Kokh et al. 2019

Authors: Kokh DB¹, Kaufmann T^1,2, Kister B^1,3, Wade RC^1,3,4,5.

Author information: ¹ Molecular and Cellular Modeling Group, Heidelberg Institute for Theoretical Studies (HITS), Heidelberg, Germany. ² Department of Physics, Heidelberg University, Heidelberg, Germany. ³ Department of Biosciences, Heidelberg University, Heidelberg, Germany. ⁴ Zentrum für Molekulare Biologie der Universität Heidelberg, DKFZ-ZMBH Alliance, Heidelberg University, Heidelberg, Germany. ⁵ Interdisciplinary Center for Scientific Computing (IWR), Heidelberg University, Heidelberg, Germany.

Corresponding author: Rebecca C. Wade(Rebecca.Wade@h-its.org) Daria B. Kokh(Daria.Kokh@h-its.org)

Journal: Frontiers in Molecular Bioscience

Download Url: https://doi.org/10.3389/fmolb.2019.00036

Citation: Kokh DB, Kaufmann T, Kister B, Wade RC(2019) Machine Learning Analysis of τRAMD Trajectories to Decipher Molecular Determinants of Drug-Target Residence Times Front. Mol. Biosci. (2019) .

DOI: https://doi.org/10.3389/fmolb.2019.00036

Licence: the Creative Commons Attribution (CC BY) license applies for all files. Under this Open Access license anyone may copy, distribute, or reuse the files as long as the authors and the original source are properly cited.

Abstract:

Drug-target residence times can impact drug efficacy and safety, and are therefore increasingly being considered during lead optimization. For this purpose, computational methods to predict residence times, τ, for drug-like compounds and to derive structure-kinetic relationships are desirable. A challenge for approaches based on molecular dynamics (MD) simulation is the fact that drug residence times are typically orders of magnitude longer than computationally feasible simulation times. Therefore, enhanced sampling methods are required. We recently reported one such approach: the τRAMD procedure for estimating relative residence times by performing a large number of random acceleration MD (RAMD) simulations in which ligand dissociation occurs in times of about a nanosecond due to the application of an additional randomly oriented force to the ligand. The length of the RAMD simulations is used to deduce τ. The RAMD simulations also provide information on ligand egress pathways and dissociation mechanisms. Here, we describe a machine learning approach to systematically analyze protein-ligand binding contacts in the RAMD trajectories in order to derive regression models for estimating τ and to decipher the molecular features leading to longer τ values. We demonstrate that the regression models built on the protein-ligand interaction fingerprints of the dissociation trajectories result in robust estimates of τ for a set of 94 drug-like inhibitors of heat shock protein 90 (HSP90), even for the compounds for which the length of the RAMD trajectories does not provide a good estimation of τ. Thus, we find that machine learning helps to overcome inaccuracies in the modeling of protein-ligand complexes due to incomplete sampling or force field deficiencies. Moreover, the approach facilitates the identification of features important for residence time. In particular, we observed that interactions of the ligand with the sidechain of F138, which is located on the border between the ATP binding pocket and a hydrophobic transient sub-pocket, play a key role in slowing compound dissociation. We expect that the combination of the τRAMD simulation procedure with machine learning analysis will be generally applicable as an aid to target-based lead optimization.

Resources

Here you can find the data used and generated in this work, along with the visualization of the main results produced.

Structural data

Representative 3D experimental structural data:

Complex of HSP90 with a resorcinol-type compound bound to loop-type conformation (shown in Fig.1A)	PDB 5J2X	Download	View
Complex of HSP90 with a recorcinol-type compound bound to the helix-type conformation of the binding site (shown in Fig.1B)	PDB 5J9X	Download	View
Complex of HSP90 with a tricyclic compound bound to helix-type conformation of the binding site (shown in Fig.1D)	PDB 2YKI	Download	View

1D/2D structure data of molecular compound used in the study:

1D structures (SMILES format):	Download
2D structures (sdf format):	Download

Experimental data

Kinetic rates of protein-ligand binding:

Experimental Kinetics data

Download

References:

[1] Amaral, M., Kokh, D. B., Bomke, J., Wegener, A., Buchstaller, H. P., Eggenweiler, H. M., et al. (2017). Protein conformational flexibility modulates kinetics and thermodynamics of drug binding. Nat. Commun. 8, 2276.

[2] Kokh, D., Amaral, M., Bomke, J., Graedler, U., Musil, D., Buchstaller, H., et al. (2018). Estimation of Drug-Target Residence Times by τ- Random Acceleration Molecular Dynamics Simulations. J. Chem. Theory Comput 14, 3859–3869.

[3] Schuetz, D. A., Richter, L., Grandits, M., Graedler, U., Buchstaller, H., Eggenweiler, H., et al. (2018b). Ligand Desolvation steers on-rate and impacts Drug Residence Time of Heat shock protein 90 ( Hsp90 ) Inhibitors. J. Med. Chem. 90, 4397–4411.

Models

Exploring clustering of dissociation trajectories by the overall protein-ligand contact similarity

All compounds (Fig. 5)	Claster composition	Helix-binding compounds (PDB 5J9X)	View	Loop-binding compounds (PDB 5J2X)	View
Indazole-type compounds (Fig. 6)	Claster composition	Indazole-type compounds (PDB 5LNZ)	View