Subgenome-informed statistical modeling of transcriptomes in 25 common wheat accessions reveals cis- and trans- regulation architectures

Common wheat is allohexaploid, where it is difficult to obtain homoeolog-distinguished transcriptome data. Lasy-Seq, a type of 3' RNA-seq, is a technology efficient at obtaining homoeolog-distinguished transcriptomes. Here we applied Lasy-Seq to obtain transcriptome data from the seedlings, second leaves, and root tips of 25 common wheat lines mainly from East Asia. Roots and seedlings were similar to each other in transcriptome profiles, but they were different from the leaves. We then asked how three homoeologous genes from different subgenomes (i.e., triads) show different levels of expression. Specifically, we examined the effects of subgenomes, lines and their interaction on the expression levels of each homoeolog triad, separately in each tissue. Of the 19,805 homoeolog triads, 51-55% showed significant effect of subgenome, suggesting cis-regulation, whereas 24-30% showed significant effect line, suggesting trans-regulation. We also found that 7.7-9.0% triads showed significant effects of the interaction. Hierarchical clustering and co-trans regulation network analysis of homoeolog triads revealed that the patterns of expression polymorphisms among the lines were shared in different genes. Our results also implied that expression variation between lines is caused by changes in a smaller number of common trans-factors. We performed GO term enrichment analysis using newly annotated and substantially improved GO annotations, which revealed that GO terms related to each tissue type function were enriched in genes expressed in the leaves and roots. Our information provides fundamental knowledge for the future breeding of plants possessing complex gene regulatory networks such as common wheat.