Using molecular simulation to improve prediction of drug export

Aims

This project aims to develop a predictive computation model for assessing the recognition process and dynamics involved in P-glycoprotein-based drug efflux, providing a new tool to reduce the use of animals typically used for this purpose.

Background

P-glycoprotein (Pgp) is an efflux protein which transports a wide variety of molecules out of the cell back into the blood stream. Pgp has a significant role in drug absorption and disposition and this can have an impact on the potential efficacy and/or toxicity of the drug. 

In the central nervous system, Pgp prevents chemicals crossing the blood-brain barrier - this gatekeeper function can be problematic for the efficacy of drugs whose target resides within the brain.

Animals are used to understand whether and how a drug will interact with Pgp. This use could be minimised with the availability of a predictive computational model.

Research details and methods

The research will examine the interaction of typical CNS-based drugs with lipid bilayers and Pgp to characterise at the atomic level, those that cross the blood-brain barrier and those that do not, plus the structural features of Pgp that are important in substrate recognition and binding susceptibility. X-ray crystal structures and molecular simulation methodologies will be used to generate a predictive computational model.

Predicting how well a drug works (efficacy) and how safe it is (toxicology) are vital components in the drug development process. Currently, these aspects are tested for within animal models (390,000 animal procedures for this aspect alone were performed in 2011 according to the Home Office). Ultimately we'd like to be in a position where predictions are accurate enough to replace the necessity for animal tests, but more realistically a reduction in testing is more achievable. There are many different things to consider when constructing computational models to make predictions for efficacy and toxicology. This proposal aims to examine the role of P-glycoprotein (Pgp), which plays a role in both efficacy and toxicology. Pgp is an efflux protein; it transports a wide variety of compounds out of the cell back into the blood stream. In the brain and central nervous system this is useful because it keeps dangerous chemicals out of the brain. However, it is also a problem for drugs whose target resides within the brain because Pgp will pump these compounds out as well. Thus, the efficacy of some compounds designed to work in the brain is poor because they never reach their target. Being able to more accurately predict whether a compound will be susceptible to Pgp efflux is thus a crucial component to predicting the efficacy of drugs. The flip side of this is that you don't always want compounds to get into the brain. For example if a drug is designed to work elsewhere in the body, it might well result in dangerous side-effects if it penetrated the brain. Thus, Pgp plays a critical role in toxicology as well. Knowledge of what dictates Pgp-drug interactions could enable us to design compounds that deliberately make them more prone to Pgp export thus improving the safety of those compounds. This proposal will make use of recent X-ray crystal structures and molecular simulation methodologies to improve predictions of Pgp susceptibility.

In the first instance, we will examine the interactions of typical CNS-type drugs with lipid bilayers and with Pgp via the following objectives: 1. to characterize at the atomistic level the interaction of a series of compounds that are classified as BBB active (BBB+) or otherwise (BBB-); 2. to elucidate which molecular features may be important in determining CNS activity in terms of bilayer permeation; 3. to examine the influence of protonation states on the free energy of permeation; 4. to investigate which aspects of the structure of Pgp may be important in substrate recognition. How is access to the binding pocket likely to be influenced by initial penetration of the compounds; 5. to use this data to generate more accurate implicit models that capture the essential features of the model but offer massive improvements in speed.

All of the above aims will be studied using state-of-the-art molecular simulation methodologies with which we have an international track record. As well as providing much-needed insight into the recognition process and dynamics involved in Pgp-based drug efflux, the results will help us to develop better empirical rules that can be incorporated into pharmacokinetic predictive tools which in turn should contribute to a reduction in animal procedures.