Ancient Gene Duplicates in Gossypium (cotton) Exhibit Near-Complete Expression Level Divergence and Signatures of Expression Level Neo- and Sub-Functionalization

Whole genome duplication (WGD) is widespread in flowering plants and is likely a driving force in angiosperm diversification. The redundancy introduced by WGD allows the evolution of novel gene interactions and functions, although the patterns and processes leading to functional diversification are poorly understood. The recently released genome sequence of Gossypium raimondii (cotton) revealed an approximately 60 million year old five to six-fold ploidy increase, from which we identified ~2,000 pairs of duplicated paralogous genes. Gene expression analyses of these duplicates revealed that 99.4% of gene pairs in G. raimondii exhibit differential expression in at least one of three tissues (petal, leaf and seed), with 93 to 94% exhibiting differential expression on a per tissue basis.  For 1,666 (85%) pairs, differential expression was observed in all tissues. These observations were mirrored in a time-series of G. raimondii seed, and separately in leaf, petal and seed of the G. arboreum, the latter indicating expression level diversification prior to the divergence of G. arboreum and G. raimondii. Extending the analysis to include a generalized linear model, revealed 92.4% of the paralog pairs exhibited expression level divergence, with most exhibiting statistically significant gene and tissue interactions. Furthermore, one sixth of all paralog pairs display complementary expression patterns in different tissues. These data indicate massive, near-complete expression level neo- and/or sub-functionalization among ancient gene duplicates. The ubiquity of neo- and/or sub-functionalization among gene pairs suggests that these processes have been essential in the maintenance of ancient gene duplicates over c. 60 mya.