WIDOCK: a reactive docking protocol for virtual screening of covalent inhibitors

Abstract

Here we present WIDOCK, a virtual screening protocol that supports the selection of diverse electrophiles as covalent inhibitors by incorporating ligand reactivity towards cysteine residues into AutoDock4. WIDOCK applies the reactive docking method (Backus et al. in Nature 534:570-574, 2016) and extends it into a virtual screening tool by introducing facile experimental or computational parametrization and a ligand focused evaluation scheme together with a retrospective and prospective validation against various therapeutically relevant targets. Parameters accounting for ligand reactivity are derived from experimental reaction kinetic data or alternatively from computed reaction barriers. The performance of this docking protocol was first evaluated by investigating compound series with diverse warhead chemotypes against KRAS^G12C, MurA and cathepsin B. In addition, WIDOCK was challenged on larger electrophilic libraries screened against OTUB2 and NUDT7. These retrospective analyses showed high sensitivity in retrieving experimental actives, by also leading to superior ROC curves, AUC values and better enrichments than the standard covalent docking tool available in AutoDock4 when compound collections with diverse warheads were investigated. Finally, we applied WIDOCK for the prospective identification of covalent human MAO-A inhibitors acting via a new mechanism by binding to Cys323. The inhibitory activity of several predicted compounds was experimentally confirmed and the labelling of Cys323 was proved by subsequent MS/MS measurements. These findings demonstrate the usefulness of WIDOCK as a warhead-sensitive, covalent virtual screening protocol.

Keywords: Covalent docking; Covalent inhibitors; Virtual screening; Warhead reactivity.

Fig. 1

Known covalent inhibitors of the proteins targeted in this study: ARS-853 for KRAS^G12C, fosfomycin for MurA, E-64 for CatB and clorgyline for MAO-A

Fig. 2

Self-docking of the acrylamide-based KRAS^G12C inhibitor 5 co-crystallized in 5V6S. Crystal structure in cyan; AD4 non-covalent docking pose in pink; pose generated by WIDOCK in orange. a Overall consensus found in the predicted binding modes with respect to the co-crystallized conformation. b Structural differences in the warhead region: I warhead conformation in the crystal structure; II warhead conformation in the standard non-covalent AD4 pose; III warhead conformation in WIDOCK pose

Fig. 3

Evaluation of WIDOCK and covalent docking in AD4 (CovAD4) against KRAS^G12C. a Docking results: colored cells represent experimental and predicted actives (in green and blue, respectively). Experimental actives were considered as those showing > 50% KRAS^G12C labeling at pH 7.5 at 100 μM concentration. Pseudo-Lennard–Jones potentials for WIDOCK were derived using experimental reactivities against BME. b Performance metrics at custom cutoffs. c ROC curves

Fig. 4

Evaluation of WIDOCK and covalent docking in AD4 against MurA and CatB. a Docking results: colored cells represent experimental and predicted actives (in green and blue, respectively). Pseudo-Lennard–Jones potentials were parametrized using experimental reactivities against GSH. b Performance metrics at custom cutoffs. c ROC curves

Fig. 5

Correlation between calculated activation energy barriers (ΔG^‡) and experimental reactivities (lnk)

Fig. 6

Comparison of ROC curves obtained using WIDOCK based on experimental (WIDOCK-exp, in orange) and calculated (WIDOCK-pred, in green) reactivity data against MurA and CatB

Fig. 7

Surface representation showing the best scoring poses obtained by docking compounds 32, 41 and 46 to MurA via different protocols. Standard non-covalent docking pose in pink; covalent docking pose in cyan; docking pose provided by WIDOCK parametrized with experimental reactivities in orange; docking pose provided by WIDOCK parametrized with predicted reactivities in green

Fig. 8

Surface representation showing the best scoring poses generated by docking compounds 44 and 46 against CatB via different protocols. Standard non-covalent docking pose in pink; covalent docking pose in cyan; docking pose provided by WIDOCK parametrized with experimental reactivities in orange; docking pose provided by WIDOCK parametrized with predicted reactivities in green

Fig. 9

Differences of the distances between the reacting ligand atom and the cysteine sulfur as obtained with WIDOCK and with the non-covalent AD4 docking. Average differences plotted against binned remaining activities are shown

Fig. 10

Evaluation of WIDOCK and covalent docking in AD4 against OTUB2 and NUDT7. a Docking poses for the compounds co-crystallized in the structures used for the virtual screening against OTUB2 (5QIV) and NUDT7 (5QHA). Crystal structure in cyan; covalent docking pose in purple; WIDOCK pose in orange. b Performance metrics at custom cutoffs. c ROC curves

Fig. 11

ROC curves for the virtual screening against OTUB2 and NUDT7 for a subset of compounds evaluated by WIDOCK with experimentally (WIDOCK-exp, in orange) and computationally derived (WIDOCK-pred, in green) reactivity parameters

Fig. 12

Active site of MAO-A with the FAD cofactor in purple, covalently bound to Cys406. Residues surrounding the active site Cys323 are shown

Fig. 13

Evaluation of WIDOCK and covalent docking in AD4 against MAO-A. For WIDOCK, results obtained by using both experimental and predicted reactivity parameters are shown. a Docking results: colored cells represent experimental and predicted actives (in green and blue, respectively). b Performance metrics at custom cutoffs. c ROC curves

Fig. 14

Best scoring conformations of compounds 32 and 45 against MAO-A obtained by different docking protocols. Non-covalent docking pose in pink; covalent docking pose in cyan; WIDOCK docking pose obtained by experimental and computed reactivity parameters in orange and green, respectively. Some residues were trimmed to aid the visualization

Fig. 15

Differences of the distances between the reacting ligand atom and the cysteine sulfur as obtained with WIDOCK and with non-covalent AD4 docking. Average differences plotted against binned remaining activities are shown