Inhibitor Synthesis and Biophysical Characterization of Protein–Ligand–Solvent Interactions An Analysis of the Thermodynamics and Kinetics of Ligand Binding to Thermolysin
In the pre-clinical development stages of most drug design campaigns, the equilibrium binding affinity of a prospective lead candidate, in the form of an IC50, Kd or ΔG° value, is the most commonly employed benchmark parameter for its effectiveness as a putative drug. Hydrogen bonding, van der Waals...
|PDF Full Text
No Tags, Be the first to tag this record!
|In the pre-clinical development stages of most drug design campaigns, the equilibrium binding affinity of a prospective lead candidate, in the form of an IC50, Kd or ΔG° value, is the most commonly employed benchmark parameter for its effectiveness as a putative drug. Hydrogen bonding, van der Waals and electrostatic interactions, as well as hydrophobic effects are among the most prominent factors that contribute to binding. In structure based design approaches, these interactions can routinely be linked to a structural motif of a drug molecule, which can greatly assist in the construction of compounds with a desired set of properties. Equilibrium binding affinity can also be expressed in terms of kinetics, were the steady-state constant Kd is defined as the ratio of the rate constants of dissociation (kd) and association (ka). The thermodynamic expression ΔG° can be subdivided into an enthalpic (ΔH°) and an entropic (–TΔS°) term. In either case, the molecular mechanisms that define the kinetics of binding or the compensation of enthalpic and entropic contributions are not fully understood. The goal of this dissertation is the in-depth investigation of the molecular processes that drive protein–ligand interactions. A special focus is set on the partitioning of thermodynamic and kinetic parameters into their respective microscopic elements. For this, the metalloprotease thermolysin (TLN) is used as a model system. This protein is well characterized and represents a robust system with excellent crystallographic properties and a thoroughly documented inhibitor class.
The first publication (Chapter 2) presents an improved strategy for the synthesis and purification of phosphonamidate peptides that are known as potent inhibitors of TLN. Due to the inherent instability of the phosphorous–nitrogen bond, the introduction of polar functional groups into the inhibitor scaffold is quite challenging. Here, a synthetic strategy is presented that minimizes the amount of hydrolysis during peptide coupling, deprotection and purification through the use of an allyl-based protection system and a solid-phase extraction (SPE) protocol for the final purification step. This allows the synthesis of highly pure TLN inhibitors incorporating a variety of functional groups for use in biophysical experiments.
In the second publication (Chapter 3), a strategy for the design of inhibitors is highlighted, which relies on the targeted design of water networks that are formed around a protein–ligand complex. Based on information from a previous study, the shape of a hydrophobic portion of a TLN ligand is altered in a way that allows a beneficial stabilization of water molecules in the first solvation layer of the complex. Supported by molecular dynamics simulations, a series of diastereomeric inhibitors is synthesized and the binding process is characterized by X-ray crystallography, isothermal titration calorimetry (ITC) and surface plasmon resonance spectroscopy (SPR). The optimization of the hydrophobic P2’ moiety results in a 50-fold affinity enhancement compared to the original methyl substituted ligand. This improvement is mainly driven by a favorable enthalpic term that originates from the stabilization of water polygons in the solvation shell.
In the follow-up study in Chapter 4, the binding signature of a series of inhibitors that place a charged and polar moiety in the solvent exposed S2’ pocket of TLN is investigated. Here, a partially hydrated ammonium group is gradually retracted deeper into the hydrophobic protein environment. From the crystal structures it is evident that the polar ligands do not recruit an increased amount of water molecules into their solvation layer when compared to related analogues that feature a purely aliphatic residue at the solvent interface. The penalty for the partial desolvation of the charged functional group, in combination with the lack of a strongly ordered water network, results in a severe affinity decrease that is driven by an unfavorable enthalpic term.
The deep, hydrophobic S1’ pocket of TLN determines the substrate specificity of the protease and is commonly addressed by high affinity inhibitors. Experimental evidence from previous studies suggests, however, that this apolar crevice is only poorly solvated in the absence of an interaction partner. With the study in Chapter 5, an attempt for the experimental analysis of the hydration state of the S1’ pocket is presented. For this, a special inhibitor is designed that transforms the protein pocket into a cavity, while simultaneously providing enough empty space for the accommodation of several water molecules. A detailed analysis of an experimentally phased electron density map reveals that the cavity remains completely unsolvated and thus, vacuous. As an intriguing prospect for the exploitation of such poorly hydrated protein pockets in drug design, the placement of an iso-pentyl moiety in the ligand’s P1’ position results in a dramatic, enthalpically driven gain in affinity by a factor of 41 000.
With a detailed structural analysis of a series of chemically diverse TLN inhibitors, the kinetics of the protein–ligand binding process are investigated in Chapter 6. From the SPR derived kinetic information, it becomes apparent that the nature of the functional group in the P2’ position of a thermolysin inhibitor has a significant impact on its dissociation kinetics. This property can be linked to the interaction between the respective functionality of a ligand and Asn112, a residue that lines the active site of the protease and is commonly believed to align a substrate for proteolytic cleavage. This residue undergoes a significant conformational change when the protein transitions from its closed state to its open form, from which a ligand is released. Interference with this retrograde induced-fit mechanism through strong hydrogen-bonding interactions to an inhibitor results in a pronounced deceleration of the dissociation process. The case of the known inhibitor ZFPLA demonstrates that a further restriction of the rotation of Asn112 by a steric barrier in the P1 position of a ligand, can reduce the rate constant of dissociation by a factor of 74 000.
Fragment-based lead discovery has become a popular method for the generation of prospective drug molecules. The weak affinity of fragments and the necessity for high concentrations, however, can result in false-positive signals from the initial binding assays that routinely plague fragment-based screening. The pursuit of such a “red herring” can lead to a significant loss of time and resources. In Chapter 7, a molecule that emerged as one of the most potent binders from an elaborate fragment screen against the aspartic protease endothiapepsin is identified as a false-positive. Detailed crystallographic, HPLC and MS experiments reveal that the affinity detected in multiple assays can in fact be attributed to another compound. This entity is formed from the initially employed molecule in a reaction cascade that results in a major rearrangement of its heterocyclic core structure. Supported by quantum chemical calculations and NMR experiments, a mechanism for the formation of the elusive compound is proposed and its binding mode analyzed by X-ray crystallography.