Small molecule affinity fingerprinting. A tool for enzyme family subclassification, target identification, and inhibitor design.

Classifying proteins into functionally distinct families based only on primary sequence information remains a difficult task. We describe here a method to generate a large data set of small molecule affinity fingerprints for a group of closely related enzymes, the papain family of cysteine proteases. Binding data was generated for a library of inhibitors based on the ability of each compound to block active-site labeling of the target proteases by a covalent activity based probe (ABP). Clustering algorithms were used to automatically classify a reference group of proteases into subfamilies based on their small molecule affinity fingerprints. This approach was also used to identify cysteine protease targets modified by the ABP in complex proteomes by direct comparison of target affinity fingerprints with those of the reference library of proteases. Finally, experimental data were used to guide the development of a computational method that predicts small molecule inhibitors based on reported crystal structures. This method could ultimately be used with large enzyme families to aid in the design of selective inhibitors of targets based on limited structural/function information.

A molecular basis for the selectivity of thiadiazole urea inhibitors with stromelysin-1 and gelatinase-A from generalized born molecular dynamics simulations.

Matrix metalloproteinases (MMPs) represent a potentially important class of therapeutic targets for the treatment of diseases such as cancer. Selective inhibition of MMPs will be required given the high sequence identity across the family and the discovery that individual MMPs also regulate the natural angiogenesis inhibitor angiostatin. In this study, we have used computational methods to model the selectivity for six thiadiazole urea inhibitors with stromelysin-1 and gelatinase-A, two homologous MMPs that have been implicated in breast cancer. From continuum Generalized Born molecular dynamics (GB-MD) and MM-GBSA analysis, we estimated ligand free energies of binding using 200 snapshots obtained from a short 40 ps simulation of the relevant protein-ligand complex. The MM-GBSA free energies, computed from the continuum GB-MD trajectories, show strong correlation with the experimental affinities (r(2) = 0.74); prior studies have employed explicit water MD simulations. Including estimates for changes in solute entropy in the binding calculations slightly diminishes the overall correlation with experiment (r2 = 0.71). Notably, in every case, the simulation results correctly predict that a given ligand will bind selectively to stromelysin-1 over gelatinase-A which is gratifying given the high degree of structural homology between the two proteins. The increased selectivity for stromelysin-1 appears to be driven by (1) increased favorable van der Waals interactions, (2) increased favorable Coulombic interactions, and (3) decreased unfavorable total electrostatic energies (Coulombic plus desolvation).

Using pharmacophore models to gain insight into structural binding and virtual screening: an application study with CDK2 and human DHFR.

This study provides results from two case studies involving the application of the HypoGenRefine algorithm within Catalyst for the automated generation of excluded volume from ligand information alone. A limitation of pharmacophore feature hypothesis alone is that activity prediction is based purely on the presence and arrangement of pharmacophoric features; steric effects remained unaccounted. Recently reported studies have illustrated the usefulness of combining excluded volumes to the pharmacophore models. In general, these excluded volumes attempt to penalize molecules occupying steric regions that are not occupied by active molecules. The HypoGenRefine algorithm in Catalyst accounts for steric effects on activity, based on the targeted addition of excluded volume features to the pharmacophores. The automated inclusion of excluded volumes to pharmacophore models has been applied to two systems: CDK2 and human DHFR. These studies are used as examples to illustrate how ligands could bind in the protein active site with respect to allowed and disallowed binding regions. Additionally, automated refinement of the pharmacophore with these excluded volume features provides a more selective model to reduce false positives and a better enrichment rate in virtual screening.

SitePrint: three-dimensional pharmacophore descriptors derived from protein binding sites for family based active site analysis, classification, and drug design.

Integrating biological and chemical information is one key task in drug discovery, and one approach to attaining this goal is via three-dimensional pharmacophore descriptors derived from protein binding sites. The SitePrint program generates, aligns, scores, and classifies three-dimensional pharmacophore descriptors, active site grids, and ligand surfaces. The descriptors are formed from molecular fragments that have been docked, minimized, filtered, and clustered in protein active sites. The descriptors have geometric coordinates derived from the fragment positions, and they capture the shape, electrostatics, locations, and angles of entry into pockets of the recognition sites: they also provide a direct link to databases of organic molecules. The descriptors have been shown to be robust with respect to small changes in protein structure observed when multiple compounds are cocrystallized in a protein. Five aligned thrombin cocrystals with an average core alpha-carbon RMSD of 0.7 A gave three-dimensional pharmacophore descriptors with an average RMSD of 1.1 A. On a larger test set, alignment and scoring of the descriptors using clique-based alignment, and a best first search strategy with an adapted forward-looking Ullmann heuristic was able to select the global minimum three-dimensional alignment in twenty-nine out of thirty cases in less than one CPU second on a workstation. A protein family based analysis was then performed to demonstrate the usefulness of the method in producing a correlation of active site pharmacophore descriptors to protein function. Each protein in a test set of thirty was assigned membership to a family based on computed active site similarity to the following families: kinases, nuclear receptors, the aspartyl, cysteine, serine, and metallo proteases. This method of classifying proteins is complementary to approaches based on sequence or fold homology. The values within protein families for correctly assigning membership of a protein to a family ranged from 25% to 80%.