Mapping Amino Acids at Protein-Membrane Interfaces to Update the Current Membrane Binding Model

Abstract

Peripheral Membrane Proteins (PMP) are essential components of multiple biological processes like lipid metabolism, membrane remodelling and other complex cellular pathways. They bind transiently to the surface of biological membranes through short helices or loops, and their Interfacial Binding Site (IBS) has been described as displaying basic and hydrophobic amino acids. However, recent studies indicate that this model is too simplistic to account for the fine-tuned lipid specificity achieved by many PMPs (Grauffel et al., JACS, 2013; Chon et al., Biochemistry, 2015). Terms like “hydrophobic spikes” or “protruding loops” have been used to describe PMP IBS. Recently we proposed a hydrophobic protrusion model for IBSs (Fuglebakk et al., Plos Comp. Biol., 2018). This model is based on the calculation of the protein convex hull and its vertices. When tested on a dataset of 300 protein families it showed good performance in discriminating membrane-binding proteins from a reference dataset. This shows that it captures a structural pattern common to PMP families. Yet it does not have the resolution to capture fine differences between families, and these differences may be decisive for their membrane specificity. In order to characterize the IBS of the PH, SH2 and C2 domains at a finer resolution we collected a curated dataset of structures and sequences from the CATH database (Sillitoe et al., NAR, 2019). We developed a statistical framework and could map the frequency of charged amino acids and aromatic amino acids at protrusions and around protruding hydrophobes in the three domains. We will present and discuss the differences between the three domains and how they relate to their function.