SRSF3 confers selective processing of miR-17-92 cluster to promote tumorigenic properties in colorectal cancer

Almost a half of microRNAs (miRNAs) in mammalian cells are generated from polycistronic primary transcripts encoding more than one miRNA. Mature miRNAs from polycistronic clusters frequently regulate complementary sets of target mRNAs. How the processing of individual miRNAs within the clusters is controlled to give rise to distinct miRNA levels in vivo is not fully understood. Our investigation of SRSF3 (Serine-Arginine Rich Splicing Factor3) regulated noncoding RNAs in pluripotent cells identified miR-17-92 cluster as a key SRSF3 target, SRSF3 binding to the CNNC motif 17-18nt downstream of the miRNA stem loop. Here we show that SRSF3 binding site context, not merely the distance from the stem loop, within primary transcript is a critical determinant of the processing efficiency of distinct miRNAs derived from the miR-17-92 cluster. SRSF3 specifically enhanced the processing of two paralog miRNAs, miR-17 and miR-20a, targeting overlapping mRNAs including the cell cycle inhibitor CDKN1A/p21. Functional analysis demonstrated that SRSF3 inhibits CDKN1A expression and promotes cell cycle and self-renewal through the miRNA processing pathway both in normal pluripotent stem cells and cancer cells. Strikingly, analysis of colorectal cancer tumour-normal pairs demonstrated that the SRSF3-regulated miRNA processing pathway is present in a large proportion of colorectal cancer patients and distinguishes poorly differentiated high-grade tumours. Our research uncovers a critical role of SRSF3 in selective processing of miR-17-92 miRNAs, which mechanistically and functionally links SRSF3 to hallmark features of cancer.