Asymmetric membrane properties through a protein lens†

Abstract

Plasma membranes are asymmetric, with each monolayer presenting specific lipid compositions and biophysical properties. Transmembrane domains (TMDs) of single-pass transmembrane proteins (spTMPs) have adapted their physico-chemical properties to these asymmetric constraints. In this study, we analysed the structural features of such TMDs across the tree of life to obtain information about their interaction with asymmetric membrane bilayers and predict species-specific membrane properties. We observed that TMDs in the plasma membranes of all eukaryotic species possess asymmetries in lipid accessible surface area (ASA), hydrophobicity, aromaticity and charge. Bacteria deviate from this trend, with strong differences between bacterial clades. Notably, TMDs in the Golgi and the endoplasmic reticulum of eukaryotic species display inverted profiles for accessible surface area, hydrophobicity and aromaticity compared to their plasma-membrane counterparts. To determine how well TMD profiles reflect average membrane properties, we performed molecular dynamics simulations of a spTMP in an asymmetric lipid bilayer whose composition approximates the human plasma membrane. The simulated spTMP was chosen to represent the average TMD properties of the human proteome. We compared the electron density profiles of the simulated asymmetric membrane to the average TMD profiles derived from the human proteome and observed that phospholipid acyl-chain density overlapped very well with TMD hydrophobicity, and phosphate group density with TMD charge. The profiles of phospholipid unsaturation in the acyl chains overlapped well with the average location of TMD phenylalanines in the cytoplasmic leaflet, while there was additional accumulation of large hydrophobic and aromatic residues in the membrane midplane, which had low acyl-chain density. This study reveals the complementarity of membrane and TMD properties in asymmetric membranes, suggesting that the properties of TMDs can be used to make predictions about the properties of their solvating membranes.