Journal of Molecular Biology
More than a Simple Lipophilic Contact: A Detailed Thermodynamic Analysis of Nonbasic Residues in the S1 Pocket of Thrombin
Introduction
Nearly two decades of drug research on small molecules inhibiting the serine protease thrombin, an established target for the prevention of cardiovascular diseases, have revealed a plethora of highly potent drug candidates.1 Most of the early development programs followed a strategy to address the S1 specificity pocket of the enzyme with basic molecular moieties to mimic the arginine side chain of the natural substrate fibrinopeptide A.2 However, as a consequence of being positively charged under physiological conditions, they generally have poor pharmacokinetic properties. In order to address this deficiency, sophisticated prodrug strategies have been developed to achieve sufficient bioavailabilty.3 Ongoing efforts to produce therapeutically useful orally available thrombin inhibitors resulted in the discovery of the uncharged m-chloro-benzylamides as promising alternatives to the prodrug strategy.4 They are assumed to form a lipophilic contact in the S1 pocket,5– 8 which has the polar carboxylate group of Asp189 at its far end. Therefore the rationalization put forward5, 6, 7, 8 to explain the pronounced binding affinities of the m-chloro derivatives as predominantly hydrophobic binding appears unsatisfactory. Detailed investigations with respect to the binding thermodynamics of these extremely promising ligands to elucidate their binding characteristics are described in the present contribution.
We present a detailed study, including X-ray crystallography and isothermal titration calorimetry, on a congeneric series of thrombin inhibitors to find possible explanations for these so far poorly understood structure–activity relationships.
An electrostatic attractive force resulting from a polarized chlorine atom above a tyrosine π-system has been demonstrated recently for chloro-aromatic inhibitors addressing the structural related S1 pocket of coagulation factor Xa.9, 10, 11, 12, 13 An analysis of crystal structures from the Protein Data Bank along with ab initio calculations showed the Cl–π interaction as a clearly attractive force, which was assigned to the dispersion force as major source of attraction.14 A better understanding of the electrostatic component of this interaction may also make medicinal chemists more aware of the possible nature of such interactions and the transfer of this knowledge to other related and ongoing drug design projects. The astonishingly high ligand efficiency of triclosane to its molecular target FabI,15, 16, 17, 18 a ligand that intuitively contradicts our keen sense for an optimal lead compound, re-emphasizes the necessity to better understand the driving forces of chloro-aromatic interactions in protein–ligand complex formation.
The characterization of the determinants for molecular recognition of small molecules by a macromolecular target is an essential requirement for any drug development project that intends to start from rational grounds.19 Whereas the foundation of these mostly non-covalent interactions is reasonably well understood, it is still a major challenge to obtain detailed knowledge about the individual thermodynamic parameters and their respective contributions to binding affinity. Nevertheless, studies providing this detailed information would allow us to predict more precisely ligand binding affinity as a prerequisite for successful in silico screening.20, 21, 22
The interplay of intra- and intermolecular forces between ligand, receptor and solute makes the binding event a process of challenging thermodynamic complexity. A promising approach to unravelling this complexity is to perform systematic studies on interacting molecules with only minor differences.23 Few systematic studies are available,24, 25, 26 and they often demonstrate paradoxical or unexpected results.27 They also emphasize that entropy can significantly affect the Gibbs free energy of binding, and that entropy is notoriously difficult to predict.28 If only the overall binding affinity, as determined by a typical bio-assay, is taken into account, the danger exists that a superficial and oversimplified interpretation of binding energetics will be made.29 Consequently, multiple biophysical methods should be utilized in order to clarify the energetics of the ligand binding process in greater depth.
The serine protease thrombin30 is an ideal model system for performing fundamental studies aimed at providing a better understanding of the energetics of the ligand–protein binding process because its active site is composed of a set of distinct, well-defined, binding pockets that can easily be addressed.31 Crystallography can be performed routinely and the large quantities of enzyme necessary for isothermal titration calorimetry (ITC) are available. Furthermore, thrombin is a target familiar to many researchers in the enzyme inhibitor field and, therefore, using it as a model system facilitates an appreciation of the information obtained as well as its application to other enzyme systems.
Here, we present a “thermodynamic mapping” of a series of D-Phe-Pro-based thrombin inhibitors (Fig. 1), that differ only in the portion binding to the S1 pocket. This ligand series allows us to follow changes in enthalpic and entropic contributions to the Gibbs free energy of binding in a stepwise fashion. A series of 26 ligands has been synthesized and their binding affinities towards human thrombin were first determined in a kinetic assay. For 12 of the most interesting compounds, the Gibbs free energy of binding ΔG0 could be factorized by ITC into an enthalpic ΔH0 and entropic contribution –TΔS0. As a reference state for thrombin before the binding event, a crystal structure of the uncomplexed enzyme was determined, showing an extended hydration pattern of the active site. Five inhibitors with wide-ranging binding affinity, even though they are structurally very similar, have been investigated by X-ray crystallography. Their binding modes elucidate the interaction pattern and hydration in the S1 specificity pocket and allow conclusions about residual ligand mobility in light of their relative differences in the temperature factor.
When combined with solubility and desolvation energy considerations, the data provide a more complete analysis of the driving forces for inhibitor binding, in which unexpected electrostatic attraction and residual entropy play important roles.
Section snippets
Analysis of crystal structures
X-ray crystallography of uncomplexed thrombin resulted in a structure with the active site of the enzyme in its hydrated state. Eight structural water molecules could be assigned to peaks in the difference electron density for the region that is usually occupied by the inhibitor series under consideration. The assignment of these water molecules during refinement was made on the basis of the initial difference electron density and on expedient temperature factors. Although the complete solution
Conclusions
We have presented a congeneric series of thrombin inhibitors, for which the thermodynamic driving forces of complex formation have been studied in detail. The differing thermodynamic binding profiles of closely related ligands could be explained by individual differences in their potential to interact with solvent and protein. In particular, our data accentuate the strong influence of desolvation energy attributed to the part of the inhibitors that is buried deeply in the S1 pocket.
Furthermore,
Synthesis of Boc-D-Phe-L-Pro
L-Proline benzyl ester (772 mg, 3.77 mmol), Boc-D-phenylalanine (999 mg, 3.77 mmol), and N-hydroxybenzotriazole (560 mg, 4.14 mmol) were dissolved in 10 ml of anhydrous dichloromethane (DCM). After 10 min 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (1.07 g, 8.29 mmol) was added, followed at 10 min later by the addition of diispropyl azodicarboxylate (8.29 mmol). The reaction mixture was stirred overnight. The DCM was removed and the residue was taken up in ethyl acetate, then washed three
Acknowledgements
We kindly acknowledge CSL Behring, Marburg, for supplying us with generous amounts of human thrombin from the production of Beriplast®.
We thank the beamline support staff at DESY and BESSY for their advice during data collection, and the BMBF (support code 05ES3XBA/5) for generously supporting travel to BESSY/Berlin.
Author Contributions: MM, MZ and DH: The design and synthesis of the compounds as well as collaborative interpretation of the binding data. BB, CG, AH and GK: The testing of the
References (54)
- et al.
The discovery of orally available thrombin inhibitors: optimisation of the P1 pharmacophore
Bioorg. Med. Chem. Lett.
(1999) - et al.
Unexpected enhancement of thrombin inhibitor potency with o-aminoalkylbenzylamides in the P1 position
Bioorg. Med. Chem. Lett.
(2003) Virtual ligand screening: strategies, perspectives and limitations
Drug Discov. Today
(2006)- et al.
Thermodynamic penalty arising from burial of a ligand polar group within a hydrophobic pocket of a protein receptor
J. Mol. Biol.
(2006) The structure of thrombin, a chameleon-like proteinase
J. Thromb. Haemost.
(2005)- et al.
Solubility and partitioning I: Solubility of nonelectrolytes in water
J. Pharm. Sci.
(1980) - et al.
A new method for the estimation of the aqueous solubility of organic compounds
J. Pharm. Sci.
(1992) - et al.
Synthetic inhibitors of bovine factor Xa and thrombin comparison of their anticoagulant efficiency
Thromb. Res.
(1989) - et al.
Processing of X-ray diffraction data collected in oscillation mode
Methods Enzymol.
(1997) - et al.
Inhibition of human alpha-thrombin by a phosphonate tripeptide proceeds via a metastable pentacoordinated phosphorus intermediate
J. Mol. Biol.
(2001)