A distinct foliar pigmentation pattern formed by activator-repressor gradients upstream of an anthocyanin-activating R2R3-MYB

SUMMARY

The emergence of novel traits is often preceded by a potentiation phase, when all the genetic components necessary for producing the trait are assembled. However, elucidating these potentiating factors is challenging. We have previously shown that an anthocyanin-activating R2R3-MYB, STRIPY, triggers the emergence of a distinct foliar pigmentation pattern in the monkeyflower Mimulus verbenaceus. Here, using forward and reverse genetics approaches, we identify three potentiating factors that pattern STRIPY expression: MvHY5, a master regulator of light signaling that activates STRIPY and is expressed throughout the leaf, and two leaf developmental regulators, MvALOG1 and MvTCP5, that are expressed in opposing gradients along the leaf proximodistal axis and negatively regulate STRIPY. These results provide strong empirical evidence that phenotypic novelties can be potentiated through incorporation into preexisting genetic regulatory networks and highlight the importance of positional information in patterning the novel foliar stripe.

In brief

Determining how novel phenotypes originate is a central question in evolution. LaFountain et al. use forward and reverse genetics approaches to identify three leaf developmental regulators that are co-opted to pattern a distinct foliar anthocyanin stripe in the monkeyflower species Mimulus verbenaceus.

Graphical Abstract

INTRODUCTION

The emergence of phenotypic novelties is one of the great enigmas of evolutionary biology.^1–5 It has been proposed that novelties arise through distinct phases, beginning with a series of mutations that set the stage for the new phenotype in a process collectively known as potentiation. Once all the necessary potentiating factors are in place, a final mutation triggers the emergence of the phenotype, known as actualization. The new phenotype may then undergo a period of adaptive refinement in response to new or changing environmental conditions.^6,7 This concept of phased emergence has been beautifully illustrated by the Escherichia coli long-term evolution experiment, wherein one population gained the rare ability to utilize citrate under aerobic growth conditions. In this population, the gain-of-expression of the CitT antiporter gene resulted in rudimentary utilization of citrate present in the growth media, actualizing the so-called Cit⁺ phenotype.⁶ However, the uptake of citrate into the cell by the CitT antiporter requires the concurrent efflux of a solute, and thus would confer only a negligible benefit without the solute supply.⁶ Backcrossing experiments with the ancestral Cit clone revealed that the phenotype was potentiated by mutations that enhanced the intracellular accumulation of C4-dicarboxylates, which were thus already available for exchange with citrate when the actualization CitT mutation occurred.⁸ It is notable that identification of the potentiating factors in this case was largely dependent on the availability of ancestral generations in frozen glycerol stocks for functional interrogation,^8,9 which is infeasible in most multicellular organisms. This underscores one of the many challenges of elucidating the early stages in the emergence of novel traits in natural systems.

A widely accepted hypothesis is that novel phenotypes in multicellular organisms are generated by co-option of existing genes or gene regulatory networks (GRNs).¹⁰ In this model, the preexisting GRNs and the mutations leading to co-option can be considered as the potentiation factors and actualization event, respectively. While numerous studies provide correlational evidence supporting this notion (see, e.g., references 11–16), few have disentangled the potentiation and actualization processes with functional validation of candidate genes. One of the most extensively studied systems is the wing patterns of Heliconius butterflies, wherein optix has been repeatedly associated with the actualization of red pigmentation and WntA was implicated as an important potentiation factor patterning optix expression.17–20 However, even in this well-studied system, how optix became co-opted into the WntA GRN and how WntA expression itself was pre-patterned remain to be elucidated. To better understand these early processes during the evolution of phenotypic novelty, additional systems are required that include recently evolved phenotypes and that are amenable to rigorous experimental analyses.

In a previous study, we reported that a novel medio-lateral stripe phenotype on the leaves of Mimulus verbenaceus (Phrymaceae) was actualized due to gain of localized expression of STRIPY, an anthocyanin-activating R2R3-MYB gene.²¹ Chemical mutagenesis of a striped M. verbenaceus inbred line revealed that multiple upstream activators and repressors position the spatial expression of STRIPY, and thus are presumed to have served as potentiating factors for actualization of the stripe pattern.²¹ In the present work, a combination of reverse and forward genetics experiments is used to identify and characterize three of these potentiating factors—one positive and two negative regulators of STRIPY expression. The upstream activator is a well-known regulatory hub for light-induced development and stress responses,^22,23 while the two negative regulators are associated with boundary formation and marginal growth in lateral organs.^24–27 We suggest that the gain of stripe-specific expression of STRIPY in the M. verbenaceus leaf most likely originated from the rewiring of this anthocyanin-activating R2R3-MYB gene into an ancestral leaf development program that includes these three upstream regulators.

RESULTS

High-light (HL) treatment causes stripe extension in the wild type (MvBL)

The medio-lateral anthocyanin stripe phenotype emerged in a population along the West Fork Trail near Sedona, Arizona, which is where the inbred line MvBL was derived.²¹ We sampled leaves from each node of MvBL plants that had just reached anthesis (8–9 weeks old) in our standard greenhouse conditions, and observed slight variation in the final position of the stripe along the proximodistal axis of mature leaves (Figures 1A and S1A). HL treatment of MvBL plants in a controlled growth chamber experiment resulted in the extension of pigmentation into the proximal section of the leaf (Figures 1B, 1C, and S2A), whereas low-light treatment had the opposite effect (Figures S2B and S2C). qRT-PCR experiments showed that STRIPY expression increased in the HL-treated plants compared to the control plants. By contrast, the other members of the MYB-bHLH-WD40 (MBW) regulatory complex that activates anthocyanin biosynthesis, WD40a and ANbHLH1,²⁸ did not show significant change in expression (Figure 1D). This result suggests that HL treatment specifically upregulates one component of the MBW complex, STRIPY. We also note that HL treatment of MvNB, a non-striped inbred line derived from the same wild population,²¹ did not induce anthocyanin stripe formation (Figure S2A), suggesting that only the MvBL STRIPY allele is responsive to HL treatment.

Figure 1. HL treatment caused stripe extension in the wild-type MvBL.

(A) Representative developmental series of leaves collected from sequential nodes of a plant that has just reached the flowering stage (i.e., first flower reaching anthesis). Nodes are ordered from oldest (base of the plant) to youngest.

(B and C) Photographs of representative control and HL-treated plants (treated for 12 days).

(D) qRT-PCR of control and HL-treated leaves. Errors bars are 1 SD from 4 biological replicates (each replicate is a pair of 15-mm leaves collected on day 28 of the treatment). Asterisk (*) indicates that the difference between the control and HL samples is statistically significant, with a p < 0.05.

Scale bars in (A)–(C), 1 cm.

MvHY5 is an upstream activator of STRIPY

Because HL upregulates STRIPY expression, we hypothesized that the master regulator of light-induced developmental processes, ELONGATED HYPOCOTYL5 (HY5), could be an upstream activator of STRIPY. HY5 has been reported to activate anthocyanin biosynthesis in a light-dependent manner in other plant systems.^29–34 Two paralogs of AtHY5 (At5G11260), denoted MvHY5a and MvHY5b, were identified in the M. verbenaceus genome that share 81% and 82% sequence identity on the nucleotide level and the amino acid (aa) level, respectively (Figures S3A and S3B). Mining previously published RNA sequencing (RNA-seq) data²¹ from three sections (proximal, stripe, distal) of the leaf suggested that neither paralog has preferential expression along the proximodistal axis (Figure 2A). To test whether the MvHY5 paralogs are necessary for STRIPY activation, we generated stable RNAi lines, wherein both MvHY5a and MvHY5b were knocked down and the leaf stripe was markedly reduced with no change in floral pigmentation (Figures 2B, 2C, S3C, and S3D). Accordingly, STRIPY expression was found to be downregulated in the HY5 RNAi lines.

Figure 2. MvHY5 is an upstream activator of STRIPY.

(A) Heatmap of normalized expression values from RNA-seq of the distal, stripe, and proximal sections cut from 18- to 20-mm MvBL leaves (n = 4 biological replicates).

(B) MvHY5 RNAi plant that has just reached anthesis.

(C) qRT-PCR showing relative expression of MvHY5a/b and STRIPY in MvBL and 3 independent MvHY5 RNAi lines.

(D) Series of leaves collected from sequential nodes of an HY5 OE transgenic plant (line 1; hereafter HY5 OE-1) that has just reached anthesis. Nodes are ordered from oldest to youngest.

(E) HY5 OE-1 whole plant.

(F) Heatmap of normalized expression values from RNA-seq for MBW complex components in MvBL and HY5 OE-1 (n = 4 biological replicates). (G) “STRIPY RNAi 3 HY5 OE-1” F1 progeny.

(H) qRT-PCR comparison of relative expression of MvHY5a and STRIPY in MvBL, HY5 OE-1, 3 “STRIPY RNAi 3 HY5 OE-1” F1 progeny (F1), and STRIPY RNAi. Error bars in (C) and (H) represent 1 SD from 3 technical replicates, and double asterisk (**) indicates that the difference between the wild-type MvBL and transgenic line is statistically significant, with a p < 0.01. Scale bars in (B), (D), (E), and (G), 1 cm.

See also Figures S3 and S4.

To further confirm MvHY5 as an upstream activator of STRIPY, we generated transgenic lines overexpressing one of the paralogs, MvHY5a, which were expected to recapitulate the extended stripe phenotype observed in the HL treatment. HY5 is known to be involved in an antagonistic relationship with CONSTITUTIVE PHOTOMORPHOGENIC1 (COP1), an E3 ubiquitin ligase that accumulates in the nucleus in dark or warm conditions^35,36 and triggers the degradation of HY5.³⁷ Therefore, to overcome the potential degradation of HY5, we prepared COP1-resistant overexpression plasmid (35S:MvHY5a(ΔN77)-YFP; hereafter, HY5 OE) by cloning MvHY5a with an N-terminal truncation that eliminates the COP1 recognition site (Figure S3B), as previously described.²⁹ The resulting transgenic lines displayed a significant increase in anthocyanin coverage per leaf area (Figures 2D, 2E, S1, and additional lines in Figures S3E and S3F) and upregulation of STRIPY (Figure 2F), as well as downstream anthocyanin biosynthetic genes F3Ha, DFR, and ANS (Figure S3G).

To test whether the leaf anthocyanin phenotype of HY5 OE is dependent on STRIPY, we crossed a strong HY5 OE line with a STRIPY RNAi line from a previous study.²¹ The resulting F1 plants carrying both transgenes lacked foliar anthocyanins (Figures 2G and 2H), thus confirming that STRIPY is necessary for HY5-mediated anthocyanin activation in M. verbenaceus leaves. Furthermore, dual luciferase and yeast one-hybrid assays suggest that MvHY5a can interact directly with the STRIPY promoter (Figure S4).

MvALOG1 is a repressor of STRIPY in the proximal section of the leaf

A previous chemical mutagenesis experiment in the MvBL genetic background yielded several loss-of-function mutants with alteration of the medio-lateral stripe pattern.²¹ One such mutant line is MV00191, which displays an extended stripe phenotype that is most pronounced in young seedlings and weakens slightly as the plant ages (Figures 3A–3C). The mutant phenotype and the observation that STRIPY is upregulated in MV00191²¹ indicates the existence of a repressor of STRIPY expression in the proximal section of the leaf. To identify this putative repressor, we performed bulk segregant analysis on an F2 population between MV00191 and MvBL. We identified a non-synonymous mutation at a highly conserved site of a gene from the Arabidopsis LSH and Oryza G1 family (Figure S5A), which we named MvALOG1. Mining the RNA-seq data of the three sections of the MvBL leaf²¹ revealed a clear preferential expression of MvALOG1 in the proximal section (Figures 3D and 3E). Knocking down this gene in MvBL by RNAi recapitulated the mutant phenotype (Figures S5B–S5E). Furthermore, qRT-PCR confirmed that STRIPY is upregulated in both the mutant and RNAi plants (Figures 3F and S5F, respectively). Together, these results suggest that MvALOG1 is a repressor of STRIPY expression in the proximal section of the leaf.

Figure 3. MvALOG1 is a repressor of STRIPY in the proximal section of the leaf.

(A and B) Young MvBL (A) and MV00191 plant (B) at the stage when the extended stripe phenotype is the strongest (~4 weeks post-sowing).

(C) MvBL and MV00191 plants that have just reached anthesis (~60 days post-sowing).

(D) Volcano plot of significantly differentially expressed genes between the proximal and distal sections of the MvBL leaf. Vertical dashed lines represent log2 foldchange ≥ 0.58.

(E) Normalized counts for MvALOG1 from RNA-seq data on the distal, stripe, and proximal sections of the MvBL leaf. Error bars represent 1 SD of 4 biological replicates.

(F) qRT-PCR of MvALOG1 and STRIPY expression in the MV00191 mutant line. Error bars represent 1 SD of 4 (MV00191) or 3 (MvBL) biological replicates, and the asterisk (*) indicates that the difference between the mutant line and MvBL is statistically significant, with a p < 0.05.

(G and H) Developmental series (G) and whole plant (H) from a class I line (ALOG1 OE-4).

(I) Heatmap of normalized expression values from RNA-seq analyses of whole leaves of MvBL, class III (ALOG1–9), and class I (ALOG1–4) plants (n = 4 biological replicates).

(J and K) Developmental series (J) and whole plant (K) from a class III line (ALOG1 OE-9).

(L) Confocal microscope image of 35S:YFP-MvALOG1 (ALOG1 OE-4) shows nuclear localization of the protein.

(M–P) GUS staining of a young shoot (M and N) and individual leaf (O and P) from a MvALOG1pro:GUS reporter line and the wild-type control.

Scale bars in (A)–(C), (H), and (K), 1 cm; scale bar in (L), 20 mm; and scale bars in (M)–(P), 1 mm.

See also Figure S5.

To further test the function of MvALOG1, we transformed MvBL with the 35S:YFP-ALOG1 plasmid, with the expectation that the anthocyanin stripe should be eliminated in the OE lines (hereafter, ALOG1 OE). The resulting ALOG1 OE lines fall into three phenotypic categories. Class I completely lacks foliar anthocyanin as expected, but also shows reduced lateral organ size (Figures 3G, 3H, and S5G). Class II displays a weak (or nearly absent) anthocyanin stripe, with no change in lateral organ size (Figure S5H). Curiously, the third class shows variable extension of the stripe into the distal section (Figures 3J, 3K, and S5I) and more anthocyanin coverage by area compared to the MvBL line (Figure S1B). qRT-PCR showed that these three classes of phenotypes were correlated with the ALOG1 transcript level in the transgenic lines, class I being the highest and class III showing only a moderate increase relative to the wild type (Figure S5J). Confocal microscopy of leaf epidermal cells from the class I OE line (ALOG1 OE-4) showed strong nuclear localized signal (Figure 3L), corroborating its role as a transcriptional regulator.

To examine whether these phenotypes are associated with other components of the MBW complex, we compared RNA-seq data from a class I and class III transgenic line with that of the wild type. Similar to the qRT-PCR results, RNA-seq data revealed that the MvALOG1 transcript level increased ~60-fold in ALOG1 OE-4 (class I) but only ~15-fold in ALOG1 OE-9 (class III) (Figures 3I and S5J). Consistent with the phenotype, STRIPY expression is downregulated to undetectable level in ALOG1 OE-4, while upregulated in ALOG1 OE-9 (Figure 3I). The expression of the other MBW components, MvANbHLH1 and MvWD40a, are comparable to that of MvBL in both class I and class III OE lines and are thus not regulated by MvALOG1 (Figure 3I).

To help visualize the spatial pattern of MvALOG1 expression in the whole leaf, we generated MvALOG1 promoter reporter (MvALOG1pro:GUS) lines. GUS signal was observed in the proximal section of very young leaves in multiple independent lines, as well as the base of the corolla in developing flower buds (Figures 3M–3P, S5K, and S5L), suggesting that MvALOG1 is expressed in the proximal area of lateral organs during their development.

TCP5 regulates STRIPY expression in the distal section of the leaf

In our previous study, we hypothesized two potential pattern formation mechanisms underlying the anthocyanin stripe in MvBL: the pattern may be formed by a one-gradient system in which an activator and a repressor of STRIPY are expressed in a proximodistal gradient, with the field of expression slightly larger in the former than the latter, or alternatively, the pattern is formed by a two-gradient system wherein the activator is uniformly expressed and STRIPY expression is inhibited by repressors with opposing gradients from the proximal and distal end, respectively.²¹ Our finding that the upstream activator of STRIPY (MvHY5) showed no expression gradient across the leaf (Figure 2A) argued against the one-gradient model. Therefore, we reasoned that the two-gradient model is more plausible and used the RNA-seq data of MvBL leaf sections to search for putative STRIPY regulators with preferential expression in the distal end.

One such candidate was a homolog of Arabidopsis TCP5 (At5G60970) called MvTCP5 (Figures 4A and 4B). AtTCP5 has been extensively characterized as a regulator of cell expansion during leaf development and is often functionally redundant with multiple paralogs^38–44 (also reviewed in references 26,27). To test whether MvTCP5 is involved in STRIPY regulation, we knocked down the function of MvTCP5 by employing a chimeric repressor system with the canonical EAR motif (35S:TCP5-SRDX; hereafter, TCP5 SRDX),⁴⁵ which was expected to repress all downstream targets of MvTCP5 and related paralogs, thereby alleviating functional redundancy.⁴⁶ The resulting TCP5 SRDX lines displayed excessive marginal growth, as well as anthocyanin production that is extended to the distal section of the leaf (Figures 4C, 4D, S1A, and S6A–S6C). Because the EAR motif converts the gene of interest into an active repressor,⁴⁵ we inferred that the MvTCP5-SRDX chimeric protein represses a yet-to-be identified inhibitor of STRIPY (IOS) in the distal section, leading to the increased anthocyanin production and STRIPY expression in the distal section (Figures 4C–4E). The endogenous MvTCP5 is likely a positive regulator of IOS, thereby negatively regulating STRIPY expression.

Figure 4. MvTCP5 activates a repressor of STRIPY in the distal section of the leaf.

(A) Volcano plot showing differentially expressed genes as determined from RNA-seq data collected using sections of 18- to 20-mm MvBL leaves. Verticaldashed lines represent log2 fold change ≥ 0.58.

(B) Normalized counts for MvTCP5 from distal, stripe, and proximal sections of the MvBL leaf as determined from RNA-seq data of MvBL leaf sections. Error bars indicate 1 SD from 4 biological replicates.

(C) Adult TCP5 SRDX-3 plant.

(D) Developmental series of leaves collected from sequential nodes of TCP5 SRDX-3 plant that has just reached anthesis. Nodes are ordered from oldest to youngest.

(E) qRT-PCR of STRIPY expression in distal section (12 mm from tip) of a pair of 25-mm leaves from MvBL and TCP5 SRDX plants. Error bars represent 1 SD from 3 biological replicates.

(F) Relative expression as measured by qRT-PCR for endogenous MvTCP5. Error bars represent 1 SD from 3 technical replicates (each replicate is a pair of 15-mm leaves).

Asterisks in (E) and (F) represent that the significance of the difference between the transgenic line and MvBL (*p < 0.05; **p < 0.01). Scale bars in (C) and (D), 1 cm.

See also Figure S6.

We further reasoned that constitutive OE of MvTCP5 driven by the 35S promoter (hereafter TCP5 OE) may increase IOS expression throughout the leaf and thus eliminate the anthocyanin stripe. However, TCP5 OE did not lead to obvious change in the stripe pattern (Figures S6D–S6F), despite the transgene being clearly functional, as evidenced by reduced leaf size and pistil length in the transgenics compared to the wild type (Figures S6D–S6H). One possible explanation is that the function of MvTCP5 requires interacting partners,⁴⁷ which might be expressed only in the distal section of the leaf. We also observed significant increases in endogenous MvTCP5 expression in the TCP5 OE lines and decreases in the TCP5 SRDX lines (Figures 4F and S6I), both of which suggest that MvTCP5 can activate its own expression, although the relevance of this self-activation to stripe formation is unclear at this point.

Interrogation of combinatorial action of STRIPY regulators

To determine the phenotypic effects of combinatorial action of these upstream regulators of STRIPY, crosses were made between the various transgenic and mutant lines (Figure 5A). These crosses revealed the following: (1) OE of ALOG1 could overcome the increased anthocyanin production that results from HY5 OE or TCP5 SRDX (first two columns of Figure 5A), leading to a complete loss of anthocyanin production in the leaf. This observation indicates that ALOG1 is a potent inhibitor of STRIPY expression, dominating other STRIPY regulators. (2) The enhancing effects of anthocyanin production of HY5 OE, TCP5 SRDX, and alog1 were largely additive, but the combination of two of these was not enough to extend the anthocyanin stripe to the whole leaf. (3) By simultaneously overexpressing the activator (HY5 OE) and removing the two negative regulators of STRIPY (alog1;TCP5 SRDX) in a triple combinatorial cross, we were able to extend the anthocyanin pigmentation into both the proximal and distal sections to nearly cover the entire adaxial surface of the leaf in the adult plant. RNA-seq analysis of whole leaves from the transgenic lines showed no significant change in expression of MvALOG1, MvTCP5, or MvHY5a when the other two genes were experimentally manipulated (Figure 5B). Taken together, these results suggest that the three regulators of STRIPY do not form hierarchical relationships from one another; they regulate STRIPY expression more or less independently, although the potency of their activation or repression of STRIPY seems to differ.

Figure 5. Combinatorial action of MvHY5, MvALOG1, and MvTCP5.

(A) Leaves from the parental lines HY5 OE, TCP5 SRDX, ALOG1 OE, and alog1 and combinatorial crosses thereof. Scale bars, 1 cm.

(B) Heatmap of normalized expression values for MvHY5a, MvALOG1, and MvTCP5 from whole-leaf RNA-seq. Each circle represents 1 biological replicate.

(C) Schematic diagram of the current genetic regulatory model underlying stripe formation.

To test whether the foliar expression patterns of these three regulators are unique to M. verbenaceus, we performed qRT-PCR experiments on leaf sections of a closely related species, M. parishii,⁴⁸ and found very similar expression patterns (Figure S7A). Intriguingly, M. parishii occasionally produced the same medio-lateral anthocyanin stripe in the first few leaves in our standard greenhouse environment (Figures S7B–S7D), although the precise conditions triggering the phenotype are unknown. Together, these observations suggest that the last common ancestor of these species was already potentiated for stripe formation.

DISCUSSION

In this study, we identified three upstream regulators of STRIPY as putative potentiating factors that form the “hidden” pre-pattern for the localized STRIPY expression. The activator is expressed throughout the leaf, whereas the two negative regulators of STRIPY are expressed in two opposing gradients along the proximodistal axis. These findings provide molecular support to the two-gradient repressor model for the stripe pattern formation (Figure 5C).

Mechanisms of MvHY5, MvALOG1, and MvTCP5 as regulators of STRIPY

HY5 is a central hub of light signaling in plants and regulates a myriad of light-induced developmental processes, including anthocyanin biosynthesis.^29–34 In some species, HY5 is known to directly activate the anthocyanin-regulating R2R3-MYB by binding to its promoter.^30,49 Our expression data and the phenotype of the “HY5 OE 3 STRIPY RNAi” cross (Figures 2C, 2F, and 2G) strongly suggest that in the M. verbenaceus leaf, HY5 also positively regulates anthocyanin biosynthesis through the R2R3-MYB gene STRIPY, most likely through direct interaction with the STRIPY promoter (Figure S4). The fact that HY5 OE increased anthocyanin coverage but not to the entire leaf (Figures 2D and S1) can be explained by two non-exclusive hypotheses: (1) additional, rate-limiting cofactors of HY5, such as BBX proteins,⁵⁰ may be required for STRIPY activation throughout the leaf, and (2) the activation potency of the HY5-cofactor complex is insufficient to override the inhibitory effect of the STRIPY repressors in the proximal and distal sections of the leaf.

While the activation of STRIPY by MvHY5 is hardly surprising, the inhibition of STRIPY expression by MvALOG1 is unexpected. The ALOGs are plant-specific, highly conserved transcription factors that have been shown to regulate diverse developmental processes in various species, including boundary formation during embryogenesis, meristem maturation, corolla tube differentiation, flowering time, and inflorescence architecture.^{24,25,51–55} However, they have not been previously implicated in the regulation of anthocyanin pigmentation, indicating that the role of MvALOG1 in the emergence of the novel foliar stripe in M. verbenaceus is unlikely to be a conserved function of ALOG proteins; instead, it represents a lineage-specific co-option through the rewiring of STRIPY into the ALOG1 target network.

ALOG proteins are generally small in size (150–200 aa), composed of a highly conserved DNA-binding domain (~130 aa) with short, disordered regions at the N and C termini (Figure S5A). They have been shown to directly bind downstream gene promoters in tomato, rice, and Arabidopsis.^56–58 A recently solved crystal structure of the ALOG DNA-binding domain revealed that it contains four alpha helices, with helix 3 directly contacting the DNA bases, and mutations in helix 3 abolish DNA binding.⁵⁷ Given that the causal mutation in the M. verbenaceus alog1 mutant leads to an aa substitution at a highly conserved residue in helix 3 (Figure S5A), it is plausible that this substitution interferes with DNA binding and represents a loss-of-function allele, which is corroborated by the similar phenotypes between the mutant and the MvALOG1 RNAi lines (Figures S5C–S5E). Despite their well-established nuclear localization^{52–54,56,58} and DNA-binding capability,^56,57 how the ALOG proteins regulate target genes remains unclear. They have been shown to either repress^53,56,58 or activate^51,59 gene transcription, potentially depending on which protein partners they interact with. This functional versatility provides a possible explanation to the disparate phenotypes of our class I and class III ALOG1 OE lines (Figures 3H and 3K): the high expression level of ALOG1 could result in homodimerization and transcriptional repression,^54,58 whereas moderate expression throughout the leaf may lead to a different protein complex that activates STRIPY expression.

MvTCP5 is a member of the class II CINCINNATA (CIN)-type TCP transcription factors, with well-documented roles in the regulation of cell expansion during lateral organ development.^26,27 The TCP5 SRDX phenotypes (Figures 4C and 4D) demonstrate that MvTCP5 is involved in anthocyanin pigment patterning in the distal section of the leaf. As the SRDX motif is expected to repress target gene expression, whereas STRIPY is upregulated in the distal portion of TCP5 SRDX leaves, we inferred that there should be an intermediate inhibitor of STRIPY (IOS) downregulated by the TCP5 SRDX chimeric protein, leading to upregulation of STRIPY.

Potentiation and actualization of the novel foliar stripe

The general function of all three upstream regulators of STRIPY in leaf development appears to be well conserved between Mimulus and Arabidopsis, which represent two major angiosperm clades (i.e., Asteridae and Rosidae) that have diverged for 125 million years.⁶⁰ For example, the MvHY5 RNAi and MvHY5 OE lines produce conspicuously lighter and darker green leaves, respectively, than the wild type (Figures 2B, 2E, and S3C–S3F), consistent with its well-characterized role in chlorophyll biosynthesis.²² MvALOG1 OE leads to reduced leaf size (Figures 3G and 3H), similar to OE of the ALOG gene LSH4 in Arabidopsis.²⁵ Silencing MvTCP5 and its paralogs using the SRDX repression motif results in exaggerated growth of leaf margins (Figures 4C and 4D), reminiscent of the phenotype of 35S:TCP3-SRDX in Arabidopsis.⁴¹ The conservation between the Mimulus and Arabidopsis homologs is also manifested in their expression patterns. In Arabidopsis, HY5 is also expressed throughout the leaf,⁶¹ whereas the ALOG genes and CIN-type TCP genes are often preferentially expressed in the basal and distal end of the lateral organs, respectively.^25,62 These similarities in protein function and expression pattern between Mimulus and Arabidopsis imply that MvHY5, MvALOG1, and MvTCP5 are components of an ancestral leaf development program, in which their spatial patterns of expression set up a perfect pre-pattern for a distinct, lateral stripe in the middle of the leaf (Figure 5C). In the simplest scenario, a series of mutations in the cis-regulatory region of STRIPY should be sufficient to rewire STRIPY into this ancestral leaf development program. In other words, MvHY5, MvALOG1, and MvTCP5-IOS are potentiation factors that became co-opted to regulate anthocyanin pigment patterning through cis-regulatory mutations in STRIPY.

Because the causal difference between the striped and the ancestral, non-striped form of M. verbenaceus was genetically mapped to STRIPY, and the STRIPY protein does not differ in function between the two forms, we have previously inferred that cis-regulatory change in STRIPY represents the actualization event underlying the ultimate emergence of the stripe.²¹ The curious observation that M. parishii occasionally produces the stripe in the first few leaves further extends our inference by suggesting two non-exclusive possibilities: (1) while the striped form of M. verbenaceus has evolved the necessary cis-elements that wire STRIPY into the MvHY5-MvALOG1-MvTCP5-IOS regulatory network, M. parishii may have similar but less optimal cis-elements in the STRIPY promoter region, allowing “leaky” expression of STRIPY in the leaf, and (2) the common ancestor of M. verbenaceus and M. parishii not only had all the potentiation factors in place but also evolved the necessary cis-elements in STRIPY; however, only the striped form of M. verbenaceus has the proper chromatin environment (e.g., DNA methylation, histone modification) at the STRIPY locus for stable STRIPY expression. The non-striped M. verbenaceus, however, may have a strong repressive chromatin environment, while that of M. parishii is slightly weaker, permitting occasional STRIPY expression. We note that in both scenarios, the actualization event involves some sort of change at the STRIPY locus, be it cis-regulatory mutation or epigenetic modification. Elucidation of the specific STRIPY cis-elements and the epigenetic states of the ancestral and derived STRIPY alleles will be critical to test these hypotheses and provide a deeper understanding of the origin of this recently evolved foliar pigmentation pattern.

Limitations of the study

In this work, we have uncovered three leaf developmental regulators, MvHY5a, MvALOG1, and MvTCP5, as potentiating factors of a novel foliar pigmentation pattern in the monkeyflower species M. verbenaceus. Although we have shown that these three regulators play key roles in patterning the localized expression of STRIPY, their specific functional mechanisms are not yet well understood. For example, the stable transformation of an MvTCP5 chimeric repression construct resulted in STRIPY expression and anthocyanin accumulation in the distal portion of the leaf, leading us to hypothesize that MvTCP5 normally activates a repressor of STRIPY (IOS) in this region. However, without knowing the identity of IOS, we do not yet have a clear mechanistic understanding of how MvTCP5 prevents STRIPY expression in the distal end of the leaf. Furthermore, the appearance of different classes of extended and reduced stripe phenotypes in the ALOG1 OE lines suggests that the mode of action of this protein may be more complex than simple direct repression. Lastly, HY5 is well known to be dependent on co-factors for functionality and fine regulatory control; its interacting partners in the activation of the foliar stripe remain to be identified. Clarification of how these players collectively govern STRIPY expression will shed light on the detailed mechanisms through which STRIPY came to be wired into the leaf development program.

STAR★METHODS

RESOURCE AVAILABILIT

Lead contact

Requests for further information should be directed to Yao-Wu Yuan (yaowu.yuan@uconn.edu).

Materials availability

Requests for Mimulus verbenaceus and Mimulus parishii resources should be directed to and will be fulfilled by Yao-Wu Yuan (yaowu.yuan@uconn.edu).

Data and code availability

The data that support the findings of this study are openly available from the National Center for Biotechnology Information Bioproject at https://www.ncbi.nlm.nih.gov/bioproject/, reference nos. SRA:PRJNA813304 and SRA:PRJNA1039270 as of the date of publication.

Code utilized for bulked segregant analysis is publicly available at https://doi.org/10.5281/zenodo.11624236.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Plant materials and husbandry

Plants were grown and maintained in the University of Connecticut Botanical Conservatory. Seeds were sown on Lambert LM-GPS germination mix and seedlings were repotted in Jolly Gardener pro-line C/20 growing mix for long-term maintenance. Plants were watered by sub-irrigation, and fertilizer was applied by top-watering thrice per week. Natural light was supplemented with sodium vapor lamps to maintain the plants at approximately 150 μmol m⁻²s⁻¹ for 16 h per day.

METHOD DETAILS

High- and low-light treatment

The light treatment experiments were conducted using two Conviron A1000 growth chambers. Both chambers were set to the same day length cycle, which consisted of 16h/day at 20°C and 8h/night at 15°C. For the high-light treatment experiment, twelve MvBL plants were equally split between two different light regimes: a control regime measured to be approximately 128 μmol m⁻²s⁻¹; and a “high-light” regime at approximately 343 μmol m⁻²s⁻¹. For the low-light experiment, six MvBL plants were placed into the “control” chamber, but three of these were placed inside of a homemade tent made of shade cloth that reduced the light intensity to ~35 μmol m⁻²s⁻¹. Light intensity was measured using an LI-250A light meter (LI-COR Biosciences, Lincoln, NE).

Bulked segregant analysis of MV00191

The EMS mutant line MV00191 displays full expansion of the stripe throughout the proximal section of the leaf on young plants.²¹ EMS mutagenesis results in numerous SNPs throughout the genome. To identify the causal SNP underlying the mutant phenotype, we generated an “MvBL x MV00191” F2 population and isolated genomic DNA from 41 individuals that were presumed to be homozygous for the causal mutation based on phenotype. These samples were pooled together with equal representation, and a small-insert library (300–500 bp) was prepared using the Illumina TruSeq DNA Nano library preparation kit. The library was then sequenced to ~70x coverage using the HiSeq4000 platform at Macrogen, Inc (New York). After reads mapping and SNP calling, to eliminate homozygous SNPs called due to assembly error or non-specific mapping, the MV00191 SNPs were compared against SNP profiles of other EMS mutants generated in the MvBL background, following the previously described approach (code available at Zenodo:11624236 or https://github.com/qslin/MiMut).⁶⁸ The homozygous variants that remained after filtering were manually examined for putative loss-of-function mutations (Table S1) and then checked against our M. verbenaceus leaf transcriptome data to confirm foliar expression. Using this method, we identified a strong candidate gene homologous to the known ALOG gene family member LIGHT SENSITIVE HYPOCOTYLS3 from Arabidopsis (locus AT2G31160, also known as ORGAN BOUNDARY1). We named this gene MvALOG1.

Plasmid construction and plant transformation

The ALOG1 and TCP5 overexpression plasmids was constructed using full-length coding DNAs and the pEarleyGate 104 and pEarleyGate 101 vectors, respectively, which drive transgene expression by the constitutive CaMV 35S promoter and includes an EYFP fluorescent marker fused in frame to either the N terminus (pEarleyGate 104) or C terminus (pEarleyGate 101) of the target gene.⁶³ RNAi knockdown was performed by cloning short coding DNA sequences (MvALOG1, 156 bp; MvHY5a, 367 bp), which were then recombined with the Gateway pB7GWIWG2(II) vector.⁶⁴ The MvALOG1pro:GUS plasmid was prepared using ~2.9 kb of sequence upstream of the start codon with recombination into the pGWB633 vector (generously provided by Dr. Tsuyoshi Nakagawa).⁶⁵ The TCP5-SRDX construct was prepared using the CRES-T chimeric repression vector (generously provided by Dr. Nobutaka Mitsuda), which includes the EAR motif for constitutive repression of all target genes.⁴⁵ Primer sequences used for plasmid constructions are listed in Table S2. Sequences of the final plasmids were verified by Sanger sequencing and/or Oxford nanopore sequencing at Plasmidsaurus (Eugene, OR).

The final plasmids were infiltrated into six to eight adult plants from the MvBL inbred line via Agrobacterium tumefaciens GV3101 using the method described in Yuan et al.⁶⁹ Flowers were hand pollinated for three weeks post-infiltration, and seeds were collected from dehiscent fruits and sown. Germinants were screened by daily application of 1:1000 dilution (v/v) of Finale herbicide (Bayer, Leverkusen, Germany) in water for 2–3 weeks.

RT-qPCR and RNA-seq of structural genes and transcription factors

To standardize the developmental stage of whole leaves selected for RNA-seq, we selected one leaf per plant from below a 5 mm flower bud on the main stem of young plants. These leaves ranged between 20 and 25 mm in length. For the RNA-seq analysis on the leaf sections, six leaves ranging from 18 to 20 mm in length were collected per plant, cut into sections, and individual sections were combined to represent each sample. For all other RNA preparations, a pair of 15 mm leaves was selected from the main stem of young plants that had just begun flowering (i.e., only one or two flowers having reached anthesis).

Fresh plant tissue was collected into liquid nitrogen and ground using a Mini-G Geno/Grinder tissue homogenizer and 4 mm stainless steel grinding balls (Spex Sample Prep, Metuchen, NJ). RNA was extracted using a Spectrum Plant Total RNA Kit (Sigma-Aldrich, Saint Louis, MO). RNA-seq analysis was performed according to the methods of LaFountain et al.²¹ Heatmaps were generated using TBtools.⁶⁷

Complementary DNA (cDNA) for the RT-(q)PCR reactions was synthesized using the GoScript Reverse Transcription system (Promega, Madison, WI). Primers used in RT-qPCR analysis are listed in Table S3. RT-qPCR experiments were conducted using a CFX96 touch real-time PCR detection system (Bio-Rad, Hercules, CA) with SYBR Green master mix (Applied Biosystems, Foster City, CA). The relative expression of target genes was normalized by the expression of the reference gene MvUBC,⁶⁹ and relative expression values were calculated using the delta-delta CT method. The statistical significance of RT-qPCR data was calculated using Welch’s T-tests with the program Microsoft Excel.

Confocal fluorescence microscopy

Microscopy was conducted using a Leica SP8 spectral confocal microscope system at the University of Connecticut Advance Light Microscopy Facility. Microscope settings were as follows: objective, 20x oil; excitation wavelength, 496 nm; laser power, 20%; emission wavelength range, 501–596 nm; and detector gain, 800 V. Images were taken on epidermal peels made from the abaxial base of 3 cm leaves.

Beta-glucuronidase (GUS) assay

Fresh tissue was immersed in 5-bromo-4-chloro-3-indolyl glucuronide (X-gluc) staining solution, which was comprised of 0.1M sodium phosphate buffer pH 7.0, 10mM EDTA, 0.1% Triton X-100, 1mM potassium ferricyanide, and 2mM X-gluc (dissolved in DMSO to make a 0.1M stock solution; Gold Biotechnologies, Saint Louise, MO) and placed into a vacuum infiltration chamber at 25 Hg pressure for 30 min in the dark. Post-infiltration, the tissue was incubated in the staining solution at 37°C for 48h, followed by several washes with 70% ethanol over a 24h period to clear the plant pigments. Blue precipitate was then observed using light microscopy.

Dual luciferase assay

The plasmid used for generating the stable HY5 OE transgenic lines (i.e., 35S:MvHY5a(ΔN77)-YFP) was used as the effector for the dual luciferase experiment. For the reporter construct, three overlapping fragments of STRIPY promoter (approximately 2.6 kb, 1.5 kb, and 0.5 kb upstream of the start codon) were cloned and recombined into a pGreen 0800II-60Luc plasmid that had been modified to include a minimal 35S promoter (generously provided by Dr. Yun Zhou)⁶⁶ and transformed into Agrobacterium tumefaciens GV3101 using the method described in Yuan et al.⁶⁹ Agrobacterium cultures containing the effector and reporter plasmids were inoculated into young Nicotiana benthamiana plants in an effector-to-reporter ratio of 5:1 by volume. The dual luciferase assay was conducted three days post inoculation using the Dual Luciferase Assay System (Promega, Madison, WI).

Yeast one-hybrid

Yeast one-hybrid was conducted using the Matchmaker Gold yeast one-hybrid library system (Takara Bio USA, Inc., Shiga, Japan) following the manufacturer’s instructions for testing one-on-one interactions.

QUANTIFICATION AND STATISTICAL ANALYSIS

Measurement of leaf size and anthocyanin coverage

The position of the stripe was compared in the MvBL, alog1, HY5 OE, and TCP5 SRDX lines by scanning leaves and taking measurements using ImageJ. To standardize the developmental stage of the plant and ensure comparable measurements, leaves were sampled from the main stem of plants that had just started flowering (i.e., having only the first 1–2 flowers reaching anthesis). Measurements included the total leaf length, and distance from distal leaf tip to both top and bottom of stripe. Additionally, the percentage of anthocyanin pigmentation by adaxial leaf surface area was determined as follows: the image was split into red, blue, and green channels; the color threshold of the green channel image was adjusted such that solely the anthocyanin pigmentation was selected; the image was converted to binary; and finally, the percentage anthocyanin coverage per leaf was determined by calculating the area.

Statistical analyses

Significant differences between wild-type (MvBL) and experimental samples were determined using Welch’s unequal variances T test as calculated using Microsoft Excel. Results in Figures 1, 2, 3, and 4, Figure S1, and Figures S4–S7 are presented as the mean value with error bars indicating +/− one standard deviation (SD). Sample size for each experiment is detailed in the corresponding figure legend. Asterisks indicate that the difference between the wild-type and experimental sample is statistically significant (*indicates p-value of <0.05, ** indicates p-value of <0.01).

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Bacterial and Virus Strains
Agrobacterium tumefaciens GV3101	Widely distributed	GV3101
E. coli TOP10 competent cells	Widely distributed	TOP10
Biological Samples
Mimulus verbenaceus alog1 mutant seed	This paper	MV00191
Mimulus verbenaceus HY5 OE x STRIPY RNAi F1 seeds	This paper	N/A
Mimulus verbenaceus HY5 OE x ALOG1 OE F1 seeds	This paper	N/A
Mimulus verbenaceus HY5 OE x TCP5 SRDX F1 seeds	This paper	N/A
Mimulus verbenaceus TCP5 SRDX x ALOG1 OE F1 seeds	This paper	N/A
Mimulus verbenaceus alog1 x TCP5 SRDX F1 seeds	This paper	N/A
Mimulus verbenaceus HY5 OE x TCP5 SRDX x ALOG1 OE triple cross	This paper	N/A
Mimulus verbenaceus HY5 OE x TCP5 SRDX x alog1 triple cross	This paper	N/A
Chemicals, Peptides, and Recombinant Proteins
X-gluc (CHX salt)	Gold Biotechnology, Inc.	Cat#G1281
Critical Commercial Assays
Dual Luciferase Reporter Assay System	Promega	Cat#E1960
Matchmaker Gold Y1H Library Screening System	Takara Bio USA, Inc.	Cat#630491
Spectrum Plant Total RNA Kit	Sigma-Aldrich	Cat#STRN250
GoScript Reverse Transcription system	Promega Corp.	Cat#A5001
SYBR Green master mix	Applied Biosystems	Cat#4367659
Deposited Data
RNA sequencing data for leaf sections	NCBI Sequence Read Archive	PRJNA813304
RNA sequencing data for whole leaves	NCBI Sequence Read Archive	PRJNA1039270
Mimulus verbenaceus v2.0 reference genome	Mimubase.org	MvBLg_v2.0
Experimental Models: Organisms/Strains
Mimulus verbenaceus inbred line “MvBL”	Yao-Wu Yuan; available on request	N/A
Mimulus verbenaceus inbred line “MvNB”	Yao-Wu Yuan; available on request	N/A
Mimulus parishii inbred line “Mpar”	Yao-Wu Yuan; available on request	N/A
Nicotiana benthamiana	Widely distributed	RRID:NCBITaxon_4100
Oligonucleotides
Primers used for building transgenic constructs, see Table S2	This paper	N/A
Primers used for RT-qPCR, See Table S3	This paper	N/A
Recombinant DNA
pEarlyGate 101	Arabidopsis Biological Resource Center; described in Earley et al.63	CD3-683
pEarlyGate 104	Arabidopsis Biological Resource Center; described in Earley et al.63	CD3-686
pB7GWIWG2(II)	https://gatewayvectors.vib.be/collection/pb7gwiwg2ii; described in Karimi et al.64	Vector ID: 1_23
pGWB633	Tsuyoshi Nakagawa; described in Nakamura et al.65	N/A
CRES-T SRDX	Nobutaka Mitsuda; described in Oshima et al.45	N/A
pGreen 0800II-60Luc	Yun Zhou; described in Zhou et al.66	N/A
35S:YFP-ALOG1 (pEarlyGate 104 plasmid)	This paper	ALOG1 OE
MvALOGI RNAi (pB7GWIWG2 plasmid)	This paper	ALOG1 RNAi
MvALOG1pro:GUS (pGWB633 plasmid)	This paper	ALOG1pro GUS
35S:MvHY5a(ΔN77)-YFP (pEarlyGate 101 plasmid)	This paper	HY5OE
MvHY5 RNAi (pB7GWIWG2 plasmid)	This paper	HY5 RNAi
35S:TCP5-YFP (pEarlyGate 101 plasmid)	This paper	TCP5 OE
35S:TCP5-SRDX	This paper	TCP5 SRDX
Software and Algorithms
ImageJ NIH; https://ImageJ.nih.gov/ij/		RRID:SCR_003070
MiMut (custom pipeline for Bulked Segregant Analysis)	Qiaoshan Lin; https://doi.org/10.5281/zenodo.11624236	N/A
TBtools	Described in Chen et al.67	RRID:SCR_023018

Highlights.

• STRIPY is an activator of anthocyanin biosynthesis in a distinct mediolateral leaf stripe

• Three leaf developmental regulators form a pre-pattern that determines STRIPY expression

• MvHY5 activates STRIPY and is expressed throughout the leaf

• MvALOG1 and MvTCP5 negatively regulate STRIPY and are expressed in opposing gradients