Predictive simulations of molecules in membranes

Knowledge of how a molecule interacts with biological membranes is essential in evaluating its safety and suitability for inclusion in the formulation of household products, such as those developed by the project's industrial partner, Unilever. The membrane-water partitioning coefficient (the logarithm of the equilibrium constant for division of a given solute between the membrane and aqueous solution) is used within risk assessments for estimating hydrophobic properties of molecules. However, accurate laboratory measurement of membrane-solute interactions is challenging and time-consuming. Moreover, membranes differ in composition (e.g. the mixture of phospholipids and the cholesterol content), which affects the interactions with the solute and multiplies the task of characterisation. Predictive computer simulations are an attractive alternative to experimental measurement. Simulations that represent all atoms explicitly are too slow for such complex systems, but coarse-grained (CG) models, which work at a judiciously chosen lower level of detail, can bridge the gap. This project will harness and further develop recent advances in CG modelling to produce scalable, predictive simulations of membrane-solute interactions. We will (1) exploit the increased accuracy and chemical space captured by next-generation CG force fields to build predictive models for membrane-solute partitioning, (2) systematically characterise the effect of membrane composition on membrane-solute interactions, (3) predict how membranes mediate interactions between species embedded with them. We will engage with regulatory bodies to help shape the priorities of the project and to ensure that developments from the research are exploited as widely as possible.