High-throughput studies of the RNA-binding protein HuR targetome

Background

Human antigen R (HuR), preferentially associates with AU-rich elements (AREs) within 3'-UTRs of mammalian mRNAs regulating their steady-state levels. HuR-regulated mRNAs encode proteins involved in multiple molecular pathways including those strongly related to progression of cardiovascular diseases (CVDs). RNA immunoprecipitation (RIP) and individual nucleotide resolution cross linking and immunoprecipitation (iCLIP) are being harnessed to characterize RNA binding sites of RBPs, including those of HuR protein in multiple previous reports. However, their sensitivity can be limited by standard RNA sequencing methods which rely on either 3’-adapter ligation or long RNA-seq. The main goal of this project was to investigate HuR-associated RNAs in human umbilical vein endothelial cells (HUVECs) using RIP and iCLIP methods coupled with ultrasensitive semi ligation-based RNA-seq library preparation method. In addition, we developed a novel approach “direct CLIP (DiCLIP)” and benchmarked it with standard iCLIP on HUVECs HuR IP samples.

Methods

Both RIP and DiCLIP pellets were preformed from paraformaldehyde fixed HUVECs lysates according to in-house developed workflows using anti-HuR as well as control IgG antibody. Upon total RNA isolation, stranded small RNA-seq libraries were prepared using 3’-polyadenylation and 5’-ligation of RNA adapters and sequenced on Illumina NextSeq 550.

Results

Multiple transcripts have been detected by RNA-seq in RIP pellets when using as little as 0.6-6 ng total RNA inputs. Furthermore, 67% RNA transcripts (8079 out of 12076 detected) were significantly (BH adj. pvalue < 0.05) enriched in anti-HuR vs. IgG RIP pellets. Downstream GO analysis showed that mRNAs strongly associated with HuR (> 3.5 LFC versus IgG background) encode proteins enriched in mTORC1, VEGF and TGF-beta signaling pathways which are related to development of atherosclerosis and other CVDs. Multiple transcripts have been detected by RNA-seq in DiCLIP pellets when using as little as 0.3-0.5 ng total RNA inputs. Furthermore, DiCLIP peaks were in close proximity/overlapping putative HuR binding motifs, while classical iCLIP peaks were 10-12 bases downstream of HuR binding sites. Finally, DiCLIP identified multiple HuR binding sites which were overlooked in classical iCLIP datasets.

Conclusion

Classical RIP coupled with ultrasensitive semi-ligation based small RNA-seq revealed multiple previously unreported mRNAs associated with HuR protein. Newly developed protocol DiCLIP enables characterization of RNA binding sites of RBPs from low sample inputs and with higher resolution as compared to classic iCLIP methodology.