Abstract
Alternative transcription initiation (ATI) appears to be a ubiquitous regulatory mechanism of gene expression in eukaryotes. However, the extent to which it affects the products of gene expression and how it evolves and is regulated remain unknown. Here, we report genome-wide identification and analysis of transcription start sites (TSSs) in various soybean (Glycine max) tissues using a survey of transcription initiation at promoter elements with high-throughput sequencing (STRIPE-seq). We defined 193,579 TSS clusters/regions (TSRs) in 37,911 annotated genes, with 56.5% located in canonical regulatory regions and 43.5% from start codons to 3' untranslated regions, which were responsible for changes in open reading frames of 24,131 genes. Strikingly, 6,845 genes underwent ATI within coding sequences (CDSs). These CDS-TSRs were tissue-specific, did not have TATA-boxes typical of canonical promoters, and were embedded in nucleosome-free regions flanked by nucleosomes with enhanced levels of histone marks potentially associated with intragenic transcriptional initiation, suggesting that ATI within CDSs was epigenetically tuned and associated with tissue-specific functions. Overall, duplicated genes possessed more TSRs, exhibited lower degrees of tissue specificity, and underwent stronger purifying selection than singletons. This study highlights the significance of ATI and the genomic and epigenomic factors shaping the distribution of ATI in CDSs in a paleopolyploid eukaryote.