Footprinting SHAPE-eCLIP reveals transcriptome-wide hydrogen bonds at RNA-protein interfaces

Discovering the interaction mechanism and location of RNA binding proteins (RBPs) on RNA is critical for understanding gene expression regulation. Here, we apply selective 2'-hydroxyl acylation analyzed by primer extension (SHAPE) on in vivo transcripts to identify transcriptome-wide footprints (fSHAPE) on RNA when compared to protein-absent transcripts. Structural analyses indicate that fSHAPE precisely detects nucleotides whose base moieties hydrogen bond with protein and that fSHAPE patterns can predict binding sites of RBPs of interest. To illustrate, we demonstrate that fSHAPE correctly identifies known and novel iron response protein RNA elements. Furthermore, we enable selective interrogation of RNA-protein complexes by integrating SHAPE and fSHAPE with crosslinking and immunoprecipitation (eCLIP) of desired RBPs. To demonstrate, histone stem loop elements and their nucleotides that hydrogen bond with stem-loop binding protein were identified by SHAPE-eCLIP and fSHAPE-eCLIP. Together, these technologies greatly expand on strategies for understanding specific cellular RNA interactions in RNA-protein complexes. Overall design: Footprinting and SHAPE structure probing of human cells was performed in conjunction with enhanced crosslinking and immunoprecipitation in order to interrogate cellular RNA-protein complexes in a targeted manner. fSHAPE was performed on multiple human cell lines and identifies hydrogen bonds that occur between RNA base moieties and interacting protein residues. fSHAPE-eCLIP was performed on K562 cells with immunoprecipitation for stem loop binding protein (SLBP) and both identifies SLBP transcript binding sites as well as bases that hydrogen bond with protein. SHAPE-eCLIP was performed on K562 cells with immunoprecipitation for SLBP and identifies SLBP transcript binding sites as well as returning RNA structure probing information in these sites. in vivo click (ic)SHAPE structure probing, performed in parallel with fSHAPE, produces base pairing probailities (i.e. "reactivities") at each nucleotide across transcripts. Please note that [1] each processed data was generated from multiple samples/replicates as indicated in the corresponding sample description field. [2] the 293T_vitro_icSHAPE.tar.gz files were generated from re-analyzing the samples GSM2075714-GSM2075717. The 293T_vivo_icSHAPE.tar.gz files were generated from re-analyzing the samples GSM2075714-GSM2075715, GSM2075718-GSM2075719 (in GSE74353; third-party re-analysis). [3] the 293tmaps.tar.gz files were generated from re-analyzing the samples GSM2075716-GSM2075719 (in GSE74353; third-party re-analysis).