RNAs interact with networks of proteins to form complexes (RNPs) that govern many biological processes, but these networks are currently impossible to examine in a comprehensive way.We developed a live-cell chemical probing strategy for mapping protein interaction networks in any RNA with single-nucleotide resolution. This RNP-MaP strategy (RNP network analysis by mutational profiling) simultaneously detects binding by and cooperative interactions involving multiple proteins with single RNA molecules. RNP-MaP revealed that two structurally related, but sequence-divergent noncoding RNAs, RNase P and RMRP, share nearly identical RNP networks and, further, that protein interaction network hubs identify function-critical sites in these RNAs. RNP-MaP identified numerous protein interaction networks within the XIST long noncoding RNA that are conserved between mouse and human RNAs and distinguished communities of proteins that network together on XIST. RNP-MaP data show that the Xist E region is densely networked by protein interactions and that PTBP1, MATR3, and TIA1 proteins each interface with the XIST E region via two distinct interaction modes; and we find that the XIST E region is sufficient to mediate RNA foci formation in cells. RNP-MaP will enable discovery and mechanistic analysis of protein interaction networks across any RNA in cells.
RESULTS
RNP-MaP ValidationTo comprehensively map protein interaction networks of an RNA of interest, we identified a cellpermeable reagent, NHS-diazirine (SDA), that can rapidly label RNA nucleotides at sites of protein binding. SDA has two reactive moieties: a succinimidyl ester and a diazirine ( Fig. 1a).Succinimidyl esters react to form amide bonds with amines such as those found in lysine side chains. When photoactivated with long-wavelength ultraviolet light (UV-A, 365 nm), diazirines form carbene or diazo intermediates 10 , which are broadly reactive to nucleotide sugar and base moieties. The two-step reaction of SDA thus crosslinks protein lysine residues with RNA with a distance governed by SDA linker length (4 Å) and lysine flexibility (6 Å). Lysine is one of the most prevalent amino acids in RNA-binding domains 11 , and photo-intermediate lifetimes are short; thus, SDA is expected to crosslink short-range RNA-protein interactions relatively independently of local RNA structure or protein properties. To perform crosslinking, live cells are treated with SDA for a short time and then exposed to UV-A light.To detect SDA-mediated RNA-protein crosslinks, we use the MaP reverse transcription technology ( Fig. 1b). SDA-treated cells are lysed and crosslinked proteins are digested to short peptide adducts. Adduct-containing RNA is then used as a template for relaxed-fidelity reverse transcription 12,13 . Under MaP conditions, reverse transcriptase reads through adduct-containing nucleotides and incorporates non-templated nucleotides into the product DNA at the site of RNAprotein crosslinks. Importantly, because transcription reads through adducts, RNP-MaP detects multiple pro...