General and efficient computational assays for antibiotic breakdown by *-lactamases.

Rising antibiotic resistance is a major problem for human health. Resistance to *-lactams, the single most important antibiotic class, often arises through *-lactam breakdown by *-lactamase enzymes. *-lactamase producing bacteria are often multi-drug resistant, and may cause untreatable infections. Assessment of *-lactamase activity is thus highly important, but performing biochemical assays for new variants of the quickly evolving *-lactamases with multiple drugs is time consuming and costly. Multi-scale computer simulations can now be used to predict if an enzyme breaks down a *-lactam; as we have demonstrated for carbapenem breakdown by Class A *-lactamases (Chem Comm, 50, 14736). Further (unpublished) work indicates that such assays can a) be made more efficient (e.g. 6-12 hrs on 1 CPU) and b) be applied to *-lactam antibiotics with different chemical structures.
This multidisciplinary project aims to develop efficient in silico assays for hydrolysis of key *-lactams by clinically relevant class A, C & D serine *-lactamases, and test their accuracy by direct comparison to experimentally measured kinetics. This is challenging as reaction mechanisms differ by class and both acylation and deacylation halves of the reaction must be assessed as either can be rate limiting. Simulations using quantum mechanics / molecular mechanics (QM/MM) methods will be used to identify detailed mechanisms and perform in silico assays. Steady-state, stopped- and quenched-flow kinetic methods will be used for experimental determination of rates for the different reaction steps (with co-supervisor Jim Spencer). The *-lactamase test set will be augmented with new enzymes and variants identified from an extensive collection of clinical isolates from Pakistan, Thailand, Vietnam by genomic and phenotypic approaches (via collaborator Prof. Timothy Walsh, Cardiff).
The project will provide training in cutting-edge techniques in multiple disciplines (computational chemistry, molecular biology/biochemistry and clinical microbiology/genomics) using state-of-the-art facilities in the context of a collaborative multi-disciplinary environment (e.g. www.bristol.ac.uk/bristolbridge). The project will make full use of Bristol's excellent high-performance computing resources (e.g. BlueCrystal Phase4, one of the largest University-owned computer clusters).
The resulting first-principle in silico assays will be general (i.e. independent of training sets etc.) and atomically detailed, so that insights will be obtained into mechanisms of resistance through *-lactamases. These insights will inform development of new *-lactam antibiotics and *-lactamase inhibitors, and such new drug leads can be evaluated using the protocols developed. We already collaborate with drug discovery companies active in this area. To further increase the impact of the research in this project, we aim to automate the in silico assays developed and make them available, e.g. through a web-server for use by non-specialists.