Convergent Evolution at the Pathway Level: Predictable Regulatory Changes during Flower Color Transitions

Convergent Evolution at the Pathway Level: Predictable Regulatory Changes during Flower Color... Abstract The predictability of evolution, or whether lineages repeatedly follow the same evolutionary trajectories during phenotypic convergence remains an open question of evolutionary biology. In this study, we investigate evolutionary convergence at the biochemical pathway level and test the predictability of evolution using floral anthocyanin pigmentation, a trait with a well-understood genetic and regulatory basis. We reconstructed the evolution of floral anthocyanin content across 28 species of the Andean clade Iochrominae (Solanaceae) and investigated how shifts in pigmentation are related to changes in expression of seven key anthocyanin pathway genes. We used phylogenetic multivariate analysis of gene expression to test for phenotypic and developmental convergence at a macroevolutionary scale. Our results show that the four independent losses of the ancestral pigment delphinidin involved convergent losses of expression of the three late pathway genes (F3′5′h, Dfr, and Ans). Transitions between pigment types affecting floral hue (e.g., blue to red) involve changes to the expression of branching genes F3′h and F3′5′h, while the expression levels of early steps of the pathway are strongly conserved in all species. These patterns support the idea that the macroevolution of floral pigmentation follows predictable evolutionary trajectories to reach convergent phenotype space, repeatedly involving regulatory changes. This is likely driven by constraints at the pathway level, such as pleiotropy and regulatory structure. anthocyanin pathway, gene expression, floral pigment evolution Introduction The prevalence of convergent evolution, that is, the emergence of similar forms in different lineages, suggests that evolution may often be repeatable, with similar outcomes resulting from “replaying the tape of life” (Gould 1990; Orgogozo 2015). Nonetheless, the extent to which these convergent phenotypes arise via similar underlying mechanisms remains an open question (Stern and Orgogozo 2009; Rosenblum et al. 2014). In some cases, convergent phenotypes evolve through distinct genetic or developmental changes (Steiner et al. 2009; Nery et al. 2016), while in others, similar mechanisms are involved (Chan et al. 2010; Martin and Orgogozo 2013). The predictable recruitment of the same mechanisms may be explained by genetic constraints, for example if only one or a few loci have the potential to influence the phenotype, for example, in tetrodotoxin resistance (Brodie and Brodie 2015) and albinism of cavefish (Protas et al. 2006). By contrast, when multiple paths can lead to the same phenotype, the repeated involvement of particular loci or types of changes suggest biases, either in mutation rates or fixation probability (Stern and Orgogozo 2008; Gompel and Prud’homme 2009; Christin et al. 2010; Streisfeld and Rausher 2011; Conte et al. 2012). Still, the ability to assess such biases is limited for many traits where the underlying developmental pathway and the potential routes to convergent phenotypes are largely unknown. Floral anthocyanin pigmentation provides a powerful system for studying phenotypic convergence and the predictability of evolution at the molecular and developmental levels. The anthocyanin branch of the large flavonoid biosynthetic pathway (hereafter the anthocyanin pathway) produces three classes of red, purple, and blue pigments (fig. 1) which comprise the most common pigments determining color in flowers and fruits (Grotewold 2006). Small changes in the anthocyanin pathway (i.e., mutations at a single locus) can radically alter floral hue and/or color intensity (Koes et al. 1994; Mol et al. 1998; Rausher 2006,, 2008), with important ecological consequences, for example, pollinator attraction (Schemske and Bradshaw 1999; Hoballah et al. 2007; Sheehan et al. 2016). Upstream steps of this pathway (enzymes CHS, CHI, F3H; fig. 1) are shared across multiple flavonoid compounds (e.g., flavones and flavonols). The subsequent steps involve the so-called branching enzymes F3′H and F3′5′H, which determine the type of anthocyanin produced, and the downstream enzymes DFR and ANS, which convert the colorless precursors into pigments. Although the entire flavonoid pathway contains dozens of genes, this group of core enzymes (fig. 1) are largely responsible for controlling metabolic flux toward anthocyanin pigments, and together determine the amount and type of anthocyanins produced (Grotewold 2006). This pathway is regulated at the level of transcription, and the regulatory genes are well known (Quattrocchio et al. 2006). In most flowering plants, the upstream steps are jointly regulated by MYB transcription factors while the downstream steps are activated by a complex of MYB, bHLH, and WD40 (MBW) transcription factors (Mol et al. 1998; Stracke et al. 2007). The regulation of the branching genes varies across species and is generally not as well understood as for the structural genes (De Vetten et al. 1999; Schwinn et al. 2006; Albert et al. 2014). Fig. 1. View largeDownload slide Simplified view of the flavonoid biosynthesis pathway adapted from Rausher (2006). Arrows indicate biochemical reactions catalyzed by enzymes (white boxes). Intermediate metabolites are enclosed in gray boxes. Abbreviations as follows: LCP, leucopelargonidin; LCC, leucocyanidin; LCD, leudodelphinidin; CHS, chalcone synthase; CHI, chalcone isomerase; F3H, flavanone 3-hydroxylase; F3′H, flavonoid 3′-hydroxylase; F3′5H, flavonoid 3′,5′-hydroxylase; DFR, dihydroflavonol 4-reductase; ANS, anthocyanidin synthase. The methylated cyanidin derivative peonidin and methylated delphinidin derivatives malvidin and petunidin not shown. The dashed arrow indicates that F3′5′H does not have 3′ hydroxylation activity in Iochrominae (Smith and Rausher 2011). Fig. 1. View largeDownload slide Simplified view of the flavonoid biosynthesis pathway adapted from Rausher (2006). Arrows indicate biochemical reactions catalyzed by enzymes (white boxes). Intermediate metabolites are enclosed in gray boxes. Abbreviations as follows: LCP, leucopelargonidin; LCC, leucocyanidin; LCD, leudodelphinidin; CHS, chalcone synthase; CHI, chalcone isomerase; F3H, flavanone 3-hydroxylase; F3′H, flavonoid 3′-hydroxylase; F3′5H, flavonoid 3′,5′-hydroxylase; DFR, dihydroflavonol 4-reductase; ANS, anthocyanidin synthase. The methylated cyanidin derivative peonidin and methylated delphinidin derivatives malvidin and petunidin not shown. The dashed arrow indicates that F3′5′H does not have 3′ hydroxylation activity in Iochrominae (Smith and Rausher 2011). In the context of the anthocyanin biosynthetic pathway, evolutionary transitions in flower color appear to occur via a small and predictable subset of possible mechanisms. For example, although losses of pigmentation could occur by inactivation along any branch of the pathway, convergent evolutionary transitions to white flowers have been exclusively linked to changes in the MYB genes that control the downstream elements of the pathway (reviewed in Sobel and Streisfeld 2013). The fixation of these regulatory mutations preserves the potential for expressing flavonoids in other tissues, where they function, for example, in protection from UV light (Ryan et al. 2001), signaling (Watkins et al. 2017), and pollen fertility (Mo et al. 1992). Similarly, repeatable patterns of genetic evolution have been described for the evolution of floral hue, where transitions from blue to red flowers frequently often involve genetic and developmental changes inferred to incur minimal pleiotropic effects (Wessinger and Rausher 2012). Nonetheless, these patterns of apparent predictability in the mechanisms of flower color evolution rely on a relatively small number of studies (e.g., five in the case of fixed gains of floral anthocyanins; Sobel and Streisfeld 2013). Here, we use the Iochrominae clade (Solanaceae) to trace the evolution of anthocyanin gene expression during repeated shifts among flower colors and quantify the extent of convergence in pathway regulation. This group of around 40 species, native to South America, displays a remarkable variety of flower colors, largely due to variation in anthocyanin pigmentation. Previous studies have shown that the ancestral flower color in the clade is blue with mainly delphinidin-based anthocyanins (Smith and Baum 2006; Ho and Smith 2016). Multiple lineages have lost these pigments to shift to white or yellow flowers (Ho and Smith 2016), and two lineages have evolved orange to red flowers with cyanidin or pelargonidin pigments (Berardi et al. 2016). In two cases, the genetic basis for these color transitions has been identified. In red-flowered Iochroma gesnerioides, the shift to pelargonidin pigments occurred via the transregulatory loss of F3′h expression, and the inactivation of F3′5′h through gene deletion (Smith and Rausher 2011). In the white flowered I. loxense, the loss of pigment production involved an R3 MYB repressor that blocks expression of multiple downstream genes (Gates et al. 2018). Although the genetic basis is unknown for the remaining color transitions in Iochrominae, several lines of evidence point to an important role of pathway regulation. First, a study of molecular evolution of three core pathway genes (the upstream Chi and F3h and the anthocyanin specific Ans), revealing consistent purifying selection across the clade, including lineages that have lost floral anthocyanins (Ho and Smith 2016). Second, flavonoids, including anthocyanins, are abundant in vegetative tissue (Berardi et al. 2016) as well as fruits (Smith SD, unpublished data), which is consistent with the broad conservation of pathway function. Finally, phylogenetic analyses point to at least one regain of floral anthocyanins following loss, suggesting functionality of the structural pathway genes was maintained, allowing renewed floral expression of the pathway (Ho and Smith 2016). In order to test the prediction that convergent flower color shifts in Iochrominae have repeatedly involved similar regulatory changes, we combined a suite of biochemical, developmental, and phylogenetic approaches. First, we profiled anthocyanin pigment production across 28 species, and estimated the evolutionary history of anthocyanin pigmentation in this clade. Next, we quantified the expression of seven pathway genes and used statistical comparative methods to investigate the relationship between gene expression and pigment phenotypes. Given the likely pleiotropic effects of changes to the early genes of the pathway on other flavonoids, we predicted that the shifts from blue to white flowers involve losses of expression of downstream elements of the pathway. By contrast, we expected that transitions among the blue, purple, and red pigments involve differential regulation of the branching enzymes. We further tested these predictions using multivariate approaches to cluster the species in gene expression space. Together our analyses reveal a tight relationship between floral expression of pathway genes, anthocyanin pigment production, and flower color evolution in Iochrominae. Results Evolutionary Shifts in Pigment Production Consistent with previous work in this system, species within Iochrominae produce all three classes of anthocyanins, and the types present are closely associated with the flower color phenotype. The blue-flowered species produce delphinidin-derived anthocyanins (fig. 2). While some blue-flowered Iochrominae make these pigments exclusively, most also synthesize small amounts (<15%) of cyanidin-derived anthocyanins (supplementary fig. S1, Supplementary Material online). The two red-flowered sister species, I. gesnerioides and I. fuchsioides produce the red anthocyanidin pelargonidin, and the orange-red flowered I. edule is the only species which produces exclusively cyanidin (supplementary fig. S1, Supplementary Material online). As predicted, six white or yellow flowered species produce no detectable anthocyanidins (supplementary fig. S1, Supplementary Material online). Fig. 2. View largeDownload slide Evolution of main floral anthocyanidin pigments in Iochrominae. (A) Phylogeny of Iochrominae, with ancestral state reconstruction of anthocyanidins from stochastic mapping, with root state fixed at delphinidin. Pies at nodes represent frequencies of node states across 1,000 simulations of character evolution. Disks at tips are colored by species main anthocyanidin pigment for each species 5 (supplementary fig. S1, Supplementary Material online). (B) Transitions between character states, summarized over 1,000 stochastic histories, with arrow width proportional to average number of transitions (see supplementary table S2, Supplementary Material online). Crosses indicate nodes with posterior probability <0.95. Fig. 2. View largeDownload slide Evolution of main floral anthocyanidin pigments in Iochrominae. (A) Phylogeny of Iochrominae, with ancestral state reconstruction of anthocyanidins from stochastic mapping, with root state fixed at delphinidin. Pies at nodes represent frequencies of node states across 1,000 simulations of character evolution. Disks at tips are colored by species main anthocyanidin pigment for each species 5 (supplementary fig. S1, Supplementary Material online). (B) Transitions between character states, summarized over 1,000 stochastic histories, with arrow width proportional to average number of transitions (see supplementary table S2, Supplementary Material online). Crosses indicate nodes with posterior probability <0.95. Analyses of trait evolution with stochastic mapping supported strong directionality in transitions among pigment types (fig. 2), similar to previous work in this and other clades (reviewed in Rausher 2008). The most common transitions were inferred to be convergent losses of ancestral delphinidin production, with stochastic mapping estimating an average of five losses across the clade (fig. 2 and supplementary table S3, Supplementary Material online). The persistence of just four unpigmented lineages suggests that one of these transitions was reversed. The earliest loss likely occurred c. 3 Ma on the branch leading to the “A” clade containing the unpigmented species I. ellipticum, I. salpoanum, I. confertiflorum, Acnistus arborescens (fig. 2A). Additional more recent shifts involved I. lehmannii, I. loxense, and D. solanacea. Stochastic mapping estimated that transitions from unpigmented to producing cyanidin occurred at least twice in the clade’s history (fig. 2B), although only one extant species, I. edule, produces this pigment exclusively. Its position, nested within the unpigmented “A” clade, suggests that it has recently regained floral anthocyanin expression. Shifts to floral pelargonidin production are inferred to be rare, likely occurring only once in the common ancestor of the two red-flowered species. Several transitions (e.g., pelargonidin to cyanidin, delphinidin to cyanidin) were never observed among the stochastic maps (fig. 2 and supplementary table S3, Supplementary Material online), indicating they are unlikely to have occurred during the history of Iochrominae. Evolution of Anthocyanin Pathway Gene Expression Our comparative analysis of gene expression revealed distinct evolutionary patterns across pathway genes. Expression levels of the upstream genes (e.g., Chs, Chi, and F3h) were similar across Iochrominae regardless of pigment composition, with log fold-change values mostly between −1 and +1, that is, transcript concentrations within an order of magnitude of the reference blue-flowered I. cyaneum (fig. 3A). There were a few exceptions; for example, Chs had low expression (log fold-change of −3) in the white-flowered I. ellipticum and F3h showed moderate down-regulation (log fold-changes of around −1.5) in several species which lack floral anthocyanins (I. ellipticum, I. confertiflorum, and D. solanacea; fig. 3A). The phylogenetic ANOVA suggested that despite these cases, species without floral anthocyanins did not have significantly lower expression for Chs, Chi, or F3h than their pigmented relatives (fig. 3B). This pattern of conservation of upstream gene expression in lineages that have lost floral anthocyanins is consistent with their continued production of other floral flavonoids, in particular flavonols (supplementary fig. S1, Supplementary Material online). These compounds are typically as abundant in flower tissues as anthocyanins and remain abundant in most white- and yellow-flowered species (Berardi et al. 2016; Larter M, Berardi A, and Smith SD, unpublished data). Fig. 3. View largeDownload slide Anthocyanin pathway gene expression in Iochrominae. (A) Gene expression for each species, with bars corresponding to each tip of the phylogeny. Colors of bars and dots at tips represent main floral anthocyanidins (supplementary fig. S2, Supplementary Material online). Bracket represents the “A clade” sensu Smith and Baum (2006). (B) Log-fold-change gene expression aggregated by floral anthocyanidin production. Pink circles represent I. edule, that makes cyanidin. The letters represent significant differences under a pairwise Tukey comparison test (alpha=0.05). Fig. 3. View largeDownload slide Anthocyanin pathway gene expression in Iochrominae. (A) Gene expression for each species, with bars corresponding to each tip of the phylogeny. Colors of bars and dots at tips represent main floral anthocyanidins (supplementary fig. S2, Supplementary Material online). Bracket represents the “A clade” sensu Smith and Baum (2006). (B) Log-fold-change gene expression aggregated by floral anthocyanidin production. Pink circles represent I. edule, that makes cyanidin. The letters represent significant differences under a pairwise Tukey comparison test (alpha=0.05). In contrast with the evolutionary conservation of expression observed for the upstream genes, the other elements of the pathway showed strong color-specific patterns of expression. In the two red-flowered species, which produce only pelargonidin, we observed extremely low expression of the branching genes F3′h and F3′5′h (with log fold change of −3 or lower, or c. 1,000 times fewer transcripts than in the reference blue species), while the remaining steps of the pathway had comparable expression levels to blue-flowered species (fig. 3B). In the unpigmented lineages, the two downstream genes Dfr and Ans showed similarly strong down-regulation (log fold change of around −3), along with the branching gene F3′5′h (fig. 3B). The retention of F3′h expression in these lineages is likely tied to the prevalence of quercetin, a flavonol that requires F3′H activity (fig. 1), in their floral tissue (supplementary fig. S1, Supplementary Material online). The cyanidin-producing species I. edule, which is nested within the unpigmented A clade, has recovered expression of F3′5′h, Dfr, and Ans, although we were unable to assess statistical significance with only a single tip (fig. 3A and B). Coexpression Patterns of the Anthocyanin Pathway Genes Given that anthocyanin pathway genes are known to be jointly regulated by MBW complexes, we used multivariate approaches to quantify coexpression and relate these patterns to floral pigmentation variation. The phylogenetic PCA (pPCA) of expression levels grouped species by primary pigment types, and thus flower color (fig. 4A). The first two PC axes explained ∼90% of the variance (fig. 4A). The first PC axis divides the pigmented and unpigmented species (fig. 4A), with high loadings of expression levels of F3′5′h, Dfr, and Ans (fig.4B). The second PC axis separates the red pelargonidin-producing species from the rest of Iochrominae, with high loadings of expression of branching genes F3′h and F3′5′h. The two centermost points, slightly detached from the blue-flowered cluster, are the red-orange I. edule and purple-spotted S. punctata. The separation of I. edule might be expected as it is the only species producing primarily cyanidin and has high F3′h expression but relatively low F3′5′h expression. S. punctata produces primarily delphinidin, but expression levels for pathway genes are generally lower than the other blue and purple taxa, perhaps due to the spotted phenotype. Fig. 4. View largeDownload slide Phylogenetic PCA of anthocyanin pathway gene expression. (A) Species scores on PC1 and PC2, with symbols detailed in box below. Symbols are filled based on pigment production as follows: red for pelargonidin, pink for cyanidin, white for delphinidin, and yellow for no anthocyanidins. Symbol outline colors indicate k-means clustering with k = 3 (as in text): black, red, and blue outlines. (B) Arrows show loadings for each variable on the PC axes, scale shown on bottom and right axes. (C) Pairwise gene expression correlation table. Circle size and color scale represent magnitude of the correlation coefficients, and stars within circles show significance level (P < 0.05, 0.01, and 0.001 for one, two, and three stars, respectively). Fig. 4. View largeDownload slide Phylogenetic PCA of anthocyanin pathway gene expression. (A) Species scores on PC1 and PC2, with symbols detailed in box below. Symbols are filled based on pigment production as follows: red for pelargonidin, pink for cyanidin, white for delphinidin, and yellow for no anthocyanidins. Symbol outline colors indicate k-means clustering with k = 3 (as in text): black, red, and blue outlines. (B) Arrows show loadings for each variable on the PC axes, scale shown on bottom and right axes. (C) Pairwise gene expression correlation table. Circle size and color scale represent magnitude of the correlation coefficients, and stars within circles show significance level (P < 0.05, 0.01, and 0.001 for one, two, and three stars, respectively). The pattern of grouping in gene expression space by pigment type was confirmed by k-means clustering. With k = 3, we recovered the same three clusters: the species that 1) make no anthocyanins, 2) those that make pelargonidin, and 3) those that make either delphinidin- or cyanidin-based anthocyanins. This latter cluster is broken when k = 4, but not into groups clearly reflecting different pigmentation phenotypes. Both subclusters contain blue-flowered species predominantly producing delphinidin, however they are slightly differentiated along PC1 of the pPCA of gene expression (supplementary fig. S3, Supplementary Material online). In addition to identifying phenotypic clusters of species, the pPCA highlights the high levels of correlated expression among groups of pathway genes (fig. 4C). The upstream genes Chs and Chi are significantly coexpressed with F3h across the phylogeny (r2 = 0.55 and 0.36; P < 0.001 and 0.02, respectively; supplementary table S4, Supplementary Material online), and these genes show no significant correlations with the remaining genes. A similar pattern is exhibited by F3′5′h and the downstream genes Dfr and Ans (r2 = 0.49–0.87, P < 0.001). Intriguingly, the expression of the branching gene F3′h shows no significant correlations with the expression of any of the other pathway genes. This result aligns with previous crossing studies indicating an independent (unidentified) transregulatory factor for this branch of the pathway in Iochroma (Smith and Rausher 2011). Discussion Regulatory Mechanisms Drive Evolution of Floral Pigmentation Intensity Our results show that convergent losses of floral anthocyanin production are driven at the pathway level by convergence in the expression of pathway genes in Iochrominae. Ancestral delphinidin production has been lost independently in four white- and yellow-flowered lineages, and in each case these transitions are associated with 1,000-fold decrease in expression of the branching gene F3′5′h and the downstream genes Dfr and Ans. These repeatable, coordinated expression shifts at the macroevolutionary level are striking given the wide range of mechanisms known to generate unpigmented flowers in natural populations and horticultural varieties (Streisfeld and Rausher 2011). For example, both regulatory and structural mutations at the Chs loci have been found to underlie white morphs of some normally pigmented species (Habu et al. 1998; Coberly and Rausher 2003; Dick et al. 2011). This contrast between the range of mechanisms that can give rise to the derived phenotype and those that underlie species level transitions suggests bias toward particular genetic paths, possibly due to mutational target size or pleiotropy (Gompel and Prud’homme 2009). Given that flavonoids play important roles in physiological processes throughout the plant (Winkel-Shirley 2001), pleiotropic effects of mutations that change flower color are particularly likely to affect their probability of fixation over short- and long-evolutionary timescales (Coberly and Rausher 2008). Although we have not identified the exact genetic changes that caused the coordinated down-regulation of F3′5′h, Dfr, and Ans in the unpigmented lineages, several lines of evidence point to a transregulatory mechanism. First, the tight coexpression of these three genes across the phylogeny suggests that they belong to a single regulatory module. Indeed, Dfr and Ans are known to be jointly regulated by an MBW transcription factor complex in a wide range of taxa (Holton and Cornish 1995; Mol et al. 1998). It appears that F3′5′h belongs to this regulatory module in Iochrominae (Gates et al. 2018) as it does many but not all eudicots (De Vetten et al. 1999; Butelli et al. 2008; Albert et al. 2014). Furthermore, losses of pigmentation in other lineages also tend to be driven by transregulatory mutations (reviewed in Wessinger and Rausher 2012; Sobel and Streisfeld 2013). These often involve the MBW complex (e.g., Schwinn et al. 2006; Hoballah et al. 2007) and lead to loss of expression in the downstream genes (Whittall et al. 2006; Cooley et al. 2011). Among the transregulators of the downstream genes, the MYB genes have emerged as the dominant players in flower pigment evolution. Indeed, changes in these transcription factors have been found to underlie every case of evolutionary gain or loss of anthocyanins in which the genetic basis has been dissected (reviewed in Sobel and Streisfeld 2013). This pattern may relate to the fact that MYBs are often highly tissue-specific, and determine the transcriptional targets for the MBW complex (Ramsay and Glover 2005). Thus either cis- or coding mutations at these loci will have limited pleiotropic effects, relative to mutations in other elements of the complex or in the pathway itself (Streisfeld and Rausher 2011). In addition to this activating role as part of the MBW complex, MYB proteins can act at transcriptional repressors of the anthocyanin pathway (Aharoni et al. 2001; Albert et al. 2014; Liu et al. 2015). Most transitions to white or yellow flowers involve loss-of-function mutations in the MYB activators (Sobel and Streisfeld 2013) although in Iochroma loxense, a gain-of-function in an MYB repressor was responsible (Gates et al. 2018). Given the similarity we observe in expression profiles, with strongest downregulation in F3′5′h, Dfr, and Ans, changes at this same MYB repressor may underlie the losses of floral pigmentation in other Iochrominae lineages. Investigating this possibility will require a multifaceted approach as the rapid evolution of MYB genes and their immense copy number (over 100 in Iochroma; Gates et al. 2016) complicates standard candidate gene experiments. Understanding the mechanisms responsible for the repeated losses of floral anthocyanins in Iochrominae has important implications for potential regains of pigmentation. Given that the structure of the pathway appears conserved (Ho and Smith 2016) and remains active in nonfloral tissues (Berardi et al. 2016; Smith SD, unpublished data), an evolutionary gain of floral anthocyanins likely requires only reactivation of the pathway in floral tissue. This could occur via downregulation of a repressor (Yuan et al. 2013) or upregulation of an activator (Streisfeld et al. 2013). Indeed, this may be the case with I. edule, which has regained floral anthocyanins following a loss roughly 5 Ma. Compared with its white and yellow relatives in the A clade, this species has elevated F3′5′h, Dfr, and Ans expression, a change that could be achieved by their shared regulator. Nonetheless, I. edule did not regain the ancestral blue delphinidin pigmentation but instead produces purple-pink cyanidin. This unique feature may represent incomplete reversal, given that the expression of F3′5′h (along with Dfr and Ans) is still lower than all of the blue-flowered species (fig. 3B). Alternately, additional changes in the pathway following the loss may have altered the consequences for reactivation of the F3′5′h-Dfr-Ans module. For example, DFR may have shifted its specificity to the less hydroxylated DHQ as cyanidin is more common in vegetative tissue (Wessinger and Rausher 2012). The functionality of F3′5′h may have also degraded since its products (myricetin and delphinidin) do not appear in vegetative tissues of A-clade species (Berardi et al. 2016; supplementary fig. S1, Supplementary Material online). Genetic Targets of Shifts in Floral Hue In contrast to shifts in the intensity of pigmentation, shifts in floral hue due to transitions between anthocyanin pigments are driven by changes to the branching genes F3′h and F3′5′h. The relative activity of these enzymes determines the abundance of the pigment precursors (the three dihydroflavonols, fig. 1) and in turn, the flux toward the three classes of pigments. Thus, shifts in F3′h and F3′5′h expression can alter which branch of the anthocyanin pathway is most active thereby changing the main floral anthocyanin pigment (Rausher 2008). Previous work in Iochroma demonstrated that the loss of both F3′H and F3′5′H activity is required for the complete transition from the ancestral delphinidin production to the derived pelargonidin production (Smith and Rausher 2011). In the red-flowered I. gesnerioides, this was achieved by flower-specific loss of F3′h expression due to a transregulatory change along with degeneration of the F3′5′h gene. Here, we find that the downregulation of F3′h is also found in the red-flowered sister species I. fuchsioides. Both species produce trace amounts of cyanidin in flowers and leaves and substantial amounts of quercetin in leaves, indicating that F3′H remains functional (Berardi et al. 2016). Additional studies will be required to determine whether I. fuchsioides, like I. gesnerioides, lacks a functional F3′5h gene. Even if I. fuchsioides has retained F3′5′h, its expression is essentially undetectable in flowers (fig. 3A). In this case, the initial loss of expression of F3′5′h in the common ancestor of I. fuchsioides and I. gesnerioides may have set the stage for the later degeneration of the coding sequence in one species (I. gesnerioides). Although the two red-flowered Iochroma are the result of a single evolutionary transition, the patterns we observe mirror those found in shifts in hue in other taxa. For example, shifts from delphinidin to cyanidin or pelargonidin production consistently involve coding and/or regulatory changes at F3′5′h (Nakatsuka et al. 2006; Takahashi et al. 2010; Hopkins and Rausher 2011; Wessinger and Rausher 2015) and exclusively regulatory changes at F3′h (Streisfeld and Rausher 2009; Des Marais and Rausher 2010). Given these patterns, Wessinger and Rausher (2012,, 2015) posited that F3′5′h is a “disposable” gene while F3′h is constrained due to its functions beyond flower color. Indeed, the flavonol quercetin, which requires F3′H activity has higher antioxidant and antiradical activity (Burda and Oleszek 2001; Hirano et al. 2001) and provides better UV protection (Ryan et al. 2001, 2002) than myricetin, which requires F3′5′H activity. Thus, the pleiotropic cost of losing F3′h function is likely higher than that for F3′5′h (Streisfeld and Rausher 2011). Macroevolutionary Signatures of Regulatory Structure Given the complex patterns of pigment production across tissues and in response to environmental cues, the anthocyanin pathway has long served as a model for studying the spatial and temporal regulation of metabolic pathways. Early studies of anthocyanin regulation revealed that steps in the pathway are often coordinately controlled by shared transcription factors, but that these relationships can differ across taxa. For instance, the anthocyanin biosynthetic genes are divided into two coexpressed clusters (“early” and “late” genes) in Eudicots, but which genes belong to those clusters varies across taxa (e.g., Martin et al. 1991; Quattrocchio 1993; Pelletier and Winkel-Shirley 1996). The discovery of these regulatory relationships has and continues to rely on detailed studies of transcription factor and structural gene expression in experimental model systems (e.g., Quattrocchio 1993; Butelli et al. 2008). Our analyses suggest that comparative studies of gene expression at the pathway level may provide a complementary approach to detecting coregulated modules. The phylogenetic PCA recovered the strong coexpression of F3′5′h, Dfr, and Ans, a group that corresponds to “late” genes in the model system Petunia (Mol et al. 1998; Albert et al. 2014). The traditional “early” genes in Petunia (Chs, Chi, F3h; Quattrocchio 1993) emerge here as a loose cluster, likely reflecting the redundant regulation of these steps by multiple MYB transcription factors (Quattrocchio et al. 2006). Finally, the expression of F3′h, which is controlled by a separate transregulator (Smith and Rausher 2011), is decoupled from other parts of the pathway. The close correspondence between the correlation structure of expression variation in these genes and the regulatory structure known from plant model systems likely reflects the predominance of transregulatory mutations in flower color evolution. If indeed the fluctuation in pigment intensity across the clade arises from changes in shared MYB transcription factors, multiple genes will show highly correlated variation at a phylogenetic scale. Equally, gene that experience independent control will lack such correlated variation. Conclusions and Future Directions Evolutionary constraints at the pathway level restrict the available trajectories between floral pigmentation phenotypes, leading to strong predictability of the mechanisms involved in phenotypic transitions. The predictable involvement of regulatory changes in branching and downstream steps of the pathway likely arises from selection for maintaining the production of fitness-related flavonoid compounds in other tissues. At the scale of an entire clade, our study supports key features of flower color evolution that have emerged from a growing number of studies in model systems. In particular, we show that evolutionary losses of anthocyanins predictably involve reversible regulatory shut-down of the late pathway genes. Moreover, changes of floral hue require reduced activity of the branching genes, although the underlying mechanism varies due to pleiotropic and regulatory constraints. One explanation for the repeatability of evolution is that while many changes can produce the same phenotypic outcome (e.g., a coding change to an early pathway gene inducing loss of floral pigments), some are filtered out through selection within populations, due to their negative pleiotropic effects (Gompel and Prud’homme 2009). Rare unpigmented morphs are found in multiple blue-flowered Iochrominae species, with previous research showing that in one case, this intraspecific variation is due to a loss-of-function mutation to the late gene Dfr (Coburn et al. 2015). Further work in several polymorphic species could shed some light on the fixation advantage of minimally pleiotropic mutations at the population level explaining the bias in mechanisms visible at longer evolutionary timescales (Stern and Orgogozo 2008; Gompel and Prud’homme 2009). Materials and Methods Study System Iochrominae are distributed across Western South America, from Columbia and Peru to Brazil and northern Argentina. Flower buds were collected from natural populations for 19 of the species. The remaining nine species were grown from seed at the UC-Boulder EBIO Greenhouses. Greenhouse plants were placed outside two to three weeks prior to flower bud collection to provide natural light (without UV filtering) to maximize the stimulation of the anthocyanin pathway. We obtained samples from at least three individuals per species, with the exception of I. ellipticum from the Galapagos Islands, for which we only have one individual, and Vassobia breviflora for which we only have two individuals. Vouchers and accession numbers are available in supplementary table S1, Supplementary Material online. Anthocyanin Identification and Quantification We separated and quantified anthocyanidins (aglycones of anthocyanins) and closely related flavonoids (flavones, flavonols) from floral tissue using High Performance Liquid Chromatography (HPLC). Following Berardi et al. (2016), we extracted anthocyanidins and other flavonoids from silica-dried floral tissue using the acid hydrolysis procedure of Harborne (1998) with minor modifications. Dried extracts were resuspended in methanol with 1% HCl (v/v), and 10 μl were injected and analyzed using an Agilent Infinity 1260 Infinity Series HPLC system with a 100 × 4.6 mm Chromolith Performance RP-18 endcapped column (EMD Chemicals, Darmstadt, Germany) and a photodiode array detector (PDA; scanning from 200 to 600 nm at a step of 2 nm). Separation of the compounds followed the protocols of Berardi et al. (2016). Peaks were detected at 520 nm for anthocyanidins and 365 nm for flavones and flavonols and identified by comparison of retention times and UV-VIS absorbance spectra with commercial standards (Extrasynthese, Genay, France). Chromatograms were visualized using Agilent ChemStation for LC 3 D Systems (Rev. B0403, Waldbronn, Germany). Concentrations were scaled by tissue weight and were estimated by comparing the area of each chromatogram peak to those from dilution series of the corresponding standards. Two of the anthocyanidins (cyanidin and delphinidin) have methylated derivatives, and these were summed for each to obtain the total amount of cyanidin- and delphinidin-derived anthocyanins, that is, matching each branch of the pathway (fig. 1). Relative amounts of the three classes (pelargonidin-, cyanidin-, and delphinidin-derived anthocyanins) were calculated to identify the principal pigment and minor pigments in each species. HPLC data for nine species in this data set were previously published in Berardi et al. (2016). Quantification of Anthocyanin Pathway Gene Expression Floral buds were collected at a stage roughly equivalent to Petunia bud Stage 5 (Pollak et al. 1993), at which time both the early and late parts of the pathway are expressed (Smith and Rausher 2011). Tissue was preserved in RNAlater (Qiagen, Valencia, CA), or flash frozen if collected from a greenhouse plant. The protocol for measuring gene expression followed previous work (Coburn et al. 2015). Briefly, total RNA was extracted using the Spectrum kit (Sigma–Aldrich, St. Louis, MO), and on-column DNase digestion (Qiagen) was used to remove genomic DNA. Following reverse transcription with the SuperScriptII Reverse Transcription kit (Life Technologies, CA), newly synthesized cDNA was used as a template for amplification of the gene fragments using previously designed primers for seven pathway genes Chs, Chi, F3h, F3′h, F3′5′h, Dfr, and Ans and a reference “housekeeping” gene, the elongation factor EF1-α. The qPCR mix contained 10 µl of SYBR green dye (Finnzymes, Espoo, Finland), 0.8 µl primers, 0.4 µl reference dye, and 0.8 µl cDNA (2 ng), and 8 µl H2O. The qPCR program proceeded with 10 min at 95°C, then 40–50 cycles of 95°C for 20 s followed by 55°C for 60 s, ending with a dissociation stage, at 72°C for 10 min. Three replicate reactions were run for each individual, and the resulting amplification curves were imported into R for analysis (R Core Team 2017). From the log-linear phase of these amplification curves, we obtained the parameters of the relative expression ratio model (Pfaffl 2001), that is, the crossing points and amplification efficiency (Peirson et al. 2003). We normalized expression of each of the pathway genes by the expression of the reference gene, ef1-α (following Pfaffl 2001) and calculated log10 fold-change in gene expression relative to a single control individual of the blue-flowered I. cyaneum. This blue color is similar to other blue-flowered Iochroma species and is thought to be the ancestral state in the clade, and we used the same individual of I. cyaneum as a reference in previous gene expression studies (Smith and Rausher 2011). In the comparison of any individual’s expression to this control individual, a fold-change of 0 means equal expression in sample and control, whereas positive or negative fold-change values indicate higher and lower expression relative to the control, respectively. Each unit of fold-change represents an order-of-magnitude change in the initial concentration of gene transcripts in the sample, again relative to the control. Stochastic Mapping of Pigment Evolution In order to understand evolutionary shifts in floral pigmentation, we reconstructed the evolution of pigment production in Iochrominae using maximum likelihood with ace {ape} (Paradis et al. 2004) and stochastic mapping (SM) using make.simmap {phytools} (Revell 2012) in R version 3.4.2 (R Core Team 2017). SM is a Bayesian approach that simulates possible histories (“maps”) of discrete character evolution on a phylogeny based on a model of character evolution (Huelsenbeck et al. 2003). Here, we employed an all-rates different model, where transition rates among all states are allowed to vary independently. The simulations can be summarized to give the posterior probability of each state at each node, as well as posterior distributions of the frequency of events, that is, the number of transitions to and from each character state across the tree. Using HPLC pigment profiling, we scored taxa as based on their primary pigment type (delphinidin, cyanidin, pelargonidin, or none), giving us four possible states. Based on our maximum likelihood reconstruction (supplementary fig. S2, Supplementary Material online) in addition to biochemical and phylogenetic analyses at the family level (Ng and Smith 2016), we fixed the root state to delphinidin for SM and completed 1,000 simulations of character evolution. For these and subsequent analyses, we used the previously published, time-calibrated, maximum clade credibility tree for Iochrominae (Smith and Goldberg 2015). Comparative Analyses of Gene Expression We used phylogenetic comparative methods to test the relationship between levels of gene expression and pigment type. We removed I. ellipticum from the data set because we only have one individual. For each of the seven pathway genes, we implemented a phylogenetic ANOVA using gls {nlme} (Pinheiro et al. 2014). For this analysis, we grouped together all species based on which pigments they make (delphinidin, pelargonidin, and none) in order to test how gene expression varies by pigment type. With only one species (I. edule) producing cyanidin as the primary pigment, we excluded this species. In order to account for phylogenetic nonindependence, we incorporated the correlation structure based on the phylogeny, with the strength of signal modeled with Pagel’s λ (Pagel 1999) using corPagel {ape}. This correlation structure, with λ estimated from the data, produced a better fit than Brownian motion or the Ornstein–Uhlenbeck model (results not shown). Next, we examined patterns of coexpression across pathway genes in the context of phenotypes using multivariate comparative methods. We implemented phylogenetic PCA (pPCA) because standard PCA in comparative settings can mislead inferences (Uyeda et al. 2015). We used the function phyl.pca {phytools} (Revell 2009), again with Pagel’s λ to account for the strength of phylogenetic signal. To estimate pairwise gene expression relationships, we extracted the phylogenetic correlation matrix using code from the phyl.pca function. These pairwise correlations are equivalent to phylogenetic generalized least squares (or PGLS) using Pagel’s λ. We then computed P values with the Bonferonni correction for multiple testing, using corr.p {psych}. Given that species with the same pigment type might be expected to group in gene expression space, we detected clusters using k-means clustering (supplementary fig. S3, Supplementary Material online), assuming two to four clusters with the {stats} package in R (R Core Team 2017). This method aims to minimize within-cluster variance while maximizing differences among clusters for a given number of clusters. Acknowledgments This work was supported by the National Science Foundation (grant DEB1355518 to S.D.S.). We thank the Servicio Nacional Forestal y de Fauna Silvestre (SERFOR) of Peru for permission to carry out investigation under permits N° 0259-2016 and N° 041-2017 and the Servicio Nacional de Áreas Naturales Protegidas (SERNANP) under permit N° 004-2013. We are also grateful to the Direccion Provincial del Ambiente Pichincha, Ecuador for permission to collect under permit N° 05-2012-IC-FAU-DPAP-MA. We thank members of the Smith lab for constructive feedback and helping to improve the article. References Aharoni A , De Vos CHR , Wein M , Sun Z , Greco R , Kroon A , Mol JNM , O’Connell AP. 2001 . The strawberry FaMYB1 transcription factor suppresses anthocyanin and flavonol accumulation in transgenic tobacco . Plant J . 28 3 : 319 – 332 . Google Scholar CrossRef Search ADS PubMed Albert NW , Davies KM , Lewis DH , Zhang HB , Montefiori M , Brendolise C , Boase MR , Ngo H , Jameson PE , Schwinn KE. 2014 . A conserved network of transcriptional activators and repressors regulates anthocyanin pigmentation in eudicots . Plant Cell 26 3 : 962 – 980 . Google Scholar CrossRef Search ADS PubMed Berardi AE , Hildreth SB , Helm RF , Winkel BSJ , Smith SD. 2016 . Evolutionary correlations in flavonoid production across flowers and leaves in the Iochrominae (Solanaceae) . Phytochemistry 130 : 119 – 127 . Google Scholar CrossRef Search ADS PubMed Brodie ED III , Brodie ED Jr. 2015 . Predictably convergent evolution of sodium channels in the arms race between predators and prey . Brain Behav Evol . 86 1 : 48 – 57 . Google Scholar CrossRef Search ADS PubMed Burda S , Oleszek W. 2001 . Antioxidant and antiradical activities of flavonoids . J Agric Food Chem . 49 6 : 2774 – 2779 . Google Scholar CrossRef Search ADS PubMed Butelli E , Titta L , Giorgio M , Mock H-P , Matros A , Peterek S , Schijlen EGWM , Hall RD , Bovy AG , Luo J , Martin C. 2008 . Enrichment of tomato fruit with health-promoting anthocyanins by expression of select transcription factors . Nat Biotechnol . 26 11 : 1301 – 1308 . Google Scholar CrossRef Search ADS PubMed Chan YF , Marks ME , Jones FC , Villarreal G , Shapiro MD , Brady SD , Southwick AM , Absher DM , Grimwood J , Schmutz J , et al. . 2010 . Adaptive evolution of pelvic reduction in sticklebacks by recurrent deletion of a Pitx1 enhancer . Science 327 5963 : 302 – 305 . Google Scholar CrossRef Search ADS PubMed Christin PA , Weinreich DM , Besnard G. 2010 . Causes and evolutionary significance of genetic convergence . Trends Genet . 26 9 : 400 – 405 . Google Scholar CrossRef Search ADS PubMed Coberly LC , Rausher MD. 2003 . Analysis of a chalcone synthase mutant in Ipomoea purpurea reveals a novel function for flavonoids: amelioration of heat stress . Mol Ecol . 12 5 : 1113 – 1124 . Google Scholar CrossRef Search ADS PubMed Coberly LC , Rausher MD. 2008 . Pleiotropic effects of an allele producing white flowers in Ipomoea purpurea . Evolution 62 5 : 1076 – 1085 . Google Scholar CrossRef Search ADS PubMed Coburn RA , Griffin RH , Smith SD. 2015 . Genetic basis for a rare floral mutant in an andean species of solanaceae . Am J Bot . 102 2 : 264 – 272 . Google Scholar CrossRef Search ADS PubMed Conte GL , Arnegard ME , Peichel CL , Schluter D. 2012 . The probability of genetic parallelism and convergence in natural populations . Proc R Soc B Biol Sci . 279 1749 : 5039 – 5047 . Google Scholar CrossRef Search ADS Cooley AM , Modliszewski JL , Rommel ML , Willis JH. 2011 . Gene duplication in mimulus underlies parallel floral evolution via independent trans-regulatory changes . Curr Biol . 21 8 : 700 – 704 . Google Scholar CrossRef Search ADS PubMed De Vetten N , Ter Horst J , Van Schaik HPP , de Boer A , Mol J , Koes R. 1999 . A cytochrome b5 is required for full activity of flavonoid 3’,5’-hydroxylase, a cytochrome P450 involved in the formation of blue flower colors . Proc Natl Acad Sci U S A . 96 2 : 778 – 783 . Google Scholar CrossRef Search ADS PubMed Des Marais DL , Rausher MD. 2010 . Parallel evolution at multiple levels in the origin of hummingbird pollinated flowers in Ipomoea . Evolution 64 7 : 2044 – 2054 . Google Scholar PubMed Dick CA , Buenrostro J , Butler T , Carlson ML , Kliebenstein DJ , Whittall JB. 2011 . Arctic mustard flower color polymorphism controlled by petal-specific downregulation at the threshold of the anthocyanin biosynthetic pathway . PLoS One 6 4 : e18230. Google Scholar CrossRef Search ADS PubMed Gates DJ , Olson BJSC , Clemente TE , Smith SD. 2018 . A novel R3 MYB transcriptional repressor associated with the loss of floral pigmentation in Iochroma . New Phytol . 217 3 : 1346 – 1356 . Google Scholar CrossRef Search ADS PubMed Gates DJ , Strickler SR , Mueller LA , Olson BJSC , Smith SD. 2016 . Diversification of R2R3-MYB transcription factors in the tomato family Solanaceae . J Mol Evol . 83 ( 1–2 ): 26 – 37 . Google Scholar CrossRef Search ADS PubMed Gompel N , Prud’homme B. 2009 . The causes of repeated genetic evolution . Dev Biol . 332 1 : 36 – 47 . Google Scholar CrossRef Search ADS PubMed Gould SJ. 1990 . Wonderful life: the Burgess Shale and the nature of history. New York (NY ): WW Norton and Company . Grotewold E. 2006 . The science of flavonoids. New York (NY ): Springer Science and Business Media . Habu Y , Hisatomi Y , Iida S. 1998 . Molecular characterization of the mutableflakedallele for flower variegation in the common morning glory . Plant J . 16 3 : 371 – 376 . Google Scholar CrossRef Search ADS PubMed Harborne JB. 1998 . Phytochemical methods: a guide to modern techniques of plant analysis . 3rd ed . London (United Kingdom ): Chapman and Hall . Hirano R , Sasamoto W , Matsumoto A , Itakura H , Igarashi O , Kondo K. 2001 . Antioxidant ability of various flavonoids against DPPH radicals and LDL oxidation . J Nutr Sci Vitaminol (Tokyo) 47 5 : 357 – 362 . Google Scholar CrossRef Search ADS PubMed Ho WW , Smith SD. 2016 . Molecular evolution of anthocyanin pigmentation genes following losses of flower color . BMC Evol Biol . 16 : 98. Google Scholar CrossRef Search ADS PubMed Hoballah ME , Gübitz T , Stuurman J , Broger L , Barone M , Mandel T , Dell’Olivo A , Arnold M , Kuhlemeier C. 2007 . Single gene-mediated shift in pollinator attraction in Petunia . Plant Cell 19 3 : 779 – 790 . Google Scholar CrossRef Search ADS PubMed Holton TA , Cornish EC. 1995 . Genetics and biochemistry of anthocyanin biosynthesis . Plant Cell 7 7 : 1071. Google Scholar CrossRef Search ADS PubMed Hopkins R , Rausher MD. 2011 . Identification of two genes causing reinforcement in the Texas wildflower Phlox drummondii . Nature 469 7330 : 411 – 414 . Google Scholar CrossRef Search ADS PubMed Huelsenbeck JP , Nielsen R , Bollback JP. 2003 . Stochastic mapping of morphological characters . Syst Biol . 52 2 : 131 – 158 . Google Scholar CrossRef Search ADS PubMed Koes RE , Quattrocchio F , Mol JNM. 1994 . The flavonoid biosynthetic pathway in plants: function and evolution . BioEssays 16 2 : 123 – 132 . Google Scholar CrossRef Search ADS Liu J , Osbourn A , Ma P. 2015 . MYB transcription factors as regulators of phenylpropanoid metabolism in plants . Mol Plant 8 5 : 689 – 708 . Google Scholar CrossRef Search ADS PubMed Martin A , Orgogozo V. 2013 . The loci of repeated evolution: a catalog of genetic hotspots of phenotypic variation . Evolution 67 5 : 1235 – 1250 . Google Scholar PubMed Martin C , Prescott A , Mackay S , Bartlett J , Vrijlandt E. 1991 . Control of anthocyanin biosynthesis in flowers of Antirrhinum-Majus . Plant J . 1 1 : 37 – 49 . Google Scholar CrossRef Search ADS PubMed Mo Y , Nagel C , Taylor LP. 1992 . Biochemical complementation of chalcone synthase mutants defines a role for flavonols in functional pollen . Proc Natl Acad Sci U S A. 89 15 : 7213 – 7217 . Google Scholar CrossRef Search ADS PubMed Mol J , Grotewold E , Koes R. 1998 . How genes paint flowers and seeds . Trends Plant Sci . 3 6 : 212 – 217 . Google Scholar CrossRef Search ADS Nakatsuka T , Nishihara M , Mishiba K , Hirano H , Yamamura S. 2006 . Two different transposable elements inserted in flavonoid 3′, 5′-hydroxylase gene contribute to pink flower coloration in Gentiana scabra . Mol Genet Genomics 275 3 : 231 – 241 . Google Scholar CrossRef Search ADS PubMed Nery MF , Borges B , Dragalzew AC , Kohlsdorf T. 2016 . Selection on different genes with equivalent functions: the convergence story told by Hox genes along the evolution of aquatic mammalian lineages . BMC Evol Biol . 16 1 : 113. Google Scholar CrossRef Search ADS PubMed Ng J , Smith SD. 2016 . Widespread flower color convergence in Solanaceae via alternate biochemical pathways . New Phytol . 209 1 : 407 – 417 . Google Scholar CrossRef Search ADS PubMed Orgogozo V. 2015 . Replaying the tape of life in the twenty-first century . Interface Focus 5 6 : 20150057. Google Scholar CrossRef Search ADS PubMed Pagel M. 1999 . Inferring the historical patterns of biological evolution . Nature 401 : 877 – 884 . Google Scholar CrossRef Search ADS PubMed Paradis E , Claude J , Strimmer K. 2004 . APE: analyses of phylogenetics and evolution in R language . Bioinformatics 20 2 : 289 – 290 . Google Scholar CrossRef Search ADS PubMed Peirson SN , Butler JN , Foster RG. 2003 . Experimental validation of novel and conventional approaches to quantitative real-time PCR data analysis . Nucleic Acids Res . 31 14 : e73. Google Scholar CrossRef Search ADS PubMed Pelletier MK , Winkel-Shirley B. 1996 . Analysis of flavanone 3-hydroxylase in Arabidopsis seedlings . Plant Physiol . 111 1 : 339 – 345 . Google Scholar CrossRef Search ADS PubMed Pfaffl MW. 2001 . A new mathematical model for relative quantification in real-time RT-PCR . Nucleic Acids Res . 29 9 : 45e – 421 . Google Scholar CrossRef Search ADS Pinheiro J , Bates D , DebRoy S , Sarkar D. 2014 . R Core Team (2014) nlme: linear and nonlinear mixed effects models . R Package Version 3 : 1 – 117 . Pollak PE , Vogt T , Mo Y , Taylor LP. 1993 . Chalcone synthase and flavonol accumulation in stigmas and anthers of Petunia hybrida . Plant Physiol. 102 3 : 925 – 932 . Google Scholar CrossRef Search ADS PubMed Protas ME , Hersey C , Kochanek D , Zhou Y , Wilkens H , Jeffery WR , Zon LI , Borowsky R , Tabin CJ. 2006 . Genetic analysis of cavefish reveals molecular convergence in the evolution of albinism . 38 : 107 – 111 . Quattrocchio F. 1993 . Regulatory genes controlling anthocyanin pigmentation are functionally conserved among plant species and have distinct sets of target genes. Plant Cell Online 5 11 : 1497 – 1512 . Google Scholar CrossRef Search ADS Quattrocchio F , Baudry A , Lepiniec L , Grotewold E. 2006 . The regulation of flavonoid biosynthesis. In: Grotewold E , editor. The science of flavonoids . New York : Springer New York . p. 97 – 122 . R Core Team. 2017. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. Available from: https://www.r-project.org/; last accessed June 11, 2018. Ramsay NA , Glover BJ. 2005 . MYB-bHLH-WD40 protein complex and the evolution of cellular diversity . Trends Plant Sci . 10 2 : 63 – 70 . Google Scholar CrossRef Search ADS PubMed Rausher MD. 2006 . The evolution of flavonoids and their genes. In: Grotewold E , editor. The science of flavonoids . New York : Springer New York . p. 97 – 122 . Rausher MD. 2008 . Evolutionary transitions in floral color . Int J Plant Sci . 169 1 : 7 – 21 . Google Scholar CrossRef Search ADS Revell LJ. 2009 . Size-correction and principal components for interspecific comparative studies . Evolution 63 12 : 3258 – 3268 . Google Scholar CrossRef Search ADS PubMed Revell LJ. 2012 . phytools: an R package for phylogenetic comparative biology (and other things) . Methods Ecol Evol . 3 2 : 217 – 223 . Google Scholar CrossRef Search ADS Rosenblum EB , Parent CE , Brandt EE. 2014 . The molecular basis of phenotypic convergence . Annu Rev Ecol Evol Syst . 45 1 : 203 – 226 . Google Scholar CrossRef Search ADS Ryan KG , Swinny EE , Markham KR , Winefield C. 2002 . Flavonoid gene expression and UV photoprotection in transgenic and mutant Petunia leaves . Phytochemistry 59 1 : 23 – 32 . Google Scholar CrossRef Search ADS PubMed Ryan KG , Swinny EE , Winefield C , Markham KR. 2001 . Flavonoids and UV photoprotection in Arabidopsis mutants . Zeitschrift Für Naturforsch C 56 ( 9–10 ): 745 – 754 . Google Scholar CrossRef Search ADS Schemske DW , Bradshaw HD. 1999 . Pollinator preference and the evolution of floral traits in monkeyflowers (Mimulus) . Proc Natl Acad Sci U S A . 96 21 : 11910 – 11915 . Google Scholar CrossRef Search ADS PubMed Schwinn K , Venail J , Shang Y , Mackay S , Alm V , Butelli E , Oyama R , Bailey P , Davies K , Martin C. 2006 . A small family of MYB-regulatory genes controls floral pigmentation intensity and patterning in the genus Antirrhinum . Plant Cell 18 4 : 831 – 851 . Google Scholar CrossRef Search ADS PubMed Sheehan H , Moser M , Klahre U , Esfeld K , Dell’Olivo A , Mandel T , Metzger S , Vandenbussche M , Freitas L , Kuhlemeier C. 2016 . MYB-FL controls gain and loss of floral UV absorbance, a key trait affecting pollinator preference and reproductive isolation . Nat Genet . 48 2 : 159 – 166 . Google Scholar CrossRef Search ADS PubMed Smith SD , Baum DA. 2006 . Phylogenetics of the florally diverse Andean clade Iochrominae (Solanaceae) . Am J Bot . 93 8 : 1140 – 1153 . Google Scholar CrossRef Search ADS PubMed Smith SD , Goldberg EE. 2015 . Tempo and mode of flower color evolution . Am J Bot . 102 7 : 1014 – 1025 . Google Scholar CrossRef Search ADS PubMed Smith SD , Rausher MD. 2011 . Gene loss and parallel evolution contribute to species difference in flower color . Mol Biol Evol . 28 10 : 2799 – 2810 . Google Scholar CrossRef Search ADS PubMed Sobel JM , Streisfeld MA. 2013 . Flower color as a model system for studies of plant evo-devo . Front Plant Sci . 4 : 1 – 17 . Google Scholar CrossRef Search ADS PubMed Steiner CC , Römpler H , Boettger LM , Schöneberg T , Hoekstra HE. 2009 . The genetic basis of phenotypic convergence in beach mice: similar pigment patterns but different genes . Mol Biol Evol . 26 1 : 35 – 45 . Google Scholar CrossRef Search ADS PubMed Stern DL , Orgogozo V. 2008 . The loci of evolution: how predictable is genetic evolution? Evolution 62 9 : 2155 – 2177 . Google Scholar CrossRef Search ADS PubMed Stern DL , Orgogozo V. 2009 . Is genetic evolution predictable? Science 323 5915 : 746 – 751 . Google Scholar CrossRef Search ADS PubMed Stracke R , Ishihara H , Huep G , Barsch A , Mehrtens F , Niehaus K , Weisshaar B. 2007 . Differential regulation of closely related R2R3-MYB transcription factors controls flavonol accumulation in different parts of the Arabidopsis thaliana seedling . Plant J . 50 4 : 660 – 677 . Google Scholar CrossRef Search ADS PubMed Streisfeld MA , Rausher MD. 2009 . Genetic changes contributing to the parallel evolution of red floral pigmentation among Ipomoea species . New Phytol . 183 3 : 751 – 763 . Google Scholar CrossRef Search ADS PubMed Streisfeld MA , Rausher MD. 2011 . Population genetics, pleiotropy, and the preferential fixation of mutations during adaptive evolution . Evolution 65 3 : 629 – 642 . Google Scholar CrossRef Search ADS PubMed Streisfeld MA , Young WN , Sobel JM. 2013 . Divergent selection drives genetic differentiation in an R2R3-MYB transcription factor that contributes to incipient speciation in Mimulus aurantiacus . PLoS Genet . 9 3 : e1003385. Google Scholar CrossRef Search ADS PubMed Takahashi R , Dubouzet JG , Matsumura H , Yasuda K , Iwashina T. 2010 . A new allele of flower color gene W1 encoding flavonoid 3’5’-hydroxylase is responsible for light purple flowers in wild soybean Glycine soja . BMC Plant Biol . 10 : 155. Google Scholar CrossRef Search ADS PubMed Uyeda JC , Caetano DS , Pennell MW . 2015 . Comparative Analysis of Principal Components Can be Misleading. Syst. Biol. 64 : 677 – 689 . Google Scholar CrossRef Search ADS PubMed Watkins J , Chapman JM , Muday GK. 2017 . Abscisic acid-induced reactive oxygen species are modulated by flavonols to control stomata aperture . Plant Physiol . 175 4 : 1807. Google Scholar CrossRef Search ADS PubMed Wessinger CA , Rausher MD. 2012 . Lessons from flower colour evolution on targets of selection . J Exp Bot . 63 16 :5741–5709. Wessinger CA , Rausher MD. 2015 . Ecological transition predictably associated with gene degeneration . Mol Biol Evol . 32 2 : 347 – 354 . Google Scholar CrossRef Search ADS PubMed Whittall JB , Voelckel C , Kliebenstein DJ , Hodges SA. 2006 . Convergence, constraint and the role of gene expression during adaptive radiation: floral anthocyanins in Aquilegia . Mol Ecol . 15 14 : 4645 – 4657 . Google Scholar CrossRef Search ADS PubMed Winkel-Shirley B. 2001 . Flavonoid biosynthesis. A colorful model for genetics, biochemistry, cell biology, and biotechnology . Plant Physiol . 126 2 : 485 – 493 . Google Scholar CrossRef Search ADS PubMed Yuan YW , Sagawa JM , Young RC , Christensen BJ , Bradshaw HD . 2013 . Genetic dissection of a major anthocyanin QTL contributing to pollinator-mediated reproductive isolation between sister species of Mimulus. Genetics 194 : 255 – 263 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Molecular Biology and Evolution Oxford University Press

Convergent Evolution at the Pathway Level: Predictable Regulatory Changes during Flower Color Transitions

Loading next page...
 
/lp/ou_press/convergent-evolution-at-the-pathway-level-predictable-regulatory-eaHwPYnc0g
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
0737-4038
eISSN
1537-1719
D.O.I.
10.1093/molbev/msy117
Publisher site
See Article on Publisher Site

Abstract

Abstract The predictability of evolution, or whether lineages repeatedly follow the same evolutionary trajectories during phenotypic convergence remains an open question of evolutionary biology. In this study, we investigate evolutionary convergence at the biochemical pathway level and test the predictability of evolution using floral anthocyanin pigmentation, a trait with a well-understood genetic and regulatory basis. We reconstructed the evolution of floral anthocyanin content across 28 species of the Andean clade Iochrominae (Solanaceae) and investigated how shifts in pigmentation are related to changes in expression of seven key anthocyanin pathway genes. We used phylogenetic multivariate analysis of gene expression to test for phenotypic and developmental convergence at a macroevolutionary scale. Our results show that the four independent losses of the ancestral pigment delphinidin involved convergent losses of expression of the three late pathway genes (F3′5′h, Dfr, and Ans). Transitions between pigment types affecting floral hue (e.g., blue to red) involve changes to the expression of branching genes F3′h and F3′5′h, while the expression levels of early steps of the pathway are strongly conserved in all species. These patterns support the idea that the macroevolution of floral pigmentation follows predictable evolutionary trajectories to reach convergent phenotype space, repeatedly involving regulatory changes. This is likely driven by constraints at the pathway level, such as pleiotropy and regulatory structure. anthocyanin pathway, gene expression, floral pigment evolution Introduction The prevalence of convergent evolution, that is, the emergence of similar forms in different lineages, suggests that evolution may often be repeatable, with similar outcomes resulting from “replaying the tape of life” (Gould 1990; Orgogozo 2015). Nonetheless, the extent to which these convergent phenotypes arise via similar underlying mechanisms remains an open question (Stern and Orgogozo 2009; Rosenblum et al. 2014). In some cases, convergent phenotypes evolve through distinct genetic or developmental changes (Steiner et al. 2009; Nery et al. 2016), while in others, similar mechanisms are involved (Chan et al. 2010; Martin and Orgogozo 2013). The predictable recruitment of the same mechanisms may be explained by genetic constraints, for example if only one or a few loci have the potential to influence the phenotype, for example, in tetrodotoxin resistance (Brodie and Brodie 2015) and albinism of cavefish (Protas et al. 2006). By contrast, when multiple paths can lead to the same phenotype, the repeated involvement of particular loci or types of changes suggest biases, either in mutation rates or fixation probability (Stern and Orgogozo 2008; Gompel and Prud’homme 2009; Christin et al. 2010; Streisfeld and Rausher 2011; Conte et al. 2012). Still, the ability to assess such biases is limited for many traits where the underlying developmental pathway and the potential routes to convergent phenotypes are largely unknown. Floral anthocyanin pigmentation provides a powerful system for studying phenotypic convergence and the predictability of evolution at the molecular and developmental levels. The anthocyanin branch of the large flavonoid biosynthetic pathway (hereafter the anthocyanin pathway) produces three classes of red, purple, and blue pigments (fig. 1) which comprise the most common pigments determining color in flowers and fruits (Grotewold 2006). Small changes in the anthocyanin pathway (i.e., mutations at a single locus) can radically alter floral hue and/or color intensity (Koes et al. 1994; Mol et al. 1998; Rausher 2006,, 2008), with important ecological consequences, for example, pollinator attraction (Schemske and Bradshaw 1999; Hoballah et al. 2007; Sheehan et al. 2016). Upstream steps of this pathway (enzymes CHS, CHI, F3H; fig. 1) are shared across multiple flavonoid compounds (e.g., flavones and flavonols). The subsequent steps involve the so-called branching enzymes F3′H and F3′5′H, which determine the type of anthocyanin produced, and the downstream enzymes DFR and ANS, which convert the colorless precursors into pigments. Although the entire flavonoid pathway contains dozens of genes, this group of core enzymes (fig. 1) are largely responsible for controlling metabolic flux toward anthocyanin pigments, and together determine the amount and type of anthocyanins produced (Grotewold 2006). This pathway is regulated at the level of transcription, and the regulatory genes are well known (Quattrocchio et al. 2006). In most flowering plants, the upstream steps are jointly regulated by MYB transcription factors while the downstream steps are activated by a complex of MYB, bHLH, and WD40 (MBW) transcription factors (Mol et al. 1998; Stracke et al. 2007). The regulation of the branching genes varies across species and is generally not as well understood as for the structural genes (De Vetten et al. 1999; Schwinn et al. 2006; Albert et al. 2014). Fig. 1. View largeDownload slide Simplified view of the flavonoid biosynthesis pathway adapted from Rausher (2006). Arrows indicate biochemical reactions catalyzed by enzymes (white boxes). Intermediate metabolites are enclosed in gray boxes. Abbreviations as follows: LCP, leucopelargonidin; LCC, leucocyanidin; LCD, leudodelphinidin; CHS, chalcone synthase; CHI, chalcone isomerase; F3H, flavanone 3-hydroxylase; F3′H, flavonoid 3′-hydroxylase; F3′5H, flavonoid 3′,5′-hydroxylase; DFR, dihydroflavonol 4-reductase; ANS, anthocyanidin synthase. The methylated cyanidin derivative peonidin and methylated delphinidin derivatives malvidin and petunidin not shown. The dashed arrow indicates that F3′5′H does not have 3′ hydroxylation activity in Iochrominae (Smith and Rausher 2011). Fig. 1. View largeDownload slide Simplified view of the flavonoid biosynthesis pathway adapted from Rausher (2006). Arrows indicate biochemical reactions catalyzed by enzymes (white boxes). Intermediate metabolites are enclosed in gray boxes. Abbreviations as follows: LCP, leucopelargonidin; LCC, leucocyanidin; LCD, leudodelphinidin; CHS, chalcone synthase; CHI, chalcone isomerase; F3H, flavanone 3-hydroxylase; F3′H, flavonoid 3′-hydroxylase; F3′5H, flavonoid 3′,5′-hydroxylase; DFR, dihydroflavonol 4-reductase; ANS, anthocyanidin synthase. The methylated cyanidin derivative peonidin and methylated delphinidin derivatives malvidin and petunidin not shown. The dashed arrow indicates that F3′5′H does not have 3′ hydroxylation activity in Iochrominae (Smith and Rausher 2011). In the context of the anthocyanin biosynthetic pathway, evolutionary transitions in flower color appear to occur via a small and predictable subset of possible mechanisms. For example, although losses of pigmentation could occur by inactivation along any branch of the pathway, convergent evolutionary transitions to white flowers have been exclusively linked to changes in the MYB genes that control the downstream elements of the pathway (reviewed in Sobel and Streisfeld 2013). The fixation of these regulatory mutations preserves the potential for expressing flavonoids in other tissues, where they function, for example, in protection from UV light (Ryan et al. 2001), signaling (Watkins et al. 2017), and pollen fertility (Mo et al. 1992). Similarly, repeatable patterns of genetic evolution have been described for the evolution of floral hue, where transitions from blue to red flowers frequently often involve genetic and developmental changes inferred to incur minimal pleiotropic effects (Wessinger and Rausher 2012). Nonetheless, these patterns of apparent predictability in the mechanisms of flower color evolution rely on a relatively small number of studies (e.g., five in the case of fixed gains of floral anthocyanins; Sobel and Streisfeld 2013). Here, we use the Iochrominae clade (Solanaceae) to trace the evolution of anthocyanin gene expression during repeated shifts among flower colors and quantify the extent of convergence in pathway regulation. This group of around 40 species, native to South America, displays a remarkable variety of flower colors, largely due to variation in anthocyanin pigmentation. Previous studies have shown that the ancestral flower color in the clade is blue with mainly delphinidin-based anthocyanins (Smith and Baum 2006; Ho and Smith 2016). Multiple lineages have lost these pigments to shift to white or yellow flowers (Ho and Smith 2016), and two lineages have evolved orange to red flowers with cyanidin or pelargonidin pigments (Berardi et al. 2016). In two cases, the genetic basis for these color transitions has been identified. In red-flowered Iochroma gesnerioides, the shift to pelargonidin pigments occurred via the transregulatory loss of F3′h expression, and the inactivation of F3′5′h through gene deletion (Smith and Rausher 2011). In the white flowered I. loxense, the loss of pigment production involved an R3 MYB repressor that blocks expression of multiple downstream genes (Gates et al. 2018). Although the genetic basis is unknown for the remaining color transitions in Iochrominae, several lines of evidence point to an important role of pathway regulation. First, a study of molecular evolution of three core pathway genes (the upstream Chi and F3h and the anthocyanin specific Ans), revealing consistent purifying selection across the clade, including lineages that have lost floral anthocyanins (Ho and Smith 2016). Second, flavonoids, including anthocyanins, are abundant in vegetative tissue (Berardi et al. 2016) as well as fruits (Smith SD, unpublished data), which is consistent with the broad conservation of pathway function. Finally, phylogenetic analyses point to at least one regain of floral anthocyanins following loss, suggesting functionality of the structural pathway genes was maintained, allowing renewed floral expression of the pathway (Ho and Smith 2016). In order to test the prediction that convergent flower color shifts in Iochrominae have repeatedly involved similar regulatory changes, we combined a suite of biochemical, developmental, and phylogenetic approaches. First, we profiled anthocyanin pigment production across 28 species, and estimated the evolutionary history of anthocyanin pigmentation in this clade. Next, we quantified the expression of seven pathway genes and used statistical comparative methods to investigate the relationship between gene expression and pigment phenotypes. Given the likely pleiotropic effects of changes to the early genes of the pathway on other flavonoids, we predicted that the shifts from blue to white flowers involve losses of expression of downstream elements of the pathway. By contrast, we expected that transitions among the blue, purple, and red pigments involve differential regulation of the branching enzymes. We further tested these predictions using multivariate approaches to cluster the species in gene expression space. Together our analyses reveal a tight relationship between floral expression of pathway genes, anthocyanin pigment production, and flower color evolution in Iochrominae. Results Evolutionary Shifts in Pigment Production Consistent with previous work in this system, species within Iochrominae produce all three classes of anthocyanins, and the types present are closely associated with the flower color phenotype. The blue-flowered species produce delphinidin-derived anthocyanins (fig. 2). While some blue-flowered Iochrominae make these pigments exclusively, most also synthesize small amounts (<15%) of cyanidin-derived anthocyanins (supplementary fig. S1, Supplementary Material online). The two red-flowered sister species, I. gesnerioides and I. fuchsioides produce the red anthocyanidin pelargonidin, and the orange-red flowered I. edule is the only species which produces exclusively cyanidin (supplementary fig. S1, Supplementary Material online). As predicted, six white or yellow flowered species produce no detectable anthocyanidins (supplementary fig. S1, Supplementary Material online). Fig. 2. View largeDownload slide Evolution of main floral anthocyanidin pigments in Iochrominae. (A) Phylogeny of Iochrominae, with ancestral state reconstruction of anthocyanidins from stochastic mapping, with root state fixed at delphinidin. Pies at nodes represent frequencies of node states across 1,000 simulations of character evolution. Disks at tips are colored by species main anthocyanidin pigment for each species 5 (supplementary fig. S1, Supplementary Material online). (B) Transitions between character states, summarized over 1,000 stochastic histories, with arrow width proportional to average number of transitions (see supplementary table S2, Supplementary Material online). Crosses indicate nodes with posterior probability <0.95. Fig. 2. View largeDownload slide Evolution of main floral anthocyanidin pigments in Iochrominae. (A) Phylogeny of Iochrominae, with ancestral state reconstruction of anthocyanidins from stochastic mapping, with root state fixed at delphinidin. Pies at nodes represent frequencies of node states across 1,000 simulations of character evolution. Disks at tips are colored by species main anthocyanidin pigment for each species 5 (supplementary fig. S1, Supplementary Material online). (B) Transitions between character states, summarized over 1,000 stochastic histories, with arrow width proportional to average number of transitions (see supplementary table S2, Supplementary Material online). Crosses indicate nodes with posterior probability <0.95. Analyses of trait evolution with stochastic mapping supported strong directionality in transitions among pigment types (fig. 2), similar to previous work in this and other clades (reviewed in Rausher 2008). The most common transitions were inferred to be convergent losses of ancestral delphinidin production, with stochastic mapping estimating an average of five losses across the clade (fig. 2 and supplementary table S3, Supplementary Material online). The persistence of just four unpigmented lineages suggests that one of these transitions was reversed. The earliest loss likely occurred c. 3 Ma on the branch leading to the “A” clade containing the unpigmented species I. ellipticum, I. salpoanum, I. confertiflorum, Acnistus arborescens (fig. 2A). Additional more recent shifts involved I. lehmannii, I. loxense, and D. solanacea. Stochastic mapping estimated that transitions from unpigmented to producing cyanidin occurred at least twice in the clade’s history (fig. 2B), although only one extant species, I. edule, produces this pigment exclusively. Its position, nested within the unpigmented “A” clade, suggests that it has recently regained floral anthocyanin expression. Shifts to floral pelargonidin production are inferred to be rare, likely occurring only once in the common ancestor of the two red-flowered species. Several transitions (e.g., pelargonidin to cyanidin, delphinidin to cyanidin) were never observed among the stochastic maps (fig. 2 and supplementary table S3, Supplementary Material online), indicating they are unlikely to have occurred during the history of Iochrominae. Evolution of Anthocyanin Pathway Gene Expression Our comparative analysis of gene expression revealed distinct evolutionary patterns across pathway genes. Expression levels of the upstream genes (e.g., Chs, Chi, and F3h) were similar across Iochrominae regardless of pigment composition, with log fold-change values mostly between −1 and +1, that is, transcript concentrations within an order of magnitude of the reference blue-flowered I. cyaneum (fig. 3A). There were a few exceptions; for example, Chs had low expression (log fold-change of −3) in the white-flowered I. ellipticum and F3h showed moderate down-regulation (log fold-changes of around −1.5) in several species which lack floral anthocyanins (I. ellipticum, I. confertiflorum, and D. solanacea; fig. 3A). The phylogenetic ANOVA suggested that despite these cases, species without floral anthocyanins did not have significantly lower expression for Chs, Chi, or F3h than their pigmented relatives (fig. 3B). This pattern of conservation of upstream gene expression in lineages that have lost floral anthocyanins is consistent with their continued production of other floral flavonoids, in particular flavonols (supplementary fig. S1, Supplementary Material online). These compounds are typically as abundant in flower tissues as anthocyanins and remain abundant in most white- and yellow-flowered species (Berardi et al. 2016; Larter M, Berardi A, and Smith SD, unpublished data). Fig. 3. View largeDownload slide Anthocyanin pathway gene expression in Iochrominae. (A) Gene expression for each species, with bars corresponding to each tip of the phylogeny. Colors of bars and dots at tips represent main floral anthocyanidins (supplementary fig. S2, Supplementary Material online). Bracket represents the “A clade” sensu Smith and Baum (2006). (B) Log-fold-change gene expression aggregated by floral anthocyanidin production. Pink circles represent I. edule, that makes cyanidin. The letters represent significant differences under a pairwise Tukey comparison test (alpha=0.05). Fig. 3. View largeDownload slide Anthocyanin pathway gene expression in Iochrominae. (A) Gene expression for each species, with bars corresponding to each tip of the phylogeny. Colors of bars and dots at tips represent main floral anthocyanidins (supplementary fig. S2, Supplementary Material online). Bracket represents the “A clade” sensu Smith and Baum (2006). (B) Log-fold-change gene expression aggregated by floral anthocyanidin production. Pink circles represent I. edule, that makes cyanidin. The letters represent significant differences under a pairwise Tukey comparison test (alpha=0.05). In contrast with the evolutionary conservation of expression observed for the upstream genes, the other elements of the pathway showed strong color-specific patterns of expression. In the two red-flowered species, which produce only pelargonidin, we observed extremely low expression of the branching genes F3′h and F3′5′h (with log fold change of −3 or lower, or c. 1,000 times fewer transcripts than in the reference blue species), while the remaining steps of the pathway had comparable expression levels to blue-flowered species (fig. 3B). In the unpigmented lineages, the two downstream genes Dfr and Ans showed similarly strong down-regulation (log fold change of around −3), along with the branching gene F3′5′h (fig. 3B). The retention of F3′h expression in these lineages is likely tied to the prevalence of quercetin, a flavonol that requires F3′H activity (fig. 1), in their floral tissue (supplementary fig. S1, Supplementary Material online). The cyanidin-producing species I. edule, which is nested within the unpigmented A clade, has recovered expression of F3′5′h, Dfr, and Ans, although we were unable to assess statistical significance with only a single tip (fig. 3A and B). Coexpression Patterns of the Anthocyanin Pathway Genes Given that anthocyanin pathway genes are known to be jointly regulated by MBW complexes, we used multivariate approaches to quantify coexpression and relate these patterns to floral pigmentation variation. The phylogenetic PCA (pPCA) of expression levels grouped species by primary pigment types, and thus flower color (fig. 4A). The first two PC axes explained ∼90% of the variance (fig. 4A). The first PC axis divides the pigmented and unpigmented species (fig. 4A), with high loadings of expression levels of F3′5′h, Dfr, and Ans (fig.4B). The second PC axis separates the red pelargonidin-producing species from the rest of Iochrominae, with high loadings of expression of branching genes F3′h and F3′5′h. The two centermost points, slightly detached from the blue-flowered cluster, are the red-orange I. edule and purple-spotted S. punctata. The separation of I. edule might be expected as it is the only species producing primarily cyanidin and has high F3′h expression but relatively low F3′5′h expression. S. punctata produces primarily delphinidin, but expression levels for pathway genes are generally lower than the other blue and purple taxa, perhaps due to the spotted phenotype. Fig. 4. View largeDownload slide Phylogenetic PCA of anthocyanin pathway gene expression. (A) Species scores on PC1 and PC2, with symbols detailed in box below. Symbols are filled based on pigment production as follows: red for pelargonidin, pink for cyanidin, white for delphinidin, and yellow for no anthocyanidins. Symbol outline colors indicate k-means clustering with k = 3 (as in text): black, red, and blue outlines. (B) Arrows show loadings for each variable on the PC axes, scale shown on bottom and right axes. (C) Pairwise gene expression correlation table. Circle size and color scale represent magnitude of the correlation coefficients, and stars within circles show significance level (P < 0.05, 0.01, and 0.001 for one, two, and three stars, respectively). Fig. 4. View largeDownload slide Phylogenetic PCA of anthocyanin pathway gene expression. (A) Species scores on PC1 and PC2, with symbols detailed in box below. Symbols are filled based on pigment production as follows: red for pelargonidin, pink for cyanidin, white for delphinidin, and yellow for no anthocyanidins. Symbol outline colors indicate k-means clustering with k = 3 (as in text): black, red, and blue outlines. (B) Arrows show loadings for each variable on the PC axes, scale shown on bottom and right axes. (C) Pairwise gene expression correlation table. Circle size and color scale represent magnitude of the correlation coefficients, and stars within circles show significance level (P < 0.05, 0.01, and 0.001 for one, two, and three stars, respectively). The pattern of grouping in gene expression space by pigment type was confirmed by k-means clustering. With k = 3, we recovered the same three clusters: the species that 1) make no anthocyanins, 2) those that make pelargonidin, and 3) those that make either delphinidin- or cyanidin-based anthocyanins. This latter cluster is broken when k = 4, but not into groups clearly reflecting different pigmentation phenotypes. Both subclusters contain blue-flowered species predominantly producing delphinidin, however they are slightly differentiated along PC1 of the pPCA of gene expression (supplementary fig. S3, Supplementary Material online). In addition to identifying phenotypic clusters of species, the pPCA highlights the high levels of correlated expression among groups of pathway genes (fig. 4C). The upstream genes Chs and Chi are significantly coexpressed with F3h across the phylogeny (r2 = 0.55 and 0.36; P < 0.001 and 0.02, respectively; supplementary table S4, Supplementary Material online), and these genes show no significant correlations with the remaining genes. A similar pattern is exhibited by F3′5′h and the downstream genes Dfr and Ans (r2 = 0.49–0.87, P < 0.001). Intriguingly, the expression of the branching gene F3′h shows no significant correlations with the expression of any of the other pathway genes. This result aligns with previous crossing studies indicating an independent (unidentified) transregulatory factor for this branch of the pathway in Iochroma (Smith and Rausher 2011). Discussion Regulatory Mechanisms Drive Evolution of Floral Pigmentation Intensity Our results show that convergent losses of floral anthocyanin production are driven at the pathway level by convergence in the expression of pathway genes in Iochrominae. Ancestral delphinidin production has been lost independently in four white- and yellow-flowered lineages, and in each case these transitions are associated with 1,000-fold decrease in expression of the branching gene F3′5′h and the downstream genes Dfr and Ans. These repeatable, coordinated expression shifts at the macroevolutionary level are striking given the wide range of mechanisms known to generate unpigmented flowers in natural populations and horticultural varieties (Streisfeld and Rausher 2011). For example, both regulatory and structural mutations at the Chs loci have been found to underlie white morphs of some normally pigmented species (Habu et al. 1998; Coberly and Rausher 2003; Dick et al. 2011). This contrast between the range of mechanisms that can give rise to the derived phenotype and those that underlie species level transitions suggests bias toward particular genetic paths, possibly due to mutational target size or pleiotropy (Gompel and Prud’homme 2009). Given that flavonoids play important roles in physiological processes throughout the plant (Winkel-Shirley 2001), pleiotropic effects of mutations that change flower color are particularly likely to affect their probability of fixation over short- and long-evolutionary timescales (Coberly and Rausher 2008). Although we have not identified the exact genetic changes that caused the coordinated down-regulation of F3′5′h, Dfr, and Ans in the unpigmented lineages, several lines of evidence point to a transregulatory mechanism. First, the tight coexpression of these three genes across the phylogeny suggests that they belong to a single regulatory module. Indeed, Dfr and Ans are known to be jointly regulated by an MBW transcription factor complex in a wide range of taxa (Holton and Cornish 1995; Mol et al. 1998). It appears that F3′5′h belongs to this regulatory module in Iochrominae (Gates et al. 2018) as it does many but not all eudicots (De Vetten et al. 1999; Butelli et al. 2008; Albert et al. 2014). Furthermore, losses of pigmentation in other lineages also tend to be driven by transregulatory mutations (reviewed in Wessinger and Rausher 2012; Sobel and Streisfeld 2013). These often involve the MBW complex (e.g., Schwinn et al. 2006; Hoballah et al. 2007) and lead to loss of expression in the downstream genes (Whittall et al. 2006; Cooley et al. 2011). Among the transregulators of the downstream genes, the MYB genes have emerged as the dominant players in flower pigment evolution. Indeed, changes in these transcription factors have been found to underlie every case of evolutionary gain or loss of anthocyanins in which the genetic basis has been dissected (reviewed in Sobel and Streisfeld 2013). This pattern may relate to the fact that MYBs are often highly tissue-specific, and determine the transcriptional targets for the MBW complex (Ramsay and Glover 2005). Thus either cis- or coding mutations at these loci will have limited pleiotropic effects, relative to mutations in other elements of the complex or in the pathway itself (Streisfeld and Rausher 2011). In addition to this activating role as part of the MBW complex, MYB proteins can act at transcriptional repressors of the anthocyanin pathway (Aharoni et al. 2001; Albert et al. 2014; Liu et al. 2015). Most transitions to white or yellow flowers involve loss-of-function mutations in the MYB activators (Sobel and Streisfeld 2013) although in Iochroma loxense, a gain-of-function in an MYB repressor was responsible (Gates et al. 2018). Given the similarity we observe in expression profiles, with strongest downregulation in F3′5′h, Dfr, and Ans, changes at this same MYB repressor may underlie the losses of floral pigmentation in other Iochrominae lineages. Investigating this possibility will require a multifaceted approach as the rapid evolution of MYB genes and their immense copy number (over 100 in Iochroma; Gates et al. 2016) complicates standard candidate gene experiments. Understanding the mechanisms responsible for the repeated losses of floral anthocyanins in Iochrominae has important implications for potential regains of pigmentation. Given that the structure of the pathway appears conserved (Ho and Smith 2016) and remains active in nonfloral tissues (Berardi et al. 2016; Smith SD, unpublished data), an evolutionary gain of floral anthocyanins likely requires only reactivation of the pathway in floral tissue. This could occur via downregulation of a repressor (Yuan et al. 2013) or upregulation of an activator (Streisfeld et al. 2013). Indeed, this may be the case with I. edule, which has regained floral anthocyanins following a loss roughly 5 Ma. Compared with its white and yellow relatives in the A clade, this species has elevated F3′5′h, Dfr, and Ans expression, a change that could be achieved by their shared regulator. Nonetheless, I. edule did not regain the ancestral blue delphinidin pigmentation but instead produces purple-pink cyanidin. This unique feature may represent incomplete reversal, given that the expression of F3′5′h (along with Dfr and Ans) is still lower than all of the blue-flowered species (fig. 3B). Alternately, additional changes in the pathway following the loss may have altered the consequences for reactivation of the F3′5′h-Dfr-Ans module. For example, DFR may have shifted its specificity to the less hydroxylated DHQ as cyanidin is more common in vegetative tissue (Wessinger and Rausher 2012). The functionality of F3′5′h may have also degraded since its products (myricetin and delphinidin) do not appear in vegetative tissues of A-clade species (Berardi et al. 2016; supplementary fig. S1, Supplementary Material online). Genetic Targets of Shifts in Floral Hue In contrast to shifts in the intensity of pigmentation, shifts in floral hue due to transitions between anthocyanin pigments are driven by changes to the branching genes F3′h and F3′5′h. The relative activity of these enzymes determines the abundance of the pigment precursors (the three dihydroflavonols, fig. 1) and in turn, the flux toward the three classes of pigments. Thus, shifts in F3′h and F3′5′h expression can alter which branch of the anthocyanin pathway is most active thereby changing the main floral anthocyanin pigment (Rausher 2008). Previous work in Iochroma demonstrated that the loss of both F3′H and F3′5′H activity is required for the complete transition from the ancestral delphinidin production to the derived pelargonidin production (Smith and Rausher 2011). In the red-flowered I. gesnerioides, this was achieved by flower-specific loss of F3′h expression due to a transregulatory change along with degeneration of the F3′5′h gene. Here, we find that the downregulation of F3′h is also found in the red-flowered sister species I. fuchsioides. Both species produce trace amounts of cyanidin in flowers and leaves and substantial amounts of quercetin in leaves, indicating that F3′H remains functional (Berardi et al. 2016). Additional studies will be required to determine whether I. fuchsioides, like I. gesnerioides, lacks a functional F3′5h gene. Even if I. fuchsioides has retained F3′5′h, its expression is essentially undetectable in flowers (fig. 3A). In this case, the initial loss of expression of F3′5′h in the common ancestor of I. fuchsioides and I. gesnerioides may have set the stage for the later degeneration of the coding sequence in one species (I. gesnerioides). Although the two red-flowered Iochroma are the result of a single evolutionary transition, the patterns we observe mirror those found in shifts in hue in other taxa. For example, shifts from delphinidin to cyanidin or pelargonidin production consistently involve coding and/or regulatory changes at F3′5′h (Nakatsuka et al. 2006; Takahashi et al. 2010; Hopkins and Rausher 2011; Wessinger and Rausher 2015) and exclusively regulatory changes at F3′h (Streisfeld and Rausher 2009; Des Marais and Rausher 2010). Given these patterns, Wessinger and Rausher (2012,, 2015) posited that F3′5′h is a “disposable” gene while F3′h is constrained due to its functions beyond flower color. Indeed, the flavonol quercetin, which requires F3′H activity has higher antioxidant and antiradical activity (Burda and Oleszek 2001; Hirano et al. 2001) and provides better UV protection (Ryan et al. 2001, 2002) than myricetin, which requires F3′5′H activity. Thus, the pleiotropic cost of losing F3′h function is likely higher than that for F3′5′h (Streisfeld and Rausher 2011). Macroevolutionary Signatures of Regulatory Structure Given the complex patterns of pigment production across tissues and in response to environmental cues, the anthocyanin pathway has long served as a model for studying the spatial and temporal regulation of metabolic pathways. Early studies of anthocyanin regulation revealed that steps in the pathway are often coordinately controlled by shared transcription factors, but that these relationships can differ across taxa. For instance, the anthocyanin biosynthetic genes are divided into two coexpressed clusters (“early” and “late” genes) in Eudicots, but which genes belong to those clusters varies across taxa (e.g., Martin et al. 1991; Quattrocchio 1993; Pelletier and Winkel-Shirley 1996). The discovery of these regulatory relationships has and continues to rely on detailed studies of transcription factor and structural gene expression in experimental model systems (e.g., Quattrocchio 1993; Butelli et al. 2008). Our analyses suggest that comparative studies of gene expression at the pathway level may provide a complementary approach to detecting coregulated modules. The phylogenetic PCA recovered the strong coexpression of F3′5′h, Dfr, and Ans, a group that corresponds to “late” genes in the model system Petunia (Mol et al. 1998; Albert et al. 2014). The traditional “early” genes in Petunia (Chs, Chi, F3h; Quattrocchio 1993) emerge here as a loose cluster, likely reflecting the redundant regulation of these steps by multiple MYB transcription factors (Quattrocchio et al. 2006). Finally, the expression of F3′h, which is controlled by a separate transregulator (Smith and Rausher 2011), is decoupled from other parts of the pathway. The close correspondence between the correlation structure of expression variation in these genes and the regulatory structure known from plant model systems likely reflects the predominance of transregulatory mutations in flower color evolution. If indeed the fluctuation in pigment intensity across the clade arises from changes in shared MYB transcription factors, multiple genes will show highly correlated variation at a phylogenetic scale. Equally, gene that experience independent control will lack such correlated variation. Conclusions and Future Directions Evolutionary constraints at the pathway level restrict the available trajectories between floral pigmentation phenotypes, leading to strong predictability of the mechanisms involved in phenotypic transitions. The predictable involvement of regulatory changes in branching and downstream steps of the pathway likely arises from selection for maintaining the production of fitness-related flavonoid compounds in other tissues. At the scale of an entire clade, our study supports key features of flower color evolution that have emerged from a growing number of studies in model systems. In particular, we show that evolutionary losses of anthocyanins predictably involve reversible regulatory shut-down of the late pathway genes. Moreover, changes of floral hue require reduced activity of the branching genes, although the underlying mechanism varies due to pleiotropic and regulatory constraints. One explanation for the repeatability of evolution is that while many changes can produce the same phenotypic outcome (e.g., a coding change to an early pathway gene inducing loss of floral pigments), some are filtered out through selection within populations, due to their negative pleiotropic effects (Gompel and Prud’homme 2009). Rare unpigmented morphs are found in multiple blue-flowered Iochrominae species, with previous research showing that in one case, this intraspecific variation is due to a loss-of-function mutation to the late gene Dfr (Coburn et al. 2015). Further work in several polymorphic species could shed some light on the fixation advantage of minimally pleiotropic mutations at the population level explaining the bias in mechanisms visible at longer evolutionary timescales (Stern and Orgogozo 2008; Gompel and Prud’homme 2009). Materials and Methods Study System Iochrominae are distributed across Western South America, from Columbia and Peru to Brazil and northern Argentina. Flower buds were collected from natural populations for 19 of the species. The remaining nine species were grown from seed at the UC-Boulder EBIO Greenhouses. Greenhouse plants were placed outside two to three weeks prior to flower bud collection to provide natural light (without UV filtering) to maximize the stimulation of the anthocyanin pathway. We obtained samples from at least three individuals per species, with the exception of I. ellipticum from the Galapagos Islands, for which we only have one individual, and Vassobia breviflora for which we only have two individuals. Vouchers and accession numbers are available in supplementary table S1, Supplementary Material online. Anthocyanin Identification and Quantification We separated and quantified anthocyanidins (aglycones of anthocyanins) and closely related flavonoids (flavones, flavonols) from floral tissue using High Performance Liquid Chromatography (HPLC). Following Berardi et al. (2016), we extracted anthocyanidins and other flavonoids from silica-dried floral tissue using the acid hydrolysis procedure of Harborne (1998) with minor modifications. Dried extracts were resuspended in methanol with 1% HCl (v/v), and 10 μl were injected and analyzed using an Agilent Infinity 1260 Infinity Series HPLC system with a 100 × 4.6 mm Chromolith Performance RP-18 endcapped column (EMD Chemicals, Darmstadt, Germany) and a photodiode array detector (PDA; scanning from 200 to 600 nm at a step of 2 nm). Separation of the compounds followed the protocols of Berardi et al. (2016). Peaks were detected at 520 nm for anthocyanidins and 365 nm for flavones and flavonols and identified by comparison of retention times and UV-VIS absorbance spectra with commercial standards (Extrasynthese, Genay, France). Chromatograms were visualized using Agilent ChemStation for LC 3 D Systems (Rev. B0403, Waldbronn, Germany). Concentrations were scaled by tissue weight and were estimated by comparing the area of each chromatogram peak to those from dilution series of the corresponding standards. Two of the anthocyanidins (cyanidin and delphinidin) have methylated derivatives, and these were summed for each to obtain the total amount of cyanidin- and delphinidin-derived anthocyanins, that is, matching each branch of the pathway (fig. 1). Relative amounts of the three classes (pelargonidin-, cyanidin-, and delphinidin-derived anthocyanins) were calculated to identify the principal pigment and minor pigments in each species. HPLC data for nine species in this data set were previously published in Berardi et al. (2016). Quantification of Anthocyanin Pathway Gene Expression Floral buds were collected at a stage roughly equivalent to Petunia bud Stage 5 (Pollak et al. 1993), at which time both the early and late parts of the pathway are expressed (Smith and Rausher 2011). Tissue was preserved in RNAlater (Qiagen, Valencia, CA), or flash frozen if collected from a greenhouse plant. The protocol for measuring gene expression followed previous work (Coburn et al. 2015). Briefly, total RNA was extracted using the Spectrum kit (Sigma–Aldrich, St. Louis, MO), and on-column DNase digestion (Qiagen) was used to remove genomic DNA. Following reverse transcription with the SuperScriptII Reverse Transcription kit (Life Technologies, CA), newly synthesized cDNA was used as a template for amplification of the gene fragments using previously designed primers for seven pathway genes Chs, Chi, F3h, F3′h, F3′5′h, Dfr, and Ans and a reference “housekeeping” gene, the elongation factor EF1-α. The qPCR mix contained 10 µl of SYBR green dye (Finnzymes, Espoo, Finland), 0.8 µl primers, 0.4 µl reference dye, and 0.8 µl cDNA (2 ng), and 8 µl H2O. The qPCR program proceeded with 10 min at 95°C, then 40–50 cycles of 95°C for 20 s followed by 55°C for 60 s, ending with a dissociation stage, at 72°C for 10 min. Three replicate reactions were run for each individual, and the resulting amplification curves were imported into R for analysis (R Core Team 2017). From the log-linear phase of these amplification curves, we obtained the parameters of the relative expression ratio model (Pfaffl 2001), that is, the crossing points and amplification efficiency (Peirson et al. 2003). We normalized expression of each of the pathway genes by the expression of the reference gene, ef1-α (following Pfaffl 2001) and calculated log10 fold-change in gene expression relative to a single control individual of the blue-flowered I. cyaneum. This blue color is similar to other blue-flowered Iochroma species and is thought to be the ancestral state in the clade, and we used the same individual of I. cyaneum as a reference in previous gene expression studies (Smith and Rausher 2011). In the comparison of any individual’s expression to this control individual, a fold-change of 0 means equal expression in sample and control, whereas positive or negative fold-change values indicate higher and lower expression relative to the control, respectively. Each unit of fold-change represents an order-of-magnitude change in the initial concentration of gene transcripts in the sample, again relative to the control. Stochastic Mapping of Pigment Evolution In order to understand evolutionary shifts in floral pigmentation, we reconstructed the evolution of pigment production in Iochrominae using maximum likelihood with ace {ape} (Paradis et al. 2004) and stochastic mapping (SM) using make.simmap {phytools} (Revell 2012) in R version 3.4.2 (R Core Team 2017). SM is a Bayesian approach that simulates possible histories (“maps”) of discrete character evolution on a phylogeny based on a model of character evolution (Huelsenbeck et al. 2003). Here, we employed an all-rates different model, where transition rates among all states are allowed to vary independently. The simulations can be summarized to give the posterior probability of each state at each node, as well as posterior distributions of the frequency of events, that is, the number of transitions to and from each character state across the tree. Using HPLC pigment profiling, we scored taxa as based on their primary pigment type (delphinidin, cyanidin, pelargonidin, or none), giving us four possible states. Based on our maximum likelihood reconstruction (supplementary fig. S2, Supplementary Material online) in addition to biochemical and phylogenetic analyses at the family level (Ng and Smith 2016), we fixed the root state to delphinidin for SM and completed 1,000 simulations of character evolution. For these and subsequent analyses, we used the previously published, time-calibrated, maximum clade credibility tree for Iochrominae (Smith and Goldberg 2015). Comparative Analyses of Gene Expression We used phylogenetic comparative methods to test the relationship between levels of gene expression and pigment type. We removed I. ellipticum from the data set because we only have one individual. For each of the seven pathway genes, we implemented a phylogenetic ANOVA using gls {nlme} (Pinheiro et al. 2014). For this analysis, we grouped together all species based on which pigments they make (delphinidin, pelargonidin, and none) in order to test how gene expression varies by pigment type. With only one species (I. edule) producing cyanidin as the primary pigment, we excluded this species. In order to account for phylogenetic nonindependence, we incorporated the correlation structure based on the phylogeny, with the strength of signal modeled with Pagel’s λ (Pagel 1999) using corPagel {ape}. This correlation structure, with λ estimated from the data, produced a better fit than Brownian motion or the Ornstein–Uhlenbeck model (results not shown). Next, we examined patterns of coexpression across pathway genes in the context of phenotypes using multivariate comparative methods. We implemented phylogenetic PCA (pPCA) because standard PCA in comparative settings can mislead inferences (Uyeda et al. 2015). We used the function phyl.pca {phytools} (Revell 2009), again with Pagel’s λ to account for the strength of phylogenetic signal. To estimate pairwise gene expression relationships, we extracted the phylogenetic correlation matrix using code from the phyl.pca function. These pairwise correlations are equivalent to phylogenetic generalized least squares (or PGLS) using Pagel’s λ. We then computed P values with the Bonferonni correction for multiple testing, using corr.p {psych}. Given that species with the same pigment type might be expected to group in gene expression space, we detected clusters using k-means clustering (supplementary fig. S3, Supplementary Material online), assuming two to four clusters with the {stats} package in R (R Core Team 2017). This method aims to minimize within-cluster variance while maximizing differences among clusters for a given number of clusters. Acknowledgments This work was supported by the National Science Foundation (grant DEB1355518 to S.D.S.). We thank the Servicio Nacional Forestal y de Fauna Silvestre (SERFOR) of Peru for permission to carry out investigation under permits N° 0259-2016 and N° 041-2017 and the Servicio Nacional de Áreas Naturales Protegidas (SERNANP) under permit N° 004-2013. We are also grateful to the Direccion Provincial del Ambiente Pichincha, Ecuador for permission to collect under permit N° 05-2012-IC-FAU-DPAP-MA. We thank members of the Smith lab for constructive feedback and helping to improve the article. References Aharoni A , De Vos CHR , Wein M , Sun Z , Greco R , Kroon A , Mol JNM , O’Connell AP. 2001 . The strawberry FaMYB1 transcription factor suppresses anthocyanin and flavonol accumulation in transgenic tobacco . Plant J . 28 3 : 319 – 332 . Google Scholar CrossRef Search ADS PubMed Albert NW , Davies KM , Lewis DH , Zhang HB , Montefiori M , Brendolise C , Boase MR , Ngo H , Jameson PE , Schwinn KE. 2014 . A conserved network of transcriptional activators and repressors regulates anthocyanin pigmentation in eudicots . Plant Cell 26 3 : 962 – 980 . Google Scholar CrossRef Search ADS PubMed Berardi AE , Hildreth SB , Helm RF , Winkel BSJ , Smith SD. 2016 . Evolutionary correlations in flavonoid production across flowers and leaves in the Iochrominae (Solanaceae) . Phytochemistry 130 : 119 – 127 . Google Scholar CrossRef Search ADS PubMed Brodie ED III , Brodie ED Jr. 2015 . Predictably convergent evolution of sodium channels in the arms race between predators and prey . Brain Behav Evol . 86 1 : 48 – 57 . Google Scholar CrossRef Search ADS PubMed Burda S , Oleszek W. 2001 . Antioxidant and antiradical activities of flavonoids . J Agric Food Chem . 49 6 : 2774 – 2779 . Google Scholar CrossRef Search ADS PubMed Butelli E , Titta L , Giorgio M , Mock H-P , Matros A , Peterek S , Schijlen EGWM , Hall RD , Bovy AG , Luo J , Martin C. 2008 . Enrichment of tomato fruit with health-promoting anthocyanins by expression of select transcription factors . Nat Biotechnol . 26 11 : 1301 – 1308 . Google Scholar CrossRef Search ADS PubMed Chan YF , Marks ME , Jones FC , Villarreal G , Shapiro MD , Brady SD , Southwick AM , Absher DM , Grimwood J , Schmutz J , et al. . 2010 . Adaptive evolution of pelvic reduction in sticklebacks by recurrent deletion of a Pitx1 enhancer . Science 327 5963 : 302 – 305 . Google Scholar CrossRef Search ADS PubMed Christin PA , Weinreich DM , Besnard G. 2010 . Causes and evolutionary significance of genetic convergence . Trends Genet . 26 9 : 400 – 405 . Google Scholar CrossRef Search ADS PubMed Coberly LC , Rausher MD. 2003 . Analysis of a chalcone synthase mutant in Ipomoea purpurea reveals a novel function for flavonoids: amelioration of heat stress . Mol Ecol . 12 5 : 1113 – 1124 . Google Scholar CrossRef Search ADS PubMed Coberly LC , Rausher MD. 2008 . Pleiotropic effects of an allele producing white flowers in Ipomoea purpurea . Evolution 62 5 : 1076 – 1085 . Google Scholar CrossRef Search ADS PubMed Coburn RA , Griffin RH , Smith SD. 2015 . Genetic basis for a rare floral mutant in an andean species of solanaceae . Am J Bot . 102 2 : 264 – 272 . Google Scholar CrossRef Search ADS PubMed Conte GL , Arnegard ME , Peichel CL , Schluter D. 2012 . The probability of genetic parallelism and convergence in natural populations . Proc R Soc B Biol Sci . 279 1749 : 5039 – 5047 . Google Scholar CrossRef Search ADS Cooley AM , Modliszewski JL , Rommel ML , Willis JH. 2011 . Gene duplication in mimulus underlies parallel floral evolution via independent trans-regulatory changes . Curr Biol . 21 8 : 700 – 704 . Google Scholar CrossRef Search ADS PubMed De Vetten N , Ter Horst J , Van Schaik HPP , de Boer A , Mol J , Koes R. 1999 . A cytochrome b5 is required for full activity of flavonoid 3’,5’-hydroxylase, a cytochrome P450 involved in the formation of blue flower colors . Proc Natl Acad Sci U S A . 96 2 : 778 – 783 . Google Scholar CrossRef Search ADS PubMed Des Marais DL , Rausher MD. 2010 . Parallel evolution at multiple levels in the origin of hummingbird pollinated flowers in Ipomoea . Evolution 64 7 : 2044 – 2054 . Google Scholar PubMed Dick CA , Buenrostro J , Butler T , Carlson ML , Kliebenstein DJ , Whittall JB. 2011 . Arctic mustard flower color polymorphism controlled by petal-specific downregulation at the threshold of the anthocyanin biosynthetic pathway . PLoS One 6 4 : e18230. Google Scholar CrossRef Search ADS PubMed Gates DJ , Olson BJSC , Clemente TE , Smith SD. 2018 . A novel R3 MYB transcriptional repressor associated with the loss of floral pigmentation in Iochroma . New Phytol . 217 3 : 1346 – 1356 . Google Scholar CrossRef Search ADS PubMed Gates DJ , Strickler SR , Mueller LA , Olson BJSC , Smith SD. 2016 . Diversification of R2R3-MYB transcription factors in the tomato family Solanaceae . J Mol Evol . 83 ( 1–2 ): 26 – 37 . Google Scholar CrossRef Search ADS PubMed Gompel N , Prud’homme B. 2009 . The causes of repeated genetic evolution . Dev Biol . 332 1 : 36 – 47 . Google Scholar CrossRef Search ADS PubMed Gould SJ. 1990 . Wonderful life: the Burgess Shale and the nature of history. New York (NY ): WW Norton and Company . Grotewold E. 2006 . The science of flavonoids. New York (NY ): Springer Science and Business Media . Habu Y , Hisatomi Y , Iida S. 1998 . Molecular characterization of the mutableflakedallele for flower variegation in the common morning glory . Plant J . 16 3 : 371 – 376 . Google Scholar CrossRef Search ADS PubMed Harborne JB. 1998 . Phytochemical methods: a guide to modern techniques of plant analysis . 3rd ed . London (United Kingdom ): Chapman and Hall . Hirano R , Sasamoto W , Matsumoto A , Itakura H , Igarashi O , Kondo K. 2001 . Antioxidant ability of various flavonoids against DPPH radicals and LDL oxidation . J Nutr Sci Vitaminol (Tokyo) 47 5 : 357 – 362 . Google Scholar CrossRef Search ADS PubMed Ho WW , Smith SD. 2016 . Molecular evolution of anthocyanin pigmentation genes following losses of flower color . BMC Evol Biol . 16 : 98. Google Scholar CrossRef Search ADS PubMed Hoballah ME , Gübitz T , Stuurman J , Broger L , Barone M , Mandel T , Dell’Olivo A , Arnold M , Kuhlemeier C. 2007 . Single gene-mediated shift in pollinator attraction in Petunia . Plant Cell 19 3 : 779 – 790 . Google Scholar CrossRef Search ADS PubMed Holton TA , Cornish EC. 1995 . Genetics and biochemistry of anthocyanin biosynthesis . Plant Cell 7 7 : 1071. Google Scholar CrossRef Search ADS PubMed Hopkins R , Rausher MD. 2011 . Identification of two genes causing reinforcement in the Texas wildflower Phlox drummondii . Nature 469 7330 : 411 – 414 . Google Scholar CrossRef Search ADS PubMed Huelsenbeck JP , Nielsen R , Bollback JP. 2003 . Stochastic mapping of morphological characters . Syst Biol . 52 2 : 131 – 158 . Google Scholar CrossRef Search ADS PubMed Koes RE , Quattrocchio F , Mol JNM. 1994 . The flavonoid biosynthetic pathway in plants: function and evolution . BioEssays 16 2 : 123 – 132 . Google Scholar CrossRef Search ADS Liu J , Osbourn A , Ma P. 2015 . MYB transcription factors as regulators of phenylpropanoid metabolism in plants . Mol Plant 8 5 : 689 – 708 . Google Scholar CrossRef Search ADS PubMed Martin A , Orgogozo V. 2013 . The loci of repeated evolution: a catalog of genetic hotspots of phenotypic variation . Evolution 67 5 : 1235 – 1250 . Google Scholar PubMed Martin C , Prescott A , Mackay S , Bartlett J , Vrijlandt E. 1991 . Control of anthocyanin biosynthesis in flowers of Antirrhinum-Majus . Plant J . 1 1 : 37 – 49 . Google Scholar CrossRef Search ADS PubMed Mo Y , Nagel C , Taylor LP. 1992 . Biochemical complementation of chalcone synthase mutants defines a role for flavonols in functional pollen . Proc Natl Acad Sci U S A. 89 15 : 7213 – 7217 . Google Scholar CrossRef Search ADS PubMed Mol J , Grotewold E , Koes R. 1998 . How genes paint flowers and seeds . Trends Plant Sci . 3 6 : 212 – 217 . Google Scholar CrossRef Search ADS Nakatsuka T , Nishihara M , Mishiba K , Hirano H , Yamamura S. 2006 . Two different transposable elements inserted in flavonoid 3′, 5′-hydroxylase gene contribute to pink flower coloration in Gentiana scabra . Mol Genet Genomics 275 3 : 231 – 241 . Google Scholar CrossRef Search ADS PubMed Nery MF , Borges B , Dragalzew AC , Kohlsdorf T. 2016 . Selection on different genes with equivalent functions: the convergence story told by Hox genes along the evolution of aquatic mammalian lineages . BMC Evol Biol . 16 1 : 113. Google Scholar CrossRef Search ADS PubMed Ng J , Smith SD. 2016 . Widespread flower color convergence in Solanaceae via alternate biochemical pathways . New Phytol . 209 1 : 407 – 417 . Google Scholar CrossRef Search ADS PubMed Orgogozo V. 2015 . Replaying the tape of life in the twenty-first century . Interface Focus 5 6 : 20150057. Google Scholar CrossRef Search ADS PubMed Pagel M. 1999 . Inferring the historical patterns of biological evolution . Nature 401 : 877 – 884 . Google Scholar CrossRef Search ADS PubMed Paradis E , Claude J , Strimmer K. 2004 . APE: analyses of phylogenetics and evolution in R language . Bioinformatics 20 2 : 289 – 290 . Google Scholar CrossRef Search ADS PubMed Peirson SN , Butler JN , Foster RG. 2003 . Experimental validation of novel and conventional approaches to quantitative real-time PCR data analysis . Nucleic Acids Res . 31 14 : e73. Google Scholar CrossRef Search ADS PubMed Pelletier MK , Winkel-Shirley B. 1996 . Analysis of flavanone 3-hydroxylase in Arabidopsis seedlings . Plant Physiol . 111 1 : 339 – 345 . Google Scholar CrossRef Search ADS PubMed Pfaffl MW. 2001 . A new mathematical model for relative quantification in real-time RT-PCR . Nucleic Acids Res . 29 9 : 45e – 421 . Google Scholar CrossRef Search ADS Pinheiro J , Bates D , DebRoy S , Sarkar D. 2014 . R Core Team (2014) nlme: linear and nonlinear mixed effects models . R Package Version 3 : 1 – 117 . Pollak PE , Vogt T , Mo Y , Taylor LP. 1993 . Chalcone synthase and flavonol accumulation in stigmas and anthers of Petunia hybrida . Plant Physiol. 102 3 : 925 – 932 . Google Scholar CrossRef Search ADS PubMed Protas ME , Hersey C , Kochanek D , Zhou Y , Wilkens H , Jeffery WR , Zon LI , Borowsky R , Tabin CJ. 2006 . Genetic analysis of cavefish reveals molecular convergence in the evolution of albinism . 38 : 107 – 111 . Quattrocchio F. 1993 . Regulatory genes controlling anthocyanin pigmentation are functionally conserved among plant species and have distinct sets of target genes. Plant Cell Online 5 11 : 1497 – 1512 . Google Scholar CrossRef Search ADS Quattrocchio F , Baudry A , Lepiniec L , Grotewold E. 2006 . The regulation of flavonoid biosynthesis. In: Grotewold E , editor. The science of flavonoids . New York : Springer New York . p. 97 – 122 . R Core Team. 2017. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. Available from: https://www.r-project.org/; last accessed June 11, 2018. Ramsay NA , Glover BJ. 2005 . MYB-bHLH-WD40 protein complex and the evolution of cellular diversity . Trends Plant Sci . 10 2 : 63 – 70 . Google Scholar CrossRef Search ADS PubMed Rausher MD. 2006 . The evolution of flavonoids and their genes. In: Grotewold E , editor. The science of flavonoids . New York : Springer New York . p. 97 – 122 . Rausher MD. 2008 . Evolutionary transitions in floral color . Int J Plant Sci . 169 1 : 7 – 21 . Google Scholar CrossRef Search ADS Revell LJ. 2009 . Size-correction and principal components for interspecific comparative studies . Evolution 63 12 : 3258 – 3268 . Google Scholar CrossRef Search ADS PubMed Revell LJ. 2012 . phytools: an R package for phylogenetic comparative biology (and other things) . Methods Ecol Evol . 3 2 : 217 – 223 . Google Scholar CrossRef Search ADS Rosenblum EB , Parent CE , Brandt EE. 2014 . The molecular basis of phenotypic convergence . Annu Rev Ecol Evol Syst . 45 1 : 203 – 226 . Google Scholar CrossRef Search ADS Ryan KG , Swinny EE , Markham KR , Winefield C. 2002 . Flavonoid gene expression and UV photoprotection in transgenic and mutant Petunia leaves . Phytochemistry 59 1 : 23 – 32 . Google Scholar CrossRef Search ADS PubMed Ryan KG , Swinny EE , Winefield C , Markham KR. 2001 . Flavonoids and UV photoprotection in Arabidopsis mutants . Zeitschrift Für Naturforsch C 56 ( 9–10 ): 745 – 754 . Google Scholar CrossRef Search ADS Schemske DW , Bradshaw HD. 1999 . Pollinator preference and the evolution of floral traits in monkeyflowers (Mimulus) . Proc Natl Acad Sci U S A . 96 21 : 11910 – 11915 . Google Scholar CrossRef Search ADS PubMed Schwinn K , Venail J , Shang Y , Mackay S , Alm V , Butelli E , Oyama R , Bailey P , Davies K , Martin C. 2006 . A small family of MYB-regulatory genes controls floral pigmentation intensity and patterning in the genus Antirrhinum . Plant Cell 18 4 : 831 – 851 . Google Scholar CrossRef Search ADS PubMed Sheehan H , Moser M , Klahre U , Esfeld K , Dell’Olivo A , Mandel T , Metzger S , Vandenbussche M , Freitas L , Kuhlemeier C. 2016 . MYB-FL controls gain and loss of floral UV absorbance, a key trait affecting pollinator preference and reproductive isolation . Nat Genet . 48 2 : 159 – 166 . Google Scholar CrossRef Search ADS PubMed Smith SD , Baum DA. 2006 . Phylogenetics of the florally diverse Andean clade Iochrominae (Solanaceae) . Am J Bot . 93 8 : 1140 – 1153 . Google Scholar CrossRef Search ADS PubMed Smith SD , Goldberg EE. 2015 . Tempo and mode of flower color evolution . Am J Bot . 102 7 : 1014 – 1025 . Google Scholar CrossRef Search ADS PubMed Smith SD , Rausher MD. 2011 . Gene loss and parallel evolution contribute to species difference in flower color . Mol Biol Evol . 28 10 : 2799 – 2810 . Google Scholar CrossRef Search ADS PubMed Sobel JM , Streisfeld MA. 2013 . Flower color as a model system for studies of plant evo-devo . Front Plant Sci . 4 : 1 – 17 . Google Scholar CrossRef Search ADS PubMed Steiner CC , Römpler H , Boettger LM , Schöneberg T , Hoekstra HE. 2009 . The genetic basis of phenotypic convergence in beach mice: similar pigment patterns but different genes . Mol Biol Evol . 26 1 : 35 – 45 . Google Scholar CrossRef Search ADS PubMed Stern DL , Orgogozo V. 2008 . The loci of evolution: how predictable is genetic evolution? Evolution 62 9 : 2155 – 2177 . Google Scholar CrossRef Search ADS PubMed Stern DL , Orgogozo V. 2009 . Is genetic evolution predictable? Science 323 5915 : 746 – 751 . Google Scholar CrossRef Search ADS PubMed Stracke R , Ishihara H , Huep G , Barsch A , Mehrtens F , Niehaus K , Weisshaar B. 2007 . Differential regulation of closely related R2R3-MYB transcription factors controls flavonol accumulation in different parts of the Arabidopsis thaliana seedling . Plant J . 50 4 : 660 – 677 . Google Scholar CrossRef Search ADS PubMed Streisfeld MA , Rausher MD. 2009 . Genetic changes contributing to the parallel evolution of red floral pigmentation among Ipomoea species . New Phytol . 183 3 : 751 – 763 . Google Scholar CrossRef Search ADS PubMed Streisfeld MA , Rausher MD. 2011 . Population genetics, pleiotropy, and the preferential fixation of mutations during adaptive evolution . Evolution 65 3 : 629 – 642 . Google Scholar CrossRef Search ADS PubMed Streisfeld MA , Young WN , Sobel JM. 2013 . Divergent selection drives genetic differentiation in an R2R3-MYB transcription factor that contributes to incipient speciation in Mimulus aurantiacus . PLoS Genet . 9 3 : e1003385. Google Scholar CrossRef Search ADS PubMed Takahashi R , Dubouzet JG , Matsumura H , Yasuda K , Iwashina T. 2010 . A new allele of flower color gene W1 encoding flavonoid 3’5’-hydroxylase is responsible for light purple flowers in wild soybean Glycine soja . BMC Plant Biol . 10 : 155. Google Scholar CrossRef Search ADS PubMed Uyeda JC , Caetano DS , Pennell MW . 2015 . Comparative Analysis of Principal Components Can be Misleading. Syst. Biol. 64 : 677 – 689 . Google Scholar CrossRef Search ADS PubMed Watkins J , Chapman JM , Muday GK. 2017 . Abscisic acid-induced reactive oxygen species are modulated by flavonols to control stomata aperture . Plant Physiol . 175 4 : 1807. Google Scholar CrossRef Search ADS PubMed Wessinger CA , Rausher MD. 2012 . Lessons from flower colour evolution on targets of selection . J Exp Bot . 63 16 :5741–5709. Wessinger CA , Rausher MD. 2015 . Ecological transition predictably associated with gene degeneration . Mol Biol Evol . 32 2 : 347 – 354 . Google Scholar CrossRef Search ADS PubMed Whittall JB , Voelckel C , Kliebenstein DJ , Hodges SA. 2006 . Convergence, constraint and the role of gene expression during adaptive radiation: floral anthocyanins in Aquilegia . Mol Ecol . 15 14 : 4645 – 4657 . Google Scholar CrossRef Search ADS PubMed Winkel-Shirley B. 2001 . Flavonoid biosynthesis. A colorful model for genetics, biochemistry, cell biology, and biotechnology . Plant Physiol . 126 2 : 485 – 493 . Google Scholar CrossRef Search ADS PubMed Yuan YW , Sagawa JM , Young RC , Christensen BJ , Bradshaw HD . 2013 . Genetic dissection of a major anthocyanin QTL contributing to pollinator-mediated reproductive isolation between sister species of Mimulus. Genetics 194 : 255 – 263 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Molecular Biology and EvolutionOxford University Press

Published: Sep 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off