Applied Biophysics for Drug Discovery

Applied Biophysics for Drug Discovery is a guide to new techniques and approaches to identifying and characterizing small molecules in early drug discovery. Biophysical methods are reasserting their utility in drug discovery and through a combination of the rise of fragment-based drug discovery and an increased focus on more nuanced characterisation of small molecule binding, these methods are playing an increasing role in discovery campaigns. 

This text emphasizes practical considerations for selecting and deploying core biophysical method, including but not limited to ITC, SPR, and both ligand-detected and protein-detected NMR.

Topics covered include:

•          Design considerations in biophysical-based lead screening

•          Thermodynamic characterization of protein-compound interactions

•          Characterizing targets and screening reagents with HDX-MS

•          Microscale thermophoresis methods (MST)

•          Screening with Weak Affinity Chromatography

•          Methods to assess compound residence time

•          1D-NMR methods for hit identification

•          Protein-based NMR methods for SAR development

•          Industry case studies integrating multiple biophysical methods

This text is ideal for academic investigators and industry scientists planning hit characterization campaigns or designing and optimizing screening strategies.

• Language: English
• Publisher: Wiley
• Release date: Jul 14, 2017
• ISBN: 9781119099505

Book preview

1

Introduction

Donald Huddler*

Computational and Structural Chemistry, GlaxoSmithKline plc, Collegeville, PA, USA,

Over the last two decades, biophysics has reemerged as a core discipline in drug discovery. Many may argue that biophysical methods never truly left discovery, but all will note the renewed present importance and central role of such methods. This reemergence is driven by three primary forces: the birth of fragment‐based drug discovery schemes, the recognition of and desire to mitigate artifacts in traditional biochemical screening, and a desire to accelerate the transition from first‐in‐class to best‐in‐class molecules by focusing on hit and lead kinetics. Each of these strategies or goals requires various information‐rich biophysical methods to experimentally execute. This text aims to summarize some of the key methods emerging from these three broad enterprises. First, though, it will map the contours of these three drivers of biophysics’ reemergence and link them to the chapters that follow.

Fragment‐based drug discovery and fragment‐based lead discovery are slightly different names for the same discovery approach: using a library of relatively small compounds to probe the surface of a target protein for binding sites. Fragment‐based discovery approaches are animated by the information theory‐based idea that relatively simple, small compounds sample chemical space more effectively than larger, more complex molecules [1, 2]. In practice, this approach drives one to develop low complexity screening libraries [3, 4]; consequently, the binding interactions with target proteins are generally very weak. Weak interactions require sensitive methods to unambiguously detect the binding event [5]. In simple bimolecular binding, the concentration of the complex is driven by the concentration of the ligand; this drives many scientists to screen their fragment libraries at relatively high concentrations. Effective screening methods must both be able to detect relatively weak interactions in the context of relatively high compound concentrations; several biophysical methods are well suited for this demanding screening campaign [6]. Various NMR approaches have been successfully applied to identify and characterize weak small molecule–protein interactions [7]. This text explores both traditional protein‐detected NMR [8] approaches in Chapters 9 and 10 and nontraditional NMR [9, 10] approaches in Chapter 8. Both approaches have merit and are usefully applicable in partially overlapping circumstances. Surface plasmon resonance (SPR) [11, 12] and microscale thermophoresis (MST) [13] have also been successfully deployed in fragment screening campaigns to detect weak interactions. Chapters 5 and 6 explore applications of MST and SPR beyond fragment‐based discovery, respectively.

A second force driving the reemergence of biophysical methods in drug discovery has been the desire to identify and eliminate high‐throughput screening hits that operate through uninteresting nuisance mechanisms. Brian Schoichet recognized and characterized some commonly observed nuisance phenomena; many of these nuisance mechanism enzymatic assay hits had weak micromolar activities and showed either a flat or highly irregular SAR [14]. Schoichet’s team determined that the aberrant behavior in biochemical screening assays was driven by poor solubility resulting in compound aggregate formation. These compound aggregates, present in extremely low concentration, serve as protein sinks, adsorbing most of the target protein, yielding what appeared to be detectable but weak inhibition [15]. His team demonstrated that many of these aggregation‐based inhibitors could be culled from screening hits by comparing activity in an assay with no or very low detergent to a high detergent assay condition. Compounds that lose activity in the high detergent assay were likely to be uninteresting nuisance hits.

Several biophysical methods complement the differential detergent biochemical assay [16]. In the biochemical assay approach, the presence of aggregates is inferred, whereas in the biophysical approaches, the aggregates are directly detected. SPR is uniquely suited such direct detection of nuisance behavior in a buffer matched to the original biochemical screening buffer [17]. Aggregated compounds generate complex binding responses that are not simple 1 : 1 interactions but rather reflect the partitioning of the aggregated compound between the free buffer and the protein captured on the sensor chip. Aggregated compounds also show complex binding to the sensor surface with no target protein captured, providing a simple, parallel means to detect nonideal interactions in real time during library screening. Hit validation workflows now commonly employ SPR, mass spectrometry, and other biophysical methods to remove nuisance mechanism hits [18].

A third trend driving the reemergence of biophysics in drug discovery is the desire to optimize kinetic or thermodynamic properties with an aim to rapidly progress from a first‐in‐class compound to a best‐in‐class compound. When comparing a first‐in‐class compound to a best‐in‐class compound, the best‐in‐class molecule generally has high selectivity for the pharmacologic target and consequently a lengthy residence time with that target [19]. Detailed understanding of compound binding kinetics [20] and inhibitory mechanism leads to better candidates with properties more like an ideal best‐in‐class compound [21]. SPR allows real‐time analysis of binding kinetics [22]; streamlined experimental approaches allow rapid compound sorting based on kinetic parameters [23]. Combining thermodynamic data with affinity and kinetic data further characterizes the intermolecular interactions, enabling detailed SAR and further compound optimization [24]. This idea is explored and different methods applied inform interaction quality in Chapters 2, 4, 7, and 11.

The text concludes with a case study in Chapter 14 that joins many of the methods and concepts discussed in earlier chapters. The Pfizer research team used a combination of traditional biochemical analysis, focused structural information derived from NMR, SPR kinetics, and NMR dynamics to optimize a Staphylococcus aureus DHFR inhibitor. Data from no one method assured success; it was the conjunction of data from the several biophysical techniques that enabled their focused, hypothesis‐driven prospective library design that ultimately yielded novel, nonacid cell‐active inhibitors. Importantly, the dynamics and kinetic data incorporated common resistance mutations, informing the library design and ultimately the candidate compounds. This discovery case study exemplifies the fully integrated discovery approach where data‐rich biophysical techniques continually inform discovery. This approach enables research teams to target transient protein conformations, protein–protein interaction surfaces, or complex enzyme targets—all examples of targets that have met will have little success with traditional high‐throughput enzymatic screening [25].

This text is a survey of contemporary biophysical methods in drug discovery. Biophysical methods report on intermolecular interactions directly with rich detail; these methods naturally complement traditional high‐throughput screening [26, 27], particularly when attacking irregular, nonenzymatic [28, 29], or membrane protein [30, 31] targets.

References

1. Leach, A. R. and Hann, M. M. Molecular complexity and fragment‐based drug discovery: ten years on. Curr. Opin. Chem. Biol. 15:489–496 (2011).

2. Hann, M. M., Leach, A. R., and Harper, G. Molecular complexity and its impact on the probability of finding leads for drug discovery. J. Chem. Inf. Comput. Sci. 41:856–864 (2001).

3. Boyd, S. M., Turnbull, A. P., and Walse, B. Fragment library design considerations. WIREs Comput. Mol. Sci. 2:868–885 (2012).

4. Lau, W. F., Withka, J. M., Hepworth, D., Magee, T. V., Du, Y. J., Bakken, G. A., et al. Design of a multi‐purpose fragment screening library using molecular complexity and orthogonal diversity metrics. J. Comput. Aided Mol. Des. 25:621 (2011).

5. Mashalidis, E. H., Sledz, P., Lang, S., and Abell, C. A three‐stage biophysical screening cascade for fragment‐based drug discovery. Nat. Protoc. 8:2309–2324 (2013).

6. Joseph‐McCarthy, D., Campbell, A. J., Kern, G., and Moustakas, D. Fragment‐based lead discovery and design. J. Chem. Inf. Model. 54:693–704 (2014).

7. Kim, H. Y. and Wyss, D. F. NMR screening in fragment‐based drug design: a practical guide. Methods Mol. Biol. 1263:197–208 (2015).

8. Dias, D. M. and Ciulli, A. NMR approaches in structure‐based lead discovery: recent developments and new frontiers for targeting multi‐protein complexes. Prog. Biophys. Mol. Biol. 116:101–112 (2014).

9. Pilger, J., Mazur, A., Monecke, P., Schreuder, H., Elshorst, B., Bartoschek, S., et al. A combination of spin diffusion methods for the determination of protein‐ligand complex structural ensembles. Angew. Chem. 54:6511–6515 (2015).

10. Cala, O. and Krimm, I. Ligand‐orientation based fragment selection in STD NMR screening. J. Med. Chem. 58:8739–8742 (2015).

11. Perspicace, S., Banner, D., Benz, J., Müller, F., Schlatter, D., and Huber, W. Fragment‐based screening using surface plasmon resonance technology. J. Biomol. Screen. 14:337–349 (2009).

12. Kreatsoulas, C. and Narayan, K. Algorithms for the automated selection of fragment‐like molecules using single‐point surface plasmon resonance measurements. Anal. Biochem. 402:179–184 (2010).

13. Jerabek‐Willemsen, M., Wienken, C. J., Braun, D., Baaske, P., and Duhr, S. Molecular interaction studies using microscale thermophoresis. Assay Drug Dev. Technol. 9:342–353 (2011).

14. McGovern, S. L., Caselli, E., Grigorieff, N., and Shoichet, B. K. A common mechanism underlying promiscuous inhibitors from virtual and high‐throughput screening. J. Med. Chem. 45(8):1712–1722 (2002).

15. McGovern, S. L., Helfand, B. T., Feng, B., and Shoichet, B. K. A specific mechanism of nonspecific inhibition. J. Med. Chem. 46(20):4265–4272 (2003).

16. Feng, B. Y., Simeonov, A., Jadhav, A., Babaoglu, K., Inglese, J., Shoichet, B. K., and Austin, C. P. A high‐throughput screen for aggregation‐based inhibition in a large compound library. J. Med. Chem. 50(10):2385–2390 (2007).

17. Giannetti, A. M., Koch, B. D., and Browner, M. F. Surface plasmon resonance based assay for the detection and characterization of promiscuous inhibitors. J. Med. Chem. 51:574–580 (2008).

18. Lee, H., Zhu, T., Patel, K., Zhang, Y.‐Y., Truong, L., Hevener, K. E., et al. High‐throughput screening (HTS) and hit validation to identify small molecule inhibitors with activity against NS3/4A proteases from multiple hepatitis C virus genotypes. PLoS One 8(10):e75144 (2013). doi:10.1371/journal.pone.0075144.

19. Copeland, R. A. The dynamics of drug‐target interactions: drug‐target residence time and its impact on efficacy and safety. Expert Opin. Drug Discov. 5:305–310 (2010).

20. Danielson, U. H. Integrating surface plasmon resonance biosensor‐based interaction kinetic analyses into the lead discovery and optimization process. Future Med. Chem. 1:1399–1414 (2009).

21. Zhang, R. and Monsma, F. Binding kinetics and mechanism of action: toward the discovery and development of better and best in class drugs. Expert Opin. Drug Discov. 5:1023–1029 (2010).

22. Day, Y. S. N., Baird, C. L., Rich, R. L., and Myszka, D. G. Direct comparison of binding equilibrium, thermodynamic, and rate constants determined by surface‐ and solution‐based biophysical methods. Protein Sci. 11:1017–1025 (2002).

23. Huber, W. A new strategy for improved secondary screening and lead optimization using high‐resolution SPR characterization of compound–target interactions. J. Mol. Recognit. 18:273–281 (2005).

24. Winquist, J., Geschwindner, S., Xue, Y., Gustavsson, L., Musil, D., Deinum, J., and Danielson, U. H. Identification of structural‐kinetic and structural‐thermodynamic relationships for thrombin inhibitors. Biochemistry 52:613–626 (2013).

25. Makley, L. N. and Gestwicki, J. E. Expanding the number of druggable targets: non‐enzymes and protein‐protein interactions. Chem. Biol. Drug Des. 81:22–32 (2013).

26. Genick, C. C., Barlier, D., Monna, D., Brunner, R., Bé, C., Scheufler, C., and Ottl, J. Applications of biophysics in high‐throughput screening hit validation. J. Biomol. Screen. 19:707–714 (2014).

27. Schiebel, J., Radeva, N., Köster, H., Metz, A., Krotzky, T., Kuhnert, M., et al. One question, multiple answers: biochemical and biophysical screening methods retrieve deviating fragment hit lists. ChemMedChem 10:1511–1521 (2015).

28. Wendt, M. D., Sun, C., Kunzer, A., Sauer, D., Sarris, K., Hoff, E., et al. Discovery of a novel small molecule binding site of human survivin. Bioorg. Med. Chem. Lett. 17:3122–3129 (2007).

29. Vassilev, L. T., Vu, B. T., Graves, B., Carvajal, D., Podlaski, F., Filipovic, Z., et al. In vivo activation of the p53 pathway by small‐molecule antagonists of MDM2. Science 303:844–848 (2004).

30. Aristotelous, T., Ahn, S., Shukla, A. K., Gawron, S., Sassano, M. F., Kahsai, A. W., et al., Discovery of β2 adrenergic receptor ligands using biosensor fragment screening of tagged wild‐type receptor. ACS Med. Chem. Lett. 4:1005–1010 (2013).

31. Christopher, J. A., Brown, J., Doré, A. S., Errey, J. C., Koglin, M., Marshall, F. H., et al. Biophysical fragment screening of the β1‐adrenergic receptor: identification of high affinity arylpiperazine leads using structure‐based drug design. J. Med. Chem. 56:3446–3455 (2013).

Note

* Current address: Widener University Delaware Law School, Wilmington, USA

2

Thermodynamics in Drug Discovery

Ronan O’Brien1, Natalia Markova2, and Geoffrey A. Holdgate3

¹ Business Development‐MicroCal, Malvern Instruments, Northampton, MA, USA

² Scientific Marketing Biosciences, Malvern Instruments, Stockholm, Sweden

³ Structure and Biophysics, Discovery Sciences, AstraZeneca, Cambridge, UK

2.1 Introduction

For the drug discovery scientist, the term thermodynamics refers to the study of the heat change that occurs when biomolecules interact. It can be measured either directly by isothermal titration calorimetry (ITC) or indirectly by using any technique that can be used to determine an affinity over a range of temperatures such as surface plasmon resonance (SPR) or fluorescence.

The change in temperature that occurs when molecules interact is, for all practical purposes, a universal phenomenon and has led to the use of ITC to study a wide variety of biomolecular interactions; these include, but are not limited to, protein–small molecule, protein–protein, protein–nucleic acid, protein–metal ion, protein–carbohydrate, nucleic acid–nucleic acid, and ion–ion interactions. The broad applicability of ITC and the exceptionally low errors in affinity determination typically observed using the technique have made it the gold standard for measuring KD [1].

In addition to being a convenient label‐free probe for studying interactions, the heat change is related to the binding enthalpy (ΔH) of the interaction and, taken together with the affinity KD, can be used to calculate the change in entropy of the process. This thermodynamic data gives insight into the non‐covalent forces responsible for driving binding and recognition. It can be used to direct SAR programs and help reveal the energetic hot spots that are key for molecular recognition and that need to be retained throughout lead optimization.

In this chapter we present an overview of the current use of thermodynamics in the drug discovery process. This includes a brief outline of the techniques employed to generate thermodynamic data as well as more detailed discussion of the complexities surrounding the data interpretation. In addition, the utility of enthalpy as a probe for binding in fragment‐based drug discovery programs and for understanding complex interactions will be highlighted.

2.2 Methods for Measuring Thermodynamics of Biomolecular Interactions

Thermodynamic data can be obtained either directly by ITC or indirectly by any method that can be used to determine a KD as a function of temperature such as SPR or fluorescence.

2.2.1 Direct Method: Isothermal Titration Calorimetry

Isothermal titration calorimeters measure the heat change that occurs when two molecules interact. Heat is liberated or absorbed as a result of the redistribution of non‐covalent bonds when the interacting molecules go from the free to the bound state. ITC monitors these heat changes by measuring the differential power required to maintain zero temperature difference between a reference and a sample cell as the binding partners are mixed.

The reference cell usually contains water or buffer, while the sample cell contains one of the binding partners and a stirring syringe that holds the other binding partner (the ligand). The ligand is injected into the sample cell, typically in 0.5–2 µl aliquots, until the ligand concentration is two‐ to threefold greater than the sample. Each ligand injection results in a heat pulse that is integrated with respect to time and normalized for concentration to generate a titration curve of kcal/mol versus molar ratio (ligand/sample). A binding model is fitted to the the resulting isotherm (data) to obtain the affinity (KD), stoichiometry (N), and enthalpy of interaction (ΔH). The Gibbs free energy (ΔG) and the change in the entropy (ΔS) upon binding can then be calculated using the relationship

(2.1)

where R is the gas constant and T is the absolute temperature in Kelvin. In addition to these parameters, it is possible to determine the change in heat capacity of an interaction (ΔCp) by determining the change in enthalpy at different temperatures (T) and using the relationship

(2.2)

2.2.2 Indirect Methods: van’t Hoff Analysis

2.2.2.1 Enthalpy Measurement Using van’t Hoff Analysis

It is possible to access enthalpy and entropy values without the need for calorimetric experiments. These thermodynamic parameters may be estimated using indirect methods, which make use of the temperature dependence of the binding affinity, by employing the van’t Hoff equation. This allows estimates of entropy and enthalpy to be made using any technique that allows the determination of the binding affinity at a range of temperatures. Equation 2.3 is an integrated form of the van’t Hoff equation, and it is clear from inspection that the enthalpy can be derived from changes in binding affinity as long as the constant pressure heat capacity change upon ligand binding (∆Cp) is known or can be fitted. The binding entropy can then be determined from the Gibbs–Helmholtz equation in the usual way. Thus, a number of alternative methods to measure KD, including SPR, microscale thermophoresis (MST), fluorescence, and radioligand binding assays can be used to determine van’t Hoff enthalpies. The experimental design should be such that binding affinities are determined over a wide temperature range (within which the protein retains its native fold) so that the enthalpy change associated with binding can then be calculated using the van’t Hoff relationship shown in Equation 2.3.

(2.3)

where the values for K1 and K2 are the dissociation constants at different temperatures, T1 and T2.

The use of the indirect van’t Hoff approach is not without potential difficulties. Firstly, binding enthalpy is itself temperature dependent, and so the inclusion of the ∆Cp term is required. Estimating ∆Cp in the absence of calorimetric data is often difficult, as deriving ∆Cp from Equation 2.3 requires true curvature in the van’t Hoff plot to be distinguishable from apparent curvature due to errors in the affinity measurement. Hence, this indirect approach requires accurate and precise KD values. Secondly, the temperature‐dependent change in ∆G often is relatively small, which makes deriving the two correlated parameters from this data quite challenging, which may result in relatively large uncertainty in the derived enthalpy compared with the direct calorimetric approach.

2.3 Thermodynamic‐Driven Lead Optimization

The observation by Ernesto Freire [2] that for two drug classes, the HIV protease inhibitors and the statins, the best‐in‐class drugs have the most favorable binding enthalpy has driven many drug discovery laboratories to include thermodynamic data in their decision‐making processes.

It has also been suggested that thermodynamic profiles could be used to identify inhibitors that were optimized for a number of properties including flexibility, to minimize drug resistance caused by rapid mutation of the target binding site [3]; specificity, to reduce side effects caused by nonspecific binding [4–6]; and solubility in water, to maximize the ligand efficiency of polar interactions [7, 8].

2.3.1 The Thermodynamic Rules of Thumb

In the last 10 years or so, a series of guidelines have emerged that have been broadly used to interpret thermodynamic data and have been proposed as key drivers for lead optimization programs [9, 10]. At the simplest level they can be summarized as:

Hydrogen bonds have a favorable enthalpy.

Hydrophobic interactions have a favorable entropy.

Conformational changes are entropically unfavorable.

By applying these guidelines the medicinal chemist can, in theory, test the success or failure of their optimization strategies. For example, if an effective hydrogen bond was successfully introduced, then one would expect to see an increase in the affinity of the interaction and a more negative enthalpy. If so, further iterations could be tested, and if not, determination of the complex structure may reveal some interesting and unexpected SAR. Equally, the success or failure of strategies to rigidify a ligand scaffold can be assessed by monitoring any reduction in unfavorable entropy of an interaction.

A good example of this type of approach, and the use of these rules of thumb, is the interpretation of the thermodynamic data for the interaction of a parent inhibitor (KNI‐10026) and two derivatives (KNI‐10007 and KNI‐10006) with plasmepsin II, an antimalarial target [11] (see Figure 2.1).

Schematic illustrating the ΔH and KD for the interaction (depicted by arrows) of the parent inhibitor KNI‐10026 (left) and two derivatives, KNI‐10007 (top right) and KNI‐10006 (bottom right) with plasmepsin II.

Figure 2.1 The ΔH and KD for the interaction of the parent inhibitor KNI‐10026 (left) and two derivatives, KNI‐10007 (top right) and KNI‐10006 (bottom right) with plasmepsin II.

Source: Freire [11]. Reproduced with permission from John Wiley & Sons, Inc.

The introduction of a hydroxyl group to the parent compound resulted in an increase in the favorable enthalpy of binding from −1.2 to −6.0 kcal/mol that is consistent with the introduction of an additional hydrogen bond. However there was a concomitant reduction in the affinity from 16 to 76 nM due to the greater entropy loss. This enthalpy–entropy compensation (EEC) is common in lead optimization and will be described in more detail elsewhere in this chapter. By changing the stereochemistry of the hydroxyl group in the second inhibitor, the affinity of the interaction was increased to 0.5 nM. In this case the enthalpic advantage of the additional hydrogen bond was maintained while minimizing the entropy loss. The differences in the change in entropy of this interaction were attributed to the additional burial of hydrophobic groups in the binding pocket for the tighter binder KNI‐10006.

Either coincidentally or because of the emergence of ITC as a convenient assay to determine the quality of a hydrogen bond, there have been a number of articles promoting enthalpy‐driven lead optimization strategies [4, 7]. It is clearly an attractive prospect to be able to quickly develop a drug with high efficacy using a combination of ITC, X‐ray crystallography, molecular modeling, and medicinal chemistry. However, more recently, and perhaps not surprisingly, examples have emerged [1] suggesting that thermodynamic lead optimization is more complex than originally thought. Here we outline a number of additional factors that need to be considered when attempting thermodynamic lead optimization.

2.3.2 Enthalpy–Entropy Compensation

EEC is a phenomenon that has been discussed in the scientific literature over many years. EEC appears to be a real and demonstrable effect that many groups have experienced, but the cause may be due to more than one effect occurring across and within the experimental measurements [9]. The basic proposal is quite simple. Consider complex formation between a target protein and a ligand. This binding event is the result of the disruption of interactions of each free partner with the solvent, forming new interactions with each other in the complex. During optimization, the structure of the ligand is modified in order to produce increased bonding interactions with the protein binding site. This will tend (generally) to make ∆H° more negative. However, by introducing further points of interaction, there tends to be an increased order in the complex as a result of the modification, producing a more unfavourable contribution to ∆S°. Often, these two opposing contributions to ∆G° tend to be of similar magnitude in many studies on biological systems. Hence, the traditional medicinal chemistry approach of building new chemical functionality into a molecule to improve the interaction with the binding site (favorable enthalpy) tends to introduce constraints to movement of the molecule and potentially the protein (unfavorable entropy). Since these two parameters oppose each other, the overall ∆G° of binding often is relatively unchanged, confounding the aim to improve affinity. Perhaps this cause of EEC is not surprising, given that both ∆H° and ∆S° themselves are dependent upon ∆Cp. However, there are also other reasons to experience real or apparent compensation. Firstly, errors in the measurement of enthalpy can give rise to equal and opposite errors in the calculated entropy. These errors may arise directly from errors in the measured ligand concentration, which has a larger effect on ∆H° than on ∆G° and also from the error in the measured parameter values due to the uncertainty in the nonlinear regression fit. Secondly, there is a relatively narrow window, both in terms of measurable binding affinities that can be accessed by ITC and in terms of those affinity values that are physiologically relevant. This narrow window that restricts the measurable range of ∆G°, but not ∆H° or ∆S°, will lead to apparent compensation when plotting ∆H° versus T∆S°.

So, the observation of EEC, real or apparent, means that changing the structure of the compound often tends to have a larger effect on ∆H° than on ∆G° (usually measured from KD or IC50 values). This has an important implication for medicinal chemistry, as there may be differences in ∆H°, but not ∆G°, for related compounds that may form very different interactions with the protein. And in the extreme case, difference in binding mode, caused by modifications to the compound, resulting in very different thermodynamic signatures, but with little change in affinity may be observed using ITC measurements. This is important so that SAR may be understood in the context of changed interactions in a common binding mode rather than being complicated by trying to understand changes across completely different sets of interactions.

EEC, and how to overcome it, thus represents a key challenge for medicinal chemistry, when the goal is to make modifications to lead compounds in order to improve the affinity for the target protein.

Additionally, the general rules of thumb described earlier (see section 2.3.1) often do not hold up to thorough scrutiny, and attempting to make modifications to lead compounds using enthalpic and entropic signatures during the optimization process is not trivial, as numerous effects may contribute to the observed values.

Given that EEC, at least to some degree, may be expected during the optimization process, how can the medicinal chemist use the available information advantageously? A combination of biophysically determined thermodynamic data and high quality structural information is still highly useful in attempting to overcome the compensatory tendency, for example, by directing hydrogen bonds to already structured regions of the protein or by utilizing multiple hydrogen bonding interactions from a single group, such that the entropic penalty has already been paid [12].

The presence of EEC suggests that even if specific interactions with potentially enthalpic signatures can be introduced, it is not clear that these will result in increases in binding affinity. Therefore, this may not be a strategy that should be used without having a strong understanding of the binding site and the flexibility and dynamics of protein in the binding site vicinity and beyond. Certainly, in the absence of this knowledge, attempting to design for improvements in affinity directly may prove more