Multiplex mapping of protein-protein interaction interfaces

We describe peptide mapping through Sp lit A ntibiotic R esistance C omplementation (SpARC-map), a method to identify the probable interface between two interacting proteins. Our method is based on in vivo affinity selection inside a bacterial host, and uses high throughput DNA sequencing results to infer the location of protein-protein interaction (PPI) interfaces. SpARC-map uses only routine microbiology techniques, with no reliance on specialized instrumentation or reconstituting protein complexes in vitro; it can be tuned to detect PPIs over a broad range of affinities; it can be multiplexed to probe multiple PPIs in parallel; its nonspecific background can be precisely measured, enabling the sensitive detection of weak PPIs. Using SpARC-map, we recover the known interface in the (p21-PCNA complex. We also use SpARC-map to probe the purinosome, the weakly bound complex of six purine biosynthetic enzymes, where no PPI interfaces are known. There, we identify interfaces that satisfy structural requirements for substrate channeling; we also identify protein surfaces that participate in multiple distinct interactions, which we validate using site-specific photocrosslinking in live human cells. Finally, we show that SpARC-map results can impose stringent constraints on outputs from machine learning based structure prediction.