Reconsidering Homeolog Silencing of Allopolyploid Gene Expressions in Arabidopsis

Allopolyploidization, or the whole genome duplication between different species, is considered to be a driving force in evolution. Recent high-throughput sequencing has allowed us to perform genome-wide quantification of homeolog-specific expression, leading to various significant findings. However, subsequent analysis considering homeolog-specific feature to find functionally significant homeologs after the quantification is still inadequate. First one is the detection of homeologs whose expression balance between homeolog pair significantly changes in two conditions. To overcome the problem, we developed a bioinformatics workflow, HomeoRoq (http://seselab.org/homeoroq/) [NAR, 42(6):e46, 2014] including homeolog-specific expression quantification and statistical test to find the homeologs whose expression ratio significantly changes over the two conditions. Second one is the detection of transcriptional silencing between homeolog pair. For the homeolog-specific silencing, much attention has been focused on the absolute expression level, and thus the homeologs have been detected by the absolute level and a threshold. However, this conventional approach does not consider the expression ratio between homeolog pair. Here, we reconsider homeolog silencing by focusing on the genome-wide distribution of expression ratio and the homeologs whose expression ratio is extreme biased. In this study, we first show that the expression pattern in allotetraploid Arabidopsis kamchatica derived from diploid A. lyrata and A. halleri can be divided into the homeologs with and without the extreme bias. We then developed a statistical method to detect the extreme bias homeologs using RNA-seq. The results showed that ~200 to ~700 homeologs were detected only as extreme bias expression and included the homeologs showing species-specific function.