Structural basis of rhomboid intramembrane protease specificity and mechanism revealed by X-ray crystallographic analysis of rhomboid-substrate peptide complexes

Sebastian Zoll, Stancho Stanchev, Jakub Began, Jan Skerle, Lucie Peclinovská, Martin Lepšík, Pavel Majer and Kvido Stříšovský

Institute of Organic Chemistry and Biochemistry, Academy of Sciences of the Czech Republic, Flemingovo n. 2, Prague, 166 10, Czech Republic

Intramembrane proteases regulate a growing number of biological processes emanating from biological membranes. The structures of bacterial homologues of all four catalytic types of these integral membrane proteins have been solved, but their mechanism is incompletely understood, mainly because of the difficulty of gaining structural information about how they bind substrates. To gain insight into substrate binding to rhomboid proteases, we have synthesised a series of covalently binding peptidyl chloromethylketone inhibitors and analysed their interactions with E.coli rhomboid GlpG enzymologically and structurally. We show that tetrapeptidyl chloromethylketones derived from a natural rhomboid substrate bind GlpG covalently in a mechanism-based and sequence-dependent, substrate-like manner. We have solved the crystal structures of several tetrapeptidyl chloromethylketones, which reveal, for the first time, the molecular interactions of a substrate peptide with an intramembrane protease. The overall binding mode differs from previous speculative models. The structures explain the observed amino acid preferences in substrates and show the S1 to S4 subsites of rhomboid active site. The S1 site is well formed and extends into a water filled cavity (“water retention site”) previously proposed to be important for acyl-enzyme hydrolysis. Unexpectedly, the S4 subsite is formed by residues from the peripheral structure of the L1 loop. Mutational and crystallographic analysis of the S4 subsite shows that it is plastically formed by three residues of the L1 loop. Given that the L1 loop region has diversified and expanded in rhomboid-like proteins including iRhoms, we propose that this element of the rhomboid fold plays a role in substrate or client protein binding in the broader rhomboid-like superfamily. Finally, using molecular dynamics and the published structural and enzymatic data we propose a model of the P4 to P3’ region of the substrate that is in contact with the active site of GlpG. The mode of binding of the rest of the transmembrane domain of the substrate is less clear and we propose two possible scenarios.