Deciphering the continuum of interactions between integral membrane proteins and cholesterol

Abstract

Cholesterol is one of the major lipid components of the plasma membrane in all mammalian cells. It has been shown to modulate membrane fluidity, assembly, organization, and dynamics and alter integral membrane protein (IMP) function. As cholesterol plays multiple roles, it is difficult to distinguish between its ability to specifically bind to IMPs and change their function from its ability to alter membrane properties around them nonspecifically. Current methods to solve this problem are sequenced-based; however, with increasing investigation sequence-based approaches seemed to be insufficient in modeling and predicting IMP-cholesterol interactions. As an alternative, this work presents a structure-based approach, RosettaCholesterol (RC), that combines physico-chemical and evolutionary properties. In order to identify similar physico-chemical features of cholesterol binding sites, the Protein Data Bank was utilized. Structural features of cholesterol binding pockets were clustered across protein families to use them as a reference set for comparing query proteins to the clusters. An evolutionary analysis was conducted to compare the conservation of the residues in cholesterol binding surfaces to their other IMP lipid-accessible counterparts. Since IMPs are continuously immersed in lipid solvent, any conserved cholesterol binding surfaces are expected to be significant. The RC adapts standard RosettaLigand by implementing a new sampling and scoring strategy. This new strategy makes it possible to recognize a ligand as a lipid by accounting for the ligand's orientation and van der Waals interactions within the membrane. The RC protocol allows the prediction of biologically relevant energetically-favorable IMP-cholesterol complexes and the quantification of the specificity of an IMP's cholesterol binding site.