Jan Brezovsky

Laboratory of Biomolecular Interactions and Transport, Adam Mickiewicz University, Poznań, Poland - Research

Cracking the code of enzyme tunnels: understanding, predicting, and engineering hidden pathways

Abstract

Molecular transport through protein tunnels governs ligand access to buried active sites [1], a process critical for understanding binding kinetics and optimizing drug residence times [2,3]. Yet capturing the full complexity of tunnel networks requires extensive sampling that generates massive trajectory datasets [4, 5], analyses of which represent a methodological challenge. To overcome this, we developed TransportTools library [6,7] capable of systematic characterization of transient tunnels and explicit tracking of molecular migrations across thousands of parallel simulations. Building on this, we developed knowledge-based seeding strategies that incorporate prior structural information about tunnel geometries into adaptive sampling workflows [5]. This approach dramatically improves sampling consistency for complete (un)binding pathways, enabling reliable estimation of koff/kon ratios that match experimental measurements even for complex systems with multiple relevant, functional tunnels. Applying this framework at scale, we generated and analyzed 1.6 μs adaptive simulations of 40 diverse enzymes (five EC and four structural classes). This dataset of tunnel dynamics covers over 450 distinct transient tunnels, detailing almost 2.5 million water transport events via more than 90 million tunnel conformations. Finally, such approaches can reveal molecular principles behind the effect of mutations in ABCG transporters [8] and drive protein engineering of N-terminal hydrolases by re-designing molecular gates controlling access to their binding site [9, 10].

References

[1] Methods Mol. Biol. 2018, 1685:25-42. [2] Med. Res. Rev. 2017, 37: 1095-1139 [3] Chem. Rev. 2013, 113: 5871–5923 [4] J. Chem. Theory Comput 2026, 22: 135-150. [5] J. Chem. Theory Comput 2024, 20: 5807-5819. [6] Bioinformatics 2022, 38: 1752-1753. [7] MethodsX 2023, 10C: 10196. [8] Cell. Mol. Life Sci 2023, 80: 105. [9] ACS Catal. 2022, 12: 6359-6374. [10] BioRxiv 2023, DOI:10.1101/2023.05.09.538545