Active Learning Guided Hit Optimization for the Leucine-Rich Repeat Kinase 2 WDR Domain Based on In Silico Ligand Binding Affinities

Olexandr Isayev,Filipp Gusev,Evgeny Gutkin,Francesco Gentile,Fuqiang Ban,S. Benjamin Koby,Fengling Li,Irene Chau,Suzanne Ackloo,Cheryl H. Arrowsmith,Albina Bolotokova,Pegah Ghiabi,Elisa Gibson,Levon Halabelian,Scott Houliston,Rachel J. Harding,Ashley Hutchinson,Peter Loppnau,Sumera Perveen,Almagul Seitova,Hong Zeng,Matthieu Schapira,Artem Cherkasov,Maria G. Kurnikova

DOI: https://doi.org/10.26434/chemrxiv-2024-jv0rx

2024-10-04

Abstract:The leucine-rich repeat kinase 2 (LRRK2) is the most mutated gene in familial Parkinson’s disease, whose mutations lead to pathogenic hallmarks of the disease. The LRRK2 WDR domain is an understudied drug target for Parkinson’s disease with no known inhibitors prior to the first phase of the Critical Assessment of Computational Hit-Finding Experiments (CACHE) Challenge. CACHE challenges are designed to attract state-of-the-art computational methods for both hit-finding and lead optimization of small molecule inhibitors to challenging protein targets. A unique advantage of the CACHE challenge is that the predicted molecules are experimentally validated in-house. Here we report on our winning submission of experimentally confirmed LRRK2 WDR inhibitor molecules, predicted from thermodynamics integration (TI) calculations performed on only 672 compounds within a chemical space of 25,171 molecules. We used a free energy molecular dynamics (MD) -based active learning (AL) workflow to optimize our two previously confirmed hit molecules. We identified 8 experimentally verified novel inhibitors out of 35 tested (23% hit rate) with a maximum affinity increase of almost 35-fold. These results demonstrate the efficacy of free energy-based active learning workflow to quickly and efficiently explore large chemical spaces while minimizing the number and length of computational simulations. This workflow is widely applicable to the screening of any chemical space for small molecule analogs with increased affinity, subject to the general constraints of RBFE calculations. The mean absolute error of TI MD calculations was 1.30 kcal/mol with respect to measured KD of hit compounds.

Chemistry

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to discover small - molecule inhibitors for the WDR domain of Leucine - rich repeat kinase 2 (LRRK2). The LRRK2 gene is the most frequently mutated gene in familial Parkinson's disease, and its mutations can lead to the pathological features of the disease. Although there have been research reports on inhibitors and PROTACs of the LRRK2 kinase domain, prior to this study, there were no small - molecule inhibitors for the LRRK2 WDR domain. Therefore, this study aims to screen out small - molecule inhibitors that can bind to the LRRK2 WDR domain from a large number of commercially available compounds through computational methods and machine - learning techniques, and optimize the binding affinities of these inhibitors to increase their potential as drug candidates. Specifically, the researchers utilized a workflow based on Active Learning (AL), combined with Relative Binding Free Energy (RBFE) calculations, to optimize two previously identified LRRK2 WDR domain - binding molecules (Hit 1 and Hit 2). Through this method, they screened out approximately 25,000 analogs from a library containing approximately 550 million commercially available compounds, and finally performed RBFE calculations on 672 analogs. Based on the calculation results, 75 molecules were selected for experimental verification, among which 35 molecules were further tested, and finally 8 new inhibitors were identified, with the binding affinity of one molecule being nearly 35 times higher than that of the initial Hit. This study not only demonstrates how to efficiently explore large - scale chemical spaces using computational methods, but also provides an important foundation for drug development with the LRRK2 WDR domain as a therapeutic target for Parkinson's disease.

Active Learning Guided Hit Optimization for the Leucine-Rich Repeat Kinase 2 WDR Domain Based on In Silico Ligand Binding Affinities

In silico screening of LRRK2 WDR domain inhibitors using deep docking and free energy simulations

CACHE Challenge #1: targeting the WDR domain of LRRK2, a Parkinson's Disease associated protein.

Multi-method computational evaluation of the inhibitors against leucine-rich repeat kinase 2 G2019S mutant for Parkinson's disease

Reaction-Based Enumeration, Active Learning, and Free Energy Calculations To Rapidly Explore Synthetically Tractable Chemical Space and Optimize Potency of Cyclin-Dependent Kinase 2 Inhibitors

Active Learning Guided Drug Design Lead Optimization Based on Relative Binding Free Energy Modeling

Discovery of potent LRRK2 inhibitors by ensemble virtual screening strategy and bioactivity evaluation

Crowdsourced mapping of unexplored target space of kinase inhibitors

Subpocket Similarity-Based Hit Identification for Challenging Targets: Application to the WDR Domain of LRRK2

CACHE (Critical Assessment of Computational Hit-finding Experiments): A public-private partnership benchmarking initiative to enable the development of computational methods for hit-finding

Discovery and Optimization of N-Heteroaryl Indazole LRRK2 Inhibitors

Concurrent Optimizations of Efficacy and Blood-Brain Barrier Permeability in New Macrocyclic LRRK2 Inhibitors for Potential Parkinson's Disease Therapeutics

The Fourth CACHE Challenge – A Comprehensive Drug Discovery Workflow to Discover Potential Inhibitors of the Tyrosine Kinase Binding Domain of Cbl-b

Leveraging multiple data types for improved compound-kinase bioactivity prediction

Identification of inhibitors for neurodegenerative diseases targeting dual leucine zipper kinase through virtual screening and molecular dynamics simulations

Protein domain-based prediction of drug/compound-target interactions and experimental validation on LIM kinases

The First CACHE Challenge – Testing Diverse Virtual Screening Scoring Methods to Identify Potential LRRK2 Binders

Active learning driven prioritisation of compounds from on-demand libraries targeting the SARS-CoV-2 main protease

Leveraging Machine Learning and AlphaFold2 Steering to Discover State-Specific Inhibitors Across the Kinome

Efficient Exploration of Chemical Space with Docking and Deep Learning

Active learning approaches in molecule pKi prediction