Combined free energy calculation and machine learning methods for understanding ligand unbinding kinetics

Abstract

The determination of drug residence times, which define the time an inhibitor is in complex with its target, is a fundamental part of the drug discovery process. Synthesis and experimental pharmacokinetics measurements are, however, expensive, and time-consuming. In this work, we aimed to obtain drug residence times computationally. Furthermore, we propose a novel algorithm to identify molecular design objectives based on ligand unbinding kinetics. We designed an enhanced sampling technique to accurately predict the free energy profiles of the ligand unbinding process, focusing on the free energy barrier for unbinding. Our method first identifies unbinding paths determining a corresponding set of internal coordinates (IC) that form contacts between the protein and the ligand, then iteratively updates these interactions during a series of biased molecular-dynamics (MD) simulations to reveal the ICs important for the whole of the unbinding process. Subsequently, we performed finite temperature string simulations to obtain the free energy barrier for unbinding using the set of ICs as a complex reaction coordinate. Importantly, we also aimed to enable further design of drugs focusing on improved residence times. To this end, we developed a supervised machine learning (ML) approach that uses as input unbiased 'downhill' trajectories from the transition state (TS) ensemble of the string unbinding path. We demonstrate that our ML method can identify key ligand-protein interactions driving the system through the TS. Some of the most important drugs for cancer treatment are kinase inhibitors. One of these kinase targets is Cyclin Dependent Kinase 2 (CDK2), an appealing target for anticancer drug development. Here, we tested our method using three different CDK2 inhibitors for potential further development of these compounds. We compared the free energy barriers obtained from our calculations with those observed in available experimental data. We highlighted important interactions at the distal ends of the ligands that can be targeted for improved residence times. Our method provides a new tool to determine unbinding rates, and to identify key structural features of the inhibitors that can be used as starting points for novel design strategies in drug discovery.

Dissociation Rate Calculation via Constant Force Molecular Simulation

“…The calculation is considerably tricky, and few studies have succeeded. It has been done only for weak binders to proteins and proteinprotein association by a number of brutally long-time MD simulations 4,59 or by indirect calculation from binding free energy and 𝑘 +,, value 60 . One exception is recent works by Wolf S. et al: They proposed the novel idea that dissipated-corrected TMD simulation is combined with onedimensional Langevin dynamics.…”

Dissociation Rate Calculation via Constant-force Steered Molecular Dynamics Simulation

“…The calculation is considerably difficult, and only a few studies have been successful. It has been performed only for weak binders to proteins and protein-protein association by brutally long-time MD simulations 4,60 or by indirect calculation from the binding free energy and 𝑘 +,, value 61 . One exception is the recent work by Wolf et al, who proposed the novel idea that dissipated-corrected targeted molecular dynamics (TMD) simulation is combined with one-dimensional Langevin dynamics.…”

Investigating the Unbinding of Muscarinic Antagonists from the Muscarinic 3 Receptor

“…We used the Gini feature importance 58 to evaluate the relevance of the features from the GBDT models, averaged across the 100 trainings to calculate their relative feature importance (RFI). To identify key features in MLP models, we removed the variance from each feature one-by-one 29 and assessed the accuracy drop they encounter when predicting outcomes with the trained models. If the accuracy of the prediction is greatly reduced when a feature is altered, the feature was considered important for the description of the TS.…”

“…The unbinding procedure was followed as described in our previously published 29 protocol. After the equilibration, a 20 ns production run without any restraints was performed.…”

“…In this work, we applied our recently developed unbinding algorithm 29 to hMR3, to investigate the dissociation of tiotropium ( 1 ) and two structurally similar ligands, N-methylscopolamin ( 2 ) and homatropine methylbromide ( 3 ) (Figure 1). The obtained unbinding pathways were refined using an adaptation of the finite temperature string method.…”