Structural basis for substrate selectivity by site-one protease revealed by studies with a small-molecule inhibitor

Site-one protease (S1P) carries out the first proteolytic step to activate membrane-bound effector proteins in the Golgi. S1P matures through an autocatalytic process that begins in the endoplasmic reticulum (ER) and culminates with the displacement of its inhibitory pro-domain by its cofactor, sterol regulatory element binding protein-regulating gene (SPRING). Spatial control of S1P activity and substrate localization underpins signaling pathways governing, among others, lipogenesis, ER stress, and lysosome biogenesis. The factors governing these pathways are activated by S1P-mediated proteolysis upon their regulated transport from the ER to the Golgi. S1P cleaves substrates with the recognition sequence RX(L/I/V)Z, where X is any residue other than Cys or Pro and Z is preferably Leu or Lys. However, the structural basis for substrate recognition by S1P has remained unknown. Here, we used the small molecule PF-429242, a competitive inhibitor of S1P, to investigate substrate recognition by the S1P/SPRING complex. We determined the structure of S1P/SPRING bound to PF-429242 and found that PF-429242 binds S1P in the same pocket that recognizes the substrate's conserved P4 Arg. Further structural analysis suggests that S1P requires a conformation change to accommodate the substrate's P2 (L/I/V) residue. We designed an S1P mutation (I308A) to reduce the steric clash at the P2 position and generated an S1P that was resistant to PF-429242 in biochemical and cell culture assays. Our findings reveal selectivity in the recognition of substrates by S1P and provide a roadmap for the rational design of improved S1P inhibitors.