Abstract
Over the last decade, there has been an increased awareness on kinetic and thermodynamic profiling of ligand-protein interactions and its linkage to the clinical effectiveness of drug candidates. Herein we provide an overview on where the pharmaceutical industry stands with kinetic and thermodynamic profiling in drug discovery. We also discuss the challenges for current technologies as well as provide our views on future prospects to use kinetic and thermodynamic profiling for prospective drug design.
Introduction
Traditional drug discovery paradigm has focused primarily on optimizing drug-target binding affinities and in vivo pharmacokinetics while developing lead compounds against a validated drug target. However, in the last decade, it has been proposed that kinetic and thermodynamic profiling of compounds may be better predictors of compound selectivity and in vivo efficacy.1-4
The relationship between a compound’s binding affinity to a drug target and kinetic parameters, in many cases, could be defined as in Figure 1. Association rate constant kon and dissociation rate constant koff are not intrinsically related to one another. Thus, they are correlated with structural features of the ligand, target and the complex in different ways. It is believed that in general, slow koff offers advantages on compound efficacy and safety profile while kon is controlled by the diffusion limit. This came from retrospective analyses indicating many best-in-class drugs possessed slowest koff in its class. One of the classical examples to illustrate this point is the well-known muscarinic M3 receptor antagonists, Tiotropium, the first once-a day bronchodilator that provided a significant advantage over other drugs in its class. Its long lasting effect was attributed to its exceptionally slow koff, which translated into a half-life of 35 hr vs 2-30 mins for other drugs in this class. It’s worth pointing out that the binding affinities of these drugs for M3 receptor are very similar.5
Taken together, kinetic profiling may provide another layer of differentiation for compounds that are otherwise indistinguishable by binding affinity. Thus, it has been advocated that kinetic profiling of compounds should be incorporated at an early stage of the drug discovery processes until a clear decision can be made on what constitutes a desirable kinetic profile for the target and the chemical series of interest.
Similarly, from a thermodynamic perspective, Equation 1 shows the relationship between binding affinity and changes in enthalpy (ΔH) and entropy (ΔS) associated with the complex formation. ΔH represents the heat associated with the making and breaking of noncovalent bonds in going from the free to the bound state. ΔS reports on the overall change in the degree of freedom of a system. On retrospective perusal of the drug thermodynamic data, it is apparent that, despite similar ΔG values, the underlying changes in ΔH and ΔS could be quite different for each interaction.
ΔG = ΔH-TΔS = RTLnKD Equation 1
Freire et al had studied the thermodynamic profiling of HIV protease drugs over the development time course of their development.4 The best in class HIV protease drug, Darunavir, had the highest ΔH contribution compared to other drugs in its class. Similarly, thermodynamic profiling of the binding of a series of statins to HMGCoA reductase also demonstrated that Rosuvastation, the best-in-class drug, had significantly optimized enthalpic contribution compared to the first approved drug in its class, Fluvastatin. The view thus was put forward that ΔH contributions to binding provides a good parameter for compound selection as it maximizes the influence of forces other than hydrophobicity. Obviously, thermodynamic data offers another level of understanding on the formation of protein-ligand interfaces in addition to kinetic profiling and binding affinity studies. Analogous to koff in kinetic profiling, it is proposed that ΔH optimization is required to achieve high efficacy and selectivity. Needless to say, thermodynamic and kinetic profiling provides an approach for optimization of inhibitors guided by independent variables (kon, koff, ΔH and ΔS), rather than by a compound variable (KD).
Figure 1. Dissociation equilibrium constant (KD) and kinetic parameters –interaction between receptor ( R ) and ligand (L) to form the complex RLThese studies suggested a more holistic approach including binding kinetic and thermodynamic data should be incorporated as early as possible in the drug discovery process, to enable medicinal chemists to identify and prioritize hit and lead compounds with best-inclass potentials. Despite the fact validation of some of the claims remains to be established.6,7 The hope is that the proposed impact of binding kinetics and thermodynamics will ultimately lead to clinical candidates. In this review, we will discuss the progresses, challenges and outlook, as an industry, in understanding and utilizing kinetic and thermodynamic profiling for prospective drug design.
Discussion
Thermodynamic and Kinetic Data Acquisition. Surface plasmon resonance (SPR) is one of the most prevalent methods used in drug discovery to obtain kinetic parameters. In SPR, the target protein is immobilized onto a sensor chip. Compounds of interest are then flown over the sensor surface. As the compounds bind to the target protein on the sensor chip, it induces a real time change in the refractive index, which is directly proportional to the mass bound at the surface. SPR is a label free technology for obtaining kon and koff rates, from which binding affinities could be calculated. It is known that the SPR signal is highly sensitive to temperature variation and changes in bulk solvent and measurements need to be performed with a fully dissolved ligand. However, it has been reported that only 7% of the available published and in-house kinetic data at Pfizer had the temperature for the determinations reported. The kinetic profile, kon and koff, of compounds within a chemical series, may very likely span a broad range of values covering several orders of magnitudes. Generally, the upper limit for kon determination by SPR is 10-6 M-1 s-1, whereas the fastest koff that can be accurately measured is approximately 10-1 s-1. This inevitably causes technical challenges for kinetic profiling of compounds with fast kon and koff rates that are beyond instrument detection limits. In our experience, it is not uncommon to have compounds with a fast kinetic profile, especially in the early stage of the drug discovery where the binding efficiency between the compound and the protein target is far from optimal. In one of the hit-to-lead programs we evaluated, more than 25% of the compounds profiled had fast kinetics thus making the selection of hits based on kinetic profile impractical in this case. It is our view that in order to fully explore the potential utility of kinetic parameters in the early stages of a drug discovery program, further advances in instrumentation technology for kinetic measurements are warranted.
For thermodynamic profiling, isothermal titration calorimetry (ITC) is the gold standard to obtain thermodynamic parameters. ITC directly measures the change in enthalpy at a constant temperature by titrating the protein target and a ligand that form an equilibrium complex at known concentrations. From an ITC experiment, binding affinity KD and ΔH can be obtained directly and ΔS can consequently be calculated. ITC allows the highly accurate determination of thermodynamic parameters with no requirement for chemical modification, labelling or immobilization.
With the increasing research on thermodynamic profiling, very recently, it has been reported that protein and compound purities could significantly influence the ITC readout in addition to other experimental parameters. We have also found chiral purity of a compound, in addition to compound purity, could distort the ITC data readout which resulted in false negative readout. These studies demonstrated that ITC experimental parameters must be rigorously monitored in the context of compound structure and purity to obtain accurate and reliable thermodynamic data.
Fast instrument advancements have been seen in the last decade, triggered partially by the increased interest and research in kinetic and thermodynamic profiling. Currently, the throughput of SPR and ITC are not comparable to traditional methods for obtaining binding affinity data. In the review by Miller et al in 2012,8 they were able to extract 398 compounds with both kon and koff values from the Pfizer biological data repository and literature sources. Overall, 1855 compounds had either kon or koff values. This is evidently nowhere near the plethora of thermodynamic affinity data available in the literature. In one of the kinetic and themodynamic profiling study we completed, while more than 800 TR-FRET binding affinities were obtained on one chemical series with fast turnaround times enabling rapid SAR iterations, less than 15% of these analogs were profiled in SPR and ITC, due to the longer testing cycles needed for data collection and processing, making the timely incorporation of this data in compound selection much less feasible. This in turn, limited the practical use of kinetic and thermodynamic information for prospective design, compared to other parameters that are easily obtainable. Nevertheless, significant instrumentation improvements have been reported in recent years. For example, Biacore 8K can potentially characterize 64 interactions in 5 hours. New development in ITC technology allows the instrument to be fully automated with capacity to run four 96-well plates unattended with improved software for data analysis. With the ever increasing interest and research in kinetic and thermodynamic profiling, we expect unprecedented instrumentation development to enable fast kinetic and thermodynamic data acquisition that is comparable to the binding affinity data acquisition in the foreseeable future.
SKR and STR Analysis. The kinetic and thermodynamic profile of structurally different compounds could reveal features and aspects of compound-protein interactions and thus aid optimization. This is defined as structure-kinetics relationships (SKR) and Structurethermodynamic relationships (STR), analogous to the traditional structure-activity relationship (SAR) studies.
The rational design of compounds with desirable or distinct kinetics signatures is still in its infancy primarily due to the lack of molecular understanding of the factors that influence binding kinetics and the difficulty in characterizing transition states of the ligand-target complexes. Despite these challenges, significant progress has been made by gaining knowledge from both the ligand and target perspectives, such as by ligand modifications with subtle changes to probe relevant pharmacophores, mutation studies, and molecular dynamic studies to establish the corresponding “anchoring” sites on the receptor, to name a few.
Many elegant and systematic SKR studies have been reported very recently. The study of a series of tertiary amine muscarinic M3 receptor antagonists by Glosshop et al9 nicely exemplified a systematic combination of both SAR and SKR studies in the early phase of drug discovery process. It was found that the gem-dimethyl substitution had a profound (>38-fold) effect on the dissociation rate. Interestingly, this difference was not observed with the corresponding binding affinities. Prioritization of this structural moiety and further chemical modification yielded a clinical candidate maintaining the gemdimethyl moiety which provided efficacious 24h bronchodilation from a single inhaled dose.
It should also be pointed out that SKR studies are not limited to koff only. Our internal data showed that kon could vary dramatically even among close analogs. Recently literature examples also suggested in certain cases kon may play an important role in downstream processes.10,11 It is also feasible that for certain compound-target protein interactions, the kinetic profile cannot be simplified as kon/koff, with more complex interaction pathways between the compound and the target protein. Systematic SKR studies should be able to help to answer these questions.
One of the challenges in utilizing STR in a predictive manner resulted from the fact that multiple factors such as hydrophobic interactions, solvation and local water structure may influence the thermodynamic readout. Good progress has been made recently in understanding the effects of these factors and how to apply this knowledge in STR studies.12 It is more fruitful to compare similar molecules which bind to the same target binding side in STR studies. The rationale was that because of the structural similarity of these compounds, the effects of many of the factors will largely cancel out.
Interplay of thermodynamic and kinetic data with other parameters. To understand the impact of kinetic and thermodynamic profiling on the time-course of target occupancy and drug effect, one should realize that this profiling is one of the several parameters that will influence drug dosing and effects. For example, many factors can influence the ultimate importance of kinetic profiling in the context of in vivo efficacy such as the concentration profile of the free drug in plasma and at the target site, the concentration of the target, competition between drug and endogenous ligand binding, target turnover etc. It will also be interesting to see how kinetic profiling interplays with thermodynamic profiling. Do compounds with better koff also possess more ΔH contribution? Do kinetic and thermodynatic profiling influence the efficacy and selectivity in similar ways? As more and more SKR and STR studies come aboard, this will enable us to answer these questions and assess the interplay of these parameters in a global and systematic way.
Conclusion
Despite all the challenges, significant progress has been made in the study of kinetic and thermodynamic profiling in drug discovery. For example, the Kinetics for Drug Discovery (K4DD) Consortium, charged with the mission to investigate major open questions related to binding kinetics with public-private partnership, enables targeting drug-binding kinetics in a holistic way. The hope is that as progress continues in these areas, the new insights gained and methods developed will translate to better decision making for selecting drug candidate, ultimately placing kinetic and thermodynamic profiling at the same or higher level as with binding affinity. In the long run, the multipronged strategy will be applied to drug discovery to achieve a thorough understanding of the interplay between structure, kinetics.