Talk abstracts

Talk on Saturday 11:20-11:40am submitted by Jeremy Sanford

Mapping the in situ RNA binding specificity of the SR protein SF2/ASF using CLIP-Seq

Jeremy Sanford (Department of Biochemistry and Molecular Biology, Indiana University School of Medicine), Xin Wang (Department of Medicine, Indiana University School of Medicine), Xinguo Wang (Center for Genomics and Bioinformatics, Indiana University), Yunlong Liu (Department of Medicine, Indiana University School of Medicine)

Abstract:
Recognition of cis-acting RNA elements by trans-acting RNA binding proteins is critical to the precise processing of RNA and for coordinated regulation of post-transcriptional gene expression. Traditional methods for mapping sites of RNA-protein interactions rely on in vitro equilibrium binding conditions focusing on single binding events. Here we describe a comprehensive method for mapping the in vivo sites of RNA-protein interactions across the human genome. We focused our initial studies on the shuttling SR protein SF2/ASF. This protein has well defined biochemical roles in post-transcriptional gene expression, however few bona fide endogenous targets have been identified. We employed cross-linking immunoprecipitation (CLIP) to purify covalent SF2/ASF-RNA complexes from HEK293T cells. A major advantage of CLIP is that photo cross-linking of intact living cells is used to capture the in situ specificity of RNA-protein interactions, without the added complications associated with chemical cross-linking (protein-protein cross-links). 45-55 nt RNA fragments corresponding to the in vivo binding sites for SF2/ASF were amplified and directly sequenced by ultra-high-throughput pyrosequencing using the 454 platform. More than 150,000 sequences encompassing 8,166 genomic regions were mapped to the human genome using BLAT. 4,481 genomic regions were represented by multiple amplicons, termed replicons. Approximately 50% of these regions mapped to non-repetitive genomic loci. These regions contain 1463 intragenic and 431 intergenic loci, respectively. We find that SF2/ASF binding sites are strongly biased towards exonic sequences and exon-intron boundaries relative to intronic positions (76% compared with 24%, respectively). We determined a mean PhastCons score for all 8,166 genomic regions identified by pyroCLIP. Genomic regions containing multiple amplicons sequences are highly conserved. Importantly, this correlation is observed for both intra- and intergenic binding sites for SF2/ASF, suggesting that SF2/ASF may be involved in processing of noncoding RNAs or novel protein coding genes. Finally clustering of SF2/ASF mRNA targets based upon primary encoded function reveals that there is a highly significant enrichment (p value = 10^-74) for genes involved in mRNA metabolism. Roles for SF2/ASF in RNA processing will be discussed based upon this novel genome wide perspective.

Keywords: RNA binding protein, genomics, high throughput sequencing