Rescoring docking hit lists for model cavity sites: predictions and experimental testing

Abstract

Molecular docking computationally screens thousands to millions of organic molecules against protein structures, looking for those with complementary fits. Many approximations are made, often resulting in low "hit rates." A strategy to overcome these approximations is to rescore top-ranked docked molecules using a better but slower method. One such is afforded by molecular mechanics-generalized Born surface area (MM-GBSA) techniques. These more physically realistic methods have improved models for solvation and electrostatic interactions and conformational change compared to most docking programs. To investigate MM-GBSA rescoring, we re-ranked docking hit lists in three small buried sites: a hydrophobic cavity that binds apolar ligands, a slightly polar cavity that binds aryl and hydrogen-bonding ligands, and an anionic cavity that binds cationic ligands. These sites are simple; consequently, incorrect predictions can be attributed to particular errors in the method, and many likely ligands may actually be tested. In retrospective calculations, MM-GBSA techniques with binding-site minimization better distinguished the known ligands for each cavity from the known decoys compared to the docking calculation alone. This encouraged us to test rescoring prospectively on molecules that ranked poorly by docking but that ranked well when rescored by MM-GBSA. A total of 33 molecules highly ranked by MM-GBSA for the three cavities were tested experimentally. Of these, 23 were observed to bind--these are docking false negatives rescued by rescoring. The 10 remaining molecules are true negatives by docking and false positives by MM-GBSA. X-ray crystal structures were determined for 21 of these 23 molecules. In many cases, the geometry prediction by MM-GBSA improved the initial docking pose and more closely resembled the crystallographic result; yet in several cases, the rescored geometry failed to capture large conformational changes in the protein. Intriguingly, rescoring not only rescued docking false positives, but also introduced several new false positives into the top-ranking molecules. We consider the origins of the successes and failures in MM-GBSA rescoring in these model cavity sites and the prospects for rescoring in biologically relevant targets.

Fig. 1

The model cavity sites. (a) Cavity binding site in T4 lysozyme L99A with benzene bound. (b) Cavity binding site in T4 lysozyme L99A/M102Q with phenol bound; the hydrogen bond with the O^ε2 oxygen of Gln102 is represented by a dashed line. (c) Cavity binding site of cytochrome c peroxidase W191G with aniline bound; the hydrogen bond with Asp235 is represented by a dashed line. The heme and an ordered water molecule are also depicted. In (a), (b), and (c), the cavities are represented by a tan molecular surface and the protein ribbons are green. Rendered with the program PyMOL.

Fig. 2

Retrospective enrichment of ligands and decoys for (a) the hydrophobic L99A cavity, (b) the polar L99A/ M102Q cavity, and (c) the anionic W191G cavity. The plots depict the percentage of known ligands (continuous lines) or decoys (dashed lines) found (y-axis) at each percentage level of the ranked database using the top 10,000 best scoring docking hits (x-axis) for L99A (a) and L99A/ M102Q (b) and the 5400 best scoring docking hits (x-axis) for CCP (c). Docking enrichment of known ligands (continuous lines) and decoys (dashed lines) are represented by the dark blue curves. PLOP enrichment of known ligands (continuous lines) and decoys (dashed lines) are represented by the pink curves. AMBERDOCK enrichment of known ligands (continuous lines) and decoys (dashed lines) are represented by green curves.

Fig. 3

Predicted and experimental ligand orientations for the hydrophobic L99A cavity. The carbon atoms of the crystallographic pose, the DOCK predicted pose, the AMBERDOCK predicted pose, and the PLOP predicted pose are colored gray, yellow, cyan, and magenta, respectively. The F_o–F_c omit electron density maps (green mesh) are contoured at 2.5–3.0σ (a) β-chlorophenetole (1), (b) 4-(methylthio)nitrobenzene (2), (c) 2,6-difluorobenzylbromide (4), (d) 2-ethoxyphenol (5), and (e) 3-methylbenzylazide (6) bound to L99A. Rendered with the program PyMOL.

Fig. 4

Predicted and experimental ligand orientations for the polar L99A/M102Q cavity site. The carbon atoms of the crystallographic, DOCK, AMBERDOCK, and PLOP predicted poses are colored gray, yellow, cyan, and magenta, respectively. Hydrogen bonds are depicted with dashed lines. The F_o−F_c electron density omit maps (green mesh) are contoured at 2.5–3.0σ. (a) n-Phenylglycinonitrile (10), (b) 2-nitrothiophene (11), (c) 2-(n-propylthio)ethanol (12), (d) 3-methylbenzylazide (6), (e) 2-phenoxyethanol (9), and (f) (R)-(+)-3-chloro-1-phenyl-1-propanol (13) bound to L99A/ M102Q. Rendered with the program PyMOL.

Fig. 5

Predicted and experimental ligand orientations for the anionic CCP cavity. The carbon atoms of the crystallographic, DOCK, AMBERDOCK, and PLOP predicted poses are colored gray, green, cyan, and orange, respectively. (a) n-Methylbenzylamine (18), (b) cyclopentane carboximidamide (19), (c) (1-methyl-1H–pyrrol-2-yl)-methylamine (20), (d) 1,2-dimethyl-1H–pyridin-5-amine (22), (e) pyrimidine-2,4,6-triamine (24), (f) 1-methyl-5-imidazolecarboxaldehyde (26), (g) 3-methoxypyridine (27), (h) 2-imino-4-methylpiperdine (28), (i) 2,4,5-trimethyl-3-oxazoline (29), and (j) 1-methyl-2-vinylpyridinium (30). Rendered with the program PyMOL.

Fig. 6

The topologically similar ligands and decoys of (a) 2-phenylpropanol and 2-phenoxyethanol (9) for L99A and (b) n-phenylhydroxylamine and o-benzylhydroxylamine (14) to L99A/M102Q.