Overcoming Roadblocks in Lead Optimization: A Thermodynamic Perspective

One of the most important goals in the development of lead compounds is the optimization of their binding affinity towards the intended target, as binding affinity is directly related to potency. Moreover, this goal needs to be accomplished in ways that do not compromise important properties like solubility or bioavailability. Recent experimental and computational developments permit monitoring of the thermodynamic forces that define the binding affinity, the enthalpy and entropy changes, thus providing a rigorous guideline to the optimization process. Having access to individual components of the binding affinity rather than its overall value accelerates the optimization process and facilitates the achievement of extremely high affinity. Binding affinity originates from different types of interactions between the drug molecule and the target protein as well as their interactions with the solvent (water). In the optimization of binding affinity, there are terms that can be controlled by the designer and terms that are beyond control. The loss of translational degrees of freedom, for example, cannot be altered. For competitive inhibitors that target the same site, the energy associated with conformational changes in the protein is usually constant and cannot be manipulated. The situation is obviously different for drugs that operate allosterically by modulating conformational changes. From the viewpoint of affinity optimization, the binding energy can be viewed as the difference between the interaction energy with the target and the desolvation energy of the drug molecule. These interactions contribute in a characteristic fashion to the enthalpy and entropy of binding, two quantities that can be measured experimentally by isothermal titration calorimetry (ITC) and that can be used to guide the optimization process. At the thermodynamic level, the binding affinity is determined by the magnitude of the Gibbs energy (ΔG), which is a function of only two terms, the enthalpy (ΔH) and the entropy (ΔS) changes. As the enthalpy and entropy contribute to the binding energy in an additive fashion (ΔG = ΔH − TΔS), it is clear that an infinite number of enthalpy and entropy values can add up to yield the same Gibbs energy value. Compounds that exhibit the same ΔG will bind to the target with the same affinity; however, compounds that are either predominantly enthalpic or entropic will differ in other aspects, as the enthalpy and entropy changes originate from different types of interactions. The enthalpy change reflects the strength of drug/target interactions in relation to those with the solvent. The favorable term arises primarily from van der Waals and hydrogen bonding interactions between drug and target. Two major terms define the binding entropy; the first one is the solvation entropy associated with the burial from the solvent of hydrophobic groups, and the second one is the conformational entropy, which usually reflects the loss of conformational degrees of freedom upon binding. From the engineering point of view, a favorable enthalpy change is obtained from good geometric complementarity between drug and target and the proper location of hydrogen bond donors and acceptors. As these interactions are stereo-specific, a favorable enthalpy change is not only an important contributor to affinity but also to selectivity. The solvation entropy change reflects a repulsion of the drug from the solvent rather than an attractive interaction with the target. This is a favorable but non-specific force proportional to the hydrophobicity of the drug. The conformational entropy change, on the other hand, usually reflects a loss of conformational degrees of freedom in the drug molecule and protein, being therefore an unfavorable term. The magnitude of the conformational entropy loss can be reduced by introducing conformational constraints in the drug molecule so that it occupies similar conformations in the free and bound states. The binding affinity of a compound can be improved by generating a favorable binding enthalpy, favorable solvation entropy, or by minimizing the unfavorable conformational entropy. Obviously, extremely high affinity is achieved when the three factors are optimized simultaneously. The degree of difficulty associated with optimizing the enthalpy is not the same as the one associated with optimizing the entropy. Historically, it has proven much easier to optimize the entropy. As the major favorable contributor is the hydrophobic effect, which is proportional to the number of non-polar groups that are buried from the solvent, the tendency throughout the years has been toward an increase in the hydrophobicity of drug candidates (1, 2). Medicinal chemists have long learned to conformationally constrain and pre-shape molecules to the geometry of the binding site, which completes the entropy optimization. According to these traditional precepts, affinity is achieved through hydrophobicity, and selectivity is achieved through shape complementarity. There is a limit, however, to the hydrophobic character that can be imparted to a compound before it becomes completely insoluble and useless as a drug molecule. At some point in the optimization process, it becomes necessary to introduce favorable enthalpic interactions if the goal is to achieve nanomolar or sub-nanomolar affinities. The exact threshold depends on the characteristics of the target site. Compounds that exhibit extremely high affinity have been shown to display both favorable entropic and enthalpic interactions (3-5). Despite the limits to affinity, a compound that derives selectivity primarily from shape complementarity is prone to lose some when confronted with homologous enzymes with structurally similar binding pockets. Even though enthalpic interactions are required for extremely high affinity and improved selectivity, the optimization of the binding enthalpy has been notoriously more difficult than the optimization of the binding entropy, the reason being that the enthalpy of desolvation of polar groups is very large and unfavorable, as shown in Table 1. Polar groups carry a desolvation penalty about one order of magnitude larger than non-polar groups. A polar group needs to establish a very good interaction with the target in order to compensate for the desolvation penalty and make a favorable contribution. For this reason, they are often engineered as solubilizers of otherwise extremely hydrophobic compounds rather than major contributors to affinity. As the major contributors to the binding enthalpy are polar groups, a common misconception is that enthalpically driven compounds must be highly polar and that consequently their bioavailability will be compromised. In fact, what is often observed experimentally is that compounds with the same number of polar groups have vastly different binding enthalpies. For example, among the HIV-1 protease inhibitors, saquinavir and TMC-126 have exactly the same number of polar groups; however, saquinavir binds to the protease with an unfavorable enthalpy of 1.5 kcal/mol, whereas TMC-126 does so with a very favorable binding enthalpy of −12 kcal/mol. To generate a favorable binding enthalpy, it is not the number of polar groups that matters but the quality of their interactions with the target. It is better to have few groups that establish strong interactions than a large number of groups mostly paying the desolvation penalty. In fact, it has been shown that there is no correlation between the enthalpic character of a compound and the Lipinski rules of five (4). Fortunately, the situation is changing on two fronts. Experimentally, ITC permits monitoring of the enthalpy and entropy changes throughout the optimization process and therefore a direct evaluation of the thermodynamic consequences of introducing different functionalities at specific sites. At the computational level, the success of the initial work dealing with the derivation of empirical correlations between binding enthalpy and structural parameters (6) has led to new ways of predicting enthalpy from structure and of predicting the enthalpic effects expected from the introduction of different functionalities into a given scaffold. One of the ongoing projects in this laboratory is the development of plasmepsin inhibitors as new anti-malarial drugs (7). Starting with the allophenylnorstatine scaffold that mimics the main cleavage site in the hemoglobin molecule of infected victims, we have been able to generate high affinity inhibitors with Ki's in the high picomolar range (7, 8). The evolution of the potency of these compounds reflects the situation encountered in most drug discovery laboratories when a given chemical scaffold begins to be optimized. Starting with hits characterized by Ki's in the micromolar range, the goal is to increase potency by three to five orders of magnitude, i.e. an increase in the Gibbs energy of binding of 4–7 kcal/mol. How can this be achieved? How do the individual components of the Gibbs energy advance? Figure 1 shows the evolution of the contribution of the enthalpy change to the total Gibbs energy of binding as the affinity of the compounds to plasmepsin II is optimized from the micromolar to the high picomolar level. It is immediately apparent that low affinity compounds can exhibit a wide range of enthalpy/entropy combinations. In other words, low affinity can be generated by essentially any combination of hydrophobic and polar interactions. As the affinity increases, the range of enthalpy/entropy combinations narrows and appears to converge to a smaller range of values as the maximal affinity is approached. For this series of plasmepsin II inhibitors based upon the allophenylnorstatine scaffold, the highest affinity was achieved with a binding enthalpy of −4.5 kcal/mol and an entropic contribution (−TΔS) of −8.8 kcal/mol. Enthalpic contribution to the Gibbs energy of binding (ΔH/ΔG) versus the logarithm of the binding affinity (Log Ka) for 71 allophenylnorstatine inhibitors of plasmepsin II. All thermodynamic parameters were determined under identical conditions (10 mm formate buffer, pH 4.0, 2% DMSO, at 25 °C). The data in Figure 1 clearly demonstrates the importance of balancing enthalpic and entropic contributions in order to maximize binding affinity and illustrates important steps in the design process. As enthalpic interactions are more difficult to engineer, enthalpically driven hits are usually easier to optimize than entropically driven ones; i.e. it is less costly energetically to introduce hydrophobic groups. A calorimetric characterization of hits identified by screening or any other method should allow the designer to recognize the nature of the forces by which the hits bind to the target. This step is crucial at these early stages, because it allows separation of those molecules that bind because they are excluded from the solvent from those that bind because they establish favorable interactions with the target. It is at the earlier stages where the spread of enthalpy/entropy combinations is maximal and where a careful decision needs to be made. It is always advantageous to choose compounds that establish good interactions with the target and thermodynamic dissection provides that information. Further down the optimization road, thermodynamic dissection indicates if the process is driven within a reasonable pathway, avoiding thermodynamic extremes that sooner or later lead the process to a roadblock and sometimes a dead end. Obviously, this task is facilitated if it is supplemented by algorithms able to predict the enthalpic or entropic consequences of introducing different chemical functionalities in the scaffold under optimization. This work was supported by grants from the National Institutes of Health GM 57144 and GM56550 and the National Science Foundation MCB0131241.