PEPTIDE PARTITIONING · Experimental measurement of the direct partitioning of peptides into lipid bilayers

The transfer of peptide segments into the lipid bilayer to form stable transmembrane helices is the crucial first step in membrane protein folding and assembly. However, the mechanisms that drive this process are not fully understood. A recent experimental assay has measured the insertion of designed peptide sequences into the endoplasmic reticulum membrane via the cellular translocon machinery. This has provided the first quantitative estimate of the translocon-to-membrane transfer free energy of polypeptide segments. However, no suitable experimental setup currently exists that allows the direct measurement of the free energy change involved in transferring peptides from water into lipid bilayers. This is because peptides that are hydrophobic enough to insert into membranes generally aggregate in solution. I propose an experimental peptide setup that directly measures the water to bilayer partitioning of a series of designed peptides that do not have this problem. These will be derived by minor re-engineering of pHLIP, a peptide based on helix C of bacteriorhodopsin. This peptide has a set of unique properties that make it ideal for the proposed study. It is soluble at neutral pH and spontaneously inserts into lipid bilayers when the pH is lowered. Remarkably, the peptide partitions as a monomer, without disrupting the bilayer, and has been shown not to aggregate either in solution or inside the bilayer. For theoreticians the proposed direct measurements are highly desirable, since they allow the calibration of current computer simulation models under identical conditions. Recent advances in algorithms and computer hardware have enabled fully converged simulations of the adsorption, folding and insertion of peptides into lipid bilayers. These studies have exposed striking differences between the various different methods used as well as currently established theory, which cannot be resolved without a suitable experimental dataset like the one proposed here.