Genomic architecture and evolution of heterostyly: new insights from Cordia subcordata (Boraginaceae)

Heterostyly is a floral polymorphism controlled by an S-locus supergene in several angiosperm families. Most heterostylous species are self-incompatible. Here, we investigate the genomic architecture of distyly in self-compatible Cordia subcordata in which incompatibility has apparently been lost. We assembled chromosome-level genomes of floral morphs and conducted population genomic analyses to locate the S-locus region. We used transcriptomic analyses of floral organs and functional validation by gene overexpression to identify genes controlling floral dimorphism. The tempo and mode of origin of S-locus genes was also investigated to determine whether gene duplication facilitated supergene assembly. The candidate S-locus in C. subcordata contained 12 genes, eight of which were restricted to the S-morph. CsGA2ox6 deactivates gibberellins and was exclusively expressed in S-morph pistils. Overexpression of CsGA2ox6 in transgenic tobacco produced flowers with shortened styles and an apparently functioning self-incompatibility system. The genomic locations of paralogs and estimations of duplication age suggested that the S-locus genes may have arisen through stepwise duplications, although an origin via segmental duplication could not be excluded. Our study revealed molecular convergence with several other distylous families in hemizygous structure and possibly in the mode of supergene origins. We also identified a molecular pathway for style-length control, likely through gibberellin deactivation by CsGA2ox6, which may have also controlled the expression of self-incompatibility in transgenic plants.