Read an Excerpt
Biophysical Approaches Determining Ligand Binding to Biomolecular Targets
Detection, Measurement and Modelling
By Alberto Podjarny, Annick Dejaegere, Bruno Kieffer The Royal Society of Chemistry
Copyright © 2011 Royal Society of Chemistry
All rights reserved.
ISBN: 978-1-84973-009-9
CHAPTER 1
Introduction: The Who, Where, Why Questions
ANNICK DEJAEGERE, BRUNO KIEFFER AND ALBERTO PODJARNY
Institut de Génétique et de Biologie Moléculaire et Cellulaire, 1, rue Laurent Fries, BP 10142, 67404, Illkirch Cedex, France
1.1 Introduction
The study of ligand binding to macromolecules of biological interest is essential for the understanding of biological processes and for the efforts to modify them, e.g. by therapeutic molecules. Several biophysical experimental techniques have been used for these studies, such as crystallography, fluorescence, calorimetry, nuclear magnetic resonance, mass spectrometry, circular dichroism, thermal shift assay and plasmon resonance. These experimental studies have been complemented by modelling of the interactions of ligands with their biological targets. During the era of genomics, proteomics and other "omics" initiatives, much emphasis was put on the search and characterisation of new targets for biologically active ligands, with comparatively little effort devoted in the recent years to gaining deeper insights into the driving forces for ligand binding. The important fundamental and practical implications of a detailed understanding of ligand–target interactions have, however, been reconsidered, in the light of new developments in the pharmaceutical industry, as well as with the advent of systemic approaches to biological function. This in turn has led a revival of biophysical techniques for studying macromolecule–ligand interactions, both in academics and in the pharmaceutical industry.
Increasing amounts of data on protein–ligand interactions, as well as sustained technological developments of the biophysical methods that are used for their study, have also changed our vision of the molecular binding event. The failure of the simple key-and-lock model to explain numerous molecular interactions called for more complex models, such as induced fit, where molecular flexibility plays a fundamental role. Although the concepts beyond induced fit are quite old, only recently did experimental methods allow for a detailed mechanistic probing that supported them. Recognition of the role of disordered regions in molecular recognition is, in this respect, revealing. A growing set of experimental evidence shows that allosteric mechanisms may exploit dynamical properties of proteins. Many of these observations were made possible thanks to the development of NMR spectroscopy, which enabled a detailed description of molecular motions and the observation of transient excited states of proteins. As the sensitivity of biophysical methods is still improving, new applications are made possible, such as the observation of binding events within cellular environments. For example, a detailed description of a binding event involving a membrane protein becomes a conceivable goal using current methods.
A strong asset of biophysical methods is their versatility: they provide detailed mechanistic information on ligand binding (kinetics, equilibrium constants, stoichiometry, structural information, binding mechanisms), and they are equally adapted to screening compound libraries for identifying binders.
Basically, biophysical methods can provide answers to the following questions: Which ligand is likely to bind to a particular target? Where would suitable ligands bind? And what is the driving force for binding? We summarise the above as the Who? Where? Why? (WWW) questions of ligand binding. The ability of biophysical methods to cover these complementary aspects of ligand binding stresses the importance of their integrated use. In particular, during the early phases of drug discovery (see Figure 1.1), biophysical characterisation has the potential to accelerate the pace of pharmaceutical research by swiftly providing better chemical starting points together with a high-quality information package that can be used to guide and expedite lead optimisation (ref. 3 and references therein).
The objective of this book is to provide the reader with a description of the basic principles of experimental and computational biophysical methods used to study binding events. Particular emphasis will be placed on gaining a deeper insight into the respective merits and limitations of the technologies involved. As a guideline to better understand the strengths and limitations of the various techniques, we will consider their respective abilities to answer the above-mentioned WWW (Who? Where? Why?) questions.
1.1.1 Who?
The "who" question has been at the heart of pharmacology since analytical chemistry methods allowed to identify chemical compounds that were responsible for a given biological effect. The fortuitous discovery of penicillin by Alexander Fleming prompted chemical industries to develop large-scale screening procedures to identify the chemical nature of potent natural products. Mostly based on activity assays, these efforts led to the discovery of a significant part of the current pharmacopoeia (which is composed of about 3000 compounds). As a side-product of this top-down screening process, large collections of chemical compounds have been designed and organised as chemical libraries. At the end of the twentieth century, the advent of molecular biology technologies together with the development of biophysical methods triggered a deep modification of the perspectives in the drug discovery process. The ability to identify, express and purify large amounts of macromolecular targets and obtain sensitive and automated measurements of ligand-binding events led to the development of high-throughput screening programs (HTS), where libraries of chemical compounds are screened to identify binders. This evolution modified the way to answer the "who" question, which focused on the detection of the binding event, aiming at discriminating high-affinity binders within large ensembles of chemicals. Several biophysical techniques were then largely improved to meet the requirements of industrial processes. These efforts resulted in considerable gains in term of sensitivity, speed, miniaturisation and reliability for most of these techniques. Recently, the need to enlarge the chemical space explored by screening methods shifted the emphasis towards the detection and characterisation of weak binding (10 µM up to 10 mM) events between small fragments and macromolecular targets. This step added more constraints on the choice of biophysical methods chosen to answer the "who" question. Robust, high-throughput detection of weak binders poses a technological challenge, and calls for elaborated controls or cross-validation of results using a combination of biophysical methods. A number of biophysical methods are available to detect and measure the binding of a ligand to a target. These methods are based on the observation of different physical effects, the nature of which defines the level of accuracy, the range of sensitivity and the degree of possible automation (see Table 1.1).
Fluorescence spectroscopy gathers many assets to answer the "who" question and has therefore been widely used in drug-discovery process. Highly sensitive fluorescence measurements could be performed in a fully automated way, allowing an easy implementation in industrial processes. Several fluorescence parameters could be monitored to detect a binding event providing intrinsic cross-validation capabilities. NMR and X-ray crystallography are naturally rather associated with the "where" question due to their ability to provide atomic resolution structures of protein-ligand complexes. However, these two structural methods have also evolved as efficient tools to perform medium-throughput ligand screening and several examples assessing their ability to answer the "who" question have been reported. X-ray crystallography relies on the crystallisation protocols to trap protein–ligand complexes, a process that can be automated to a large extent. Given the large amount of information contained in a single set of measurements, X-ray crystallography provides one of the most complete and integrated ways of answering the "who" question, though this method is prone to false negative hits, in particular when low-affinity binders are screened. In the situation of weak binders screening, NMR provides a powerful complementary tool. Performed in solution, using label-free ligands, NMR has the unique ability to detect binding events with the least possible perturbation of the system. Despite continuous improvement in the sensitivity and automation, the use of X-ray crystallography and NMR structural methods for ligand screening is affected by the need of high amounts of material. Several screening procedures included therefore some a-priori filtering using virtual docking strategies, attempting to reduce the number of molecules to analyse. Continuous training of docking algorithms on well-defined systems is constantly improving the accuracy of these procedures and their integration within experimental screening protocols is a major issue of the next years.
Surface plasmon resonance (SPR) and mass spectrometry provide alternative ways to search for binding events. SPR, which is a label-free method, requires immobilising the target on a surface. Several measurements can be performed using the same chip allowing the method to be adapted to medium-throughput screening processes. SPR is adapted to detect weak binding events and provides information about the kinetic parameters of the interaction. Mass spectrometry is also a label-free method that probes the molecular interactions in the gas phase. Its very high sensitivity and straightforward data analysis allow the use of this technique in highly automated environments. Circular dichroism is usually not used as a primary screening method but rather in the validation step, where its ability to monitor the binding event under a large range of conditions constitutes a valuable tool in the hit-to-lead step. The large consumption of material requested for calorimetry studies restricts its use for screening purposes, although the availability of the energetic parameters of a binding event is recognised as very valuable information in the hit-to-lead (Figure 1.1) process.
1.1.2 Where?
In order to answer the question of where a ligand binds, we need structural information. The two experimental techniques capable of providing atomic resolution structural information are X-ray crystallography and NMR. Since these techniques are also used to check ligand binding, experimental structural information concerning the "where" question is often obtained as a corollary of screening studies.
1.1.2.1 Crystallography
The determination of the structure of ligands bound to a protein goes back to the early days of protein crystallography, with the heme containing proteins Haemoglobin and Myoglobin. The first deposited structures of a protein bound to synthetic ligands dates to 1976, with the study of binding of chloromethyl ketone substrate analogues to crystalline papain.
The crystallographic determination of the structure of protein-ligand complexes had an immediate impact on the pharmaceutical industry. The first marketed drug for which crystallography played a significant role was Captopril (Bristol-Myers Squibb), an inhibitor of the angiotensin converting enzyme ACE, marketed in 1981. The development of this drug was based on the structure of the complex of the closely related zinc protease bovine carboxy-peptidase A in complex with benzylsuccinate. The first successful drug to reach clinical use based on a structure-based drug-design cycle (see Figure 1.1) was Trusopt (generic dorzolamide), an inhibitor of carbonic anhydrase II, marketed in 1995. The discovery of Trusopt followed a series of X-ray crystallographic studies of complexes of carbonic anhydrase II with acetazolamide and other sulfonamides.
Ligands usually bind in "beautiful sites", which are designed for natural ligands in receptors and enzymes. However, they can also bind in "exosites" or "allosteric" sites, which can be far from the natural binding pockets. Ligands can also induce conformational changes in the protein, creating their own binding sites.
To determine the binding site by protein crystallography, a crystal of the ligand–protein complex is obtained, diffraction data is measured and then ligands are observed in electron-density maps. These "difference maps" are usually calculated once the protein structure has been determined and refined. The crystallographer then interprets the electron-density in terms of the chemical structure of the ligand, and the full structure of the ligand–protein complex is then refined.
Technological advances mainly fostered by structural genomics efforts and investments have turned protein crystallography into a fast, highly automated technique which is particularly suited for the study of series of complexes between a particular protein target and low molecular weight ligands. The use of X-ray crystallography has increased exponentially as structure determination of complexes has been streamlined with the use of improved crystallisation techniques, synchrotron sources and new softwares. It has been applied to many pharmaceutical targets, in particular to HIV inhibitors, kinases and phosphatases.
In order to crystallise the protein–ligand complex, two alternative techniques are possible. The ligand is either crystallised at the same time as the protein, in a "co-crystallisation" experiment, or an existing crystal of the protein is soaked with a solution containing the ligand, in a "soaking" experiment. Each technique has its advantages and disadvantages, and the one chosen depends on each particular case.
The "where" question is the least demanding in terms of data quality and resolution. Even fairly low resolution (around 3.5 Å) may be sufficient to position the ligand. However, at this resolution many questions remain, like the exact ligand orientation, local structural changes induced by the ligand or the involvement of solvent molecules. Therefore, higher resolution data, around 2.0 Å, is needed for an accurate positioning of the ligand and the surrounding protein residues and water molecules. This is especially necessary if the ligand has to be modified to increase its affinity and selectivity.
The interpretation of electron-density maps can be straightforward for the case of strongly binding inhibitors at a resolution around 2.0 Å. For this case, software for automatic interpretation has been developed. However, at lower resolutions or for weakly binding inhibitors, careful inspection is necessary to avoid incorrect interpretations.
X-ray analysis of protein–ligand complexes is prone to false negatives, in particular when dealing with weak binders. The high protein and ligand concentrations used in co-crystallisation experiments are, in principle, highly favourable for the study of weak complexes. However, protein–protein contacts in the crystal can block access to a ligand binding site (in a soaking experiment) or compete out a ligand (in a co-crystallisation experiment). Furthermore, complex formation can be adversely affected by the crystallisation conditions (pH, high salt or high concentration of an organic solvent), and the solubility of some compounds can be exceedingly low under the crystallisation conditions.
In order to overcome these problems, fragment screening is being developed. This concept includes the identification of small molecules, or "fragments", that bind to the molecular target with weak overall affinity but high ligand efficiency, and the subsequent optimisation of these fragments on the basis of structural information, using fragment growing, merging or linking strategies.
1.1.2.2 NMR
The ability to obtain protein structure from NMR data has been demonstrated by the pioneering work of Kurt Wühtrich in the early 1980s. Although the same approach was used to determine the solution structures of complexes between small ligands and proteins, the real advent of protein–ligand complexes came with the possibility to enrich recombinant proteins with stable, magnetically active isotopes such as nitrogen 15 and carbon 13. The resulting simplification of protein signal coupled to the ability to selectively observe the labelled molecule triggered the development of powerful NMR methods for protein–ligand complexes structure determination. Multi-dimensional hetero-nuclear correlation spectra are used to monitor changes of the resonance frequency upon ligand addition to a solution of labelled protein. Provided that the resonance frequencies have been assigned to the corresponding amino acid and that the three-dimensional structure of the protein is known, the detection of a subset of shifting frequencies allows the identification of the binding pocket, a method widely known as chemical shift mapping. This method gained considerable attention in 1996 when Stephen Fesik demonstrated the possibility to design highly potent ligands using chemical shift driven optimisation of fragments binding to adjacent pockets, an approach widely known as "SAR by NMR". These experiments were based on 1H-15N HSQC correlation spectra, which provide roughly one correlation peak per amino acid, and therefore a true spectral identity card of the protein. The conceptual and practical simplicity of this method explains its very fast adoption by the community of people involved in ligand design, establishing NMR as an efficient tool to get structural information on low-affinity complexes. Indeed, for these complexes, the fast exchange between the bound and free states of the protein averages the resonance frequencies, which can therefore be used as a direct measurement of the complex formation. Interaction between high molecular weight complexes and small ligands may also be studied using chemical shift approaches with appropriate isotopic labelling strategies that aim at reducing the spectral complexity and strengthening the NMR signal.
(Continues...)
Excerpted from Biophysical Approaches Determining Ligand Binding to Biomolecular Targets by Alberto Podjarny, Annick Dejaegere, Bruno Kieffer. Copyright © 2011 Royal Society of Chemistry. Excerpted by permission of The Royal Society of Chemistry.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
