Abstract
Genes containing multiple pre-mRNA cleavage and polyadenylation sites, or polyA sites, express mRNA isoforms with variable 3' untranslated regions (UTRs). By systematic analysis of human and mouse transcriptomes, we found that short 3'UTR isoforms are relatively more abundant when genes are highly expressed whereas long 3'UTR isoforms are relatively more abundant when genes are lowly expressed. Reporter assays indicated that polyA site choice can be modulated by transcriptional activity through the gene promoter. Using global and reporter-based nuclear run-on assays, we found that RNA polymerase II is more likely to pause at the polyA site of highly expressed genes than that of lowly expressed ones. Moreover, highly expressed genes tend to have a lower level of nucleosome but higher H3K4me3 and H3K36me3 levels at promoter-proximal polyA sites relative to distal ones. Taken together, our results indicate that polyA site usage is generally coupled to transcriptional activity, leading to regulation of alternative polyadenylation by transcription.