We have shown that molecular simulations can predict the ability of enzymes to break down different antibiotics, distinguishing enzymes that destroy carbapenems rapidly (and so cause bacterial resistance to these ‘last resort’ antibiotics) from enzymes that do not. By analysing the chemical reactions of carbapenem breakdown in different bacterial enzymes, we showed that a fundamental physical property, the electric field, is a crucial factor in determining how fast the breakdown reaction happens. This insight will help in predicting and combating bacterial antibiotic resistance, e.g. in the development of new antibiotics and of inhibitors to block the enzymes that break down antibiotics. We have made these tools freely available. Combining simulations with experiments, we also showed what factors are responsible for the antibiotic breakdown ability of an enzyme that is a globally significant cause of antibiotic resistance.
Another focus of PREDACTED is the temperature dependence of enzyme-catalysed reactions. We have shown (combining simulations with experiments) that the factors that make enzymes different from simple chemical catalysts (in particular, the non-linearity of their temperature dependence) are complex, but can be understood and modelled. Our models explain previously puzzling experimental observations. They provide a new conceptual framework for understanding how enzymes evolve to function at different temperatures. This is fundamentally important in how organisms adapt to ecological niches, and how they may respond to climate change. Our framework also provides a route to engineering enzymes for activity at low temperatures, as biocatalysts for low temperature, less energy-intensive processes for biotechnology and industry.
We have also applied the simulation tools we are developing (including non-equilibrium molecular dynamics simulations) to proteins from the SARS-CoV-2 virus. With experimental collaborators, and as part of an international consortium, we investigated the virus spike protein and identified new features relating to function and infectivity. For the SARS-CoV-2 main protease, we found features relevant to drug resistance. We apply high performance and cloud computing. We have developed new approaches to combine simulation and interactive virtual reality to investigate binding to proteins. We are applying all of these methods to understand, predict and design enzyme activity and are testing these predictions experimentally.
