USING AI TO DRIVE DRUG DISCOVERY FOR MEMBRANE PROTEIN TARGETS
Lead supervisor: Prof Phil Biggin, Department of Biochemistry, Oxford
Commercial partner: Vertex Pharmaceuticals, Milton site, Oxfordshire
This project aims to develop tools, which would accelerate the discovery for classically undruggable, inherently flexible membrane protein targets such as the Cystic Fibrosis Transmembrane Regulator (CFTR). The project will form the basis of a brand-new collaboration between Professor Biggin’s group and Vertex Pharmaceuticals.
Cystic fibrosis is caused by any one of several defects in the CFTR protein and affects (or impacts) at least 70,000 people world-wide; in the UK alone, there are 10,000 people in the CF UK registry. There is currently no known cure and median life expectancy is less than 40 years (37 years). Several mutations have been identified where the channel is correctly trafficked but remains functionally compromised and cannot transport chloride. Vertex first developed Ivacaftor, (Kalydeco™) a CFTR potentiator that increases chloride transport. However, the mechanism of action remains elusive and it does not work for all mutations associated with CFTR. The most common mutation is deltaF508, which is treated by two additional Vertex medicines, Orkambi and Symdeko.
Cryo-EM structures of first the zebrafish and now human CFTR in an apo, inactive state have been resolved to 3.9 Å resolution, providing for the first time an opportunity to not only investigate the mechanism by which these medicines work but also to put future drug design work on a more solid, structural footing, specifically for drugs that act at sites within the membrane. Recently (Oct 2018), another cryo-EM (chicken) structure in the activated form was solved to 4.3 Å resolution providing the opportunity to use computational methods to explore the mechanistic differences between the inactive and active forms.
The project builds on the combined expertise of Vertex and Professor Biggin’s laboratory. This computational project will develop and apply advanced free energy calculations and machine learning techniques. Both aspects will be key to understanding how CFTR, and indeed other membrane proteins, can be targeted in a rational, structural fashion. In addition, the substantial experience of Vertex in terms of developing small molecules for such targets, alongside assay and synthesis expertise, will allow us to validate the computational predictions performed in this project in a prospective manner.
The overall aims of the project are:
- To develop and validate new methodologies for membrane protein-drug interactions, building on the previous work in the Biggin group, with particular focus on free energy calculations and rapid pose prediction using deep-learning.
- To assess the dynamic behaviour of target proteins, such as, but not limited to, CFTR, and their interaction with small molecules in bilayer systems that mimic the in vivo composition.
This project combines several related disciplines. No previous computational experience is necessary, but an up to date awareness of machine-learning and/or molecular simulation would be advantageous. A good background in chemistry would also be useful.