helpfully, Democritus (460–370 bce) proposed that all matter was made up of small indivisible particles called atoms (atomism). Around the same time, Plato (428–348 bce) was being less helpful but slightly prophetic. He merged geometry with the four elements of the universe to propose that water was one of five (Platonic) solids: tetrahedron, cube, octahedron, dodecahedron, and icosahedron. Specifically, Plato proposed that water is an icosahedron, which explained its ability to flow across (cubic) the Earth. It was not until 1781 that Henry Cavendish showed water to consist of hydrogen (inflammable air) and oxygen (dephlogisticated air) by burning a mixture of the two [22]. This discovery was built further upon by Amadeo Avogadro, who was the first to clearly differentiate between atoms and molecules, and that the latter must be represented by an empirical formula. He also hypothesized that equal volumes of gas at equal pressure contain the same amounts of molecules regardless of the nature of the gas, and from this was able to repeat Cavendish’s experiment to show that water was two parts hydrogen and one part oxygen [23]. Moving into the late 1800s and the early 1900s, a combination of Röntgen’s discovery of X‐rays, Einstein’s concept of a photon, and Compton’s confirmation of X‐ray scattering by electrons ultimately led to the development of X‐ray crystallography. This led Bragg and others to determine the structure of ice. In the solid state, water possesses tetrahedral geometry [24] in which oxygen atoms are 2.75 Å from each other. There was, after all, an element of truth to what Plato had thought: water was not an icosahedron, but four water molecules did make a tetrahedron. That same decade saw the nucleation of the idea of hydrogen bonding (H‐bonding). In 1920 Latimer and Rodebush proposed that a “free pair of electrons might be able to exert sufficient force on a hydrogen held by a pair of electrons on another water molecule to bind the two molecules together.”

1.2  The Shape of Water

However, it was not until 1939 and the publication of Pauling’s The Nature of the Chemical Bond that the H‐bond began to meet with widespread acceptance. Hexagonal ice, the form of all natural snow and ice on Earth, has the perfect tetrahedral structure with bond–bond, bond–lone pair, and lone pair–lone pair angles of 109.47° [25]. The reason for the high boiling point, large heat of vaporization, high heat capacity, and high surface tension of water was evident. Partially through the development of X‐ray crystallography, liquids began to be thought of as either dense ordered gases or imperfect disordered crystals, and with its H‐bonding network, water was a prime candidate for this line of thinking. For example, an influential model by Bernal and Fowler treated water as a point charge with tetrahedral geometry whose structure was akin to that of disordered quartz [26]. As we shall shortly see (vide infra), this idea of water possessing considerable structure quickly moved into adjacent areas of research. The Bernal and Fowler paper influenced thinking through the 1950s. Thus, Erying [27] proposed a liquid state structure of water consisting of crystalline close packing threaded with many dislocations. According to this model, molecules escaping the close packing could wander almost gas‐like between the clusters. Similarly, Pauling proposed a disordered crystal based on the clathrate structure, while Bernal proposed a model of a random H‐bonded network where water molecules gave rise to connected four‐ and seven‐membered rings. The last half‐century or so has seen a dipartite approach to studying water. On the one hand, there have been the continued improvements in established spectroscopic techniques as well as the development of new spectroscopies [28]. The suite now available is extensive and includes Raman/Raman–­multivariate curve resolution (Raman‐MCR), IR, 2D‐IR, sum‐frequency generation techniques, and multiple X‐ray and neutron approaches such as scattering, small‐angle scattering, X‐ray absorption spectroscopy (XAS), and X‐ray emission spectroscopy (XES). Complementing this approach has been the development of computational chemistry. Thus, the combination of increasingly powerful computers, more accurate water models (e.g. TIP4P‐Ew) [29], and new strategies has added greatly to our understanding of water and solvation at the molecular level [30]. We now know that the mean H─O bond length in liquid water is 0.97 Å, the mean HOH bond angle 106°, and the mean negative charge on the O atom ~70% of an elementary charge, with each positively charged H atom sharing the neutralizing charge. We also know that although water makes four H‐bonds in the solid state, as a liquid it forms ~3.6 H‐bonds on average. However, this number varies according to the specific analytical technique used. Furthermore, there is still some debate regarding the symmetry of the H‐bonds in liquid water. In the gas phase, water forms minimum energy clusters [31–33]; however, it is clear from simulations that these structures do not persist in bulk solution at ambient temperature. Rather, the current consensus is that water possesses primarily tetrahedral structure  [34, 35], although this is not universally agreed upon. Thus  to pick one controversial example, XAS suggests that ~80% of water ­molecules have one strong and one weak hydrogen‐bonded O─H group, such as what may  occur in cyclic pentamers or hexamers [36]. In contrast, the

3

4

1  Water Runs Deep

r­ emaining ~20% of the molecules in this model reside in four‐hydrogen‐bonded tetrahedrally coordinated clusters. We also know that water H‐bonds break and form on the timescale of tens of femtoseconds to picoseconds. They are weaker than the H‐bonds in ice; the oxygen atoms in liquid water are 2.8–2.9 Å apart, but X‐ray and neutron scattering still show evidence of a distorted tetrahedral structure [37]. As for the remaining ~0.4 H‐bonds, evidence points toward virtually all molecules with such dangling H‐bonds returning to an H‐bonding partner within 200 fs. In other words, dangling H‐bonds are intrinsically unstable [38]. Assuming the consensus is that water does not form rings or specific clusters, but instead possesses a slightly defective tetrahedral structure, what remains to be developed is a good understanding of how this lattice, composed of exceedingly short‐lived H‐bonds and containing ~10% defects, leads to the physical properties of water observed at the molecular and bulk levels. In the interim, it is still common to hear enduring if rather vague terms such as “flickering tetrahedral clusters” to describe the structure of water.

1.3 ­The Matrix of Life as a Solvent Many common organic solvents are large and, in a structural sense, relatively homogeneous; they are relatively nonpolar, and correspondingly the range of their different solvent–solute interactions is relatively small and intrinsically similar to their own solvent–solvent interactions. As a result, chemists can frequently ignore the existence of organic solvents, but not water. Water is exceptionally small: a third of the volume of dichloromethane and one sixth of the volume of toluene. Even dissolved oxygen is large relative to water. Moreover, water is highly polar; recall the −0.7 charge on the oxygen and +0.35 charge on each hydrogen atom. These lead to an agile, nimble, and cohesive pack animal and en masse an excellent solvent; it can dissolve both simple ions and hydrocarbons with a solubility range of ~50 orders of magnitude [39]. The small size, high polarity, and presence of two H‐bond acceptors and two H‐bond donors mean that water is uniquely dense with supramolecular motifs. This combination of features leads to a unique combination of properties, some of which are extreme and many of which are not. For example, water has a significant dipole moment of 1.84 D and a high dielectric constant of εr = 78; but hexamethylphosphoric acid triamide (HMPT) has a dipole moment of 5.55 D, and N‐methylformamide has a dielectric constant of 182. Similarly, water has an α‐value (Kamlet–Taft solvent parameter) of 1.17, indicating that it is a weaker H‐bond donor than many halogenated alcohols, and a very middling β‐value of 0.47, indicating it is a weaker H‐bond acceptor than solvents such as pyridine. That stated, by the measure of solvatochromic dyes, water is the extreme polar solvent. For example, it has the largest ETN of common solvents (a measure of polarity based on the transition energy for the longest wavelength absorption band of a betaine dye, which for water is by definition 1.00) . The extensive H‐ bonding network in liquid water means that it also has an exceptionally high heat capacity: its value of 75.2 J mol−1  K−1 (at 25 °C) is second only to ammonia

1.3  The Matrix of Life as a Solvent

(80.8 J mol−1 K−1) for heteroatomic species. This has ramifications not only for aquatic life on Earth but also for the fundamental properties of solvation. Water is also highly cohesive. This means that it has remarkably high melting and boiling points for its size (cf. the boiling points of H2S, H2Se, H2Te, and H2Po are −60.3, −41.3, −4, and 36.1 °C, respectively). It also has an exceptional surface tension: 72.8 mN m−1, a value much higher than any other common liquid with the exception of mercury (472 mN m−1). This high surface tension means that water is not a good wetting agent, which is good news for the platypus but frustrating for the analysis of the solvation of nonpolar solutes. Compared to common organic solvents, water is also quite viscous, although its value (8.94 × 10−4 Pa s) is dwarfed by larger molecules with multiple H‐bonding groups such as glycerol. Water also has a small but significant ionization constant: at 25 °C and zero ionic strength, Kw = 1 × 10−14. This means that even at neutral pH water is not chemically pure; there is hydronium (H3O+) and hydroxide (OH−) or their corresponding higher ions such as H5O2+ (Zundel cation) or H9O4+ (Eigen cation) to consider also. This adds another level to the complexities of water. For example, it has been known for a long time that, via proton hopping processes, hydronium diffuses at twice the rate of hydroxide ion. Why is this? Very recent work suggests that it is because hydronium diffusion is concerted, whereas hydroxide diffusion is stepwise [40]. However, this has yet to be confirmed by others. So, by many metrics water is an exceptional solvent, yet by others, less so. The ineffable uniqueness of water is that it is composed of the first and third most abundant elements in the universe, which combined lead to a molecule of unusual physical and chemical properties. As a result, given a suitable temperature window, water can be expected to be the ubiquitous solvent of life. There is a problem however. Many of the aforementioned properties cannot easily be scaled to the molecular level, and it is down at the molecular or atomistic level that supramolecular chemists would really like to fully understand water. How are solutes solvated? How extensive is the water–water H‐bonding around a solute? How does this solvation influence the thermodynamics or kinetics of host–guest complexation? There is no easy way to “translate” surface tension, wettability, cohesive energy density, etc. to the molecular level. Yet this is an eternal problem with studying aqueous supramolecular chemistry (and one reason why computational chemistry has been such a boon for the field). This means that given sufficient time hypotheses can become fact. To pick one classic exemplar, consider a 1945 landmark from Frank and Evans [41]. In this paper the authors describe examining the solvation thermodynamics of small nonpolar molecules in water. In expanding their own and other’s work, they concluded that the entropic cost of dissolving small nonpolar organics in water arose because they modify “the water structure in the direction of greater crystallinity – the water, so to speak, builds a microscopic iceberg around it.” This idea that the solvation shell of dissolved nonpolar molecules is highly structured was subsequently used by Kauzmann in a model for the thermodynamics of protein folding [42]. In large part thanks to this paper that this general idea can still be found within undergraduate biochemistry textbooks. However, as we will expand upon below, the extent to which this is true is unclear.

5

6

1  Water Runs Deep

1.4 ­Solvation Thermodynamics While the customary viewpoint of supramolecular chemists is from the perspective of a molecular host, water scientists consider water to be the host of interest. Thus, rather than considering host and guest desolvation and host–guest complex formation, water scientists consider how water (the host) responds to the addition of a solute (the guest). In this section, we mostly opt to take this “reverse” perspective of the water scientist and discuss what happens when a solute is taken from the gas phase into the aqueous phase. When applied to solvation processes, Gibbs (Bogoliubov and/or Feynman) inequalities rigorously describe and set upper and lower bounds for the thermodynamics of transfer from the gas to solution phase [43]. As we describe below, Gibbs inequalities, combined with Linear Response Theory (that relates and defines solute–solvent interaction energies), lead to some remarkable thermodynamic relationships. For example, both van der Waals and electrostatic solute–water interactions are well described by the Linear Response Theory, and one fascinating conclusion is that for the latter, there is nearly perfect cancellation of ion–water and water reorganization contributions to experimental hydration entropies. Thus, ion–water interactions formed upon dissolution invariably produce a decrease in entropy, but the resulting release of heat dissipated to the surrounding solution leads to an entropy increase of nearly equal and opposite magnitude. It is this entropy cancellation that is evidently responsible for the near equivalence of the experimental hydration entropies of noble gases and isosteric halide ions, despite the initial conclusion that the latter would strongly coopt and organize water molecules into a highly defined solvation shell. More generally, it is this negative ion–water interaction entropy and countering positive water reorganization entropy that is responsible for near equivalence of the experimental hydration entropies of ionic, polar, and nonpolar solutes. This suggests that in the binding of a charged amphiphile or ion to a nonpolar pocket of a host, the changes in solvation of the charge group are not reflected in the experimental ΔS for complexation [44, 45]. To fully understand solvation, it is important to consider all the solute–­solvent and solvent–solvent interactions involved, and a convenient way to do so is to break the process down into three steps: the formation of a cavity the size and shape of the solute in water (a hard sphere is ideal), the “switching on” of van der Waals forces in the solute, and lastly the inclusion of electrostatic forces. Figure 1.1 illustrates this thought process, showing the change in water structure 0 and solute–solvent (uv) interaction energy before (Euv ) and after “equilibration” (Euv) of each step. As has been summarized elsewhere [43], the Linear Response 0 Theory links these two energies (Euv and Euv) by a linear function, with dispersive and electrostatic interactions falling into very different linear response regimes. The abrupt insertion of a hard‐sphere potential into pure water results in (by 0 definition) an initial, infinite solute–solvent interaction energy (Euv ), cav which decreases to approximately zero once equilibrated. Gibbs inequality dictates that the free energy of cavity formation is invariably positive (0 ≤ ΔG(cav) ≤ ∞). Furthermore, a combination of experimental and simulation data reveals that the thermodynamic functions of cavity formation (ΔG(cav), ΔH(cav), and ΔS(cav)) are

1.4  Solvation Thermodynamics Unequilibrated

Equilibrated

Step 1 Cavity formation (hard sphere)

° ≈∞ Euv(cav)

≫

Euv(cav) ≈ 0

≈

Euv(cav)

Step 2 van der Waals interactions

° Euv(cav)

Step 3 Electrostatic interactions

° Euv(elec) ≈0

≫

Euv(elec)

Figure 1.1  Three‐step visualization of a solvation process for a solute (green circles) and its surrounding hydration shell of water molecules (blue circles with arrows depicting the dipole of each water). Step 1: cavity formation. Steps 2 and 3: introduction of van der Waals and electrostatic interactions, respectively. Each step consists of unequilibrated and equilibrated states, where the former depicts the system after the introduction or change to the solute but before the response of the solvation shell. The equilibrated state corresponds to the system after the hydration shell has responded to the change. Solute–water interaction energy descriptions for each state are derived from the Linear Response Theory and Gibbs inequalities [43].

nonlinear [46]. In addition, the cavity formation enthalpy in water is markedly temperature-dependent, a consequence of the fragility of the water H‐bond network surrounding a nonpolar (hard‐sphere) solute. It is this temperature dependence that dictates the large heat capacity changes seen in the solvation of nonpolar guests [47]. As we will see in the next section, host molecules greatly attenuate the free energy requirements of cavity formation. In step 2 the van der Waals forces between solute and solvent are “switched on.” van der Waals forces are largely insensitive to the orientations of the water in the solute solvation shell, and Gibbs inequalities lead to an extremely restrictive constraint on dispersive contributions to solvation, namely, that 0 Euv disp Euv G disp , and Suv(disp)  ≈  0. In other words, dispersion disp interactions are expected to contribute enthalpically but not entropically to the solvation of nonpolar molecules.

7

8

1  Water Runs Deep

In contrast, “switching on” electrostatic interactions reorientates the solvation shell. Before equilibration the solute–water interaction energy is expected to be small (Euv(elec) ≈ 0) because of the random orientation of the solvation shell water molecules and the weak van der Waals forces. However, once “switched on,” the interaction energy is expected to be strongly negative. In this case, Gibbs inequalities and Linear Response Theory lead to two important conclusions. First, that solute–solvent interaction energy is negative but less negative than the overall change in free energy (Euv(elec) ≤ ΔG(elec) ≤ 0). Second, that electrostatic interactions are not expected to significantly contribute to experimental hydration entropies since ΔS(el ec) ≈ Suv(elec) + ΔSvv(elec) ≈ 0. This is why the hydration entropies of halide ions are essentially indistinguishable from that of noble gases [48]. However, the large free energy of halide ion hydration directly reveals the very large, canceling solute–solvent and solvent–solvent reorganizing entropies (ΔG(elec) ≈ TSuv(elec) ≈  − TΔSvv(elec)). Similar though less dramatic effects can be seen with neutral polar molecules [46]. These important points noted that what supramolecular chemists are adept at gathering are experimental thermodynamic data for guest complexations. How does such data relate to the individual solute–solvent and solvent–solvent enthalpy and entropy changes? As Eq. (1.1) shows, the free energy of solvation (ΔG) can be expressed solely by solute–solvent terms, specifically the solute– solvent interaction energy (Euv) and entropy (TSuv): G

Euv TSuv (1.1)

In contrast, it can be shown that the corresponding enthalpy (ΔH) and entropy (TΔS) changes contain additional solvent reorganization contributions (Eqs. (1.2) and (1.3)): H

Euv

Evv

Pv (1.2)

T S TSuv T Svv (1.3) where Euv is the overall solute–solvent interaction energy, ΔEvv is the overall change in solvent–solvent interaction energy upon dissolution, P is the pressure, v̄ is the solute partial molar volume, TSuv is the overall solute–solvent entropy contribution, and TΔSvv is the overall change in solvent–solvent entropy contribution upon dissolution. Note that at ambient pressure, the pressure–volume work (Pv̄) associated with increasing the volume of the system by an amount equivalent to the change in the partial molar volume of the host–guest complex is negligibly small, so ΔH ≈ ΔU  =  Euv + ΔEvv. Thus, the familiar ΔG  =  ΔH − TS clearly implies that the solvent reorganization enthalpy and entropy must precisely compensate (as ΔEvv = TΔSvv). It is also possible to calculate the intrinsic (gas phase) free energy of host–guest complexation, given realistic interaction potentials and structures for the host, guest, and complex. Additionally, for rigid hosts and guests, the corresponding enthalpy and entropy of binding contain additional trivial ideal gas terms (for flexible systems there are also complicating terms arising from the binding‐ induced conformational change of the host and/or guest). Such calculations combined with solution thermodynamic data and Eqs. (1.1)–(1.3) mean that it is

1.5  The Three Effects

theoretically possible to parse out all the different contributions to host–guest binding thermodynamics. In summary, because of weak entropic effects associated with van der Waals forces between a nonpolar group and water, and because of strong but compensation entropic effects with ions, the influence of water on host–guest binding free energies may be attributed entirely to binding‐induced changes in solute– water interactions. It is important to note, however, that the corresponding binding enthalpy and entropy may also be significantly influenced by changes to water–water interactions (such as the solute‐induced breaking of water–water H‐bonds) upon host–guest binding.

1.5 ­The Three Effects In aqueous solutions chemistry, there are two very familiar phenomena: the hydrophobic effect and the Hofmeister effect. We will, however, add a third here: the reverse (or inverse) Hofmeister effect. The hydrophobic effect is the observed tendency of nonpolar substances to aggregate in an aqueous solution and exclude water molecules; to reiterate the perfunctory adage, oil and water do not mix. We will discuss the inaccuracy of this adage and the details of the hydrophobic effect in Section 1.5.1. For biochemists and organic chemists concerned with organic solutes in water, the Hofmeister effect is most commonly and simply expressed as how salts modulate the hydrophobic effect. As we will see, some salts can increase the solubility of (macro)molecules, while others can decrease it, i.e. apparently by making the solute more hydrophobic. Finally, to make matters more complex, there is also the reverse Hofmeister effect. This is the phenomenon whereby salts that upon initial inspection would be expected to cause the solute under study to become more water soluble but instead induce the (macro)molecule to aggregate and precipitate out of solution. We will discuss the Hofmeister and reverse Hofmeister effects in Sections 1.5.2 and 1.5.3, respectively. It is important to appreciate that while the hydrophobic effect and Hofmeister effects are usually discussed separately, current evidence suggests that the hydrophobic effect, the salting‐in Hofmeister effect, and the reverse Hofmeister effect are rooted in common non‐covalent interactions that occur between weakly hydrated solutes. This point is reinforced by considering the overview in Figure  1.2. In this simplified picture, molecules are considered to partake in only two kinds of non‐covalent interactions: van der Waals and electrostatic interactions. This allows a general ordering of species going from left to right, from weakly to strongly solvated. All of the phenomena in question rely on relatively poor solvation to allow species‐specific interaction between pairs of molecules. A pair of neopentanes will display the hydrophobic effect (although it is arguable that a bigger nonpolar surface is actually required for association in water to occur; vide infra). On the other hand, a neopentane and perchlorate (or tetramethylammonium) will display the salting‐in Hofmeister effect; association of the two will increase the solubility of neopentane in water. The typical anionic Hofmeister series is shown in the upper “branch” of the figure; moving

9

1  Water Runs Deep

F–

Ho

ei of m

H c ni

Neopentane

I–

io

ic ect f ion An ter ef s i e fm

st er

se rie

Br–

s

Cl–

SCN–

An

10

ClO4– Hydrophobic effect

Neopentane

Reverse or inverse Hofmeister effect

C + fme ationi NMe4 iste c r ef fec t

Ho

Increasing hydration

Cs+

Li+

Figure 1.2  A unified view of the relationship between the hydrophobic effect, the salting‐in Hofmeister effect, and the reverse Hofmeister effect. In this scheme, solutes are ordered in a continuum from nonpolar (i.e. in water, hydrophobic) on the extreme left to hard highly hydrated ions on the far right. In this simplified model, solutes such as neopentane are considered to form only van der Waals interactions with water and other solutes, while solutes such as fluoride form only electrostatic interactions with water and other species. Between these two extremes, from left to right, van der Waals interactions are slowly decreasing, while electrostatic interactions are increasing. The combination of these two interaction types lead to the salting‐in Hofmeister effect between neutrals and anions (upper “branch,” typical series of anions shown), with the strongest interactions involving the soft, polarizable anions depicted by the more intense arrows. The near‐transparent arrow between neopentane and fluoride is for illustrative purposes only; anionic Hofmeister effects between such species are essentially nonexistent except at exceedingly high concentrations. Analogous interactions between neutrals and cations can lead to the cationic Hofmeister effect (lower “branch”), with the same gradation in arrow suggestive of the intensity of the effect. As depicted in the vertical, interactions between nonpolar molecules lead to the hydrophobic effect, while interactions between cations and anions can lead to the reverse Hofmeister effect between weakly solvated ions. As with the salting‐in Hofmeister effect, reverse Hofmeister effects existing between hard ions such as fluoride and lithium are essentially nonexistent at moderate concentrations. Note that other interactions, such as reverse Hofmeister effects involving tetramethylammonium and iodide, are not shown for clarity.

1.5  The Three Effects

from left to right, the anions are increasingly more strongly solvated and decreasingly capable of interacting with, for example, a neopentane solute. Between the anionic and cationic “branches,” the reverse Hofmeister effect is manifested; if a solute is cationic and poorly solvated, then exchange of its anion with a more weakly solvated anion can lead to aggregation and precipitation via charge neutralization. Again, the more weakly hydrated the two species, the stronger the effect. 1.5.1  The Hydrophobic Effect Understanding the hydrophobic effect is key to understanding how organic molecules behave in water, their solubility, how they associate, and how they fold. For lucid discussions on water and the hydrophobic effect, readers are directed to reviews by Ball [20], Ben‐Amotz [49], Blokzijl and Engberts [50], Sharp and Vanderkooi [51], Dill and coworkers [52], and Chandler [53]. Supramolecular chemists beware. Ever since the paper by Frank and Evans [41] suggesting “icebergs” around nonpolar solutes, there has been a continuous debate about the existence of highly structured water around nonpolar solvents. Yet, as has been stated previously, to date there is no good reason to suppose that water does organize around such solutes, and some evidence to indicate that it does not [20]. In short, the jury is still out. This uncertainty has unfortunately meant that the hydrophobic effect is frequently rolled out to explain a whole range of results and phenomena, deus ex machina. Science is bringing an increasing number of tools to bear on the topic of the hydrophobic effect; however a note of caution is needed regarding one of the most powerful tools for molecular‐level detailed analysis of water and solvation: computational chemistry. Specifically, the prevalence of off‐the‐shelf computational packages can lead to the assumption that in aquo atomistic molecular dynamics (MD) simulations of contemporary host molecules are routine. This is not the case. Consider first the limitations of water models. One of the best indicators of this is that the melting points of commonly used water models range from −127.55 to 0.75 °C [54]. That water models are approximations is also evident in the fact that there is no readily usable (for MD simulations) models that can faithfully replicate the phase diagram of water; the closer to perfect a water model is made, the more computationally costly it becomes. This is not to say that water models are not to be trusted; most function exceedingly well when measured against one or a small set of metrics. So, if an accurate potential of mean force or free energy of binding is required, TIP4P‐Ew is an excellent choice [30]. However, water contemporary models are not sufficiently accurate that they easily satisfy multiple metrics or, importantly, will accurately reveal serendipitous explanations of physical or chemical properties. Limits to computational power also mean that for MD simulations a “complex” system might consist of homogeneous surfaces or shapes such as parallel plates, tubes, or hemispherical cavities (vide infra). Relatedly, calculations involving the average contemporary host–guest system are exceedingly computationally demanding, even when polarizability is ignored; with current computer power, the inclusion of polarizability limits atomistic MD simulations to molecules much smaller

11

12

1  Water Runs Deep

than  the average contemporary host. Finally, force fields for atoms are largely biased toward those commonly found in Nature. If a researcher wishes to study the binding of most polyatomic inorganic species, she or he may have to build the requisite force fields. In short, although computational chemistry has much to offer the study of aqueous solutions, its limitations are real. These issues noted that even in their absence the hydrophobic effect is difficult to precisely define because it is intimately linked to the hydration of each individual nonpolar solute; even if organic molecules were only composed of carbon and hydrogen, the hydrophobic effect would be exceedingly complex. Consider, for example, a dimerization process that is not possible to empirically determine: that of methane. Calculations are key here. The free energy of dimerization in water is equivalent to the sum of the interaction free energy in the gas phase and the corresponding water‐mediated contribution. If the “methane” monomers are in fact hard spheres, their intrinsic free energy of interaction is zero. In such a case, it is possible to calculate a water‐mediated interaction of ≈RT. In contrast, methane dimerization shows a well depth of 1.2 kJ mol−1 (~1/2 RT at 300 K) and a water‐mediated contribution that can either be attractive or repulsive, depending on the calculation [55–60]. Indeed, the magnitude and sign of the water‐mediated contribution is remarkably sensitive to the precise solute–water interactions [61]. Hence, the weak attraction between water and aliphatic groups means that they are on the knife edge between hydrophilic and hydrophobic regimes [49]; in water, alkyl chains are frequently pushed together, but sometimes they can be pulled apart; it all depends on the precise solute and if present, the precise nature of salts or buffers. One current rule of thumb is that ~1 nm2 of hydrocarbon surface area must be desolvated in order for binding or assembly to out‐compete thermal fluctuations (RT) and that below this value, smaller molecules are held apart by water [61]. This delicate balance arises because: (i) of the complex (and unknown) thermodynamics of non‐spherical cavity formation, (ii) the van der Waals solute–solute interactions are generally weak, and (iii) with polar or charged molecules, there are almost compensating enthalpy and entropy contributions (see Section 1.4). This sets water apart from other solvents; it is as if that sometimes it is there and sometimes it is not. To put another way, binding to a host in organic solvents is commonly inhibited by the solvent. As a result, simple competition (guest vs. solvent) means that binding is weak relative to the gas phase. The weak interactions between solute and water mean, however, that water is not a good competitor and binding is enhanced relative to organic solvents [62]. Irrespective of the precise solvation details, it is paramount to recall this point: the hydrophobic effect is relative to other solvents, not the gas phase; there is no magical force that can “turbocharge” binding beyond the intrinsic gas phase affinity. Unsurprisingly perhaps, this knife‐edge solvation means that the thermodynamics of the hydrophobic effect are mostly dependent on temperature. Thus, while the slightly positive hydration free energies (ΔG) is only weakly dependent, the ΔH and TΔS terms are highly temperature-dependent. Inevitably then, at some temperature ΔH  =  0 (i.e. ΔG  =   − TΔS) and TΔS  =  0 (i.e. ΔG  =  ΔH). Furthermore, as ΔS =  − (∂ΔG/∂T)P, ΔG must attain a maximum when TΔS = 0. This means we must take care when stating that a process is “driven” by enthalpy

1.5  The Three Effects

or entropy. In other words, it is not useful to define a binding event in water as being driven by either the “classical” (entropically driven) or “nonclassical” (enthalpically driven) hydrophobic effect. Entropy changes are not a good thermodynamic signature of this phenomenon. Relatedly, while hydration of an alkane inevitably results in an increase in the heat capacity (ΔCP = dH/dT) of the solution, this phenomenon is not unique to water. Hence this change in ΔCP should not be used to identify the hydrophobic effect. Interestingly, it may, however, be the case that the sizable temperature dependence of ΔCP itself may be a unique signature. A simple view of how solvation is intimately tied to the shape of a solute is to consider how the number of dangling H‐bonds of a solvation shell water – those that are H‐bonded to a solute rather than another water – varies according to its local environment [63]. Little is known about dangling H‐bonds, but the use of Raman‐MCR spectroscopy, which allows the unique solvation shell water molecules around solutes to be probed, suggests that for aromatic surfaces at least, OH–π bonds are 20% weaker than H‐bonds of bulk water but are more flexible. In other words, dangling H‐bonds are entropically favored but less enthalpically favored than H‐bonds to the bulk. Figure 1.3 shows how the number of dangling H‐bonds increases with increasing size of convex solutes and how it increases further with concave surfaces. In other words, strongly positive curvature allows water to form its full complement of water–water H‐bonds, whereas the ultimate in negative curvature – the fully encapsulating sphere – forces a water to form four dangling H‐bonds. The small size of water means that it is capable of finding itself in all of these situations [64].

A

D

A

D

D

D D

A

(a)

D

(d)

A

D 0

1

(b)

D

(c)

1

D

2

(e)

3

(f)

4

Figure 1.3  Idealized representations of the solvation of different shape solutes: small convex (a), large convex (b), flat (c), shallow concave (d), deep concave (e), and fully encapsulating (f ) (half of surface removed for visualization). In each example, a water molecule is shown along with an idealized number of strong H‐bonds (donor, D; acceptor, A) to other water molecules in the solvation shell of the solute. The corresponding numbers of dangling H‐bonds, i.e. H‐bonds to the solute, are shown in the green boxes.

13

14

1  Water Runs Deep

To a first approximation, guest molecules are convex, and it was this shape that was the first utilized by Frank and Evans to probe the hydrophobic effect [41]. Building on Kauzmann’s idea that the hydrophobic effect is dependent on the size of the solute [42], Stillinger was the first to demonstrate that small solutes are solvated differently than large ones [65]. The most successful model for describing how the hydrophobic effect changes with size is the Lum–Chandler–Weeks theory [53, 66]. This unified and general theory of solvation of small and large apolar species in water reveals that in the case of small solutes, the H‐­bonding network of water molecules around the solute is distorted yet complete. The result is a hydration layer denser than that seen in the bulk; the solute is said to be wetted. In these cases, it is found that ΔGsolv scales with the volume of the solute. With solutes of diameter ~1 nm or larger, the distortion of the H‐bond network around the solvents reaches a critical point, and the network breaks. Dangling H‐bonds to the solute begin to form, and as a result the cohesive forces of the water molecules are depleted. This leads to a dewetting of the surface of the solute, where the distance between the solute surface and the average water in the first solvation shell increases relative to the O⋯O distance in bulk water [53]. In such cases, ΔGsolv is dominated by interfacial free energetics and scales with the surface area. This wetting/dewetting transition around 1 nm in diameter leads to an entropy–enthalpy crossover. ΔGsolv is dominated by entropy at small solute size, but enthalpy at sizes greater that ~1 nm in diameter. This solvation crossover has been investigated in depth computationally [67– 72], but it has been hard to verify experimentally because of the low solubility of alkanes. Nevertheless, evidence of dewetting has been obtained by high‐energy X‐ray reflectivity measurements of the interface between water and octadecylsilane monolayers [73, 74], and evidence of a thermodynamic crossover has been obtained from single‐molecule force spectroscopic studies of hydrophobic polymers [75, 76]. Furthermore, a combination of femtosecond 2D‐IR spectroscopy and femtosecond polarization‐resolved vibrational pump–probe spectroscopy has shown a size‐dependent correlated slowing of the vibrational frequency dynamics and orientational mobility of solvation shell water molecules around nonpolar groups [77]. Relatedly, Raman‐MCR spectroscopy of simple alcohols has also provided evidence of a transition, with size‐dependent (and temperature-dependent) changes in solvation shells – from those with greater tetrahedral structure than bulk water to those with dangling H‐bonds [78]. Similar results have also been observed with carboxylic acids and tetraalkylammoniums [79]. Intriguingly, the degree of water structure or the change in the nature of the solvation shell is dependent on whether the solubilizing group is a neutral alcohol, a negatively charged carboxylate, or a positively charged ammonium. Evidently, there is much to learn regarding how functional groups influence nonpolar group hydration, and the complications associated with introducing a charge into a guest solute are exquisitely illustrated by computational work examining the introduction of positive charge into helium‐like particles with diameters ranging from 0 to 30 Å [80]. As anticipated, for the uncharged particle, ΔGsolv became more positive with increasing size, and this was attenuated by the introduction of positive charge. The hydrophobic to hydrophilic crossover was found to be 0.4e, where, interestingly, ΔGsolv was found to be ~0 at all diameters.

1.5  The Three Effects

In other words, unlike hydrophobic solutes, the 0.4e charged solute showed no entropy–enthalpy crossover. Zero curvature, i.e. a flat surface, represents the transition between the convexity of a guest and the concavity of a host, and not surprisingly their hydration is very similar to that of large convex solutes. For example, X‐ray reflectivity measurements and MD simulations revealed significant dewetting of the surface of a crystalline monolayer of n‐C36H74 and the creation of a 1.0 Å wide vacuum layer at the surface [81]. Furthermore, vibrational sum‐frequency generation spectroscopy has demonstrated that D2O near a hydrophobic surface has enhanced orientation and structure and stronger H‐bond interactions relative to the D2O/air interface consistent with dangling H‐bonds and water dipoles perpendicular to the hydrophobic surface [82]. The hydration of a singular flat surface has however only tangential relation to the solvation of either a host or a guest. Of greater interest to the supramolecular community is the substantial body of computational work examining the solvation of parallel plates, which gives an inkling to the solvation of hosts. Sufficiently separated, each plate is hydrated as if in total isolation. However, as separation of the two hydrated plates decreases, so full solvation of the intervening space becomes increasingly difficult. At a critical distance, a drying transition emerges. In other words, the average density of water is found to be lower than the bulk. If water molecules are counted as a function of time, the intervening space is found to oscillate between empty and fully solvated and/or be occupied by an intermediate state of partial solvation. Finally, at a critical distance, the space between the plates fully dries, and hydrophobic collapse occurs. It has been found that this critical separation distance between two nonpolar plates is linearly proportional to their interfacial area [83]. For plates of 2 nm in diameter, the critical distance is ~1 nm. Furthermore, in work supporting this general finding, it was found that plates of area of 1 nm2 only lead to drying transitions at a distance of less than 0.9 nm [84]. Phase transitions in intervening water between plates have also been observed computationally [85]. For example, changing the distance between nonpolar surfaces revealed a liquid–solid phase transition in which the solid phase has a bilayer amorphous structure and a fully connected H‐bond network. This freezing is observed at a separation distance of ~1 nm and was found to be entropically unfavorable but enthalpically favored. Similarly, MD simulations of water confined between pairs of nonpolar or polar nanoscale plates revealed the effects of pressure on the hydration of the intervening space [86]. When water was confined between the nonpolar plates, capillary evaporation occurs between the plates at low pressure, with the smaller separations between plates eliciting drying at higher pressures. Furthermore, at select distances and pressure, the intervening water crystallized into a bilayer ice structure. In contrast, water confined by hydrophilic plates remained in the liquid phase at all pressures and distances studied. Interestingly, it has also been observed that nonpolar/polar patterning on plates plays a fundamental role on inert‐plate hydration [87]. For example, an analysis of five pairs of plates containing equal amounts of nonpolar and polar groups arranged in different patterns revealed both qualitative and  quantitative differences between them. Thus, if all the nonpolar area on

15

16

1  Water Runs Deep

each plate was grouped in the center, dewetting between the plates was observed. However, if the same area of nonpolarity was spaced out evenly on the surface, no dewetting at the same plate separation was seen. In pairing plates that were polar and nonpolar, respectively, it was the latter that was found to dominate wetting/dewetting. Overall there is some ambiguity with the study of parallel plates, in large part because of the water models and precise protocols used. However, it is evident that the larger the surface areas of the plates, the larger the critical distance before drying transitions and/or water structure changes are observed and that in all cases at some minimal critical distance, complete dewetting occurs. The hydration of concave (negatively curved) surfaces (Figure 1.3) is where the exceedingly small size of water and its ability to solvate the tightest of surfaces comes to the fore. This is the domain of water‐soluble hosts; however the most detailed molecular‐level understanding of such surfaces arguably comes from in silico studies of carbon nanotubes. Solvation of the inner bore of nanotubes is similar to that seen with parallel plates, but their negative curvature often results in more extreme examples of structured water or dewetting. For example, with narrow carbon nanotubes of ~8 Å diameter, water wires are observed in which each water molecule has ~2 dangling H‐bonds [88, 89]. As a result of the limited H‐­bonding between these molecules, only a small reduction in the van der Waals attraction between water and tube is sufficient to induce complete dewetting. Water flow through such nanotubes has been shown to occur in bursts and is limited only by the barriers of entry and egress. In other words, the tube itself is essentially frictionless, and the flow rate is nearly independent of its length [90]. Counter­intuitively perhaps, it has been determined that as the diameter of the bore is increased, water transport rates decrease. This has been attributed to increased H‐bonding between water molecules [91]. In general, with slightly wider nanotubes, phase transitions in the occupying water are observed [92, 93]. Thus, widening a tube can convert bore water from a gas‐like state to an ice‐like state or even to stacked layers of pentagons or hexagons [94, 95]. It is thus perhaps not surprising that there is evidently a non­monotonic relationship between diameter and the relative ΔG of filling. Thus, whether or not this filling is dominated by enthalpy or entropy depends on the bore diameter; solvation of 1.1–1.2 nm tubes was observed to be dominated by enthalpy, while solvation of smaller or larger bores was slightly dominated by entropy. Furthermore, with the exception of 1.1–1.2 nm nanotubes, confinement within the carbon nanotubes leads to an increase in the translational entropy of bound water. Again, however, these sorts of effects are model dependent. Relatedly, in silico studies have also been carried out on wholly artificial hosts composed of homogeneous nonpolar surfaces. Thus, in early work Monte Carlo simulations of the thermodynamic stability of water clusters inside smooth graphene‐like spherical cavities and the fullerenes C140 and C180 revealed thermodynamically stable water clusters composed of 3–9 water molecules [96]. The smallest stable water cluster, observed in a 1 nm diameter spherical cavity, was a trimer held together by three H‐bonds. With a slightly smaller diameter of 0.9 nm, a thermodynamically unstable dimer was observed.

1.5  The Three Effects

By way of another example, 0.8 nm diameter hemispherical pockets “carved” out of neutral particles aligned in a hexagonal close‐packed grid have also been studied. These pockets were shown to have average water densities lower than the bulk as a result of fluctuations between empty and filled states [97]. The 0.8 nm diameter pocket was deemed close to a critical size for promoting transitions between gas‐like and liquid‐like phases. Further work comparing this cavity size with a smaller 0.5 nm diameter pocket revealed that although both possessed vapor‐like regions of reduced water density, the smaller pocket possessed a substantial desolvation barrier, whereas the 0.8 nm cavity did not [98, 99]. A more detailed study estimated ΔG°, ΔH°, and TΔS° along a concavity guest binding trajectory as a function of the charge of the host and guest [100, 101]. This revealed a range of thermodynamic signatures in which water enthalpic or entropic contributions drove cavity guest binding or rejection. For example, the binding of a neutral guest to a neutral host was driven by enthalpy but was entropically penalized. Overall, bound water molecules with dangling H‐ bonds were attributed to complexation. The same groups calculated changes in the 1D‐ and 2D‐IR spectra during guest binding, suggesting a strategic route to correlating changes in water structure with identifying signature vibrational spectra. Collaborating earlier work from the Rick group with a cavitand host [102], the wholly artificial pockets exhibited solvated–desolvated oscillations, the magnitude and timescale of which were modulated by an approaching guest [103]. These results suggest that nonpolar guest binding to concavity does not necessarily require a dissociative (SN1‐like) mechanism, but rather can follow a triggered dissociative mechanism in which the pocket spontaneously evacuates on the approach of the guest. As will be discussed in Chapter 2, the supramolecular community has generated a multitude of host families possessing concavity (negative curvature) for guest binding. The diversity within this family, combined with the synthetic prowess of supramolecular chemists, offers countless opportunities to probe the hydrophobic effect (and the Hofmeister effect; vide infra). As should be apparent, there are still many open questions pertaining to the driving forces behind guest complexation in water. However, many fundamental points are well understood. First, although small, water is not a good competitor for a host pocket, and hence guest binding is enhanced relative to organic solvent [62] (recall the hydrophobic effect is relative to other solvents, not the gas phase). It is also well established what the key role of a water‐soluble host is to template cavitation in water. Consider, for example, how difficult it is to make a cavity in water, say, for a 10 (non‐hydrogen)‐atom molecule of volume ~300 Å3. Revised scaled particle theory calculations [104, 105], combined with an analytical equation of state for cavity formation derived from experimental and simulation results [46], reveal that the free energy of formation of a ~300 Å3 cavity is ~100 kJ mol−1. This is a virtually unsurmountable energy requirement for the non‐covalent interactions formed in a host–guest complex to counter. However, as Rick has shown, the free energy of desolvation of such a cavity in a host is only ~20 kJ mol−1 [102]. This idea was suggested many years ago by Bender and coworkers [106] and has also been recently observed in cucurbiturils [107–109]. Thus, to paraphrase Cram, the concave structure of water‐soluble hosts prepays the free energy costs to

17

18

1  Water Runs Deep

promote water cavitation. In other words, the primary role of a water‐soluble host is the energetically reasonable formation of nothingness in water. Hosts template water cavitation. How do enthalpy and entropy factor in to this templation of a cavity? Figure 1.3 tacitly states a partial answer to this: while the hydration of small convex solutes is entropically costly, the solvation of concavity is enthalpically expensive. As just alluded to, the idea that it is enthalpically costly to solvate concavity was first expressed by Bender in 1967, who stated that “water molecules in the cavity cannot form their full complement of H‐bonds as a result of steric restrictions” [106]. Emerging from this has been the idea of “high‐energy” water inside the hosts [107–109] and that this is responsible for what has been called the “nonclassical hydrophobic effect.” However, caution is warranted here. First is with regard to the two points of nomenclature. As previously mentioned, the extreme temperature dependence of ΔH and TΔS means that the terms “classical” and “nonclassical” hydrophobic effects are rather arbitrary; the knife‐edge hydration of nonpolar molecules means that a spontaneous complexation event can switch from enthalpically promoted to entropically promoted with only a small change in temperature (or even within replications of a single experiment). Parenthetically, if associations dominated by enthalpy are termed “nonclassical” and those dominated by entropy “classical,” what are complexation events driven by both to be called? The second nomenclature point is that the term “high energy” means different things to different chemists. Unsurprisingly, there are examples in the literature where “high energy” is used in a free energetic sense and an enthalpic sense. Hence, attributing the “nonclassical hydrophobic effect” to “high‐energy water” is imprecise and unhelpful. There are also fundamental problems with the concept of “high‐energy water.” For example, since the chemical potentials or partial molar Gibbs free energy of all water molecules in an equilibrated system are necessarily the same, using the term in the context of the standard definition of free energy is misleading. Furthermore, there are issues that the field has yet to deal with even if “high energy” is strictly used to mean “high enthalpy.” Thus, while in the most general terms the solvation of a convex guest is entropically penalized and the solvation of a concave host is enthalpically penalized, a key open question is how these two factors combine when a concave host and convex guest are desolvated to form a complex. At first glance it would seem that as host–guest complexations are mostly exothermic, that concavity desolvation dominates over convexity desolvation. However, this has yet to be determined unequivocally. Furthermore, what of entropy? The term “high‐energy water” sidesteps how entropy factors in. Do hosts also promote cavity formation in water by organizing bound water exceptionally well? Water binding to a pocket reduces its entropy of translation, but, being unable to form a full complement of H‐ bonds, a bound water also possesses a countering increase in its entropy of rotation. How these factors contribute to the overall entropy of binding is still unclear. For a simple cavity in water, enthalpy and entropy contribute to the overall free energy cost in a size‐ and temperature‐dependent manner [46, 104, 105].

1.5  The Three Effects

Unfortunately, what we know about cavity formation in water is not likely to directly pertain to what happens in the cavity of a host. All it does is serve as a warning as to how complex host pocket desolvation is likely to be. There is much to learn about how enthalpy and entropy combine to affect cavity solvation and how this changes as a function of cavity size. The complexity of thermodynamic analyses aside, hosts do promote cavitation in water. Thus, until a clearer thermodynamic picture of the solvation of concave hosts is forthcoming, discussions of cavitation are arguably best described in the context of drying transitions. Along these lines, mapping the water occupancy of different cavities, for example, with software such as WaterMap [110], which uses explicit water MD simulations and inhomogeneous solvation theory [111] to calculate the enthalpy, entropy, and free energy of water molecules within a solvated binding site relative to bulk water [112], may be helpful to begin to understand how cavity shape affects solvation and hence guest binding. In summary, the fundamentals of the hydrophobic effect are exceedingly complex. There is much to learn about the factors that control the solvation thermodynamics of both hosts and guests and the spectroscopic properties of their solvation shells. But there are also significant opportunities to take what is currently known about water and the hydration of nonpolar (and polar) molecules and apply it to understanding and engineering host–guest complexation and assembly processes. 1.5.2  The Hofmeister Effect In reference to Figure  1.2, the second prominent phenomenon observed in aqueous solution is the Hofmeister effect [113–116]. All life as we know it involves salty solutions, and therefore understanding how salts interact with water and how they interact with other solutes is key to a myriad of different sciences. For example, a general understanding of the supramolecular properties of ions would improve computational modeling and hence reduce costs and  speed up ligand identification in computer‐aided drug discovery [117]. Similarly, an improved understanding of the supramolecular properties of chloride ions would improve our understanding of ion transport and diseases such as cystic fibrosis [118]. The Hofmeister effect was first reported between 1887 and 1898 in a series of papers published by Franz Hofmeister, the most important one of which was titled Concerning regularities in the protein‐precipitating effects of salts and the relationship of these effects to the physiological behavior of salts [119, 120]. This paper described the ordering of various salts according to their ability to precipitate egg white proteins from aqueous solutions. The salts used had either a common cation or anion and therefore allowed for the ready separation of cationic or anionic effects. This series of ions is now known as the lyotropic or Hofmeister series, but in truth there is no reason to limit studies to salts from Hofmeister’s original selection; Hofmeister’s choice of salts was based on availability, not on what can and what cannot induce the Hofmeister effect.

19

20

1  Water Runs Deep Salting out –

F ~

SO42–

Salting in –

–

–

–

> MeCO2– > Cl > NO3 > Br > I > CIO4– > SCN

–

(a) Salting out +

Cs >

Rb+

>

Salting in NH4+

+

+

+

> K > Na > Li > Ca

2+

> Mg

2+

> Zn2+

(b)

Figure 1.4  Typical ordering of anions (a) and cation (b) in the Hofmeister series.

Subsequent to Hofmeister’s initial discovery, an extensive body of work has been built up. What has intrigued so many scientists is that irrespective of the dependent variable being probed in an experiment, an ordinal sequence such as that shown in Figure 1.4 is observed. Every conceivable bulk property of water or aqueous solution has been investigated at some time, but arguably the most important (or prominent) experiments concerned how salts affected proteins or other biomacromolecules. Hence salting‐out salts usually induce the precipitation of a protein, while salting‐in salts will typically increase its solubility. In other words, species such as F− appears to increase the hydrophobic effect, while I− appears to weaken it. Studies probing the effects of cations vs. anions reveal that the anionic Hof­ meister effect is more prominent. As a result, the reproducibility of the anionic Hofmeister series in Figure 1.4 is much stronger. There are three important reasons as to why this is so. First, there are far more inorganic anions than metal cations. Second, in general anions are larger than cations; the second largest monovalent metal cation, Cs+, is approximately the same size as the second smallest halide Cl−. On average, this larger size leads to a greater charge diffusivity, a weaker hydration shell (though importantly, not a lower free energy of hydration), and the possibility of closer non‐covalent interactions with other solutes. These points mean that not only are the Hofmeister effects stronger with anions, but also by virtue of sheer numbers, they are more prevalent. A third factor behind the prominence of anionic Hofmeister effects is that biomacromolecules are by far the number one target in Hofmeister studies, and in these the negatively charged groups (carboxylates, phosphates, and sulfates) tend to be more strongly solvated than their complementary cationic groups (primarily ammonium and guanidinium). As a result, when a salt is added to the solution of a solute, it is the anion that is more likely to affect it. The model Hofmeister formulated to account for these observations was based on the theory of electrolytic dissociation developed by Arrhenius and Ostwald, and tried to link the observed ordering of the ions with their strength of hydration or, as it was known at the time, their water‐absorbing effects. The implication was that a salt either “pulled” water from the hydration shell of a protein or

1.5  The Three Effects

it did not. Over time, this idea morphed into the theory of salts being either water structure makers or water structure breakers, and key to this development was that, due to technological restraints at the time, researchers were limited to studying bulk phenomena. Thus, the seminal work of Jones and Dole measuring the viscosity of solutions of strong electrolytes culminated in what is now called the Dole–Jones equation, an expression describing the relationship between the viscosity of a solution and the concentration of solute [121]. Extending this work (and that of Bernal and Fowler describing water structure  [26]), Cox and Wolfenden subsequently observed that the sign of the coefficient characterizing solute–solvent interactions in the Dole–Jones equation (the β‐coefficient) is correlated with the temperature coefficient of the electrical mobility of an ion [122]. Furthermore, they logically tied the mobility of an ion to the local viscosity of the water molecules in its solvation shell and formulated the idea that ions with positive (negative) β‐coefficients “polymerized” (“depolymerized”) the water. This idea was expanded upon by Gurney, who discussed ions as bringing local order or disorder to water [123]. Interestingly, Gurney mostly used the wording “local order or disorder” in the early pages of his text but in the latter sections tended to drop “local” and just discuss the effects of ions as inducing order or disorder. This was perhaps an omen. For many subsequent years, the idea that ions are either water structure makers or breakers became dogma, and it has only been in the last three decades or so that new developments in spectroscopic, computational, and macromolecular design strategies have downgraded the role of water‐mediated ion effects on proteins (vide infra) [114]. But before going into the Hofmeister effect further, it is important to highlight two heavily utilized terms in the field: “kosmotrope” [124] and “chaotrope” [125]. Originally these terms were coined to describe the effect ions had on biomacromolecular structure. Thus, salting‐out salts were called kosmotropes (from the Greek noun kosmos or order) because they were noted to stabilize the fold of proteins. In contrast, salting‐in salts, or chaotropes, were noted to destabilize the fold of a protein and hence bring chaos. However, these terms have gone through a linguistic shift in their use and meaning and have subsequently become synonymous with the idea of water structure making and breaking [113]. Like the hydrophobic effect these terms have come to be used deus ex machina to describe the effects of salts. This is an unfortunate state of affairs for, as we will discuss, there is increasing evidence that at moderate concentration salts do no greatly modify the structure of water. Hence, describing salts as kosmotropes/water structure makers or chaotropes/water structure breakers is ambiguous at best and quite possibly wholly incorrect. Avoiding these issues, terms such as “salting in” and “salting out” have become more popular (although they too have issues; vide infra). So, what are the roots of the Hofmeister effect? Our current understanding of salting‐out salts is relatively straightforward. Thus a 5 M solution of NaSO4 requires so many water molecules of solvation that there is insufficient water left to solubilize the protein. Hence it is precipitated from solution. On the other hand, salting‐in salts are now thought to directly interact with a protein to weaken its fold and apparently attenuate the hydrophobic effect [114] and can do so at

21

22

1  Water Runs Deep

much lower concentrations, e.g. 2 mM in the case of a synthetic host–guest pair [126]. There are two reasons why direct ion–solute interactions are now favored over water‐mediated interactions. First, with improvements in spectroscopy and calorimetry, it is now evident that salts do not influence the structure of bulk water. For example, dielectric relaxation data show that there are six slow water molecules hydrating the Mg2+ ion in Mg(ClO4)2, one slow water molecule associated with hydrating SO42− in Cs2SO4, but eighteen slow water molecules hydrating the ions of MgSO4; there is synergy in ion pairing of divalent ions. Importantly, however, there is no evidence that this synergy extends beyond the solvation shell [127–129]. Similarly, Raman spectroscopy and Monte Carlo simulations have determined that an ion exerts little influence beyond this solvation shell [130], while a combination of vibrational sum‐frequency spectroscopy and surface potential measurements upon a monolayer has done likewise [131]. Finally, pressure perturbation calorimetry, which measures the heat transfer resulting from a pressure change above the sample solution, has failed to reveal bulk changes to water structure upon the addition of salts [132]. Briefly, if bulk water structure is changed by a salt, the sign of how the heat capacity (CP) changes as a function of pressure (P) at constant temperature (∂CP/∂P)T should correlate with whether a salt is a water structure maker (negative) or a water structure breaker (positive). However, no such correlation has been found. Taken together, these results suggest that the salting-in Hofmeister effect does not involved significant water‐ mediated effects and therefore must involve direct ion–­ macromolecule interactions [133]. So, if anions do not influence solutes through water, where do anions bind in proteins? NMR, thermodynamic, and MD studies of an uncharged 600‐residue elastin‐like polypeptide, (VPGVG)120, found that weakly hydrated anions complex primarily to binding sites composed of an amide nitrogen atom and an adjacent α‐carbon atom [134]. In addition, weaker anion association to the nonpolar side chains of the polypeptide was also observed. Similar affinities of weakly hydrated anions have also been seen at the liquid water surface, frequently used as a surrogate for nonpolar surfaces [135]. Furthermore, such nonpolar to anion interactions have been quantified using NMR and ITC in a host–guest system [44, 136]. These studies revealed significant affinity of these types of anions for nonpolar concavity, as well as a salting‐out‐type effect induced by cation complexation to solubilizing carboxylate groups. All of this data points to the salting‐in phenomenon of the Hofmeister effect being engendered by the weak solvation of anions allowing them to interact with nonpolar surfaces. There are likely equivalent (but weaker) interactions involving cations, but these have not been explored to nearly the same extent. Key to progress here is most certainly a thorough understanding of the hydration of ions. To highlight just one example, even if the free energy of hydration data available in the literature was comprehensive (it is not), it would not be sufficient. What is needed is a thorough understanding of the “plasticity” of a solvation shell, i.e. how energetically feasible is it to push some water molecules in the hydration shell of an ion aside and allow non‐covalent interactions with other solutes. Only once this has been determined can the supramolecular properties of an ion be more thoroughly explained.

1.5  The Three Effects

1.5.3  The Reverse Hofmeister Effect Weakly solvated anions can also induce the reverse Hofmeister effect, and, not surprisingly, this phenomenon can also teach us much about living systems. For example, the reverse Hofmeister effect has been linked to protein deposition in the cardiovascular system and thrombosis [137], as well as Alzheimer’s [138] and Parkinson’s disease [139]. As depicted in Figure  1.2, whereas weakly hydrated ion interactions with nonpolar groups induce the salting‐in phenomena, the same anions can induce the reverse Hofmeister effect by ion pairing. There is an important ramification associated with this point. Namely, the existence of the reverse Hofmeister effect means that the Hofmeister effect in general can only readily be explained by direct ion–solute interactions; it is hard to reconcile how a salt might be a water structure breaker in the presence of one protein, but a water structure maker in the presence of another. The reverse Hofmeister effect also emphasizes a problem with the salt‐in/salt‐out nomenclature, viz that salts such as NaI can be both. The most prominent example of the reverse Hofmeister effect is the enzyme lysozyme [140–142]. The precipitation of lysozyme follows the Hofmeister series at high pH or high ionic strength, but under neutral and acidic conditions, when most/all acidic and basic groups are protonated, there is an apparent reversal of the Hofmeister series [143, 144]. In other words, weakly hydrated anions salt out lysozyme better than strongly hydrated ones. NMR and MD studies with tripeptide models have shown that the reverse Hofmeister effect in proteins such as lysozyme is rooted in charge neutralization [145]. In other words, weakly solvated anions can closely associate with positive charges on a protein and induce precipitation. This has ramifications for the one third of the proteome consisting of proteins with a pI > 7 [146]. It is useful to note here the relationship between the reverse Hofmeister effect and the solubility of small ammonium compounds in water. Thus, it is well known that to make an ammonium salt more water soluble, the chloride (Cl−) salt might be formed, whereas if organic solvent solubility is required, the perchlorate (ClO4−) salt might be targeted. Organic chemists have been content with this rule of thumb and have therefore not investigated this phenomenon further. However, it seems that the study of the solubility of small ammonium ions in water could lead to a new level of understanding of the reverse Hofmeister effect in biomacromolecules. The latency of the reverse Hofmeister effect is apparent even with simple salts [127]. Thus, dielectric relaxation spectroscopy, far‐infrared (terahertz) absorption spectroscopy, femtosecond mid‐infrared spectroscopy, and X‐ray spectroscopy and scattering, as well as MD simulations, have all revealed that it is inappropriate to think of salts forming statistical mixtures of fully hydrated ions [127]. Rather, ion pairing and ion clustering are in fact common. For example, even monovalent salts such as alkali halides or carboxylates form solvent‐separated ion pairs (M+⋯H2O⋯H2O⋯A−), solvent‐shared ion pairs (M+⋯H2O⋯A−), and, to a small extent, contact ion pairs (M+⋯A−). Population distributions are shifted toward contact ion pairs in the case of higher valency ions, but unfortunately, we have not

23

24

1  Water Runs Deep

yet formed a thorough understanding of how ion specific these distribution profiles are, nor the extent of specific ion‐pairing effects. More importantly, weakly hydrated anions such as thiocyanate (SCN−) and tetra‐n‐alkylammoniums (R4N+) can form clusters or aggregates. For example, in 1 M solutions of KSCN, about 20−30% of anions are in clusters of an average size of two to three (with an approximately equal number of cations) [147]. It is therefore now apparent that aggregates involving CsSCN or Me4NSCN, or those involving more polarizable anions, would likely demonstrate larger aggregates or aggregates at lower concentrations. Along these lines of thought, recent studies of a host–guest system revealed specific anion binding sites and affinities in the host and allowed anion binding to be traced to aggregation and ultimately (at high enough concentration) precipitation of the host and its complexes [126]. Some time ago, Collins formulated what he called the law of matching water affinities [148], a conceptual framework that states that the charge density of an ion is the important physical variable for specificity of ion pairing. Thus, Collins used hydration enthalpies [149] to build a rule of thumb that states the free energy of ion pairing tends to be more favorable if the anion and cation are of similar size, rather than if one ion is small and hard, while the other is large and polarizable. This rule of thumb has been difficult to unequivocally verify, and much work still needs to be carried out. Undoubtedly, an expanded and stronger framework would have to take into account findings from reverse Hofmeister effect studies. There is still much to learn here.

1.6 ­Conclusions and Future Work The three effects, the hydrophobic effect, the Hofmeister effect, and the reverse Hofmeister effect, are intrinsically linked to each other by relatively weak solvation of solutes and the high cohesiveness of water. In studying the hydrophobic effect, scientists look to describe the attraction between nonpolar molecules in an aqueous environment. In contrast, investigating the salting‐in Hofmeister effect turns the attention toward “greasy” polarizable ions interacting with similarly greasy neutrals. Evidence points to anions being more important, but “greasy” cations have not been investigated as thoroughly. Finally, the focus of the reverse Hofmeister effects concerns how the same “greasy” anions and cations associate with each other. Arguably, a significant handicap impeding progress on all three fronts has been the tendency to investigate structurally complex macromolecules such as proteins. It was the proteinaceous world that first identified the Hofmeister effects, and proteins have also been a staple of examining the hydrophobic effect, but as all three effects may or may not be present in one protein, and as these phenomena often counter one another, the big picture has been exceedingly difficult to visualize (let alone understand). The supramolecular community has much to offer here. Reductionist philosophies have their role to play in science, and from the authors’ perspective, it is the reductionist viewpoint of supramolecular chemists that is needed to move aqueous supramolecular chemistry forward. They cannot do so alone of course; the water community – a diverse group of researchers from many different branches

­  References

of science – is the water expert. But it is the supramolecular chemistry community who has a broad set of tools to offer the water community, tools, it should be added, that are mostly brand new to majority of the latter. It is up to individual scientists, both within the water and supramolecular communities, to decide how they should reach out across the divide and bridge these two areas [28]. But build bridges they must. There are obviously countless possibilities here, but one anecdotal example familiar to the authors may be illustrative. Approximately one quarter of the average ~$2.6 billion needed to bring a new drug to market arises from preclinical studies to identify ligands with suitable drug‐like properties and receptor affinities. Consequently, improving ligand identification is key. The Drug Design Data Resource (D3R; https://drugdesigndata.org/) aims to “advance the technology of computer‐aided drug discovery through the interchange of high quality protein–ligand datasets and workflows, and by holding community‐wide, blinded prediction challenges.” These blind challenges with proteins are, to say the least, challenging. Hence, one relatively new component of the D3R is the Statistical Assessment of the Modelling of Proteins and Ligands (SAMPL) [150]. The main thrust of the annual SAMPL exercise consists of blind challenges in which a supramolecular chemistry group collects but does not disseminate thermodynamic data for host–guest complexation targets, while computational groups attempt to calculate the host–guest affinities a priori. The ultimate release of the empirical data allows computational chemists to evaluate the success of their latest strategies. By such endeavors, the D3R hopes to ultimately improve ligand affinity predictions such that rapid and cheap computer calculations can replace the majority of early lead screening and syntheses from the drug development pipeline. The annual SAMPL exercise is but one example of how the two communities can create new knowledge that would otherwise be hard to obtain. By such endeavors science gains much better control of the aqueous world while simultaneously pushing human knowledge further into unknown territories.

­Acknowledgments The authors gratefully acknowledge the support of the National Institutes of Health (GM 125690). N. E. E. also acknowledges the Louisiana Board of Regents for a graduate student fellowship (LEQSF(2016‐21)‐GF‐12). The authors also express their gratitude to Dor Ben‐Amotz for helpful discussions and Paolo Suating for aid with the graphics in this chapter.

­References 1 Trigger, B.G. (2003). Understanding Early Civilizations: A Comparative Study.

Cambridge University Press.

2 Wright, H.T. (1990). Rise of civilizations: Mesopotamia to Mesoamerica.

Archaeology 42: 46–48, 96–100.

25

26

1  Water Runs Deep

3 Lagasse, P., Goldman, L., Hobson, A., and Norton, S. (eds.) (2000). Civilization.

In: The Columbia Encyclopedia, 6e. Visible Ink Press.

4 Gleick, P.H. (2006). Water and terrorism. Water Policy 8: 481–503. Peter, H. and Gleick, M.H. (2012). Water Conflict Chronology. Washington, DC: 5

Island Press.

6 Gupta, V.K., Ali, I., Saleh, T.A. et al. (2012). Chemical treatment technologies for

waste‐water recycling – an overview. RSC Adv. 2: 6380–6388.

7 Karagiannis, I.C. and Soldatos, P.G. (2008). Water desalination cost literature:

review and assessment. Desalination 223: 448–456.

8 Sethi, S.K. and Manik, G. (2018). Recent progress in super hydrophobic/

9

10 11 12 13 14 15 16 17 18 19 20 21 22

23 24 25 26 27

hydrophilic self‐cleaning surfaces for various industrial applications: a review. Polym. Plast. Technol. Eng. 1–21. Snyder, P.W., Lockett, M.R., Moustakas, D.T., and Whitesides, G.M. (2013). Is it the shape of the cavity, or the shape of the water in the cavity? Eur. Phys. J.: Spec. Top. 223: 853–891. Whitesides, G.M. (1999). Complexity in chemistry. Science 284: 89–92. Rousseau, D.L. (1992). Case‐studies in pathological science. Am. Sci. 80: 54–63. Davenas, E., Beauvais, F., Amara, J. et al. (1988). Human basophil degranulation triggered by very dilute antiserum against IgE. Nature 333: 816–818. Maddox, J., Randi, J., and Stewart, W.W. (1988). “High‐dilution” experiments a delusion. Nature 334: 287–291. Lippincott, E.R., Stromberg, R.R., Grant, W.H., and Cessac, G.L. (1969). Polywater. Science 164: 1482–1487. Derjaguin, B.V. (1970). Superdense water. Sci. Am. 223: 52–71. Vonnegut, K. (1963). Cat’s Cradle. Dell. Ovchinnikova, K. and Pollack, G.H. (2009). Can water store charge? Langmuir 25: 542–547. Ovchinnikova, K. and Pollack, G.H. (2009). Reply to comment on “Can water store charge?”. Langmuir 25: 11202. Corti, H.R. and Colussi, A.J. (2009). Do concentration cells store charge in water? Comment on “Can water store charge?”. Langmuir 25: 6587–6589. Ball, P. (2008). Water as an active constituent in cell biology. Chem. Rev. 108: 74–108. Israelachvili, J. and Wennerstrom, H. (1996). Role of hydration and water structure in biological and colloidal interactions. Nature 379: 219–225. West, J.B. (2014). Henry Cavendish (1731–1810): hydrogen, carbon dioxide, water, and weighing the world. Am. J. Physiol.: Lung Cell. Mol. Physiol. 307: L1–L6. Lipeles, E.S. (1983). The chemical contributions of Amadeo Avogadro. J. Chem. Educ. 60: 127–128. Barnes, W.H. (1929). The crystal structure of ice between 0 °C and −183 °C. Proc. R. Soc. London, Ser. A 125: 670–693. Kuhs, W.F. and Lehmann, M.S. (1983). The structure of the ice Ih by neutron diffraction. J. Phys. Chem. 87: 4312–4313. Bernal, J.D. and Fowler, R.H. (1933). A theory of water and ionic solution, with particular reference to hydrogen and hydroxyl ions. J. Chem. Phys. 1: 515–548. Eyring, H., Ree, T., and Hirai, N. (1958). Significant structures in the liquid state Proc. Natl. Acad. Sci. U.S.A. 44: 683.

­  References

28 Cremer, P.S., Flood, A.H., Gibb, B.C., and Mobley, D.L. (2017). Collaborative

29

30 31 32

33

34 35 36 37 38 39 40

41

42 43

44 45

46

routes to clarifying the murky waters of aqueous supramolecular chemistry. Nat. Chem. 10: 8–16. Horn, H.W., Swope, W.C., Pitera, J.W. et al. (2004). Development of an improved four‐site water model for biomolecular simulations: TIP4P‐Ew. J. Chem. Phys. 120: 9665–9678. Clark, G.N.I., Cappa, C.D., Smith, J.D. et al. (2010). The structure of ambient water. Mol. Phys. 108: 1415–1433. Keutsch, F.N. and Saykally, R.J. (2001). Water clusters: untangling the mysteries of the liquid, one molecule at a time. Proc. Natl. Acad. Sci. U.S.A. 98: 10533–10540. Perez, C., Muckle, M.T., Zaleski, D.P. et al. (2012). Structures of cage, prism, and book isomers of water hexamer from broadband rotational spectroscopy. Science 336: 897–901. Cole, W.T., Farrell, J.D., Wales, D.J., and Saykally, R.J. (2016). Structure and torsional dynamics of the water octamer from THz laser spectroscopy near 215 μm. Science 352: 1194–1197. Smith, J.D., Cappa, C.D., Wilson, K.R. et al. (2004). Energetics of hydrogen bond network rearrangements in liquid water. Science 306: 851–853. Head‐Gordon, T. and Johnson, M.E. (2006). Tetrahedral structure or chains for liquid water. Proc. Natl. Acad. Sci. U.S.A. 103: 7973–7977. Wernet, P., Nordlund, D., Bergmann, U. et al. (2004). The structure of the first coordination shell in liquid water. Science 304: 995–999. Soper, A.K. and Phillips, M.G. (1986). A new determination of the structure of water at 25 °C. Chem. Phys. 107: 47–60. Eaves, J.D., Loparo, J.J., Fecko, C.J. et al. (2005). Hydrogen bonds in liquid water are broken only fleetingly. Proc. Natl. Acad. Sci. U.S.A. 102: 13019–13022. Sharp, K.A. (2001). Water: structure and properties. In: Encyclopedia of Life Sciences. Chichester: Wiley. Chen, M., Zheng, L., Santra, B. et al. (2018). Hydroxide diffuses slower than hydronium in water because its solvated structure inhibits correlated proton transfer. Nat. Chem. 10: 413–419. Frank, H.S. and Evans, M.W. (1945). Free volume and entropy in condensed systems III. Entropy in binary liquid mixtures; partial molal entropy in dilute solutions; structure and thermodynamics in aqueous electrolytes. J. Chem. Phys. 13: 507–532. Kauzmann, W. (1959). Some factors in the interpretation of protein denaturation. Adv. Protein Chem. 14: 1–63. Ben‐Amotz, D. and Underwood, R. (2008). Unraveling water’s entropic mysteries: a unified view of nonpolar, polar, and ionic hydration. Acc. Chem. Res. 41: 957–967. Sokkalingam, P., Shraberg, J., Rick, S.W., and Gibb, B.C. (2016). Binding hydrated anions with hydrophobic pockets. J. Am. Chem. Soc. 138: 48–51. Assaf, K.I., Ural, M.S., Pan, F. et al. (2015). Water structure recovery in chaotropic anion recognition: high‐affinity binding of dodecaborate clusters to γ‐cyclodextrin. Angew. Chem. Int. Ed. 54: 6852–6856. Ben‐Amotz, D. (2005). Global thermodynamics of hydrophobic cavitation, dewetting, and hydration. J. Chem. Phys. 123: 184504.

27

28

1  Water Runs Deep

47 Ben‐Amotz, D. and Widom, B. (2006). Generalized solvation heat capacities.

J. Phys. Chem. B 110: 19839–19849.

48 Ben‐Naim, A. (1987). Solvation Thermodynamics. New York: Plenum Press. 49 Ben‐Amotz, D. (2016). Water‐mediated hydrophobic interactions. Annu. Rev.

Phys. Chem. 67: 617–638.

50 Blokzijl, W. and Engberts, J.B.F.N. (1993). Hydrophobic effects. Opinions and

facts. Angew. Chem. Int. Ed. Engl. 32: 1545–1579.

51 Sharp, K.A. and Vanderkooi, J.M. (2010). Water in the half shell: structure of

water, focusing on angular structure and solvation. Acc. Chem. Res. 43: 231–239.

52 Southall, N.T., Dill, K.A., and Haymet, A.D.J. (2002). A view of the hydrophobic

effect. J. Phys. Chem. B 106: 521–533.

53 Chandler, D. (2005). Interfaces and the driving force of hydrophobic assembly.

Nature 437: 640–647.

54 Vega, C., Sanz, E., and Abascal, J.L. (2005). The melting temperature of the most

common models of water. J. Chem. Phys. 122: 114507.

55 Koga, K. (2013). Osmotic second virial coefficient of methane in water. J. Phys.

Chem. B 117: 12619–12624.

56 Smith, D.E. and Haymet, A.D.J. (1993). Free energy, entropy, and internal energy

57

58

59

60

61 62

63 64

65

66

of hydrophobic interactions: computer simulations. J. Chem. Phys. 98: 6445–6454. Gupta, R. and Patey, G.N. (2012). Aggregation in dilute aqueous tert‐butyl alcohol solutions: insights from large‐scale simulations. J. Chem. Phys. 137: 034509. Ghosh, M.K., Uddin, N., and Choi, C.H. (2012). Hydrophobic and hydrophilic associations of a methanol pair in aqueous solution. J. Phys. Chem. B 116: 14254–15260. Morrone, J.A., Li, J., and Berne, B.J. (2012). Interplay between hydrodynamics and the free energy surface in the assembly of nanoscale hydrophobes. J. Phys. Chem. B 116: 378–389. Makowski, M., Czaplewski, C., Liwo, A., and Scheraga, H.A. (2010). Potential of mean force of association of large hydrophobic particles: toward the nanoscale limit. J. Phys. Chem. B 114: 993–1003. Ben‐Amotz, D. (2015). Hydrophobic ambivalence: teetering on the edge of randomness. J. Phys. Chem. Lett. 6: 1696–1701. Houk, K.N., Leach, A.G., Kim, S.P., and Zhang, X. (2003). Binding affinities of host–guest, protein–ligand, and protein–transition‐state complexes. Angew. Chem. Int. Ed. 42: 4872–4897. Hillyer, M.B. and Gibb, B.C. (2016). Molecular shape and the hydrophobic effect. Annu. Rev. Phys. Chem. 67: 307–329. Lynden‐Bell, R.M., Giovambattista, N., Debenedetti, P.G. et al. (2011). Hydrogen bond strength and network structure effects on hydration of non‐polar molecules. Phys. Chem. Chem. Phys. 13: 2748–2757. Stillinger, F.H. (1973). Structure in aqueous solutions of nonpolar solutes from the standpoint of scaled‐particle theory. In: The Physical Chemistry of Aqueous System (ed. R.L. Kay), 43–60. New York: Plenum Press. Lum, K., Chandler, D., and Weeks, J.D. (1999). Hydrophobicity at small and large length scales. J. Phys. Chem. B 103: 4570–4577.

­  References

67 Rajamani, S., Truskett, T.M., and Garde, S. (2005). Hydrophobic hydration from

68

69 70 71

72

73 74

75 76 77

78 79

80

81

82 83 84 85

small to large lengthscales: understanding and manipulating the crossover. Proc. Natl. Acad. Sci. U.S.A. 102: 9475–9480. Athawale, M.V., Jamadagni, S.N., and Garde, S. (2009). How hydrophobic hydration responds to solute size and attractions: theory and simulations. J. Chem. Phys. 131: 115102. Mittal, J. and Hummer, G. (2008). Static and dynamic correlations in water at hydrophobic interfaces. Proc. Natl. Acad. Sci. U.S.A. 105: 20130–20135. Ashbaugh, H.S. (2009). Entropy crossover from molecular to macroscopic cavity hydration. Chem. Phys. Lett. 477: 109–111. Huang, X., Margulis, C.J., and Berne, B.J. (2003). Do molecules as small as neopentane induce a hydrophobic response similar to that of large hydrophobic surfaces? J. Phys. Chem. B 107: 11742–11748. Gallicchio, E., Kubo, M.M., and Levy, R.M. (2000). Enthalpy–entropy and cavity decomposition of alkane hydration free energies: numerical results and implications for theories of hydrophobic solvation. J. Phys. Chem. B 104: 6271–6285. Poynor, A., Hong, L., Robinson, I.K. et al. (2006). How water meets a hydrophobic surface. Phys. Rev. Lett. 97: 266101. Mezger, M., Reichert, H., Schoder, S. et al. (2006). High‐resolution in situ X‐ray study of the hydrophobic gap at the water‐octadecyl‐trichlorosilane interface. Proc. Natl. Acad. Sci. U.S.A. 103: 18401–18404. Li, I.T. and Walker, G.C. (2011). Signature of hydrophobic hydration in a single polymer. Proc. Natl. Acad. Sci. U.S.A. 108: 16527–16532. Garde, S. and Patel, A.J. (2011). Unraveling the hydrophobic effect, one molecule at a time. Proc. Natl. Acad. Sci. U.S.A. 108: 16491–16492. Bakulin, A.A., Pshenichnikov, M.S., Bakker, H.J., and Petersen, C. (2011). Hydrophobic molecules slow down the hydrogen‐bond dynamics of water. J. Phys. Chem. A 115: 1821–1829. Davis, J.G., Gierszal, K.P., Wang, P., and Ben‐Amotz, D. (2012). Water structural transformation at molecular hydrophobic interfaces. Nature 491: 582–585. Davis, J.G., Zukowski, S.R., Rankin, B.M., and Ben‐Amotz, D. (2015). Influence of a neighboring charged group on hydrophobic hydration shell structure. J. Phys. Chem. B 119: 9417–9422. Pascal, T.A., Lin, S.T., Goddard, W. III, and Jung, Y. (2012). Stability of positively charged solutes in water: a transition from hydrophobic to hydrophilic. J. Phys. Chem. Lett. 3: 294–298. Jensen, T.R., Ostergaard Jensen, M., Reitzel, N. et al. (2003). Water in contact with extended hydrophobic surfaces: direct evidence of weak dewetting. Phys. Rev. Lett. 90: 086101. Strazdaite, S., Versluis, J., Backus, E.H., and Bakker, H.J. (2014). Enhanced ordering of water at hydrophobic surfaces. J. Chem. Phys. 140: 054711. Huang, X., Margulis, C.J., and Berne, B.J. (2003). Dewetting‐induced collapse of hydrophobic particles. Proc. Natl. Acad. Sci. U.S.A. 100: 11953–11958. Sharma, S. and Debenedetti, P.G. (2012). Free energy barriers to evaporation of water in hydrophobic confinement. J. Phys. Chem. B 116: 13282–13289. Koga, K. (2002). Solvation forces and liquid–solid phase equilibria for water confined between hydrophobic surfaces. J. Chem. Phys. 116: 10882–10889.

29

30

1  Water Runs Deep

86 Giovambattista, N., Rossky, P.J., and Debenedetti, P.G. (2006). Effect of pressure

87

88 89 90 91

92 93 94

95

96 97

98

99

100 101 102 103

104

on the phase behavior and structure of water confined between nanoscale hydrophobic and hydrophilic plates. Phys. Rev. E 73: 041604. Hua, L., Zangi, R., and Berne, B.J. (2009). Hydrophobic interactions and dewetting between plates with hydrophobic and hydrophilic domains. J. Phys. Chem. C 113: 5244–5253. Hummer, G., Rasaiah, J.C., and Noworyta, J.P. (2001). Water conduction through the hydrophobic channel of a carbon nanotube. Nature 414: 188–190. Kofinger, J., Hummer, G., and Dellago, C. (2011). Single‐file water in nanopores. Phys. Chem. Chem. Phys. 13: 15403–15417. Kalra, A., Garde, S., and Hummer, G. (2003). Osmotic water transport through carbon nanotube membranes. Proc. Natl. Acad. Sci. U.S.A. 100: 10175–10180. Ohba, T., Kaneko, K., Endo, M. et al. (2013). Rapid water transportation through narrow one‐dimensional channels by restricted hydrogen bonds. Langmuir 29: 1077–1082. Takaiwa, D., Hatano, I., Koga, K., and Tanaka, H. (2008). Phase diagram of water in carbon nanotubes. Proc. Natl. Acad. Sci. U.S.A. 105: 39–43. Ohba, T. (2014). Size‐dependent water structures in carbon nanotubes. Angew. Chem. Int. Ed. 53: 8032–8036. Pascal, T.A., Goddard, W.A., and Jung, Y. (2011). Entropy and the driving force for the filling of carbon nanotubes with water. Proc. Natl. Acad. Sci. U.S.A. 108: 11794–11798. Mashl, R.J., Joseph, S., Aluru, N.R., and Jakobsson, E. (2003). Anomalously immobilized water: a new water phase induced by confinement in nanotubes. Nano Lett. 3: 589–592. Vaitheeswaran, S., Yin, H., Rasaiah, J.C., and Hummer, G. (2004). Water clusters in nonpolar cavities. Proc. Natl. Acad. Sci. U.S.A. 101: 17002–17005. Setny, P. and Geller, M. (2006). Water properties inside nanoscopic hydrophobic pocket studied by computer simulations. J. Chem. Phys. 125: 144717. Setny, P. (2007). Water properties and potential of mean force for hydrophobic interactions of methane and nanoscopic pockets studied by computer simulations. J. Chem. Phys. 127: 054505. Setny, P. (2008). Hydrophobic interactions between methane and a nanoscopic pocket: three dimensional distribution of potential of mean force revealed by computer simulations. J. Chem. Phys. 128: 125105. Baron, R., Setny, P., and McCammon, J.A. (2010). Water in cavity‐ligand recognition. J. Am. Chem. Soc. 132: 12091–12097. Setny, P., Baron, R., and McCammon, J.A. (2010). How can hydrophobic association be enthalpy driven? J. Chem. Theory Comput. 6: 2866–2871. Ewell, J., Gibb, B.C., and Rick, S.W. (2008). Water inside a hydrophobic cavitand molecule. J. Phys. Chem. B 112: 10272–10279. Setny, P., Baron, R., Michael, Kekenes‐Huskey, P. et al. (2013). Solvent fluctuations in hydrophobic cavity‐ligand binding kinetics. Proc. Natl. Acad. Sci. U.S.A. 110: 1197–1202. Ashbaugh, H.S. and Pratt, L.R. (2006). Colloquium: scaled particle theory and the length scales of hydrophobicity. Rev. Mod. Phys. 78: 159–178.

­  References

105 Huang, D.M., Geissler, P.L., and Chandler, D. (2001). Scaling of hydrophobic

solvation free energies. J. Phys. Chem. B 105: 6704–6709.

106 Van Etten, R.L., Sebastian, J.F., Clowes, G.A., and Bender, M.L. (1967).

107

108

109

110

111 112

113

114 115 116 117

118

119 120

121

122

Acceleration of phenyl ester cleavage by cycloamyloses. A model for enzymic specificity. J. Am. Chem. Soc. 89: 3242–3253. Biedermann, F., Uzunova, V.D., Scherman, O.A. et al. (2012). Release of high‐energy water as an essential driving force for the high‐affinity binding of cucurbit[n]urils. J. Am. Chem. Soc. 134: 15318–15323. Biedermann, F., Nau, W.M., and Schneider, H.J. (2014). The hydrophobic effect revisited – studies with supramolecular complexes imply high‐energy water as a noncovalent driving force. Angew. Chem. Int. Ed. 53: 11158–11171. Biedermann, F., Vendruscolo, M., Scherman, O.A. et al. (2013). Cucurbit[8]uril and blue‐box: high‐energy water release overwhelms electrostatic interactions. J. Am. Chem. Soc. 135: 14879–14888. Guimaraes, C.R. and Mathiowetz, A.M. (2010). Addressing limitations with the MM‐GB/SA scoring procedure using the WaterMap method and free energy perturbation calculations. J. Chem. Inf. Model. 50: 547–559. Lazaridis, T. (1998). Inhomogeneous fluid approach to solvation thermodynamics. 1. Theory. J. Phys. Chem. B 102: 3531–3541. Fox, J.M., Kang, K., Sherman, W. et al. (2015). Interactions between Hofmeister anions and the binding pocket of a protein. J. Am. Chem. Soc. 137: 3859–3866. Ball, P. and Hallsworth, J.E. (2015). Water structure and chaotropicity: their uses, abuses and biological implications. Phys. Chem. Chem. Phys. 17: 8297–8305. Jungwirth, P. and Cremer, P.S. (2014). Beyond Hofmeister. Nat. Chem. 6: 261–263. Tobias, D.J. and Hemminger, J.C. (2008). Chemistry. Getting specific about specific ion effects. Science 319: 1197–1198. Lo Nostro, P. and Ninham, B.W. (2012). Hofmeister phenomena: an update on ion specificity in biology. Chem. Rev. 112: 2286–2322. Scannell, J.W., Blanckley, A., Boldon, H., and Warrington, B. (2012). Diagnosing the decline in pharmaceutical R&D efficiency. Nat. Rev. Drug Discovery 11: 191–200. Childers, M., Eckel, G., Himmel, A., and Caldwell, J. (2007). A new model of cystic fibrosis pathology: lack of transport of glutathione and its thiocyanate conjugates. Med. Hypotheses 68: 101–112. Hofmeister, F. (1888). Zur Lehre von der Wirkung der Salze. Arch. Exp. Pathol. Pharmakol. 24: 247–260. Kunz, W., Henle, J., and Ninham, B.W. (2004). “Zur Lehre von der Wirkung Der Salze” (about the science of the effect of salts): Franz Hofmeister’s historical papers. Curr. Opin. Colloid Interface Sci. 9: 19–37. Jones, G. and Dole, M. (1929). The viscosity of aqueous solutions of strong electrolytes with special reference to barium chloride. J. Am. Chem. Soc. 51: 2950–2964. Cox, W.M. and Wolfenden, J.H. (1934). The viscosity of strong electrolytes measured by a differential method. Proc. R. Soc. London, Ser. A 145: 475–488.

31

32

1  Water Runs Deep

123 Gurney, R.W. (1953). Ionic Processes in Solution. New York: McGraw‐Hill. 124 Collins, K.D. and Washabaugh, M.W. (2009). The Hofmeister effect and the

behaviour of water at interfaces. Q. Rev. Biophys. 18: 323–422.

125 Hamaguchi, K. and Geiduschek, E.P. (1962). The effect of electrolytes on the

stability of the deoxyribonucleate helix. J. Am. Chem. Soc. 84: 1329–1338.

126 Jordan, J.H., Gibb, C.L.D., Wishard, A. et al. (2018). Ion–hydrocarbon and/or

127 128

129 130

131 132

133 134 135 136 137 138

139

140

141

142

ion–ion interactions: direct and reverse Hofmeister effects in a synthetic host. J. Am. Chem. Soc. 140: 4092–4099. van der Vegt, N.F., Haldrup, K., Roke, S. et al. (2016). Water‐mediated ion pairing: occurrence and relevance. Chem. Rev. 116: 7626–7641. Omta, A.W., Kropman, M.F., Woutersen, S., and Bakker, H.J. (2003). Negligible effect of ions on the hydrogen‐bond structure in liquid water. Science 301: 347–349. Tielrooij, K.J., Garcia‐Araez, N., Bonn, M., and Bakker, H.J. (2010). Cooperativity in ion hydration. Science 328: 1006–1009. Smith, J.D., Saykally, R.J., and Geissler, P.L. (2007). The effects of dissolved halide anions on hydrogen bonding in liquid water. J. Am. Chem. Soc. 129: 13847–13856. Gurau, M.C., Lim, S.M., Castellana, E.T. et al. (2004). On the mechanism of the Hofmeister effect. J. Am. Chem. Soc. 126: 10522–10523. Batchelor, J.D., Olteanu, A., Tripathy, A., and Pielak, G.J. (2004). Impact of protein denaturants and stabilizers on water structure. J. Am. Chem. Soc. 126: 1958–1961. Zhang, Y. and Cremer, P.S. (2006). Interactions between macromolecules and ions: the Hofmeister series. Curr. Opin. Chem. Biol. 10: 658–663. Rembert, K.B., Paterova, J., Heyda, J. et al. (2012). Molecular mechanisms of ion‐specific effects on proteins. J. Am. Chem. Soc. 134: 10039–10046. Petersen, P.B. and Saykally, R.J. (2006). On the nature of ions at the liquid water surface. Annu. Rev. Phys. Chem. 57: 333–364. Carnegie, R.S., Gibb, C.L.D., and Gibb, B.C. (2014). Anion complexation and the Hofmeister effect. Angew. Chem. Int. Ed. 53: 11498–11500. Haynes, C.A. and Norde, W. (1994). Globular proteins at solid/liquid interfaces. Colloids Surf., B 2: 517–566. Klement, K., Wieligmann, K., Meinhardt, J. et al. (2007). Effect of different salt ions on the propensity of aggregation and on the structure of Alzheimer’s Aβ(1‐40) amyloid fibrils. J. Mol. Biol. 373: 1321–1333. Irvine, G.B., El‐Agnaf, O.M., Shankar, G.M., and Walsh, D.M. (2008). Protein aggregation in the brain: the molecular basis for Alzheimer’s and Parkinson’s diseases. Mol. Med. 14: 451–464. Broide, M.L., Tominc, T.M., and Saxowsky, M.D. (1996). Using phase transitions to investigate the effect of salts on protein interactions. Phys. Rev. E 53: 6325–6335. Wentzel, N. and Gunton, J.D. (2007). Liquid–liquid coexistence surface for lysozyme: role of salt type and salt concentration. J. Phys. Chem. B 111: 1478–1481. Ishimoto, C. and Tanaka, T. (1977). Critical behavior of a binary mixture of protein and salt water. Phys. Rev. Lett. 39: 474–477.

­  References

143 Reis‐Kautt, M.M. and Ducruix, A.F. (1989). Relative effectiveness of various

144 145 146

147 148 149 150

ions on the solubility and crystal growth of lysozyme. J. Biol. Chem. 264: 745–748. Zhang, Y. and Cremer, P.S. (2009). The inverse and direct Hofmeister series for lysozyme. Proc. Natl. Acad. Sci. U.S.A. 106: 15249–15253. Paterova, J., Rembert, K.B., Heyda, J. et al. (2013). Reversal of the Hofmeister series: specific ion effects on peptides. J. Phys. Chem. B 117: 8150–8158. Schwartz, R., Ting, C.S., and King, J. (2001). Whole proteome pI values correlate with subcellular localizations of proteins for organisms within the three domains of life. Genome Res. 11: 703–709. Bian, H., Wen, X., Li, J. et al. (2011). Ion clustering in aqueous solutions probed with vibrational energy transfer. Proc. Natl. Acad. Sci. U.S.A. 108: 4737–4742. Collins, K.D. (1997). Charge density‐dependent strength of hydration and biological structure. Biophys. J. 72: 65–76. Morris, D.F.C. (1968). Ionic radii and enthalpies of hydration of ions. In: Structure and Bonding, 63–82. Berlin: Springer. Yin, J., Henriksen, N.M., Slochower, D.R. et al. (2017). Overview of the SAMPL5 host–guest challenge: are we doing better? J. Comput.‐Aided Mol. Des. 31: 1–19.

33

35

2 Water‐Compatible Host Systems Frank Biedermann Institute of Nanotechnology (INT), Karlsruhe Institute of Technology (KIT), Hermann‐von‐Helmholtz‐Platz 1, 76344, Eggenstein‐Leopoldshafen, Germany

2.1 ­General Overview Devising supramolecular systems that are operational in water and reach the performance of natural systems is among the most intellectually challenging tasks in chemistry to date [1]. In addition to this curiosity‐driven research, such supramolecular systems promise to find applications in the medicinal and materials areas. Water‐compatible host systems are the basic supramolecular building blocks that are needed to realize the desired applications. From a structural point of view, the systems can be classified into acyclic and macrocyclic hosts, which each share certain fundamental characteristics. Nevertheless, from a practitioner’s point of view, this didactical pleasant structural distinction is not particularly helpful. For application‐driven interests, it may be preferable to classify water‐ compatible hosts according to their preferred binding partners, e.g. metal or organic cations, inorganic or organic anions, or neutral polar or apolar organic species. We are utilizing a mixed approach and will introduce in this chapter some of the most utilized water‐compatible hosts according to their structural features and further group them with respect to their typical guest binding characteristics. In some cases, this distinction is somewhat artificial because many hosts do not only bind one class of guests. Nevertheless, we believe that this rough division provides a larger overview over important water‐compatible hosts that are reoccurring in many of the subsequent chapters in this book where their binding to peptides and proteins (Chapter 3), nucleotides (Chapter 4), carbohydrates (Chapter 5), and ions (Chapter 6) will be discussed in more detail. Moreover, receptors based on coordination compounds are not considered here but will be presented in Chapter 7. At the end of this chapter, we will give a short comparison between the different host classes and provide practical information that we hope the practitioners will find useful for choosing a water‐compatible host when having a specific application idea in mind. The guideline and the benchmark for water‐compatible host systems is, as so often in supramolecular chemistry, Nature. Proteins, for example, achieve Supramolecular Chemistry in Water, First Edition. Edited by Stefan Kubik. © 2019 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2019 by Wiley-VCH Verlag GmbH & Co. KGaA.

36

2  Water‐Compatible Host Systems N23

D128

Tyr 43

S27 L25 W108

O H

O S45 V47

T90 L110

Asn 23

Y43

W79

Asp 128 H

N

O

W120

H

H

O

H N

Ser 27

O

N H

O

O H

Ser 45

S S88

N49 Ser 88

A86

(a)

(b)

O

H

Asn 49

O O

H N H

Figure 2.1  X‐ray crystallography structure of the streptavidin–biotin binding site (PDB 1mk5), where the ligand binding cavity is visualized by the gray mesh and the side chains of residues within 4 Å from the bound biotin are shown as sticks (a). Panel (b) shows the binding pattern in a schematic fashion. Source: Adapted with permission from Ref. [4]. Copyright 2007 Elsevier.

impressive binding affinities in water by burying the substrate in a structurally well‐defined cleft or cavity inside the protein structure  [2]. One out of many impressive protein‐type hosts is streptavidin that displays one of the highest affinities known for a small molecule (biotin) in aqueous media (Ka = 3 × 1013 M−1) [3]. On account of its reliably strong and selective complex formation, the (strept) avidin–biotin pair has found a wide range of supramolecular, biochemical, and biotechnological applications. Figure  2.1 illustrates the binding pattern inside the active site. Noteworthy, the release of energetically frustrated water molecules from the streptavidin binding pocket was suggested to provide a major driving force for binding that complements the energy gained through hydrogen bonding between the host and the guest [5]. Achieving this finely tuned affinity and selectivity with supramolecular hosts is not trivial but also not impossible. The example in Figure 2.1 indicates that the key to success is creating a cavity where the substrates find appropriate binding partners combined with favorable contributions to complex stability deriving from cavity desolvation.

2.2 ­Acyclic Systems 2.2.1  Acyclic Molecular Recognition Units Natural receptors typically fully encapsulate their targets in suitable pockets, whereas acyclic artificial hosts can be conformationally rather flexible and do not fully engulf the guest molecule. Supramolecular chemists sometimes give preference to such open structures because they are synthetically easier to access than their (macro)cyclic counterparts. However, this advantage is often counterbalanced by a lower degree of preorganization, a smaller host–guest contact area,

2.2  Acyclic Systems

and a missing contribution to binding from the release of “high‐energy” cavity water. Thus, acyclic hosts typically show lower binding strengths in aqueous media than macrocyclic systems. Nevertheless, sizeable and practically useful affinities can be reached when multiple non‐covalent binding motifs are simultaneously exploited, for instance, the combination of Coulomb interactions and hydrogen bonding. A representative example is the guanidiniocarbonyl pyrrole (GCP)‐derived zwitterionic binding motif (Figure  2.2a) that was introduced by Schmuck, which shows a dimerization constant of 170 M−1 in water and can be utilized to induce hydrogel formation [6]. The analogous GCP‐based tris‐­cationic receptor (Figure 2.2b) binds to the carboxylate groups of N‐acetylated amino acids in aqueous media (Ka ≈ 103 M−1) even in the presence of salts [7]. One‐ armed flexible peptide receptors that were derived from a cationic GCP binding motif efficiently bind anionic tetrapeptides in water (Ka up to 105 M−1) through electrostatic and hydrogen bonding interactions (see Chapter 3). One of the biggest assets of these acyclic receptors in comparison with macrocyclic systems is the possibility to optimize their performance by using combinatorial chemistry [8]. The affinities of acyclic receptors for their targets can often be improved by utilizing multivalent (polyvalent) binding motifs, especially for larger guests that contain several target sites. Multivalent systems benefit from the additive binding enthalpy of each binding site and from the reduced translational/rotational entropy cost of complex formation as compared with monovalent binding. A conformational entropy loss of the multivalent system can, however, occur when O O 2 O

N H H

O H N H N H

H NH H N H O

N H

(a)

NH2 MeO2C

NH3 R AcHN

O H H N N

O

N H H

Ka = 170 M–1

O

O H N H

N H HN

O H N N H H HN N H O H

1.80 2.19

1.79

1.64

Ka = 103 M–1

1.87 2.11

O

(b) Figure 2.2  Acyclic GCP‐derived receptors can bind their targets with a sufficiently high binding strength in aqueous media when multiple non‐covalent motifs such as Coulomb interactions and hydrogen bonding are exploited. A zwitterionic self‐complementary binding motif that dimerizes in aqueous media (a) [6] and a tris‐cationic receptor that binds N‐acetyl amino acids even in the presence of salts (b) are shown [7]. The geometry‐optimized structure of the N‐acetyl alanine complex shows several bifurcated hydrogen bonds (distances in Å). Source: Adapted with permission from Ref. [7]. Copyright 2005 American Chemical Society.

37

38

2  Water‐Compatible Host Systems

flexible linker molecules are involved, reducing the propensity for simultaneous binding (rigid linkers are often synthetically less accessible and may lead to strained binding geometries, which also lower the overall affinity). One instructive example is a trimeric vancomycin derivative that binds a trimeric d‐Ala‐d‐Ala ligand [9]. The multivalency effect leads to a remarkably strong complex formation (Ka = 2 × 1016 M−1 in phosphate buffered saline), which can be linked to the simultaneous formation of 15 hydrogen bonds. For the monovalent system, an affinity of Ka = 6 × 105 M−1 was observed. 2.2.2  Molecular Tweezers Molecular tweezers (Figure 2.3 shows a representative example) were introduced by Schrader and Klärner as water‐soluble hosts that selectively bind positively charged amino acids, i.e. lysine (Lys), arginine (Arg), and their derivatives [11, 12]. The remarkable high selectivity of these receptors originates from a combination of size‐selective binding (threading mechanism), electrostatics (the tweezer interior is electron rich), the hydrophobic effect, and dispersive interactions. O O O P OH

H3N

X

NHR L -lysine (subunit)

O

NH2

O P OH O O 2 Na

H2N

O

N H

X

NHR L-arginine (subunit) R = protecting group or peptide X = OR or NH-peptide

(a)

Tyr213 Thr217

Lys214

Leu218

(b)

(c)

Figure 2.3  Chemical structure of a molecular tweezer that selectively binds lysine (Lys, Ka = 5 × 104 M−1), arginine (Arg, Ka = 2 × 104 M−1), and peptides/proteins with sterically accessible Lys and Arg residues in physiological buffer (a). Calculated structure of a tweezer– lysine complex (b), and crystal structure of a tweezer complex with a single accessible lysine residue of protein 14‐3‐3 (c). Source: (b,c) Reproduced with permission from Ref. [10]. Copyright 2013 Macmillan Publishers Limited.

2.2  Acyclic Systems

Initially, tweezers with methylphosphonate groups were developed, which were found to complex, e.g. Nα‐acetyl lysine methyl ester with a Ka = 2 × 104 M−1 in D2O, and maintained an affinity of Ka = 4 × 103 M−1 even in 25 mM NaH2PO4 buffer [13]. Similar binding affinities were observed for peptides containing sterically accessible Lys and Arg residues. Positively charged histidine and biogenic amines such as dopamine, noradrenaline, and adrenaline were only weakly bound (Ka  1016 M−1 in water‐saturated chloroform) that it proved to be extremely challenging to remove the cations after the synthesis of the hosts. The lack of water solubility has, however, so far prevented the use of these compounds in the aqueous environment. Orthoester cryptands were introduced in 2015 by von Delius as a new macrobicyclic family, demonstrating that the design of ethylene oxide‐based cation binding hosts has not lost its charm despite 50 years of research [39]. Crown ethers bind a wide range of metal cations and organic guests bearing  ammonium groups. They are most widely used as phase‐transfer catalysts because they can pick up a cation from the aqueous phase and transfer it, together with its now desolvated and thus reactive “naked” anion to the organic phase [40]. The classic example is the 18‐crown‐6‐mediated dissolution of KMnO4 in benzene. The denticity of the crown ethers influences the affinity trends. For example, 18‐crown‐6 has a high affinity for K+, 15‐crown‐5 for Na+, and 12‐crown‐4 for Li+. Cation binding is in general stronger in less competitive organic solvents than in water, e.g. the K+ complex of 18-crown-6 has a Ka of 1 × 106 M−1 in methanol vs. a Ka of 120 M−1 in H2O [32].

43

44

2  Water‐Compatible Host Systems

Azacrown ethers are “softer” than the parent crown ethers and thus prefer “softer” metal cations as binding partners, e.g. Hg2+. This can be a problem if azacrown receptors are desired for the recognition of “hard” alkali metal cations such as Na+ and K+. In these circumstances, alternative cation binding motifs with higher inherent affinities are required. One option is to use spherands, but the large synthetic effort and the lack of water solubility hamper their use. Cryptands are a much more promising choice since they display higher binding affinities than crown ethers for metal cation complexation, e.g. cryptand‐2.2.2 binds K+ with a Ka of 2 × 105 M−1 in water [32]. Nevertheless, K+ binding is strongly attenuated in water by a factor of ≥104 in comparison with methanol [41]. Cryptands are also more cation specific due to their more rigidified structure and shielded cavity (hole‐size fitting concept [29]). Targeting ammonium functionalities, as found in amino acids and neurotransmitters, by crown ether derivatives was widely attempted because K+ and NH4+ have almost identical radii [41]. Unfortunately, the affinities of crown and azacrown ethers to ammonium ions are very low in aqueous media. Moreover, “functionalized” ammonium ions do not sterically fit inside cryptands, albeit a comparably weak interaction was reported between amino acids and cryptand‐2.2.2 (Ka approx. 103 M−1 in methanol; for comparison, Ka for NH4+ that can be fully encapsulated is 2 × 108 M−1) [42]. When considering the use of crown ethers and cryptands for biological applications, one has to keep in mind that crown ethers were found to be acutely and chronically toxic when administered orally to mice (LD50 are 3.2 g kg−1 for 12‐crown‐4, 1.0 g kg−1 for 15‐crown‐5, and 0.7 g kg−1 for 18‐crown‐6) [43]. Crown ethers are also lethal to Escherichia coli bacteria on account of the facilitated transmembrane cation transport [44]. Cryptands such as cryptand‐2.2.2. (aka Kryptofix‐2.2.2) are even more toxic than crown ethers (LD50 = 30–40 mg kg−1 in rodent) [45]. Their use in chelation therapy or positron‐emission tomography (PET) imaging applications thus requires a careful dosing [46]. 2.3.2 Bambus[n]urils Bambus[n]urils (n = 4, 6) (Figure 2.6a) were recently discovered by the group of Sindelar and were found to be one of the most potent receptors for inorganic anions, both in organic solvents and water [49]. The hexameric bambus[6]uril macrocycles with benzyl substituents were the most effective, while the tetrameric form was inactive. The binding preference of the water‐soluble bambus[6] uril followed the order I− > Br− ≫ Cl− > F− with affinities in phosphate‐buffered aqueous solution of Ka  =  1 × 107, 1 × 105, 9 × 102, and 1 × 102 M−1, respectively. Moreover, large anions such as ClO4− (6 × 107 M−1), BF4− (4 × 106 M−1), PF6− (2 × 106 M−1), NO3− (5 × 105 M−1), ReO4− (3 × 104 M−1), and IO4− (7 × 103 M−1) were also strongly bound [47]. C–H⋯anion interactions and the hole‐size fitting concept [29] certainly play an important role in explaining the high affinities and anion binding selectivity. In addition, the unusual large binding enthalpies observed (e.g. ΔH° up to −84 kJ mol−1 for iodide binding in water [47] and only −63 kJ mol−1 in chloroform [50]) may originate from the involvement of high‐ energy water release from the host cavity [51].

2.3  Macrocyclic Receptors that Bind Charged Guests O RN

R N

R N

O N O

N NO N N R N O R NN RNO O N N R N R N O N N N N N R N R O O N O R

O N

R

R=

CO2H

(a) S O

R NO N O N N N

R S

N S

R N

R OR O N S N N N

O R

(b)

S

N

R=

S

CO2H

Figure 2.6  Chemical structures and crystal structures of (a) water‐soluble anion binding macrocycles bambus[6]uril and its 1 : 1 complex with chloride and (b) biotin[6]uril and its 1 : 1 complex with iodide. Source: (a) Reproduced with permission from Ref. [47]. Copyright 2015 John Wiley & Sons. (b) Reproduced with permission from Ref. [48]. Copyright 2014 Royal Society of Chemistry.

The structurally related biotin[6]urils (Figure 2.6b) show similar binding trends for inorganic anions but display generally lower affinities than bambus[n]urils [52, 53]. Biotin[6]uril esters were found to be effective anionophores, where the transport of less hydrophilic anions (NO3−) was favored over hard, strongly hydrated anions (SO42−) because of the soft C–H⋯anion binding motif [52]. If a sensitive signal transduction method besides NMR can be developed for bambus[n]urils and biotin[n]urils, they may see future applications for anion sensing at biologically relevant concentrations. 2.3.3 Calix[n]arenes Calix[n]arenes (n ≥ 4; commonly n = 4, 5, 6, 8) are macrocycles formed by the condensation reaction of a phenol and formaldehyde. The 3D structure of their cone conformation, characterized by a wider upper rim and a narrower lower rim (Figure 2.7a), resembles a basket, vase, or cup. This conformation inspired the name calixarene, which is a combination of the Greek “calix” (=vase) and “arene”, referring to the aromatic building blocks. Calix[n]arenes also adopt several other conformations because the rotational barrier around the methylene

45

46

2  Water‐Compatible Host Systems

R

R

RÒ RÒ

R

R

OR′

OR′

(a)

R

R

upper rim R R

OR′ OR′ OR′ Lower rim OR′

NH3

HO O

O O

O

O

Dopamine

O3S SO3

O3S O3S

(b) NaO3S NaO3S

HO

Ka = 1 × 103 M–1

SO3

SO3Na SO3Na N

HO

OH OH

N

M2V

OH

CX4

(c)

CX4·M2V

CX5·M2V

Ka = 9 × 104 M–1

Ka = 3 × 105 M–1

Figure 2.7  Parent structure of a calix[4]arene and representation in the “cup”‐shaped structure (a). Both the upper and lower rims can be functionalized by suitable aliphatic and aromatic groups R and R′. The installation of sulfonate groups near the lower rim leads to amphiphilic calix[n]arenes that display millimolar binding affinities for dopamine and other aromatic amines in water (b). Negatively charged p‐sulfonatocalix[n]arenes CXn (n = 4–8) are cation‐selective, water‐soluble hosts. They bind strongly to methyl viologen M2V, aka paraquat, in aqueous media (c). Source: (a,c) Reproduced with permission from Ref. [55]. Copyright 2007 American Chemical Society. (b) Reproduced with permission from Ref. [54]. Copyright 2015 Springer Nature.

2.3  Macrocyclic Receptors that Bind Charged Guests

bridge is small. For instance, calix[4]arene features four distinct up–down conformations: (i) cone (C4v symmetry or C2v in the less symmetric pinched‐cone conformation), (ii) partial cone (Cs), (iii) 1,2 alternate (C2h), and (iv) 1,3 alternate (D2d). Larger calix[n]arenes can adopt a significantly larger number of conformations. Importantly, it was found that calix[n]arenes can be fixed into the cone conformation by suitably large substituents at the lower rim that cannot pass through the ring. Alternatively, large substituents such as tert‐butyl or sulfonate placed at the upper rim in combination with intramolecular hydrogen bonding at the lower rim can also favor the cone conformer. Calix[n]arenes were first and probably unintendedly prepared by Adolf von Baeyer at the end of the nineteenth century by mixing aldehydes, including formaldehyde, with phenols in a strongly acidic solution. The obtained tars could, however, not be separated or structurally characterized at that time, such that calix[n]arenes remained unrecognized. It was mainly through the systematic work by Gutsche starting in the 1970s that the structures and behaviors of the calix[n]arene tetramer, hexamer, and octamer were identified [56]. This large body of work also resulted in reliable and scalable synthetic protocols for the synthesis and functionalization at the upper and lower rims. The host–guest chemistry of calix[n]arenes was widely studied in organic solvents, in the solid state, and on surfaces. Interesting observations such as the water transport mechanism within crystals of calix[n]arenes have been made [57], and many applications of functionalized calix[n]arenes as enzyme mimics or catalysts were reported [58, 59]. Besides, calix[n]arenes have been widely used as sensors for cations and small molecules such as gases [60, 61]. How­ ever,  the use of calix[n]arenes as water‐compatible hosts is comparably less well investigated although the interest is progressively increasing, for example, in the contexts of stimuli–response materials, drug delivery systems, multifunctional nanoplatforms, and supramolecular catalysis [62, 63]. Several water‐soluble calix[n]arenes have been synthesized by installing ionic or polar residues on either the upper or lower rim of the calix[n]arene scaffold. Among these, sulfonated calix[n]arenes are the most popular hosts for host–guest binding studies [64]. For instance, it was found that a calix[5] arene with alkyl sulfonate substituents at the lower rim (Figure  2.7b) bound to  monoamine neurotransmitters when present below the CMC. Above the CMC, the host–guest binding characteristics were more complicated because aggregated calix[n]arenes were involved. The binding affinities for dopamine, tyramine, and phenethylamine were found to be moderate (approx. 103 M−1) in  spite of the extended hydrophobic cavity formed by the five tert‐butyl‐­ substituted subunits [54, 65]. Interestingly, binding was always strongly enthalpically favored and entropically disfavored. Anionic p‐sulfonatocalix[n]arenes (SCXn) are commercially available, highly water soluble, and not very prone to aggregation, rendering them well suited as aqueous hosts. Besides, SCXn are biocompatible, which makes them potentially useful for diverse life sciences and pharmaceutical applications. Unlike many other simple calix[n]arenes, SCXn reliably bind to guest molecules inside their cavities. In particular, the host–guest chemistry of SCXn with permanently positively charged organic guests or metal cations has been investigated in water. For

47

48

2  Water‐Compatible Host Systems

instance, methyl viologen is strongly bound with Ka ≈ 105 M−1 by both SCX4 and SCX5 in phosphate buffer (Figure 2.7c) [55]. Basic organic guests are also bound, and complexation is accompanied by shifts of the pKa values by 1–2 units [66]. By contrast, the aqueous binding strength of non‐charged organic guests with calix[n]arenes is unimpressive and much lower than that of cucurbit[n]urils (Section 2.4.2), e.g. Ka = 310 M−1 for benzene with a p‐tert‐butyl‐calix[4]arene [67], Ka  =  80 M−1 for benzaldehyde with SCX4 [68], and Ka  =  4 × 103 M−1 for cyclopentanone with SCX6 [69]. The comparably weak affinity of SCXn for hydrophobic guests can be understood by the rather strong binding of  cavity water molecule(s) in the aromatic, widely open cavities of these receptors, imposing a significant cavity desolvation penalty prior to guest binding [67, 70]. Cationic guests are more tightly bound by calix[n]arenes through additional cation–π interactions, and SCXn can thus be used as cation‐selective binders in aqueous media. Indeed, the preferential binding of cationic guests such as histamine by SCX4 has been exploited for the monitoring of biomembrane uptake processes, both in vitro and even in living cells through an indicator displacement assay [71]. 2.3.4 Pillar[n]arenes In 2008, the group of Ogoshi reported a new class of macrocyclic hosts, which became known as pillar[n]arenes (n = 5–10) (Figure 2.8) [74]. On account of their facile synthesis, versatile functionality, interesting host–guest properties, and original supramolecular assembly characteristics, these hydroquinone‐based hosts are now widely used in supramolecular chemistry [72, 75]. Their naming OR

n

RO RO

RHO

OR OR

RO OR

OR

RO OR

(a)

RO

(b)

(c)

Figure 2.8  General structure of pillar[n]arenes and structure of pillar[5]arene (a). The crystal structure shows the conformation of decamethoxypillar[5]arene in the solid state (b). Crystal structure of a pillar[5]arene derivative that encapsulates four water molecules (c). An extended version of this host functions as a single‐molecule transmembrane channel, transporting water molecules across a vesicle lipid bilayer. Source: (b) Reproduced with permission from Ref. [72] Copyright 2012 Royal Society of Chemistry. (c) Reproduced with permission from Ref. [73]. Copyright 2012 American Chemical Society.

2.3  Macrocyclic Receptors that Bind Charged Guests

was inspired by the shape resemblance to the symmetrical pillars that constitute the Parthenon in Athens [76]. Occasionally, but less preferable the term “pillarene” is used. The repeating units of pillar[n]arenes are phenolic moieties similar to those in calix[n]arenes (Section 2.3.3) or resorcinol‐based hosts such as deep cavitands (Section 2.4.3). However, the repeating units in pillar[n]arenes are connected by methylene bridges in para positions, whereas calix[n]arenes and deep cavitands are linked by methylene bridges in meta positions. This connectivity difference leads to large structural differences: the cavity of pillar[n]arene is cylindrically shaped with symmetric openings at both ends (akin to cucurbit[n] urils) (Section  2.4.2), while calix[n]arenes and deep cavitands are conical in shape, possessing only one opening. Pillar[5]arenes have a cavity diameter of ca. 4.7 Å, whereas the cavity size of pillar[6]arenes reaches about 6.7 Å. Larger pillar[n]arenes (n = 8–10) display deformed cavities in the solid state, allowing them in some cases to bind two guests simultaneously in two adjacent cavities. As a consequence of the electron‐rich building blocks and the para electron density enhancement, pillar[n]arenes show binding preference for electron‐deficient guests, including positively charged pyridinium, viologen, and ammonium or imidazolium salts. In addition, several neutral electron‐deficient species are also strongly bound, potentially with contributions from C–H⋯O(N) and C–H⋯π interactions. While most of their properties and applications are in the solid state and in organic solvents, water‐soluble pillar[n]arenes and their host–guest properties are becoming the forefront of research. The ease of derivatization enabled the preparation of pillar[n]arenes with pending negatively or positively charged side chains [75]. Expectedly, these highly charged derivatives were found to be relatively strong binders for respective oppositely charged guests (see Chapter 6). Due to the absence of a large hydrophobic driving force in combination with sizeable electrostatic effects, pillar[n]arenes are also more selective guest binders than, for instance, cyclodextrins or cucurbit[n]urils, which opens application potential in complex (biological) environments [75]. In fact, water‐soluble carboxylated pillar[n]arenes (n = 6 or 7), which appear to be relatively nontoxic, may be of use for drug delivery and biodiagnostics because of their high solubility and their ability to complex relevant bioactive compounds and drugs [77, 78]. Besides, many aqueous self‐assembly applications of pillar[n]arenes have already been described, e.g. the solubilization of fullerenes [79] and the formation of micelles or vesicles, which may be used for drug/siRNA co‐delivery [80]. Water‐soluble pillar[n]arenes can also be obtained by attaching  nonionic oligo(ethylene oxide) chains, which form homogeneous aqueous solutions at room temperature but show a lower critical solution temperature (LCST), i.e. they aggregate upon increase of the temperature [81]. This property can be exploited for the design of temperature‐responsive supramolecular systems. Pillar[n]arene‐based photoswitching systems show promising prospects, for instance, by exploiting the preferential binding of E over Z azobenzene derivatives [82]. Other sophisticated applications include the use of carbohydrate‐­ functionalized pillar[n]arenes as multivalent glycoclusters to control bacterial cell agglutination via the glycocluster–­lectin binding mechanism [83]. Besides, tubular‐shaped pillar[n]arenes are promising candidates for artificial channel

49

50

2  Water‐Compatible Host Systems

builders, enabling the transport of molecules through membranes. In a proof‐ of‐concept study, this functionality has already been demonstrated for pillar[5] arene‐based transmembrane water channels (Figure 2.8c) [73].

2.4 ­Macrocyclic Receptors that (also) Bind Non‐charged Organic Guests Several hosts have been discovered, often by serendipity, that show high binding affinities for organic guests also in aqueous media. These high‐affinity binders generally feature a macrocyclic preorganized conformation with a well‐shielded cavity. In some instances, higher binding strength in water than in organic solvents, also in the case of hydrophilic guests, was observed. This can often be attributed to the energy gain on account of the release of energetically frustrated water molecules from the host cavity [51, 84]. In the following section, selected classes of macrocycles for organic guests and some of their properties are discussed. 2.4.1 Cyclodextrins Cyclodextrins (CDs) are cyclic oligosaccharides containing n 1,4‐linked α‐d‐­ glucopyranoside units (α‐CD, n = 6; β‐CD, n = 7; γ‐CD, n = 8) [85, 86]. Figure 2.9a shows the structure of β‐CD. CDs were discovered in the late nineteenth century by the pharmacist and chemist Antoine Villiers who studied the action of enzymes on various carbohydrates, including potato starch [87]. At the beginning of the twentieth century, the investigations into the bacterial production of CDs were intensified by Franz Schardinger, who became known as the “founding father” of CD chemistry because he was the first to describe their fundamental properties and to study their host–guest chemistry. However, it was only in the late 1940s when the cyclic structure was confirmed by the at that time emerging X‐ray structure analysis. Still to date, CDs are mainly produced by Bacillus macerans acting through the enzyme cyclodextrin glucosyltransferase on amylose, the linear component of starch, rendering them available on the multiton scale. The CD cavity represents a hydrophobic microenvironment. Thus, CDs preferentially capture suitably sized hydrophobic guest molecules in their cavity from aqueous media. In fact, much debate had occurred over the exact binding mechanism. Szejtli revised and unified the views of Cramer, Saenger, and Bergeron on the inclusion mechanism and concluded that cavity water molecules in CDs are in an unfavorable energy state due to the polar–apolar interactions and are thus readily displaced by a less polar guest, leading to an enthalpic gain of binding [85, 87]. This model was the prototype for rationalizing the binding mechanism for cucurbit[n]urils and other host–guest complexes in water [51, 84]. Generally, CDs display much lower binding constants than cucurbit[n]urils in aqueous media, approaching a Ka of ca. 105 M−1 for certain adamantane derivatives with β‐CD [88]. The affinity for organic guests is nevertheless sufficiently high such that many self‐assembled structures can be obtained through CD complexes as supramolecular anchors. Some representative examples are the formation of protein heterodimers [89, 90], non‐covalently connected micelles

2.4  Macrocyclic Receptors that (also) Bind Non‐charged Organic Guests

O HO

HO

HO

O HO

OH O

HO O HO

OH O OH

HO

(b)

O

O OH OH

O

(a)

O O O H H

OH H H O OO

OH H O O O

O

OH

0.57 nm

0.78 nm

0.95 nm

6 glucose units

7 glucose units

8 glucose units

α-Cyclodextrin

β-Cyclodextrin

γ-Cyclodextrin

1 : 1 CD-guest complex

0.78 nm

O

OH

2 : 1 CD-guest complex

Figure 2.9  Chemical structure of β‐cyclodextrin and schematic representations of the different homologs α‐CD (n = 6), β‐CD (n = 7), and γ‐CD (n = 8) (a) and schematic drawings of 1 : 1 and 2 : 1 cyclodextrin–guest inclusion complexes (b).

[91], hollow spheres [92], and voltage‐responsive vesicles [93]. More recently it  was discovered that boron clusters show markedly higher affinities for CDs (Ka > 105 M−1 for both β‐CD and γ‐CD) than contemporarily used adamantanes and ferrocenes [94]. These complexes have been successfully tested as supramolecular connectors for bioactive surfaces [95] and for indicator displacement applications in sensing [96]. Depending on the size ratio between CD and guest, 1 : 1 and 2 : 1 complexes are commonly formed (Figure 2.9b). For instance, encapsulation of camphor by two α‐CD was found to be highly cooperative, and water molecules may have a structural role, acting as “glue” between the two α‐CD molecules in forming the complex [97]. Unlike many other hosts described in this chapter, CDs are chiral, such that selective binding of guest enantiomers should be expected and is commonly observed. For instance, the aforementioned binding of camphor by two α‐CD shows a binding strength that is by a factor of two larger for (+)‐­camphor than for (−)‐camphor at ambient temperature [97]. For that reason, stationary

51

52

2  Water‐Compatible Host Systems

phases are often utilized in chromatographic methods that have been functionalized with CDs as chiral recognition units to separate enantiomers [98, 99]. Furthermore, chiral separations are also commonly carried out by capillary electrophoresis where CDs are added into the run buffer [100]. Another important property of CDs is their high water solubility due to their large number of hydroxy groups. The water solubilities of α‐, β‐, and γ‐CD at ambient conditions are approximately 13%, 2%, and 26% (w/w), respectively, which corresponds to approximately 19 g l−1 or ca. 17 mM for β‐CD. The comparably low aqueous solubility of β‐CD can be pinpointed to the formation of an intramolecular hydrogen bond network between the secondary hydroxy groups, which are therefore not in strong contact with the solvent molecules. Literally thousands of CD derivatives have been reported that have variable ring size and random or site‐specific chemical functionalization [101]. A few representative examples are shown in Figure  2.10. These derivatives are typically more water soluble than native CDs and have different host–guest binding characteristics. Many randomly functionalized CDs are used in the industrial and pharmacological praxis. The random substitution pattern is therefore an application advantage because it lowers the crystallization propensity and thus indirectly ensures that the respective CD materials remain in solution. However, these host mixtures are also extremely complex and diverse, which can be an issue for systematic host–guest binding studies. For instance, there are 221 − 1 distinct derivatives possible when considering random functionalization of the 21 hydroxy groups of β‐CD (not even counting potential stereoisomers!). In order to enable approval for drug use, commercial suppliers of randomly functionalized CDs therefore provide a range of the degree of functionalization, e.g. n units per glucose unit. Nevertheless, one needs to keep potential batch‐to‐ batch differences in mind when utilizing such materials. Conversely, there are several monofunctionalized CDs containing tosylate, amino groups, azide groups, etc. for further functionalization. These compounds are structurally defined and are therefore useful for the linking of CDs to polymers, dyes, or surfaces. CDs are biocompatible, which has enabled their use as pharmaceutical excipients and in agrochemicals, fragrances, and foods [101]. CDs also show resistance OR″ O O O O R R′

R″O

O R″O

O OR′ OR

O RO

O R′O

OR″ O RO O

OR′ O OR O R″O

R′O R R′ O O O

OR′ O R O O

OR″

–OR

–OR′

–OR″

Chemical name

–OCH3

–OH

–OCH3

Dimethyl-β-CD

–OCH3

–OCH3

–OCH3

Trimethyl-β-CD

–OAc

–OAc

–OAc

Triacetyl-β-CD

Random: –OCH3 or –OH OR″ Random: –OCH2CH2OH or –OH

O

Random: –OCH(OH)CH3 or –OH

Randomly methylated β-CD Hydroxyethyl-β-CD 2-Hydroxypropyl-β-CD

Random: –O(CH2)4SO3Na or –OH Sulfobutylether-β-CD

Figure 2.10  Representative substituted β‐CD derivatives that are widely used and mostly commercially available.

2.4  Macrocyclic Receptors that (also) Bind Non‐charged Organic Guests

to enzymatic degradation by higher organisms, and they are therefore excreted intact via the kidney when injected intravenously into humans. On the other hand, bacterial and fungal amylases can degrade CDs, allowing them to be used in agrochemical formulations or consumer products because they do not accumulate in the environment (water, soil). The toxicities of CDs are relatively low (LD50 > 0.7 g kg−1 in mice) [102] where the most problematic homolog is β‐CD (at high concentrations) because it can extract cholesterol and other lipid membrane components, causing hemolysis of erythrocytes. Nevertheless, CDs are the most biocompatible molecular hosts available. One can exploit the different CD homologs and derivatives for tailoring the binding strength and the binding mode. For instance, prostaglandin E2 is bound by α‐CD, β‐CD, and γ‐CD, but the binding modes largely differ (Figure 2.11a) [101]. Due to the small cavity size, α‐CD primarily encapsulates aliphatic chains, whereas β‐CD is large enough to include also aromatic rings. γ‐CD can complex O

COOH

COOR

O RO HO

OR

Alprostadil, α-CD (prostaglandin medication)

HO

N

H N

α-CD–prostaglandin E2

S O

N

MeO O

O

COOH HO

Cl OH

β-CD–prostaglandin E2

ONa

H N Cl

Diclofenac, 2HP-γ-CD (anti-inflammatory drug) O RO

O RO

COOH HO

(a)

OMe

Omeprazole, β-CD (proton pump inhibitor)

H H

OH

γ-CD–prostaglandin E2

OR

H

O

(b)

Hydrocortisone, 2HP-β-CD (cortisol medication)

Figure 2.11  Proposed models of inclusion complexes between prostaglandin E2 and α‐CD, β‐CD and γ‐CD (a), and representative approved and marketed cyclodextrin drug complexes (2HP = 2‐hydroxypropyl) (b). Source: (a,b) Reproduced with permission from Ref. [101]. Copyright 2004 Springer Nature.

53

54

2  Water‐Compatible Host Systems

aliphatic and aromatic substances; commonly two guests are simultaneously bound in order to fill the cavity of the γ‐CD host. The encapsulation of guest molecules by CDs, resulting in a marked solubility increase of hydrophobic guests (drugs, pesticides, fragrances, aromas, etc.) and in their protection from chemical degradation, is the basis for most industrial CD applications. In Figure 2.11b, selected drugs are shown whose formulations with CDs have been approved and marketed. For instance, the combination of α‐CD and prostaglandin PGE1, sold as alprostadil alphadex, provides a more metabolic stable medication than PGE1 alone. Thus, it allows for parenteral use (as opposed to the required intra‐­ arterial administration of PGE1 alone) for the treatment of peripheral vascular complications [101]. Furthermore, the rather strong affinity of β‐CD for cholesterol (Ka = 2 × 104 M−1) is exploited in biochemical studies to deplete cellular cholesterol levels [103, 104]. Unfortunately, other steroidal hormones and drugs without a long alkyl chain are less strongly bound. Extra‐cavity complexes are also common, and CDs can self‐assemble into nanosized aggregates, which assists their solubilizing power [105]. With the contrary purpose, cross‐linked CD‐based materials are being developed for the removal of steroidal hormones and other pharmaceutical residues from waste and drinking water (micropollutant capture), where again the host–guest chemistry of CDs with a wide range of hydrophobic organic guests is exploited [106, 107]. CDs are suitable components for hydrogel formation, where one also takes advantage of the multivalency effect. Several CD‐functionalized polymers were, for example, reported that gelate with guest‐functionalized polymers in water, resulting in a number of functional materials (see Chapter  8) [108]. The CD‐ based binding motif was also employed, mostly in a multivalent fashion, for the spatially controlled and reversible (redox)stimuli‐responsive linking of several target molecules such as proteins, dendrimers, and nanoparticles onto surfaces [109]. This approach was termed “molecular print board” [110, 111]. 2.4.2 Cucurbit[n]urils The glycoluril‐based cucurbit[n]urils (n = 5–8, 10, 14) are particularly well‐suited hosts in aqueous media due to their exceptionally high binding strengths [76, 112–116]. The parent cucurbit[6]uril (CB6) was synthesized by Behrend and coworkers as a sparingly soluble “condensation product” already in 1905, and until today synthetic protocols for CBn macrocycles closely resemble this first report of the acid‐catalyzed condensation of glycoluril and formaldehyde. It took almost 80 years before Mock and coworkers determined the “pumpkin‐shaped” structure of CB6, which motivated the name choice cucurbit[n]urils in analogy to the botanical family Cucurbitaceae. Due to the insolubility of CB6 in virtually all organic solvents and poor solubility in water, only a limited number of groups, e.g. those of Mock and Buschmann, engaged in CB6 host–guest binding studies at that time. With the beginning of the millennium, the groups of Kim, Isaacs, and Day achieved the synthesis and isolation of other homologs, including CB5, CB7, CB8, and CB10 (Figure 2.12a). Recently, the structure of the yet largest CBn member (CB14) has been reported, which was found to adopt a twisted conformation [118]. For aqueous applications, the comparably high solubility of CB7

2.4  Macrocyclic Receptors that (also) Bind Non‐charged Organic Guests

in water (approx. 10 mM) and high affinities for many organic guests are great assets, supporting the emergence of many CB7‐based supramolecular studies and applications. CBn macrocycles also exhibit other beneficial properties that are of practical relevance. First, CBn macrocycles are inert toward many chemicals, e.g. acids, bases, oxidation, and reduction reagents. They are also photochemically inert. Moreover, despite their chemical inertness, there are pathways with which CBn macrocycles can be functionalized, e.g. via stepwise buildup of functionalized CBn derivatives from tailor‐made monomers or via controlled oxidative hydro­ xylation of CBn macrocycles [119, 120]. The key intermediate resulting from the latter method can be further functionalized in several ways (Figure 2.12b). Acyclic CBn derivatives that were introduced by Isaacs and coworkers as highly water‐soluble hosts for aliphatic and aromatic guests, including drugs and carbon nanotubes, also allow structural changes (Figure 2.12c) [121]. CBn and acyclic CBn derivatives are nontoxic such that ex vivo and in vivo use is possible. Indeed, a biotin‐functionalized CB7 can be utilized for targeted oxaliplatin drug delivery to cancer cells [122]. In addition to solubilization enhancement, complexation by CB7 and acyclic CBn also improves the chemical stability OO

OO O N N N NN N

O

N N N N NN

N N NN N N O O O

h

NN N N N N O

OO

x

d

(a)

(b)

d (Å)

h (Å)

0

CB5

4.4

9.1

68

1

CB6

5.8

9.1

142

2

CB7

7.3

9.1

242

3

CB8

8.8

9.1

367

5

CB10

11.7

9.1

691

O N

N

O

N

N

O

O

N

N

N

NH3 SO3Na

O

N

N

N

N

HH O

S

O

N

OH

O

O H

SO3Na

R=

x

O

(c)

CBn

OO O O OO N N N N N NN N NN N N R N N NN N N N N N N NN O O O OO O SO3Na

V (Å3)

x

O

N

N

N

N

H O

O

O SO3Na

Figure 2.12  Chemical structures of important members of the CBn macrocycle family and size estimations for their maximum inner cavity diameter d, their height h, and their cavity volume V (values were taken from Refs. [116, 117]) (a), examples of functionalized CBn derivatives (b), and representative structure of an acyclic cucurbituril (c).

55

56

2  Water‐Compatible Host Systems

and availability of drugs [123]. It is hoped that these CBn derivatives complement CDs as delivery vehicles in real‐world applications in the near future. The binding affinities of the CBn family for organic guests in water are in the range of 103–109 M−1 [76, 112–117, 124]. Guests include not only simple aliphatic hydrocarbons but also biologically relevant molecules such as amino acids, peptides, neurotransmitters, hormones, drugs, and toxins. Moreover, also technologically and environmentally important nonfunctional molecules such as  perfluorinated hydrocarbons are strongly bound by CBn macrocycles and their derivatives. Indeed, almost all organic molecules that fit into the cavity of the different CBn hosts are appreciably strongly bound in water; only guests that require the immersion of a negatively charged moiety into the CBn cavity are an exception [125]. For instance, a wide range of steroids such as testosterone and β‐estradiol can be bound with high affinities by CB8 (Ka ≥ 106 M−1) [126], significantly stronger than by other available water‐compatible hosts, e.g. CDs. Positively charged analytes show generally higher affinities (factor of 10–100) than their non‐charged counterparts, which can be attributed to attractive ion– dipole interactions of cations with the carbonyl‐fringed CBn portals [116]. Even very hydrophilic amino sugars, e.g. glucosamine and galactosamine, are bound in their protonated form by CB7 with remarkably high affinities of up to 104 M−1 in water [127]. The binding of CB7 and of CB7–dye complexes to certain proteins (e.g. insulin) has also been reported [128]; such interactions were found to modulate the activity of enzymes (e.g. type II endonuclease) [129, 130]. Metal cations are also relatively strongly bound by the CBn portals, with binding affinities in the 10–100 M−1 range. This general property of CBn is the cause of the pronounced medium dependence of CBn‐analyte binding constants, which typically decrease markedly in saline buffers. The high binding affinities of the CBn macrocycles can be attributed to the strong enthalpic gain caused by the release of “high‐energy water” from the ­confined and hydrophobic CBn cavities upon guest binding. After release, the expelled cavity water molecules can “enjoy” a stronger hydrogen bonding network in the aqueous bulk [51, 84, 131]. CB7 is the current record holder for the highest affinities for organic molecules in water (Figure 2.13a), reaching a spectacular value of 7 × 1017 M−1 (−ΔG° = 100 kJ mol−1) for a diamantane derivative [132]. Different ferrocene and adamantane derivatives also show very high binding affinities. This has been utilized for the design of a supramolecular Velcro that can strongly but reversibly bind two surfaces together under water with a higher strength than commercial 3M double‐sided tape (Figure  2.13b) [133]. CB7–ferrocene host–guest chemistry was also exploited for the non‐­covalent tethering of biological recognition units (e.g. peptides, proteins) to surfaces. This provided control over the adhesion of living cells [134]. A supramolecular fishing approach for plasma membrane proteins was also introduced based on the CB7–ferrocene couple [135]. A different application, which was put forward by the Nau group, is the use of CBn complexes in indicator displacement assays, allowing the monitoring of biochemical processes, e.g. enzymatic reactions and membrane permeation profiles of organic species [136]. Various supramolecular hydrogels were also prepared through cucurbit[n]uril‐ based host–guest complexes. For instance, the functionalization of hyaluronic

2.4  Macrocyclic Receptors that (also) Bind Non‐charged Organic Guests

N NH3

Ka = 2 × 1014 M–1

N

ΔH° = –81 kJ mol–1 –TΔS° = 0 kJ mol–1

Ka = 7 × 1017 M–1

OH

N

Fe

N Fe

Fe N

Ka = 3 × 109 M–1 (a)

ΔH° = –90 kJ

mol–1

–TΔS° = 36 kJ

mol–1

Fc-[Si]

Ka = 4 × 1012 M–1

Ka = 3 × 1015 M–1

ΔH° = –90 kJ

mol–1

ΔH° = –90 kJ mol–1

–TΔS° = 18 kJ

mol–1

–TΔS° = 2 kJ mol–1

Underwater adhesion based on host–guest supramolecular complex External stimuli

CB7-[Si]

Velcro

O H

(b)

CB7

N

N CH2 H N CH2

N O

Fe 7

Fc

N H

Fc@CB7

Figure 2.13  Cucurbit[7]uril forms extremely strong host–guest complexes with charged and non‐charged organic guests in water on account of a strong enthalpic driving force for binding (a). Principle of the under water adhesion of two surfaces through a supramolecular Velcro. First, the covalent functionalization of one surface with the host CB7 and of the other surface with the guest ferrocene was carried out (not shown). When both surfaces were brought into contact under water, strong complexation‐mediated adhesion occurred. The image shows that even a small area of the supramolecular Velcro sufficed to carry the weight of a 2 kg weight under water (b). Source: (a) Reproduced with permission from Ref. [51]. Copyright 2017 American Chemical Society. (b) Adapted with permission from Ref. [133]. Copyright 2013 John Wiley & Sons.

acid chains with CB6 as the host and diaminohexane as a complementary, strongly binding guest (Ka > 106 M−1 in water) was carried out. Mixing of a solution of the so‐decorated polymers led to efficient hydrogel formation, even in the presence of living cells [137]. The mechanical strength of the hydrogel could be readily adjusted by the addition of competitive binders. The gel was also shown to be biocompatible after injection under the skin of mice. The injectability and the stimuli responsiveness of such hydrogels, which can be utilized to control drug delivery and excretion of the gel components after use from the body, are big assets in comparison with covalently linked gels [138]. In addition, 2D

57

58

2  Water‐Compatible Host Systems

polymers [139, 140] and hollow nanocapsules [141, 142] were prepared by CBn‐ based materials. The assembly of CBn macrocycles on the surface of metal nanoparticles was shown to enable surface enhanced Raman scattering (SERS) sensing applications and catalysis in aqueous media [143, 144]. The cavity of CB8 is large enough to host two aromatic guests efficiently (Figure  2.14a) [113]. Typical guests are pairs of tryptophane moieties or the combination of methyl viologen and a non‐charged aromatic compound such as naphthalene or indole/tryptophan. CB8 can thus act as a “molecular glue,” mediating the selective binding of two complementary guests. Discrete CB8‐based assemblies, for instance, molecular loop locks [146], self‐assembled dendrimers [147], peptide–peptide [148], protein–polymer [149], protein–peptide [150], and protein–protein [151] conjugates have been described. These ternary complexes are also of use for material assemblies in water, e.g. of supramolecular polymers [152] and networks/hydrogels [153], and for the stimuli‐responsive assembly of nanoparticles [154]. Furthermore, they were exploited for the conjugation of biomolecules or cells to surfaces [155]. CB8‐based materials design does not only circumvent the cumbersome functionalization of the hosts in favor of more facile functionalization of both the first and second guest, but it also enables switching of complex formation and thus control over the spatiotemporal assembly state. Redox stimuli (e.g. by reduction of the viologen‐type guests) and light triggers (e.g. by E/Z isomerization of azobenzene guests) are particularly convenient to apply [156]. For instance, a light‐switchable supramolecular polymer has been prepared by CB8‐mediated complex formation [157]. Moreover, CB8‐accelerated photodimerization, e.g. of anthracene‐type guests, can (reversibly) “freeze” the structural arrangement of the self‐assembled material [158], which was utilized for a light stimulus‐induced cargo release from CBn‐polymer‐based microdroplets [159]. Notably, CB8 complex formation and the potentially associated assembly process can be conveniently monitored by the appearance of charge‐transfer bands in the visible region of the absorbance spectrum (Figure 2.14b) [145], the emergence of induced circular dichroism (i.c.d.) bands (for chiral guests such as amino acids and peptides) [160], or by the quenching of the emission of the first or the second guest [161]. The aqueous assembly of CB8 and suitable dicationic dyes is thus also a promising strategy for the design of associative binding chemosensors (Figure  2.14c) [112]. Prototype systems for the detection and identification of aromatic compounds at micromolar concentrations and the monitoring of enzymatic reactions were reported [112, 160, 161]. 2.4.3  Deep Cavitands Deep‐cavity cavitands, in short deep cavitands, are resorcin[4]arene‐based hosts to which additional (aromatic) rings have been added either as bridges between the aromatic subunits or as substituents in the 2‐position of the resorcinol rings [76, 162]. This design deepens the binding pocket of the resorcin[4]arene, explaining the name “deep cavitands” of these hosts. Deep cavitands are bowl shaped and their cavity is closed at one end. This enclosed cavity thus enforces much more constraints on the orientation and conformation of guest molecules

2.4  Macrocyclic Receptors that (also) Bind Non‐charged Organic Guests NHR R R R N

or R′ R = OH, NH2, halogen O-alkyl, O-polymer

N R

R = alkyl, aryl

CB8

(a)

O

+

N H R = OH, peptide R′ = H, peptide

+

OH

OH

CN

Br

OH

HO

OH

OH

OMe

OH

OH

OH

OH

OH

O NH2

O

MeO OH

(b)

Analyte

CB8

Dye

Dye

Dye

Analyte

″Communication″

(c)

Figure 2.14  The large macrocycle cucurbit[8]uril can simultaneously bind one equivalent of each a dicationic aromatic guest and a non‐charged second guest, resulting in the formation of a 1 : 1 : 1 ternary complex where both aromatic guests are hold in a face‐to‐face π–π‐ stacking arrangement (a). The combination of CB8, methyl viologen, and aromatic second guests (each at 0.5 mM in water) leads to “charge‐transfer colors” on account of ternary complex formation [145] (b). CB8‐mediated ternary complexes can be exploited to construct associative binding assays for sensing purposes, where analyte‐specific spectroscopic responses such as charge‐transfer absorption bands, analyte‐specific emission quenching, or induced circular dichroism (i.c.d.) can be utilized for analyte differentiation (c). Source: (c) Reproduced with permission from Ref. [112]. Copyright 2018 John Wiley & Sons.

than the cavities of cucurbit[n]urils, pillar[n]arenes, polyaromatic molecular tubes, or CDs because extended guests cannot stick out at both portals. Deep cavitands can thus be expected to show pronounced size‐selective guests binding. Furthermore, the “lid” renders the cavity more hydrophobic because contact of cavity water molecules with bulk water is reduced [163]. Cavitands can be made water soluble by installation of multiple (mostly charged) groups in the vicinity of the cavity while taking care that the cavity is not filled or otherwise affected by the side chains. For instance, the positively charged deep cavitand shown in Figure 2.15a displays a good solubility in aqueous media and strongly binds anionic guests such as adenosine triphosphate (ATP) [164]. Extending the cavity walls of resorcinarenes using bridges between the resorcinol subunits yields velcrands (cavitands with ortho‐substituted aromatic units between the resorcinol moieties) and benzal‐bridged cavitands [76, 162, 166]. A  representative water‐soluble velcrand with benzimidazole bridges is shown in  Figure  2.15b. Velcrands usually exist as a mixture of the “kite” form and the “vase” form, the latter of which has a deep concave cavity that allows the

59

60

2  Water‐Compatible Host Systems

H2N NH2

H2N NH2 H2N

NH2

NaO2C CO2Na

H2N NH2

N HN

O

OO

O O

OO

N

R

O

R

R

N

HN

O O

R

NaO2C NH CO2Na N

O

O Et

R = (CH2)3(OCH2CH2)4OCH3

(a)

O O

Et

Et

NH

O

Et

(b)

2×

Velcrand

(c)

O O

2×

O

Kite

O

O O

O

O

O

O

Capsule

O

O

O

O

O O

O

O

OO

O

O

Vase

O

O

Hydrophobic cavity

(d)

O

O

O

O

O

O O

O

Figure 2.15  A water‐soluble deep cavitand that strongly binds to ATP in protic and aqueous media (a). Deep cavitands can cause helical coiling of long alkane chains, here shown for a benzimidazolone-bridged deep cavitand and SDS as guest (b). Deep cavitands with aromatic walls can interconvert between a flat “kite” shape and a “vase” shape (c). Both forms can dimerize to a velcrand dimer (left) or to a capsule (right), depending on the solvents and guests present. The “octa‐acid” cavitand possesses a well‐shielded hydrophobic binding pocket in which size‐matching organic guests can be bound (d). A capsule with a 2 : 1 host–guest stoichiometry forms when larger guests are bound, e.g. long alkanes or steroids. This full encapsulation inside the capsule can lead to a substantial distortion of the guests. Source: (a) Reproduced with permission from Ref. [164]. Copyright 2014 John Wiley & Sons. (b) Reproduced with permission from Ref. [162]. Copyright 2007 Royal Society of Chemistry. (c) Reproduced with permission from Ref. [165]. Copyright 2016 Nature America. (d) Reproduced with permission from Ref. [166]. Copyright 2015 Royal Society of Chemistry.

encapsulation of guests (Figure 2.15c). Expectedly, the conformational equilibrium between both forms is strongly depended on the structures of the host and guest(s) and the polarity of the solvent (in water, most velcrands preferentially adopt the kite form, which dimerizes to minimize the solvent exposed surface). The benzimidazole‐linked cavitand in Figure 2.15b was shown to host surfactant‐­ like molecules like SDS by forming 1  :  1 complexes. In these complexes, the alkane chains of the guests coil into helical conformations in order to fit well into the deep cavity and to minimize the contact with bulk water. Thus, deep cavitands can be utilized to enforce energetically disfavored conformations of

2.4  Macrocyclic Receptors that (also) Bind Non‐charged Organic Guests

n‐alkanes. The polar functional groups of amines, esters, carboxylates, alcohols, etc. tend to situate themselves at the portal region of the deep cavitands such that they can be in contact with bulk solvent and possibly interact with charged/ polar residues that are arranged along the cavity rim. This feature can be utilized to compress α,ω‐difunctionalized alkanes into U‐ or J‐shaped conformations [76]. In these conformations, the two terminal functional groups are held in close spatial proximity, providing the prospects for interesting chemistry. Indeed, it was shown that a velcrand with a benzimidazolone bridge can efficiently enhance macrocyclization processes, yielding lactams that are otherwise only accessible through high‐dilution synthesis [167]. The water‐soluble “octa‐acid” cavitand shown in Figure 2.15d is sparingly soluble in neutral aqueous solutions but highly soluble at pH >7 when its carboxylate side chains are deprotonated. This host possesses an approximately 0.8 nm wide and 0.8 nm deep hydrophobic pocket that is closed at the bottom. In comparison with velcrands, it is also more preorganized and thus a fruitful workhorse for the identification of general concepts of host–guest complexation in water and systematic investigations of solvent effects. Octa‐acid shows strong binding of hydrophobic species (e.g. adamantanes and hydrocarbons) in water [76, 166], which may be driven by the release of poorly hydrogen‐bonded cavity water molecules and strengthened by direct host–guest interactions (e.g. dispersion interactions). Octa‐acid binds hydrophobic guests generally stronger than the similarly sized β‐CD. For example, it possesses an affinity of Ka = 1 × 106 M−1 for adamantanecarboxylate, whereas β‐CD displays an order of magnitude lower affinity for this guest. Indeed, the binding pocket of octa‐acid is better shielded and more hydrophobic than that of CDs, which can rationalize this observation. With a Ka of ca. 108 M−1, the steroid (+)‐dehydroisoandrosterone is one of the strongest binding guests for octa‐acid in aqueous media to date. However, in comparison with cucurbit[n]urils, even octa‐acid falls short in terms of binding strength in water. Guest binding of octa‐acid was found to be correlated to the Hofmeister effect caused by chaotropic/kosmotropic ions [168]. This finding was explained by the pronounced affinity of weakly coordinating anions such as iodide and perchlorate for this host, which leads to a reduction of the apparent binding strength of hydrophobic guests due to the competitive binding mechanism [169]. In addition to 1 : 1 host–guest complex formation with monomeric cavitands, 2 : 1 and 2 : 2 host–guest complexes of deep cavitands in general and with octa‐ acid in particular are also common. Capsule formation causes the rigidification of the cavitands and thus yields a well‐defined and robust cavity geometry, which increases the scope of these hosts for size and shape‐selective binding of guests. Small alkanes form stable 2 : 2 host–guest complexes with octa‐acid capsules in water [166]. Helical motifs were observed for longer alkanes (C11–C14) that are bound in a 2 : 1 host–guest stoichiometry; C18–C23 alkanes adopt a U‐shaped conformation (Figure 2.15d). For alkane guests longer than C26, the two halves of the capsules are pushed away from each other, disrupting the capsule. Naturally, the size‐selective binding of octa‐acid‐based capsules has motivated their use for physical separations and kinetic resolutions. For instance, when a mixture of hydrocarbon gases was added to the headspace above the aqueous solution of the

61

62

2  Water‐Compatible Host Systems

host, the strongest binding alkane was preferentially sequestered from the gas mixture and pulled into the aqueous phase [166]. For instance, n‐propane and n‐butane were separated from each other in this way – the capsules chose with high selectivity n‐butane as guest (2 : 2 complexes). Similarly, it was found that esters can be separated from each other when a kinetic resolution by saponification was performed in the presence of the capsules. The stronger binding ester was preferentially bound and therefore efficiently protected from the hydrolysis reaction. Binding of photoactive guests inside the capsule was also successfully exploited to control the outcome of photochemical reactions such as the Norrish type I reaction of the guests [170]. 2.4.4  Molecular Tubes Anthracene‐based molecular tubes have been reported by the group of Yoshi­zawa. They exhibit an orthogonal conformation between anthracene panels and linked meta‐phenylene linkers with additional substituents (Figure  2.16a) and thus possess a tubular hydrophobic cavity [171, 174]. These hosts are shape persistent and inherently fluorescent, which makes them promising candidates for supramolecular applications. Several water‐compatible anthracene tubes have been prepared by attachment of charged/polar side chains, and their supramolecular host–guest recognition properties were investigated in aqueous media [175, 176]. Expectably, hydrophobic guests such as alkanes and aromatic species are strongly bound, which can often be conveniently monitored by changes of the emission properties of the hosts. Interesting size‐selective binding properties  were observed. For instance, unsaturated methyl nervonate (Figure  2.16b) was bound with 100% selectivity over its fully saturated analog methyl tetracosanoate (H3C(CH2)22COOCH3) [171]. Similarly, hydrophobic steroids such as sodium cholate were strongly bound (Ka = 4 × 104 M−1) in aqueous media by an acridinium‐based molecular tube in an entropically driven process. For a hydrophobic, non‐charged polyaromatic molecular tube, an enthalpic driving force for binding, presumably on account of “high‐energy” water release was observed [177]. In some cases, such as for esculin (Figure 2.16b), 2 : 1 ternary complex formation upon simultaneous binding of two guests was witnessed, further enriching the possible host– guest chemistry that can be explored with these fascinating hosts [171]. To date, one of the largest hurdles to overcome is the low‐yielding synthesis of some of these molecular tubes. However, gram‐scale synthetic pathways have become available, [178] and these polyaromatic receptors will surely have a bright future ahead. Interesting naphthalene‐based molecular tubes were introduced by the Jiang group (Figure 2.16c). These hosts combine a hydrophobic interior with endo‐­ oriented NH‐amide groups as hydrogen bond donors [172, 173]. Thus, these tubes appear to combine the nonclassical hydrophobic effect as driving force for binding (ΔH° ≪ 0 was observed even for hydrophilic guests) with the guest binding selectivity that is provided by the hydrogen bonding unit. Indeed, hydrophilic 1,4‐dioxane was surprisingly strongly bound in aqueous media (Ka = 1 × 104 M−1) as a consequence of two NH⋯O hydrogen bonds established

2.4  Macrocyclic Receptors that (also) Bind Non‐charged Organic Guests R

R

R

R

R = O(CH2)3SO3 R

R

R

(a)

R Ka = 4 × 104 M–1 in H2O

OH O

HO HO

O

OH HO

O

O

HO

ONa

H

OH H Sodium cholate O

Esculin

OMe

Methyl nervonate

(b) O O

R = CH2COO

O O

+

Ka = 1 × 104 M–1 in H2O

O

Achiral macrocycle

Chiral epoxide

O O

H N O O OR N RO H OR RO

(c)

(d)

O

OH

Chiral complex

Ka = 4 × 104 M–1 in H2O

X-ray crystal structure

In water

Hydrogen bond

i.c.d. signal

Figure 2.16  Schematic representation of a water‐compatible polyaromatic molecular tube. Many more different designs can be found in the literature [171] (a). Typical guests that form binary (sodium cholate, methyl nervonate) or ternary complexes (esculin) with polyaromatic tubes in water (b). An endo‐functionalized molecular tube combines the hydrophobic effect as driving force for binding with hydrogen bonding recognition for selectivity control. Remarkable high binding affinities were observed for polar guests such as 1,4‐dioxane [172] (c). Binding of chiral epoxides leads to induced circular dichroism (i.c.d.) that can be utilized for chirality sensing and ee determination in aqueous media (d). Source: (d) Reproduced with permission from Ref. [173]. Copyright 2017 American Chemical Society.

63

64

2  Water‐Compatible Host Systems

between the endo‐functionalized host and 1,4‐dioxane [172]. Moreover, tetrahydrofuran (THF) and tetrahydropyran (THP), although more hydrophobic than 1,4‐dioxane, were more weakly bound, with Ka = 230 and 70 M−1, respectively, presumably because only one NH⋯O hydrogen bond was established for  each guest and some repulsive NH⋯CH2 contact may occur. Besides, 1,4‐­dioxane was clearly distinguished from 1,3‐dioxane (210 M−1) for which the establishment of two hydrogen bonds was not possible because of a poor geometric fit. This host design is very promising because it allows for the selective binding of guests beyond hydrophobicity as selection criterion, while most other hydrophobic hosts such as CBn macrocycles always show binding preference for the most hydrophobic guests. The directional hydrogen bonding contact also establishes a defined host–guest binding geometry, which can lead to i.c.d. effects without averaging out effects typically observed for chiral molecules that are conformationally more flexible. In addition, the host can provide an optical response even for spectroscopically silent guests. Indeed, chirality sensing and ee determination of a range of chiral epoxides was feasible with these endo‐functionalized naphthalene‐based hosts in aqueous media, capitalizing on i.c.d. effects (Figure 2.16d) [173].

2.5 ­Practitioner’s Guidelines for Choosing a Water‐Compatible Host When choosing a suitable water‐compatible host for a specific supramolecular application in mind, some key features are worth considering. 2.5.1  Guest Binding Affinity and Selectivity One usually tries to pick a host that has a sufficiently high affinity (Ka) in water for the binding partner at the given concentration range, where the binding relation Ka(complex)−1 = Kd(complex) ≈ c(guest) is fulfilled. Systems with Kd(complex) ≪ c(guest) are associated with a low degree of complex formation, while systems with Kd(complex) ≫ c(guest) may not exhibit a dynamic, stimuli‐responsive behavior. For most of the hosts discussed in this chapter, typical binding affinities are given in the corresponding sections. For instance, if a strong binding of a hydrophobic non‐charged or positively charged organic guest is desired, then cucurbit[n]urils (Section 2.4.2) are often the best starting point. For negatively charged hydrophobic organic guests, CDs (Section 2.4.1) are typically better candidates. Crown ethers and cryptands (Section 2.3.1) are strong (hard) metal cation binders, while bambus[n]urils (Section 2.3.2) are high‐affinity binders for a range of weakly coordinating anions. Azacrown ethers (Section 2.3.1) are useful for the binding of “soft” metal cations in aqueous media. Second, the binding selectivity of the host is often of great importance. Obviously, rather selective binding hosts are preferred when guest binding in complex media containing competitive binders, e.g. salts or metabolites, is required. For instance, the pyrrole‐based tris‐cationic receptor presented in Section 2.2.1 is useful for the selective recognition of an “oxoanion site” as found

2.5  Practitioner’s Guidelines for Choosing a Water‐Compatible Host

in amino acid carboxylates and peptides. Cryptands are suitable highly selective metal cation binders (Section 2.3.1), while the selective recognition of organic cationic species can be achieved by calix[n]arenes (Section 2.3.3) and pillar[n] arenes (Section  2.3.4). Molecular tweezers (Section  2.2.2) are useful for the selective recognition of peptides and proteins containing sterically accessible basic Lys and Arg residues. Deep cavitands (Section 2.4.3) are suitable for size‐ selective binding of organic guests. Some molecular tubes (Section  2.4.4) are useful for recognizing hydrophobic guest molecules with additional hydrogen bonding motifs, e.g. epoxides. For chiral guest recognition and separation, CDs (Section 2.4.1) are often a suitable platform. Guests with aromatic moieties can be selectively bound by CB8/auxiliary guest receptor complexes (Section 2.4.2). Low selective hosts can be utilized in less complex media or may be even desired as general binders in certain applications such as the removal of micropollutants from water (e.g. CDs, cucurbit[n]urils) or as general drug delivery platforms (micelles) (Section 2.2.4). 2.5.2 Availability/Scalability If the use of a water‐compatible host beyond fundamental research is considered, then it is crucial that the host is commercially available and/or that a facile and scalable synthetic route exists. Micelles and vesicles can be readily prepared from a very wide range of commercially available surfactants. CD homologs with different cavity sizes are easily obtained by enzymatic processes (Section 2.4.1) and are thus commodity materials. Crown ethers, cryptands, bambus[n]urils, calix[n] arenes, pillar[n]arenes, and cucurbit[n]urils are produced by simple synthetic approaches, and the yields can be relatively high if the reactions are performed under optimized reaction conditions such as in the presence of templates or acids or bases and at suitable temperatures. Importantly, these water‐compatible hosts are based on low‐cost starting materials such as glycol (crown ethers), phenols (calix[n]arenes and pillar[n]arenes), urea/glyoxal (CBn), and formaldehyde (calix[n]arenes, pillar[n]arenes, and CBn) such that their production at industrial scales is possible. To date, several (aza)crown ethers, cryptands, CDs, CBn, and calix[n]arenes can be purchased at moderate prices. The syntheses of deep cavitands are more elaborate, but preparative procedures have been developed that allow for gram‐scale production of a water‐soluble velcrand in six linear steps [165]. The synthetic routes to molecular tweezers and polyaromatic tubes are to date rather lengthy, somewhat restricting their use by non-expert groups. 2.5.3 Functionality It is often desirable that host compounds can be functionalized or derivatized in order to modify their molecular properties, e.g. solubility, biocompatibility, host–guest recognition characteristics, etc. Functionalizable hosts also enable preparation of immobilized systems that are needed for materials or sensor applications. CDs have many reactive hydroxy groups through which a variety of functional groups can be introduced. In addition, several organo‐soluble CDs have been obtained by introducing protecting groups (e.g. methyl and

65

66

2  Water‐Compatible Host Systems

benzyl groups); these organo‐soluble CDs can be subjected to a wider variety of organic reactions than the parent CDs. Several functionalized CDs are commercially available but significantly more expensive than native CDs. Likewise, functionalized calix[n]arenes and pillar[n]arenes can be readily prepared starting from their reactive phenolic moieties, which is possible because of their good solubility in most organic solvents. There are no reactive moieties in simple crown ethers and cryptands; thus functionalization steps on the respective linear precursors prior to the macrocyclization are required. Azacrowns are examples for readily functionalizable crown ethers but show generally weaker affinities for “hard” cations than parent crown ethers. The difficulty to functionalize CBn remains a significant practical issue for the CBn chemistry. Most protocols involve hydroxylation of preformed CBn macrocycles. Alternatively, monofunctionalized CBn can be obtained by macrocyclization of glycoluril oligomers with functionalizable units. Some functionalizable bambus[n]urils, molecular tweezers, deep cavitands, and polyaromatic molecular tubes were also reported. 2.5.4 Solubility The aqueous solubility is obviously one of the key features of water‐­compatible supramolecular hosts. Native CDs and many CD derivatives exhibit good solubility not only in water (α‐CD, up to 150 mM; β‐CD, up to 17 mM; γ‐CD, up to 180 mM) but also in highly polar solvents such as dimethyl sulfoxide (DMSO) and N,N‐dimethylformamide (DMF). The solubility of CDs in water is markedly increasing at higher temperature. Parent calix[n]arenes are only soluble in organic solvents (CHCl3, THF, DMF, etc.), but water‐soluble calix[n]arene derivatives are known. In particular, sulfonated calix[n]arenes are excellently soluble in aqueous media and are commercially available (Section  2.3.3). Likewise, water‐soluble pillar[n]arenes with charged side chains and amphiphilic pillar[n]arenes have been reported (Section 2.3.4). Crown ethers and cryptands are readily soluble in many organic solvents and in water on account of their oligo(ethylene oxide) building blocks; charged crown ether derivatives can be even more soluble in water (Section  2.3.1). Aqueous solubility (and organo‐­ solubility) is one of the largest problems of the CBn macrocycles. The poor solubility (≤100 μM in water) of the first discovered CB6 is so low that CBn host– guest chemistry was carried out in rather unattractive solvent mixtures, e.g. in aqueous formic acid, for decades. Some CB6 derivatives are significantly more soluble in both aqueous and organic media. The cucurbit[n]uril homologs CB5 and CB7 are much more soluble (≥10 mM) in water than CB6. Besides, buffer solutions generally increase the solubility of CBn on account of cation binding to the CBn portals. CB8 and CB10 are again less soluble (≤100 μM in water), which can cause problems. Fortunately, CB8 can be solubilized in the presence of suitable “first guests” such as paraquat (up to approx. 1 mM). The respective complexes allow for the binding of a second aromatic guest (Section  2.4.2). Highly water‐soluble deep cavitands, polyaromatic molecular tubes, and molecular tweezers are also known.

­  References

2.5.5 Biocompatibility/Toxicity For a potential real‐world application of water‐compatible molecular hosts where the compounds come in contact with humans and biosystems, those systems should be selected that are nontoxic and environmentally benign. Many biocompatible surfactants are known; others such as polyfluorinated ones can be problematic. CDs are generally biocompatible and little or nontoxic (depending on the ring size) and are thus widely used as expedient for drug solubilization and drug delivery applications (Section 2.4.1), as additives in the food industry, for the stabilization of flavors, and for the elimination of undesired tastes or other undesired compounds such as cholesterol. In general, CDs are by far the most widely studied macrocyclic hosts in terms of biocompatibility and toxicity aspects. Most calix[n]arenes showed good biocompatibility in animal models, but unfortunately only a limited number of derivatives have been assessed for their toxicity or immunological responses. Functionalized calix[n]arenes are being considered for pharmacology and cancer chemotherapy applications. Similarly, pillar[n]arenes are believed to be biocompatible and nontoxic, but the number of studies that have been conducted is still very limited. Crown ethers and even more so cryptands are toxic. Orthoester cryptands are believed to be more biocompatible because they can hydrolyze easily (Section 2.3.1). Cucurbit[n]urils are likely biocompatible although CBn and CBn complexes can bind to certain proteins and therefore affect enzymatic functions. CBn and several acyclic CBn derivatives were tested and found to be nontoxic, such that ex vivo and in vivo use of CBn‐based materials and chemosensors is in reach (Section 2.4.2). Molecular tweezers appear to be biocompatible and nontoxic, at least in the expected therapeutic window for the treatment of protein aggre­gation and neurodegenerative diseases (Section 2.2.2). For bambus[n]urils and ­foldamers, there is still insufficient toxicologic data available.

­References 1 Snyder, P., Lockett, M., Moustakas, D., and Whitesides, G. (2014). Is it the shape

2

3

4 5

of the cavity, or the shape of the water in the cavity? Eur. Phys. J.: Spec. Top. 223: 853–891. Houk, K.N., Leach, A.G., Kim, S.P., and Zhang, X. (2003). Binding affinities of host–guest, protein–ligand, and protein–transition‐state complexes. Angew. Chem. Int. Ed. 42: 4872–4897. Weber, P.C., Wendoloski, J.J., Pantoliano, M.W., and Salemme, F.R. (1992). Crystallographic and thermodynamic comparison of natural and synthetic ligands bound to streptavidin. J. Am. Chem. Soc. 114: 3197–3200. Laitinen, O.H., Nordlund, H.R., Hytönen, V.P., and Kulomaa, M.S. (2007). Brave new (strept)avidins in biotechnology. Trends Biotechnol. 25: 269–277. Young, T., Abel, R., Kim, B. et al. (2007). Motifs for molecular recognition exploiting hydrophobic enclosure in protein–ligand binding. Proc. Natl. Acad. Sci. U.S.A. 104: 808–813.

67

68

2  Water‐Compatible Host Systems

6 Schmuck, C. and Wienand, W. (2003). Highly stable self‐assembly in water: ion

pair driven dimerization of a guanidiniocarbonyl pyrrole carboxylate zwitterion. J. Am. Chem. Soc. 125: 452–459. 7 Schmuck, C. and Geiger, L. (2005). Efficient complexation of N‐acetyl amino acid carboxylates in water by an artificial receptor: unexpected cooperativity in the binding of glutamate but not aspartate. J. Am. Chem. Soc. 127: 10486–10487. 8 Schmuck, C. (2006). How to improve guanidinium cations for oxoanion binding in aqueous solution? The design of artificial peptide receptors. Coord. Chem. Rev. 250: 3053–3067. 9 Rao, J., Lahiri, J., Weis, R.M., and Whitesides, G.M. (2000). Design, synthesis, and characterization of a high‐affinity trivalent system derived from vancomycin and l‐Lys‐d‐Ala‐d‐Ala. J. Am. Chem. Soc. 122: 2698–2710. 10 Bier, D., Rose, R., Bravo‐Rodriguez, K. et al. (2013). Molecular tweezers modulate 14‐3‐3 protein–protein interactions. Nat. Chem. 5: 234–239. 11 Klärner, F.‐G. and Schrader, T. (2013). Aromatic interactions by molecular tweezers and clips in chemical and biological systems. Acc. Chem. Res. 46: 967–978. 12 Schrader, T., Bitan, G., and Klärner, F.‐G. (2016). Molecular tweezers for lysine and arginine – powerful inhibitors of pathologic protein aggregation. Chem. Commun. 52: 11318–11334. 13 Fokkens, M., Schrader, T., and Klärner, F.‐G. (2005). A molecular tweezer for lysine and arginine. J. Am. Chem. Soc. 127: 14415–14421. 14 Talbiersky, P., Bastkowski, F., Klärner, F.‐G., and Schrader, T. (2008). Molecular clip and tweezer introduce new mechanisms of enzyme inhibition. J. Am. Chem. Soc. 130: 9824–9828. 15 Lump, E., Castellano, L.M., Meier, C. et al. (2015). A molecular tweezer antagonizes seminal amyloids and HIV infection. eLife 4: e05397. 16 Hecht, S. and Huc, I. (2007). Foldamers: Structure, Properties and Applications. Weinheim: Wiley‐VCH. 17 Zhang, D.‐W., Zhao, X., Hou, J.‐L., and Li, Z.‐T. (2012). Aromatic amide foldamers: structures, properties, and functions. Chem. Rev. 112: 5271–5316. 18 Juwarker, H., Suk, J.‐m., and Jeong, K.‐S. (2009). Foldamers with helical cavities for binding complementary guests. Chem. Soc. Rev. 38: 3316–3325. 19 Prince, R.B., Barnes, S.A., and Moore, J.S. (2000). Foldamer‐based molecular recognition. J. Am. Chem. Soc. 122: 2758–2762. 20 Li, X., Wu, Y.‐D., and Yang, D. (2008). α‐Aminoxy acids: new possibilities from foldamers to anion receptors and channels. Acc. Chem. Res. 41: 1428–1438. 21 Meudtner, R.M. and Hecht, S. (2008). Helicity inversion in responsive foldamers induced by achiral halide ion guests. Angew. Chem. Int. Ed. 47: 4926–4930. 22 Chandramouli, N., El‐Behairy, M.F., Lautrette, G. et al. (2016). Polar solvent effects on tartaric acid binding by aromatic oligoamide foldamer capsules. Org. Biomol. Chem. 14: 2466–2472. 23 Suk, J.‐m. and Jeong, K.‐S. (2008). Indolocarbazole‐based foldamers capable of binding halides in water. J. Am. Chem. Soc. 130: 11868–11869. 24 Ferrand, Y. and Huc, I. (2018). Designing helical molecular capsules based on folded aromatic amide oligomers. Acc. Chem. Res. 51: 970–977.

­  References

25 Riess, J.G. (2002). Fluorous micro‐ and nanophases with a biomedical

perspective. Tetrahedron 58: 4113–4131.

26 Sletten, E.M. and Swager, T.M. (2016). Readily accessible multifunctional

fluorous emulsions. Chem. Sci. 7: 5091–5097.

27 Lawrence, M.J. and Rees, G.D. (2012). Microemulsion‐based media as novel

drug delivery systems. Adv. Drug Delivery Rev. 64: 175–193.

28 Trinh, H.M., Joseph, M., Cholkar, K. et al. (2017). Nanomicelles in diagnosis and

29 30 31

32

33

34

35 36 37 38

39

40 41 42

43

drug delivery. In: Emerging Nanotechnologies for Diagnostics, Drug Delivery and Medical Devices, 45–58. Boston: Elsevier. Schneider, H.‐J. and Yatsimirsky, A.K. (2008). Selectivity in supramolecular host–guest complexes. Chem. Soc. Rev. 37: 263–277. Pedersen, C.J. (1967). Cyclic polyethers and their complexes with metal salts. J. Am. Chem. Soc. 89: 7017–7036. Ouchi, M., Inoue, Y., Kanzaki, T., and Hakushi, T. (1984). Molecular design of crown ethers. 1. Effects of methylene chain length: 15‐ to 17‐crown‐5 and 18‐ to 22‐crown‐6. J. Org. Chem. 49: 1408–1412. Gokel, G.W., Leevy, W.M., and Weber, M.E. (2004). Crown ethers: sensors for ions and molecular scaffolds for materials and biological models. Chem. Rev. 104: 2723–2750. Zheng, B., Wang, F., Dong, S., and Huang, F. (2012). Supramolecular polymers constructed by crown ether‐based molecular recognition. Chem. Soc. Rev. 41: 1621–1636. Gokel, G.W., Barbour, L.J., Ferdani, R., and Hu, J. (2002). Lariat ether receptor systems show experimental evidence for alkali metal cation–π interactions. Acc. Chem. Res. 35: 878–886. Salorinne, K. and Nissinen, M. (2008). Calixcrowns: synthesis and properties. J. Inclusion Phenom.Macrocyclic Chem. 61: 11–27. Lehn, J.‐M. (1978). Cryptates: the chemistry of macropolycyclic inclusion complexes. Acc. Chem. Res. 11: 49–57. Schneider, H.‐J. (2012). Applications of Supramolecular Chemistry. Boca Raton: CRC Press. Cram, D.J., Kaneda, T., Helgeson, R.C. et al. (1985). Host–guest complexation. 35. Spherands, the first completely preorganized ligand systems. J. Am. Chem. Soc. 107: 3645–3657. Brachvogel, R.‐C., Maid, H., and von Delius, M. (2015). NMR studies on Li+, Na+ and K+ complexes of orthoester cryptand o‐Me2‐1.1.1. Int. J. Mol. Sci. 16: 20641. Weber, W.P. and Gokel, G.W. (2012). Phase Transfer Catalysis in Organic Synthesis. New York: Springer. Izatt, R.M., Bradshaw, J.S., Nielsen, S.A. et al. (1985). Thermodynamic and kinetic data for cation–macrocycle interaction. Chem. Rev. 85: 271–339. Buschmann, H.‐J., Schollmeyer, E., and Mutihac, L. (1998). The complexation of amino acids by crown ethers and cryptands in methanol. J. Inclusion Phenom. Macrocyclic Chem. 30: 21–28. Hendrixson, R.R., Mack, M.P., Palmer, R.A. et al. (1978). Oral toxicity of the cyclic polyethers – 12‐crown‐4, 15‐crown‐5, and 18‐crown‐6 – in mice. Toxicol. Appl. Pharmacol. 44: 263–268.

69

70

2  Water‐Compatible Host Systems

44 Tso, W.‐W., Fung, W.‐P., and Tso, M.‐Y.W. (1981). Variability of crown ether

toxicity. J. Inorg. Biochem. 14: 237–244.

45 Baudot, P., Jacque, M., and Robin, M. (1977). Effect of a diaza‐polyoxa‐

46 47

48 49 50 51 52

53 54

55

56 57 58 59 60 61 62

63

macrobicyclic complexing agent on the urinary elimination of lead in lead‐ poisoned rats. Toxicol. Appl. Pharmacol. 41: 113–118. Mewis, R.E. and Archibald, S.J. (2010). Biomedical applications of macrocyclic ligand complexes. Coord. Chem. Rev. 254: 1686–1712. Yawer, M.A., Havel, V., and Sindelar, V. (2015). A bambusuril macrocycle that binds anions in water with high affinity and selectivity. Angew. Chem. Int. Ed. 54: 276–279. Lisbjerg, M., Jessen, B.M., Rasmussen, B. et al. (2014). Discovery of a cyclic 6 + 6 hexamer of d‐biotin and formaldehyde. Chem. Sci. 5: 2647–2650. Lizal, T. and Sindelar, V. (2018). Bambusuril anion receptors. Isr. J. Chem. 58: 326–333. Havel, V. and Sindelar, V. (2015). Anion binding inside a bambus[6]uril macrocycle in chloroform. ChemPlusChem 80: 1601–1606. Biedermann, F. and Schneider, H.‐J. (2016). Experimental binding energies in supramolecular complexes. Chem. Rev. 116: 5216–5300. Lisbjerg, M., Valkenier, H., Jessen, B.M. et al. (2015). Biotin[6]uril esters: chloride‐selective transmembrane anion carriers employing C–H⋯anion interactions. J. Am. Chem. Soc. 137: 4948–4951. Lisbjerg, M., Nielsen, B.E., Milhoj, B.O. et al. (2015). Anion binding by biotin[6] uril in water. Org. Biomol. Chem. 13: 369–373. Arena, G., Pappalardo, A., Pappalardo, S. et al. (2015). Complexation of biologically active amines by a water‐soluble calix[5]arene. J. Therm. Anal. Calorim. 121: 1073–1079. Guo, D.S., Wang, L.H., and Liu, Y. (2007). Highly effective binding of methyl viologen dication and its radical cation by p‐sulfonatocalix[4,5]arenes. J. Org. Chem. 72: 7775–7778. Gutsche, C.D. (2008). Calixarenes: An Introduction. Cambridge: RSC. Thallapally, P.K., Lloyd, G.O., Atwood, J.L., and Barbour, L.J. (2005). Diffusion of water in a nonporous hydrophobic crystal. Angew. Chem. Int. Ed. 44: 3848–3851. Rebilly, J.‐N. and Reinaud, O. (2014). Calixarenes and resorcinarenes as scaffolds for supramolecular metallo‐enzyme mimicry. Supramol. Chem. 26: 454–479. Homden, D.M. and Redshaw, C. (2008). The use of calixarenes in metal‐based catalysis. Chem. Rev. 108: 5086–5130. Diamond, D. and McKervey, M.A. (1996). Calixarene‐based sensing agents. Chem. Soc. Rev. 25: 15–24. Sharma, K. and Cragg, P. (2011). Calixarene based chemical sensors. Chem. Senses 1: 1–18. Arena, G., Casnati, A., Contino, A. et al. (1999). Water‐soluble calixarene hosts that specifically recognize the trimethylammonium group or the benzene ring of aromatic ammonium cations: a combined 1H NMR, calorimetric, and molecular mechanics investigation. Chem. Eur. J. 5: 738–744. Liu, Y., Li, P.‐Y., and Chen, Y. (2016). Function‐directed supramolecular assembly based on water soluble calixarenes. In: Calixarenes and Beyond (ed. P. Neri, J.L. Sessler and M.‐X. Wang), 511–524. New York: Springer.

­  References

64 Guo, D.‐S. and Liu, Y. (2014). Supramolecular chemistry of p‐sulfonatocalix[n]

arenes and its biological applications. Acc. Chem. Res. 47: 1925–1934.

65 Gattuso, G., Notti, A., Pappalardo, S. et al. (2014). Recognition in water of

66

67 68

69

70

71 72 73 74

75

76

77 78 79

80

81

bioactive substrates by a sulphonato p‐tert‐butylcalix[5]arene. Supramol. Chem. 26: 597–600. Shinkai, S., Araki, K., and Manabe, O. (1988). Does the calixarene cavity recognise the size of guest molecules? On the ‘hole‐size selectivity’ in water‐ soluble calixarenes. J. Chem. Soc., Chem. Commun. 187–189. Notestein, J.M., Katz, A., and Iglesia, E. (2006). Energetics of small molecule and water complexation in hydrophobic calixarene cavities. Langmuir 22: 4004–4014. Rehm, M., Frank, M., and Schatz, J. (2009). Water‐soluble calixarenes – self‐ aggregation and complexation of noncharged aromatic guests in buffered aqueous solution. Tetrahedron Lett. 50: 93–96. Guo, D.‐S., Uzunova, V.D., Assaf, K.I. et al. (2015). Inclusion of neutral guests by water‐soluble macrocyclic hosts – a comparative thermodynamic investigation with cyclodextrins, calixarenes and cucurbiturils. Supramol. Chem. 28: 384–395. Hontama, N., Inokuchi, Y., Ebata, T. et al. (2009). Structure of the calix[4] arene‐(H2O) cluster: the world’s smallest cup of water. J. Phys. Chem. A 114: 2967–2972. Norouzy, A., Azizi, Z., and Nau, W.M. (2015). Indicator displacement assays inside live cells. Angew. Chem. Int. Ed. 54: 792–795. Cragg, P.J. and Sharma, K. (2012). Pillar[5]arenes: fascinating cyclophanes with a bright future. Chem. Soc. Rev. 41: 597–607. Hu, X.‐B., Chen, Z., Tang, G. et al. (2012). Single‐molecular artificial transmembrane water channels. J. Am. Chem. Soc. 134: 8384–8387. Ogoshi, T., Kanai, S., Fujinami, S. et al. (2008). para‐Bridged symmetrical pillar[5]arenes: their Lewis acid catalyzed synthesis and host–guest property. J. Am. Chem. Soc. 130: 5022–5023. Ogoshi, T., Yamagishi, T.‐a., and Nakamoto, Y. (2016). Pillar‐shaped macrocyclic hosts pillar[n]arenes: new key players for supramolecular chemistry. Chem. Rev. 116: 7937–8002. Murray, J., Kim, K., Ogoshi, T. et al. (2017). The aqueous supramolecular chemistry of cucurbit[n]urils, pillar[n]arenes and deep‐cavity cavitands. Chem. Soc. Rev. 46: 2479–2496. Wheate, N.J., Dickson, K.‐A., Kim, R.R. et al. (2016). Host–guest complexes of carboxylated pillar[n]arenes with drugs. J. Pharm. Sci. 105: 3615–3625. Zhang, H., Ma, X., Nguyen, K.T., and Zhao, Y. (2013). Biocompatible pillararene‐ assembly‐based carriers for dual bioimaging. ACS Nano 7: 7853–7863. Li, H., Chen, Q., Schönbeck, C., and Han, B.‐H. (2015). Sugar‐functionalized water‐soluble pillar[5]arene and its host–guest interaction with fullerene. RSC Adv. 5: 19041–19047. Yincheng, C., Kui, Y., Peng, W. et al. (2014). Cationic vesicles based on amphiphilic pillar[5]arene capped with ferrocenium: a redox‐responsive system for drug/siRNA co‐delivery. Angew. Chem. Int. Ed. 53: 13126–13130. Ogoshi, T., Shiga, R., and Yamagishi, T.‐a. (2012). Reversibly tunable lower critical solution temperature utilizing host–guest complexation of pillar[5]arene with triethylene oxide substituents. J. Am. Chem. Soc. 134: 4577–4580.

71

72

2  Water‐Compatible Host Systems

82 Ogoshi, T., Kida, K., and Yamagishi, T.‐a. (2012). Photoreversible switching of

83

84

85 86 87 88 89

90

91 92

93

94

95

96

97

98

the lower critical solution temperature in a photoresponsive host–guest system of pillar[6]arene with triethylene oxide substituents and an azobenzene derivative. J. Am. Chem. Soc. 134: 20146–20150. Yu, G., Ma, Y., Han, C. et al. (2013). A sugar‐functionalized amphiphilic pillar[5] arene: synthesis, self‐assembly in water, and application in bacterial cell agglutination. J. Am. Chem. Soc. 135: 10310–10313. Biedermann, F., Nau, W.M., and Schneider, H.‐J. (2014). The hydrophobic effect revisited – studies with supramolecular complexes imply high‐energy water as a noncovalent driving force. Angew. Chem. Int. Ed. 53: 11158–11171. Szejtli, J. (1998). Introduction and general overview of cyclodextrin chemistry. Chem. Rev. 98: 1743–1754. Dodziuk, H. (2006). Cyclodextrins and Their Complexes: Chemistry, Analytical Methods, Applications. Weinheim: Wiley‐VCH. Crini, G. (2014). Review: a history of cyclodextrins. Chem. Rev. 114: 10940–10975. Palepu, R. and Reinsborough, V. (1990). β‐Cyclodextrin inclusion of adamantane derivatives in solution. Aust. J. Chem. 43: 2119–2123. Zhang, L., Wu, Y., and Brunsveld, L. (2007). A synthetic supramolecular construct modulating protein assembly in cells. Angew. Chem. Int. Ed. 46: 1798–1802. Uhlenheuer, D.A., Wasserberg, D., Nguyen, H. et al. (2009). Modulation of protein dimerization by a supramolecular host–guest system. Chem. Eur. J. 15: 8779–8790. Guo, M. and Jiang, M. (2009). Non‐covalently connected micelles (NCCMs): the origins and development of a new concept. Soft Matter 5: 495–500. Chen, D. and Jiang, M. (2005). Strategies for constructing polymeric micelles and hollow spheres in solution via specific intermolecular interactions. Acc. Chem. Res. 38: 494–502. Yan, Q., Yuan, J., Cai, Z. et al. (2010). Voltage‐responsive vesicles based on orthogonal assembly of two homopolymers. J. Am. Chem. Soc. 132: 9268–9270. Assaf, K.I., Ural, M.S., Pan, F. et al. (2015). Water structure recovery in chaotropic anion recognition: high‐affinity binding of dodecaborate clusters to γ‐cyclodextrin. Angew. Chem. Int. Ed. 54: 6852–6856. Neirynck, P., Schimer, J., Jonkheijm, P. et al. (2015). Carborane‐β‐cyclodextrin complexes as a supramolecular connector for bioactive surfaces. J. Mater. Chem. B 3: 539–545. Assaf, K.I., Suckova, O., Al Danaf, N. et al. (2016). Dodecaborate‐functionalized anchor dyes for cyclodextrin‐based indicator displacement applications. Org. Lett. 18: 932–935. Schmidtchen, F.P. (2002). The anatomy of the energetics of molecular recognition by calorimetry: chiral discrimination of camphor by α‐cyclodextrin. Chem. Eur. J. 8: 3522–3529. Xiao, Y., Ng, S.‐C., Tan, T.T.Y., and Wang, Y. (2012). Recent development of cyclodextrin chiral stationary phases and their applications in chromatography. J. Chromatogr. A 1269: 52–68.

­  References

99 Ward, T.J. and Ward, K.D. (2012). Chiral separations: a review of current topics

and trends. Anal. Chem. 84: 626–635.

100 Nielen, M.W.F. (1993). Chiral separation of basic drugs using cyclodextrin‐

modified capillary zone electrophoresis. Anal. Chem. 65: 885–893.

101 Davis, M.E. and Brewster, M.E. (2004). Cyclodextrin‐based pharmaceutics:

past, present and future. Nat. Rev. Drug Discovery 3: 1023.

102 Frank, D., Gray, J., and Weaver, R. (1976). Cyclodextrin nephrosis in the rat.

Am. J. Pathol. 83: 367–382.

103 Yvan‐Charvet, L., Pagler, T., Gautier, E.L. et al. (2010). ATP‐binding cassette

104

105 106

107

108

109

110

111 112 113 114 115 116 117

transporters and HDL suppress hematopoietic stem cell proliferation. Science 328: 1689–1693. Wang, N., Lan, D., Chen, W. et al. (2004). ATP‐binding cassette transporters G1 and G4 mediate cellular cholesterol efflux to high‐density lipoproteins. Proc. Natl. Acad. Sci. U.S.A. 101: 9774–9779. Messner, M., Kurkov, S.V., Flavià‐Piera, R. et al. (2011). Self‐assembly of cyclodextrins: the effect of the guest molecule. Int. J. Pharm. 408: 235–247. Alsbaiee, A., Smith, B.J., Xiao, L. et al. (2015). Rapid removal of organic micropollutants from water by a porous β‐cyclodextrin polymer. Nature 529: 190–194. Schäfer, A.I., Stelzl, K., Faghih, M. et al. (2018). Poly(ether sulfone) nanofibers impregnated with β‐cyclodextrin for increased micropollutant removal from water. ACS Sustainable Chem. Eng. 6: 2942–2953. Harada, A., Takashima, Y., and Nakahata, M. (2014). Supramolecular polymeric materials via cyclodextrin‐guest interactions. Acc. Chem. Res. 47: 2128–2140. Nijhuis, C.A., Boukamp, B.A., Ravoo, B.J. et al. (2007). Electrochemistry of ferrocenyl dendrimer‐β‐cyclodextrin assemblies at the interface of an aqueous solution and a molecular printboard. J. Phys. Chem. C 111: 9799–9810. Huskens, J., Deij, M.A., and Reinhoudt, D.N. (2002). Attachment of molecules at a molecular printboard by multiple host–guest interactions. Angew. Chem. Int. Ed. 41: 4467–4471. Auletta, T., Dordi, B., Mulder, A. et al. (2004). Writing patterns of molecules on molecular printboards. Angew. Chem. Int. Ed. 43: 369–373. Sinn, S. and Biedermann, F. (2018). Chemical sensors based on cucurbit[n]uril macrocycles. Isr. J. Chem. 58: 357–412. Barrow, S.J., Kasera, S., Rowland, M.J. et al. (2015). Cucurbituril‐based molecular recognition. Chem. Rev. 115: 12320–12406. Assaf, K.I. and Nau, W.M. (2015). Cucurbiturils: from synthesis to high‐affinity binding and catalysis. Chem. Soc. Rev. 44: 394–418. Masson, E., Ling, X., Joseph, R. et al. (2012). Cucurbituril chemistry: a tale of supramolecular success. RSC Adv. 2: 1213–1247. Lagona, J., Mukhopadhyay, P., Chakrabarti, S., and Isaacs, L. (2005). The cucurbit[n]uril family. Angew. Chem. Int. Ed. 44: 4844–4870. Nau, W.M., Florea, M., and Assaf, K.I. (2011). Deep inside cucurbiturils: physical properties and volumes of their inner cavity determine the hydrophobic driving force for host–guest complexation. Isr. J. Chem. 51: 559–577.

73

74

2  Water‐Compatible Host Systems

118 Cheng, X.‐J., Liang, L.‐L., Chen, K. et al. (2013). Twisted cucurbit[14]uril.

Angew. Chem. Int. Ed. 52: 7252–7255.

119 Dong, N., He, J., Li, T. et al. (2018). Synthesis and binding properties of

120 121

122

123

124

125

126

127

128 129 130

131

132 133 134 135

monohydroxycucurbit[7]uril: a key derivative for the functionalization of cucurbituril hosts. J. Org. Chem. 83: 5467–5473. Kim, K., Selvapalam, N., Ko, Y.H. et al. (2007). Functionalized cucurbiturils and their applications. Chem. Soc. Rev. 36: 267–279. Ma, D., Hettiarachchi, G., Nguyen, D. et al. (2012). Acyclic cucurbit[n]uril molecular containers enhance the solubility and bioactivity of poorly soluble pharmaceuticals. Nat. Chem. 4: 503–510. Cao, L., Hettiarachchi, G., Briken, V., and Isaacs, L. (2013). Cucurbit[7]uril containers for targeted delivery of oxaliplatin to cancer cells. Angew. Chem. Int. Ed. 52: 12033–12037. Macartney, D.H. (2011). Encapsulation of drug molecules by cucurbiturils: effects on their chemical properties in aqueous solution. Isr. J. Chem. 51: 600–615. Shetty, D., Khedkar, J.K., Park, K.M., and Kim, K. (2015). Can we beat the biotin–avidin pair? cucurbit[7]uril‐based ultrahigh affinity host–guest complexes and their applications. Chem. Soc. Rev. 44: 8747–8761. Kolman, V., Kulhanek, P., and Sindelar, V. (2010). Inclusion of carboxyl function inside of cucurbiturils and its use in molecular switches. Chem. Asian J. 5: 2386–2392. Lazar, A.I., Biedermann, F., Mustafina, K.R. et al. (2016). Nanomolar binding of steroids to cucurbit[n]urils: selectivity and applications. J. Am. Chem. Soc. 138: 13022–13029. Jang, Y., Natarajan, R., Ko, Y.H., and Kim, K. (2014). Cucurbit[7]uril: a high‐ affinity host for encapsulation of amino saccharides and supramolecular stabilization of their α‐anomers in water. Angew. Chem. Int. Ed. 53: 1003–1007. Chinai, J.M., Taylor, A.B., Ryno, L.M. et al. (2011). Molecular recognition of insulin by a synthetic receptor. J. Am. Chem. Soc. 133: 8810–8813. Hennig, A., Ghale, G., and Nau, W.M. (2007). Effects of cucurbit[7]uril on enzymatic activity. Chem. Commun. 1614–1616. Parente Carvalho, C., Norouzy, A., Ribeiro, V. et al. (2015). Cucurbiturils as supramolecular inhibitors of DNA restriction by type II endonucleases. Org. Biomol. Chem. 13: 2866–2869. Biedermann, F., Uzunova, V.D., Scherman, O.A. et al. (2012). Release of high‐energy water as an essential driving force for the high‐affinity binding of cucurbit[n]urils. J. Am. Chem. Soc. 134: 15318–15323. Cao, L., Šekutor, M., Zavalij, P.Y. et al. (2014). Cucurbit[7]uril·guest pair with an attomolar dissociation constant. Angew. Chem. Int. Ed. 53: 988–993. Ahn, Y., Jang, Y., Selvapalam, N. et al. (2013). Supramolecular velcro for reversible underwater adhesion. Angew. Chem. Int. Ed. 52: 3140–3144. Brinkmann, J., Cavatorta, E., Sankaran, S. et al. (2014). About supramolecular systems for dynamically probing cells. Chem. Soc. Rev. 43: 4449–4469. Lee, D.‐W., Park, K.M., Banerjee, M. et al. (2011). Supramolecular fishing for plasma membrane proteins using an ultrastable synthetic host–guest binding pair. Nat. Chem. 3: 154–159.

­  References

136 Dsouza, R.N., Hennig, A., and Nau, W.M. (2012). Supramolecular tandem

enzyme assays. Chem. Eur. J. 18: 3444–3459.

137 Park, K.M., Yang, J.A., Jung, H. et al. (2012). In situ supramolecular assembly

138

139

140

141

142

143 144

145

146

147

148

149

150

151

and modular modification of hyaluronic acid hydrogels for 3D cellular engineering. ACS Nano 6: 2960–2968. Appel, E.A., Loh, X.J., Jones, S.T. et al. (2012). Ultrahigh‐water‐content supramolecular hydrogels exhibiting multistimuli responsiveness. J. Am. Chem. Soc. 134: 11767–11773. Zhang, K.‐D., Tian, J., Hanifi, D. et al. (2013). Toward a single‐layer two‐ dimensional honeycomb supramolecular organic framework in water. J. Am. Chem. Soc. 135: 17913–17918. Kim, J., Baek, K., Shetty, D. et al. (2015). Reversible morphological transformation between polymer nanocapsules and thin films through dynamic covalent self‐assembly. Angew. Chem. Int. Ed. 54: 2693–2697. Kim, D., Kim, E., Lee, J. et al. (2010). Direct synthesis of polymer nanocapsules: self‐assembly of polymer hollow spheres through irreversible covalent bond formation. J. Am. Chem. Soc. 132: 9908–9919. Yun, G., Hassan, Z., Lee, J. et al. (2014). Highly stable, water‐dispersible metal‐nanoparticle‐decorated polymer nanocapsules and their catalytic applications. Angew. Chem. Int. Ed. 53: 6414–6418. Taylor, R.W., Coulston, R.J., Biedermann, F. et al. (2013). In situ SERS monitoring of photochemistry within a nanojunction reactor. Nano Lett. 13: 5985–5990. Wu, Y., Lan, Y., Liu, J., and Scherman, O.A. (2015). Catalytic polymeric nanocomposites via cucurbit[n]uril host–guest interactions. Nanoscale 7: 13416–13419. Biedermann, F. and Scherman, O.A. (2012). Cucurbit[8]uril mediated donor‐ acceptor ternary complexes: a model system for studying charge‐transfer interactions. J. Phys. Chem. B 116: 2842–2849. Jeon, W.S., Kim, E., Ko, Y.H. et al. (2005). Molecular loop lock: a redox‐driven molecular machine based on a host‐stabilized charge‐transfer complex. Angew. Chem. Int. Ed. 44: 87–91. Wang, W. and Kaifer, A.E. (2006). Electrochemical switching and size selection in cucurbit[8]uril‐mediated dendrimer self‐assembly. Angew. Chem. Int. Ed. 45: 7042–7046. Heitmann, L.M., Taylor, A.B., Hart, P.J., and Urbach, A.R. (2006). Sequence‐ specific recognition and cooperative dimerization of N‐terminal aromatic peptides in aqueous solution by a synthetic host. J. Am. Chem. Soc. 128: 12574–12581. Biedermann, F., Rauwald, U., Zayed, J.M., and Scherman, O.A. (2011). A supramolecular route for reversible protein–polymer conjugation. Chem. Sci. 2: 279–286. Gubeli, R.J., Sonzini, S., Podmore, A. et al. (2016). Selective, non‐covalent conjugation of synthetic peptides with recombinant proteins mediated by host–guest chemistry. Chem. Commun. 52: 4235–4238. Uhlenheuer, D.A., Young, J.F., Nguyen, H.D. et al. (2011). Cucurbit[8]uril induced heterodimerization of methylviologen and naphthalene functionalized proteins. Chem. Commun. 47: 6798–6800.

75

76

2  Water‐Compatible Host Systems

152 Liu, J., Tan, C.S.Y., Lan, Y., and Scherman, O.A. (2016). Aqueous polymer

153

154

155

156 157

158

159

160

161

162 163 164

165 166 167 168

self‐assembly based on cucurbit[n]uril‐mediated host–guest interactions. Macromol. Chem. Phys. 217: 319–332. Appel, E.A., Biedermann, F., Rauwald, U. et al. (2010). Supramolecular cross‐ linked networks via host–guest complexation with cucurbit[8]uril. J. Am. Chem. Soc. 132: 14251–14260. Stoffelen, C., Voskuhl, J., Jonkheijm, P., and Huskens, J. (2014). Dual stimuli‐ responsive self‐assembled supramolecular nanoparticles. Angew. Chem. Int. Ed. 53: 3400–3404. Cavatorta, E., Verheijden, M.L., van Roosmalen, W. et al. (2016). Functionalizing the glycocalyx of living cells with supramolecular guest ligands for cucurbit[8]uril‐mediated assembly. Chem. Commun. 52: 7146–7149. Tian, F., Jiao, D., Biedermann, F., and Scherman, O.A. (2012). Orthogonal switching of a single supramolecular complex. Nat. Commun. 3: 1207. del Barrio, J., Horton, P.N., Lairez, D. et al. (2013). Photocontrol over cucurbit[8]uril complexes: stoichiometry and supramolecular polymers. J. Am. Chem. Soc. 135: 11760–11763. Biedermann, F., Ross, I., and Scherman, O.A. (2014). Host–guest accelerated photodimerisation of anthracene‐labeled macromolecules in water. Polym. Chem. 5: 5375–5382. Yu, Z., Zhang, J., Coulston, R.J. et al. (2015). Supramolecular hydrogel microcapsules via cucurbit[8]uril host–guest interactions with triggered and UV‐controlled molecular permeability. Chem. Sci. 6: 4929–4933. Biedermann, F. and Nau, W.M. (2014). Noncovalent chirality sensing ensembles for the detection and reaction monitoring of amino acids, peptides, proteins, and aromatic drugs. Angew. Chem. Int. Ed. 53: 5694–5699. Biedermann, F., Hathazi, D., and Nau, W.M. (2015). Associative chemosensing by fluorescent macrocycle‐dye complexes – a versatile enzyme assay platform beyond indicator displacement. Chem. Commun. 51: 4977–4980. Biros, S.M. and Rebek, J. Jr. (2007). Structure and binding properties of water‐soluble cavitands and capsules. Chem. Soc. Rev. 36: 93–104. Ewell, J., Gibb, B.C., and Rick, S.W. (2008). Water inside a hydrophobic cavitand molecule. J. Phys. Chem. B 112: 10272–10279. Sebo, L., Diederich, F., and Gramlich, V. (2000). Tetrakis(phenylamidinium)‐ substituted resorcin[4]arene receptors for the complexation of dicarboxylates and phosphates in protic solvents. Helv. Chim. Acta 83: 93–113. Mosca, S., Yu, Y., and Rebek, J. Jr. (2016). Preparative scale and convenient synthesis of a water‐soluble, deep cavitand. Nat. Protoc. 11: 1371–1387. Jordan, J.H. and Gibb, B.C. (2015). Molecular containers assembled through the hydrophobic effect. Chem. Soc. Rev. 44: 547–585. Mosca, S., Yu, Y., Gavette, J.V. et al. (2015). A deep cavitand templates lactam formation in water. J. Am. Chem. Soc. 137: 14582–14585. Gibb, C.L.D. and Gibb, B.C. (2011). Anion binding to hydrophobic concavity is central to the salting‐in effects of Hofmeister chaotropes. J. Am. Chem. Soc. 133: 7344–7347.

­  References

169 Gibb, C.L.D., Oertling, E.E., Velaga, S., and Gibb, B.C. (2015). Thermodynamic

170 171

172

173

174 175

176

177

178

profiles of salt effects on a host–guest system new insight into the Hofmeister effect. J. Phys. Chem. B 119: 5624–5638. Ramamurthy, V. (2015). Photochemistry within a water‐soluble organic capsule. Acc. Chem. Res. 48: 2904–2917. Yazaki, K., Catti, L., and Yoshizawa, M. (2018). Polyaromatic molecular tubes: from strategic synthesis to host functions. Chem. Commun. 54: 3195–3206. Huang, G.‐B., Wang, S.‐H., Ke, H. et al. (2016). Selective recognition of highly hydrophilic molecules in water by endo‐functionalized molecular tubes. J. Am. Chem. Soc. 138: 14550–14553. Wang, L.‐L., Chen, Z., Liu, W.‐E. et al. (2017). Molecular recognition and chirality sensing of epoxides in water using endo‐functionalized molecular tubes. J. Am. Chem. Soc. 139: 8436–8439. Yoshizawa, M. and Klosterman, J.K. (2014). Molecular architectures of multi‐anthracene assemblies. Chem. Soc. Rev. 43: 1885–1898. Hagiwara, K., Akita, M., and Yoshizawa, M. (2015). An aqueous molecular tube with polyaromatic frameworks capable of binding fluorescent dyes. Chem. Sci. 6: 259–263. Hagiwara, K., Otsuki, M., Akita, M., and Yoshizawa, M. (2015). Polyaromatic molecular tubes with a subnanometer pore and the guest‐induced emission enhancement behavior. Chem. Commun. 51: 10451–10454. Shunsuke, K., Masahiro, Y., Munetaka, A. et al. (2018). Hydrophilic oligo(lactic acid)s captured by a hydrophobic polyaromatic cavity in water. Angew. Chem. Int. Ed. 57: 3706–3710. Kiyonori, K., Masafumi, O., Kohei, Y. et al. (2018). Alkylation‐, heating‐, and doping‐induced emission enhancement of a polyaromatic tube in the solid state. Chem. Asian J. 13: 515–519.

77

79

3 Artificial Peptide and Protein Receptors Joydev Hatai and Carsten Schmuck University of Duisburg‐Essen, Institute of Organic Chemistry, Universitätsstraße 7, 45141 Essen, Germany

3.1 ­Introduction The specific recognition of a peptide or protein underlies many biological pro­ cesses. Artificial receptors that bind to the same targets, thereby inhibiting pro­ tein–protein interactions, thus open new perspectives in drug discovery. In this chapter, a selection of artificial receptors is presented that efficiently recognize peptide sequences or bind to protein surfaces. Furthermore, the detection and discrimination of various proteins using cross‐reactive sensor arrays (so‐called chemical “noses” or “tongues”) followed by data deconvolution through chemo­ metric methods are discussed. The final part of the chapter deals with approaches to develop unimolecular sensing systems that allow the detection of proteins or protein families in complex biological mixtures.

3.2 ­Peptide Recognition Artificial receptors that selectively bind to a specific amino acid residue or a short peptide sequence under biological conditions are attractive targets in supramolecular chemistry. Whereas numerous receptors have been reported that allow peptide recognition in organic solvents, examples of the successful recognition of peptides in aqueous solution are still limited. The problems asso­ ciated with the recognition of a peptide in water are the low hydrogen bonding strength and the weakness of ion‐pair interactions. Thus, molecular recognition relying on only hydrogen bonding interactions usually does not work in aqueous media. To overcome these limitations and to reinforce the interactions, addi­ tional binding motifs must be introduced into the receptors. Such modifications can rely on binding modes such as hydrophobic interactions, ion‐pair interac­ tions, cation–π interactions, multiple complementary H‐bonds, or metal coordi­ nation. In the following sections, examples of artificial receptors for peptides in water are presented whose binding modes are based on this strategy. Supramolecular Chemistry in Water, First Edition. Edited by Stefan Kubik. © 2019 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2019 by Wiley-VCH Verlag GmbH & Co. KGaA.

80

3  Artificial Peptide and Protein Receptors (a) NHAc NHAc NHAc NHAc AA1, AA2, AA3 = Asn, Arg, Glu, His, Leu, Lys, Pro, Ser, Trp, Tyr AA3 AA3 AA3 AA3 2 2 2 AA AA2 R = CH CH OEt AA AA 2 2 O AA1 AA1 AA1 AA1 O Gly Gly Gly Gly N O Linker = H NH NH HN NH O 3.1

RO OR OR O Linker HN

(b) HO

N

O BB1

BB2

BB3

N H

O

O

O

N

O

H N

N O

Figure 3.1  General structures of the 1000‐membered peptidocalix[4]arene library 3.1 (a) and the rhodamine 101‐labeled guests (b).

3.2.1 Calixarenes Calixarenes are characterized by a three‐dimensional cup or bucket shape with a wider upper rim, a narrower lower rim, and a central annulus. The hydrophobic cavity can encapsulate smaller molecules or ions, while the lower and upper rims can easily be decorated with various functionalities. Calixarenes are thus attrac­ tive scaffolds for the development of receptors for amino acids and peptides whose binding mode can involve both hydrophobic and ion‐pair interactions. Hioki and coworkers synthesized a library of four‐armed hydrophilic peptidoca­ lix[4]arenes of the general structure 3.1 (Figure 3.1a) decorated with linear tetra­ peptides containing polar, acidic, or basic amino acids along the upper rim [1]. The properties of these calixarenes were assessed by using an on‐bead assay and seven different tripeptides as substrates, labeled with the fluorescent dye rhodamine 101 and varying in the type and sequence of the amino acid residues (Figure 3.1b). The binding studies were performed at pH 6.9 in phosphate buffer containing 5% Triton X‐100 to suppress the interaction of the dyes with the resin beads. Brightly fluores­ cent beads were picked and analyzed spectroscopically. These investigations showed that neutral peptides did not bind to 3.1. Expectedly, acidic receptors preferentially bound to basic peptides and vice versa. For e­ xample, an acidic tripeptide substrate containing the sequence Ser‐Asp‐Asp (BB1‐BB2‐BB3) bound more strongly to a basic receptor carrying an Arg residue than to receptors containing Lys or His. These findings suggested that binding was dominated by electrostatic interactions. 3.2.2  Guanidiniocarbonyl Pyrroles Schmuck et al. designed the receptor 3.2 for the selective and specific detection of small peptide sequences in aqueous media [2]. 3.2 comprised a Ser‐Lys dipeptide

3.2  Peptide Recognition

containing a naphthyl residue at the C‐terminal end and a guanidiniocarbonyl pyrrole (GCP) moiety at the N‐terminus (Figure 3.2a), which is weakly basic and therefore only partially protonated under physiological conditions (pKa 6–7). The interactions between 3.2 and various N‐acetylated amino acids, dipeptides, and tripeptides were studied. Nuclear magnetic resonance (NMR) titrations and mass spectrometry analyses confirmed the formation of 1 : 1 complexes between 3.2 and dipeptides. UV titrations in 20% dimethyl sulfoxide (DMSO) and buffered water (5 mM bis‐Tris buffer, pH  =  6.0) showed that 3.2 interacted only weakly with amino acids such Ac‐l‐Ala‐OH or Ac‐l‐Phe‐OH (Ka  NDP > NMP, which

119

120

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

can be ascribed to the larger number of negative charges present in trinucleo­ tides. Nevertheless, other intermolecular forces such as hydrogen bonding also play significant roles. The interplay of these different but somehow related forces is highlighted in an early structure of an adduct formed between pyroph­ osphate (P2O74−, PPi) and tetraprotonated spermine (H4(4.3)4+) in which the anion forms a hydrogen bond network with the charged ammonium groups and water molecules [24]. NH2

H2N 4.1

H N

H2N

NH2

H2N

N H

4.2

H N

NH2

4.3

Chart 4.3 

The relevance of electrostatic interactions was evidenced in an early report by Tabushi et al. [25] They found that 1,4‐diazabicyclo[2.2.2]octane (DABCO), whose two nitrogen atoms were quaternized with stearyl chains (4.4; Chart 4.4), was able to very efficiently transfer at pH values of 3–5 adenosine diphosphate (ADP) from aqueous solutions into a chloroform phase. The selectivity found for the transfer of ADP over AMP was ascribed to the formation of two salt bridges between the phosphate groups and the dicationic receptor.

NH HN

4.4 O

NH

HN

NH

N C18H37

H37C18 N

NH

HN

O

O

HN

NH

HN

NH HN

HN

NH

NH HN

NH H HN N

4.5

4.6

HN NH

N NH H HN

N H

NH

HN

NH HN

4.8

4.9

N H

NH NH

HN HN

H N 4.10

4.7 N H

NH NH

HN HN

N H

NH NH

HN HN

O

O 4.11

Chart 4.4 

4.12

N NH

NH

HN

NH

HN

HN

NH

HN

4.13

4.14

4.3  Nucleotide Receptors

The research groups of Lehn and Kimura provided evidence for the strong interaction between polyammonium azamacrocycles and the nucleotides ATP, ADP, and AMP [1, 2]. Lehn’s group reported the interaction of the macrocycles [24]aneN6 (4.5), [32]aneN8 (4.6), and [27]aneN6O3 (4.7) with AMP, ADP, and ATP (Chart 4.4). The values of the binding constants obtained by analysis of the pH‐metric titration curves varied from 3.4 to 9.1 logarithmic units, following the sequence AMP  log K2 (log K1 = 5.36, log K2 = 2.89, log K3 = 3.85). This unusual trend was rationalized by assuming that all three arms of the receptor contributed to AMP binding in the 1  :  1 complex. Binding of the second guest required the receptor to produce an appropriate binding site that allowed accommodating the second guest without pronounced repulsion between the two anions. The third AMP was then expected to more easily find a suitable receptor arrangement for binding, explaining the increase in the binding constant of the third step. 4.3.2  Receptors with Aromatic Units As outlined in the introduction, the aromatic rings of nucleotides represent an important anchor point for binding through π‐stacking interactions. Accordingly, Lehn and coworkers [34] prepared receptors 4.26 and 4.27 (Chart 4.7) in which one or more acridine moieties were attached to the parent receptor O‐Bisdien (4.25) [35].

123

124

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials O

NH

HN

4.25 R1 = R2 = H (O-Bisdien) N R2

R1 N NH

O

HN

N

4.26 R1 = H, R2 = X N H

X=

4.27 R1 = R2 = X

Chart 4.7  1

H NMR experiments showed that 4.26 and 4.27 formed stronger complexes with ATP than with 4.25 due to the additional π‐stacking interactions between the acridine rings and the nucleobase. Lehn and coworkers also prepared a series of bis‐intercaland compounds, for example, receptor 4.28 and acyclic analogs, which primarily targeted the nucleobase group of the substrates (Chart 4.8) [36]. Evidence for nucleobase intercalation came from UV–vis spectroscopy. The stabilities of the complexes were shown to depend on the nucleobase, while the stability constants obtained for nucleotides having the same base but differing in the length of the phosphate chain were practically the same. For instance, the log Ka values for the interaction of AMP, ADP, and ATP with 4.28 were 3.79, 3.84, and 3.91, respectively, in spite of the different net charges of the guests. S O

H N N

N Me OO

R S O N Me

O 4.28

N H

N H

H N

H N

NH

NH

4.30

N H N 4.29a R = (CH2)4 4.29b R = (CH2)6 4.29c R = p-C6H4

H2N

H 2N

N H H N

NH

NH

4.31

Chart 4.8 

The importance of π‐stacking in nucleotide binding was even more pro­ nounced in the case of receptors 4.29a–c (Chart 4.8). For example, the affinity of 4.29a for ATP, ADP, and AMP amounted to log Ka = 5.80, 5.65, and 5.38, respec­ tively [37]. Monocyclic azacyclophanes with appropriate polyamine chains can behave as multipoint binders of nucleotides, as shown for 4.30 (Chart 4.8) by using 1H NMR and potentiometric techniques [38, 39]. While the phosphate chain served as an electrostatic binding point and hydrogen bond acceptor, the nucleobase stacked with the aromatic part of the macrocycle. These interactions resulted in upfield shifts of the signals of the anomeric and the aromatic protons of the nucleotide substrate in the 1H NMR spectra. They were likely the reason for the

4.3  Nucleotide Receptors 7 5

% complexed ATP

log Keff

100

4.30·ATP

6 4.31·ATP

4 3 2 1

(a)

4.30·ATP

80

Free ATP

60 40 4.31·ATP

20 0

2

4

6

pH

8

10

(b)

2

4

6

pH

8

10

Figure 4.5  (a) Plots of log Keff vs. pH for the systems 4.30∙ATP and 4.31∙ATP and (b) of the overall amount of complexed ATP in a mixed system containing ATP, 4.30 and 4.31, in equimolar amounts. Source: Adapted with permission from Ref. [19]. Copyright 1981 American Chemical Society.

about one order of magnitude higher Keff found for 4.30∙ATP in comparison with 4.31∙ATP between pH 4 and 7 (Figure 4.5a). The observed differences in stability were also consistent with the higher percentage of ATP complexed by 4.30 than by 4.31 (Figure 4.5b). Nevertheless, it has to be stressed that 4.30 and 4.31 are not strictly comparable due to the presence of primary amino groups in 4.31 that have higher hydration energies. Several acyclic receptors, containing either one or two naphthalene or anthra­ cene groups appended to the polyamine chains, were also constructed to take advantage of π‐stacking interaction in nucleotide binding [28, 40]. Examples are compounds 4.32–4.35 (Chart 4.9). Among the receptors sharing the same polyamine chain, the receptor with one anthrylmethyl substituent interacted stronger with ATP than the one with one naphthylmethyl group due to the more efficient π‐stacking induced by the larger anthracene unit. However, the recep­ tor with one anthrylmethyl group exhibited comparable affinity as the receptor with naphthalene units at both chain ends. NMR studies showed that π‐­stacking contributed to binding throughout the pH range where complex formation was observed. Nuclear Overhauser effect (NOE) experiments confirmed the prox­ imity of the aromatic rings of the receptors and ATP and allowed for proposing a model for the binding mode, combining Coulomb attraction, hydrogen bond­ ing, and π‐stacking between the aromatic groups. 4.32 R1 = H, R2 = Nap R1

NH

N H

N H

HN

4.33 R1 = R2 = Nap R2

4.34 R1 = H, R2 = Ant

Nap =

Ant =

4.35 R1 = R2 = Ant

Chart 4.9 

Bencini and Tripier reported the acyclic receptors 4.36 and 4.37 in which two tetraamine fragments were interconnected by m‐xylene or 2,6‐bis(methyl) pyridine units, respectively (Chart 4.10). Both receptors formed stable com­ plexes with ATP at neutral pH [41]. Although stacking interactions occurred in

125

126

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials R

H2N

N H

N H

H N

X

H N

N H

N H

NH2

4.36 R = H, X = CH 4.37 R = H, X = N 4.38 R = H, X =

Chart 4.10 

the complexes according to NMR studies, the percentages of ATP and trip­ olyphosphate (TPP) bound to the receptors did not differ much throughout the pH range suitable for complex formation. Both receptors were flexible enough to organize themselves around the nucleotides, maximizing the number of attachment points in the interaction with the substrates. The introduction of the anthracene ring in 4.38 provided a means for using fluorescence spectros­ copy for the detection of nucleotides [42]. However, although the large anthra­ cene ring should favor π‐stacking with ATP, no improvement of complex stability was observed in comparison with the corresponding complex of 4.37, which was attributed by the authors to the aggregation of the receptor prior to ATP binding. Phenanthroline‐based monocyclic receptors exhibited an interesting multi­ point binding ability for nucleotides [43, 44]. Receptor 4.39 (Chart 4.11), in particular, showed good selectivity for ATP over the other triphosphate nucleo­ tides such as TTP, cytidine 5′‐triphosphate (CTP), and GTP. Molecular dynam­ ics calculations indicated that ATP adopted a conformation in the complex with 4.39 that allowed for the simultaneous involvement of π‐stacking, charge–charge interactions, and hydrogen bond contacts. Furthermore, the quenching of the fluorescence of 4.39 upon addition of ATP allowed for the selective sensing of this guest in aqueous solution.

NH N

N H

HN 4.39

N NH

H N

HN

Chart 4.11 

Tripier, Bencini, and coworkers prepared the series of receptors 4.40–4.45 in  which 1,4,7,10‐tetrazadecane units, either reinforced or not, were connec­ ted  through benzene or pyridine spacers (Chart 4.12) [45–47]. They obtained

4.3  Nucleotide Receptors

NH HN

NH HN NH

N

N N

N

X

HN

4.40 (X = CH)

NH

NH2

N

NH2

4.41 (X =N)

HN N N

N

N

N

N

N H

H2N

N

H2N

N

N

N HN

N 4.42

H N

NH2 4.43

N

NH HN

N

N

N

N

N

HN

N H

NH

NH2

NH

N N

N N NH 4.44

N N

N

HN H N

N

N 4.45

N

Chart 4.12 

spec­troscopic evidence for the participation of π‐stacking interactions in the binding of receptors 4.40 and 4.44 to AMP, ADP, and ATP. Additionally, the protona­tion of the pyridine ring in 4.41 at around pH 2 reinforced affinity with respect to 4.40. The oligomeric 4.42 behaved as a double proton sponge, yielding a H2(4.42)2+ species, which was protonated at the peripheral secondary amines and could not be deprotonated even at high pH values. The introduction of two further protons occurred at each of the lateral macrocycles, and the resulting tetraprotonated receptor interacted effectively with ATP and ADP, forming 1 : 1 adducts. However, the stability of these adducts relied exclusively on charge– charge interactions and hydrogen bonding because the shape of this receptor precluded the involvement of the central macrocycle and of the aromatic units in the binding. 31P NMR and molecular dynamics suggested that the central and lateral phosphate group of ATP mainly interacted with the protonated lateral rings. The comparison of the properties of 4.43–4.45 (Chart 4.12) offered interest­ ing  insight into the relative contributions of charge–charge interactions,

127

128

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

hydro­gen  bonding, and preorganization in the binding of nucleotides. The receptors showed comparable stepwise stability constants in ATP binding but in different pH domains. These differences could be rationalized on the basis of spectro­scopic studies. The large affinity of 4.45 was mainly due to its high rigidity that compensates its lower protonation degree. The cyclic natures of the substituents in 4.44 and 4.45 facilitated interactions at acidic pH, whereas the  extensive pro­tonation of 4.43 containing the acyclic substituents caused repulsion between the chains and a receptor conformation presumably not optimal for binding. Receptor 4.46 (Chart 4.13), which can be classified as a “double‐aza‐­scorpiand", contained two “scorpiand” subunits linked by a 2,9‐dimethylphenanthroline core [48]. Apart from the amino groups of the lateral macrocycles and the secondary amino groups of the central chains, 4.46 featured two additional aromatic groups for interactions with nucleotides.

N NH

N

N

HN

N NH

NH N

NH

NH

N 4.46

N NH

N

HN

NH

N NH

NH

N

NH N

4.47

Chart 4.13  1

H NMR and molecular dynamics studies showed that the receptor preferred a  compact conformation in its non‐complexed form with the phenanthroline ring stacked with the pyridine rings. This conformation was preserved until the phenanthroline ring was protonated (addition of the seventh proton) after which the receptor opened because of charge repulsions between the positive charges of the ammonium groups. 1H NMR spectroscopy and molecular dynamics cal­ culations indicated that the interactions of [H4(4.46)]4+ or [H5(4.46)]5+ with ATP also caused the disruption of the phenanthroline–pyridine stacking with the concomitant formation of phenanthroline–adenine stacked pairs. Receptor 4.46 and its analog 4.47 with a pyridine spacer between the “scorpiand” fragments furthermore exhibited an interesting selectivity for nucleoside mono­ phosphates [49]. While 4.46 showed a strong preference for AMP over the other nucleoside monophosphates (NMPs) at pH  5, 4.47 had a rare selectivity toward cytosine monophosphate (CMP) with respect to other NMPs, with the CMP/UMP (uridine 5′‐monophosphate) selectivity approaching 2 orders of magnitude. Arranz‐Mascarós et al. designed and studied the interaction of a tren‐derived receptor functionalized with a pyrimidine residue 4.48 toward AMP, ADP, and ATP. Interestingly, 4.48 (Chart 4.14) selectively recognized ADP over ATP because of a subtle combination of binding forces, involving both electrostatic interactions and π‐stacking between the purine and pyrimidine moieties [50].

4.3  Nucleotide Receptors

NH2

O HO H2N

N

N

N

N

N

N H

N

NH2

N

N N

N

N

N

4.49 (Ph2Me6N6)

4.48

N

N

N

N

N

4.50 (Ph2Pip2Me4N8)

Chart 4.14 

A study reported by Bazzicalupi et al. highlighted the importance of electro­ static attraction, hydrogen bonding, and π‐stacking in the interaction of the receptors Ph2Me6N6 (4.49) and Ph2Pip2Me4N8 (4.50)(Chart 4.14) with ADP and ATP [51]. The aromatic moieties, the short ethylene chains connecting the amino groups, the peripheral methyl groups, and the piperazine rings in 4.50 rendered both macrocycles rather rigid. Protonation of these ligands, followed by means of potentiometric titrations and NMR studies, involved distribution of four pro­ tons along the benzylic amino groups. In the case of 4.50, an additional proton resided on the amino group between the piperazine rings. The respective tetra‐ and pentacationic forms [H4(4.49)]4+ and [H5(4.50)]5+ had very similar dimen­ sions as illustrated in Figure 4.6. The close structural relationship of [H4(4.49)]4+ and [H5(4.50)]5+ suggested that they should have similar nucleotide selectivities, while nucleotide affinity of 4.50 was expected to be higher than that of 4.49 because of the higher charge state. Rather surprisingly, while ATP and ADP formed stable complexes with 4.49, 4.50 did not bind these nucleotides under the same conditions. The crystal structures of the [H4(4.49)]4+ and [H5(4.50)]5+ cations provided a possible expla­ nation for this behavior (Figure 4.7). In the crystal structure of [H4(4.49)]4+, the four ammonium groups were arranged in a convergent arrangement, allowing this receptor to form four salt bridges with the phosphate chain of the nucleotide. The crystal structure of [H5(4.50)]5+, on the other hand, showed a divergent orientation of the five ammo­ nium groups. Only two of ammonium groups converged, rendering them suita­ ble anchor points for the phosphate chains of the substrates. Apparently, this was

NH

HN 7.2 Å

N NH

(a)

[H4(4.49)]4+

HN

NH N

N

HN

NH

(b)

N

7.1 Å

HN HN

N

[H5(4.50)]5+

Figure 4.6  Proton distribution in [H4(4.49)]4+ (a) and in [H5(4.50)]5+ (b) and dimensions of the respective cavities.

129

130

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

(a)

(b)

Figure 4.7  Lateral and top views of the crystal structures of [H4(4.49)]4+ (a) and [H5(4.50)]5+ (b). The orientations of the protonated N–H groups are indicated by the arrows.

not enough to sufficiently stabilize the respective complexes. Hence, charge– charge attraction, although the driving force for most ion pairing processes in water, is not always sufficient to stabilize complexes with polyammonium recep­ tors. The reinforcement of salt bridges by hydrogen bonds can be key to achieve efficient anion binding. Bianchi and coworkers reported the crystal structure of a complex formed by the terpyridinophane macrocycle 4.51 (Figure 4.8a) with TTP that illustrated many of the binding modes observed in the interaction of nucleotides with receptors containing aromatic functions [52]. In this structure, three hydrogen bonds between the triphosphate chain of TTP and protonated nitrogen atoms of 4.51 can be seen (Figure  4.8b). Additional interactions encompassed one

O(14) N

N

N HN

NH

HN

NH N

N

N(10)

N

4.51 (a)

(b)

Figure 4.8  Structure of receptor 4.51 (a) and crystal structure of the complex between [(H4(4.51)]4+ and TTP (b).

4.3  Nucleotide Receptors

hydro­gen bond between a carbonyl oxygen of TTP and one ammonium group of 4.51, one CH⋯π interaction involving a TTP carbon atom and the ligand’s pyridine unit, and one O⋯π interaction between the TTP carbonyl O(14) atom and the N(10) in the pyridine ring of 4.51. In combination, these interactions contrib­uted to the stability between [H4(4.51)]4+ and TTP in water, which, with a log Ka of 4.6, was significantly higher than the stability of nucleotide complexes with polyazamacrocycles bearing the same positive charge and having a comparable size but lacking aromatic groups. 4.3.3  Metal Complexes as Nucleotide Receptors Metal ions are involved in the coordination of ATP in a number of metalloen­ zymes. As an example, Figure 4.9 shows that a Zn2+ ion participates in ATP bind­ ing in the active center of the pyridoxal kinase enzyme [53]. Kimura and coworkers demonstrated that the coordination of nucleobases to Zn2+ ions can also be achieved by simple mononuclear receptors derived from the Zn2+ complex of cyclen 4.52 (1,4,7,10‐tetraazacyclododecane, [12]aneN4; Chart 4.15) [54–59]. In their initial work, the authors reported a potentiometric study about the ability of [Zn∙4.52]2+ to interact with the deoxyribonucleotides 2′‐­deoxyadenosine (dA), 2′‐deoxyguanosine (dG), 2′‐deoxycytidine (dC), 2′‐deoxythymidine (dT), uridine (U), and 3′‐azido‐3′‐deoxythymidine (AZT). [Zn∙4.52]2+ had a good selectivity for dT, U, and AZT and the related derivatives Ff (ftorafur, 5‐fluoro‐1‐ (tetrahydro‐2‐furyl)uracil) and riboflavin. Spectroscopic studies indicated that binding occurred through the deprotonated N(3) imide group of the pyrimidine‐ derived substrates and that it did not involve the phosphate group as in the enzyme shown in Figure 4.9. Other nucleosides containing an amino group in place of the carbonyl group of dT, like dG, or ones lacking the imide group did not bind to [Zn∙4.52]2+, which was ascribed to steric hindrance in the case of dG or to the lack of required Lewis basic site. The proposed mode of binding was Ser187A Ser187A

O

K+ 2.1 2.6 2.4

Zn2+

HO

2.5

HO N R 2.2 2.1

Val226A

(a)

Val226A R

O

H

H

Zn2+

O OO OO O– P P P O– O O O – –

N

N

N H

O

Zn403A K+

N N

H

(b)

Figure 4.9  Crystal structure illustrating the binding of ATP within the active center of pyridoxal kinase (a) and schematic representation showing the interactions in this complex (b). Source: Adapted from PDB 1LHR.

131

132

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

HN Zn2+ HN

N Zn2+ NH

NH

HN

[Zn·4.53]2+

N N

N H

N

N H

H N

H N

Zn2+

Zn2+

[Zn·4.54]2+

H N HN

Zn2+ N H

N H [Zn·4.52]2+

H N

H N

H N

N

NH

N N

Zn2+

N

N H

N

Zn2+

NH

N H

[Zn·4.55]2+

Chart 4.15 

confirmed by an X‐ray crystal structure of the mixed complex of [Zn∙4.52]2+ with AZT, which showed the metal ion coordinated in a distorted square pyrami­ dal geometry to the four nitrogen atoms of the macrocycle and to the deproto­ nated imide group (Figure 4.10a). The position of the two pyrimidine carbonyl groups suggested possible direct or indirect hydrogen bonds to the amine groups of cyclen.

(a)

(b)

(c)

Figure 4.10  Crystal structure showing the mixed complexes AZT∙[Zn∙4.52]2+ (a), methylthymine∙[Zn∙4.53]2+ (b), and 2′‐deoxyguanidine∙[Zn∙4.52]2+ (c).

4.3  Nucleotide Receptors

To improve the binding ability of these Zn2+ complexes, the Kimura group also prepared cyclen derivatives with aromatic substituents, for example, receptor [Zn∙4.53]2+, that carried an intercalating acridine moiety [55]. Poten­tiometric as well as UV-vis and NMR spectroscopic studies revealed that the acridine moiety produced enhanced binding with dT derivatives due to the combination of π‐stacking interactions between the acridine and the nucleobase. In addition, the complex was stabilized by the coordination of the deprotonated thymine imide group to the metal center and by hydrogen bonding between the nucle­ obase carbonyl groups and amino groups of the cyclen ring. The crystal struc­ ture of the mixed complex methylthymine∙[Zn∙4.53]2+ (Figure 4.10b) revealed a square pyramidal coordination geometry of the metal. One of the carbonyl oxy­ gen atoms formed a direct hydrogen bond with a cyclen N–H proton, while the other one bound to the opposite N–H amino group indirectly through a water molecule. In contrast to [Zn∙4.52]2+, [Zn∙4.53]2+ proved to be also able to interact with dG. The crystal structure of the corresponding complex displayed a binding mode involving Zn2+ coordinated in a distorted square pyramidal fashion to the four nitrogen atoms of the cyclen ring and to N(7) of the purine ring (Figure 4.10c). Stacking interactions and hydrogen bonding between oxygen O(6) of the purine ring and an NH group of cyclen completed the interactions. Solution studies showed, however, that [Zn∙4.53]2+ was still selective for thymidine. In order to obtain receptors for the recognition of NMP, NDP, and NTP, the Kimura group also developed the dimetallic and trimetallic Zn2+‐containing systems [Zn2∙4.54]4+ and [Zn3∙4.55]6+ (Chart 4.15). Efficient recognition of thy­ midine and uridine nucleotides such as 3′‐dTMP, 5′‐dTMP, 2′‐UMP, 3′‐UMP, 5′‐UMP, 5’-dTDP (2’-deoxythymidine 5’‐diphosphate), 5′‐dTTP (2′‐deoxythymidine 5′‐triphosphate), 3′‐azido‐3′‐deoxythymidine 5′‐monophosphate (AZTMP), or 3’-azido‐3′‐deoxythymidine 5′‐diphosphate (AZTDP) was, for example, achieved with [Zn2∙4.54]4+ and the analogous receptor containing a m‐xylylene junction between the cyclen units. Moreover, the trinucleotide d(TpTpT) was found to be selectively bound by [Zn3∙4.55]6+. Related substituted cyclen complexes were also used for the optical detection of nucleotides in water (see Section 4.4). Lin and coworkers developed receptors comprising phenanthroline units with appended amine or dipicolylamine chains [60, 61]. The Zn2+ complexes of these ligands were shown to bind more strongly to nucleotides in water than the protonated metal‐free receptors, which is commonly observed for such sys­ tems. An example for the recognition of nucleotides through the formation of a mixed complex of AMP with a polyamine‐based acyclic receptor (4,7,10,13‐ tetrazahexadecane‐1,16‐diamine) and Cu2+ has also been reported [62]. The recognition of adenine nucleotides with the Zn2+ complex of receptor 4.56 was demonstrated by Roelens and coworkers [63] This receptor combined an imi­ nodiacetate fragment as the Zn2+ binding site with a polyamine macro­cycle con­ taining two trans‐1,2‐aminocyclohexane units and a pyrrole ring. Potentiometric and NMR studies titrations revealed large affinities of [Zn∙H34.56]5+ for ATP and ADP (ADP: log Ka  =  5.15, ATP: log Ka  =  6.93). The authors concluded that the bridging of the two anion binding sites of [Zn∙H34.56]5+ by the diphosphate and triphosphate chains of the substrates was key for the observed binding efficiency.

133

134

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

They also emphasized that this system displayed the highest binding affinity for triphosphates reported to date for a Zn2+ complex in water (Chart 4.16).

N H

HN – OOC – OOC

HN 4.56

NH

N

NH

Chart 4.16 

The last example in this section should illustrate nucleotide binding by certain metallocages reported by the Fujita group [64]. Here, the metal ions were not involved in the actual recognition event but helped to stabilize the receptor. The study addressed the stabilization of complementary base pair formation of mon­ onucleotides and dinucleotides in water by the screening of the base pairs from the solvent in the interior of suitable cages. In this context, 4.57 (Figure 4.11a) provided an appropriate environment for the incorporation of the AMP-UMP anti‐Hoogsteen base pair in water. When the analogous expanded cage 4.58 (Figure  4.11b) was used, the stabilization of dinucleotide thymidyl‐(3′,5′)‐2′‐ deoxyadenosine pair was achieved. Complex formation in solution was confirmed by NMR spectroscopy, while crystal structures provided structural information about the complexes in the solid state (Figure 4.11c). 4.3.4  Catalytic Aspects An intriguing property of some polyammonium‐based receptors is their ability  to mimic the catalytic properties of ATPases or kinases. Among these receptors, the 24‐membered macrocycle 4.25 (Chart 4.7) exhibited the high­ est  enhancements of ATP cleavage to produce ADP and inorganic phosphate [35, 65–69]. In contrast to other polyammonium receptors, ATP cleavage mediated by 4.25 was not only observed at acidic but also at neutral pH values at which the spontaneous hydrolysis of ATP was enhanced by a factor of 100. Monitoring the course of the reaction with 31P NMR spectroscopy revealed the formation of a transient phosphorylated species, which later was found to represent a derivative of 4.25 containing a phosphoramidate group. The pro­ posed mechanism of ATP hydrolysis is depicted in Figure 4.12. Figure 4.12 shows that one of the central amines of 4.25 is not protonated at neutral pH and performs the nucleophilic attack on the terminal phosphorus atom of ATP to yield the phosphoramidate intermediate (step B). The electro­ static and hydrogen bonding interactions of the triphosphate group of the

4.3  Nucleotide Receptors N

N 4.57

Pt N

N N N

N N

(a)

N

N

N Pt

Pt = N

4.58

Pt N

(b)

N N Pt

N H2 N

N Pt N

Pt N

N N

N

N O

N

O

N

N Pt

O

HO

OH

O

O – O P O O –

OH

O HO O

OH

N

H N

Pt

N N

N

N

HO

HO

H N H

H2 N

N

N

O O P O – O

Pt N

N N

N N Pt

–

N

N H

N N

O N

O O P O

N O H N HO O H H N H O OH O N – N O O P O N H N N O N O O HO

–

OH

OH

(c)

Figure 4.11  Structures of the metallocages 4.57 (a) and 4.58 (b) together with the respectively bound base pairs and front and top view of the crystal structure of 4.57 with the AMP–UMP base pair included (c).

nucleotide with the protonated amino groups of 4.25 place the γ‐phosphorus atom and the non‐protonated amine in a suitable position to facilitate this attack (step A). The final steps of the catalytic cycle are the dissociation of the ADP complex (step C) and the hydrolysis of the phosphoramidate (step D), not neces­ sarily in this order. Studies of ATP cleavage conducted with the series of azacycloalkanes 4.10 (Chart 4.4) and 4.15–4.20 (Chart 4.5) revealed that the size of the macrocycle was key for activity [28]. The rates of hydrolysis were dependent on ATP con­ centration and pH. The studies at pH 3 and 20 °C with [21]aneN7 (4.15) had to

135

136

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials H2PO4–

H

N H

D

H

H2O

H NH N

O O

H H H N HN H N H

ATP A

O O O O P P Adenosine P O O O O O H H H N H H N NH HN O H H N N O H H O

HO

O P O N H NH H N H

O O

H H H N HN H N H

O O O P P Adenosine HO O O O O P O H H N H H N NH HN O H H N N O H H O

ADP

C

B

Figure 4.12  Catalytic cycle of the ATP hydrolysis in the presence of 4.25 at neutral pH.

be carried out at an initial ATP concentration of 10−5 M at 20 °C. At higher tem­ peratures and/or ATP concentrations, the reaction rates were too fast to allow monitoring by high‐performance liquid chromatography (HPLC) or NMR. All the other macrocycles showed considerably lower rate accelerations with 4.16 being the second most active one. The four larger macrocycles 4.17–4.20 or the smaller 4.10 were clearly less active. At pH 7, the reactions were slower than at pH 3, indicating a reduced participation of general acid catalysis. The highest activity of 4.15, particularly at low pH values, was attributed to several factors. First, the magnitude of the interaction did not seem to be key since the larger macrocycles that formed stronger complexes were less effec­ tive. Second, in light of the formation of the phosphoramidate intermediate, which was experimentally observed also for 4.15, the nucleophilicity of the amino groups should be critical. Factors that favorably influence nucleophi­ licity are (i) a small overall positive charge at a given pH, (ii) the hydrophobic­ ity of the environment where the reaction occurs, and (iii) the presence of Lewis basic sites. As the degree of protonation of 4.15 was lower at a given pH than that of the larger macrocycles, its higher activity was plausible. However, the same reasoning should cause 4.10 to be even more active, but this was not the case. Therefore, also other factors that affect the electronic and stereochemical matching between receptor and substrate had to play a role. A comparison of the geometrical features of 4.10, 4.15, and 4.25 is made in Table 4.1.

4.3  Nucleotide Receptors

Table 4.1  Dimensions of the major and minor axes for the average ellipsoid of polyammonium macrocycles 4.10, 4.15, and 4.25 determined by X‐ray diffraction analyses. Receptor

Major axis(Å)

Minor axis(Å)

H4(4.10)4+

7.682

6.212

H6(4.10)6+

7.647

6.236

4+

H4(4.15)

7.653

6.725

H6(4.25)6+

10.005

6.714

Table 4.1 shows that the most effective receptors 4.15 and 4.25 (4.15 in the tetraprotonated form H4(4.15)4+ and 4.25 in the hexaprotonated one H6(4.25)6+) had similar dimensions along the minor axes. This distance seemed to be opti­ mal for orienting the terminal phosphate group of ATP close to an unprotonated amino group [70], providing a rationale for the similar activities of 4.15 and 4.25 and suggesting geometric parameters to play a decisive role for catalytic activity. The relevance of the size of the macrocycles and their charge density on the rate of ATP hydrolysis was later tested by using the N‐methylated derivatives 4.22 (Chart 4.6), 4.59, and 4.60 (Chart 4.17) [71, 72].

N NH

HN

NH

N H

N N

NH

HN

NH

N

H N

4.59

NH

4.60

HN

NH

HN

HN

NH

HN

NH

HN

NH

HN

NH OH HN NH

HN

NH

HN

HN

HN

4.62

HN

4.61

NH

NH NH

N

NH

4.63

4.64

Chart 4.17 

While dimethylated 4.59 was a less efficient catalyst that the non‐methylated 4.10 (Chart 4.4), the tetramethylated 4.22 (Chart 4.6) produced a significant rate

137

138

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

enhancement. The slightly larger trimethylated 4.60 was again less efficient. These effects were attributed to the changes in the optimal conformations of the macrocycles brought about by the functionalization, potentially in combination with solvation effects. Additional evidence supporting the decisive role played by cavity size in ATP hydrolysis was provided by the series of cyclophane receptors 4.61–4.63 (Chart 4.17) [73]. In this series, the ortho‐ and para‐disubstituted subunits 4.61 and 4.63, comprising 20‐ and 22‐membered rings, performed significantly poorer than the 21‐membered 4.62 with the meta‐substituted phenylene unit. The lack of activity was particularly noticeable in the case of the ortho derivative 4.61. 31P NMR studies revealed that 4.62 produced rate enhancements compa­ rable with those of 4.25 and slightly lower than those of 4.15. Interestingly, the reaction mediated by 4.62 was not only efficient but also very specific since it stopped at the formation of ADP and did not proceed significantly further. Another macrocycle with a 21‐membered ring was 4.64 (mPh22222; Chart 4.17), differing from 4.62 by the additional phenolic hydroxy group in the aromatic subunit. 4.64 produced rate accelerations of ATP hydrolysis from acidic pH values to ca. 7, although not as strongly as 4.62. At pH 7 and above, the reac­ tion significantly slowed down, which was attributed to the deprotonation of the OH group that not only reduced the affinity of the receptor for ATP due to charge repulsion but also had a detrimental effect on the conformation of the receptor, further reducing ATP affinity. O‐Bisdien 4.25 (Chart 4.7) exhibited also other enzyme‐like activities, for example, the  ability to mediate the transfer of a phosphoryl group to other substrates. Importantly, this reaction required the presence of metal ions, reminiscent of the behavior of ATP phosphotransferases, hydrolases, and synthetases, which also require metal cofactors. In the case of 4.25, the reaction proceeded via the phosphoramidate that was also formed during ATP hydrolysis [74]. The phosphate group of this intermediate was then transferred to an orthophosphate anion, yielding pyrophosphate. The course of this reaction was highly dependent on the nature of the metal ion [75, 76]. This phosphoryl transfer was discovered when investigating the effect of metal ions on the ATP hydrolysis promoted by 4.25. While the addition of Ca2+ and Mg2+ favorably affected the formation of the intermediate phosphoramidate, only Ca2+ caused a significant acceleration of ATP hydrolysis, almost doubling the first‐order rate constant found for free 4.25 under the same experimental conditions. Ln3+ also led to a considerable rate enhancement, whereas no appar­ ent effect was observed in the presence Mg2+. The presence of Zn2+ or Cd2+ decreased the rate of hydrolysis. The most striking finding in these investigations was the appearance of pyrophosphate under the influence of Mg2+, Ca2+, or Ln3+ at pH 4.5, which was not formed in the absence of these metals. This result dem­ onstrated the transfer of the phosphoryl group from the intermediate phospho­ ramidate onto an orthophosphate anion released during ATP hydrolysis. Hosseini and Lehn then showed that this process allowed 4.25 to catalyze the synthesis of ATP from acetyl phosphate and ADP in dilute aqueous solution at neutral pH in the presence of Mg2+ as promoter [77]. Mertes and coworkers demonstrated that

4.3  Nucleotide Receptors

4.25 also activated formate in the presence of ATP and Ca2+ or Mg2+, yielding a macrocycle formylated at the central nitrogen atom as the product [78]. The presence of a non‐coordinated metal ion can play a decisive role in the acceleration of ATP hydrolysis. Evidence for this was obtained by using receptor 4.65 (Terpy2222; Chart 4.18) [79]. This ligand formed both mono‐ and binuclear Zn2+ complexes, depending on the metal–ligand ratio and the pH. The first Zn2+ ion was bound to the terpyridine moiety, while the second one occupied the opposite side of the macrocycle at pH values at which the amino groups were not protonated. Both complexes interacted with ATP but did not show any pro­ nounced effect on ATP hydrolysis. The mononuclear tetraprotonated complex [Zn∙H44.65]6+ together with ATP and excess free Zn2+, however, gave rise to fast ATP hydrolysis to produce ADP and phosphate. The analysis of the time depend­ ence of the 31P NMR spectra recorded at different pH values showed that tri‐ and pentaprotonated forms of the same receptor were inactive. The hydrolytic process proceeded through the formation of a phosphoramidate intermediate, which was then rapidly hydrolyzed to hydrogen phosphate. The hydrolysis rate (kobs = 3.2 × 10−2 min−1 at pH 4) was among the highest observed for ATP dephos­ phorylation promoted by polyammonium receptors. H2N

H2N

H2N

NH2 N

N N

N H

HN

N

HN N

H N

HN

H2N

NH2

N

N

N

H2N

N

N N

NH2

N N

H2N N

H2N NH2

NH2 N

N 4.65 (Terpy2222)

NH2

N

H2N

NH2

NH2

4.66

Chart 4.18 

The fact that ATP cleavage took place only in the presence of excess Zn2+ ions, following second‐order kinetics, suggested the stabilization of the transition state by this metal ion, likely through coordination of the metal by amino groups of 4.65 and the γ‐phosphate group of ATP. Alternatively, also the phosphorami­ date intermediate could be stabilized by coordination to the metal, accounting for the observed relatively high amount of this intermediate accumulating during the reaction. This system thus represented a unique example for an ATP dephos­ phorylation process promoted by the simultaneous action of a metal complex, which mainly served to anchor the anionic substrate, and a second metal as a

139

140

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

cofactor that assisted in the transfer of a phosphoryl group from ATP to an amino group of the receptor. Bianchi and coworkers showed that the G3 poly(ethyleneimine)‐based den­ drimer 4.66 (Chart 4.18) enhanced or inhibited the dephosphorylation of ATP, depending on the solution pH [80]. Potentiometric studies provided evidence for the formation of adducts between 4.66 and nucleotides over a broad pH range of 2.5–11.0. In alkaline solutions, the dephosphorylation rate was increased by a factor of 6, while the ATP cleavage was inhibited around 30% at pH 3. The rate enhancement at high pH was attributed to the presence of free primary amino groups in 4.66 that engaged in the hydrolytic cleavage of ATP by mediating transphosphorylation. By contrast, all peripheral primary amino groups were protonated at acidic pH, allowing 4.66 to strongly bind ATP in the inner region of the dendrimer but preventing it from forming phosphoramidates. Protection of ATP in this complex thus even caused a reduction of the rate of spontaneous ATP hydrolysis.

4.4 ­Nucleotide Sensing 4.4.1  General Aspects Because of the biological importance of nucleotides, efforts were also directed at developing methods for their optical detection in aqueous solution. So far, ATP has been mostly considered as substrate in this context because of its high rele­ vance in cellular processes. The respective optical chemosensors can either have the signaling unit incorporated as an integral part into the receptor or have it bound to the receptor non‐covalently. In the latter case, the displacement of the chromophoric unit from the receptor leads to a change in the optical proper­ ties [81, 82]. For more details about structural and functional aspects of optical chemosensors, the reader is referred to Chapter 12. 4.4.2  UV–vis Sensing The Hayashita group reported the γ‐cyclodextrin‐derived receptor 4.67 capa­ ble of recognizing ATP with high selectivity over other nucleotides [83]. This receptor contained a dipicolylamine (dpa)‐azobenzene unit complexed to Cu2+ (Figure 4.13). The interaction with ATP caused a characteristic blue shift of the UV–vis absorption band that could be used to discriminate ATP over other polyphosphate anions. The binding mode likely involved the incorporation of the adenine moiety in the cyclodextrin cavity, while the triphosphate group was coordinated to the Cu2+‐dpa complex. The polythiophene‐derived copolymer 4.68 bearing additional anthracene moieties was designed to self‐assemble upon mixing with ATP [84]. This aggre­ gation was driven by π‐stacking of the anthracene ring and the nucleobase, a process associated with a clearly visible color change from yellow to deep violet. The observed optical response derived from the change of the random coil con­ formation of the polymer to an ordered conformation after ATP addition. 4.68 could thus be used to discriminate by naked‐eye ATP from various other

4.4  Nucleotide Sensing

N

N 4.67 N

O

N O

N

N

N Cu2+

NH ATP N γ-CD

N

O N HO H N N

O N

OH O

N Cu2+

O O O P O O O P P O O O O

H2N

Figure 4.13  Structure of ATP chemosensor 4.67 and schematic representation of its interactions with ATP.

nucleo­tides in solution (Figure 4.14). In addition to the colorimetric response, also the fluorescence emission of 4.68 was quenched upon addition of ATP, whereas quenching was less effective for other nucleotides. The detection limit of this system for ATP amounted to 2.3 nM.

O

Cl

N

NH

O S

S

m

m/n = 10 : 1.5

n

4.68

(a)

ATP

O

ADP

AMP

UTP

UMP

PO4

HPO4

H2PO4

Cl

4.68

(b)

Figure 4.14  Schematic structure of 4.68 (a) and color changes associated with the addition of various anions to aqueous solutions of this polymer (b). Source: (b) Adapted with permission from Ref. [84]. Copyright 2015 Royal Society of Chemistry.

141

142

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

A system derived from self‐assembled diacetylenes that were stabilized by  photopolymerization allowed for the colorimetric sensing of ATP and PPi (Chart 4.19) [85]. The respective polydiacetylene vesicles 4.69 were decorated with metal Zn2+–cyclen complexes. They changed their color from blue to red upon addition of ATP or PPi, which was attributed to a conformational change of the polydiacetylene backbone. Other anions failed to induce a spectral change. These vesicles could also be used to detect ATP or PPi by using fluorescence spectroscopy. In buffered solution, the vesicles showed a weak emission band the intensity of which significantly increased when 4.69 was treated with ATP or PPi, rendering this system a dual colorimetric and fluorometric optical sensor like 4.68. HN

H N Zn

N O

OH

(CH2)7

O

O

OH

(CH2)7

O

(CH2)7

Pyrocatechol violet (PV) HO

2+

NH

HO

(CH2)8

(CH2)8

4.69

OH SO3

O

OH (CH2)7

N Zn2+ N

(CH2)8

O

O

Zn2+ N

N

N

N

(CH2)8 [Zn2·4.70]

Chart 4.19 

An alternative approach to optically detect anions involves the use of indicator displacement assays (see Chapter  12). This approach was used in conjunction with the dizinc complex [Zn2∙4.70]3+ (Chart 4.19) [86]. This receptor formed a complex with pyrocatechol violet (PV), characterized by an absorbance band in the UV–vis spectrum at 620 nm. Upon addition of AMP, the dye was displaced from [Zn2∙4.70]3+, causing the appearance of the band of free PV in the UV–vis spectrum at 440 nm. The system allowed the detection of AMP in real-time and the screening of phosphodiesterase inhibitors. 4.4.3  Fluorescence Sensing Among the optical methods for detecting nucleotide anions, fluorescence has particular advantages such as high sensitivity, specificity, and the potential to use it in cellular systems. Early examples of fluorescent probes for nucleotide sensing were the anthracenyl‐containing linear polyamines shown in Chart 4.9 [29, 40]. They exhibited quenching at acidic pH upon the interaction with ATP. The comparison of the steady‐state fluorescence titration curves with the pH‐ dependent species distributions indicated that quenching was due to protona­ tion of the adenine N(3) atom. This protonation rendered the adenine ring

4.4  Nucleotide Sensing

electron deficient, permitting a more efficient stacking with the naphthalene or anthracene rings of the receptor. Moreover, steady‐state fluorescence and time‐ correlated single‐photon counting analysis of a receptor comprising a naphtha­ lene and an anthracene fluorophore at both ends of a linear tetraamine showed that the energy transfer between the naphthalene and anthracene chromo­ phores was not prevented by the presence of ATP. In order to generate func­ tional materials for ATP sensing, linear polyamines end‐functionalized with fluorophores (naphthalene or indole) were grafted onto the surface of boehmite nanoparticles. Fluorescence emission studies showed that these materials retained the sensing behavior in aqueous solution of the non‐immobilized ­polyamine‐based analogs [87, 88]. The peptide‐based receptor 4.71 containing a naphthalimide fluorophore and two terminal guanidiniocarbonyl pyrrole groups allowed ATP sensing in vitro and in cell cultures (Chart 4.20) [89]. The receptor showed a weak florescence emission in solution, but ATP binding produced a pronounced enhancement of the emission, which was attributed to the folding of 4.71 around the nucleotide. O N

O

O

N

H3N

O

H N

O N H

O

H N

O N H

O NH

NH HN H2N

HN HN

NH2 O

NH3

NH O O

HN HN

O NH2

4.71

O H2N

NH NH2

Chart 4.20 

The scorpion‐like polyamine receptors 4.72a–d (Chart 4.21) formed stable complexes with nucleotides in water stabilized by electrostatic, π‐stacking, and hydrogen bonding interactions [90]. The studies demonstrated that the anthra­ cene‐based receptors exhibited greater binding selectivity for GTP than for ATP and UTP, while the pyridine‐based receptors exhibited a higher affinity for ATP. The interaction of the anthracene‐based receptors 4.72a and 4.72b with ATP resulted in an increase of the fluorescence emission, while their interaction

143

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

with GTP quenched the fluorescence. This optical response allowed the dis­ crimination of both nucleotides.

H N

NH NHR N

N

R=

4.72a

NH

4.72b N

H N

N

H N

4.72c

4.72d

Chart 4.21 

An alternative approach for nucleotide sensing relied on the formation or dis­ sociation of highly emitting pyrene excimers. An example for a receptor emplo­ ying the first mechanism is the pincer 4.73 (Figure 4.15a) [91]. Both of its pyrene moieties were π‐stacked in the absence of nucleotides in solution, giving rise to an intensive excimer band in the fluorescence spectrum at 475 nm. For compari­ son, the emission band of monomeric pyrene resided at 375 nm (Figure 4.15b). (a)

O N

N N N

N

GTP

N N

4 Br

– ATP

4.73

N

N

O O O – P P P O O O O O O O HO – – – N

N N

NH2

N O N HO

N

N N

N

N O O

NH2 OH OH

(b)

N N

N

O O – P P P O O O O O O O– – – N N N N

N N

NH

N

N N

Fluorescence intensity (a.u.)

144

750 600 450 300 150 0 350

400

450 500 Wavelength (nm)

550

600

Figure 4.15  Proposed binding mode of pincer‐type receptor 4.73 with ATP and GTP (a) and fluorescence titration of 4.73 with ATP in 20 mM aqueous 2‐[4‐(2‐hydroxyethyl)piperazin‐1‐yl] ethanesulfonic acid (HEPES) buffer (pH 7.4) (b). Source: (b) Adapted with permission from Ref. [91]. Copyright 2009 American Chemical Society.

4.4  Nucleotide Sensing

Upon addition of ATP to an aqueous solution of 4.73, the excimer band decreased in favor of the monomer emission band. This effect was attributed to the forma­ tion of pyrene–adenine–pyrene sandwiches and the simultaneous excimer dis­ sociation. Other nucleoside triphosphates such as GTP, CTP, UTP, and TTP interacted only from the outside with the pyrene–pyrene excimer. These nucleo­ tides also quenched the excimer emission with an efficiency trend ATP ~ GTP > TTP ~ UTP > CTP, but only ATP induced an enhancement of the emission band of monomeric pyrene. AMP and ADP produced small effects, enabling selective ATP sensing. This probe was furthermore used in ATP staining experiments and to monitor ATP hydrolysis by apyrase. The opposite strategy, namely, the nucleotide‐induced formation of pyrene excimers, was the basis of the sensing mechanism of receptor [Zn∙4.74]2+ (Figure 4.16) [92, 93]. The Zn2+‐cyclen moiety of 4.74 interacted with thymine‐ containing nucleotides as described in Section 4.3.3 for [Zn∙4.52]2+ (Chart 4.15), while the phosphate groups hydrogen bonded to the pendant protonated amino groups. As a consequence, the pyrene residues of the receptor came in close contact upon binding of thymidine phosphates, resulting in a strong excimer emission. The nucleotide‐induced formation of emissive pyrene excimers has further­ more been achieved by using an intermolecular strategy. Here, the pyrene units  of two molecules of receptor [Zn∙4.75]2+ (Figure  4.17), which contained a  Zn2+–dipicolylamine moiety as recognition motif, were brought together by the formation of ternary complexes in which the nucleotide phosphate groups coordinated to the metal centers [94]. Interestingly, the receptor only exhibited a  large enhancement of the excimer band when interacting with ATP. In con­ trast,  ADP hardly induced an excimer emission, which was attributed to the incorporation of the nucleobase between the two pyrene units in this complex. AMP did not induce significant changes in the fluorescence spectrum of [Zn∙4.75]2+.

λex = 350 nm

λem = 375 nm

λem = 470 nm

λex = 350 nm NH O N

HN

Zn2+ O

N

NH

[Zn·4.74]2+

H H N

O

O

N TNP

pH = 7.4

2+

HN Zn

OH HO

O

NH

N

O N H

O

N

N

O

O O

O

O O P O

N H H

Figure 4.16  Proposed binding mode of thymidine phosphates to [Zn∙4.74]2+.

n

O n = 1–3

145

146

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials Zn2+ N [Zn·4.75]2+

No interaction

ADP

N

O

N N

O H

ATP

N

N H O

N

N

N

H2N HN

Zn2+ O – P P O H O O O – – O

O

N

O

AMP

O

N N

H N

N H

O

N N

O

Zn2+

P

O

OH N

HO

O O P P O O OO OO – – – O H Zn2+ N N N

–

Zn2+ N

N

N

O

O

N

N NH2 N

N

Figure 4.17  Schematic representation of the binding of AMP, ADP, and ATP to the receptor [Zn∙4.75]2+.

Sasaki and coworkers designed the receptor [Zn∙4.76]2+ (Figure  4.18a) for 8‐oxo-2′‐deoxyguanosine (8‐oxo-dG) whose oxidized guanine residue induces transversion mutations during DNA replication [95]. The 9‐hydroxy‐1,3‐­ diazaphenoxazine‐2‐one moiety in [Zn∙4.76]2+ served as a fluorescent reporter HN

NH Zn2+

HN

N

O

N N

O

(a)

N H

O O

N H

[Zn·4.76]2+

N H H N

N

HN Zn2+ N N H

(b)

O

O

O

H

O H

N H

O

O

N

H N

N

N H

O O O P O O P O P O O O O

O

O N

N O

OH

Figure 4.18  Structure of receptor [Zn∙4.76]2+ (a) and schematic representation of its interaction with 8‐oxo-dGTP (b).

4.5  Soft Materials Incorporating Nucleotides, Nucleosides, and Nucleobases

and recognition unit for the nucleobase, whereas the Zn2+–cyclen unit tightly bound to the nucleotide’s phosphate groups (Figure 4.18b). The fluorescence of  [Zn∙4.76]2+ was efficiently and selectively quenched by 8‐oxo-dGTP (2′‐­deoxyguanosine 5′‐triphosphate), allowing this compound to be used as a chemosensor for this noncanonical nucleotide.

4.5 ­Soft Materials Incorporating Nucleotides, Nucleosides, and Nucleobases Nucleotides and their derivatives are particularly useful for the assembly of soft materials because of their characteristic structural features. The built‐in nucle­ obases feature characteristic hydrogen bond donor and acceptor motifs and large aromatic surfaces that mediate hydrophobic and π‐stacking interactions, whereas the sugar units and phosphate groups can engage in hydrogen bonding and/or electrostatic interactions. The structural variability of these systems can even be extended by also considering synthetic nucleotides. The corresponding assem­ blies are attractive for applications in drug delivery, catalysis, diagnostics, or tis­ sue regeneration. In this section, examples of materials in which nucleotides play an important role are presented. The most extensively studied materials in this context are polymers and gels, which are assembled from suitable building blocks whose intermolecular inter­ actions lead to supramolecular polymerization or network formation (see also Chapter 8) [96–98]. Oda and coworkers studied the ability of the complementary nucleotides UMP and AMP to gelate in water upon the addition of bis‐ or mono‐ quaternary ammonium ions (Figure 4.19a) [99]. The two nucleobases assembled in the respective hydrogel by triple hydrogen bond formation, while the nega­ tively charge phosphate groups were neutralized by the quaternary ammonium ions. Other cations such as metal ions could also be used to initiate the gela­ tion process. Zn2+ ions, for example, selectively induced gel formation of AMP, whereas neither adenine alone nor adenosine or ATP produced a gel under the same conditions (Figure 4.19b) [100]. The binding of guest species to supramolecular recognition sites incorpo­ rated into elastic polymers can produce dimensional changes of macroscopic dimensions that can be used for different applications such as in actuators, sensors, process control, and especially drug release systems. In this context, Schneider and coworkers synthesized suitable precursors by reaction of poly(methyl methylacrylate) with diethylenetriamine and various amounts of long‐chain alkylamines [101, 102]. Transparent thin films of the elastic poly­ mers were preconditioned in buffer solution and cut into small pieces. Reversible changes in the dimensions as large as 126% were observed when adding AMP. UV–vis spectroscopy permitted monitoring the depletion of the AMP in the supernatant solution and its absorption by the polymer [101, 102]. Guanosine‐based materials have been known for a long time, but they found widespread use only after the discovery of G‐quadruplexes. G‐quadruplexes are formed by the stacking of two or more G‐quartets, which are composed of four

147

148

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials HO N

H2n+1Cn N

N

O O P O O

H2n+1Cn N

N

O

H H

OH

O O P O O

(a)

HO

O

P

O

O

(b)

HO

N N N

OH O O P O O

O

N

N CnH2n+1

O

N

HO

N

N

O

NH2

N O

O

O

P

N

O

O

O

O

NH2 HO

P

N N

N NH2

O O

Zn2+ O

HO

N

HO

Zn2+

O

Zn2+ O

HO

P

H N

N O

O

H

N CnH2n+1

OH

HO

O

N

N

HO

O

N

O

N

O

HO

H

O H

HO

N N

O O P O O

O

N

N

O

OH

N N N

NH2 N

Figure 4.19  Schematic representation of hydrogels formed from AMP, UMP, and bis‐ quaternary ammonium ions (a) and AMP and Zn2+ ions (b).

guanine bases held together by hydrogen bonding and templation by alkali metal ions. Different acyclic or cyclic polyamines were shown to significantly increase the tendency of guanosine 5′‐monophosphate (GMP) to form hydro­ gels as proven by rheological studies and scanning electron microscopy. The results from wide‐angle X‐ray scattering suggested that gel formation involved the assembly of G‐quartets [103]. Similarly, GMP also formed hydrogels when mixed with lanthanide ions within the pH region of 2–6. This gel comprised a three‐dimensional network of intertwining one‐dimensional nanofibers formed from stacked G‐quartets that surrounded the lanthanide ions (Figure  4.20). Precipitation occurred above pH 6 because of a change in the degree of proto­ nation of GMP [104]. Borate anions formed stable hydrogels with guanosine in the presence of potassium ions [105]. In this case, the formation of the borate–sugar ester bonds facilitated the potassium‐templated assembly of the G‐quartets (Figure 4.21). Hybrid organic–inorganic gels were constructed based on a polyhedral orga­ nosilicon compound (polyhedral oligomeric silsesquioxane, POSS) decorated with ammonium groups to mediate water solubility, naphthyridines as reporter

4.5  Soft Materials Incorporating Nucleotides, Nucleosides, and Nucleobases O

+

= Ln3+

N

O O P O O

O HO

NH

N

NH2 =

N

OH

pH > 6

pH 2–6

R N

H

N

N

H

O

H N H

N H

N

N O H N

O N N

N H

H N H

O

H

N N

H

O O P O O

N R

R= O HO

Nanoparticles

R N

N

Ln3+

N N R

H N

OH

G-quartet-based nanofibers

Figure 4.20  Mechanism of the lanthanide‐induced formation of G‐quartets and hydrogels from GMP. Source: Adapted with permission from Ref. [102]. Copyright 2018 Springer Nature.

RO OH RO RO

O B

O O

N

H

N

H N H

N

N

O

H

N

O O RO

B

O

N

N H N

O

N

O N

OH

H H

K+

N

O

N O

H

H N

OR

O

N

N

N

HO

B

O

H N H

H

N

N

O

O

B O

OR OR

Hydrogel

OH

OR

Figure 4.21  Structure of the borate‐containing G‐quartet and schematic representation of the corresponding hydrogel. Source: Adapted with permission from Ref. [103]. Copyright 2014 American Chemical Society.

149

150

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

groups, and nucleobases as recognition units [106]. The hydrophobic environ­ ment generated by the POSS unit was shown to enhance the interactions of the naphthyridine moieties and the nucleoside. These gels selectively interacted with the nucleotides complementary to the nucleobases attached to the core units, as evidenced from the quenching of the naphthyridine fluorescence. This effect was strongest for triphosphates and decreased in the order NTP > NDP > NMP > N.

4.6 ­Biomedical Applications Nucleosides are of key importance in living systems. They are usually located in pools in prokaryotic and eukaryotic organisms where their concentrations are closely regulated. The dysregulation of these pools during DNA/RNA replication and repair is linked to pathological conditions and, hence, to disorders and dis­ eases. Thus, the quantification in real-time of cellular nucleoside polyphosphates is essential to assist the understanding of these pathological phenomena and the modes of action of drugs. An elegant fluorescence‐based system to detect ATP in real-time was the  probe 4.77 containing rhodamine as fluorophore, a linker with a basic amino group, and a triphenylphosphonium ion as mitochondrial targeting group (Chart 4.22) [107]. This probe showed high sensitivity toward ATP with an 81‐fold fluorescence enhancement and a detection range of 0.1−10 mM. The molecule interacted with ATP by π‐stacking between the aromatic units of 4.77 and the nucleobase. In addition, electrostatic interactions caused an addi­ tional stabilization of the complex. 4.77 efficiently accumulated in mitochon­ dria and thus allowed real‐time detection of mitochondrial ATP concentration changes. Another strategy to visualize ATP generated in the mitochondria focused on the formation of covalent bonds between a boronic acid and the ribose OH groups of the target nucleotide. In this context, the probes 4.78a–c were developed, each containing a boronic acid group attached to a rhodamine moiety that was positively charged at physiological pH [108]. The ring‐closed structures of 4.78a–c were not fluorescent, but the interactions with ATP, involving the combination of boronic ester formation, π‐stacking between the xanthene and the adenine moiety, and electrostatic interactions between the amino and the triphosphate groups (Figure 4.22), induced ring opening and a strong fluorescence. The molecule selectively and rapidly responded to ATP at intracellular concentrations and localized in the mitochondria, exhib­ iting good biocompatibility and membrane penetration. Apart from nucleoside sensing, materials made of nucleotides are of great interest in tissue engineering and drug delivery because of their unique prop­ erties and easy preparation. In this context, G‐quartet hydrogels were pre­ pared, comprising a mixture of nucleoside derivatives. Specifically, a hybrid gel containing guanosine derivatives and the antiviral agent acyclovir was prepared under the influence of borate ions as described previously (see Section  4.5) [109]. The incorporation of 5′‐deoxy‐5′‐iodoguanosine (5′‐IG) into the gel network allowed triggering the release of the drug because the

4.7  Challenges and Future Perspectives O N N

O

H N

N H

P

N

O

4.77 O N HO B N

O N HO

N

O

B OH

OH N

4.78a

N

O 4.78b OH B OH

O N N

O

N

4.78c

Chart 4.22 

H N

B(OH)2

O N

O

ATP N

N

O

N

N

N H2N

O

N

O B O O N

N

OP3O74–

Figure 4.22  Proposed mechanism of the interactions of probes 4.78a–c with ATP.

intramolecular cyclization of 5′‐IG caused the destabilization of the G‐­quartet and the simultaneous dissociation of the gel (Figure 4.23).

4.7 ­Challenges and Future Perspectives Nucleotides were among the first classes of natural substrates targeted with the concepts of supramolecular chemistry, and research in this area has not lost rel­ evance over the years. Most of the receptors used for the recognition and detec­ tion of nucleotides in water were based on polyammonium ions, which can be structurally varied in a wide range. As seen in some of the above examples, the

151

152

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials O

O N

HN H2N I

N

N

B(OH)4–

O

N

N H2N

O

N

N O

H2N

Cyclization HO

O

OH

Cyclization

N

N O

O

B RO OR 5′-cG

5′-IG

N

HN

OH

Acyclovir

Gel breakdown and drug release

Figure 4.23  G‐quartet‐derived hydrogel containing acyclovir and 5′‐IG subunits. The intramolecular cyclization of 5′‐IG destabilized the gel, thus inducing the progressive release of the drug. Source: Adapted with permission from Ref. [109]. Copyright 2016 Royal Society of Chemistry.

interactions between the interacting partners mostly combined charge–charge attractions with hydrogen bonding. In addition, π‐stacking and other intermo­ lecular forces played a role in receptors containing suitable recognition elements such as aromatic residues. The combination of these interactions typically led to high affinity and in some cases selectivity even in water. Nucleotide recognition in water therefore seems to be less difficult than the recognition of other sub­ strates. In spite of this success, the field still offers many challenges such as the effective discrimination of NMPs, NDPs, and NTPs or the discrimination of nucleotides according to their nucleobase. The interest in such systems is high because of the biological relevance of nucleotide‐based substrates. Current research focuses on the construction of biomaterials from nucleotide‐derived building blocks and the development of nucleotide markers that allow the early detection of diseases. This work builds on the lessons learned from previous research on the supramolecular chemistry of nucleotides and develops this knowledge further into practical biomedical applications.

­Acknowledgment Financial support by the Spanish Ministerio de Economía y Competitividad and FEDER funds from the EU (Project CTQ2016‐78499‐C6‐1‐RED, CONSIDER CTQ2017-90852-RED, and Unidad de  Excelencia MDM 2015‐0538) and

References

Generalitat Valenciana (Project PROMETEOII2015‐002) is gratefully acknow­ ledged. I.P. thanks UVEG for an Atracció al Talent PhD grant. J. G.‐G. is indebted to the University of Bordeaux for an IDEX Postdoctoral Fellowship.

­References 1 Dietrich, B., Hosseini, M.W., Lehn, J.‐M., and Sessions, R.B. (1981). Anion

2

3 4 5 6 7

8

9

10

11 12

13

14

15

receptor molecules. Synthesis and anion‐binding properties of polyammonium macrocycles. J. Am. Chem. Soc. 103: 1282–1283. Kimura, E., Yatsunami, T., and Kodama, M. (1982). Macromonocyclic polyamines as biological polyanion complexons. 2. Ion‐pair association with phosphate and nucleotides. J. Am. Chem. Soc. 104: 3182–3187. Lipscomb, W.N. and Sträter, N. (1996). Recent advances in zinc enzymology. Chem. Rev. 96: 2375–2434. Bonora, M., Patergnani, S., Rimessi, A. et al. (2012). ATP synthesis and storage. Purinergic Signal. 8: 343–357. Alberts, B., Johnson, A., Lewis, J. et al. (2002). Molecular Biology of the Cell. Garland Science. Communi, D. and Boeynaems, J.M. (1997). Receptors responsive to extracellular uracil nucleotides. Trends Pharmacol. Sci. 18: 83–86. Pullarkat, S.T. and Lenz, H.J. (2001). Thymidylate synthase gene polymorphism determines response and toxicity of 5‐FU chemotherapy. Pharmacogenomics J. 1: 65–70. Pontarin, G., Ferraro, P., Valentino, M.L. et al. (2006). Mitochondrial depletion and thymidine phosphate pool dynamics in a cellular model of mitochondrial neurogastrointestinal encephalomyopathy. J. Biol. Chem. 281: 22720–22728. Rampazzo, C., Fabris, S., Franzolin, E. et al. (2007). Mitochondrial thymidine kinase and the enzymatic network regulating thymidine triphosphate pools in cultured human cells. J. Biol. Chem. 282: 34758–34769. Ashley, N., Adams, S., Slama, A. et al. (2007). Defects in maintenance of mitochondrial DNA are associated with intramitochondrial nucleotide imbalances. Hum. Mol. Genet. 16: 1400–1411. Beavo, J.A. and Brunton, L.L. (2002). Cyclic nucleotide research – still expanding after half a century. Nat. Rev. Mol. Cell Biol. 3: 710–718. Seifert, R., Schneider, E.H., and Bähre, H. (2015). From canonical to non‐ canonical cyclic nucleotides as second messengers: pharmacological implications. Pharmacol. Ther. 148: 154–184. Kalia, D., Merey, G., Nakayama, S. et al. (2013). Nucleotide, c‐di‐GMP, c‐di‐ AMP, cGMP, cAMP, (p)ppGpp signaling in bacteria and implications in pathogenesis. Chem. Soc. Rev. 42: 305–341. Carell, T., Kurz, M.Q., Müller, M. et al. (2018). Non‐canonical bases in the genome: the regulatory information layer in DNA. Angew. Chem. Int. Ed. 57: 4296–4312. Roy, B., Depaix, A., Périgaud, C., and Peyrottes, S. (2016). Recent trends in nucleotide synthesis. Chem. Rev. 116: 7854–7897.

153

154

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

16 Lapa, S.A., Chudinov, A.V., and Timofeev, E.N. (2016). The toolbox for modified

aptamers. Mol. Biotechnol. 58: 79–92.

17 Lee, I. and Berdis, A.J. (2010). Non‐natural nucleotides as probes for the

18 19

20 21

22

23

24 25

26

27

28

29

30

31

mechanism and fidelity of DNA polymerases. Biochim. Biophys. Acta 1804: 1064–1080. Amendola, V., Bergamaschi, G., and Miljkovic, A. (2018). Azacryptands as molecular cages for anions and metal ions. Supramol. Chem. 30: 236–242. García‐Espańa, E., Belda, R., González, J. et al. (2012). Receptors for nucleotides. In: Supramolecular Chemistry: From Molecules to Nanomaterials, vol. 3 (ed. P.A. Gale and J.W. Steed), 1225–1257. Chichester: Wiley‐VCH. Nakai, C. and Glinsmann, W. (1977). Interactions between polyamines and nucleotides. Biochemistry 16: 5636–5641. Bunce, S. and Kong, E.S.W. (1978). The interactions between nucleic acids and polyamines: I. High resolution carbon‐13 and hydrogen‐1 nuclear magnetic resonance studies of spermidine and 5′‐AMP. Biophys. Chem. 8: 357–368. Gasowska, A. (2003). Interaction centres of purine nucleotides: adenosine‐5′‐ diphosphate and adenosine‐5′‐triphosphate in their reactions with tetramines and Cu(II) ions. J. Inorg. Biochem. 96: 346–356. Bencini, A., Blanchi, A., García‐España, E. et al. (1999). Proton coordination by polyamine compounds in aqueous solution. Coord. Chem. Rev. 188: 97–156. Labadi, I., Sillanpää, R., and Lönnberg, H. (1992). Crystal structure of sperminium diphosphate hexahydrate. J. Chem. Soc., Dalton Trans. 765–767. Tabushi, I., Kobuke, Y., and Imuta, J. (1981). Lipophilic diammonium cation having a rigid structure complementary to pyrophosphate dianions of nucleotides. Selective extraction and transport of nucleotides. J. Am. Chem. Soc. 103: 6152–6157. Bianchi, A., Micheloni, M., and Paoletti, P. (1988). Supramolecular interaction between adenosine 5′‐triphosphate (ATP) and polycharged tetraazamacrocycles. Thermodynamic and 31P NMR studies. Inorg. Chim. Acta 151: 269–272. Bianchi, A., Micheloni, M., Nanini, V. et al. (1985). Dicopper(II) complex of the large polyazacycloalkane 1,4,7,10,13,16,19,22‐octaazacyclotetracosane (bistrien). Synthesis, crystal structure, electrochemistry, and thermodynamics of formation. Inorg. Chem. 24: 1182–1187. Bencini, A., Bianchi, A., García‐España, E. et al. (1992). Potential ATPase mimics by polyammonium macrocycles: criteria for catalytic activity. Bioorg. Chem. 20: 8–29. Albelda, M.T., Bernardo, M.A., García‐España, E. et al. (1999). Thermodynamics and fluorescence emission studies on potential molecular chemosensors for ATP recognition in aqueous solution. J. Chem. Soc., Perkin Trans. 2 2545–2549. Bencini, A., Bianchi, A., Antonio, B. et al. (1991). N,N′,N″,N‴‐(2‐ Aminoethyl)‐1,4,8,11‐tetraazacyclotetradecane (TAEC) as a polyammonium receptor for anions. J. Chem. Soc., Perkin Trans. 2 1445–1451. Andrés, A., Aragó, J., Bencini, A. et al. (1993). Interaction of hexaazaalkanes with phosphate type anions. Thermodynamic, kinetic, and electrochemical considerations. Inorg. Chem. 32: 3418–3424.

­  References

32 Czirok, J.B., Bojtár, M., Hessz, D. et al. (2013). Aminonaphthalimide‐based

33 34

35

36

37

38

39

40

41

42

43

44

45

dipodal imidazolium/triazole receptors for fluorescent sensing of nucleoside polyphosphates. Sens. Actuators, B 182: 280–287. Albelda, M.T., García‐España, E., Jiménez, H.R. et al. (2006). Cu2+ and AMP complexation of enlarged tripodal polyamines. Dalton Trans. 4474–4481. Hosseini, M.W., Blacker, A.J., and Lehn, J.‐M. (1990). Multiple molecular recognition and catalysis. A multifunctional anion receptor bearing an anion binding site, an intercalating group, and a catalytic site for nucleotide binding and hydrolysis. J. Am. Chem. Soc. 112: 3896–3904. Hosseini, M.W., Lehn, J.‐M., Duff, S.R. et al. (1987). Synthesis of mono‐ and difunctionalized ditopic [24]N6O2 macrocyclic receptor molecules. J. Org. Chem. 52: 1662–1666. Claude, S., Lehn, J.‐M., Schmidt, F., and Vigneron, J.P. (1991). Binding of nucleosides, nucleotides and anionic planar substrates by bis‐intercaland receptor molecules. J. Chem. Soc., Chem. Commun. 1182–1185. Čudić, P., Žinić, M., Tomišić, V. et al. (1995). Binding of nucleotides in water by phenanthridinium bis(intercaland) receptor molecules. J. Chem. Soc., Chem. Commun. 1073–1075. Aguilar, J.A., García‐España, E., Guerrero, J.A. et al. (1995). Multifunctional molecular recognition of ATP, ADP and AMP nucleotides by the novel receptor 2,6,10,13,17,21‐hexaaza[22]metacyclophane. J. Chem. Soc., Chem. Commun. 2237–2239. Aguilar, J.A., García‐España, E., Guerrero, J. et al. (1996). Synthesis and protonation behaviour of the macrocycle 2,6,10,13,17,21‐hexaaza[22] metacyclophane. Thermodynamic and NMR studies on the interaction of 2,6,10,13,17,21‐hexaaza[22]metacyclophane and on the open‐chain polyamine 4,8,11,15‐tetraazaoctadecane‐1,1. Inorg. Chim. Acta 246: 287–294. Albelda, M.T., Aguilar, J., Alves, S. et al. (2003). Potentiometric, NMR, and fluorescence‐emission studies on the binding of adenosine 5′‐triphosphate (ATP) by open‐chain polyamine receptors containing naphthylmethyl and/or anthrylmethyl groups. Helv. Chim. Acta 86: 3118–3135. Delépine, A.S., Tripier, R., Le Bris, N. et al. (2009). ATP and triphosphate anions complexation by bis‐linear tetraamines: strong interactions at neutral pH. Inorg. Chim. Acta 362: 3829–3834. Pouessel, J., Abada, S., Le Bris, N. et al. (2013). A new bis‐tetraamine ligand with a chromophoric 4‐(9‐anthracenyl)‐2,6‐dimethylpyridinyl linker for glyphosate and ATP sensing. Dalton Trans. 42: 4859–4872. Bazzicalupi, C., Biagini, S., Bencini, A. et al. (2006). ATP recognition and sensing with a phenanthroline‐containing polyammonium receptor. Chem. Commun. 4087–4089. Bazzicalupi, C., Bencini, A., Biagini, S. et al. (2009). Exploring the binding ability of phenanthroline‐based polyammonium receptors for anions: hints for design of selective chemosensors for nucleotides. J. Org. Chem. 74: 7349–7363. Delépine, A.S., Tripier, R., and Handel, H. (2008). Cyclen‐based bismacrocycles for biological anion recognition. A potentiometric and NMR Study of AMP, ADP and ATP nucleotide complexation. Org. Biomol. Chem. 6: 1743–1750.

155

156

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

46 Bencini, A., Biagini, S., Giorgi, C. et al. (2009). A tris‐macrocycle with proton

47

48 49

50

51

52

53

54

55

56

57

58

sponge characteristics as efficient receptor for inorganic phosphate and nucleotide anions. Eur. J. Org. Chem. 5610–5621. Delépine, A.S., Tripier, R., Le Baccon, M., and Handel, H. (2010). From flexible to constrained tris(tetraamine) ligands: synthesis, acid‐base properties, and structural effect on the coordination process with nucleotides. Eur. J. Org. Chem. 5380–5390. González, J., Llinares, J.M., Belda, R. et al. (2010). Tritopic phenanthroline and pyridine tail‐tied aza‐scorpiands. Org. Biomol. Chem. 8: 2367–2376. González‐García, J., Tomić, S., Lopera, A. et al. (2015). Aryl‐bis‐(scorpiand)‐aza receptors differentiate between nucleotide monophosphates by a combination of aromatic, hydrogen bond and electrostatic interactions. Org. Biomol. Chem. 13: 1732–1740. Arranz‐Mascarós, P., Bazzicalupi, C., Bianchi, A. et al. (2011). Binding and recognition of AMP, ADP, ATP and related inorganic phosphate anions by a tren‐based ligand containing a pyrimidine functionality. New J. Chem. 35: 1883–1891. Bazzicalupi, C., Bencini, A., Bianchi, A. et al. (1997). Basicity properties of two paracyclophane receptors. Their ability in ATP and ADP recognition in aqueous solution. J. Chem. Soc., Perkin Trans. 2 775–781. Bazzicalupi, C., Bencini, A., Bianchi, A. et al. (2008). Polyfunctional binding of thymidine 5′‐triphosphate with a synthetic polyammonium receptor containing aromatic groups. Crystal structure of the nucleotide‐receptor adduct. J. Am. Chem. Soc. 130: 2440–2441. Li, M.H., Kwok, F., Chang, W.R. et al. (2002). Crystal structure of brain pyridoxal kinase, a novel member of the ribokinase superfamily. J. Biol. Chem. 277: 46385–46390. Shionoya, M., Kimura, E., and Shiro, M. (1993). A new ternary zinc(II) complex with [12]aneN4(=1,4,7,10‐tetraazacyclododecane) and AZT (=3′‐azido‐3′‐ deoxythymidine). Highly selective recognition of thymidine and its related nucleosides by a zinc(II) macrocyclic tetraamine complex with novel complementary associations. J. Am. Chem. Soc. 115: 6730–6737. Shionoya, M., Ikeda, T., Kimura, E., and Shiro, M. (1994). Novel “multipoint” molecular recognition of nucleobases by a new zinc(II) complex of acridine‐ pendant cyclen (cyclen = 1,4,7,10‐tetraazacyclododecane). J. Am. Chem. Soc. 116: 3848–3859. Aoki, S., Honda, Y., and Kimura, E. (1998). The first selective and efficient transport of imide‐containing nucleosides and nucleotides by lipophilic cyclen‐ zinc(II) complexes (cyclen = 1,4,7,10‐tetraazacyclododecane). J. Am. Chem. Soc. 120: 10018–10026. Kikuta, E., Murata, M., Katsube, N. et al. (1999). Novel recognition of thymine base in double‐stranded DNA by zinc(II)‐macrocyclic tetraamine complexes appended with aromatic groups. J. Am. Chem. Soc. 121: 5426–5436. Aoki, S. and Kimura, E. (2000). Highly selective recognition of thymidine mono‐ and diphosphate nucleotides in aqueous solution by ditopic receptors zinc(II)‐bis(cyclen) complexes (cyclen = 1,4,7,10‐tetraazacyclododecane). J. Am. Chem. Soc. 122: 4542–4548.

­  References

59 Aoki, S. and Kimura, E. (2004). Zinc‐nucleic acid interaction. Chem. Rev. 104:

769–787.

60 Guo, Y., Ge, Q., Lin, H. et al. (2003). Recognition promoted by Zn2+ between

61

62

63

64

65 66

67

68

69

70

71

72

73

phenanthroline bridging polyaza ligands and nucleotides – Zn2+ acts as “messenger” between the receptor and substrate. J. Mol. Recognit. 16: 102–111. Guo, Y., Ge, Q., Lin, H. et al. (2003). The different roles of metal ions and water molecules in the recognition and catalyzed hydrolysis of ATP by phenanthroline‐containing polyamines. Biophys. Chem. 105: 119–131. Aguilar, J., Díaz, P., Escartí, F. et al. (2002). Cation and anion recognition characteristics of open‐chain polyamines containing ethylenic and propylenic chains. Inorg. Chim. Acta 339: 307–316. Francesconi, O., Gentili, M., Bartoli, F. et al. (2015). Phosphate binding by a novel Zn(II) complex featuring a trans‐1,2‐diaminocyclohexane ligand. Effective anion recognition in water. Org. Biomol. Chem. 13: 1860–1868. Sawada, T., Yoshizawa, M., Sato, S., and Fujita, M. (2009). Minimal nucleotide duplex formation in water through enclathration in self‐assembled hosts. Nat. Chem. 1: 53–56. Hosseini, M.W. and Lehn, J.‐M. (1987). Binding of AMP, ADP, and ATP nucleotides by polyammonium macrocycles. Helv. Chim. Acta 70: 1312–1319. Hosseini, M.W., Lehn, J.‐M., and Mertes, M.P. (1983). Efficient molecular catalysis of ATP‐hydrolysis by protonated macrocyclic polyamines. Helv. Chim. Acta 66: 2454–2466. Hosseini, M.W., Lehn, J.‐M., Maggiora, L. et al. (1987). Supramolecular catalysis in the hydrolysis of ATP facilitated by macrocyclic polyamines: mechanistic studies. J. Am. Chem. Soc. 109: 537–544. Blackburn, G.M., Thatcher, G.R.J., Hosseini, M.W., and Lehn, J.‐M. (1987). Evidence for potophosphatase catalysed cleavage of adenosine triphosphate by a dissociative‐type mechanism within a receptor‐substrate complex. Tetrahedron Lett. 28: 2779–2782. Bethell, R.C., Lowe, G., Hosseini, M.W., and Lehn, J.‐M. (1988). Mechanisms of the ATPase‐like activity of the macrocyclic polyamine receptor molecule [24] N6O2. Bioorg. Chem. 16: 418–428. Mertes, M.P. and Bowman‐Mertes, K. (1990). Polyammonium macrocycles as catalysts for phosphoryl transfer: the evolution of enzyme mimic. Acc. Chem. Res. 23: 413–418. Andres, A., Bazzicalupi, C., Bencini, A. et al. (1994). 1,10‐Dimethyl‐1,4,7,10,13,16‐ hexaazacyclooctadecane L and 1,4,7‐trimethyl‐1,4,7,10,13,16,19‐heptaazacyclo­ henicosane L1: two new macrocyclic receptors for ATP binding. Synthesis, solution equilibria and the crystal structure of (H4L)(ClO4)4. J. Chem. Soc., Dalton Trans. 2367–2373. Bencini, A., Bianchi, A., Giorgi, C. et al. (1996). Effect of nitrogen methylation on cation and anion coordination by hexa‐ and heptaazamacrocycles. Catalytic properties of these ligands in ATP dephosphorylation. Inorg. Chem. 35: 1114–1120. Aguilar, J.A., Descalzo, A.B., Díaz, P. et al. (2000). New molecular catalysts for ATP cleavage. Criteria of size complementarity. J. Chem. Soc., Perkin Trans. 2 1187–1192.

157

158

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

74 Hosseini, M.W. and Lehn, J.‐M. (1985). Cocatalysis: pyrophosphate synthesis

75

76

77

78

79 80

81 82 83

84

85

86

87

88 89

from acetylphosphate catalysed by a macrocyclic polyamine. J. Chem. Soc., Chem. Commun. 1155–1157. Yohannes, P.G., Mertes, M.P., and Bowman‐Mertes, K. (1985). Pyrophosphate formation via a phosphoramidate intermediate in polyammonium macrocycle/ metal ion‐catalyzed hydrolysis of ATP. J. Am. Chem. Soc. 107: 8288–8289. Yohannes, P.G., Plute, K.E., Bowman‐Mertes, K., and Mertes, M.P. (1987). Specificity, catalysis, and regulation: effects of metal ions on polyammonium macrocycle catalyzed dephosphorylation of ATP. Inorg. Chem. 26: 1751–1755. Hosseini, M.W. and Lehn, J.‐M. (1991). Supramolecular catalysis of adenosine‐ triphosphate synthesis in aqueous‐solution mediated by a macrocyclic polyamine and divalent metal‐cations. J. Chem. Soc., Chem. Commun. 451–453. Jahansouz, H., Himes, R.H., Jiang, Z. et al. (1989). Formate activation in neutral aqueous solution mediated by a polyammonium macrocycle. J. Am. Chem. Soc. 111: 1409–1413. Bazzicalupi, C., Bencini, A., Bianchi, A. et al. (2005). A zinc(II)‐based receptor for ATP binding and hydrolysis. Chem. Commun. 2630–2632. Bazzicalupi, C., Bianchi, A., Giorgi, C. et al. (2015). ATP dephosphorylation can be either enhanced or inhibited by pH‐controlled interaction with a dendrimer molecule. Chem. Commun. 51: 3907–3910. Pina, F., Bernardo, M.A., and García‐España, E. (2000). Fluorescent chemosensors containing polyamine receptors. Eur. J. Inorg. Chem. 2143–2157. Hargrove, A.E., Nieto, S., Zhang, T. et al. (2011). Artificial receptors for the recognition of phosphorylated molecules. Chem. Rev. 111: 6603–6782. Fujita, K., Fujiwara, S., Yamada, T. et al. (2017). Design and function of supramolecular recognition systems based on guest‐targeting probe‐modified cyclodextrin receptors for ATP. J. Org. Chem. 82: 976–981. Cheng, D., Li, Y., Wang, J. et al. (2015). Fluorescence and colorimetric detection of ATP based on a strategy of self‐promoting aggregation of a water‐soluble polythiophene derivative. Chem. Commun. 51: 8544–8546. Amilan Jose, D., Stadibauer, S., and König, B. (2009). Polydiacetylene‐based colorimetric self‐assembled vesicular receptors for biological phosphate ion recognition. Chem. Eur. J. 15: 7404–7412. Han, M.S. and Kim, D.H. (2003). Visual detection of AMP and real‐time monitoring of cyclic nucleotide phosphodiesterase (PDE) activity in neutral aqueous solution. Chemosensor‐coupled assay of PDE and PDE inhibitors. Bioorg. Med. Chem. Lett. 13: 1079–1082. Aucejo, R., Alarcón, J., Soriano, C. et al. (2005). New sensing devices part 1: indole‐containing polyamines supported in nanosized boehmite particles. J. Mater. Chem. 15: 2920–2927. Aucejo, R., Díaz, P., García‐España, E. et al. (2007). Naphthalene‐containing polyamines supported in nanosized boehmite particles. New J. Chem. 31: 44–51. Maity, D., Li, M., Ehlers, M., and Schmuck, C. (2017). A metal‐free fluorescence turn‐on molecular probe for detection of nucleoside triphosphates. Chem. Commun. 53: 208–211.

­  References

90 Inclán, M., Albelda, M.T., Carbonell, E. et al. (2014). Molecular recognition of

nucleotides in water by scorpiand‐type receptors based on nucleobase discrimination. Chem. Eur. J. 20: 3730–3741. 91 Xu, Z., Singh, N.J., Lim, J. et al. (2009). Unique sandwich stacking of pyrene‐ adenine‐pyrene for selective and ratiometric fluorescent sensing of ATP at physiological pH. J. Am. Chem. Soc. 131: 15528–15533. 92 Zeng, Z. and Spiccia, L. (2009). OFF‐ON fluorescent detection of thymidine nucleotides by a zinc(II)‐cyclen complex bearing two diagonal pyrenes. Chem. Eur. J. 15: 12941–12944. 93 Zeng, Z., Torriero, A.A.J., Bond, A.M., and Spiccia, L. (2010). Fluorescent and electrochemical sensing of polyphosphate nucleotides by ferrocene functionalised with two ZnII(TACN)(pyrene) complexes. Chem. Eur. J. 16: 9154–9163. 94 Xu, Q., Lv, H., Lv, Z. et al. (2014). A pyrene‐functionalized zinc(II)‐BPEA complex: sensing and discrimination of ATP, ADP and AMP. RSC Adv. 4: 47788–47792. 95 Fuchi, Y., Fukuda, T., and Sasaki, S. (2016). Synthetic receptor molecules for selective fluorescence detection of 8‐oxo‐dGTP in aqueous media. Org. Biomol. Chem. 14: 7949–7955. 96 Peters, G.M. and Davis, J.T. (2016). Supramolecular gels made from nucleobase, nucleoside and nucleotide analogs. Chem. Soc. Rev. 45: 3188–3206. 97 Schmuck, C. and Wienand, W. (2001). Self‐complementary quadruple hydrogen‐bonding motifs as a functional principle: from dimeric supramolecules to supramolecular polymers. Angew. Chem. Int. Ed. 40: 4363–4369. 98 Lopez, A. and Liu, J. (2017). Self‐assembly of nucleobase, nucleoside and nucleotide coordination polymers: from synthesis to applications. ChemNanoMat 3: 670–684. 99 Wang, Y., Desbat, B., Manet, S. et al. (2005). Aggregation behaviors of gemini nucleotide at the air‐water interface and in solutions induced by adenine‐uracil interaction. J. Colloid Interface Sci. 283: 555–564. 100 Liang, H., Zhang, Z., Yuan, Q., and Liu, J. (2015). Self‐healing metal‐ coordinated hydrogels using nucleotide ligands. Chem. Commun. 51: 15196–15199. 101 Schneider, H.J., Kato, K., and Strongin, R.M. (2007). Chemomechanical polymers as sensors and actuators for biological and medicinal applications. Sensors 7: 1578–1611. 102 Schneider, H.J. and Strongin, R.M. (2009). Supramolecular interactions in chemomechanical polymers. Acc. Chem. Res. 42: 1489–1500. 103 Belda, R., García‐España, E., Morris, G.A. et al. (2017). Guanosine‐5′‐ monophosphate polyamine hybrid hydrogels: enhanced gel strength probed by z‐spectroscopy. Chem. Eur. J. 23: 7755–7760. 104 Zhang, J., Li, X., Sun, X. et al. (2018). GMP‐quadruplex‐based hydrogels stabilized by lanthanide ions. Sci. China Chem. 61: 604–612. 105 Peters, G.M., Skala, L.P., Plank, T.N. et al. (2014). A G4·K+ hydrogel stabilized by an anion. J. Am. Chem. Soc. 136: 12596–12599.

159

160

4  Recognition, Transformation, Detection of Nucleotides and Aqueous Nucleotide‐Based Materials

106 Jeon, J.H., Kakuta, T., Tanaka, K., and Chujo, Y. (2015). Facile design of

organic‐inorganic hybrid gels for molecular recognition of nucleoside triphosphates. Bioorg. Med. Chem. Lett. 25: 2050–2055. 107 Tan, K.Y., Li, C.Y., Li, Y.F. et al. (2017). Real‐time monitoring ATP in mitochondrion of living cells: a specific fluorescent probe for ATP by dual recognition sites. Anal. Chem. 89: 1749–1756. 08 Wang, L., Yuan, L., Zeng, X. et al. (2016). A multisite‐binding switchable 1 fluorescent probe for monitoring mitochondrial ATP level fluctuation in live cells. Angew. Chem. Int. Ed. 55: 1773–1776. 09 Plank, T.N. and Davis, J.T. (2016). A G4·K+ hydrogel that self‐destructs. Chem. 1 Commun. 52: 5037–5040.

161

5 Carbohydrate Receptors Anthony P. Davis University of Bristol, School of Chemistry, Cantock’s Close, Bristol, BS8 1TS, UK

5.1 ­Introduction Of all the substrates one might wish to bind in water, carbohydrates are among the most challenging. The first requirement of a receptor is that it should distinguish between substrate and solvent – if this cannot be achieved, no binding will occur. The archetypal carbohydrates (glucose, galactose, etc.) are probably the most water‐like (hydromimetic) of target molecules. Their sole functional groups are OH and –O–, both present in water. Indeed the only difference between a carbo­ hydrate and a cluster of water molecules is the exact arrangement of OH groups and the presence of small patches of aliphatic CH. High affinities are certainly possible, as observed from Nature. For example, some bacteria produce periplas­ mic proteins, which bind quite strongly to simple monosaccharides such as glu­ cose, showing Ka = 106–107 M−1 [1]. However, as discussed below, they seem to be very highly optimized. The largest class of carbohydrate‐binding proteins, the lectins, bind remarkably weakly [2]. Affinities to monosaccharides are especially low, often below 103 M−1 and rarely >104 M−1 [3]. One has the impression that even Nature requires a special effort to bind carbohydrates strongly. To add to the challenge, achieving selectivity is difficult. There are many saccharide units, and they tend to differ subtly from each other; for example, the common monosac­ charides glucose 5.1, galactose 5.2, and mannose 5.3 (Figure 5.1) are stereoiso­ mers, differentiated only by the configurations at one or two chiral centers. Aside from the theoretical interest, there are practical reasons why carbohy­ drate receptors have been sought. The most prominent applications relate to glucose and the management of diabetes. There is great interest in developing sensors, which could continuously monitor glucose levels in the blood of diabet­ ics [4, 5]. If one can make a receptor, a sensor becomes possible in principle (although turning binding into a signal is not necessarily trivial) (see Chapter 12). This objective has dominated much of the work on boronic acid–based r­ eceptors, discussed in Section 5.4. A glucose receptor might also be exploited in glucose‐ dependent insulin delivery systems [6] or in glucose‐responsive insulin [7]. Beyond glucose, there are many other targets of interest. When combined in Supramolecular Chemistry in Water, First Edition. Edited by Stefan Kubik. © 2019 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2019 by Wiley-VCH Verlag GmbH & Co. KGaA.

162

5  Carbohydrate Receptors

HO HO

HO

OH O OH

HO

OH

5.1

OH O

HO

OH O

HO

HO OMe 5.5 Methyl α-D-glucoside

HO HO

HO

OMe NHAc

HO HO

5.3

OMe

OH O AcHN

OH

OMe

HO

OH NHAc 5.8 N-Acetyl-Dgalactosamine

5.7 N-Acetyl-Dglucosamine (GlcNAc) OH O

OH O

O HO

OH

OH

5.9

5.10

5.11

GlcNAc-β-OMe

GlcNAc-α-OMe

D-Cellobiose

HO HO

OH O OH

O HO

OH O

HO HO

OH

OH

5.12 D-Lactose

HO HO

OH O

O HO NHAc

OH O OH NHAc

OH O

HO

OH NHAc

HO HO

OMe

OH

5.4 Methyl β-D-glucoside

OH O

HO HO

OH O

HO HO

D-Mannose

OH 5.6 Methyl β-D-galactoside

OH O

OH O

HO HO

OH OH 5.2 D-Galactose

D-Glucose

HO HO

OH O

OH

OH O HO

O HO 5.13 D-Maltose

HO HO

OH O OH

OH

OH O NH3 OH

5.14

5.15

D-N,N′-Diacetylchitobiose

D-Glucosamine

Figure 5.1  Selected carbohydrates used for binding studies in water.

oligosaccharides, carbohydrates provide Nature with its most densely packed information storage system [8], which is used extensively to label cells, proteins, etc. [9]. This raises the possibility of receptors with biological activity, as diag­ nostic agents or as tools for glycobiological research. Broadly speaking, research on carbohydrate receptors may be divided into three categories. The first, covered in Section  5.2, employs organic molecular systems that rely purely on non‐covalent interactions. Early work of this type was performed largely in organic solvents, where the problems are generally less severe [10]; distinguishing between a carbohydrate and, for example, chloroform is not especially difficult. However, in recent years there has been more focus on aqueous solvent systems, with some degree of success. This approach has the advantage of biomimicry  –  aside from potential applications, the results shed light on the somewhat mysterious process of natural carbohydrate recognition. A second, smaller group of systems, described in Section 5.3, exploit interactions

5.2  Organic Molecular Receptors

between metal ions and carbohydrate functional groups. These may also be con­ sidered biomimetic, as metal ions are involved in a significant subset of lectins. Finally, covered in Section 5.4 are receptors that contain boron atoms and exploit the reversible formation of covalent B─O bonds [11, 12]. These are not biomi­ metic, but tend to be simple, accessible, and practical (within certain limits).

5.2 ­Organic Molecular Receptors This class of receptor relies on the standard non‐covalent interactions, which operate between covalently assembled molecules. A good starting point for dis­ cussion is the most powerful known family of monosaccharide receptors, the bac­ terial periplasmic proteins. Figure 5.2 shows the crystal structure of glucose in the binding site of one of these proteins. Immediately apparent is the concentrated network of hydrogen bonds, which completely surround the substrate; the pro­ tein undergoes a major conformational change on binding [13], which is probably necessary to achieve the high affinity (5 × 106 M−1). However, also important is the presence of aromatic side chains sandwiching the glucose and making contact with CH groups [14, 15]. The CH–π interactions between these moieties are intrinsically weak, but in water they become significant because they are rein­ forced by the hydrophobic effect  –  water molecules do not interact well with either surface and need to be released (see Chapter 1). Indeed, it has been pro­ posed that carbohydrate binding sites are generally inhospitable to water despite the presence of many polar groups, a consequence of their “polyamphiphilic” nature [16]. The issue of whether carbohydrate recognition is driven by polar or hydrophobic interactions has caused controversy in the past, but the message from Figure 5.2 seems clear – designs for carbohydrate receptors should focus on

Asn

Phe Asp

H2O Asp

Arg

Asn

Asp His

Trp

Figure 5.2  Crystal structure of glucose complexed to a monosaccharide‐binding periplasmic protein from Escherichia coli (see Ref. [1]). CH groups on the carbohydrate make contact with aromatic side chains (tryptophan and phenylalanine; shown in green), while substrate oxygens form 12 hydrogen bonds to a water molecule (purple) and eight polar side chains (3 × aspartate, blue; 3 × asparagine, orange; 1 × histidine, cyan; and 1 × arginine, magenta).

163

164

5  Carbohydrate Receptors

complementarity between host and guest, considering both polar and apolar fea­ tures. If the design perfectly matches the saccharide target, it is less likely to com­ plement a cluster of water molecules, and strong binding can be achieved. It should be said that complementarity is not the only design issue. Water solu­ bility is an obvious requirement of a receptor to operate in aqueous solution. Moreover, mere solubility is often insufficient. For reliable binding studies, the receptor should be monomeric in water, so that affinities can be determined by straightforward calculations and complications with spectroscopic techniques (especially nuclear magnetic resonance (NMR)) can be avoided. For organic mol­ ecules with hydrophobic surfaces, especially aromatic surfaces, there is often a strong tendency to self‐associate. Monomolecularity may therefore be difficult to achieve. 5.2.1  Acyclic Receptors If carbohydrate receptors require preorganized polar and apolar components, the simplest candidates are likely to be acyclic structures in which a rigid apolar core is surrounded by polar (water‐solubilizing) substituents. Such structures bear some resemblance to lectins, which generally possess cleft rather than cav­ ity binding sites [9]. The synthetic versions can be effective, though generally with more complex and/or unusual carbohydrates. The pairing of complex sub­ strates with simple receptors is not as paradoxical as it may seem. Large sub­ strates are intrinsically easier to bind due to greater surface area, and unusual substituents also moderate the challenge. Charged functional groups are espe­ cially helpful, as electrostatic interactions can be deployed, while apolar units (even methyl groups) also make the problem easier. Compounds 5.16–5.20 (Chart 5.1) serve as examples of such receptors. In accord with the above discussion, all were directed at less‐challenging targets. The anthracene derivative 5.16, with a net twofold positive charge, formed weak complexes with anionic oligosaccharides such as sialyl Lewis X (5.21) and GM3 (5.22) (Ka ~ 100 M−1) [17]. Pentacationic 5.17 bound polyanionic heparin 5.23 quite strongly [18], but this may simply result from multiple electrostatic interac­ tions; a polycationic micellar aggregate was even more powerful [19]. Anionic 5.18 was ineffective for uncharged substrates but showed a Ka of 90 M−1 for glu­ cosamine 5.15. Receptor 5.19 succeeded with two neutral substrates, methyl β‐d‐glucoside 5.4 and cellobiose 5.11; NMR studies suggested sequential Ka val­ ues of 2 M−1 (1 : 1) and 72 M−1 (1 : 2) for the former and 305 M−1 (1 : 1) and 66 M−1 (1 : 2) for the latter. Pyrene‐based “platforms” 5.20 are more “three dimen­ sional,” in that the polyionic substituents are preorganized so as to create two well‐defined binding sites. Binding constants of the complexes of anionic 5.20a with mannosamine 5.24 and cationic 5.20b with sialoside 5.25 were 3000 and 1300 M−1, respectively [20]. Complexation on both faces was detected, mimick­ ing lectin multivalency [9]. Methyl β‐d‐glucoside 5.4 and methyl β‐d‐galactoside 5.6 both showed measurable affinities of ~40 M−1, while galactose 5.2 (excep­ tionally) was bound with Ka ~20 M−1. It is possible that developments based on 5.19/5.20 may lead to lectin‐like performance, although enclosed binding sites may always be needed for really high affinities.

5.2  Organic Molecular Receptors R

R NH2

O2C

H2 N

NH2

CO2

H2N

N

O

N H

N H

NH3

NH2

O N H

S

CO2

HO O

O O2C

O

CO2

N

O

N H

CO2

HO

OH

O2C

COX

X O

5.20a X = HN

CO2

X H2N

CO2

OH OH O

5.21 Sialyl Lewis X (SLeX)

HO CO2

O

AcHN HO OH

O

X

OMe

AcHN HO OH

5.19

O

O

CO2 O

5.25

COX

XOC

OH

HO

5.18

XOC

OH 5.24

O

N

O CO2

NH3 O

HO HO

O NH

O

O

N

O O2C

NH2

5.17

5.16 R = (CH2)3NHC(NH2)NH2

X

N H

NH3

O

OH

O2C

O AcHN HO OH

NH2

O

O

N H HN

O2C O

NH2 NH2

OH O O

OH

O OH HO

5.22 GM3

NH2

OR NHAc

OH

OH OH O

NH2

O

HOOH

NH

5.20b X = HN

OH

O OH O

O HO

SO3

O OH

OR

SO3 O O HN O SO3

5.23 Heparin

Chart 5.1  Ar

Ar N NH

5.26 Ar =

PO3(NH4)2

HN

5.27 Ar =

N Ar

SO3Na

Ar

Chart 5.2 

A number of porphyrin‐based platform structures have also been investigated, leading to reports of high affinities even with simple monosaccharides. For example, UV–vis titrations on tetraphosphonate 5.26 (Chart 5.2) and glucose and galactose were analyzed to give remarkable Ka values of ~20 000 M−1 for both substrates [21]. However, studies on the closely related 5.27 revealed slow UV–vis spectral changes, due to changes in aggregation state, which interfere

165

166

5  Carbohydrate Receptors

with binding experiments [22]. The activities of these porphyrins therefore remain uncertain. Enclosed binding sites can be achieved with acyclic structures, provided they fold in the correct fashion. An early attempt to “sandwich” carbohydrates in water involved the dipeptide 5.28 (Chart 5.3), employing tryptophan to optimize CH–π interactions [23]. Weak binding was observed but only to the relatively easy substrate maltotriose 5.29 (Ka = 8 M−1). In later work, a number of fold­ amers built from larger, more rigid components have been studied. The hexaca­ tionic oligoamide 5.30 was used to target heparin, showing Ka = 5 × 105 M−1 [24]. Oligoresorcinol 5.31 was found to bind a number of oligosaccharides, although quantification was hampered by complex stoichiometries [25]. Most recently, the water‐soluble phenol–pyridine oligomer 5.32 was investigated as a mono­ saccharide receptor using induced circular dichroism [26]. Weak unquantified binding was observed of glucose 5.1 and N‐acetylglucosamine 5.7, and, more H N H

NH3 O

N H

HO HO

NH H

OH O HO

CO2

OH O

O HO

HO

5.29 Maltotriose

5.28 NH3

H2N

N H

H N

4

OH O OH OH

NH3

H3N NH2

O HO

NH3 O

S

O

H N

O

S

H N O

NH2

H N

O

4N

O

H

NH2

5.30 OH HO

HO

HO

OH HO

OH HO

OH HO

OH HO

OH HO

OH

OH HO

5.31 O OH

7

OH

N

CH2(CH2CH2O)8Me 5.32

Chart 5.3 

O

5

CH2(CH2CH2O)8Me

OH

5.2  Organic Molecular Receptors

significantly, an affinity of 2000 M−1 was measured for glucosamine 5.5. None of these foldamers are known to provide a preorganized, geometrically defined binding site, but the approach seems to have potential. 5.2.2  Macrocyclic Receptors As in other areas of supramolecular chemistry, macrocyclic structures have been employed to create well‐defined clefts and cavities for carbohydrate recognition. The first biomimetic carbohydrate receptors (i.e. operating in water though non‐ covalent interactions) were macrocycles 5.33 (Chart 5.4), developed by Aoyama and coworkers [27]. These calixarene‐type hosts, prepared from resorcinols and a sulfonate‐containing aldehyde, provide apolar clefts and probably act mainly via hydrophobic/CH–π interactions. They did not appear to bind the common hexoses (glucose, galactose, etc.) but showed measurable affinities for fucose 5.35 (Ka = 2, 6, and 8 M−1 for 5.33a–c, respectively); fucose is less hydrophilic than most carbohydrates so is a relatively easy substrate. It was later shown that deprotonation of 5.33a raised the binding constant to 26 M−1 [28]. This is con­ sistent with the involvement of CH–π interactions, which should increase with the electron density in the π system [29]. Another early system, macrocycle 5.34, also employed a calixarene‐type architecture [30]. Again, the common hexoses failed to bind, but the receptor was more successful with glycosides. The favored substrate was methyl α‐d‐mannoside 5.36, which was bound with Ka = 75 M−1. X H

HO

H O

O R

R

R

R

NaO3S NaO3S

H O

X

OH OH OH HO

X O

H

O H

5.33a X = H

O

H

O H

SO3Na

OH HO OH OH NaO3S

SO3Na

NaO3S

X

SO3Na

SO3Na

R = CH2CH2SO3Na

5.34

5.33b X = Me R = CH2CH2SO3Na 5.33c X = OH R = CH2CH2SO3Na

O OH OH 5.35

OH OH

HO HO HO

OH O 5.36

OMe

Chart 5.4 

Since these early results, a diverse group of macrocyclic carbohydrate recep­ tors have been reported. Perhaps surprisingly, cucurbit[7]uril 5.37 (Chart 5.5) proved an effective receptor for amino sugars, despite lacking aromatic surfaces or internal H‐bonding groups [31]. Glucosamine 5.15 and galactosamine 5.38

167

168

5  Carbohydrate Receptors

were bound with Ka = 4400 and 16000 M−1, respectively, giving complexes that could be studied by X‐ray crystallography [32]. However, to put this result in context, some other amines are bound by 5.37 with extreme affinities, suggesting that the cucurbituril is more of an amine receptor than carbohydrate receptor. OO NNN NN

O N N

N N N NN

N N

O

N NNN N

N N

N NN N N

O

OO O

O O

O N N

HO HO

OH O NH3

OH

OO

O

5.37

5.38

Chart 5.5 

A quite different family of structures, exemplified by 5.39 (Chart 5.6), were explored using dynamic combinatorial chemistry through disulfide interchange in the presence of potential guests [33, 34]. Some of the affinities measured were remarkably high – for example, 5.39 was reported to bind galactoside 5.40 with Ka = 8000 M−1, based on isothermal titration calorimetry (ITC). Unfortunately, this value was not corroborated by another technique, and there are no struc­ tural data on the complexes. In the absence of more information, these systems hold promise but probably require further investigation. NH2

O N NH O O HO

NH2 S

NH

S

HO

O HN

NH

HO

HN S

NH O O

S

N

OH O HO

OMe

5.40

HO 5.39

Chart 5.6 

Finally, a number of macrocyclic systems have employed pairs of anthracene units to create enclosed amphiphilic binding sites. The tetralactams 5.43 (Figure 5.3) were developed as part of the program on “temple” receptors dis­ cussed more fully in Section 5.2.3. Like the polycyclic temples described later, they were designed to target all‐equatorial carbohydrates such as β‐glucose, providing NH groups to hydrogen bond to equatorial –OH and aromatic sur­ faces (anthracenes in this case) to complement axial CH in a framework solubi­ lized by dendrimeric polycarboxylates. However, unlike the other temples they are prepared in a single cyclization step from commercial diamine 5.41 and the spacer reagents 5.42 (Figure 5.3). Despite their simple architecture, receptors 5.43 are quite effective. Prototype 5.43a was found to bind glucose with Ka = 56

5.2  Organic Molecular Receptors O O

NH2

C6F5O +

H2N

C6F5O

O

Cyclize

Xp

Deprotect

O

O

NH O NH O

5.43

CO2

CO2 CO2

O

N

CO2

O O

X

X

5.41 5.42 Xp = O-t-butyl-protected solublizing group

5.43a X = HN

NH

O

CO2

CO2

N

O

CO2 CO2

O N O

CO2

CO2

NH O2C

O

CO2 CO2

O

5.43b X = HN

N H NH

O N

O CO2 O2C

CO2

N O O N

NH

CO2 CO2

O2C

O

O

N

N

O2C O C 2

CO2

O

O O

CO2

N

H N

5.43c X = HN

CO2

NH O

O

N H

CO2 CO2

O 2C CO2 CO2

CO2

Figure 5.3  Synthesis and structures of macrocyclic receptors 5.43.

M−1 and good selectivity against galactose and mannose (Table 5.1) [35]. Complexation was detected and quantified by 1H NMR, ITC, and a significant change in fluorescence with potential for glucose sensing. X‐ray crystallogra­ phy confirmed that the glucose was bound through hydrogen bonding and CH–π interactions (Figure 5.4) [36]; at the time of writing, this is the only crys­ tal structure of a biomimetic carbohydrate receptor, bound or unbound. Due to the ease of synthesis, it was feasible to explore the effect of varying the side chains [37]. Table 5.1 includes results for 5.43b and 5.43c with 18 and 36 car­ boxylates, respectively. The additional carboxylates yielded a minor improve­ ment in binding to glucose and major increases in affinities to glucosamine 5.15, reaching 7000 M−1 for 5.43c. Alternatively, the two anthracenes have been linked via 1,8‐diaminocarbazole units to give macrocycle 5.44 (Chart 5.7), solubilized by four phosphonate groups [38]. Binding studies on 5.44 were complicated by a strong tendency to dimerize, as well as variable protonation states at neutral pH. However, careful analysis of the data yielded encouraging affinities for methyl glycosides. In par­ ticular, methyl α‐fucoside 5.45 was bound with a “median binding concentra­ tion” (BC050: a parameter designed to reflect overall binding ability) of ~0.5 mM.

169

170

5  Carbohydrate Receptors

Table 5.1  Association constants Ka for binding of carbohydrates in water to anthracene‐based monocycles 5.43a–c. Ka(M−1)b) Carbohydratea)

5.43a

5.43b

5.43c

d‐Glucose 5.1

56

90

70

d‐Galactose 5.2

4

7

3

d‐Mannose 5.3

~0

~0

~0

Methyl β‐d‐glucoside 5.4

96

115

92

Methyl α‐d‐glucoside 5.5

6

N‐Acetyl‐d‐glucosamine 5.7

9

31

33

d‐Glucosamine 5.15

160

2400

7000

a) See Figure 5.1. b) Measured by 1H NMR titration in D2O.

2.1

2.2 2.3

2.1

Figure 5.4  X‐ray crystal structure of 5.43a bound to β‐d‐glucose. Anthracene units are shown in space‐filling mode; side chains are omitted. Four intermolecular hydrogen bonds, lengths 2.1–2.3 Å, are shown as green broken lines. PO32–

N H 2–O

3P

HN

NH

OMe O

NH

NH

3P

5.44

Chart 5.7 

HO

OH 5.45

NH 2–O

PO32–

OH

5.2  Organic Molecular Receptors

Affinities for all other substrates, including the all‐equatorial methyl β‐glucoside 5.4, were significantly lower. The preference for a non‐all‐equatorial substrate contrasts with 5.43 and other “temples” (see Section 5.2.3), suggesting that this architecture might be more versatile than first thought. 5.2.3  Macropolycyclic Cage Receptors Carbohydrates possess complex three‐dimensional structures and require cor­ responding binding sites for full complementarity. Encapsulation is difficult to achieve with acyclic or monocyclic structures, so macropolycyclic architectures may need to be considered. The author’s group has studied a number of such systems targeting the all‐equatorial family of carbohydrates (glucose 5.1, β‐­ glucosides such as 5.4, cellodextrins such as 5.11, GlcNAc 5.7, etc.). The general approach has been to create a cavity bounded by two parallel aromatic sur­ faces and a number of polar spacer groups. As illustrated in Figure 5.5 for ­monosaccharide binding, the aromatic surfaces can participate in CH–π interac­ tions with the axial carbohydrate CH, reinforcing the hydrophobic effect, while the spacers can hydrogen bond to equatorial polar substituents. We refer to the structures as “temple” receptors, after the cartoon depiction in Figure 5.5. The first implementation of this approach was the biphenyl‐based tricyclic cage 5.46 (Chart 5.8) [39]. The design employs isophthalamide spacers with polycarboxylate side chains as later used in 5.43 (see above). In a first series of binding experiments, 5.46 was found to complex glucose with Ka = 9 M−1 and fair selectivity against galactose and mannose (Table 5.2). Though very weak, this association represents the first clear example of binding to a common hexose in water through non‐covalent interactions. In later work, 5.46 was tested against GlcNAc 5.7 and its β‐methyl glycoside 5.9, revealing that the β‐GlcNAc unit bound much more strongly than β‐glucosyl [40]. Significantly, the affinity for GlcNAc‐β‐OMe 5.9, at 630 M−1, was very similar to that of wheat germ aggluti­ nin (WGA) for the same substrate. WGA is the lectin commonly used to bind Apolar (aromatic) surfaces

H H

Polar spacer units

HO HO HO

H

H H O H Y

OR H

Hydrogen bonds Hydrophobic/CH–π interactions Y = OH, NHAc, etc.

Figure 5.5  Generalized design for “temple” monosaccharide receptors, targeting all‐equatorial substrates.

171

172

5  Carbohydrate Receptors

O

O O

O

O

O O

HH N HN N

O O

O

O

O O

O O

NH

O O

O

O

O

O

NH O NH

O

O O O

O

O

HN

5.47

O

AcHN

H N

O O

O

O

O

O OH O

O

O O

5.46

HO HO

O O

O

O

O HN O N H

O O

N HH N NH

O

O

O

O

Val-Pro-Thr-Ser-Gly-NH3

N Ser-Ala-Asn-Met-CONH2 H

Chart 5.8 

Table 5.2  Association constants Ka for binding of carbohydrates in water to receptor 5.46. Carbohydratea)

Ka(M−1) for binding to 5.46b)

d‐Glucose 5.1

9

d‐Galactose 5.2

2

d‐Mannose 5.3

≤2

Methyl β‐d‐glucoside 5.4

28

GlcNAc 5.7 (α:β = 64 : 36)

56c),d)

GlcNAc‐β‐OMe 5.9

630c)

N‐Acetyl‐d‐galactosamine 5.8

2

a) b) c) d)

See Figure 5.1. Measured by 1H NMR titration in D2O. Slow exchange on the 1H NMR chemical shift time scale. Binding to α anomer not detected, implying Ka ~ 150 M−1 for β anomer.

β‐GlcNAc units, so this result showed potential for matching at least some natu­ ral competitors. β‐O‐GlcNAc is a dynamic posttranslational modification of pro­ teins so has special significance as a target [41]. Receptor 5.46 showed an encouraging affinity of Ka = 1040 M−1 for the glycopeptide 5.47, suggesting applications in the detection and manipulation of modified proteins. However,

5.2  Organic Molecular Receptors Y HN

H X N O O X

O

O

X

O

O

O

O HN O

Y N H H N X

O

O

O

NH O

N H Y

N H Y

HH N N O O HN O

X

X

N HO H O N O N O H O H N O H N O

O

O H

N

N

5.48 X = NHC(CH2OCH2CH2CO2)3

H

O

5.49

Y = F, OH, O , OMe, OEt, OPr, OBu O X O

O

N O

O H

N H

N O

H N H

O

X

H N

H N

X

O

O

5.50

Chart 5.9 

binding to simpler models was weaker [42], so that any usage would presumably be context dependent. In attempts to tune the binding properties of 5.46, a number of alterations were made to the structural design. In particular, a variety of substituents were added to the biphenyl para positions, as in 5.48 (Chart 5.9) [43, 44]. This modi­ fication should have minimal effects on the cage conformation but can change the electron density in the aromatic surfaces, potentially modulating CH–π interactions. Somewhat surprisingly, all substituents enhanced binding to glu­ cose, even the electron‐withdrawing F (expected to reduce CH–π interactions) [45]. The optimal substituent was found to be Y = OPr, resulting in Ka = 60 M−1. The study concluded that variations in CH–π interactions may have some sig­ nificance but were superimposed on other effects, possibly relating to cavity hydration. Other variations included the introduction of alternative pyrrolo spacers, as in 5.49 [46], and simplifying the structure (tricyclic to bicyclic) as in 5.50 [47]. The former led to relatively minor effects, while the latter to slightly lower affinities for most substrates (as might be expected). In addition to biphenyl‐based monosaccharide receptors, terphenyl‐based sys­ tems have also been developed for disaccharide targets. The original design 5.51 (Chart 5.10) employed a tetracyclic architecture with meta‐terphenyl compo­ nents, designed to ensure an open cavity [48]. Tested against cellobiose 5.11 and N,N‐diacetylchitobiose 5.14, both with all‐equatorial substitution patterns,

173

174

5  Carbohydrate Receptors

X

H

O

HN

NH O

N

H

O

O

HN O

X H

O

O

X

N

O O

O

O

N

N H

O O

H

5.51

O X

N H

O

X

O

X

NH

O

N

O

O

O

O

N H Y

O2C

CO2 CO2

O

O HN NH O

O

NH

X O

CO2 CO2

O CO2

5.54

O

O HO

CO2

NH O N H NH

NH

X

OH

X = HN

HN

O

OH O

X

5.52 Y = H 5.53 Y = OMe

NH

O

HO HO

O

NH O

O

O

O

O

O HN O HN Y

X = NHC(CH2OCH2CH2CO2)3

X

Y N HH N X

Y HH N X N

CO2

O 2C OH O OH 5.55

O HO n

OH O OH

OH

HO HO

OH O NH3

O HO

OH O NH3

O HO n

OH O

OH

NH3

5.56

Chart 5.10 

receptor 5.51 showed Ka ~600 and 120 M−1, respectively. Selectivity against other substitution patterns was excellent. For example, lactose 5.12, with just one axial OH, was bound with a Ka of just 12 M−1. The tricyclic structure 5.52 was originally avoided because of fears that the cavity might close through a twisting motion. When eventually prepared, along with methoxy analog 5.53, it was surprisingly effective. Affinities for cellobiose rose to 3300 M−1, while, inter­ estingly, binding to glucose was almost undetectable [49]. Receptors 5.46 and 5.51 were effectively the first lectin mimics (“synthetic lectins”) [50] and presented an opportunity to investigate aspects of lectin behavior. In particular, it was possible to test the effect of solvent on carbohy­ drate binding, including media in which proteins would not be stable. It was found that, as expected, addition of methanol to (protected) receptor in chloro­ form weakened binding dramatically. However, less predictable was the discov­ ery that the addition of methanol to water also diminished binding. These results implied that carbohydrate recognition in water does not rely solely on hydrogen bonding, but that displacement of water (i.e. the hydrophobic effect) is also significant [51].

5.2  Organic Molecular Receptors

H

H

Figure 5.6  The evolution from (twisted) biphenyl to (flat) pyrene as roof/floor units for temple receptors.

The tendency of biphenyls to twist, and the success of condensed aromatic components in 5.43, prompted a move to pyrene‐based systems (Figure 5.6). Pyrenes are similar in size to biphenyls but rigidly planar, maximizing the poten­ tial for CH–π interactions to multiple axial CH, and also present greater hydro­ phobic surface area. Incorporating pyrenes into polycyclic cages is relatively challenging, if only because of solubility problems, but the rigidity of the condensed aromatic lowers the risk of cavity collapse. Bicycle 5.54 (Chart 5.10), with just three spacers, was predicted to have an open cavity with larger portals than, for example, tri­ cycle 5.48. When synthesized, receptor 5.54 was found to perform modestly with monosaccharides, but remarkably well with all‐equatorial oligosaccharides (Table 5.3) [52]. Among the cellodextrins, affinities rose to 12 000 M−1 for cello­ tetraose 5.55 (n = 2). The success with these large substrates, far too big to be enclosed by 5.54, clearly suggests a threaded pseudorotaxane geometry, illus­ trated in Figure 5.7 for 5.54 and cellopentaose 5.55 (n = 3). NOE NMR studies supported this conclusion, while atomic force microscopy (AFM) images implied that 5.54 can multiply thread onto cellulose 5.55 (n = large) and chitosan 5.56. The results suggest that related receptors might be useful for modifying or manipulating these polysaccharides, perhaps inducing cellulose to dissolve in water. Table 5.3  Selected association constants Ka for binding of carbohydrates in water to pyrene‐based bicyclic receptor 5.54. Carbohydratea)

Ka(M−1)

d‐Glucose 5.1

120

d‐Galactose 5.2

18

GlcNAc‐β‐OMe 5.9

270

d‐Cellobiose 5.11

3 900

d‐Cellotriose 5.55 (n = 1)

5 200

d‐Cellotetraose 5.55 (n = 2)

12 000

d‐Cellopentaose 5.55 (n = 3)

8 800

d‐Cellohexaose 5.55 (n = 4)

8 700

d‐Chitobiose 5.56 (n = 0)

19 000

a) See Figure 5.1.

175

176

5  Carbohydrate Receptors

Figure 5.7  Model of receptor 5.54 and cellopentaose 5.55 (n = 3) in the pseudorotaxane geometry indicated by NOE data. Pyrene units are shown in space‐filling mode, water‐ solubilizing side chains are omitted; isophthalamide spacers are colored gold.

The pyrene‐based tricyclic system 5.57 (Chart 5.11) bears an obvious relation­ ship to 5.46 and was clearly an attractive target. Although it presented special synthetic difficulties, these were eventually overcome by using an undirected tri­ X

HN

NH

O

O

X

X

O

O

HN

O

O

X

X

O

N H

N H

NH

HN

NH

HN

NH

O

O

O

O

+

HN O

NH

X

O

HN

NH

O NH

O

O

O

O

O

O

O

O

O

H N H

O

X O

N

O

O

H N

H N

X

X

N H

N

O

N

H

O

HN O

H N

X

X

N H

H O N H

O

CO2 X

H N

O

O

O O

5.59

O2C

5.60

CO2 CO2

O X = HN

NH O

CO2 CO2

N H NH

CO2

O CO2 O2C

Chart 5.11 

X

HN

O

O

H

O

O

5.58

O N

O

O

5.57

X

X O

CO2

5.2  Organic Molecular Receptors

Table 5.4  Association constants Ka for binding of carbohydrates in water to pyrene‐based tricyclic receptors 5.57 and 5.58. Ka(M−1)b) Carbohydratea)

5.57

5.58

d‐Glucose 5.1

120c)

190c)

Methyl β‐d‐glucoside 5.4

1440c)

1 180,b) 1 230c)

N‐Acetyl‐d‐glucosamine 5.7

—

520,b) 520c)

GlcNAc‐β‐OMe 5.9

2 100,b) 2 200c)

18 200,b) 16 600c)

GlcNAc‐α‐OMe 5.10

—

1 550,b) 1 520c)

a) See Figure 5.1. b) Measured by 1H NMR titration in D2O. c) Measured by ITC in H2O.

cyclization procedure that also generated the “staggered” analog 5.58 [53]. Separating 5.57 and 5.58 was challenging, but once achieved both could be investigated. These molecules were difficult to study by NMR due to slow exchange between conformations, but where affinities could be measured with confidence, they were often exceptional (Table 5.4). Most notably, the staggered receptor 5.58 was found to bind GlcNAc‐β‐OMe 5.9 with Ka ~18 000 M−1. The glycopeptide 5.47 was also tested. Curiously this seemed to interact more favora­ bly with “eclipsed” isomer 5.57, giving Ka = 67 000 M−1. Presumably, this high affinity is partly due to extra‐cavity interactions. In any case, it suggests that use­ ful affinities and selectivities could be achievable for GlcNAc glycosides in spe­ cific environments, a potentially valuable development for glycobiology. Finally, it is possible to extend the variety of temple receptors by desymmetri­ zation, such that roof and floor are no longer identical. An initial study combined pyrenyl and biphenyl units in bicyclic system 5.59 (Chart 5.11) [54]. Although the binding properties of 5.59 were not distinctive, resembling a compromise between 5.50 and 5.54, the methodology is extendable to a fairly wide range of potential receptors. More interesting is the tricyclic receptor 5.60, obtained from a tetra‐aminopyrene unit and the well‐known 1,3,5‐triethyl‐2,4,6‐ tris(aminomethyl)benzene [55]. This system possesses a chiral framework and was prepared as a 1 : 1 mixture with its enantiomer [56]. Although the racemate could not be resolved, the two enantiomers could be studied simultaneously by 1 H NMR with some substrates. N‐Acetyl‐d‐glucosamine 5.7 gave especially sig­ nificant results: one enantiomer of the receptor bound with Ka = 1280 M−1, and the other with Ka = 81 M−1. The enantioselectivity of 16 : 1 would be significant in most areas of supramolecular chemistry, let alone carbohydrate binding in water. Meanwhile, the affinity of the stronger binding enantiomer for GlcNAc is three times higher than for the lectin WGA and is the highest yet reported for biomimetic recognition of an uncharged, underivatized monosaccharide. 5.60, or its enantiomer, also holds the record for glucose at 250 M−1. Unfortunately, in the absence of a resolution method, it is not possible to say which enantiomer is

177

178

5  Carbohydrate Receptors O HN

NH

NH

NH2 O

S

O O

N

HN

5.61

N

NH O

O

O O

NH

O

NH S O

HN

O

HN O

Chart 5.12 

effective for either substrate. The affinities and selectivities are therefore difficult to rationalize. However, the results support the view that for a chiral substrate, a chiral binding site is likely to prove more complementary. The temple receptors are specifically aimed at all‐equatorial carbohydrates and may not prove adaptable to other geometries. An interesting bicyclic alternative is the bowl‐shaped molecule 5.61 (Chart 5.12), with one apolar surface and a polar peptidic ring [57]. Although this architecture could be suitable for binding a wide range of substrates, the only positive result reported for 5.61 was weak binding to cellobiose 5.11 (Ka = 8 M−1). Nonetheless, the general approach looks attractive and may be worth pursuing further.

5.3 ­Metal Complexes as Carbohydrate Receptors The presence of vicinal diol units in carbohydrates, capable of chelation, suggests a role for metal ions in carbohydrate recognition. Indeed, it is quite common for natural lectins to exploit divalent metal ions in this way, especially Ca2+ [58]. Applying the approach to synthetic receptors is challenging, as the metal must be bound firmly and reliably before other structural features are incorporated. For this reason, perhaps, studies on carbohydrate‐binding metal complexes have mostly involved relatively simple systems. However, useful results have been obtained in some cases. For example, Striegler has reported a series of dinuclear copper(II) complexes, e.g. 5.62 and 5.63 (Chart 5.13), which bind to sugars under H N N

Cu

O 5.62

Chart 5.13 

3+

H N Cu

+

N N

RO

O Cu 5.63 R =

N O

Cu O

OR

O 3

5.3  Metal Complexes as Carbohydrate Receptors N N

N N

+

5.64

OMe HO

O

N

O

O Eu

HO

O

O

N

O

N

O

O

O O

O O

5.66

Eu

N OMe

OH 5.65 NH

O O

Pd AcO OAc

O

5.67

Chart 5.14 

basic conditions [59, 60]. Transition metals such as copper are clearly advanta­ geous as they are readily “fixed” in water using soft ligand atoms. Impressively, complex 5.62 binds mannose with Ka ≈ 104 M−1 in water, although this requires pH 12.4, which presumably deprotonates the substrate. Square planar palladium(II) complexes have also been exploited, with binding detected by displacement of the fluorophore 5.65 (Chart 5.14) [61]. Interestingly, the performance was quite strongly dependent on ligand structure. Complex 5.64 gave especially promising results; the addition of glucose to 5.64 and 5.65 at pH = 7.4 caused large increases in fluorescence, implying formation of a glu­ cose–Pd–ligand assembly. Lanthanide ions have been employed in the form of 5.66 and, more recently, as the simple ethylenediamine tetra‐acetic acid (EDTA) complex 5.67. The former showed a strong increase in fluorescence when exposed to certain sialylated oligosaccharides, including the ganglioside GM1 [62]. The latter responded to simple monosaccharides through circularly polar­ ized luminescence, measured on a Raman optical activity spectrometer [63]. For biomimetic recognition employing Ca2+ complexes, there is currently no synthetic solution. However, there is a family of secondary metabolites that dem­ onstrates the possibility, the pradimycins and benanomycins [64, 65]. A well‐ studied example is pradimycin A 5.68 (Chart 5.15). In the presence of Ca2+, 5.68 has been shown by ITC to bind methyl α‐d‐mannoside 5.36 with Ka = 10 000 M−1, with weaker binding of 260 M−1 at a secondary site. Structural information has been obtained using solid‐state NMR spectroscopy [65, 66]. Mannosides are important targets due to their role in immunology, so the natural product 5.68 could provide a useful lead for future design work. Finally, metal ions can serve as structural elements in receptors, without pro­ viding the main driving force for binding. The macrocycle 5.69 (Chart 5.16) appears to behave as a hydrophobic cavity capable of binding largely through CH–π interactions. This cavity shows a remarkable affinity for sucrose 5.70, at Ka = 1170 M−1 [67]. Simple monosaccharides did not form complexes, nor did several other disaccharides. The selectivity was attributed to a particular com­ plementarity between sucrose, which is relatively compact in shape, and the toroidal cavity of the receptor. Metal ions may also be used to create 3D arrays of

179

180

5  Carbohydrate Receptors CO2H NH

O O

HO OH

MeO

OH O

OH 5.68

O HO

O O

O NHMe OH

HO HO

Chart 5.15 

N

Pt

N

MeO

OMe O

MeO

O

O

O O

OMe

O

MeO

OMe

HO HO HO HO

O O HO

OH O OH OH

5.70 N

Pt

N

5.69

Chart 5.16 

binding sites in metal–organic frameworks (MOFs). The MOF NU‐1000, in which pyrenyl units are linked by zirconium‐based nodes, shows selective absorption of cellobiose 5.11 and lactose 5.12, but not glucose 5.1 or maltose 5.13 [68].

5.4 ­Boron‐Based Receptors The chelate effect is also exploited in a second family of receptors that, unusually for “supramolecular” systems, rely on covalent bond formation. These are the boronic acids and esters that bind saccharides in aqueous solution through the rapid formation of cyclic boronates (Figure 5.8). This approach has been inten­ sively researched, mainly because some levels of binding are easy to achieve and the route to applications is generally more obvious. The area is well served by reviews [5, 11, 69], including a book [12], so that coverage here will be brief and selective.

5.4  Boron‐Based Receptors HO

OH B OH 5.71

–H2O

+

B

O 5.72

HO

+OH–

+OH–

HO

HO OH B OH 5.73

O

–H2O

+ HO

HO O B O 5.74

Figure 5.8  Cyclic boronate formation from ethane‐1,2‐diol and a boronic acid.

As shown in Figure 5.8, the interaction between a boronic acid and a vicinal diol is complicated, potentially involving both trigonal and tetrahedral species (depending on pH) [12, 70]. Binding is strongly favored by the tetrahedral geom­ etry; the equilibrium constant for the formation of 5.72 from 5.71 is up to five orders of magnitude smaller than that for the conversion of 5.73 to 5.74. This can be understood by considering the O─B─O bond angles in the trigonal spe­ cies. In boronic acid 5.71, this angle is around 120°, whereas in cyclic ester 5.72 it is compressed to 113°. The formation of 5.74 from 5.73 does not engender strain and is therefore far more favorable. Measured binding constants between carbohydrates and simple boronic acids are therefore strongly dependent on conditions. Thus, early measurements for phenylboronic acid 5.75 (Chart 5.17) and glucose/fructose in water gave Ka = 110/4400 M−1 [71], whereas a later study in pH 7.4 phosphate buffer gave Ka = 4.6/160 M−1 [72]. Despite this variation in absolute values, the preference for fructose 5.76 over glucose 5.1 is a pervading tendency for boron‐based receptors. As mentioned earlier, the monitoring of physiological glucose levels is a key aim in the area of carbohydrate recognition [5]. Much work has been devoted to apply­ ing boron‐based systems to this end, with particular emphasis on fluorescence signaling. Initial systems relied on conjugating the boron atom to a fluorophore, as in 5.77 [73]. Shinkai and James then took an important step by connecting a fluorophore to boron via a pyramidal nitrogen atom, as in 5.78 [74]. The nitrogen atom provides a mechanism for fluorescence transduction via photoinduced

B(OH)2 5.75

Chart 5.17 

HO

HO OH O

B(OH)2 N

OH HO 5.76

5.77

5.78

B(OH)2

181

182

5  Carbohydrate Receptors

B(OH)2

N

(HO)2B

N

OMe OMe

N

N 5.79

B(OH)2

B(OH)2

5.80

Chart 5.18 

e­ lectron transfer (PET), in which the amine lone pair is more or less available to quench the excited state of the anthracene (see Chapter 12). The nitrogen–boron interaction has two effects. Firstly, it moderates binding to c­ arbohydrates so that it occurs over a useful pH range. Secondly, it is different in complexed and uncom­ plexed species, resulting in fluorescence signaling. The nature of the B–N interac­ tion and the exact mechanism of signaling are still under discussion [75], but the fluorophore–N–B motif has been widely successful in carbohydrate sensing [5]. Having established carbohydrate sensing, the key issue was to bias selectivity toward glucose (and in particular, away from fructose). Success was achieved by adding a second N–B unit to 5.78, giving 5.79 (Chart 5.18). This receptor is able to bind carbohydrates through formation of two boronic acid ester units, pro­ vided the geometry is favorable. Glucose did indeed fit the binding site well, with Ka = 4000 M−1 in methanol/aqueous buffer, 1 : 2. Fructose was bound with a Ka of just 320 M−1 in the same medium. The chiral analog 5.80 was also studied. Selectivity for glucose was lost, fructose being slightly favored, but this receptor did show modest enantioselectivity (~3 : 1 for fructose) [76]. Given the success of 5.79 and 5.80, which are both quite flexible, it seems likely that rigid bifunctional boronates could show very high selectivities. An interesting approach was pioneered by Drueckhammer and coworkers. First, they modeled the complex 5.81 (Chart 5.19) between p‐tolylboronic acid and α‐d‐glucose. Then, they employed the tolyl C─Me bonds (highlighted in bold in Chart 5.19) as vectors for input into the molecular design program CAVEAT. By matching the vectors to linkages to potential scaffolds, the program generated

O HO

B

HO B OH

O O HO

O O B OH O

O OH 5.81

Chart 5.19 

H

5.82

HO B OH

5.4  Boron‐Based Receptors

the ­candidate bis‐boronic acid 5.82. The receptor was synthesized and tested against glucose and other monosaccharides in 30% methanol/buffer. Addition of glucose gave a significant fluorescence response, implying Ka ~40 000 M−1, with excellent selectivity (e.g. 400 : 1 vs. galactose) [77]. Perhaps surprisingly, this sys­ tem and/or approach has not been developed further since the original publica­ tion in 2001, and there may be further potential that could be realized. A key advantage of boron‐based carbohydrate receptors is the virtual guar­ antee of binding. Organic molecules chosen at random will probably not bind saccharides in water to any measurable degree. Careful design is required and this encourages long‐term focus on specific substrates (vide infra). However, molecules containing boronic acid/esters are almost certain to bind, and this allows more generalized, less targeted approaches. In particular it becomes reasonable to build libraries with boronic components and screen for binding to different targets; with boron present the chance of a null result is very low. For example, the groups of Hall and Thompson/Lavigne have both developed peptide‐based “boronolectin” libraries and screened for binding to carbohy­ drate epitopes. In the course of their work, Hall’s group tested a variety of boronic moieties for carbohydrate binding and found that the boroxole 5.83 (Chart 5.20) showed favorable properties [78]. This was then incorporated in a peptide library of general form 5.84, which were screened for binding to the

O O

R1 5.83

N H

HO

OH OH O O

O N H

O

OH O OR

O R

R2

O O R3

N H

O

N H

B OH

HO B

NH

HN

O H N

O O 5.86

Chart 5.20 

H N

5.84 R1 = solublizing group and linker R2, R3 = variable (20 possibilities in each position)

AcHN 5.85

HN

NH

O H N

O

HO

O

B OH O

OH

HO B

B OH

N H

H N O

O

O O N H

OMe

O

183

184

5  Carbohydrate Receptors

Thompson–Friedenreich (TF) disaccharide 5.85 [79]. The TF antigen is pre­ sent in a large proportion of human cancers and thus is an important target. The 400‐member library was screened using an enzyme‐linked immunosorb­ ent assay (ELISA), and one member 5.86 was found to be exceptionally active (estimated Ka = 2000 M−1) while also showing high selectivity for the target. The Thompson/Lavigne group has prepared larger bead‐bound libraries using split‐and‐mix methodology, introducing the boronic acid units through a reductive amination step [80]. Selected members showed potential for differ­ entiating between a series of glycoproteins [81]. Anslyn used a similar method to generate boronolectins for the analysis of simple sugars, with detection by indicator uptake. Pattern‐based sensing was demonstrated in real‐world sam­ ples using an array of eight bead‐bound receptors [82]. Finally, it is also possible to incorporate boron in nucleic acids. Wang and cow­ orkers conducted systematic evolution of ligands by exponential enrichment (SELEX) experiments using the boron‐containing reagent 5.87 (Chart 5.21) to replace thymine triphosphate in the polymerase chain reaction (PCR) amplifica­ tions [83]. The target used for selection was a glycosylated protein, and the resulting aptamers showed remarkably high affinities for this substrate. The method seems attractive as a general approach to boronolectins, albeit with mac­ romolecular structures.

5.5 ­Conclusions Arguably there are two main motivations for the study of synthetic carbohydrate receptors: firstly, an improved understanding of natural carbohydrate recogni­ tion (and, more generally, of molecular recognition in water) and, secondly, the development of applications such as glucose sensing and oligosaccharide profil­ ing. The systems discussed in Section  5.2, and to some extent 5.3, have made useful contributions to the first objective. The successful deployment of the hydrophobic effect, CH–π interactions, hydrogen bonding, electrostatic forces, and metal coordination implies that all these factors are important and confirms that we can come quite close to mimicking Nature by using designed, abiotic molecules. However, there are still important limitations; notably, while strate­ gies are available for binding the all‐equatorial family of carbohydrates, other stereochemistries are less well served. Moreover, while some results point to potential applications (e.g. glucose sensing, solubilization of cellulose, detection of OGlcNAc in proteins), none of the published systems seems ready for use in the real world. In contrast, research on the boron‐based receptors discussed in Section  5.4 has always been application focused (understandably, given that the binding is not biomimetic). Translation to the marketplace requires that problems inherent to this approach must be addressed. For example, trivalent boron compounds are  often susceptible to oxidation, especially by hydrogen peroxide (which is often present in biological samples) and selectivity remains an issue. However, for ­glucose monitoring, good progress has been made, to the point where an

5.5 Conclusions B(OH)2 N

O

N N N

N

O NH O NH

5.87

O O O O P O P O P O O O O

N

O

O OH

Chart 5.21 

implantable boron‐based sensor is now in late stages of testing [84]. It will be interesting to discover whether this momentum will be maintained or whether the biomimetic approach might prove competitive in the longer term. Addendum (October 2018). Though correct when originally written, the first part of the above conclusion about biomimetic receptors may now be revised. Very recently, the author’s group has reported the bicyclic cage 5.88 (Chart 5.22), which binds glucose in water with a Ka of 19 000 M−1 and very high selectivity (~100 : 1 vs. galactose, ≥1000 : 1 vs. a range of non‐carbohydrates) [85]. This molecule does seem suitable for real‐world use, and applications are currently under investigation.

O

O O O

O

HN H N

O O

O

O

O O

O

O

O

O

HN

O

O

O

HN

O NH HN

O O O

O O

Chart 5.22 

O

O

O

O O

5.88

OH O O

O N H NH

N H O

O O O O

O O

NH

O

NH

O

HN HN

NH

O

O HN

NH O NH

O

O

NH NH

O O

NH

HN

O O

O

NH

NH

NH

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O O

O

O O

O

185

186

5  Carbohydrate Receptors

­References 1 Quiocho, F.A. (1989). Protein‐carbohydrate interactions: basic molecular

features. Pure Appl. Chem. 61: 1293–1306.

2 Ambrosi, M., Cameron, N.R., and Davis, B.G. (2005). Lectins: tools for the

molecular understanding of the glycocode. Org. Biomol. Chem. 3: 1593–1608. 3 Toone, E.J. (1994). Structure and energetics of protein‐carbohydrate complexes. Curr. Opin. Struct. Biol. 4: 719–728. 4 Gifford, R. (2013). Continuous glucose monitoring: 40 years, what we’ve learned and what’s next. ChemPhysChem 14: 2032–2044. 5 Sun, X.L. and James, T.D. (2015). Glucose sensing in supramolecular chemistry. Chem. Rev. 115: 8001–8037. 6 Wu, Q., Wang, L., Yu, H.J. et al. (2011). Organization of glucose‐responsive systems and their properties. Chem. Rev. 111: 7855–7875. 7 Bakh, N.A., Cortinas, A.B., Weiss, M.A. et al. (2017). Glucose‐responsive insulin by molecular and physical design. Nat. Chem. 9: 937–943. 8 Laine, R.A.A. (1994). Calculation of all possible oligosaccharide isomers both branched and linear yields 1.05 × 10(12) structures for a reducing hexasaccharide – the isomer‐barrier to development of single‐method saccharide sequencing or synthesis systems. Glycobiology 4: 759–767. 9 Solis, D., Bovin, N.V., Davis, A.P. et al. (2015). A guide into glycosciences: how chemistry, biochemistry and biology cooperate to crack the sugar code. Biochim. Biophys. Acta 1850: 186–235. 10 Davis, A.P. and Wareham, R.S. (1999). Carbohydrate recognition through noncovalent interactions: a challenge for biomimetic and supramolecular chemistry. Angew. Chem. Int. Ed. 38: 2978–2996. 11 Wu, X., Li, Z., Chen, X.X. et al. (2013). Selective sensing of saccharides using simple boronic acids and their aggregates. Chem. Soc. Rev. 42: 8032–8048. 12 James, T.D., Phillips, M.D., and Shinkai, S. (2006). Boronic Acids in Saccharide Recognition. Cambridge: RSC. 13 Marvin, J.S. and Hellinga, H.W. (1998). Engineering biosensors by introducing fluorescent allosteric signal transducers: construction of a novel glucose sensor. J. Am. Chem. Soc. 120: 7–11. 14 Asensio, J.L., Arda, A., Canada, F.J., and Jimenez‐Barbero, J. (2013). Carbohydrate‐aromatic interactions. Acc. Chem. Res. 46: 946–954. 15 Hudson, K.L., Bartlett, G.J., Diehl, R.C. et al. (2015). Carbohydrate‐aromatic interactions in proteins. J. Am. Chem. Soc. 137: 15152–15160. 16 Lemieux, R.U. (1996). How water provides the impetus for molecular recognition in aqueous solution. Acc. Chem. Res. 29: 373–380. 17 Billing, J., Grundberg, H., and Nilsson, U.J. (2002). Amphiphilic anthracene‐ amino acid conjugates as simple carbohydrate receptors in water. Supramol. Chem. 14: 367–372. 18 Bromfield, S.M., Barnard, A., Posocco, P. et al. (2013). Mallard blue: a high‐ affinity selective heparin sensor that operates in highly competitive media. J. Am. Chem. Soc. 135: 2911–2914.

­  References

19 Bromfield, S.M., Posocco, P., Chan, C.W. et al. (2014). Nanoscale self‐assembled

20

21

22

23

24 25

26

27

28

29 30

31

32

33

multivalent (SAMul) heparin binders in highly competitive, biologically relevant, aqueous media. Chem. Sci. 5: 1484–1492. Carter, T.S., Mooibroek, T.J., Stewart, P.F.N. et al. (2016). Platform synthetic lectins for divalent carbohydrate recognition in water. Angew. Chem. Int. Ed. 55: 9311–9315. Kral, V., Rusin, O., Charvatova, J. et al. (2000). Porphyrin phosphonates: novel anionic receptors for saccharide recognition. Tetrahedron Lett. 41: 10147–10151. Renney, C.M., Fukuhara, G., Inoue, Y., and Davis, A.P. (2015). Binding or aggregation? Hazards of interpretation in studies of molecular recognition by porphyrins in water. Chem. Commun. 51: 9551–9554. Otsuki, J., Kobayashi, K., Toi, H., and Aoyama, Y. (1993). Oligosaccharide‐ peptide interaction. binding of maltodextrin to Trp–Trp via sugar‐bisindole intercalation. Tetrahedron Lett. 34: 1945–1948. Choi, S., Clements, D.J., Pophristic, V. et al. (2005). The design and evaluation of heparin‐binding foldamers. Angew. Chem. Int. Ed. 44: 6685–6689. Goto, H., Furusho, Y., and Yashima, E. (2007). Double helical oligoresorcinols specifically recognize oligosaccharides via heteroduplex formation through noncovalent interactions in water. J. Am. Chem. Soc. 129: 9168–9174. Ohishi, Y., Abe, H., and Inouye, M. (2017). Saccharide recognition and helix formation in water with an amphiphilic pyridine‐phenol alternating oligomer. Eur. J. Org. Chem. 6975–6979. Kobayashi, K., Asakawa, Y., Kato, Y., and Aoyama, Y. (1992). Complexation of hydrophobic sugars and nucleosides in water with tetrasulfonate derivatives of resorcinol cyclic tetramer having a polyhydroxy aromatic cavity: importance of guest‐host CH‐pi interactions. J. Am. Chem. Soc. 114: 10307–10313. Yanagihara, R. and Aoyama, Y. (1994). Enhanced sugar‐binding ability of deprotonated calix[4]resorcarene in water: balance of CH‐pi interaction and hydrophobic effect. Tetrahedron Lett. 35: 9725. Carver, F.J., Hunter, C.A., Livingstone, D.J. et al. (2002). Substituent effects on edge‐to‐face aromatic interactions. Chem. Eur. J. 8: 2848–2859. Poh, B.‐L. and Tan, C.M. (1993). Contribution of guest–host CH‐pi interactions to the stability of complexes formed from cyclotetrachromotropylene as host and alcohols and sugars as guests in water. Tetrahedron 49: 9581–9592. Jang, Y., Natarajan, R., Ko, Y.H., and Kim, K. (2014). Cucurbit[7]uril: a high‐ affinity host for encapsulation of amino saccharides and supramolecular stabilization of their alpha‐anomers in water. Angew. Chem. Int. Ed. 53: 1003–1007. Wang, J., Jang, Y., Khedkar, J.K. et al. (2016). How does solvation affect the binding of hydrophilic amino saccharides to cucurbit[7]uril with exceptional anomeric selectivity? Chem. Eur. J. 22: 15791–15799. Rauschenberg, M., Bomke, S., Karst, U., and Ravoo, B.J. (2010). Dynamic peptides as biomimetic carbohydrate receptors. Angew. Chem. Int. Ed. 49: 7340–7345.

187

188

5  Carbohydrate Receptors

34 Rauschenberg, M., Bandaru, S., Waller, M.P., and Ravoo, B.J. (2014). Peptide‐

based carbohydrate receptors. Chem. Eur. J. 20: 2770–2782.

35 Ke, C., Destecroix, H., Crump, M.P., and Davis, A.P. (2012). A simple and

36

37

38

39

40 41

42

43 44

45 46

47

48 49

50

accessible synthetic lectin for glucose recognition and sensing. Nat. Chem. 4: 718–723. Mandal, P.K., Kauffmann, B., Destecroix, H. et al. (2016). Crystal structure of a complex between beta‐glucopyranose and a macrocyclic receptor with dendritic multicharged water solubilizing chains. Chem. Commun. 52: 9355–9358. Destecroix, H., Renney, C.M., Mooibroek, T.J. et al. (2015). Affinity enhancement by dendritic side chains in synthetic carbohydrate receptors. Angew. Chem. Int. Ed. 54: 2057–2061. Francesconi, O., Martinucci, M., Badii, L. et al. (2018). Biomimetic synthetic receptor selectively recognising fucose in water. Chem. Eur. J. 24: 6828–6836. Klein, E., Crump, M.P., and Davis, A.P. (2005). Carbohydrate recognition in water by a tricyclic polyamide receptor. Angew. Chem. Int. Ed. 44: 298–302. Ferrand, Y., Klein, E., Barwell, N.P. et al. (2009). A synthetic lectin for O‐linked beta‐N‐acetylglucosamine. Angew. Chem. Int. Ed. 48: 1775–1779. Slawson, C., Copeland, R.J., and Hart, G.W. (2010). O‐GIcNAc signaling: a metabolic link between diabetes and cancer? Trends Biochem. Sci. 35: 547–555. Corzana, F., Fernandez‐Tejada, A., Busto, J.H. et al. (2011). Molecular recognition of beta‐O‐GlcNAc glycopeptides by a lectin‐like receptor: binding modulation by the underlying Ser or Thr amino acids. ChemBioChem 12: 110–117. Barwell, N.P., Crump, M.P., and Davis, A.P. (2009). A synthetic lectin for beta‐glucosyl. Angew. Chem. Int. Ed. 48: 7673–7676. Barwell, N.P. and Davis, A.P. (2011). Substituent effects in synthetic lectins – exploring the role of CH‐pi interactions in carbohydrate recognition. J. Org. Chem. 76: 6548–6557. Cockroft, S.L. and Hunter, C.A. (2007). Chemical double‐mutant cycles: dissecting non‐covalent interactions. Chem. Soc. Rev. 36: 172–188. Joshi, G. and Davis, A.P. (2012). New H‐bonding patterns in biphenyl‐based synthetic lectins; pyrrolediamine bridges enhance glucose‐selectivity. Org. Biomol. Chem. 10: 5760–5763. Howgego, J.D., Butts, C.P., Crump, M.P., and Davis, A.P. (2013). An accessible bicyclic architecture for synthetic lectins. Chem. Commun. 49: 3110–3112. Ferrand, Y., Crump, M.P., and Davis, A.P. (2007). A synthetic lectin analog for biomimetic disaccharide recognition. Science 318: 619–622. Sookcharoenpinyo, B., Klein, E., Ferrand, Y. et al. (2012). High‐affinity disaccharide binding by tricyclic synthetic lectins. Angew. Chem. Int. Ed. 51: 4586–4590. Davis, A.P. (2009). Synthetic lectins. Org. Biomol. Chem. 7: 3629–3638.

­  References

51 Klein, E., Ferrand, Y., Barwell, N.P., and Davis, A.P. (2008). Solvent effects

52 53

54

55

56

57

58 59 60 61 62

63 64 65 66 67 68

in carbohydrate binding by synthetic receptors: implications for the role of water in natural carbohydrate recognition. Angew. Chem. Int. Ed. 47: 2693–2696. Mooibroek, T.J., Casas‐Solvas, J.M., Harniman, R.L. et al. (2016). A threading receptor for polysaccharides. Nat. Chem. 8: 69–74. Rios, P., Carter, T.S., Mooibroek, T.J. et al. (2016). Synthetic receptors for the high‐affinity recognition of O‐GlcNAc derivatives. Angew. Chem. Int. Ed. 55: 3387–3392. Mooibroek, T.J., Crump, M.P., and Davis, A.P. (2016). Synthesis and evaluation of a desymmetrized synthetic lectin: an approach to carbohydrate receptors with improved versatility. Org. Biomol. Chem. 14: 1930–1933. Hennrich, G. and Anslyn, E.V. (2002). 1,3,5‐2,4,6‐functionalized, facially segregated benzenes – exploitation of sterically predisposed systems in supramolecular chemistry. Chem. Eur. J. 8: 2218–2224. Rios, P., Mooibroek, T.J., Carter, T.S. et al. (2017). Enantioselective carbohydrate recognition by synthetic lectins in water. Chem. Sci. 8: 4056–4061. Reenberg, T., Nyberg, N., Duus, J.O. et al. (2007). Specific recognition of disaccharides in water by an artificial bicyclic carbohydrate receptor. Eur. J. Org. Chem. 5003–5009. Weis, W.I. and Drickamer, K. (1996). Structural basis of lectin‐carbohydrate recognition. Ann. Rev. Biochem. 65: 441–473. Striegler, S. and Dittel, M. (2003). A sugar discriminating binuclear copper(II) complex. J. Am. Chem. Soc. 125: 11518–11524. Striegler, S. and Gichinga, M.G. (2008). Disaccharide recognition by binuclear copper(II) complexes. Chem. Commun. 5930–5932. Janowski, V. and Severin, K. (2011). Carbohydrate sensing with a metal‐based indicator displacement assay. Chem. Commun. 47: 8521–8523. Alpturk, O., Rusin, O., Fakayode, S.O. et al. (2006). Lanthanide complexes as fluorescent indicators for neutral sugars and cancer biomarkers. Proc. Natl. Acad. Sci. U.S.A. 103: 9756–9760. Wu, T., Prusa, J., Kessler, J. et al. (2016). Detection of sugars via chirality induced in europium(III) compounds. Anal. Chem. 88: 8878–8885. Nakagawa, Y. and Ito, Y. (2012). Carbohydrate‐binding molecules with non‐ peptidic skeletons. Trends Glycosci. Glycotechnol. 24: 1–12. Nakagawa, Y., Doi, T., Taketani, T. et al. (2013). Mannose‐binding geometry of pradimicin A. Chem. Eur. J. 19: 10516–10525. Nakagawa, Y., Doi, T., Masuda, Y. et al. (2011). Mapping of the primary mannose binding site of pradimicin A. J. Am. Chem. Soc. 133: 17485–17493. Yamashina, M., Akita, M., Hasegawa, T. et al. (2017). A polyaromatic nanocapsule as a sucrose receptor in water. Sci. Adv. 3: e1701126. Yabushita, M., Li, P., Bernales, V. et al. (2016). Unprecedented selectivity in molecular recognition of carbohydrates by a metal‐organic framework. Chem. Commun. 52: 7094–7097.

189

190

5  Carbohydrate Receptors

69 Jin, S., Cheng, Y.F., Reid, S. et al. (2010). Carbohydrate recognition by

boronolectins, small molecules, and lectins. Med. Res. Rev. 30: 171–257.

70 Peters, J.A. (2014). Interactions between boric acid derivatives and saccharides

71 72 73

74

75

76

77

78

79

80

81

82

83

in aqueous media: structures and stabilities of resulting esters. Coord. Chem. Rev. 268: 1–22. Lorand, J.P. and Edwards, J.O. (1959). Polyol complexes and structure of the benzeneboronate ion. J. Org. Chem. 24: 769–774. Springsteen, G. and Wang, B.H. (2002). A detailed examination of boronic acid‐diol complexation. Tetrahedron 58: 5291–5300. Yoon, J. and Czarnik, A.W. (1992). Fluorescent chemosensors of carbohydrates – a means of chemically communicating the binding of polyols in water based on chelation‐enhanced quenching. J. Am. Chem. Soc. 114: 5874–5875. James, T.D., Sandanayake, K.R.A.S., and Shinkai, S. (1994). Novel photoinduced electron‐transfer sensor for saccharides based on the interaction of boronic acid and amine. J. Chem. Soc., Chem. Commun. 477–478. Chapin, B.M., Metola, P., Vankayala, S.L. et al. (2017). Disaggregation is a mechanism for emission turn‐on of ortho‐aminomethylphenylboronic acid‐ based saccharide sensors. J. Am. Chem. Soc. 139: 5568–5578. James, T.D., Sandanayake, K., and Shinkai, S. (1995). Chiral discrimination of monosaccharides using a fluorescent molecular sensor. Nature 374: 345–347. Yang, W., He, H., and Drueckhammer, D.G. (2001). Computer‐guided design in molecular recognition: design and synthesis of a glucopyranose receptor. Angew. Chem. Int. Ed. 40: 1714–1718. Dowlut, M. and Hall, D.G. (2006). An improved class of sugar‐binding boronic acids, soluble and capable of complexing glycosides in neutral water. J. Am. Chem. Soc. 128: 4226–4227. Pal, A., Berube, M., and Hall, D.G. (2010). Design, synthesis, and screening of a library of peptidyl bis(boroxoles) as oligosaccharide receptors in water: identification of a receptor for the tumor marker TF‐antigen disaccharide. Angew. Chem. Int. Ed. 49: 1492–1495. Bicker, K.L., Sun, J., Lavigne, J.J., and Thompson, P.R. (2011). Boronic acid functionalized peptidyl synthetic lectins: combinatorial library design, peptide sequencing, and selective glycoprotein recognition. ACS Comb. Sci. 13: 232–243. Bicker, K.L., Sun, J., Harrell, M. et al. (2012). Synthetic lectin arrays for the detection and discrimination of cancer associated glycans and cell lines. Chem. Sci. 3: 1147–1156. Edwards, N.Y., Sager, T.W., McDevitt, J.T., and Anslyn, E.V. (2007). Boronic acid based peptidic receptors for pattern‐based saccharide sensing in neutral aqueous media, an application in real‐life samples. J. Am. Chem. Soc. 129: 13575–13583. Li, M.Y., Lin, N., Huang, Z. et al. (2008). Selecting aptamers for a glycoprotein through the incorporation of the boronic acid moiety. J. Am. Chem. Soc. 130: 12636–12638.

­  References

84 Mortellaro, M. and DeHennis, A. (2014). Performance characterization of an

abiotic and fluorescent‐based continuous glucose monitoring system in patients with type 1 diabetes. Biosens. Bioelectron. 61: 227–231. 5 Tromans, R.A., Carter, T.S., Chabanne, L. et al. (2019). A biomimetic receptor 8 for glucose. Nat. Chem. 11: 52–56.

191

193

6 Ion Receptors Luca Leoni1, Antonella Dalla Cort1, Frank Biedermann2, and Stefan Kubik 3 1

Università La Sapienza, Dipartimento di Chimica, Piazzale Aldo Moro 5, 00179 Roma, Italy Karlsruhe Institute of Technology, Institute of Nanotechnology, Hermann‐von‐Helmholtz‐Platz 1, 76344 Eggenstein‐Leopoldshafen, Germany 3 Technische Universität Kaiserslautern, Fachbereich Chemie ‐ Organische Chemie, Erwin‐Schrödinger‐Straße, 67663 Kaiserslautern, Germany 2

6.1 ­Introduction Ion receptors should selectively and ideally strongly bind a positively or negatively charged substrate, either an inorganic ion or an ion featuring additional organic substituents. Such receptors can target the charged group of the substrate by using the strongest type of non‐covalent interactions, namely, Coulomb attraction and/or coordinative interactions. Designing ion receptors that function in water therefore may seem to be straightforward at first sight, certainly less challenging than, for instance, designing receptors for carbohydrates, which often do not differ substantially from a cluster of water molecules (Chapter 5). Many charged substrates are moreover biologically or environmentally relevant, rendering the development of ion receptors also worthwhile. Examples are alkali metal ions that have pharmacological properties (lithium) or are responsible for maintaining the membrane potential of cells (sodium and potassium), alkaline earth metal ions that are components of the bone structure (calcium) and responsible for hard water (calcium and magnesium), and many transition metals that are involved in biochemical processes or contribute to the environmental pollution. Important organic cations are the neurotransmitters acetylcholine, dopamine, noradrenaline, and adrenaline or side‐chain methylated lysine and arginine derivatives that play key roles in controlling gene activity and expression. Anionic substrates range from simple inorganic anions such as nitrate and phosphate, which are responsible for the eutrophication of water bodies, over toxic anions such as cyanide and arsenate to biologically relevant anions such as chloride, sulfate, and, most importantly, phosphate. In addition, many substrates and cofactors in biochemistry are anionic, containing either carboxylate or phosphate ester groups. Thus, there are a variety of ionic substrates for which Supramolecular Chemistry in Water, First Edition. Edited by Stefan Kubik. © 2019 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2019 by Wiley-VCH Verlag GmbH & Co. KGaA.

194

6  Ion Receptors

r­eceptors could be developed, but what are the potential applications of such receptors, and is their design really straightforward? 6.1.1  Potential Applications for Ion Receptors The different roles of charged species in biological processes render ion receptors certainly interesting for medical applications. Receptors that permit the specific detection of a target in a biological matrix could be used, for example, in medical imaging (Chapter 13). On the other hand, receptors that bind to an ion involved in a biological process could alter its course or outcome and therefore serve as therapeutics [1]. It has to be considered, however, that the environment in which biological chemistry takes place substantially differs from bulk water. One reason for this is that recognition events typically occur on the surfaces or the inside of proteins or membranes where the polarity is significantly lower than in bulk water. Moreover, the high concentration of macromolecules and other solutes causes the water concentration in the cytosol of cells to be significantly lower than the 55 M of water. Supramolecular receptors therefore do not necessarily have to be active in bulk water to exhibit biological activity. For receptors that are active in water, other fields of application are probably more important. Such receptors can, for example, play key roles in analytical applications that require the detection and constant monitoring of charged species in water, usually at low concentrations [2]. Ion receptors could moreover be used for the extraction of toxic or radioactive cations and anions from water or the recycling of valuable raw materials, providing means for water purification and decontamination [3]. Selectivity in substrate recognition is a decisive aspect in these contexts because the target often represents one component in a relatively complex mixture of many charged and therefore competing species. For analytical purposes, the aspect of selectivity can be elegantly addressed by the construction of sensor arrays, but this approach requires elaborate equipment (Chapter 12). The use of simple supramolecular probes that selectively sense a target substrate by changing the color, for instance, therefore represents an attractive alternative [4]. Examples of ions whose detection and extraction is either an important goal of environmental remediation or even economically interesting are the radioactive uranyl cation UO22+, toxic transition metal ions such as mercury or cadmium, pertechnetate that is used in imaging or occurs in nuclear waste, sulfate that prevents the efficient vitrification of radioactive tank wastes, or phosphate that is a valuable and nonrenewable base material for fertilizers. It has to be noted that extraction processes mostly involve heterogeneous systems or the transfer of the target ion into or across an organic liquid phase so that also in this application the recognition event does not necessarily have to take place in bulk water. 6.1.2  Binding Modes of Ion Receptors Figure 6.1 schematically illustrates the different steps involved in the complexation of an ion by a receptor in water. They involve dehydration (of parts) of the binding partners and the formation of a hydrated complex. The intrinsic

6.1 Introduction

+

Hydrated receptor (R)

+

Hydrated ion (I)

Hydrated complex (RI)

Released water molecules

Figure 6.1  Steps associated with the binding of an ion by a receptor in water.

r­ eceptor–ion interactions can vary, depending on the actual type of interactions responsible for binding. Among the interactions useful for ion recognition in water, coordinative interactions between a Lewis‐basic receptor and a cation or a Lewis‐acidic receptor and an anion are the most efficient ones because of their covalent but dynamic nature. Ionic interactions could be expected to be the second‐best choice, but they are only strong in the gas phase and in nonpolar media. In water, the charges of ions are effectively screened by the high permittivity of the environment and by hydration [5]. As a consequence, most oppositely charged ions have no pronounced propensity to assemble in water at the millimolar concentrations normally used for binding studies. Ion pairing does, however, occur at higher concentrations (see Chapter 1). Water molecules thus strongly attenuate even very strong types of interactions. To obtain insight on how these effects influence complex formation, it is helpful to quantitatively relate the Gibbs free energy of binding ΔG° to the intrinsic free energy of complexation in the gas phase ΔG°int and the free energies of hydration of the receptor Ghyd (R ) , the ion Ghyd (I) , and the receptor–ion complex Ghyd (RI) . The respective treatment yields Eq. (6.1) [5], which shows that the binding process only becomes exergonic if Gint overcompensates the difference between the free energies of hydration of the complex and of its components Ghyd (RI) Ghyd (R ) Ghyd (I) : G

Gint

Ghyd (RI)

Ghyd (R )

Ghyd (I) (6.1)

In Eq. (6.2), the enthalpic and entropic contributions to hydration are separated: G

Gint H hyd (RI) H hyd (R) H hyd (I) T Shyd (RI) Shyd (R) Shyd (I) (6.2)

Hydration enthalpies are typically strongly negative, particularly those of ions with a high charge density and/or multiple charges, and mediated to a smaller extent by an adverse entropic term as shown for selected ions in Table 6.1 [6]. Strongly negative Gibbs free energies of hydration thus result, which cannot easily be overcompensated by a single attractive ionic interaction (under the assumption that the complex is less strongly solvated than the individual components). To achieve ion binding in water, one could therefore substantially strengthen the

195

196

6  Ion Receptors

Table 6.1  Gibbs free energies of hydration Ghyd , hydration enthalpies H hyd , and hydration entropies Shyd of selected cations and anions at 298 K [6].a) Cation +

Li

Na K

−481

+

−375

+

Cs

−304

+

Mg Ca Ba

Ghyd

−258

2+

−1838

2+

−1515

2+

−1258 +

NMe4

+

NBu4

b)

−175

b)

−121

Hhyd −531 −416 −334 −283 −1949 −1602 −1332 −218 −227

Shyd −142 −111

F

−

Cl

−510

−137

−347

−367

−75

−

−321

−336

−59

−59

−

−252 −205 −144 −356

Shyd

−472

Br I

Hhyd

−

−74 −331

Ghyd

Anion

−283

−291

−36

−

−287

−311

−66

−

−306

−312

−76

−214

−246

−57

−473

−522

−166

−1090

−1035

−200

SCN NO3

−

ClO4

−

H2PO4 2−

SO4

a) Ghyd , and H hyd in kJ mol−1, Shyd in J K−1 mol−1. b) Calculated from H hyd and Shyd .

intrinsic receptor–ion interactions Gint by, for example, incorporating multiple charges into the receptor and combining Coulomb attraction with other types of interactions. Respective receptors can indeed be rather efficient, as the examples of sulfonatocalix[4]arenes (see Section 6.2.3.1) or polyammonium‐based receptors (see Section 6.3.2.2) demonstrate. The question that arises is whether strategies exist to achieve ion recognition in case the direct receptor–ion interactions are not sufficiently effective. Equation (6.2) does indeed indicate possibilities. An overall negative ΔG° can also result if favorable enthalpic or entropic contributions from receptor dehydration overcompensate the enthalpy required to desolvate the ion. Large positive entropic contributions to binding can, for example, result from the release of ordered water molecules from the receptor cavity upon guest entry in combination with the entropically favored ion desolvation. Receptor dehydration can also be enthalpically beneficial if it is associated with the release of water molecules that feature less than the optimal number of hydrogen bonds in the cavity as in the case of cucurbit[7]urils (see Section 6.2.1.3). In addition, properties of the ions related to the Hofmeister effect and the reverse Hofmeister effect discussed in Chapter 1 can exert further effects on receptor–ion interactions as demonstrated by the binding of the “superchaotropic” dodecaborate clusters to γ‐cyclodextrin (see Section 6.3.4) or the binding of chaotropic anions inside the cavity of a water‐soluble cavitand and how it is mediated by other anions in solution (see Section 6.3.3). All of these effects are characteristic for the medium water, where the enthalpically or entropically favored reorganization of solvent molecules contributes strongly to complex stability. Water is therefore not a challenging medium for molecular recognition per se. It also offers opportunities that are absent in other solvents, which are, however, not easy to control.

6.2  Cation Receptors

Designing ion receptors can thus be based on different strategies and, consequently, is actually not straightforward. In this chapter, representative examples of ion receptors are presented to illustrate the different approaches. Wherever possible, information about the enthalpic and entropic contributions to binding is provided, which usually affords better insight than stability constants alone into which factors are responsible for complex formation and into the underlying modes of complex formation [7]. These examples were selected from the vast amount of literature with the aim to illustrate the general strategies that can be used to achieve strong and in some cases selective binding of ionic species in water.

6.2 ­Cation Receptors Supramolecular chemistry has been concerned with cation receptors since the seminal work of Pedersen, Lehn, and Cram on crown ethers, cryptands, and spherands, respectively. Since then, cation receptor chemistry represents an important research field, which has also addressed the challenge of developing systems that can efficiently work in water. For the purpose of this overview, cations are classified into three categories: electropositive inorganic cations, transition metal ions, and organic cations. Electropositive cations prefer to bind to receptors with hard binding sites such as the oxygen atoms of ethers or carbonyl groups. Prototypic receptors could derive from crown ethers or spherands. Most transition metal ions feature a rich coordination chemistry based on their interactions with ligands containing one or more Lewis‐basic centers. Although the recognition of toxic transition metals such as lead, cadmium, or mercury is highly relevant in water in the context of environmental monitoring, these ions will not be considered as substrates in this chapter because they can be recognized using the concepts of coordination chemistry without the need to invoke supramolecular chemistry. We refer in this section to transition metals only if they are part of the receptor. Analytical supramolecular strategies to detect transition metal ions in water are presented in Chapter 12. The third category of cations comprises organic cations such as protonated amines or quaternary ammonium ions. Although these ions can come in different structures, they are usually softer than electropositive metal ions, allowing cation–π interactions, hydrogen bonding, or dispersive interactions to ­contribute to complex stability. In all cases, water molecules can exert solvent effects that contribute to complex stability. 6.2.1  Neutral Receptors 6.2.1.1  Crown Ethers and Cryptands

Since their discovery by Pedersen in 1967, crown ethers have developed into one of the best studied classes of cation receptors. Many different crown ether‐ derived cation receptors are known, comprising unsubstituted or substituted crown ethers of different sizes or compounds in which a crown ether moiety is

197

198

6  Ion Receptors

combined with other functional units. Lehn later complemented the crown ether family with bicyclic or tricyclic cryptands. Extensive work in the area revealed several structure–affinity relationships. For example, acyclic systems have a lower affinity than cyclic ones, with cation affinity depending on the size match of the crown ether cavity and the cation. For example, 12‐crown‐4 has the largest affinity for Li+, 15‐crown‐5 for Na+, and 18‐crown‐6 for K+. Moreover, doubly charged cations are bound more strongly than singly charged cations of similar size. The affinity for crown ethers in water is, however, significantly lower than in less competitive organic solvents. Even going from methanol to water decreases the stability of crown ether–metal complexes by a factor of 103–104 as reflected, for example, in the drop of stability of the K+ complex of 18‐crown‐6 from a log Ka of 6 in methanol to a log Ka of 2 in water [8]. This reduction of affinity is partly due to the weakening of the ion–dipole interactions in the solvent with the higher permittivity. Another aspect is that crown ethers are not well preorganized for cation binding in water because the oxygen atoms along the ring preferentially adopt arrangements that allow interactions with the hydrating water molecules (Figure 6.2). Accordingly, bicyclic cryptands that possess a more rigid and better defined cavity form more stable complexes in water. For example, the K+ complex of cryptand‐2.2.2 6.1 (Figure 6.2) has a log Ka of 5.3 in water [8]. 6.2.1.2 Cyclodextrins

Cyclodextrins recognize a wide variety of different substrates in water, including the ionic ones [9]. While there are some examples of complexes in which anionic groups are incorporated into the cyclodextrin cavity (see Section 6.3.4), the tendency of cations to enter this cavity is low. Indeed, the inner surface of cyclodextrins has a weak positive potential, which does not favor the incorporation of a positively charged guest. Cationic guests for cyclodextrins therefore typically contain a larger nonpolar substituent, which is bound inside the cavity. 1‐ Adamantyltrimethylammonium is an example, which binds to β‐cyclodextrin with a log Ka of 3.6 in H2O at pH 8.6. For comparison, the complex of 1‐adamantanecarboxylate has a log Ka of 4.0 under the same conditions with a similar thermodynamic signature of complex formation, showing that binding is dominated by the incorporation of the adamantyl unit in the cyclodextrin ring. Similarly, the β‐cyclodextrin complexes of ferrocenecarboxylate and ferrocenylalkyldimethylammonium have log Ka values of 3.3 and 3.6, respectively [9]. O

O O

O O O

O

O

O

Preferred conformation in water

O

O N

O

O O

Preferred conformation in the complex

N O

O

O

O 6.1

Figure 6.2  Conformational flexibility of crown ethers and structure of cryptand‐2.2.2 (6.1).

6.2  Cation Receptors

6.2.1.3 Cucurbiturils

The converging arrangement of carbonyl groups at the two openings of cucurbiturils renders these macrocycles ideally predisposed for the complexation of cationic guests [10]. Indeed, alkali metal ions and alkaline earth metal ions were shown to form complexes with cucurbit[6]uril 6.2 (Figure 6.3) in 1 : 1 (v/v) formic acid/water with binding constants log Ka ranging between 2.4 and 3.2 [11]. The low selectivity and the absence of an obvious trend that correlates binding affinity with the size and/or the charge of the ion indicated, however, that the arrangement of the C═O groups is not optimal for metal coordination. Nevertheless, the solubility of 6.2 is typically higher in solutions containing salts such as Na2SO4, LiCl, KCl, CsCl, or CaCl2 at millimolar concentrations, which was attributed to interactions of the corresponding metal ions with the carbonyl groups at both portals of 6.2 [10]. In contrast to metal ions, organic amines, or diamines exhibited pronounced affinities for 6.2 and clear selectivities. Figure 6.3 shows how the log Ka values of the complexes of 6.2 with n‐alkylamines and α,ω‐ diamines depended on the lengths of the respective alkyl chains in 1  :  1 (v/v) formic acid/water [12]. In this solvent, the amines were fully protonated, and complex formation therefore involved binding of the respective ammonium ions. According to the graph in Figure 6.3, n‐butylamine formed the most stable complex among the n‐ alkyl amines, while a chain length of five to six methylene units between the head groups was optimal for α,ω‐diamines. These results were attributed to the complementarity between the cavity size of 6.2 and the length of the hydrophobic chains of the guests: the butyl group had the optimal length to fill the cavity of 6.2 if the protonated amino group of n‐butylamine was bound at one portal. Shorter alkyl amines required the additional hydration of the cavity by solvent molecules, while the alkyl chains of longer alkyl amines projected into the solution at the portal of 6.2 opposite to the one where binding of the ammonium

N N NN N N O O O OO

6.3

O N N N NN N N N N N NN OO

N N O

N N N N NN

6

NN N N N N

5

OO O

O

O O O N N NNN N N N N

O

7

O

O

N NN N N OO

log Ka

6.2

OO

OO O N N N NN N

4 3 2 1 0

1

2

3 4 5 6 7 Chain length n

8

9 10

Figure 6.3  Structures of cucurbit[6]uril (6.2) and cucurbit[7]uril (6.3) and graph showing how the log Ka values of the complexes of 6.2 with n‐alkylamines CH3(CH2)nNH2 (●) and α,ω‐diamines H2N(CH2)nNH2 (■) in 1 : 1 (v/v) formic acid/water depend on the lengths of the corresponding alkyl chains.

199

200

6  Ion Receptors

group took place and solvation of this part of the alkyl chain was energetically unfavorable. In the case of α,ω‐diamines, five or six methylene units between the head groups were optimal to allow the simultaneous binding of both ammonium groups to the two portals of 6.2 in a strain‐free arrangement. The graph in Figure 6.3 also shows that very high stability constants approaching a log Ka of 7 could be easily achieved under these conditions, indicating that cucurbiturils are privileged hosts for the binding of organic cations in aqueous solution. The larger homolog of 6.2, cucurbit[7]uril 6.3, displayed even higher affinities: log Ka values up to 15.5, thus exceeding the stability of the biotin–avidin complex, were observed, for example, for the complex between 6.3 and 1,1′‐ bis(trimethylammoniomethyl)ferrocene in water (Figure  6.4) [13]. Later it was shown that adamantane and bicyclo[2.2.2]octane derivatives with ammonium groups reached similar high affinities [14]. An attomolar dissociation constant (log Ka = 17.9) was determined for the complex between 6.3 and a diamantane quaternary diammonium ion [15]. Formation of these complexes in water was invariably associated with a substantial gain in enthalpy accompanied by smaller favorable or unfavorable entropic contributions. Work by Biedermann and Nau showed that this thermodynamic signature can be explained by the release of water molecules from the cavity of 6.3 upon complex formation [16]. This cavity‐­ bound water features less than the average number of hydrogen bonds found in bulk water. The recovery of hydrogen bonds of these water molecules therefore explains the pronounced exothermicity of complex formation. Cucurbituril 6.3 is unique in that the overall number of hydrogen bonds regained, while releasing the cavity‐bound water (number of included water molecules × hydrogen bond deficit per included water molecule) is particularly large, but the binding of other receptors also benefits from the release of “high‐energy water,” which is therefore a general contributor to complex stability and we will come back later to this aspect in Section 6.3.3. The larger cucurbit[8]uril is able to bind two aromatic guest simultaneously with high binding affinities (both stepwise binding constants 103–106 M−1 in water) [17]. Typical guest pairs for 1 : 2 homoternary complexes are two ­positively

N

NH3

H3N

N

H2N

Fe H3N

N

N log Ka

15.5

14.3

15.7

17.9

H° (kJ mol–1)

–87.0

–65.3

–84.1

Not determined

T S° (kJ mol–1)

+2.0

–16.3

–5.9

Not determined

Figure 6.4  Structures of guest molecules that form highly stable complexes with 6.3 and thermodynamic parameters associated with their formation in water at 298 K.

6.2  Cation Receptors

charged tryptophan moieties or two (positively charged) anthracene moieties. Even more common are 1 : 1 : 1 heteroternary complexes where one dicationic species such as paraquat acts as the first guest and a non‐charged aromatic compound such as naphthalene as the other one. 6.2.1.4 Cavitands

The Rebek group introduced in 1997 the deep cavitand 6.4a (Chart 6.1), which adopted a C4v symmetric vaselike conformation in organic solvents, stabilized by intramolecular hydrogen bonds between the NH and C═O groups along the rim [18]. This receptor was shown to bind hydrophobic guest molecules such as adamantyl derivatives in chloroform. Guest exchange was slow on the NMR timescale because it required disruption of several hydrogen bonds and the concomitant folding away of a side wall. These cavitands were subsequently converted into water‐soluble derivatives by replacing their hydrophobic “feet” with sugar‐derived substituents (6.4b), oligoethylene units (6.4c), or other solubilizing groups [19–21]. Intramolecular hydrogen bonding was not efficient enough to stabilize these cavitands in their vase conformations in water. They therefore preferred the kite form, which further dimerized to bury the lipophilic surfaces from the solvent. Appropriate guest molecules, including ammonium ions, caused the cavitands to rearrange into the vase conformation, whereby the hydrophobic parts of the guests were preferentially incorporated into the cavity, while the positively charged ammonium groups were oriented toward the cavity opening. Interestingly, guest exchange was slow on the NMR timescale also under these conditions. R2 R2 O O 2HN H R N R2 NH

O

R2 O R2 R2 H OO HN N O H NH N H N

O O O

O

O

O

O R1

R1

R1

6.4a (R1 = C11H23, R2 = C7H15) 6.4b (R1 =

N N N

NaO2C CO2Na

R2

N

NH CO2Na N N

HN O

O

O

R1

HO O

N HN

NaO2C

O

O

O O

O Et

Et

NH

O

Et Et

6.4d 2 OH , R = Et) OH OH

6.4c (R1 = (CH2)3(OCH2CH2)4OMe, R2 = Me)

Chart 6.1 

The cavitand 6.4d containing four solubilizing groups along the cavity opening (rendering it negatively charged and not neutral) existed in a vase conformation

201

202

6  Ion Receptors

in water also in the absence of guests, likely thanks to water molecules holding the side walls in the converging arrangement by bridging the imidazole nitrogen atoms. This cavitand bound acetylcholine with a log Ka > 4 in D2O, whereby the trimethylammonium group of the guest was included into the cavitand cavity [19]. Interestingly, dodecyltrimethylammonium bromide also bound to 6.4d with an affinity similar to that of acetylcholine but with the alkyl chain included into the cavity where it adopted a helical arrangement. The high affinity of 6.4d for substrates with trimethylammonium groups even under physiological conditions allowed the transport of cationic substrates by cavitand‐mediated endocytosis into living cells [22]. 6.2.2  Negatively Charged Receptors 6.2.2.1 Cyclophanes

Inoue and coworkers investigated the interaction of the anionic cyclophanes 6.5a–c (Chart 6.2) with phenethylamine, tyramine, and dopamine, which belong to the family of catecholamine neurotransmitters [23]. O OOC

N H

N N

H N

OOC O

O

O X

X

COO

N H

6.5a

6.5b

N COO O

NH3

R2

N

H N

R1

Phenetylamine (R1 = R2 = H) Tyramine (R1 = H, R2 = OH)

6.5c

Dopamine (R1 = R2 = OH)

Chart 6.2 

The respective binding studies were performed at pD 8.0 at which the carboxylate groups of the hosts were in the anionic form and the amino groups of the guests protonated to allow for electrostatic interactions. The binding was modest with stability constants Ka around 20 M−1 and little selectivity between the structurally similar phenethylamines. In an attempt to systematically evaluate the contribution of cation–π interactions to the cation affinity of cyclophane‐derived hosts, the Dougherty group quantified the affinity of receptor 6.6 (Chart 6.3) and related derivatives toward more than 70 potential guests in 10 mM aqueous borate buffer (pH 9) [24]. In this receptor, the peripheral carboxylate groups mainly served to mediate water solubility and were not expected to contribute to a large extent to complex stability because of their arrangement away from the ­cavity. Host 6.6 was found to strongly interact with ammonium ions, with the binding constants log Ka varying between 3 and 6, depending on the structure  of the guest. The stability of the N‐methylquinolinium complex of 6.6 amounted to a log Ka of 6.2, for example, and that of the acetylcholine complex to a log Ka of 4.5.

6.2  Cation Receptors OOC COO S

O

O

OOC COO

S

OOC S S

OOC OOC O

O S OOC COO 6.6

COO S

OOC

COO

S S

S S

COO

S S OOC

COO

6.7

6.8

(Mixture of stereoisomers)

(Mixture of stereoisomers)

Chart 6.3 

Dithiol analogs of the 9,10‐ethenoanthracene building blocks that were employed for the construction of 6.6 were used by Sanders and Otto to identify cation‐binding cyclophanes in a dynamic combinatorial chemistry approach. The templation of a dynamic library generated from these and other dithiols with ammonium ions afforded the cyclophanes 6.7 and 6.8 as mixtures of stereoisomers [25]. The cation affinity of these cyclophanes was similar to those of 6.6. The log Ka of the complex of 6.8 with the N‐methylisoquinolinium ion amounted to 5.3 and that with acetylcholine to 3.8, for example. In most cases, binding was enthalpically favorable (cation–π interactions) and promoted by entropy (solvent effects). In a related approach, the Waters groups used the same building blocks to identify receptors that bind with high affinity to methylated side chains of lysine or arginine, allowing them to serve as markers for the identification of histone posttranslational modifications (PTMs) [26]. Interestingly, templation of the respective dynamic library with the dipeptide Ac‐LysMe3‐Gly‐ NH2 afforded the same cyclophane 6.8 that was also found by Sanders and Otto. The affinity of 6.8 for H3 histone tail peptides with a trimethylated lysine residue amounted to a log Ka of 4.6 in 10 mM phosphate buffer (pH 8.5). Di‐ and mono‐methylated analogs were bound weaker with log Ka values of 4.2 and 3.8, respectively. This work was subsequently extended to dimethylated arginine as substrate [27] and to a wider range of structurally related receptors [28]. Based on similar 9,10‐bridged anthracene motifs, the water‐soluble macrocyclic tris‐triptycene 6.9 (Chart 6.4), termed 2,6‐helic[6]arene, was developed [29]. This compound was shown to bind phosphonium ions in water. The complex of the tetramethylphosphonium cation had a stability log Ka of 5.8, while the complex with the n‐hexyltrimethylammonium cation was 1 order of magnitude more stable (log Ka 6.7), indicating that hydrophobic interactions contributed to complex stability. The host–guest interactions in combination with solvent effects caused complex formation to be strongly exothermic with favorable or unfavorable entropic contributions depending on the guest.

203

204

6  Ion Receptors

RO

O OR

RO

OR (CH2)3

OR

O RO

RO

OR O (CH2)3

O (CH2)3

O

OR

RO O

RO OR 6.10 (R = CH2COO

6.9 R = CH2CO2NH4

)

Chart 6.4 

6.2.2.2 Cryptophanes

The Collet group investigated the interaction of quaternary ammonium ions with water‐soluble cryptophanes [30]. They showed that the tetramethylammonium ion interacted with cryptophane 6.10 (Chart 6.4) with a log Ka of 5.4 in D2O (pD = 6.5). Binding was believed to be due to a combination of cation–π interactions and the release of water molecules from the host and the guest. 6.2.2.3 Calixarenes

Sulfonatocalixarenes such as 6.11a (Chart 6.5) and other anionic calixarenes have found widespread applications. The Coleman group, for example, studied their affinity for amino acids, peptides, and proteins and the biological effects associated with the binding of these biorelevant substrates [31]. Sulfonatocalixarenes were also shown to bind to neurotransmitters such as acetylcholine in water [32]

OOC R

NaO3S NaO3S

SO3Na SO3Na

OR OR OR RO

NaO3S NaO3S

SO3Na

OH OH OH HO

6.11a R = H

6.12a R = H

6.11b R = OC4H9

6.12b R = OMe 6.12c R = CN 6.12d R = CONH2

Chart 6.5 

NaO3S NaO3S

SO2 SO3Na NH

OH OH OH HO 6.13

6.2  Cation Receptors

or nerve agents such as VX [33]. The Liu group studied the interaction of 6.11a with bisguanidinium guests, which are important for treatment of hyperglycemia in patients with non‐insulin‐dependent diabetes mellitus [34], and with viologens such as the herbicides paraquat and diquat [35]. One has to consider that the unavoidable counterions (usually Na+) of sulfonatocalixarenes potentially interfere in substrate recognition and that the competition of such cations may be even more pronounced in buffered solution. Bakirci et  al. indeed showed by a displacement assay that metal ions interact with 6.11a with binding constants between 80 and 280 M−1 for monocations and higher binding constants for ions with multiple charges [36]. Thus, most reported binding constants should be regarded as apparent binding constants, which could eventually end up to be different by using a competitive model for their ­evaluation [37]. Alkylation of the phenol OH groups of 6.11a strongly influences the affinity and the thermodynamics of cation binding, a relationship that was systematically investigated by Liu and Nau by comparing the affinities of 6.11a and of the tetrabutylated derivative 6.11b (Chart 6.5) toward a series of cationic and neutral guests [38]. They showed that the tetramethylammonium ion bound preferentially to 6.11a, which was attributed to the almost spherical shape of the calixarene cavity that allowed the cation to be included efficiently. The butylated derivative 6.11b, in contrast, preferred a pinched‐cone conformation not well suited to incorporate bulkier cations. The inferior fit of the NMe4+ ion into the cavity of 6.11b caused the favorable enthalpic contribution to complex formation to be lower than that observed for the corresponding complex of 6.11a, whereas the binding entropies were both favorable but small and of comparable size. The Hof groups showed that sulfonatocalixarenes can be used to identify posttranslational protein modification [39]. The parent compound 6.11a recognized trimethyllysine with a log Ka of 4.6 in D2O containing 100 mM Na2HPO4/ NaH2PO4 at pH 7.7. Binding was enthalpically favorable and accompanied by a small favorable entropic component, with the extent of the enthalpic driving force correlating with the degree of lysine methylation [40]. Based on the assumption that additional aromatic substituents should be beneficial for complex formation, a phenyl ring was incorporated into 6.11a to yield 6.12a (Chart 6.5). 6.12a did indeed bind to trimethyllysine with an improved log Ka of 4.9. The weaker affinity of the substituted derivatives 6.12b, 6.12c, and 6.12d suggested, however, that the influence of the appended substituents mainly derived from hydrophobicity rather than cation–π interactions. Further work showed that calixarene 6.13 disrupted the interaction of the reader domain of the helicase DNA‐binding protein 4 (CHD4) to histone 3 [41]. 6.2.2.4 Pillararenes

The most prominent representative of pillararenes, pillar[5]arene, features five hydroquinone‐derived subunits whose aromatic faces are preferentially arranged almost parallel to the main axis of the ring, causing the two openings of the cavity to have the same diameters, similar to cucurbiturils but different from ­calixarenes. Thus, guest molecules can be threaded through the ring, which can interact with

205

206

6  Ion Receptors

the electron‐rich aromatic subunits and also with the substituents appended to them. Typical guest molecules contain electron‐deficient aromatic systems or cationic groups to allow for aromatic and cation–π interactions. The binding of charged guests can be strengthened by attaching oppositely charged groups to the aromatic subunits of pillararenes, which also mediate water solubility. Accordingly, anionic water‐soluble pillararenes display affinity for cationic guests in water, while anions can be bound by pillararenes with cationic solubilizing groups. The anionic pillar[5]arene 6.14 (Chart 6.6) was shown to bind paraquat [42] and amino acids with positively charged side chains (e.g. Arg, Lys, and His; log Ka ~ 3 in D2O) [43]. Increasing the cavity size and the number of negatively charged substituents in the periphery caused paraquat affinity to improve from a log Ka of 4.9 for 6.14 to a log Ka of 8 for 6.15 [44] and further to a remarkable log Ka of 9.5 for 6.16 in water [45]. Complex stability was believed to be due to a combination of electrostatic interactions between the negatively charged groups along the pillararene rims and the cationic guests as well as hydrophobic and π–π stacking interactions. Non‐charged aliphatic alcohols and aromatic guests were only weakly bound in aqueous solution (Ka ≤ 500 M−1), potentially also because the contribution of the “high‐energy water” release is comparably small for pillararenes [16]. OR

OR

OR

RO

RO RO O R

RO

RO

OR R O

RO

OR RO

OR OR

RO

RO

OR

OR

RO

RO

OR

6.14 (R = CH2COONH4)

RO 6.15 (R = CH2COONa)

OR OR

OR OR

RO

RO

OR

OR

RO

OR

RO RO 6.16 (R = CH2COONH4)

Chart 6.6 

6.2.2.5  Molecular Tweezers

Schrader demonstrated the binding of bisphosphonates such as 6.17, 6.18, and 6.19 (Chart 6.7) to amino alcohols (as their ammonium salts), for example, adrenaline, or guanidinium groups by the combination of electrostatic interactions and multiple hydrogen bonds [46]. In the case of 6.19, the binding motif of the complex with the guanidinium group of arginine resembled the “arginine fork” postulated as a key element in RNA–protein recognition of the AIDS virus. Unfortunately, the respective interactions turned out to be too weak to survive in water; the respective binding studies were mostly performed in DMSO. Bell et al. showed that the rigidification of such dianionic structures combined with the introduction of further binding sites improved the affinity for guanidinium groups to such an extent that complex formation was also observed in

6.2  Cation Receptors

2NBu4 MeO O O P

O

OMe P O MeO O P O

6.17

2NBu4

MeO O O P O OMe P O

2NBu4

O

O

OMe P O O

O

6.19 O

OH

O

O OMe P O O

N

O

OH NH3

6.20

O

4Li

N N

O

OH

2K N

O O P MeO O

O

6.18 O

O

OMe O P O

Adrenaline

6.21

O P O OMe

Chart 6.7 

water [47]. In the corresponding receptor 6.20 (Chart 6.7), two oxygen atoms on the carboxylate groups and the two nitrogen atoms in the naphthyridine moiety served as hydrogen bond acceptors for the protons of the guanidinium group. 6.20 was water‐soluble and was shown by isothermal titration calorimetry (ITC) to bind arginine in an exothermic process with a log Ka of 3.0. The complex with lysine, whose side‐chain amino groups was structurally less compatible with the cleft of 6.20, was bound weaker by about 1 order of magnitude (log Ka 2.1). Another possibility to transfer the affinity of bisphosphonate 6.18 for amino alcohols into water comprised the shielding of the binding site from the surrounding medium by incorporation into a macrocyclic structure. This concept was realized by using receptor 6.21 (Chart 6.7), containing two bis(phosphonate) moieties on opposite sides of the ring [48]. 6.21 expectedly formed 2 : 1 complexes with two guest molecules incorporated into the cavity. The two binding steps associated with complex formation were typically similar in magnitude, amounting to a log Ka of 3.1 for the 1 : 1 complex between 6.21 and adrenaline, for example, and an overall log Ka of 6.2 for the respective 2 : 1 complex. A related macrocyclic tetraphosphonate bound lysine in water with a log Ka of 3.1 in the form of a 1 : 1 complex in which the two amino groups of the guest bridged two pairs of phosphonate groups of the host [49]. The concept of introducing phosphonate or phosphate groups along an aromatic scaffold was extended to the molecular clips and tweezers developed in the Klärner group. These compounds were developed to bind electron‐poor aromatic and cationic guests between their aromatic side walls in chloroform [50]. The introduction of phosphonate groups into the central subunits afforded the analogs 6.22 and 6.23 (Chart 6.8), which allowed binding studies in water. The anionic groups in these receptors should allow 6.22 to engage in similar interactions as those in 6.18. Indeed, 6.22 interacted with benzylammonium ions in DMSO, but the absence of significant effects of complex formation on the resonances of the aromatic guest protons in the 1H NMR spectra indicated that the guest interacted with the anionic groups from the outside and did not enter

207

208

6  Ion Receptors

O O P

O O P

O

O

O O

O

P

O

2NBu4 6.22

O

P

O 2Li 6.23

Chart 6.8 

the cavity [51]. Pyridinium derivatives, however, bound strongly to 6.22 by intercalation between the aromatic side walls; the complex between 6.22 and N‐ methylnicotinamide, for example, had a log Ka in D2O of 4.9. Interestingly, also NAD+ or NADP were recognized with appreciable affinity, and it was assumed that the nicotinamide moiety of both substrates bound between the receptor side walls, while the adenine residue interacted with one side wall from the outside. The phosphonate groups thus mainly served as solubilizing groups, potentially also mediating electrostatic interactions to some extent, but not as actual binding sites as in 6.17–6.19. The term tweezer therefore referred to the arrangement of the aromatic units in 6.22 and 6.23. The more spherical tweezer 6.23 efficiently recognized the cationic groups in the side chains of lysine and arginine in water by threading them through the receptor opening [52]. Binding was more efficient for lysine that was bound with a log Ka of 4.4 in D2O in the form of the protected derivative AcLysOMe. In 25 mM phosphate buffer, the log Ka dropped to 3.6. Arginine was bound slightly less efficiently, with the log Ka of the TsArgOEt complex amounting to 3.9 in D2O. Interestingly, also short peptide sequences with Lys or Arg residues could be bound effectively. The respective complexes were believed to be partly stabilized by hydrogen bond formation between the external phosphonate moieties and NH groups along the peptide chain. Because of the unique ability of tweezers such as 6.23 to interact with lysine and arginine residues exposed on protein surfaces, it was tested whether these compounds could prevent the aggregation or even dissolve aggregates of pathological proteins such as those involved in Alzheimer’s disease, Parkinson’s disease, and other disorders with highly promising results [53]. 6.2.2.6  Acyclic Cucurbiturils

The Isaacs group introduced carefully designed acyclic cucurbiturils as a potent new receptor family. Examples are compounds 6.24 and 6.25 (Chart 6.9), which feature a central glycoluril tetramer and two terminal aromatic moieties containing appended sulfonate groups to mediate water solubility. These compounds were shown to be nontoxic and able to enhance the water solubility, chemical

6.2  Cation Receptors NaO3S

O

O

O

N

N

N

N

O

N

N

N

N

H NaO3S

O

NaO3S

O

N

N

N

N

N

N

6.24

O

O

O

O

O

O

N

N

N

N

N

N

O

N

N

N

N

HH O

SO3Na

H

O

N

O

N

O

N

O

N HH

H NaO3S

O

N

N

N

N

O

SO3Na

O

SO3Na 6.25

H O

O

O

SO3Na

Chart 6.9 

stability, and bioavailability of a number of drugs [54]. Medicinal applications of these compounds are therefore intensively sought for. In addition, acyclic cucurbiturils were also shown to bind cations in water, albeit not as strongly as their cyclic counterparts. For example, the stability of the acetylcholine complex of 6.25 in 20 mM NaH2PO4 buffer (pH = 7.4) amounted to 5.3 [55]. 6.2.3  Metal‐Containing Receptors Metals can serve as organizing elements in ion receptors to stabilize an appropriate spatial arrangement of organic ligands around the cavity and can simultaneously mediate electrostatic interactions with the bound guest. In the case of cation receptors, attractive interactions require that the total positive charges of the metal centers are compensated or even overcompensated by surplus negative charges of the ligands. Two major types of metal‐containing receptors have been realized that can serve for cation recognition, macrocyclic and cage‐type receptors. 6.2.3.1 Metallacycles

In macrocyclic receptors, the metal centers stabilize a cyclic arrangement of oxygen atoms along the cavity, which engage in interactions with electropositive metal ions. Because of their relationship with crown ethers, the respective metal‐ containing counterparts have been termed metallacrowns. They were first reported in 1989 by Pecoraro and Lah [56] and can often be synthesized in excellent yields from simple starting materials [57]. One example is receptor 6.26a described by the Severin group. It assembled as an overall neutral compound when treating [(cymene)RuCl2]2 with 3‐hydroxy‐2‐pyridone in the presence of base (Chart 6.10) [58]. When shaking an aqueous solution containing a mixture of LiCl (50 mM) together with a large excess of competing salts (1 M of each NaCl, KCl, CsCl, MgCl2, and CaCl2) with a solution of 6.26a in chloroform, the quantitative extraction of only LiCl was observed in spite of the high hydration enthalpy of Li+ and Cl− (Table 6.1) and the significantly easier dehydration of the other alkali metals. The exclusive formation of the LiCl adduct therefore indicated a particularly high selectivity of 6.26a.

209

210

6  Ion Receptors

O

R

O

Ru N O

N

O O

Ru

O

O

O

R

O

NH O

NH

O Ru N

HN

O

O

O HN O

R

O

O

O 6.26a R = H 6.26c R = 6.26b R =

N

6.27

N

[Ga4·6.276]12– (

= Ga3+)

N H

Chart 6.10 

Given the pharmacological relevance of lithium salts, efforts were directed at increasing the water solubility of 6.26a. To this end, tertiary amino groups were introduced in the periphery of the metallacrown that should mediate solubility at neutral pH without interfering in the self‐assembly process [59]. The respective receptor 6.26b complexed Li+ in buffered aqueous solution (100 mM phosphate buffer, pH 7.0) with a Li+/Na+ selectivity of 10 000/1 and a log Ka of 3.4, 3 orders of magnitude higher than the Li+ affinity of 12‐crown‐4. The binding constant was found to depend on the nature of the metal center and on the pH [60]. The analog 6.26c with three piperazine substituents bound Li+ with a log Ka of 4.8 in water, which is among the highest ever reported Li+ affinities in this medium [61]. 6.2.3.2  Coordination Cages

Highly effective cation‐binding coordination cages were developed in the Raymond group. The cage [Ga4·6.276]12− (Chart 6.10) in which six bis‐bidentate catecholamides 6.27 bridge four Ga3+ centers at the vertices of a tetrahedron is one example [62]. The 12 positive charges of the four metal ions are overcompensated by the 24 negative charges of the ligands, rendering the cage overall 12‐fold negatively charged. As a consequence of the tris‐bidentate coordination at each metal center, [Ga4·6.276]12− is chiral and is typically formed during the assembly as a racemate. It is water‐soluble and features a hydrophobic cavity with a volume between 0.35 and 0.5 nm [3], sufficiently large to accommodate a variety of organic monocations such as protonated amines or quaternary ammonium or phosphonium ions. Protonated amines can be efficiently stabilized within the cavity where they exist as ammonium ions even if the pH value of the surrounding medium would normally lead to deprotonation. In quantitative terms, this stabilizing effect results in a shift of the effective basicity of the amines by 2–4 orders of magnitude. As a consequence, [Ga4·6.276]12− can

6.3  Anion Receptors

c­atalyze acid‐catalyzed reactions involving positively charged intermediates even at basic pH, as outlined in Chapter 14. Microcalorimetric investigation provided information about the thermodynamics of the interaction of [Ga4·6.276]12− with NEt4+ ions, revealing striking differences for external and internal complexation [63]. External binding was favored by attractive electrostatic interactions between the guests and the exterior surface of the cage (ΔH°  TfO− > ClO4− > CB11H12− > Tf2N− > BF4− > I− > NO3−.

N N N O

N

NH

NH

N

N

N N

N

P H

N

N

N N [Fe4 6.284]

12+

(

=

Fe2+)

N [Ni4 6.296]8+ (

= Ni2+)

Chart 6.11 

Another example of a coordination cage with inwardly directed binding sites is  [Ni4·6.296]8+ (Chart 6.11), which contained the urea‐derived ligand 6.29 to bridge the four Ni2+ centers [72, 73]. This cage fully saturated the hydrogen bond acceptor sites of an included sulfate anion by forming twelve hydrogen bonds

213

214

6  Ion Receptors

between the urea NH groups and the sulfate oxygen atoms. These hydrogen bonds in combination with electrostatic effects mediated by the four positively charged nickel(II) ions induced high sulfate affinity in water. Complex stability could not be determined exactly, but precipitation experiments indicated that the complex has a log Ka > 6.8 in water. 6.3.1.2  Tetraazamacrocycle‐Based Receptors

The affinity and selectivity of metal‐containing receptors in which the anion directly coordinates to the metal center strongly depends not only on the metal itself but also on the ligand that serves to incorporate the metal into the receptor framework. Macrocyclic polyamines such as 1,4,8,11‐tetraazacyclotetradecane 6.30 (cyclam) and 1,4,7,10‐tetraazacyclododecane 6.31 (cyclen) are particularly useful in this context (Chart 6.12). Both compounds form highly stable complexes with Cu2+ and Zn2+ in water ([Cu·6.30]2+, log Ka  =  27.2; [Zn·6.30]2+, log Ka  =  15.5; [Cu·6.31]2+, log Ka  =  24.8; [Zn·6.31]2+, log Ka  =  16.2) [74]. The respective complexes typically have square pyramidal geometries with the nitrogen atoms at the base and one apically bound anion (Figure 6.5). Their high thermodynamic stability ensures that they are not demetalated even in the presence of an excess of anions and the lability of the metal–anion bonds allows facile anion exchange [68]. HN

NH

HN

NH

HN

NH

6.30 HN

6.31

NH

Chart 6.12 

Monometallic cyclen complexes of lanthanide(III) ions were extensively investigated in the Parker group as optical anion probes (see Chapter 12) [75]. Most cyclen‐derived anion receptors feature, however, two or more anion‐binding units. The Kimura group, for instance, appended [Zn·6.31]2+ moieties to an aromatic core and showed that the dizinc(II) complex [Zn2·6.32]4+ (Chart 6.13) R

R [Zn2

R

6.32]4+

(R =

R1)

R [Zn3

R

6.33]6+

[Cu3 6.34]6+ (R = R2) R1

R2

= N HN

Zn2+

NH

NH

Chart 6.13 

H N

(R = R1)

HN Cu 2+ N

HN

N H

= N 2+ NH Cu HN NH

[Cu2

6.35]4+

N H Cu2+ N HN

6.3  Anion Receptors

bound p‐nitrophenyl phosphate with a log Ka of 4.0, about 1 order of magnitude more strongly than [Zn·6.31]2+ (log Ka  =  3.3). The most stable complex was formed by the tritopic analog [Zn3·6.33]6+ that recognized the same substrate with a log Ka of 5.8 at slightly acidic pH (pH  7. Based on these findings, Hamachi developed a variety of optical probes for phosphoproteins and nucleoside polyphosphates, biological assays to monitor kinase, phosphatase or glycosyltransferase reactions, and an inhibitor assay for the phosphoprotein–protein surface interaction. The Smith group showed that receptor [Zn2·6.43]4+ can also interact with bilayer membrane surfaces enriched in anionic phospholipids such as phosphatidylserine [89]. This system was used to establish fluorescent and flow cytometry assays to detect mammalian cells undergoing apoptosis (see Chapter 13.5.1). Work in the Jolliffe group showed that cyclic and linear peptides with ZnDPA units are versatile receptors for the recognition of the pyrophosphate anion [90]. Binding was mostly studied by using indicator displacement assays to allow optical detection of the binding event. These investigations showed that the disubstituted cyclic peptide [Zn2·6.44]4+ (Chart 6.16) bound pyrophosphate highly selectively over ADP, ATP, or citrate. Anions such as acetate, nitrate, sulfate, iodide, or phosphate did not interfere and even monophosphate esters, which were shown to interact with [Zn2·6.42]4+ and [Zn2·6.43]4+, did not produce color changes. The stability of the pyrophosphate complex proved to be too high to allow quantification (log Ka > 9). The apparent stability constants of the complexes with ADP, ATP, and citrate in 5 mM HEPES (pH 7.4) in the presence of 145 mM NaCl amounted to 7.4, 7.2, and 5.7, respectively. The weaker binding of these anions was attributed in part to their lower overall charges in comparison to pyrophosphate in combination with steric effects of the cyclopeptide side chains that interfered with the larger organic anions.

N

N

N

N

N N

N

Zn2+

Zn2+

N

Zn2+ N

O

O N O

N

N N

[Zn2 6.42]4+

N

Zn2+ N

N

[Zn2 6.43]4+

O

N N

HN

NH N

Zn2+ N

Zn2+

O

N HN NH O

O

N O

[Zn2 6.44]4+

Chart 6.16 

The variation of the structure of 6.44 by changing the positions of the ZnDPA units along the ring, the nature of the other substituents, or ring size had little effect on pyrophosphate affinity although it usually resulted in the modulation of the affinity for other anions [90]. The Jolliffe group subsequently also showed that linear peptides with ZnDPA units, which were synthetically much easier accessible than cyclopeptides, could also be used for pyrophosphate recognition. Importantly, pyrophosphate recognition proved to be so efficient that it also

217

218

6  Ion Receptors

worked in Krebs saline, a medium that mimics biological fluids, in some cases even with improved selectivity over competing anions such as ADP and ATP. 6.3.1.4  Tris(2‐aminoethyl)amine and Tris(2‐pyridylmethyl)amine‐Based Receptors

Tris(2‐aminoethyl)amine 6.45 and tris(2‐pyridylmethyl)amine 6.46 (Chart 6.17) contain one arm more than the ligands described in the previous section. They thus occupy four coordination sites on a metal, which prevents the formation of higher complexes. The Cu2+ and Zn2+ complexes of 6.45 have stability constants log Ka of 18.8 and 14.6, respectively [91], whereas a log Ka of 16.2 was determined for the Cu2+ complex of 6.46 and a log Ka of 11.8 for the respective Zn2+ complex [92]. Thus, the metal complexes of these ligands are significantly more stable than those of 6.36 and 6.37. NH2 NH2 N

NH2

6.45

N N N

N

6.46

Chart 6.17 

Both ligands mostly form trigonal bipyramidal complexes with Cu2+ and Zn2+ in which the equatorial positions and one apical position are occupied by the ligand nitrogen atoms and the second apical position by the counterion [68]. These complexes are therefore ideally predisposed for anion recognition. Indeed, the zinc(II) complex [Zn·6.46]2+ bound phenyl phosphate in 50 mM HEPES buffer (pH 7.0) with a log Ka of 3.6 [93]. The metal centers in the structurally more elaborate receptors [Cu·6.47]5+ and [Cu·6.48]5+ (Chart 6.18), developed in the Anslyn group, were expected to act as the primary binding site, and the additional substituents were introduced to reinforce complex stability by Coulomb attraction and hydrogen bonding [94]. Indeed, with log Ka values of 3.9 and 4.8, the hydrogen phosphate affinities of, respectively, [Cu·6.47]5+ and [Cu·6.48]5+ were higher than that of [Cu·6.46]2+ (log Ka = 3.0) in 2 vol% methanol/water (5 mM HEPES buffer, pH 7.4). The guanidinium and ammonium groups in the substituted receptors moreover influenced the thermodynamics of complex formation differently. For both receptors, binding was observed to be exothermic and favored entropically. However, in the case of [Cu·6.47]5+, enthalpy dominated the complex formation, and in the case of [Cu·6.48]5+, entropy [94]. These effects were attributed to differences in the solvation of the positively charged functional groups of both receptors. The ammonium groups in [Cu·6.47]5+ were expected to feature tighter bound water molecules than the guanidinium groups in [Cu·6.48]5+, causing desolvation of the former to be enthalpically more difficult. Both receptors bound arsenate and phosphate with similar affinity but did not or not strongly interact with other anions such as acetate, sulfate, nitrate, hydrogencarbonate, or chloride.

6.3  Anion Receptors

Mononuclear copper(II)‐containing receptors were also designed that allowed fluoride recognition in water in spite of the strong hydration of the fluoride anion. Fluoride recognition required, however, the incorporation of the binding site into the interior of a calix[6]arene ring as in [Cu·6.49]5+ since the simple copper complex [Cu·6.46]2+ (Chart 6.18) did not interact with fluoride [95]. The stability of the fluoride complex of [Cu·6.49]5+ amounted to a log Ka of 4.9 in 30 mM MES buffer (pH 5.9), which was significantly larger than that of the chloride complex (log Ka = 1.9). According to computations, fluoride was bound in the complex together with two water molecules. NH

HN HN M

N 2+

N N

H2N

HN N Cu2+

N

N

N

[Zn·6.46]2+ (M = Zn) [Cu·6.46]2+ (M = Cu)

H3N NH3

HN

NH

N

N

NH

NH

N

O

HN

O O

O

O

N Cu2+

NH Cu2+ N

[Cu·6.47]5+

H3N

N

N

NH

N

[Cu·6.48]5+

O

N

[Cu·6.49]5+

Chart 6.18 

Bicyclic systems containing two tris(2‐aminoethyl)amine units as head groups formed the basis for dinuclear anion receptors. These complexes benefitted from the high stability of the underlying metal–ligand interactions. For example, formation of one of the earliest dicopper(II) complexes investigated in this context, compound [Cu2·6.50]4+ (Chart 6.19) [96], was associated with stepwise equilibrium constants log Ka of 17.4 and 10.0 for the first and the second Cu2+ coordination steps, respectively (perchlorate as counterion).

O

HN

NH

HN

NH

N Cu2+ Cu2+ N HN NH O HN NH O

N Cu2+ Cu2+ N HN NH

[Cu2·6.50]4+

[Cu2·6.51]4+

HN

NH

N

N

N Zn2+ Zn2+ N HN NH HN

NH

[Zn2·6.52]4+

Chart 6.19 

[Cu2·6.50]4+ bound chloride with a log Ka of 3.6. The highest affinity was observed for the hydroxo complex of [Cu2·6.50]4+, however, which had a log Ka of

219

220

6  Ion Receptors

11.6, causing this complex to dominate the binding equilibria [96]. This exceptionally high stability was attributed to the hydrogen bonding of the OH− proton to the oxygen atoms in the side chains of the ligand. Indeed, replacing the linkers in [Cu2·6.50]4+ with linkers not featuring hydrogen bond acceptors as in [Cu2·6.51]4+ (Chart 6.19) afforded more versatile anion receptors [97]. [Cu2·6.51]4+ also formed the corresponding hydroxo complex at pH 8. In this case, the addition of other anions caused the displacement of the OH− anion. With a log Ka of 4.8, azide formed the most stable complex, which was attributed to the optimal fit of the linear azide between the metal ions. Other anions studied, even the twofold charged SO42− or the strongly coordinating SCN−, were bound considerably less strongly, leading to the conclusion that the host did not recognize the donor tendencies or the shape, but the bite length of the anionic guest. Systematic work in the Fabbrizzi group showed that anion selectivity of these cascade complexes could be modulated by adapting the length and rigidity of the linkers between the two tris(2‐aminoethyl)amine units to the guest [68, 98]. An example for a hybrid receptors containing additional coordination sites along a macrocyclic scaffold is the dizinc(II) complex [Zn2·6.52]4+ (Chart 6.19), deriving from [Zn2·6.39]4+ with two additional 2‐methylquinoline substituents [99]. The weak fluorescence of [Zn2·6.52]4+ in aqueous solution increased 21‐ fold upon the addition of 1 equiv. of pyrophosphate anions, which were bound with a log Ka of 6.2. This response was linear up to a 10 μM pyrophosphate concentration, with the detection limit amounting to 300 nM. Other anions, including phosphate, phenyl phosphate, adenosine monophosphate (AMP), ADP, and ATP, interacted weakly by at least 1 order of magnitude and produced no fluorescence response, rendering pyrophosphate detection by [Zn2·6.52]4+ highly selective. 6.3.1.5 Miscellaneous

Dalla Cort et al. showed that the uranyl cation UO22+ is sufficiently Lewis acidic to allow for anion binding in water. Based on earlier work in which the anion affinities of uranyl salophen receptors were investigated in organic media, they developed the water‐soluble analog [UO2·6.53] (Chart 6.20) that was shown to bind fluoride in 10 mM aqueous MOPS buffer (pH 7.5) with a log Ka of 2.1. Pyrophosphate, ADP, and ATP were bound even stronger with affinity constants log Ka > 4 [100]. The incorporation of analogous apolar receptors into the interior of micelles also afforded water‐compatible fluoride binders with fluoride affinities log Ka of ca. 5 [101].

HO

OH

O HO

O OH

Chart 6.20 

N N UO2 O O

HO O HO

OH O OH

[UO2 6.53]

6.3  Anion Receptors

6.3.2  Positively Charged Receptors Covalently assembled anion receptors also often feature positively charged groups that contribute to substrate recognition. Examples are ammonium, guanidinium, and imidazolium groups, all of which can be found in receptors working in water. Receptors without additional hydrogen bond donors, such as those containing only quaternary ammonium groups, interact with anions by Coulomb attraction in combination with solvation effects. Additional hydrogen bond donors typically render anion binding stronger and more selective. The modes of interaction of anions to these groups are depicted in a schematic fashion in Figure 6.6. Figure 6.6 shows that guanidinium groups can simultaneously form two hydrogen bonds to an anion, whereas ammonium groups can only form one. Moreover, with pKa values typically ranging between 11 and 13, guanidinium groups remain protonated over a wide pH range. Amines, in contrast, are less basic and their interaction with anions is therefore more strongly pH dependent. A third aspect is that hydration of ammonium groups differs profoundly from that of guanidinium groups, which has consequences for the thermodynamics of binding as reflected, for example, in the behaviors of receptors [Cu·6.47]5+ and [Cu·6.48]5+ (see Section 6.3.1.4). Electrostatic interactions and hydrogen bonding to a single positively charged group are normally insufficient for noticeable anion binding in water and most receptors therefore contain a combination of binding sites. Hydrogen bonding can also involve interactions with polarized C–H groups as those in imidazolium moieties, but the number of such systems that work in water is small. An aspect that generally affects anion binding of positively charged receptors is the unavoidable presence of counterions, which compete with the actual substrates as already mentioned in the previous section. 6.3.2.1  Receptors with Quaternary Ammonium Groups

Anion‐binding studies comprising receptors with quaternary ammonium ions as primary binding sites for anions provided information about the intrinsic strength of ion pairing in water. Schneider and coworkers showed that salt bridges contribute to the overall Gibbs free energy of binding with −5 ± 1 kJ mol−1 at an ionic strength of 0.02 M [102]. Extrapolation to an ionic strength of zero causes the ΔG° increment for an ion pair in water to increase to −8 ± 1.7 kJ mol−1 [103]. This value correlates with a Ka of 25 M−1, showing that a single ion‐pairing

H N

N

R

H N HN

(a)

(b)

(c)

H N R

H R

R N

N R

(d)

Figure 6.6  Binding modes of anions to quaternary ammonium ions (a), protonated amines (b), guanidinium (c), and imidazolium moieties (d) (R = alkyl, aryl).

221

222

6  Ion Receptors

interaction in water is insufficient to produce appreciable complex stability even under ideal conditions. That anion binding can nevertheless be achieved in water with receptors containing quaternary ammonium ions was shown by Schmidtchen. The receptors 6.54a and 6.55 (Chart 6.21) developed in this context were based on macrotricyclic systems with four ammonium groups serving as bridge heads [104]. Intramolecular charge repulsion prevented these receptors from collapsing, thus rendering their cavities well predisposed for the incorporation of an anion. Both compounds indeed proved to bind halides in water with stability constants log Ka ranging between 0.5 and 3.0. Binding selectivity correlated with the size of the cavity, with the smaller receptor 6.54a showing highest affinity for bromide in water (0.1 M aqueous sodium tosylate), whereas chloride and iodide were bound somewhat less efficiently. The larger receptor 6.55, on the other hand, bound iodide best. Carboxylates, carbonate, phosphate, and phosphate esters also interacted with 6.54a and 6.55, and it was observed that anion affinities correlated with the number of negative charges on the anion, confirming that the binding mode was primarily based on electrostatic interactions. R N

R N

N

N R

N R

6.54a R = CH3 6.54b R =

N

N

N

COO

6.55

Chart 6.21 

Anion affinity of these hosts was mediated to some extent by the type and concentration of the supporting electrolyte used for the binding studies, indicating that other ions present in solution competed with the actual substrates. This competition could be elegantly avoided by the introduction of negatively charged groups, yielding the zwitterionic host 6.54b [105]. As expected, anion affinity of 6.54b was slightly higher than that of 6.54a. Van’t Hoff analyses showed that bromide and iodide were bound in exothermic reactions with un­favorable entropic contributions (Br−: ΔH = −31.8 kJ mol−1; TΔS° = −10.0 kJ mol−1; I− = ΔH° = −69.9 kJ mol−1; TΔS° = −42.3 kJ mol−1; T = 25 °C). These trends were partly attributed to the differences in the hydration energies of both anions: ­bromide has a more negative hydration enthalpy than iodide, which explained why the enthalpic gain during bromide binding was smaller than during iodide

6.3  Anion Receptors

­ inding. It is also evident that these thermodynamic parameters were strikingly b similar as those obtained for hosts whose anion binding in water is accompanied with the release of “high‐energy water.” Similar factors could therefore also influence anion recognition of such polycationic systems. 6.3.2.2  Amine‐Based Receptors

Macrocyclic polyamines, examples of which are compounds 6.56a–c shown in Chart 6.22, are the anion‐binding counterparts of crown ethers. The replacement of ether groups along a macrocycle with amino groups has, however, more profound consequences than just reversing the charge of the preferred substrate.

NH HN

NH HN NH

NH

NH

HN

NH

HN

HN NH HN

6.56a

NH HN

6.56b

N H

N H

HN

NH

HN

NH

HN NH

H N

HN

6.56c

Chart 6.22 

First, amines are basic, rendering the receptors charged in water so that anion binding has a stronger electrostatic component than cation binding of crown ethers. Second, the degree of protonation of macrocyclic polyamines and in turn their binding properties is pH-dependent. Binding typically becomes stronger, the larger the degree of protonation, i.e. the lower the pH. One has to consider, however, that anions can also be involved in protonation equilibria, causing their overall charge to decrease at lower pH. Thus, anion binding proceeds in a certain pH window and binding studies therefore usually involve potentiometric titrations, which provide information about the interacting species coexisting in solution at a certain pH and their affinities. Third, (protonated) amines are hydrogen bond donors, causing complex stability to benefit from a further and more directed component. Binding studies involving a large number of structurally diverse p ­ olyammonium‐ based receptors yielded a set of general guidelines [106]. With regard to the protonation sequence of the amino groups along the receptor scaffold, the following rules apply: the ammonium groups in a partially protonated polyamine tend to be arranged at the largest possible distance to reduce charge repulsion. Therefore, nonadjacent amino groups are protonated first, and the least basic ones last if at all. It is important in this context that secondary amino groups have a larger propensity to accept a proton than tertiary amino groups. The introduction of tertiary amines into strategic position of a macrocyclic ­system

223

224

6  Ion Receptors

therefore offers a means of controlling the protonation pattern. Also, the ­distance of the amino groups along the receptor scaffold strongly affects the conditions required for protonation. The closer the amines are, the more difficult it becomes to achieve a higher degree of protonation. Protonation more­ over  affects the receptor conformation because ammonium groups tend to move away from each other, causing the molecular framework of the receptor to expand upon protonation. Finally, ammonium groups along macrocyclic or poly­ macrocyclic systems preferentially adopt orientations with the protons located outside of the cavity. As a consequence, the receptor has to undergo a confor­ mational reorganization to allow for hydrogen bonding to an included anion. How these factors affect anion affinity has been systematically investigated by using complex anions such as the octahedral complexes [Fe(CN)6]4− and [Co(CN)6]3− and the planar complex [Pt(CN)4]2− as guests. Comparison of the [Fe(CN)6]4− and [Co(CN)6]3− affinities of a receptor provides information, for example, about the influence of the charge of the anion on complex stability without strong interference by other factors because of the close relationship of these anions in terms of geometry and solvation. These studies showed that (i) the complexes of the higher charged [Fe(CN)6]4− anion are more stable than those of [Co(CN)6]3−, (ii) complex stability increases with increasing protonation degree of the receptor, and (iii) for a given protonation degree, complex stability increases with increasing charge density, causing smaller macrocycles to form more stable complexes than larger ones and macrocyclic receptors to form more stable complexes than acyclic ones. The decrease of affinity with increasing size of the macrocycles continues until the ring allows full incorporation of the anion. Once this change in binding mode becomes possible, a further increase of ring size causes the complex to become more stable. The microcalorimetric characterization of the interactions of macrocyclic polyamines with oxoanions revealed that these straightforward correlations partly break down when the substrates engage in hydrogen bonding interactions with the receptor [107]. Changing the receptor architecture from macrocyclic to macrobicyclic has further consequences on anion affinity and selectivity [108]. Whereas macrocyclic polyamines typically exhibit only weak affinity for halides, halide binding to macrobicyclic polyamines can be very strong. The azacryptand 6.57 (Chart 6.23), for example, practically only interacted with fluoride with an impressive affinity log Ka of 10.6 determined for H66.576+ (0.1 M aqueous NaOTs) [109]. In contrast, complexation of chloride occurred only after the addition of the sixth proton, NH HN

NH

HN

NH HN

NH

NH HN

NH

N

N

6.57

Chart 6.23 

HN

HN

NH

HN

NH

HN

NH

HN

NH

N

N

6.58

NH

NH N

N

6.59

O

HN

N

N O O 6.50

HN HN

6.3  Anion Receptors

which explained the impressive F−/Cl− selectivity of 108. The decrease of affinity in the order F− > Cl− > Br− > I− was still observed for 6.58, but it reversed for 6.59, most probably because iodide fitted best into the cavity of the largest host [109, 110]. The azacryptand 6.50 whose dicopper(II) complex [Cu2·6.50]4+ strongly interacted with OH− (see Section 6.3.1.4), bound halides, azide, nitrate, sulfate, phosphate, and pyrophosphate in its metal‐free form after protonation [111]. Among the singly charged anions, affinity of the hexaprotonated receptor was highest for azide (log Ka = 4.3), most probably because of the perfect fit of the linear anion into the ellipsoidal cavity of the host. Even fluoride was bound with a slightly smaller log Ka of 4.1. The binding constants between H66.506+ and oxoanions with a higher charge were, however, significantly larger (SO42−: log Ka = 4.9; HPO42−: log Ka = 5.5; P2O74−: log Ka = 10.3). Bicyclic receptors with aromatic residues between head groups were also reported and shown to interact in their protonated forms with halides, oxoanions, and (di)carboxylates [112]. Although the cage 6.60a (Chart 6.24) developed by Delgado differed structurally from the azacryptands in Chart 6.23, it exhibited related properties. This receptor bound a variety of anions in water, including halides and acetate, but possessed a particular selectivity for tetrahedral oxoanions [113]. Its sulfate affinity increased when going from the pentaprotonated to the hexaprotonated form, whereas the H2PO4− complex became less stable upon the addition of the sixth anion. This behavior was interpreted in terms of repulsive interactions between the fully protonated form of 6.60a and the protons on the dihydrogenphosphate anion. Indeed, replacement of the benzene units in 6.60a with pyridyl units whose ring nitrogen atoms could interact as hydrogen bond acceptors with partially protonated anions caused the H2PO4− affinity of the respective receptor 6.60b to slightly increase when adding the sixth proton [114].

HN

NH

HN

NH

HN

HN

N HN HN

NH

HN HN

NH

6.60a

6.60b

HN

NH

N

HN HN

NH N

HN HN HN

NH

6.60c

Chart 6.24 

6.60b therefore stabilized the dihydrogenphosphate anion even with all aliphatic amino groups protonated. Conversely, the pyrrole‐containing receptor 6.60c oriented further hydrogen bond donors inside the receptor cavity and therefore possessed a higher sulfate affinity than 6.60a [115]. 6.3.2.3  Guanidine‐Based Receptors

Guanidinium cations are only weakly hydrated in water. Thus, dehydration is neither very costly in terms of enthalpy nor does it have a large entropic a­ dvantage.

225

226

6  Ion Receptors

Both aspects have characteristic consequences on the thermodynamics of anion binding in comparison to receptors containing ammonium groups as seen already for receptors [Cu·6.47]5+ and [Cu·6.48]5+ (see Section 6.3.1.4). The first anion receptors containing guanidinium groups were developed in the Lehn group. Macrocycles 6.61a and 6.61b (Chart 6.25) are examples, which can be regarded as the guanidinium analogs of polyammonium‐derived receptors [116]. They interacted only weakly with anions, phosphate recognition of 6.61a and 6.61b was associated with log Ka values of 1.7 and 2.2, for example. NH2 N H

N H

H N

H N

H2N

NH

HN

NH

NH2

HN H N

NH2

H N NH2

6.61a

6.61b

Chart 6.25 

Subsequent work in the area mainly involved conformationally rigidified guanidinium derivatives [117]. The bis(guanidinium) derivative 6.62 (Chart 6.26) developed by Schmidtchen and coworker was shown to bind HPO42− with a log Ka of 3.0 in water [118]. This receptor also bound phosphate esters and AMP, albeit with a lower affinity. The tripodal receptor 6.63 with three protonated 2‐ aminoimidazolidine substituents arranged around an aromatic core recognized citrate in water with a log Ka of 3.8 in D2O at pD 7.4 (log Ka = 2.1 in 0.1 M phosphate buffer at pH 7.4) [119]. Dicarboxylates, monocarboxylates, or phosphates were bound with considerably lower affinity. The analog of 6.63 containing ammonium groups instead of the protonated 2‐aminoimidazolidine units bound to citrate under the same conditions with a log Ka of 3.3, demonstrating that in this case the guanidinium groups produced a slightly larger affinity. This result was attributed to the contributions of the extensive hydrogen bonding between the carboxylate groups of the guest and the guanidinium groups of 6.63 to complex stability.

N HO

N H

N N H

O

O

N H

N H

OH

H N N H

H N H N NH

H N

H H N N N H

6.62

Chart 6.26 

6.63

6.3  Anion Receptors

The Schmuck group introduced the guanidiniocarbonylpyrrole (GCP) group as a versatile and potent motif for carboxylate recognition in water. The prototype receptor 6.64 (Chart 6.27) was shown to bind to bind N‐acetylated amino acids in 40 vol% water/DMSO [120].

R O N

H

N H

O H

N H

N

H N H

O H

O

H2N

R N H

O H

NH2 H2N

N H

N

O

H N

H

H

NH NH

NH2 O N H HN

H2N

HN O NH

H N

O

HN

NH2 O

HN O

6.64

6.65 R = CH2(OCH2CH2)3OH

6.66

Chart 6.27 

The self‐complementarity of the deprotected analog 6.65 caused it to dimerize in water with a dimerization constant of 170 M−1 [121]. The tripodal receptor 6.66 interacted with citrate in water with a log Ka of 5.2 [122], showing that in terms of performance, this receptor ranged between 6.63 (Chart 6.26) (log Ka  =  3.8 in D2O, pH 7.4) developed by Anslyn, and the metal‐containing receptor 6.34 (Chart 6.13) (log Ka = 5.6 in 0.01 M HEPES buffer, pH 7.0) developed by Fabbrizzi. For other applications of the GCP moiety, see Chapter 3. 6.3.2.4  Imidazolium‐Based Receptors

Macrocyclic receptors containing triazolium or imidazolium units found widespread applications in anion coordination chemistry [123], but only a few systems with imidazolium groups worked in water. One is the “Texas‐sized” molecular box 6.67 (Chart 6.28). This compound interacted with ammonium decanoate in water with a binding constant log Ka of 5.4 for the respective 1 : 1 complex [124]. The complex dissociated upon addition of ATP, which bound stronger to 6.67 than the carboxylate. The carboxylate affinity of 6.67 was used to control the morphology of a copolymer containing carboxylate groups in its side chains and to revert the effect by adding ATP. Moreover, hydrogels cross‐ linked with a derivative of 6.67 with two additional carboxylate groups were shown to allow the extraction of inorganic anions such as NO3− or SO42− or anionic dyes from water [125]. The extracted cargo was released after separation of the hydrogel by adding HCl. The tetrakisimidazolium receptor 6.68 (Chart 6.28) signaled the presence of sulfate in water by a strong increase of fluorescence [126]. Sulfate complexation involved the formation of a 2 : 1 complex in which the anion was sandwiched between two receptor rings. The direct anion–receptor interactions comprised eight hydrogen bonds between the oxygen atoms of the anion and the converging

227

228

6  Ion Receptors O

N

N

N

N

O

O

N

N

O

N

N

N

N

N

N O

N

N

N

N

N

N

N

N

O N

N

O

N

N

O

O

6.67

O

6.68

Chart 6.28 

imidazolium protons. In addition, intermolecular aromatic interactions between the benzene and imidazolium moieties of the two receptor units in combination with charge‐assisted hydrogen bonds between the peripheral chains and outer ring protons were expected to contribute to the stability of the complex. Its stability amounted to 8.6 × 109 M−2 in 10 mM HEPES buffer at pH 7.0. 6.3.3  Negatively Charged Receptors That a negatively charged receptor should bind an anionic guest may seem counterintuitive at first sight. However, there are several receptors containing carboxylate groups that possess substantial anion affinity. Anion binding in these systems takes place in deep and well‐shielded cavities into which the carboxylate groups are unable to protrude. These groups therefore do not compete with the guests but mainly mediate receptor solubility. One example is the cavitand 6.69 (Chart 6.29) developed by Gibb and co­worker [127]. The overall eight carboxylate groups of this so‐called octa‐acid deep cavitand mediate water solubility at basic pH. The cavity interior, which is mostly lined by electron‐rich aromatic rings, does not seem to be well suited for anion binding, but the electrostatic potential surface of 6.69, shown in

O O

O

O

O O

O

O

O

O

O

O

O

O

O O

O

O

Chart 6.29 

O

O

O

OO

O

O

O

O O

O

O

O

O 6.69

6.3  Anion Receptors

(a)

(b)

(c)

(d)

Figure 6.7  Electrostatic potential surfaces of derivatives of receptors 6.69 (carboxylate groups replaced by protons), 6.70a (R = Me), 6.71 (R = Me), and 6.74a. The surfaces were generated by using Spartan 10 (Wavefunction, Inc.) and mapping AM1 electrostatic potentials onto surfaces of a molecular electron density of 0.002 electron Å−1 followed by color coding. In all surfaces, the potential energy values range between −120 and +120 kJ mol−1, with red and blue signifying values greater or equal to the absolute maximum in negative and positive potential, respectively.

Figure 6.7a, indicates that the methine protons of the benzal units, which converge into the cavity, cause parts of the interior to feature a positive potential. Cavitand 6.69 was mainly used as a host for apolar aliphatic compounds and as a reaction vessel to control transformations of such guests [127]. In most of these cases, guest complexation in conjunction with hydrophobic effects caused two molecules of 6.69 to self‐assemble. An exception was the complex between 6.69 and adamantane carboxylate (AC), which featured the carboxylate group at the cavity entrance. This polar group mediated receptor hydration in this region of  the complex, weakened hydrophobic interactions, and thus prevented self‐ assembly. Although one could argue that 6.69 acted as an anion receptor in this case because the carboxylate group of AC was deprotonated at the pH of the binding studies (10 mM phosphate buffer, pH 11.3), complex formation did not directly involve the anionic group of the guest. The ability of 6.69 to interact with other anions became evident while studying the influence of various sodium salts on AC binding. In these studies, the anions in the salts were varied from strongly kosmotropic ones such as fluoride and sulfate to anions at the opposite end of the Hofmeister series, namely, iodide, thiocyanate, and perchlorate [128]. Kosmotropic anions caused a slight increase in complex stability, consistent with their tendency to strengthen the hydrophobic effect underlying binding of AC to 6.69 (see Chapter  1). Chaotropic anions, on the other hand, weakened complex stability by mainly reducing the exothermicity of binding. AC complexation became endothermic and entirely due to the pronounced favorable entropy, for example, in the presence of perchlorate. The weakening of AC binding in the presence of chaotropic anions was attributed to their ability to compete with AC by occupying the cavity of 6.69. Although the anions were typically not bound very strongly, the perchlorate complex had a log Ka of 2.0, for example, they could still produce pronounced effects if their concentration was high enough. Perchlorate binding to 6.69 turned out to be mediated by salts in a similar manner as AC binding: kosmotropic anions strengthened, while chaotropic anions weakened it [129]. The latter effect was again attributed to the competitive binding of the chaotropic anions, whereas the reinforcement of the perchlorate

229

230

6  Ion Receptors

complex by kosmotropic anions was related to the increase of the ionic strength of the solution upon salt addition. Higher salt concentrations caused an increase of the degree to which sodium ions ion‐pair with the carboxylate groups surrounding the cavity of 6.69, resulting in the attenuation of the net charge of the receptor and, as a consequence, in anion binding to become stronger. In addition to ClO4−, a series of other anions, namely, Cl3CCO2−, PF6−, MeSO2S−, ReO4−, TfO−, IO4−, BH3CN−, Cl2CHCO2−, SCN−, N(CN)2−, and I−, were shown to interact with 6.69 [130]. Under the experimental conditions (50 mM sodium phosphate buffer, pH 11.5), complex formation was consistently exothermic and further characterized by adverse entropy contributions. The negative complexation entropies were attributed to the binding of highly organized water molecules along with the anions. Perchlorate, for example, kept on average 3.1 water of the hydration shell when bound to 6.69. Because of the combination of anion and water binding, the observed negative binding enthalpies could not be easily attributed to receptor desolvation alone. Molecular dynamics simulation in fact indicated cavity hydration of 6.69 to be favored by enthalpy but penalized by entropy [131]. Desolvating the cavity should thus be endothermic, suggesting that the observed exothermicity of anion binding was due to an interplay of desolvation phenomena and direct receptor–anion interactions. The authors concluded that binding of partially hydrated anions could be a promising strategy to achieve anion binding in water, circumventing the often energetically costly anion dehydration. Another polyanionic anion receptor is the hexameric glycoluril derivative 6.70a (Chart 6.30), introduced by Sindelar and coworkers [132]. The methylene bridges of 6.70a are arranged along the equator with the glycol units alternatively pointing into opposite directions. Their convex surfaces are oriented toward the cavity interior where they produce a substantial positive electrostatic potential (Figure 6.7b). O RN

R N

R N

O

S O

N N O N NO N N R R N N O R NN RNO O N R N R N O N N N N N R N R O O N O R 6.70a R =

6.70b R =

O

R NO N O N N N

R S

N S

R N

R OR O N S N N N

O R

O

3

6.71 R =

S

N

S

COO

COO

Chart 6.30 

Such bambus[6]urils were shown to possess very high affinity for anions in organic media. These binding properties were essentially retained in water [133]. In this medium, the water‐soluble derivatives 6.70a containing 12 carboxylate

6.3  Anion Receptors

groups in the periphery bound the strongly hydrated fluoride anion with a stability constant log Ka of 2.0. Affinity increased in the order F− 1 mm

+ +++

Metabolism, labeled molecules

γ‐rays Single‐photon emission computed tomography (SPECT)

No limit

>1 mm

+ +++

Metabolism, labeled molecules

Optical imaging

Light

1–2 cm

3 mm

+ +++

Molecular events, labeled molecules

Ultrasound imaging

Sound waves

mm to cm

>50 μm

++ ++

Anatomy, organ function

Source: Adapted from Lu and Yuan 2015 [1] and Silva et al. 2005 [2].

13.2  Structure and Supramolecular Properties of Molecular Probes

13.2 ­Structure and Supramolecular Properties of Molecular Probes 13.2.1 Structure Targeted molecular probes often have three core structural components: a targeting unit, a linker, and a reporter group (Figure  13.2). The targeting unit is chosen to recognize a specific receptor, enzyme, or other biomolecule, which is characteristic to the disease of interest. The reporter group provides a signal that is detected using an imaging modality, and it is tethered to the targeting unit by a linker that may vary in length, flexibility, and reactivity. The size and architecture of the conjugate can vary widely, ranging from a small molecule with low molecular weight to a nanoscale particle that is coated with many copies of the targeting unit. 13.2.2 Linkers The linker, or tether between the reporter group and the targeting unit, plays a significant role in the overall performance of the probe. The length of the linker is one of the many important properties that require optimization during probe design. For example, if the linker is short, it brings the reporter group within close proximity of the targeting unit. This may cause steric hindrance and therefore reduce binding affinity for the target. Another important property of the linker is its rigidity and flexibility, which may affect folding and self‐aggregation of the probe. In some cases, the linker can be modified to improve aqueous solubility. In other cases, the overall charge and hydrophobicity of the linker may impact affinity with off‐target sites and may affect the probe’s ability to cross the cellular membrane.

Reporter

Linker

CT, MRI, PET, SPECT, optical

Stability Cleavable Rigid/flexible

Targeting unit Vitamins, peptides, antibodies, aptamers

Target Receptors, enzymes, biomolecules

Figure 13.2  The general structure of molecular probes for medical imaging has a reporter group and a targeting unit, connected by a linker.

503

504

13  Probes for Medical Imaging

13.2.3  Reporter Groups Some imaging modalities such MRI and CT use nontargeted contrast agents to achieve an enhanced signal of specific anatomical features (e.g. blood pooling agents). Reporter groups that are commonly used in medical imaging are listed in Table 13.2. However, these nontargeted agents are not useful when cellular or functional information is needed. Linking a contrast agent to a targeting unit produces a targeted probe with high specificity. When selecting the reporter group, it is important that it aligns with the biological parameters of the target site (e.g. target size and depth, circulation half‐life, target concentration). This is discussed in greater detail in the following section. 13.2.4  Design Aspects 1. Stability: The ideal molecular probe should not undergo chemical degradation in the presence of light, temperature variations, or prolonged storage. Once inside the body, there are a range of biochemical processes that can degrade a probe’s chemical structure. Probe degradation after dosing results in low target signal, higher dose administration, and potential toxicity to the patient. On the other hand, probe clearance from the body either by excretion or metabolic decomposition is a desirable pharmacokinetic property. Thus, the design goal is to ensure the probe is stable enough to survive the duration of the imaging experiment. Certain functional groups are known to be susceptible to enzymatic oxidation, hydrolysis, or isomerization. For example, cytochrome P450 enzymes are responsible for oxidative transformation of 70–80% of drugs and are also likely to degrade most molecular probes [3]. Similarly, probes that have phosphate ester or amide bonds are susceptible to degradation by nucleases or proteases, respectively. 2. Toxicity: Molecular probes are typically employed as tracers at a low concentration that greatly minimizes the chances of undesired toxicity. In addition, many imaging protocols do not involve repeated dosing of the probe, eliminating the concern of chronic toxicity or abhorrent immune response. Thus, the typical toxicity question for an imaging probe study is whether a single low dose of probe will cause acute toxicity to the subject. A related factor to consider is the vehicle used to deliver the probe. The use of any organic solvent must be minimized, as many solvents such as methanol and dimethyl sulfoxide Table 13.2  Commonly used reporter groups in medical imaging. Modality

Reporter

MRI

Gd3+, Mn2+, superparamagnetic iron oxide (SPIO)

PET

18

SPECT

111

Optical

Near‐infrared (NIR) fluorophore bioluminescence

CT

Gold nanoparticles (AuNPs), iodine

F, 64Cu, 67Ga In, 99mTc, 123I, 86Y

13.2  Structure and Supramolecular Properties of Molecular Probes

exhibit a degree of cellular toxicity. Additionally, diagnostic probes that contain radioactive metal cations should receive special attention to be sure that the metals are not stripped from the probe structure and accumulate in anatomical regions such as bone where they become toxic [4]. 3. Size: The size of a molecular probe can greatly influence important pharmacokinetic properties such as diffusion to the site of interest and rate of cell internalization. In some cases, there may be little scope to control the size of a molecular imaging probe, as altering the reporter group or targeting unit may reduce signal or affinity, respectively. However, minor alterations to the linker lengths within cancer probes have proven to substantially influence the tumor penetration and route of excretion. While the ideal probe size is highly dependent on the target tissue, it is understood that molecules