Floral color is not as simple as it once seemed.

A false start mutation produces reduced protein and flower color, highlighting the role of mutations affecting protein translation in phenotypic evolution and variation.

One of the major questions biologists continue to investigate is how evolution proceeds. Specifically, what are the genetic mechanisms that control phenotypic change, and are the mutations responsible for these changes always similar or different? These questions become more compelling when predicting genetic mechanisms for parallel or convergent evolution: Are the same phenotypes always induced by identical genetic changes, or are there different routes to the same phenotype?

Floral color is an ideal phenotype for studying these questions because it is a broadly distributed trait often with an essential role in reproductive divergence. By dissecting the genetic mechanisms of floral color gains, losses, and shifts, researchers have established a handful of highly tractable systems for the study of floral color including Petunia (, ), Antirrhinum (), Phlox (), Penstemon (), Ipomoea (), and Mimulus ().

Over the past two decades, researchers have been able to investigate the genetic mechanisms of floral color changes with more quantitative techniques of identifying the responsible genes and proteins, analyzing gene expression and molecular genetic identification of candidate genes. This has reframed the classification of causal or contributing mutations into cis- or trans-regulatory and gain- or loss-of-function mutations. With the advent of more accessible genome sequencing and high-throughput technology, assessing suspected gene networks and potential mutational targets for genetic mechanisms like floral color is realistic and where the Mimulus (monkeyflower) system has excelled (Fig. 1). Floral color evolution is not as “simple” as it was originally thought to be, despite the deceivingly simple phenotypic transitions between colors or gains and losses.

Fig. 1.

Monkeyflowers come in all sorts of shapes, sizes, and colors.

The molecular genetics of floral color in monkeyflowers is much more complicated than originally thought, with complex networks of transcriptional activators and repressors to create the deeply red and lightly pink colored species. Mimulus parishii, a small, self-pollinating and light-pink colored species, developed a previously unidentified way to disable the gene that allows for deep pigmentation through a false-start protein, reducing protein translation instead of gene expression. Photo credit: Mei Liang.

What is particularly compelling about the wildflower genus Mimulus is that it has effectively been used to study ecological and evolutionary processes as well as developmental genetics (). Mimulus species with distinct floral colors have been associated with different floral pigments and pollinators, allowing for greater understanding of the relative importance of mutations leading to phenotypic change and speciation. Not only does Mimulus use combinations of anthocyanins (the most common type of pigments for red, pink, and blue colors) and carotenoids (yellow and orange pigments) (), but there is often extensive pigment patterning in the form of nectar guide spots, lines, and spatial patterning (). Furthermore, transgenic techniques can be performed with relative ease in Mimulus to allow for targeted experiments to understand genetic mechanisms of floral color.

The M. lewisii species complex provides a small set of genotypically similar species that have either high concentrations of anthocyanins and carotenoids (dark red pigmentation in hummingbird-pollinated M. cardinalis and M. verbenaceus) or low concentrations of anthocyanins (light pink bee-pollinated M. lewisii and selfing M. parishii) in petal lobes. A simple explanation of this evolutionary pattern would be increased expression of a single MYB transcription factor to induce dark anthocyanin pigmentation and decreased expression of the same MYB transcription factor to induce lighter anthocyanin pigmentation. Instead, the difference in anthocyanin concentration between red M. cardinalis and pink M. lewisii is attributed to ROSE INTENSITY1 (ROI1), a MYB repressor that is more highly expressed in light pink M. lewisii. In this issue of Science Advances, Liang et al. investigate whether the parallel evolution of light pink color at the phenotypic level in M. parishii is controlled by the same genetic mechanism and series of mutational events.

Liang et al. first take a classical approach to determine the genetic basis of floral color in M. parishii by comparing gene expression of MYB transcription factor ROI1 between two dark red and two light pink species (including light pink M. lewisii). ROI1 expression of M. parishii was expected to be high, similar to M. lewisii (since they share the same floral color). However, M. parishii ROI1 expression was just as low as its sister species that are dark red, signaling low repressor activity. This initial result suggests that the genetic mechanism of light pink color in M. parishii is in fact different than in sister species M. lewisii and thus not by action of the ROI1 anthocyanin repressor.

To identify which gene(s) are implicated in the low levels of anthocyanin pigments, the light pink M. parishii was crossed to dark red M. cardinalis, followed by genetic mapping to identify a single location in the genome responsible for dark red versus pale pink anthocyanin pigmentation. Within the identified region, Liang et al. found PETAL LOBE ANTHOCYANIN (PELAN), which is a known MYB transcription factor that is necessary to activate anthocyanin biosynthesis in Mimulus (). The implication of PELAN is curious, since high expression of this anthocyanin activator is more obviously associated with the deep red pigmentation of sister species M. cardinalis. In the light pink sister species M. lewisii, PELAN is necessary for anthocyanin pigment activation but anthocyanin expression is curtailed by the repressor ROI1, which is not highly expressed in light pink M. parishii. If PELAN expression goes unchecked by ROI1 repressor, how is M. parishii not deeply pigmented like M. cardinalis?

Drawing on lessons from the floral color literature, the light pink phenotype of M. parishii must then be due to either lower PELAN expression or altered protein function. Yet, through a series of molecular genetic and transgenic experiments, Liang et al. demonstrate that the M. parishii PELAN gene expression is unremarkable and that the predicted coding sequence of the protein appears to be functional when experimentally overexpressed. These are perplexing results, but if gene expression is unremarkable, yet a reduced phenotype is observed (in the form of light pink versus red color), then some mutation must still be present that interferes with protein translation from mRNA. Expanding the view to the areas directly outside of the protein-coding part of the gene, a single nucleotide mutation in the 5′UTR (the untranslated region just before the “start” of the protein coding section of a gene) was discovered in M. parishii. While we typically view mutations to the protein-coding sections of genes as extremely important in inducing phenotypic differences, mutations to the flanking regions are often just as important with major phenotypic effect. The single mutation creates a “false-start” signal to the protein translation machinery, which then translates from the wrong starting position and creates a small, nonfunctional protein with only 10 amino acids. This single mutation therefore does not induce errors in gene expression (transcription and mRNA processing) or protein function if translated correctly but instead induces a mistranslation of the PELAN protein, interfering with proper anthocyanin biosynthetic pathway activation and resulting in light pink instead of dark red floral color.

Traditional approaches to floral color gains, losses, and shifts have consistently revealed differences in gene expression (transcription) in pigment biosynthetic genes and transcription factors (if they can be identified), with many characterized cis-regulatory or coding sequence mutations to transcription factors. Not finding compelling evidence for the usual suspects, Liang et al. demonstrated that something so simple as an alternate start codon that encodes an incredibly small protein can be responsible for major phenotypic change. This work is important because it will push forward not only research in floral color genetics but also research into the mechanisms of phenotypic evolution as well as speciation. Many researchers who study phenotypic evolution based on phenotypic characterization and gene expression are reluctant to dive into protein work since it is often a completely different set of skills. However, this paper demonstrates that investigating relative protein abundance is not insurmountable and is possibly more important than previously thought.