The MYB transcription factor RcMYB1 plays a central role in rose anthocyanin biosynthesis

Abstract Rose (Rosa hybrida) is one of most famous ornamental plants in the world, and its commodity value largely depends on its flower color. However, the regulatory mechanism underlying rose flower color is still unclear. In this study, we found that a key R2R3-MYB transcription factor, RcMYB1, plays a central role in rose anthocyanin biosynthesis. Overexpression of RcMYB1 significantly promoted anthocyanin accumulation in both white rose petals and tobacco leaves. In 35S:RcMYB1 transgenic lines, a significant accumulation of anthocyanins occurred in leaves and petioles. We further identified two MBW complexes (RcMYB1-RcBHLH42-RcTTG1; RcMYB1-RcEGL1-RcTTG1) associated with anthocyanin accumulation. Yeast one-hybrid and luciferase assays showed that RcMYB1 could active its own gene promoter and those of other EBGs (early anthocyanin biosynthesis genes) and LBGs (late anthocyanin biosynthesis genes). In addition, both of the MBW complexes enhanced the transcriptional activity of RcMYB1 and LBGs. Interestingly, our results also indicate that RcMYB1 is involved in the metabolic regulation of carotenoids and volatile aroma. In summary, we found that RcMYB1 widely participates in the transcriptional regulation of ABGs (anthocyanin biosynthesis genes), indicative of its central role in the regulation of anthocyanin accumulation in rose. Our results provide a theoretical basis for the further improvement of the flower color trait in rose by breeding or genetic modification.


Introduction
Anthocyanins are secondary metabolites of f lavonoid biosynthetic pathways and are present in most f lowering plants [1].
Anthocyanins not only give f lowers and other organs rich colors, but also prevent damage from drought, ultraviolet radiation, low temperature stress,diseases,pests, and other adverse factors [2][3][4][5].
The MYB-TFs usually form MYB-bHLH-WD40 (MBW) complexes with bHLH (basic helix-loop-helix) and WDR (WD40repeat) proteins, and these complexes can promote or inhibit anthocyanin biosynthesis [18]. The R3 repeats [with a conserved motif (D/E) LX2 (R-K) X3LX6LX3R] in MYB-TFs interact with the N-terminal MYB interacting region (MIR) of bHLHs, and together with WDR proteins form the MBW complex [18,24]. In anthocyanin synthesis, there are usually multiple sets of MBW complexes that synergize or antagonize anthocyanin synthesis. In Arabidopsis, the four MYB-TFs PAP1-4 can form MBW complexes with bHLHs (TT8/GL3/EGL3) and WDR (TTG1) to further enhance anthocyanin biosynthesis synergistically [25]. In orchid (Phalaenopsis equestris), PeMYB2/11/12 independently activate the expression of PeDFR, and the presence of PebHLH1 further activates the expression of PeDFR and enhances red pigmentation [26]. When MdMYB10 was co-transformed with bHLHs in apple, the activity of MdDFR promoter was significantly enhanced. [13]. In woodland strawberry (Fragaria vesca), coexpression of FvMYB10 and FvbHLH33 can more strongly activate the promoters of FvDFR and FvUFGT to promote anthocyanin biosynthesis [27]. The small R3-MYB, which has a single R3 domain and lacks a putative activation domain, can inhibit MBW activity, and in general MYB inhibitors can competitively bind bHLH to reduce the activity of the MBW complex and decrease anthocyanin synthesis [28]. For example, IbMYB44 in Ipomoea batatas inhibits its own transcriptional activity by competitively binding bHLH2 in the MBW (MYB340-bHLH2-NaC56) complex, thereby inhibiting anthocyanin biosynthesis [29]. The MYB inhibitor MYBL2 can competitively bind to bHLHs (MYC1/EGL3/GL3/TT8), and these interactions can disrupt the MBW complex or the binding of the complex to the promoters of ABGs, resulting in decreased anthocyanin accumulation [30].
Only a few studies have reported that MYB-TFs may be involved in the anthocyanin biosynthesis of rose. RrMYB113 and RhMYB10 were supposed to play a role in anthocyanin synthesis [31,32]. RrMYB5 and RrMYB10 are induced by wounding, and their expression leads to increased accumulation of anthocyanins and proanthocyanidins [33]. Therefore, the MYB-TFs and the MBW complex regulating f lower color in roses have not been studied in depth.
Previously, we performed a detailed analysis of R. hybrid 'Rhapsody in Blue' petal RNA-seq data (accession number: PRJNA885821) and identified genes involved in rose petal color regulation. We noticed that the FPKM value of RcMYB1 increased during f lower coloration, suggesting thatzit may promote anthocyanin accumulation in rose f lowers. Here, we analysed the regulation of RcMYB1 and two MBW complexes on rose pigment formation to explore the molecular regulation mechanism of rose color formation. These results provide new information for the regulation of rose color and will be useful for breeding roses and possibly other f lowering plants with a wider range of f lower colors.

RcMYB1 encodes an SG6 R2R3 MYB TF
A phylogenetic tree of RcMYB1 and other members of the MYB transcription factor family in Arabidopsis and some Rosaceae species was constructed. In the tree, RcMYB1 was in subgroup 6 (SG6), whose members are known to promote anthocyanin accumulation in plants (Fig. S1, see online supplementary material). In addition, RcMYB1 showed high homology with RrMYB113 of rugosa rose (Rosa rugosa), RiMYB10 of red raspberry (Rubus idaeus), FaMYB10 of strawberry (F. × ananassa), and FvMYB10 of woodland strawberry (F. vesca) (Fig. S1 To study the transcript profile of RcMYB1 during petal coloration of rose, we selected the petals of seven developmental stages of 'Old Blush' f lowers (Figs S1-S7, see online supplementary material) (Fig. 1A) and determined the RcMYB1 transcript levels and anthocyanin contents. The transcript levels of RcMYB1 gradually increased as anthocyanins accumulated, with the highest transcript level at S5, and then gradually decreased as the anthocyanins degraded (Fig. 1B). Thus, the transcriptional profile of RcMYB1 was consistent with the trends in anthocyanin accumulation during Figs S1-S7, see online supplementary material (Fig. 1C).
Next, we determined the transcript levels of RcMYB1 in rose f lowers with different colors, including blue-purple, red, green, yellow, and white (Fig. 1D). Interestingly, although RcMYB1 showed relatively high transcript levels in red roses, they were equally high in green, yellow, and white roses, but low in several blue-purple roses (Fig. 1E). The yellow f lower is conferred by carotenoids, and the white rose studied here had a strong fragrance. Thus, we speculate that RcMYB1 may also be involved in the biosynthesis of volatile aromatic compounds and carotenoids.
To determine where RcMYB1 functions in cells and whether it has transcriptional activation activity, we performed a subcellular localization analysis and a trans-acting activity assay using a yeast system. In subcellular localization analysis, the RcMYB1-GFP signal was localized in nucleus cells of Nicotiana benthamiana leaves, consistent with its role as transcription factor (Fig. 1F). In the yeast system, RcMYB1-BD yeast cells and positive control VP16-BD also grew well on selective medium (SD/−Trp/-His), while negative control (BD) could not grow on selective medium (Fig. 1G). These results provided further evidence that RcMYB1 functions as an activating transcription factor in the nuclei.

RcMYB1 promotes anthocyanin accumulation in rose petals and tobacco (Nicotiana tabacum) leaves
To clarify the function of RcMYB1 in anthocyanin accumulation, we performed transient overexpression experiments in Rosa hybrida 'Gabriel' petals and tobacco (N. tabacum) leaves. At 5 d after transformation, the white petals in the control (harboring the empty vector, EV) showed a slight red color, while those transiently overexpressing RcMYB1 had changed to deep red ( Fig. 2A) and their anthocyanin content was significantly increased (Fig. 2B). The groups overexpressing RcMYB1 also showed significantly increased transcript levels of ABGs (RcCHSa, RcCHSc, RcCHI, RcF3H, RcF3 H, RcDFR, RcANS, RcUFGT, and RcGT1) (Fig. 2C). To determine the major anthocyanin components in red petals upon RcMYB1 overexpression, we examined the anthocyanin composition by liquid chromatography-mass spectrometry (LC-MS). The red color was mainly due to the increased contents of cyanidin and cyanidin derivatives (Fig. 2D). In tobacco leaves, the leaf areas injected with 35S:RcMYB1 showed a red color and accumulated anthocyanins ( Fig. 2E and F). Further, we used virus-induced gene silencing (VIGS) to silence RcMYB1 in 'Old Blush'. Our results showed that petal color was significantly lightened and anthocyanin content was significantly reduced after the silencing of RcMYB1 (Fig. 2G, H, I).

Significant accumulation of anthocyanins in RcMYB1 transgenic lines in rose
To further clarify the function of RcMYB1 in rose anthocyanin biosynthesis, we transferred the full-length coding sequence of RcMYB1 into the pCAMBIA2300 vector to obtain 35S:RcMYB1 transgenic rose lines (OE-1 and OE-2) (Fig. 3A). The rose transgenic lines were confirmed by amplifying partial sequences of the pCAMBIA2300 vector with RcMYB1 by polymerase chain reaction (PCR) from genomic DNA, as well as by western blotting with an anti-GFP antibody ( Fig. 3B and C). In the RcMYB1-OE lines, the leaves, petioles and stems significantly

RcMYB1 physically interacts with RcTTG1 and RcbHLH42 or RcEGL1 to form MBW complexes
The MBW complex plays a core role in regulating anthocyanin biosynthesis. Therefore, we further identified the possible MBW complexes related to anthocyanin biosynthesis in rose. The phylogenetic tree was constructed using rose bHLH family proteins and reported bHLH proteins (AtMYC1, AtEGL3, AtGL3, AtTT8) related to anthocyanin biosynthesis in Arabidopsis. The tree showed that RcbHLH42 and RcEGL1 were closely related to these four bHLH proteins in Arabidopsis (Fig. S4, see online supplementary material). Another phylogenetic tree was constructed using RcbHLH42, RcEGL1 and other reported bHLH proteins related to anthocyanin biosynthesis in other species. That tree showed that RcbHLH42 and RcEGL1 were closely related to bHLH3 and bHLH33 in strawberry, respectively (Fig. S5A, see online supplementary material). Multiple sequence alignment analysis revealed conserved bHLH motifs in RcbHLH42 and RcEGL1 (Fig. S6, see online supplementary material). According to the amino acid sequence of AtTTG1 in Arabidopsis, we selected the WDR gene, RcTTG1, with the highest homology from rose. A phylogenetic tree was constructed with the TTG1 proteins related to anthocyanin biosynthesis reported for different plant species and showed that RcTTG1 was closely related to FaTTG1 in strawberry (Fig. S5B, see online supplementary material). Multiple sequence alignment analysis showed that RcTTG1 contained a conserved tandem repeat WD motif (Fig. S7, see online supplementary material). To further determine whether RcbHLH42, RcEGL1, and RcTTG1 are related to anthocyanin biosynthesis, we examined their transcript levels in f lowers at different stages of development. The transcript levels of RcbHLH42 and RcTTG1 increased during the early stage of anthocyanin accumulation, while RcEGL1 showed a decreasing trend (Fig. S5C, see online supplementary material). Nevertheless, we found that the transcript levels of these three genes were significantly elevated in the transgenic lines (Fig. S5D).
To determine where RcbHLH42, RcEGL1, and RcTTG1 function in the cells, we performed subcellular localization. The results showed that the GFP signal of RcEGL1-GFP fusion protein was localized in the nucleus, while RcbHLH42-GFP and RcTTG1-GFP fusion proteins were localized in the nucleus and cytoplasm of N. benthamiana epidermal cells (Fig. S5E, see online supplementary material).
To determine whether RcMYB1 could form an MBW complex with RcTTG1, RcbHLH42, or RcEGL1, we performed yeast twohybrid assays (Y2H) assays. Directed Y2H assays validated that RcMYB1 interacted with the two bHLH proteins (RcbHLH42 and RcEGL1) and the WDR protein RcTTG1 (Fig. 4A), and RcTTG1 also interacted with the two bHLH proteins (RcbHLH42 and RcEGL1).
To verify whether RcMYB1, the two bHLHs (RcbHLH42 or RcEGL1) and RcTTG1 HAT1 interact with each other in vivo, we conducted bimolecular f luorescence complementation (BiFC) assays. When RcMYB1-nYFP was coinfiltrated with RcbHLH42-cYFP or RcbEGL1-cYFP or RcTTG1-cYFP, and when RcTTG1-nYFP was coinfiltrated with RcbHLH42-cYFP or RcbEGL1-cYFP into N. benthamiana leaves, the strong YFP f luorescence was located in the nucleus (Fig. 4B). The interactions between RcMYB1 and RcbHLH42 or RcEGL1 or RcTTG1, and between RcTTG1 and RcbHLH42 or RcEGL1 were also confirmed by Co-IP assays (Fig. 4C). These results indicated that RcMYB1, two bHLHs (RcbHLH42 and RcEGL1) and RcTTG1 can interact with each other to form MBW complexes.

Two MBW complexes positively regulate anthocyanin accumulation in a functionally redundant manner
To determine whether RcbHLHs (RcbHLH42, RcEGL1), RcTTG1, and the two MBW complexes were involved in anthocyanin accumulation, we performed transient expression and transgenic assays in tobacco. Transient expression of RcbHLHs (RcbHLH42, RcEGL1) and RcTTG1 alone did not promote anthocyanin accumulation in N. tabacum leaves ( Fig. S8A and B, see online supplementary material), while tobacco leaves co-infected with genes encoding components of two MBW complexes, RcMYB1, RcbHLH42 and RcTTG1 (MBT) and RcMYB1, RcEGL1 and RcTTG1 (MET) accumulated more anthocyanins compared with those transiently expressing RcMYB1 alone. There was no significant difference in anthocyanin accumulation between two lines containing different MBW complexes ( Fig. S8A and B, see online supplementary material).
We further verified this result in the white petals of rose f lowers. Transient overexpression of RcbHLH42 and RcEGL1 alone did not promote anthocyanin accumulation ( Fig. 5A and B), Those expressing each of the MBW complexes accumulated more anthocyanins than did the line expressing RcMYB1 alone, but there was no significant difference in anthocyanin content between the two MBW complex lines ( Fig. 5A and B). In contrast to the results of transient overexpression in tobacco leaves, white petals transiently expressing RcTTG1 alone showed a significant increase in anthocyanin accumulation ( Fig. 5A and B). Further analyses revealed that transient overexpression of RcTTG1, but not RcbHLHs (RcbHLH42 or RcEGL1), increased the transcript levels of ABGs (Fig. 5C). The transcript levels of ABGs were higher in the groups expressing the two MBW complexes than in those expressing RcMYB1 and RcTTG1 alone (Fig. 5C). We detected higher transcript levels of both EBGs and LBGs in the groups expressing the two MBW complexes than in the group expressing RcMYB1 alone. Previous studies have shown that MBW complexes generally regulate LBGs. Thus, we hypothesized that the MBW complexes could enhance the expression of RcMYB1 to regulate the expression of EBGs. In addition, there were no significant differences in phenotype and regulation of ABGs between the two lines expressing MBW complexes, indicating that the two complexes are functionally redundant.
The phenotypes of transgenic tobacco lines also confirmed our results. The 35S:RcbHLH42 and 35S:RcEGL1 transgenic lines had the same f lower color as the wild-type (WT), with no significant accumulation of anthocyanins (

RcMYB1 alone or in MBW complexes activates its own promoter and those of other anthocyanin biosynthesis genes
To further investigate the role of RcMYB1 in regulating ABGs, we performed Y1H and dual-luciferase reporter (LUC) assays. When pLacZi-proRcMYB1, proRcCHSa, proRcCHSc, proRcCHI, proRcF3H, proRcF3 H, proRcDFR, proRcANS, proRcUFGT or proRcGT1 fusion constructs were co-expressed with pB42AD-RcMYB1 in yeast cells, the yeast cells turned blue when screened on the selective medium (SD/−Trp/-Ura) supplemented with X-gal. The negative control (BD) failed to turn blue on the selective medium (Fig. 6A).
These results showed that RcMYB1 is able to bind not only to its own promoter but also to those of a wide range of ABGs (EBGs and LBGs).
Next, we performed LUC experiments to determine whether RcMYB1 and the two MBW complexes activate the promoters of RcMYB1 and ABGs. The results showed that RcMYB1 can activate its own promoter, and the transcriptional activity was further enhanced by the two MBW complexes (Fig. 6B). ChIP-qPCR further confirmed the interaction between RcMYB1 and its own promoter in vivo (Fig. 6C). In addition, RcMYB1 also significantly activated the promoters of all of EBGs and LBGs. The two MBW complexes further enhanced the promoter activity of LBGs, but not those of EBGs (Fig. 6B). These results suggested that RcMYB1 can activate its own promoter and those of a wide range of ABGs, while the two MBW complexes enhance the promoter activity of RcMYB1 and LBGs.

RcMYB1 participate in carotinoids and volatile aroma metabolism but not in PA metabolism
Some studies have indicated that MYB-TFs can regulate the accumulation of anthocyanins and proanthocyanidins (PA), and also play an important regulatory role in a variety of secondary metabolic pathways. We found that RcMYB1 has a high expression level in yellow f lowers and fragrant white f lowers ( Fig. 1D and E). Furthermore, we found potential MYB binding sites in these metabolic pathways genes promoters. Therefore, we studied whether RcMYB1 was involved in these metabolic pathways. Our results showed that the expression of Leucidin reductase (LAR) or Anthin reductase (ANR), key enzymes involved in PA synthesis [30], were not increased after overexpression of RcMYB1 and two MBW complexes (Fig. S9A, see online supplementary material). Y1H and Luc assays also indicated that RcMYB1 is not involved in transcriptional regulation of RcLAR and RcANR ( Fig. S10A and B, see online supplementary material).
Lycopene beta cyclase (LCYB) and lycopene εcyclase (LCYE) are key branching points in the carotenoid synthesis pathway [34]. Interestingly, we found that the expression level of RcLYCB, RcLYCE-1, and RcLYCE-2 were significantly increased after overexpression of RcMYB1, and further enhanced by two MBW complexes (Fig. S9B, see online supplementary material). In order to further study the relationship between RcMYB1 and carotenoid synthesis, transient overexpression was performed in R. hybrida 'Lady of Shalott', which had light orange petals. The results showed that the petals showed a deeper orange-red color after overexpression of RcMYB1 (Fig. 7A), and the contents of carotenoids and anthocyanins in petals were significantly increased (Fig. 7B). Further, UPLC was used to determine the changes of α-carotene and β-carotene contents after overexpression RcMYB1. The results showed that both α-carotene and β-carotene contents were significantly increased after RcMYB1 overexpression in R. hybrida 'Lady of Shalott' (Fig. 7C). To further verify whether RcMYB1 directly regulated the transcription of RcLYCB, RcLYCE-1 and RcLYCE-2, Y1H, Luc, and ChIP-qPCR assays were conducted. Our results showed that RcMYB1 could directly bind and activate the promoter of RcLYCB, RcLYCE-1, and RcLYCE-2 ( Fig. 7D and E). ChIP-qPCR also showed that RcMYB1 could significantly enrich the binding sites on promoters of RcLYCB, RcLYCE-1, and RcLYCE-2 (Fig. 7F, G, H).
In modern roses, RhNUDX1 is involved in the formation of geraniol, the main rose fragrance compound [35]. RcEGS1 is responsible for eugenol biosynthesis in rose [36]. The expression of RcEGS1 and RcNUDX1 was also increased after overexpression of RcMYB1 and two MBW complexes (Fig. S9C, see online supplementary material). In order to further investigate whether RcMYB1 is involved in the metabolism of aromatic compounds, GC-TOFMS analysis was performed on the transient overexpression RcMYB1 in R. hybrida 'Gabriel'. After overexpression of RcMYB1, the relative contents of the main aromatic substances phenylpropanoids and monoterpenoids were significantly increased (Fig. 8A and B). Geraniol belongs to monoter-penoids, which is synthesized by RcNUDX1. In our results, we also found a significant increase in the relative content of geraniol (Fig. 8C). However, eugenol was not found in our determination results. Eugenol is in phenylpropanoids, which commonly present in rose stamens but not in rose petals [36]. Further, Y1H, Luc, and ChIP-qPCR assays were conducted to determine whether RcMYB1 directly regulated RcNUDX1 and RcEGS1. Our results showed that RcMYB1 could directly bind and activate the promoter activity of RcNUDX1 and RcEGS1 (Fig. 8D, E, F, G).
Overexpression of RcMYB1 strongly promoted the accumulation of anthocyanins in white petals, leaves, and petioles (Figs 2  and 3).
In our study, we found that RcMYB1 is involved in an autoregulatory feedback loop to regulate the accumulation of anthocyanins (Fig. 6). A similar autoregulatory feedback loop has also been reported in apple MdMYB10 and monkeyf lower (Mimulus lewisii) PELAN (Petal Lobe Anthocyanin) [13,49]. Some inhibitory R2R3-MYBs have also been shown to be under autoregulatory control. Arabidopsis MYB4 can inhibit its own transcription by binding to its own promoter as part of a negative autoregulatory loop [50]. Similar to rice (Oryza sativa) OsMYB4 [51] and petunia MYB27 [19].

RcbHLH42, RcEGL1, and RcTTG1 are involved in the regulation of anthocyanin accumulation by interacting with RcMYB1
In this study, we constructed a phylogenetic tree containing all rose bHLH family proteins as well as documented bHLH proteins (MYC1/EGL3/GL3/TT8) related to anthocyanin biosynthesis in Arabidopsis. We found that only RcbHLH42 and RcEGL1 were closely related to these four bHLH proteins in Arabidopsis (Fig. S4, see online supplementary material). bHLHs typically play a role in the formation of MBW complexes and enhance MYB activity but do not promote anthocyanin biosynthesis alone [52]. Consistent with the finding that FhGL3L and FhTT8L of F. hybrida cannot regulate anthocyanin biosynthesis independently of endogenous MYB proteins [53], we did not observe anthocyanin accumulation after overexpressing RcbHLH42 and RcEGL1 alone in tobacco leaves and rose petals ( Fig. 5A; Fig. S8A, see online supplementary material). However, in Arabidopsis, the mutation of TT8 reduced anthocyanin and made seed coat colorless [54]. As RcMYB1 can independently activate EBGs and LBGs and can strongly promote the accumulation of anthocyanin in tobacco leaves and white rose petals, we think that the accumulation process of anthocyanin in rose petals mediated by RcMYB1 may not depend on RcbHLH42 and RcEGL1. In addition, after forming the MBW complex with RcMYB1 and RcTTG1, the transcriptional activity of RcMYB1 was enhanced by two MBW complexes (MBT, MET), so RcbHLH42 and RcEGL1 may play a role as enhancers-mediated by RcMYB1-in anthocyanin accumulation in rose petals.
A few WDR are involved in anthocyanin synthesis; on example is the WD40 protein TTG1, which has been isolated from a number of plant species [55]. In Arabidopsis, lack of seed coat pigment and decrease of anthocyanin accumulation in ttg1 mutant [56]. In rice, the anthocyanin content in the osttg1 mutant was significantly reduced in various organs [57]. Overexpression of CsWD40 in tobacco resulted in significant increase of anthocyanin content in petals [4]. In our study, anthocyanins accumulated significantly after RcTTG1 was overexpressed in white rose petals (Fig. 5A), which showed that RcTTG1 played an important role in the accumulation of anthocyanins in rose.
In the MBW complex, bHLH proteins usually interact with both MYB and WDR as linkers, while WDR may not interact with MYB. In apple, MdTTG1 interacts with bHLH but not with MYB proteins [58]. In Paeonia qiui, PqMYB113 interacts with PqbHLH1 but not with PqWD40 [52]. In tomato, SlAN11 (WDR) interacts with bHLH but not with the MYB protein in the MBW complex [59]. However, in our study, we found that RcTTG1 can interact not only with RcbHLHs but also with the RcMYB1 protein (Fig. 5). The WD40 protein interaction with MYB protein has also been reported in recent studies. For example, in woodland strawberry, FcTTG1 can interact with FcMYB114, FcMYB123, and FcbHLH42 [60]; in eggplant SmMYB86 interacts with SmTTG1 [61]; in Medicago truncatula MtWD40-1 interacts with MtPAR or MtLAP1 [62]; and in kiwifruit AcWDR1 interacts with AcMYBF110 [48].
The MBW complex is the key factor in the regulation of LBGs [28]. In Arabidopsis, EBGs are activated by coactivator-independent and functionally redundant R2R3-MYBs (MYB11/12/111), whereas activation of LBGs requires a MBW complex [25]. Similar to our results, we found that two MBW complexes significantly enhanced the activity of LBGs but had only a slight effect on EBGs activity (Fig. 6B). However, some studies have revealed that all ABGs are regulated by MBW complexes, such as in maize (Zea mays) [63], F. hybrida [46], rice (O. sativa) [64], and sweet potato [65]. These results suggest that MBW complexes may have different regulatory mechanisms for anthocyanin accumulation.
In addition, we found that the transcript levels of RcbHLH42 and RcEGL1 were not associated with anthocyanin accumulation during rose f lower development, consistent with previous observations for LcbHLH1/2/3 in L. chinensis [66]; for MdbHLH33/3 in apple [13]; for VvMYC1 in grape [45]; and for FhGL3L in F. hybrida [53]. RcEGL1 was only highly expressed at the S1 stage and gradually decreased with the development of f lower buds, while RcbHLH42 was highly expressed at the S3 and S4 stages (Fig. S5C, see online supplementary material), so we speculated that the MET complex might play a role in anthocyanin synthesis at an earlier stage than the MBT complex. In addition, we also found that the expression of RcEGL1 increased after MBT complex overexpression, and the expression of RcbHLH42 also increased after MET overexpression (Fig. S11, see online supplementary material). Therefore, we speculated that there might be cooperation in the regulatory mechanism in these two MBW complexes.

RcMYB1 is involved in other secondary metabolic pathways
So far, the roles of R2R3-MYBs in plants are mainly (∼70%) in regulating plant metabolic processes, including the benzenoid, phenylpropanoid, terpenoid, and glucosinolate (GSL) pathways [67]. In this study, we found that RcMYB1 is involved in multiple secondary metabolic pathways. Previous studies have found that MYB regulates anthocyanin and proanthocyanidin biosynthesis. Such as MtMYB5/14 in M. truncatula [68], MdMYB9/11 in apple [69], RrMYB5/10 in R.rugosa [33]. However, in this study, we found that RcMYB1 did not regulate the transcription of RcLAR and RcANR (Figs S8A and S9, see online supplementary material).
Interestingly, we found that RcMYB1 regulates the biosynthesis of carotenoids and aromatic volatiles. Carotenoids are closely related to the coloration of yellow, orange red or red f lowers [70]. Lycopene beta cyclase (LCYB) and lycopene εcyclase (LCYE) are key branching points in the carotenoid synthesis pathway [34]. In this study, we transient overexpressed RcMYB1 in R. hybrida 'Lady of Shalott', and found that petals changed from light orange to orange-red (Fig. 7A), further we confirmed that RcMYB1 can directly regulate the transcription of RcLYCB, RcLYCE-1, and RcLYCE-2, thereby mediating the accumulation of α-carotene and β-carotene (Fig. 7). Many studies have reported that R2R3 MYB TFs can directly regulate carotenoid biosynthesis. For instance, in M. truncatula, WHITE PETAL1 (WP1) directly regulates the expression of MtLYCe and MtLYCb [71]. AdMYB7 can activate the promoter of AdLCYB in A. chinensis [34]. Similarly, in Camellia sinensis [72], in Rhyncholaeliocattleya [73], and in M. lewisii [74] have also been reported.
In addition, many R2R3 MYB-TFs have also been found to contribute to volatile aroma metabolism and control the emission of volatile substances [75]. A regulatory network of three R2R3 MYB-TFs, ODORANT1 (ODO1), EMISSION OF BENZENOIDS I (EOBI), and EOBII, can directly activate the transcription of many genes in the shikimic acid pathway and its downstream f loral synthesis pathway [76,77]. Expression of AtPAP1 not only promotes the synthesis of phenylpropanoid and terpenoid odour compounds but also promotes the accumulation of surfactants in R. hybrida, tobacco, and petunia [78]. Aroma components such as aldehydes, phenylpropanoids, and terpenes increased significantly in SlMYB75-OE tomato fruits [79]. In A. thaliana, MYB3 binds to the promoter of the gene cinnamate 4-hydroxylase (C4H) to negatively regulate gene expression and inhibit phenylpropanoid synthesis [80]. PhMYB4 has also been reported to inhibit the expression of PhC4H in P. hybrida [81]. In rose, a 1R-MYB TF, RhMYB1 may be a putative gene involved in the biosynthesis of rose scent [82]. Here, we provide evidence that a R2R3 MYB TF, RcMYB1, is involved in the biosynthesis of eugenol and geraniol in rose (Fig. 8). Phenylpropanoids and monoterpenoids are the main f loral compounds of rose. After overexpression of RcMYB1, we found that the content of phenylpropanoids and monoterpenoids showed significant increases, and the content of geraniol, catalyzed by RcNUDX1, also significantly increased. Our determination results did not detect eugenol, catalyzed by RcEGS1, possibly because eugenol is normally present in the stamens of rose [36]. Interestingly, the contents of many phenylpropanoids and monoterpenoids were significantly increased after overexpression of RcMYB1, suggesting that RcMYB1 may also be widely regulated in the metabolism of aromatic compounds.

Conclusion
In conclusion, we revealed that R2R3-MYB RcMYB1 plays a central role in anthocyanin biosynthesis and also contributes to the metabolism of carotenoids and aromatic volatiles. Two bHLH proteins (RcbHLH42 and RcEGL1) can bind RcMYB1 and RcTTG1 to form the MBW complex, mainly regulating LBGs to regulate anthocyanin biosynthesis.

Cloning of the RcMYB1, RcbHLHs (RcbHLH42 and RcEGL1), and RcTTG1 genes and the promoters of RcMYB1 and anthocyanin biosynthesis genes
The full-length sequences of RcMYB1, RcbHLH42, RcEGL1, and  Table S1 (see online supplementary material).

Sequence alignment and phylogenetic analysis
The full-length amino acid sequences of RcMYB1, RcbHLH42, RcEGL1, and RcTTG1 in rose were aligned using CLUSTAL OMEGA. The amino acid sequences of MYB, bHLHs, and WD40 previously identified were from NCBI GenBank. RcMYB1, RcbHLH42, RcEGL1, and RcTTG1 were functionally classified by phylogenetic tree analysis. MEGA11 (https://megasoftware.net/) and iTOL (https:// itol.embl.de/) were used to build a maximum likelihood tree to analyse the phylogenetic relationship between protein.

RT-qPCR
Total RNA was extracted using the SteadyPure Plant RNA Extraction Kit (Accurate Biology, Hunan, ChangSha, China). qPCR was performed using Hifair ® III 1st Strand cDNA Synthesis SuperMix (Yeasen, Shanghai, China) PerfectStart ® Green qPCR SuperMix (Transgen Biotech, Beijing, China) was used for RT-qPCR. The 2 − Ct method was used, with RcActin as the internal parameter. The primers of RT-qPCR are shown in Table S2 (see online supplementary material).

Transient overexpression in R. hybrida 'Gabriel', R. hybrida 'Lady of Shalott', and N. tabacum
For the transient assay in R. hybrida 'Gabriel' and R. hybrida 'Lady of Shalott' petal discs, A. tumefaciens-mediated vacuum infiltration was used. OD 600 = 1.0, as described above. The petals were immersed in the infection solution, then treated in a vacuum of −80 kPa for 5 minutes, twice, placed in a filter paper petri dish containing 1% sucrose at 18 • C for 6 d. Each treatment contained three experimental replicates, and approximately 100 petal discs were used from five f lowers.
For the transient assay in N. tabacum leaves, the above different combinations of strains infected leaves, 7 days sampling.

Transformation of R. hybrida 'Novalis' and N. tabacum
Transformation of R. hybrida 'Novalis' was performed via A. tumefaciens-mediated transformation using leaf-derived embryogenic callus [83]. Brief ly, sterile young leaves were used to induce callus on the callus induction medium (CIM) and further induce somatic embryos on the embryos induction medium (EIM). A. tumefaciens cells containing pCAMBIA 2300-RcMYB1 were collected and suspended at OD 600 = 0.5. Then they were washed three times in an infiltrating solution at −80 kPa for 15 min and cocultured on a co-culture medium (CM) at 22 • C dark for 4 days. Somatic embryos were cultured on screening medium supplemented with kanamycin and timentin, and positive transgenic plants were identified by genomic PCR, RT-q PCR, and Western blotting.
A. tumefaciens cells containing RcMYB1, RcbHLH42, RcEGL1, or RcTTG1 were transformed into N. tabacum leaves using the leaf disc method [84]. Transgenic plants were selected for each construct in a medium containing kanamycin and timentin. Then transfer to the greenhouse until f lowering. Positive transgenic (genomic PCR and RT-q PCR) T1 progeny tobacco plants were identified.

YIH
The CDS of RcMYB1 was inserted into the pB42AD vector, and the promoter and anthocyanin synthesis genes of RcMYB1 were inserted into the pLacZi vector. Strain Y1HGold (Clontech, Mountain View, CA, USA) was used. PEG/LiAc method was used for conversion. The transformed yeast cells were inoculated into (SD/−Trp/-Ura) medium with or without X-Gal. Use the primers in Table S3 (see online supplementary material).

Y2H
Insert CDS of RcMYB1, RcbHLH42, RcEGL1, and RcTTG1 into the pGADT7 or pGBKT7 carrier. The plasmid was transformed into Y2HGold (Clontech, Mountain View, CA, USA) cells and then grown on SD/−Leu/−Trp medium. The potential physical interactions between proteins were further tested on SD/−Trp/−Leu/-His/−Ade medium. The primers used are shown in Table S3 (see online supplementary material).

BiFCassay
The coding sequences of RcMYB1 and RcTTG1 were constructed into PXY106, and the coding sequences of RcbHLH42, RcEGL1, and RcTTG1 were constructed into PXY104. The primers are in Table S3 (see online supplementary material). As mentioned above, infect the N. benthamiana leaves. After 3 d infection, the cells were stained with 150 μg/mL DAPI. Yellow f luorescent protein (YFP) f luorescence and DAPI f luorescence (Olympus FV3000 confocal scanning microscope) were observed at excitation wavelengths of 505 nm and 340 nm, respectively.

Co-IP assay
RcMYB1 and RcTTG1 were cloned into PEG104 with Flag tag by Gateway™ LR Clonase™ II Enzyme mix. RcMYB1, RcbHLH42, and RcEGL1 were cloned into PEG104 with Myc tag. Primers are in Table S3 (see online supplementary material). In Co-IP, A. tumefaciens cells containing different structures were collected, as described above, OD 600 = 1.0, and then infiltrated N. benthamiana leaves. After 3 days, N. benthamiana leaves were ground at a low temperature to extract protein, then incubated at 4 • C for 4 h in the presence of anti-Flag or anti-Myc antibody conjugated beads. They were rinsed four times in the wash buffer. The protein was isolated in 8% SDS-PAGE gel and analysed by Western blot.
For the determination of anthocyanins, liquid chromatography-mass spectrometry (LCMS) assays were performed on a Shimadzu LCMS-2020 (Shimadzu, Kyoto, Japan). Chromatographic separation was performed on an Agilent Zorbax SB-C18 column (4.6 * 250 mm, 3 μm). Eluent A (0.1% formic acid solution): Eluent B (0.1% formic acid acetonitrile solution) = 95:5. The f low rate was 1.2 mL/min, the column temperature was 30 • C for 3 min, and the detection wavelength was 254 nm. Electrospray ionization (ESI) voltage was set at 0.9 KV, capillary temperature at 350 • C, jacket gas pressure at 45 psi, and auxiliary gas at 10 psi. According to the retention time, the anthocyanin composition was determined, and then the anthocyanin equivalent was quantified.

Carotenoid extraction and quantification
A 0.5 g petal sample was crushed and 2 mL anhydrous ethanol (containing 0.1%BHT) was added, then a water bath at 80 • C for 5 min, adding 100 μL KOH solution (80%w/v). In a water bath at 80 • C for 15 min, 1 mL purified water and 1 mL n-hexane were added, after vortex mixing, centrifuge at 3000 rpm for 5 min, transfer the supernatant, add 1 mL n-hexane into the residue again for extraction, combine the two supernatants after centrifugation, and then steam dry at 30 • C under mild nitrogen. Redissolve with 0.2 mL methanol solution for detecting. The sample extracts were analysed at 450 nm using an UPLC system with DAD detector (UPLC, U3000; Thermo, Karlsruhe, Germany). The analytical conditions were as follows, UPLC: column, YMC Carotenoid S-3 μm (150 * 4.6 mm), column temperature, 40 • C; f low rate, 1.0 mL/min; injection volume, 2 μL; solvent system, MeOH:

Identification of the volatile compounds
Volatiles were collected from petal discs by solid-phase microextraction (SPME) methods. Incubate 10 min at 50 • C, extract 30 min, and deadsorb 5 min in GC injection mouth. After extraction, the compounds were analyzed with GC-TOFMS (Pegasus BT, Leco, St. Joseph, MI, USA). Conditions of GC-TOFMS: the temperature of DB-WAX (30 m × 250 μm × 0.25 μm) is 250 • C, the interface temperature is 290 • C and the ion source temperature is 230 • C. The initial temperature was 40 • C for 3 min, then increased to 70 • C at 3 • C/min, increased to 180 • C at 5 • C/min, and finally increased to 240 • C at 10 • C/min for 7 min.

Dual-luciferase reporter assay
The LUC analysis was performed with minor modifications according to the method described by He et al. [87]. CDS of RcMYB1, RcbHLH42, RcEGL1, and RcTTG1 were inserted into pCAMBIA 2300 vector to form effectors. The promoters of RcMYB1 and ABGs are inserted into pGreen II-0800 to form the reporter vector. The primers used are in Table S3 (see online supplementary material). The plasmid was introduced into A. tumefaciens strain (containing the pSoup helper plasmid). Different A. tumefaciens cells composed of effector and reporter are infected into the N.benthamiana leaves. Three day, 20 mg/mL D-lucifin (Goldbio, St Louis, MO, USA) was sprayed on the leaves and images were collected using a cooled low-light CCD imager (Tanon-4200, Shanghai, China).

ChIP-qPCR
ChIP-qPCR assays were performed as described by Bowler et al. [86]. The 5 g leaf tissues from RcMYB1-GFP-OE1 and GFP-OE plants were treated with 1% (v/v) formaldehyde to cross-link the protein-DNA complexes. The protein-DNA complexes were immunoprecipitated with an anti-GFP antibody (Cat. no. ab290; Abcam, Cambridge, UK). The immunoprecipitated DNA was purified and analysed using RT-qPCR. Use primers in Table S3 (see online supplementary material). article in accordance with the policy described in the Instructions for Authors. G.H., R.Z., S.J., and H.W. conducted the experiments. G.H. and R.Z. analysed the data and wrote the manuscript. All authors read and approved the manuscript.

Data availability
All relevant data in this study are provided in the article and its supplementary file.