Three-dimensional modelling of membrane protein complexes

Membrane proteins make up about 30% of known proteins and over half of current drug targets. Knowledge of their structure is invaluable for the design of new effective drugs. Experimental structure elucidation is difficult and costly, however. Computational methods are routinely employed to predict the three-dimensional structure of a protein of interest.

Membrane proteins live in a very different physico-chemical environment from soluble proteins. They are surrounded by lipid molecules with strongly hydrophobic fatty acid tails and polar or charged head groups. Most existing structure prediction methods do not take these differences into account and tend to perform suboptimally on membrane proteins. Recently, we have developed specialised software to deal specifically with the modelling of an individual membrane protein: iMembrane (to annotate a protein's membrane insertion), MEDELLER (to predict a protein's core structure) and PyFREAD (to model the loops).

The next step from modelling a single chain is to model the entire complex. This will involve the integration of computational methods (see above) with various kinds of experimental data, for example electron density maps from low-resolution X-ray crystallography and electron microscopy, or molecular envelopes from small-angle X-ray scattering experiments. The main goal of the project is to be able to predict the static 3D structure of an entire membrane protein complex and then insert it into a molecular dynamics-type framework to simulate its interaction with other proteins and ligands. Ultimately, this information could then serve to inform higher level models.

Funding: 

This research is funded by the 2020 Science project.