Evolution of novel genetic programs: insights from flower and fruit development

Abstract

An outstanding question in evolutionary biology is how genetic programs (interaction between multiple genes or their products) that define novel phenotypes evolve. There are two major ways such genetic programs that define novel phenotypes can evolve. First, by a de novo assembly of previously non-interacting genes during the origin of the novel phenotype. Secondly, by co-option (re-deployment, re-recruitment) of existing programs from other functions towards defining the novel phenotype. Monosymmetry of flowers is a novel phenotype that has evolved at least 130 times from polysymmetric flowers during the diversification of flowering plants. Flower monosymmetry in the order Lamiales is defined by an interaction of CYCLOIDEA, RADIALIS, DIVARICATA, and DIVARICATA and RADIALIS interacting factors (CYC–RAD–DIV–DRIF). This interaction is best understood in the Lamiales species Antirrhinum majus. The evolutionary history of the CYC–RAD–DIV–DRIF genetic program that defines Lamiales flower monosymmetry is unclear. It is an open question whether this genetic program was assembled de novo near the base of Lamiales during the evolution of flower monosymmetry or it was co-opted from a different function. We find evidence that the CYC–RAD–DIV–DRIF genetic program, which is crucial for defining the novel phenotype of flower monosymmetry in Lamiales, was likely co-opted from a different function, possibly fruit and ovary development. We come to this conclusion through a comparative analysis between representative taxa from the Lamiales and its close relative the Solanales. This suggests that the evolution of flower monosymmetry in Lamiales may have been facilitated by the availability of the CYC–RAD–DIV–DRIF genetic program. This genetic program was likely co-opted in defining flower monosymmetry in Lamiales for the following reasons. This program was ancestrally involved in regulating cell size: and this function makes it a likely candidate for defining a novel phenotype that has variously shaped petals and aborted stamens. Also, the CYC–RAD–DIV–DRIF genetic program functions by a competitive interaction between RAD and DIV proteins—the two competing proteins could easily be co-opted in defining the two morphologically distinct regions of the monosymmetric flowers. Little is known about the regulators that affect the transcription of CYC in Lamiales. We find predictive, bioinformatics-based evidence for a Lamiales-specific transcriptional autoregulation of CYC, suggesting that the evolution of flower monosymmetry in this lineage was associated with an autoregulation-mediated sustained, stable, and high transcription of CYC. Our data elucidates the evolutionary origin of the CYC–RAD–DIV–DRIF genetic program that defines flower monosymmetry in Lamiales. This interaction was likely co-opted en bloc from fruit and ovary development, with additional lineage-specific changes in gene regulation leading to transcriptional autoregulation of CYC. Entangled in this question is the fact that ovaries and fruits of many species in the tribe Antirrhineae (order Lamiales), including Antirrhinum majus, are monosymmetric—with unequal dorsal and ventral locules. We determine the genetic and micromorphological basis of this phenotype and also estimate its evolutionary history. We identify at least five evolutionary transitions from polysymmetric to monosymmetric ovaries, and at least seven reversals to polysymmetry across the tribe. Ovary monosymmetry in Antirrhinum and its closest relatives is likely controlled non-cell autonomously by a CYC–RAD interaction. CYC upregulates RAD expression in dorsal petals and stamens in early stages of development, and RAD protein from these tissues migrates to the dorsal locule to increases cell proliferation in dorsal ovary wall, causing the ovary to be monosymmetric.