The role of VvMYBA2r and VvMYBA2w alleles of the MYBA2 locus in the regulation of anthocyanin biosynthesis for molecular breeding of grape (Vitis spp.) skin coloration.

In grape, MYBA1 and MYBA2 at the colour locus are the major genetic determinants of grape skin colour, and the mutation of two functional genes (VvMYBA1 and VvMYBA2) from these loci leads to white skin colour. This study aimed to elucidate the regulation of grape berry coloration by isolating and characterizing VvMYBA2w and VvMYBA2r alleles. The overexpression of VvMYBA2r up-regulated the expression of anthocyanin biosynthetic genes and resulted in higher anthocyanin accumulation in transgenic tobacco than wild-type (WT) plants, especially in flowers. However, the ectopic expression of VvMYBA2w inactivated the expression of anthocyanin biosynthetic genes and could not cause obvious phenotypic modulation in transgenic tobacco. Unlike in VvMYBA2r, CA dinucleotide deletion shortened the C-terminal transactivation region and disrupted the transcriptional activation activity of VvMYBA2w. The results indicated that VvMYBA2r positively regulated anthocyanin biosynthesis by forming the VvMYBA2r-VvMYCA1-VvWDR1 complex, and VvWDR1 enhanced anthocyanin accumulation by interacting with the VvMYBA2r-VvMYCA1 complex; however, R44 L substitution abolished the interaction of VvMYBA2w with VvMYCA1. Meanwhile, both R44 L substitution and CA dinucleotide deletion seriously affected the efficacy of VvMYBA2w to regulate anthocyanin biosynthesis, and the two non-synonymous mutations were additive in their effects. Investigation of the colour density and MYB haplotypes of 213 grape germplasms revealed that dark-skinned varieties tended to contain HapC-N and HapE2, whereas red-skinned varieties contained high frequencies of HapB and HapC-Rs. Regarding ploidy, the higher the number of functional alleles present in a variety, the darker was the skin colour. In summary, this study provides insight into the roles of VvMYBA2r and VvMYBA2w alleles and lays the foundation for the molecular breeding of grape varieties with different skin colour.

Introduction

Anthocyanins, the plentiful category of plant flavonoids, are the major pigments responsible for coloration in many plant species (Grotewold, 2006; Holton and Cornish, 1995; Payyavula et al., 2013; Petroni and Tonelli, 2011). In many plants, anthocyanin biosynthesis is regulated by the MYB‐bHLH‐WD40 (MBW) transcription complex formed by MYB, bHLH (also known as MYC) and WD40 (also called WDR) proteins (Baudry et al., 2004; Koes et al., 2005; Patra et al., 2013). The MYB superfamily is one of the most plentiful categories of transcription factors (TFs) in plants, which is classified into four categories (1R‐, 3R‐, 4R‐ and R2R3‐MYB) according to the amount of highly conserved repeats in the MYB domain (Dubos et al., 2010). Among these four categories, R2R3‐MYB is the largest family, with 126 members reported in Arabidopsis and further grouped into 22 subfamilies based on conserved motifs (Stracke et al., 2001). The R2R3‐MYB family plays an important role in the regulation of anthocyanin biosynthesis (Allan et al., 2008). Many R2R3‐MYB regulators have been proved as transcriptional activators of anthocyanin biosynthetic genes in numerous plant species such as Arabidopsis (Borevitz et al., 2000; Gonzalez et al., 2008), maize (Paz‐Ares et al., 1987), grapevine (Deluc et al., 2008; Walker et al., 2007), apple (Ban et al., 2007; Espley et al., 2007), tomato (Mathews et al., 2003), petunia (Quattrocchio et al., 1999), potato (Jung et al., 2005, 2009; Zhang et al., 2009), pepper (Borovsky et al., 2004) and sweet potato (Chu et al., 2013). In contrast, some R2R3‐MYB TFs have been identified as repressors of the anthocyanin biosynthetic pathway in several plant species, such as Arabidopsis (Jin et al., 2000), strawberry (Aharoni et al., 2001), grapevine (Cavallini et al., 2015), apple (Lin‐Wang et al., 2011), ginkgo (Xu et al., 2014) and petunia (Albert et al., 2011).

Grapevine is considered as an important fruit crop due to its high economic value and cultivated worldwide (Nile et al., 2013). Skin colour is an essential quality trait of grape berries, which not only determines the market value of table grapes but also influences the processing applications and processed product quality (Kayesh et al., 2013). At present, the skin colour of grape berry has become more diversified owing to human choice and hybridization. Anthocyanin component and content are responsible for colour variation of berry skin among white (greenish‐yellow or yellowish‐green), red and black. The existence or absence of anthocyanins in berry skin has been shown to segregate as a monogenic trait determined by a locus in linkage group 2 (Doligez et al., 2002; Fischer et al., 2004; Salmaso et al., 2008). Walker et al. (2006) reported that two MYB loci (MYBA1 and MYBA2) located on the second chromosome control berry colour in grapevine by regulating anthocyanin biosynthesis. Because these two loci are in very close genetic linkage (Azuma et al., 2011), they can be considered as a part of a single haplotype (allele). The origin of white‐skinned grape berries was clarified to be the result of the destruction of two functional genes (VvMYBA1 and VvMYBA2) from the MYBA1 and MYBA2 loci. The transcriptional inactivation of VvMYBA1 caused by the insertion of Gret1 (Ty3‐gypsy‐type retrotransposon) in its promoter region resulted in a non‐functional allele (VvmybA1a; Kobayashi et al., 2004, 2005). Recent genetic evidences indicated that white‐skinned grape varieties were homozygous for VvmybA1a, whereas coloured‐skinned varieties contained at least one type of the functional allele (VvmybA1c or VvmybA1b; Azuma et al., 2007; Kobayashi et al., 2004; Lijavetzky et al., 2006; This et al., 2007). Walker et al. (2007) reported that two single nucleotide polymorphism (SNP) mutations in the coding region of VvMYBA2 disrupted its regulatory function. In Vitis labrusca grapes, three functional MYB genes – VlmybA1‐3, VlmybA1‐2 and VlmybA2 – have been identified at the correspondent colour locus (Kobayashi et al., 2002; Koshita et al., 2008). Various combinations between five alleles (VvmybA1a, VvmybA1b, VvmybA1c, VlmybA1‐3 and VvmybA1) of the MYBA1 locus and four alleles (VvMYBA2r, VvMYBA2w, VlmybA2 and VlmybA1‐2) of the MYBA2 locus led to multiple haplotypes in different grape varieties (Figure S1; Ban et al., 2014).

In grapevine (Vitis spp.), numerous structural genes in the flavonoid pathway have been identified (Boss et al., 1996; Downey et al., 2003; Kennedy et al., 2000); however, their transcriptional regulation is still far from being clarified. Recent studies showed that several R2R3‐MYB genes, such as VvMYBA1 (Kobayashi et al., 2002, 2004), VvMYB5a (Deluc et al., 2006), VvMYBPA1 (Bogs et al., 2007), VvMYB5b (Deluc et al., 2008), VvMYBF1 (Czemmel et al., 2009), VvMYBPA2 (Terrier et al., 2009) and VvMYBPAR (Koyama et al., 2014), are associated with the regulation of anthocyanin and tannin biosynthesis. In addition, the VvMYBA2r and VvMYBA2w alleles have been partially characterized, and VvMYBA2r is thought to specifically regulate the gene expression of VvUFGT, which encodes an enzyme responsible for the conversion of anthocyanidins to anthocyanins (Walker et al., 2007). However, there is limited information regarding the detailed functional characterization of VvMYBA2r and VvMYBA2w, as well as their potential regulatory networks involved in the formation of berry colours. Moreover, the precise relationship between genotypes (colour locus and MYB haplotype) and phenotypes (berry colour) requires further investigation. In this study, we comprehensively characterized the role of VvMYBA2r and VvMYBA2w proteins by performing bioinformatics analysis, expression pattern surveys, subcellular localization, dual‐luciferase reporter assay, transactivation activity analysis, stable and transient expression analyses and regulatory network investigations (yeast two‐hybrid and bimolecular fluorescence complementation). More interestingly, we explored the relationship between MYB haplotypes and coloration using 213 germplasms from different Vitis species and two interbreeding populations as plant materials. Our results might provide a potential strategy for the early prediction of colour diversification during the classical cross‐breeding of grapes and an important theoretical basis for the molecular breeding of grape varieties with different berry colour. In addition, our study provides insight into the role of VvMYBA2r and VvMYBA2w alleles of the MYBA2 locus in the regulation of anthocyanin biosynthesis.

Results

Isolation and bioinformatics analysis of VvMYBA2r and VvMYBA2w

VvMYBA2r and VvMYBA2w, encoding putative R2R3‐MYB TFs, were isolated, respectively, from a cDNA library constructed using ‘Heimeiren’ and ‘Shine Muscat’ grapes. The open reading frame (ORF) of VvMYBA2r encodes a protein of 344 amino acid (aa) residues with a predicted molecular weight (Mw) of 39.1 kD and an isoelectric point (pI) of 7.78. The ORF of VvMYBA2w encodes a protein of 265 aa residues with a predicted Mw of 30.3 kD and pI of 9.57. The alignment of VvMYBA2r and VvMYBA2w with other identified R2R3‐MYB TFs from several plant species revealed the existence of R2R3 repeats (DNA‐binding domain) in the N‐terminal regions of these proteins (Figure 1a). In addition to the well‐conserved R2R3 domain, two additional distinct motifs were observed within the sequence alignment of VvMYBA2r and VvMYBA2w: a bHLH interacting motif ([D/E]Lx2[R/K]x3Lx6Lx3R), which is involved in the interactions with bHLH TFs and is present in the majority of R2R3‐MYB TFs (Grotewold et al., 2000; Stracke et al., 2001), and a motif 6 (KPRPR[S/F]F), which is often found in MYB TFs that could regulate anthocyanin biosynthesis. The functional characteristics of VvMYBA2r and VvMYBA2w were elucidated by modelling the structure of R2R3 domain in VvMYBA2r/2w on the Phyre2 website (http://www.sbg.bio.ic.ac.uk/phyre2/html/) using the Arabidopsis R2R3‐MYB TF as the template model (PDB code: 6kks; Figure 1b, upper panel; Kelley et al., 2015; Wang et al., 2020). The R2R3 domain in VvMYBA2r/2w had 58% sequence identity with Arabidopsis R2R3‐MYB protein. The resulting model was very similar to the template model, with a root‐mean‐square deviation (RMSD) of superimposed Cα of 1.2 Å for 100 aligned residues. Structural comparison with the AtMYB‐DNA complex revealed the possible VvMYBA2r/2w‐DNA interaction (Figure 1b, bottom panel). The arginine residue at position 44 (R44) of VvMYBA2r was located in the backbone of the DNA duplex and could be involved in DNA binding. In contrast, in VvMYBA2w, R44 was replaced by a neutral leucine residue (named R44L), which did not have a positive side chain, implying that R44L is likely to modify the interaction of the TF with the DNA backbone and disrupts the tertiary structure of the TF. A phylogenetic tree was also constructed using amino acid sequences of VvMYBA2r and VvMYBA2w and MYBs of other plant species, such as Arabidopsis thaliana, Malus domestica, Oryza sativa, Gossypium hirsutum, Solanum tuberosum, Prunus avium, Fragaria vesca, Pyrus communis, Solanum lycopersicum, Lotus japonicus, Picea mariana, Petunia hybrida, Antirrhinum majus, Diospyros kaki, Trifolium arvense, Trifolium repens and Medicago sativa (Figure 1c). According to the phylogenetic tree, VvMYBA2r showed the closest relationship with VvMYBA2w with a 75.29% sequence similarity. The other MYB TFs from the grapevine, including VvMYBF1 (Czemmel et al., 2009), VvMYBF2 or VvMYBPA1 (Bogs et al., 2007) and VvMYBPA2 (Terrier et al., 2009), were categorized into other clusters, indicating possible functional divergence between these proteins and VvMYBA2r/2w.

Figure 1

The three‐dimensional structural model of the R2R3 domain and phylogenetic analysis of VvMYBA2r and VvMYBA2w. (a) Multiple alignment of VvMYBA2r and VvMYBA2w and other R2R3‐MYB transcription factors from various plant species. Identical residues are shown in black, conserved residues in dark grey, and similar residues in light grey. The position of the R2R3‐type MYB domain is indicated below the alignment, and the C1 motif is highlighted by black boxes. The bHLH interacting motif is indicated above the alignment, and motif 6 is indicated below the alignment. (b) The R2R3 domains of VvMYBA2r and VvMYBA2w were modelled using Phyre2. The R2R3 domain of VvMYBA2r (left panel) and VvMYBA2w (right panel) was coloured by rainbow. R44 may participate in DNA binding. Grey indicates Arabidopsis R2R3‐type MYB, hot pink indicates VvMYBA2r amino acids 6‐118, and green indicates VvMYBA2w amino acids 6‐110. R44 and R44L were both coloured yellow. The DNA duplex was coloured light blue. (c) Phylogenetic relationships between both VvMYBA2r and VvMYBA2w and other R2R3‐MYB transcription factors from grapevine and other plant species. Phylogenetic and evolutionary analyses were performed using the neighbour‐joining method using MEGA version 7.0 (Tamura et al., 2011). The numbers next to the nodes are bootstrap values from 1000 replicates. GenBank accession numbers are as follows (in parentheses): VvMYBA2r (DQ886419), VvMYBA2w (DQ886420), VlMYBA1‐2 (AB427164), VlMYBA2 (AB073013), VlMYBA1‐3 (AB427165), VvMYBA1a (AB111100), VvMYBA1b (AB111101), VvMYBA1c (AB242302), VvMYBA1SUB (DQ345539), VvMYBA1BEN (AB442010), VvMYBA12 (NM_001281001.1), VvMYBF1 (GQ423422.1), VvMYB5a (AAS68190), VvMYB5b (Q58QD0), VvMYBPA1 (AM259485), VpMYBA1 (KC342651.1), GhMYBA1 (KF749431.1), StMYBA1‐1 (KP317177.1), PaMYBA1 (JN166079.1), MdMYBA1 (GU270471.1), FvMYBA1 (EU155163.1), PcMYB10 (KC993864), ZmP (P27898), SlMYB12 (ACB46530), AtMYB11 (NP_191820), LjMYB12 (BAF74782), VvMYBPA2 (ACK56131), MdMYB22 (DQ074470), VvMYBF2 (GSVIVT00033103001), AtMYB5 (U26935), PmMYBF1 (AAA82943), AtMYB3 (AAF18515), AtMYB32 (NP_179263), AtMYB7 (NP_179263), AtMYB4 (AAC83582), AtMYB123/TT2 (Q9FJA2), OsMYB3 (D88619), ZmC1 (AAA33482), AtMYB75/PAP1 (AAG42001), AtMYB113 (NP_176811), AtMYB90/PAP2 (AAG42002), PhAN2 (AAF66727), SlANT1 (AAQ55181), PhMYB1 (Z13996), AmMIXTA (CAA55725), AtMYB122 (NP_177548), AtMYB28 (NP_200950), AtMYB29 (NP_196386), AtMYB0/GL1 (P27900), AtMYB66/WER (CAC01874), AtMYB101 (NP_180805), AtMYB65 (NP_187751), AtMYB33 (NP_850779), AtMYB23 (NM_123397), and AtMY114 (NP_176812), GhMYB38 (AAK19618), DkMYB2 (BAI49719), LjTT2a, (BAG12893), LjTT2b (BAG12894), LjTT2c (BAG12895), TaMYB14 (AFJ53053), TrMYB14‐1 (AFJ53048), MsMYB14‐1 (AFJ53055), DkMYB4 (BAI49721), PhPH4 (AAY51377), OsMYB4 (BAA23340), AtMYB111 (NP_199744), AtMYB12 (NP_182268). I: Anthocyanins, II: Trichome development, III: General flavonoid pathway regulation, IV: Proanthocyanidin (PA) clade 1, V: PA clade 2, VI: Flavonol/Phlobaphene clade, VII: C2 repressor motif clade, VIII: Cell shape, IX: Glucosinolate clade, X: Anther development.

Expression patterns of VvMYBA2r, VvMYBA2w, VvMYCA1, VvWDR1 and flavonoid biosynthetic enzymes during grape berry development

The spatio‐temporal expression patterns of VvMYBA2r, VvMYCA1 and VvWDR1, as well as the structural genes of the flavonoid biosynthetic pathway, were examined using ‘Heimeiren’ berries at different development stages using quantitative real‐time reverse‐transcriptase PCR (qRT‐PCR). In this study, the entire process of grape berry development, from fruit setting to the maturation phase, was classified into six stages (Figure S2A). The transcript level of VvMYBA2r was low from 15 to 30 days after anthesis (DAA), but continuously increased thereafter until maturity (Figure S2B). However, the transcript level of VvMYBA2w was lower than that of VvMYBA2r throughout the berry developmental stages of white ‘Shine Muscat’ cultivar (Figure S3). VvMYBA2r and VvUFGT were highly co‐expressed, peaking at 95 DAA (Figure S2B). Correlation analysis also revealed that the transcript level of VvMYBA2r showed highly significant positive correlation with that of VvUFGT (r = 0.983, P < 0.01; Table S1), which is consistent with the results of Niu et al. (2016). The VvLDOX transcript abundance, which was maintained at high levels from 60 to 95 DAA and peaked at 95 DAA (Figure S2B), also showed a highly significant positive correlation with VvMYBA2r (r = 0.981, P < 0.01; Table S1). The expression levels of VvF3H1 and VvOMT were low before veraison, but high until maturity, peaking at 60 DAA. The transcript levels of VvF3H1 and VvOMT showed positive correlation with that of VvMYBA2r (r = 0.930 and 0.913, P < 0.01 and <0.05, respectively; Table S1). VvLAR1, VvLAR2 and VvANR were highly expressed before veraison, but exhibited continuous decreasing trends until maturity, as previously reported (Bogs et al., 2005). VvLAR1 and VvLAR2 both had a significant negative correlation with VvMYBA2r transcript abundance (Table S1). The VvMYCA1 transcript level continuously increased during berry development, peaking at 75 DAA, and then declining at 95 DAA. The expression of VvWDR1 mildly fluctuated throughout the berry development process, although it decreased at 30 and 95 DAA as well.

VvMYBA2r and VvMYBA2w are both localized to the nucleus

VvMYBA2r and VvMYBA2w were predicted to be localized to the nucleus via the WoLF PSORT Prediction online software (https://www.genscript.com/wolf‐psort.html). The prediction was validated by fusing full‐length ORFs of VvMYBA2r and VvMYBA2w to the N‐terminal of green fluorescence protein (GFP) driven by CaMV 35S promoter, generating fusion proteins 35S:VvMYBA2r‐GFP and 35S:VvMYBA2w‐GFP, respectively. The vector containing GFP alone (35S: GFP) was regarded as control. The two fusion proteins and the control were agroinfiltrated into the leaf epidermis of Nicotiana benthamiana using Agrobacterium‐mediated transformation. Fluorescence microscopy exhibited that the control was uniformly dispersed throughout the cell (Figure 2), whereas the 35S:VvMYBA2r‐GFP and 35S:VvMYBA2w‐GFP fusion proteins were detected exclusively in the cell nucleus (Figure 2).

Figure 2

Subcellular localization of VvMYBA2r and VvMYBA2w proteins. Epidermal cells of Nicotiana benthamiana leaves were transiently transformed with 35S:GFP (a), used as a control, 35S:VvMYBA2r‐GFP (b) and 35S:VvMYBA2w‐GFP (c), respectively. Images under fluorescence (left), bright field (middle) and the merged images are shown on the right. Bar: 20 μm.

Ectopic expression of VvMYBA2r, but not of VvMYBA2w, enhances anthocyanin accumulation in transgenic tobaccos

The ability of VvMYBA2r and VvMYBA2w to regulate anthocyanin biosynthesis in grapevine was determined by ectopically expressing the coding sequences of VvmybA2r and VvmybA2w in tobacco plants. Transformed tobacco plants were screened using PCR to confirm the existence of each transgene (Figure 3s). Semi‐quantitative PCR was used to evaluate their expression levels in the leaf, corolla and stamen of the transgenic plants. The amount of anthocyanin was considerably greater in the corollas and stamens of tobacco plants overexpressing VvMYBA2r (Figure 3b, h) than in the WT plants (Figure 3a, g). However, anthocyanin accumulation was not significantly different between the VvMYBA2w‐overexpressing lines and WT plants (Figure 3f, l). Notably, a substantial delay in anther dehiscence was observed in the transgenic lines compared with that in the WT plants, which is consistent with the findings of Deluc et al. (2006, 2008). Cross‐section observations of the anthers showed differences in the lignification network of the endothecial cell wall, which is responsible for stomium breaking and the resultant release of pollens, between the transgenic lines and WT plants. Fewer endothecial cells developed lignified fibres, which were frequently incomplete (Figure 3n, o, p, q, r) in the transformed lines than in the WT plants, which exhibited intact fibres covering the entire radial walls (Figure 3m). Thus, the phenomenon can be attributed to the difference in the lignified fibres of endothecial cells. Five VvMYBA2r‐overexpressing lines (R1, R3, R4, R7 and R9) accumulated significantly higher levels of anthocyanin in the leaves, corollas and stamens, whereas no significant difference in anthocyanin content was detected in the five lines overexpressing VvMYBA2w (Figure 3u, v, w) compared with that in the WT plants. In the corolla of five VvMYBA2r‐overexpressing lines, the average anthocyanin content was 6.18 ± 0.47 mg/g fresh weight (FW), with the highest anthocyanin content noted in the R3 line (7.21 ± 0.60 mg/g FW; Figure 3v), which was nearly 3.5 times the content in the corollas of the WT plant (2.08 ± 0.42 mg/g FW). Therefore, the regulation of anthocyanin biosynthesis differed between VvMYBA2r and VvMYBA2w.

Figure 3

Phenotypes and anthocyanin content of transgenic tobacco plants transformed with VvMYBA2r, VvMYBA2r, VvMYBA2r, VvMYBA2r and VvMYBA2w. Corolla and stamen of wild‐type (WT) plant (a, g), transgenic tobacco overexpressing VvMYBA2r (B, H), VvMYBA2r (C, I), VvMYBA2r (D, J), VvMYBA2r (e, k) and VvMYBA2w (f, l), respectively. Cross‐sections of anthers from WT (m), VvMYBA2r (n)‐, VvMYBA2r (o)‐, VvMYBA2r (p)‐, VvMYBA2r (q)‐ and VvMYBA2w (r)‐ overexpressing transgenic tobacco plants. Red boxed areas are magnified, and the magnifying area of respective figures is shown below. Arrows indicate endothecial cells developing complete (m) or incomplete (n, o, p, q and r) lignified fibres. Bars = 50 mm (upper), 250 μm (middle) and 70 μm (bottom). The expression level of transgenes in three tested tissues of transgenic tobacco plants (s). Actin was used as a quantitative control. 35S was used for screening positive plants. CK represents clear water used as negative control. (t) Diagrammatic representation of VvMYBA2w, VvMYBA2r and its three mutations (VvMYBA2rM, VvMYBA2rD and VvMYBA2rM+D) showing the location of sequence differences. The amount of amino acids in each domain is shown below the sequence, and coloured boxes show the regions that were identical. The white star represents the position of mutation leading to the R44L alteration. The black star represents the position of the deletion of CA dinucleotide. The C‐terminal domain (CR) was repeated in VvMYBA2r (CR1 and CR2). Anthocyanin content was quantified using spectrophotometry in leaf (u), corolla (v) and stamen (w) from five independent transgenic lines of each construct. Error bars are the Standard error (SE) for three replicate extracts per line. Statistical significance was determined using one‐way ANOVA; significant differences between means (LSD, P < 0.05) are shown as different lowercase letters above the bars. Red dividing lines are used to differentiate the different parts of the gel images.

Unlike in VvMYBA2r, the VvMYBA2w aa sequence has two non‐synonymous SNP mutations. One of them is a non‐conserved aa change in arginine residue at position 44 (R44), and the second is the deletion of a dinucleotide (CA) in the reading frame at aa 258 (Figure 1a; Walker et al., 2007). The effects of the two non‐synonymous mutations (R44L substitution and CA dinucleotide deletion) within VvMYBA2w on anthocyanin biosynthesis in V. vinifera as well as the effect of either of the mutations on gene activity were determined by mutating VvMYBA2r to generate three novel constructs (VvMYBA2r, VvMYBA2r and VvMYBA2r), each containing one of the differences (Figure 3t). Subsequently, the three constructs were ectopically expressed in tobacco. Compared to those in WT plants, slight phenotypic differences were found in the corollas of the transgenic lines overexpressing VvMYBA2r (Figure 3c, i) and VvMYBA2r (Figure 3d, j). However, no significant phenotypic change was observed in the corollas and stamen of the VvMYBA2r +D‐overexpressing lines (Figure 3e, k) compared with those in WT plants. Moreover, the anthocyanin contents of the three tested tissues in the tobacco plants overexpressing VvMYBA2r and VvMYBA2r were significantly lower than those of the VvMYBA2r‐overexpressing tobacco lines. Furthermore, no significant difference in anthocyanin contents was noted between corollas and leaves of VvMYBA2r‐ and VvMYBA2r‐overexpressing tobacco lines (Figure 3u, v). Interestingly, the corollas of the transgenic lines overexpressing VvMYBA2r accumulated lower levels of anthocyanin than those of the VvMYBA2r‐ and VvMYBA2r‐overexpressing tobacco lines. However, no significant difference was noted in anthocyanin contents of the three tested tissues between the VvMYBA2r and VvMYBA2w‐overexpressing tobacco lines (Figure 3u, v, w). These results suggest that either mutation seriously impacts its effect on regulating anthocyanin biosynthesis, although neither had such a severe effect as VvMYBA2w, indicating that the two alterations might be additive in their effects.

VvMYBA2r, but not VvMYBA2w, activates the anthocyanin biosynthetic genes in the corollas of transgenic tobaccos

The efficacy of VvMYBA2r and VvMYBA2w to regulate anthocyanin biosynthesis, as well as the effect of the non‐synonymous SNP mutations of VvMYBA2w on transcriptional regulatory activity, was elucidated by quantifying the transcript levels of several structural genes from the anthocyanin biosynthesis pathway, including NtUFGT, NtCHI, NtCHS, NtDFR, NtF3H, NtF3′H, NtOMT and NtANS, in the corolla of transgenic tobacco plants. These structural genes were expressed at higher levels in VvMYBA2r‐overexpressing transgenic tobacco plants than in WT plants (Figure S4). Among these genes, NtUFGT (by 2.94‐ and 3.64‐fold), NtOMT (by 3.13‐ and 3.99‐fold) and NtANS (by 3.74‐ and 3.96‐fold) were significantly up‐regulated in the R1 and R3 transgenic lines, respectively, compared with that in WT plants. Thus, VvMYBA2r could enhance the expression levels of these anthocyanin biosynthesis genes. However, the eight genes, including NtUFGT, NtOMT and NtANS, were not induced significantly in the W1 and W2 transgenic lines compared with that in WT plants. In VvMYBA2r‐overexpressing tobacco lines (RM2 and RM5), the expression levels of these genes were significantly lower than those in VvMYBA2r‐overexpressing tobacco lines (R1 and R3) and similar to those in VvMYBA2r‐overexpressing tobacco lines (RD3 and RD4). These results also suggested that either mutation seriously affected the efficacy of VvMYBA2r to regulate anthocyanin biosynthesis. In addition, the transcript levels of the eight genes in the VvMYBA2r‐overexpressing tobacco lines (RM+D1 and RM+D3) were lower than those in the VvMYBA2r‐ and VvMYBA2r‐overexpressing tobacco lines and not significantly different from those in the VvMYBA2w‐overexpressing tobacco lines, further indicating that the two mutations within VvMYBA2w were additive in their effects. Furthermore, the expression profiles of these structural genes in the corollas of transgenic tobacco lines were highly correlated with their anthocyanin contents.

CA dinucleotide deletion within VvMYBA2w shortens its C‐terminal transcriptional activation region and disrupts its transcriptional activation activity

Apart from nuclear localization, transcriptional activation ability is another defining feature for transcription factors. Yeast system was used to confirm the capacities of VvMYBA2r and VvMYBA2w to activate transcription. For this purpose, pGBKT7 (BD)‐VvMYBA2r and BD‐VvMYBA2w fusion plasmids were transformed into the yeast strain Y2HGold. The growth of the cells was compared with that of cells transformed with the negative control (pGBKT7) on the same selection medium synthetic dropout (SD)/‐Trp and of cells transformed with the positive control (pCL1) on the SD/‐Leu medium. The yeast Y2HGold cells transformed with each vector could grow on the SD medium (Figure 4a, upper panels), indicating the reliability of the yeast system. The yeast cells transformed with the negative control and BD‐VvMYBA2w fusion plasmid could not grow on the SD/‐His‐Ade medium, whereas the yeast cells transformed with pCL1 and BD‐VvMYBA2r fusion plasmids survived on this medium (Figure 4a, middle panels). The yeast cells transformed with pCL1 and BD‐VvMYBA2r fusion plasmid turned blue when they were cultured on the SD/‐His‐Ade medium with 20 mM X‐α‐Gal (Figure 4a, bottom panels). In addition, the β‐galactosidase activity was measured using O‐nitrophenyl‐β‐D‐galactopyranoside (ONPG) as substrate. The yeast cells transformed with VvMYBA2w effector showed a very weak activity similar to those transformed with the empty vector used as negative control (Figure 4b). In contrast, VvMYBA2r could strongly activate the transcription of the LacZ gene. These results indicated that VvMYBA2r possessed transactivation activity, whereas VvMYBA2w did not. The difference of transactivation properties between VvMYBA2r and VvMYBA2w was investigated by inserting three mutations (VvMYBA2r, VvMYBA2r and VvMYBA2r) of VvMYBA2r were inserted into the pGBKT7 vector to generate BD‐VvMYBA2rM, BD‐VvMYBA2rD and BD‐VvMYBA2rM+D fusion plasmids, respectively. The yeast cells transformed with BD‐VvMYBA2rD and BD‐VvMYBA2rM+D fusion plasmids could not grow and exhibited very weak β‐Gal activity similar to that of cells transformed with the negative control (Figure 4a, b), whereas the yeast cells transformed with BD‐VvMYBA2rM fusion plasmid survived on the SD/‐His‐Ade selection medium and could strongly activate the transcription of the LacZ gene, indicating that CA dinucleotide deletion disrupted the transactivation activity of VvMYBA2w. This was further verified by segmenting VvMYBA2r into VvMYBA2rΔNterm (the truncated protein corresponds to amino acids 1‐257) and VvMYBA2rΔCterm (the truncated protein corresponds to amino acids 258‐344) based on the second non‐synonymous SNP mutation site. The yeast cells transformed with BD‐VvMYBA2rΔNterm could not grow on the SD/‐His‐Ade medium, whereas those transformed with BD‐VvMYBA2rΔCterm survived, indicating that the region for transcriptional activation function was located in the C‐terminal region of VvMYBA2r. These results indicated that the non‐synonymous SNP mutation (CA dinucleotide deletion) shortened the C‐terminal transcriptional activation region and thus disrupted the transcriptional activation activity of VvMYBA2w.

Figure 4

CA dinucleotide deletion within VvMYBA2w shortens its C‐terminal transcriptional activation region and disrupts its transcriptional activation activity. (a) Transcriptional activation capacity of VvMYBA2w, VvMYBA2r, and the two truncated VvMYBA2r peptides (VvMYBA2rΔNterm and VvMYBA2r ΔCterm), as well as the three VvMYBA2r mutations (VvMYBA2rM, VvMYBA2rD and VvMYBA2rM+D). Growth of yeast cells (strain Y2HGold) transformed with positive (pCL1) and negative (pGBKT7) control vectors and fusion vectors harbouring VvMYBA2r, VvMYBA2w, VvMYBA2rΔNterm (amino acids 1‐257), VvMYBA2rΔCterm (amino acids 258‐344), VvMYBA2r, VvMYBA2r and VvMYBA2r on SD/‐His‐Ade with or without X‐α‐Gal. (b) Yeast transactivation assay of the LacZ reporter gene using VvMYBA2w, VvMYBA2r, VvMYBA2rΔNterm, VvMYBA2rΔCterm, VvMYBA2rM, VvMYBA2rD and VvMYBA2rM+D. Activation of LacZ by the respective constructs was determined by measuring β‐galactosidase activity. Each value is the mean ± standard error (SE) of two independent yeast transformations and each experiment included three measurements (* P < 0.05 vs negative control). MEL1 UAS, Melibiose 1‐GAL4 Upstream Activating Sequence; mp, minimal promoter; pADH1, the promoter of alcohol Dehydrogenase 1; GAL4 DBD, GAL4 DNA‐binding domain.

VvMYBA2r and VvWDR1 both interact with VvMYCA1 in vitro and in vivo

As anthocyanin biosynthesis is speculated to be regulated by the MBW transcription complex (MYB, bHLH, and WD40), the physical interactions among VvMYBA2r, VvMYCA1 and VvWDR1 were detected using the yeast two‐hybrid (Y2H) system. The VvMYBA2rΔNterm fragment was cloned into pGBKT7 vector, and no auto‐activation was confirmed on the SD/‐His‐Ade medium with X‐α‐Gal. The AD‐VvMYCA1 fusion plasmid was co‐transformed with the individual construct harbouring either BD‐VvMYBA2rΔNterm or BD‐VvWDR1 fusion plasmid. The yeast cells could grow on the SD/‐Trp‐Leu (DDO) medium (Figure 5a). Yeast cells of positive control (pGBKT7‐53 plus pGADT7‐T) and co‐transformants with two combinations (BD‐VvMYBA2rΔNterm plus AD‐VvMYCA1, and BD‐VvWDR1 plus AD‐VvMYCA1) could grow on the SD/‐Trp‐Leu‐His‐Ade (QDO) medium containing 125 ng/mL aureobasidin A (AbA) and turn blue in the presence of X‐α‐Gal, whereas the negative control combinations (pGBKT7‐lam plus pGADT7‐T, BD‐VvVvMYBA2rΔNterm plus AD, BD plus AD‐VvMYCA1, BD‐VvWDR1 plus AD, and BD plus AD) could not grow on the same screening medium. In addition, the yeast cells co‐transformed with either BD‐VvMYBA2rΔNterm plus AD‐VvMYCA1 or BD‐VvWDR1 plus AD‐VvMYCA1 could strongly activate the transcription of the LacZ gene similar to that in the positive control (Figure 5b). These data indicated that VvMYBA2r and VvWDR1 both physically interact with VvMYCA1 in yeast cells. The interaction among VvMYBA2r, VvWDR1 and VvMYCA1 was also verified by performing an in vivo interaction assay using bimolecular fluorescence complementation (BiFC). Yellow fluorescent protein (YFP) fluorescence was observed in the nuclei of the cells co‐transformed with two combinations (YNE‐VvMYBA2r plus YCE‐VvMYCA1 and YNE‐VvWDR1 plus YCE‐VvMYCA1). In contrast, no YFP fluorescence was observed in the control combinations including the empty pSPYCE‐35S (YCE) vector plus either YNE‐VvMYBA2r or YNE‐VvWDR1, and the empty pSPYNE‐35S (YNE) plus either YCE or YCE‐VvMYCA1 (Figure 5c). These results indicated that VvMYBA2r and VvWDR1 both interacted with VvMYCA1 in vivo in plant cells. Thus, VvMYBA2r, VvMYCA1 and VvWDR1 were potentially able to form the MBW transcription complex in grapevine.

Figure 5

VvMYBA2r and VvWDR1 both interact with VvMYCA1 in vitro and in vivo. (a) All the constructs together with the positive control (pGBKT7‐53 plus pGADT7‐T) and negative control (pGBKT7‐lam plus pGADT7‐T) were transformed into yeast strain Y2HGold. Yeast clones co‐transformed with different constructs were grown on SD/‐Trp‐Leu (DDO) or SD/‐Trp‐Leu‐His‐Ade (QDO) medium containing 125ng/mL AbA with or without X‐α‐Gal. The blue colour indicates X‐α‐Gal activity on the QDO medium. (b) β‐galactosidase assay in yeast. Transformed yeasts were tested for LacZ activation. Negative and positive two‐hybrid controls refer to the manufacturer’s instructions. β‐galactosidase activity results are the mean of three measurements of three independent yeast clones. Error bars indicate standard error (SE). (c) BiFC assay shows that each of VvMYBA2r, VvWDR1 and VvMYBA2rD interacts with VvMYCA1 in vivo in the tobacco leaf epidermis. The controls were pSPYNE‐35S (YNE)‐VvMYBA2r plus pSPYCE‐35S (YCE), YNE‐VvMYBA2rD plus YCE, YNE‐VvWDR1 plus YCE, YNE plus YCE‐VvMYCA1 and YNE plus YCE. The representative images of the epidermal cells under fluorescence (YFP), bright field (DIC) and merged are shown. Bars: 30 μm.

R44L substitution abolishes the interaction of VvMYBA2w with the bHLH partner

Y2H assay was also performed to explore the ability of VvMYBA2w to physically interact with the bHLH partner. Unlike VvMYBA2r, VvMYBA2w could not interact with the VvMYCA1 protein in yeast (Figure 5a). These results along with those of multiple sequence alignment indicated that the R44L substitution and/or CA dinucleotide deletion potentially affected the interaction of VvMYBA2w with the VvMYCA1 protein. This was verified by co‐transforming the AD‐VvMYCA1 fusion plasmid with the individual construct harbouring either BD‐VvMYBA2rM, BD‐VvMYBA2rD or BD‐VvMYBA2rM+D fusion plasmid. As shown in Figure 5a, b, the yeast cells co‐transformed with the BD‐VvMYBA2rD and AD‐VvMYCA1 fusion plasmids could grow on the SD/‐Trp‐Leu‐His‐Ade (QDO) medium containing 125 ng/mL AbA and turn blue in the presence of X‐α‐Gal, and could strongly activate the transcription of the LacZ gene similar to that in the positive control. However, the yeast cells co‐transformed with either BD‐VvMYBA2rM or VvMYBA2rM+D with AD‐VvMYCA1 fusion plasmids could not survive under the same condition and only showed very weak β‐Gal activity similar to that of the negative control. Thus, the R44L substitution abolished the ability of VvMYBA2w to form a binary complex with VvMYCA1 protein. This finding was also confirmed by performing the BiFC assay. YFP fluorescence was observed in the nuclei of the cells co‐transformed with YNE‐VvMYBA2rD and YCE‐VvMYCA1, similar to that in cells co‐transformed with YNE‐VvMYBA2r and YCE‐VvMYCA1 (Figure 5c). In contrast, no YFP fluorescence was observed in the cells co‐transformed with YCE‐VvMYCA1 and each of VvMYBA2rM, VvMYBA2rM+D and VvMYBA2w, similar to that in cells co‐transformed with the negative control (YNE plus YCE and YNE‐VvMYBA2rD plus YCE). Taken together, these results indicate that the R44L substitution abolishes the interaction capabilities of VvMYBA2w with the VvMYCA1 protein partner, which subsequently affects the regulation of VvUFGT gene expression. Similar results regarding loss of interaction ability with the bHLH partner was noted in grapevine VvMYB5b with a R69L substitution.

R44L substitution and CA dinucleotide deletion both seriously impair the ability of VvMYBA2w to activate transcription of an anthocyanin pathway structural gene

The effect of R44L substitution and CA dinucleotide deletion in VvMYBA2w on activating the transcription of anthocyanin pathway structural genes in the absence or presence of bHLH proteins was investigated by conducting a dual‐luciferase (LUC) assay (Figure 6a). VvMYBA2r alone could induce the VvUFGT promoter by approximately five‐fold, whereas the activity of LUC was not significantly induced by VvMYBA2w alone compared with that by the empty vector. However, the activity of LUC was seriously inhibited when each of VvMYBA2rM and VvMYBA2rD was used, unlike when VvMYBA2r was used alone, implying either mutation could impair the ability of VvMYBA2r to activate the transcription of VvUFGT. Interestingly, the LUC activity was lower with VvMYBA2rM+D alone than VvMYBA2rM and VvMYBA2rD alone, similar to the activation capacity with VvMYBA2w alone, implying two alterations might be additive in their effects. The MYB proteins co‐expressed with bHLH protein partners could significantly enhance the capacity of MYB to activate the transcription of the anthocyanin pathway structural genes (Hichri et al., 2011). The bHLH protein ENHANCER OF GLABRA3 (EGL3) participated in the regulation of anthocyanin pathway in Arabidopsis and was used as a positive control in this study. VvMYBA2r in combination with VvMYCA1 significantly induced this promoter by approximately 60‐fold (Figure 6b), suggesting that VvMYCA1 could significantly enhance the capacity of VvMYBA2r to activate the VvUFGT transcription. Furthermore, VvMYBA2rD in combination with VvMYCA1 could activate the VvUFGT promoter by approximately nine‐fold, which is more than that with VvMYCA1 alone, further indicating that CA dinucleotide deletion impairs the ability of VvMYBA2r to activate the transcription of the VvUFGT gene despite the interaction of VvMYBA2rD and VvMYCA1 proteins. However, the promoter activity did not increase significantly when either VvMYBA2rM, VvMYBA2rM+D or VvMYBA2w with VvMYCA1 was used compared with that when VvMYCA1 was used alone. In addition, stronger activation capacities were measured when EGL3 was used instead of VvMYCA1. Thus, the dual‐luciferase assay revealed that either mutation seriously impairs the ability of VvMYBA2w to activate the transcription of VvUFGT and the two alterations might be additive in their effects.

Figure 6

Dual‐luciferase assay probes the effects of VvMYBA2w, VvMYBA2r and its three mutations (VvMYBA2r, VvMYBA2r and VvMYBA2r +D) on activating the VvUFGT promoter from grapevine in N. benthamiana leaves. (a) Schematic diagrams of the vectors used for the dual‐luciferase assay. In the reporter vectors, the promoter of VvUFGT gene was fused to the LUC reporter gene. P35S, 35S promoter; T35S, 35S Terminator; REN, Renilla luciferase; LUC, Firefly luciferase; RB and LB, T‐DNA borders; attR1 and attR2, GATEWAY recombination sequences. (b) VvMYBA2w, VvMYBA2r and its three mutations activate the promoter of grapevine VvUFGT gene determined by dual‐luciferase assay in tobacco leaves. The bHLH protein ENHANCER OF GLABRA3 (EGL3) participated in the regulation of anthocyanin pathway in Arabidopsis and was used as a positive control in this study. Error bars are the standard error (SE) for three independent experiments with at least four replicate reactions. Statistical significance was determined using one‐way ANOVA; significant differences between means (LSD, P < 0.05) are shown as different lowercase letters above the bars.

VvWDR1 enhances anthocyanin biosynthesis by interacting with the VvMYBA2r‐VvMYCA1 complex

The function of VvMYBA2r and VvMYBA2w in regulating anthocyanin biosynthesis was determined by inserting the ORFs of VvMYBA2r and VvMYBA2w into the transient expression vector pEAQ‐HT and transiently infiltrating into the leaves of N. tabacum via Agrobacterium‐mediated transformation. The tobacco leaves showed no colour change 12 days after infiltration in leaf patches with VvMYBA2w alone, as well as after co‐infiltration with VvMYBA2w and VvMYCA1. Anthocyanin accumulation was not observed after eight days of infiltration of leaf patches with the ternary complexes containing VvMYBA2w, VvMYCA1 and VvWDR1; however, tobacco leaf chlorosis was observed in the leaf patches 12 days after infiltration. The tobacco leaves showed no colour change four days after infiltration with VvMYBA2r alone, but chlorosis was observed in the infiltrated leaf patches after six days. When VvMYBA2r and VvMYCA1 were overexpressed together, tobacco leaf chlorosis was observed in the infiltrated leaf patches at four days, which deepened from six to twelve days. Interestingly, when VvWDR1 was infiltrated into the leaves together with VvMYBA2r and VvMYCA1, stronger and earlier anthocyanin accumulation was observed (Figure S5). Thus, the Y2H, BiFC and tobacco transient expression assays suggested that VvWDR1 could enhance anthocyanin biosynthesis by interacting with the VvMYBA2r‐VvMYCA1 complex.

VvMYBA2w specifically interacts with VvSAP5 and VvUBE2A in vitro and in vivo

To elucidate the potential functions of VvMYBA2w, a Y2H assay was performed to screen the potential proteins that may function in a complex with VvMYBA2w. The transcriptional activation assay displayed that the full‐length VvMYBA2w could not activate the reporter gene, suggesting that VvMYBA2w could not autoactivate itself (Figure 4). Therefore, a bait (BD‐VvMYBA2w) was used for screening a grape cDNA library. In all, 42 potential VvMYBA2w‐interacting candidates were obtained. After sequencing, 13 potential VvMYBA2w‐interacting proteins, including ripening‐related proteins (VvGrip22, VvPP2C and VvGFH2), anthocyanin biosynthesis‐related enzyme (VvCYP93A3), a receptor‐like kinase (VvTMK4) and stress‐associated proteins (VvSAP5, VvCAT and VvPPO), were targeted (Table S2), because of the presence of more than one copy for the candidate proteins. Six interesting genes (VvSAP5, VvCACYBP, VvGFH2, VvPP2C, VvUBE2A and VvGrip22) obtained from Y2H screening were ligated into pGADT7 (AD) vector and co‐transformed with BD‐VvMYBA2w to yeast Y2HGold cells for the verification of protein–protein interactions (Figure S6A). The yeast cells of the positive control (pGBKT7‐53 plus pGADT7‐T) and co‐transformants with BD‐VvMYBA2w plus either AD‐VvUBE2A or AD‐VvSAP5 could grow on the QDO medium containing 125 ng/mL AbA and turn blue in the presence of X‐α‐Gal (Figure S6B). However, the yeast cells co‐transformed with BD‐VvMYBA2r plus either AD‐VvUBE2A or AD‐VvSAP5 could not grow on the QDO medium, indicating that VvMYBA2r could not interact with VvSAP5 and VvUBE2A in vitro (Figure S7). In addition, the yeast cells harbouring the controls (pGBKT7‐lam plus pGADT7‐T, BD‐VvMYBA2w plus AD, BD plus AD‐VvUBE2A, BD plus AD‐VvSAP5 and BD plus AD) could not grow on the QDO medium (Figure S6B). Subsequently, the BiFC assay was further performed to confirm the results of the Y2H assay. YFP fluorescence was observed in the cell nuclei of N. benthamiana leaf epidermis co‐transformed with vectors containing either YNE‐VvMYBA2w plus YCE‐VvSAP5 or YNE‐VvMYBA2w plus YCE‐VvUBE2A (Figure S6C). In contrast, no YFP fluorescence was noted in the leaf epidermal cells co‐transformed with the controls (YNE plus YCE, YNE plus YCE‐VvSAP5, YNE plus YCE‐VvUBE2A and YNE‐VvMYBA2w plus YCE). Thus, both Y2H and BiFC assays showed that VvMYBA2w specifically interacted with VvSAP5 and VvUBE2A in vitro and in vivo.

Rapid identification of genotypes and haplotype compositions of MYBA1 and MYBA2 loci in 213 grape germplasms

To explore fruit colouring characteristics of grapevine (Vitis spp.), the pericarp colour at the fruit maturation stage of 213 grape germplasms preserved at the Zhengzhou National Grape Germplasm Resources Garden of Chinese Academy of Agricultural Sciences (CAAS) was investigated at the fruit maturation stage. The pericarp colour at maturity was classified into seven categories: greenish‐yellow, yellowish‐green, pink, red, purplish‐red, purplish‐black and bluish violet (Figure S8), among which greenish‐yellow and yellowish‐green were defined as white. Among the germplasms, the number of purplish‐red grapes was the largest with 65 members, such as, ‘Muscat Humburg’ and ‘Damina’. Moreover, 49, 36, 22, 21, 12 and 8 germplasms were annotated as purplish‐black, greenish‐yellow, yellowish‐green, red, pink and bluish violet, respectively.

Based on the investigation of the colour traits of 213 grape germplasms, the genotypes of the MYBA1 and MYBA2 loci were identified rapidly by PCR using specific primer sequences. The haplotype compositions of the MYBA1 and MYBA2 loci of 211 grape germplasms except ‘Yuanruihei’ and ‘Olarra Queen’ could be directly determined on the basis of the identification results of their genotypes. The genotypes of MYBA1 (VvmybA1a and VvmybA1c) and MYBA2 (VvMYBA2r and VvMYBA2w) loci were heterozygous in both ‘Yuanruihei’ and ‘Olarra Queen’. Thus, their haplotype compositions may be either ‘phased phase’ HapA/HapC‐N (AC‐N) or ‘repulsive phase’ HapG/HapC‐Rs (GC‐Rs) (Figure S9). In view of the close linkage between MYBA1 and MYBA2 loci, as well the difference of MYB gene separation after self‐crossing in the ‘phased phase’ and ‘repulsive phase’ varieties, self‐crossing populations of ‘Yuanruihei’ and ‘Olarra Queen’ were generated for identifying the haplotype compositions of ‘Yuanruihei’ and ‘Olarra Queen’. Furthermore, the genotypes and haplotype compositions of the self‐crossing populations of ‘Yuanruihei’ and ‘Olarra Queen’ were identified rapidly by PCR using specific primer sequences (Figures S10 and S11). In the self‐crossing population of ‘Yuanruihei’, nine seedlings only contained HapA; eight, only HapC‐N; and 16, both HapA and HapC‐N (Table S3). In the self‐crossing population of ‘Olarra Queen’, six seedlings only contained HapA; seven, only HapC‐N; and 12, both HapA and HapC‐N (Table S4). Therefore, the haplotype composition of ‘Yuanruihei’ and ‘Olarra Queen’ was thought to be both HapA/HapC‐N (AC‐N, abbreviated form). Thus, the haplotype compositions of MYBA1 and MYBA2 loci were determined for all the 213 germplasms, and 18 haplotype composition types (A, AB, AC‐Rs, AC‐RsE1, AC‐RsE2, AE1, AE1E2, AE2, AF, CN, C‐NE2, C‐Rs, C‐RsE1E2, F, G, GC‐N, GF and AC‐N; abbreviated form) were identified (Table S5).

Statistical analysis indicated that the prediction results of berry colour according to the genotypes and haplotypes corresponded to their actual berry colour in 99.1% identified germplasms. The relationship analysis between haplotype compositions and berry colours showed that the haplotype compositions consisted of HapB or HapC‐Rs tended to confer red or purplish‐red colour to grape berry, whereas HapE2 or HapC‐N appeared in high frequencies in the black‐skinned grape varieties (Figure 7, Table S6). With regard to ploidy, the higher the number of functional alleles was present in a variety, the darker was the grape skin colour. The haplotype compositions of V. vinifera were found to be different from those for hybrids of V. vinifera and V. labrusca, and the major haplotype composition in coloured grapes from V. vinifera was AC‐Rs (46.23%), whereas the dominant haplotype compositions in the hybrids of V. vinifera and V. labrusca were AE1 (21.90%) and AE1E2 (31.43%; Table S7).

Figure 7

The relationship between the haplotype composition of MYBA1 and MYBA2 loci and fruit colour in grapevines. (a) Type and proportion of haplotypes in V. vinifera. (b) Type of fruit colour of each haplotype composition in V. vinifera. (c) Type and proportion of haplotypes in the hybrids of V. vinifera and V. labrusca. (d) Type of fruit colour of each haplotype composition in the hybrids of V. vinifera and V. Labrusca. (e) Relationship between skin coloration and haplotype compositions of MYBA1 and MYBA2 loci in the 213 grape germplasms.

Investigation of genetic separation of MYB haplotype regulating fruit coloration using interbreeding populations

The genetic separation of MYB haplotype was investigated using two interbreeding populations (‘Muscat Humburg’× ‘Crimson seedless’ and ‘Qiuhongbao’ × ‘Cuibao seedless’) in this study. The haplotype composition of three red‐skinned parents (‘Muscat Humburg’, ‘Crimson seedless’ and ‘Qiuhongbao’) was HapA/HapC‐Rs (AC‐Rs, abbreviated form), whereas that of ‘Cuibao seedless’ was homozygous HapA (HapA/HapA; Table S5). According to Mendel’s law, the expected segregation of ‘Muscat Humburg’ (2N; HapA/HapC‐Rs) × ‘Crimson seedless’ (2N; HapA/HapC‐Rs) would be 1 HapA/HapA:2 HapA/HapC‐Rs:1 HapC‐Rs/HapC‐Rs. Similarly, the progenies from ‘Qiuhongbao’ (2N; HapA/HapC‐Rs) and ‘Cuibao seedless’ (2N; HapA/HapA) cross would be segregated as 1 HapA/HapA:1 HapA/HapC‐Rs. After that, the genotype and haplotype compositions of two interbreeding populations without the knowledge of berry colour were identified using PCR analysis (Tables S8 and S9; Figures S12 and S13). In the interbreeding population of ‘Muscat Humburg’ × ‘Crimson seedless’, 18, 32 and 19 individuals, respectively, had haplotype composition of HapA/HapA, HapA/HapC‐Rs and HapC‐Rs/HapC‐Rs. Similarly, in the interbreeding population of ‘Qiuhongbao’ × ‘Cuibao seedless’, 25 and 38 progenies contained HapA/HapA (A, abbreviated form) and HapA/HapC‐Rs, respectively. The chi‐square tests revealed that the actual segregation ratios of two interbreeding populations fit the expected distributions (Table 1). The berry skin colour was predicted according to the genotypes and haplotype compositions. To verify this, the berry colour of the progenies was visually assessed at the harvest period (Figures S14 and S15). The berry colour of the above‐mentioned interbreeding populations was consistent with that predicted based on the haplotype compositions of MYBA1 and MYBA2 loci, implying the feasibility and authenticity of the early prediction of berry colour during the classical cross‐breeding, which would be useful in the molecular breeding for grape skin colour traits.

Table 1

Segregation of the haplotypes in the hybrid progeny of the ‘Muscat Humburg’ × ‘Crimson seedless’, and ‘Qiuhongbao’ × ‘Cuibao seedless’ cross

Parents	Haplotype composition	Total number of progenies	Number of progenies			Expected ratio	x 2 value	P‐value
‘Muscat Humburg’ (♀) × ‘Crimson seedless’ (♂)	HapA/HapC‐Rs × HapA/HapC‐Rs	69	HapA/HapA	HapA/HapC‐Rs	HapC‐Rs/HapC‐Rs	1:2:1	0.391	0.82
			18	32	19
‘Qiuhongbao’ (♀) × ‘Cuibao seedless’ (♂)	HapA/HapC‐Rs × HapA/HapA	63	HapA/HapA		HapA/HapC‐Rs	1:1	2.29	0.13
			25		38

Discussion

Amino acid sequence features of VvMYBA2r, VvMYBA2w and other R2R3‐MYB proteins within the anthocyanin‐related clade

In A. thaliana, anthocyanin biosynthesis is regulated by four MYB TFs belonging to the anthocyanin‐related clade; all of them act at different developmental stages and in response to various stimuli (Dubos et al., 2010; Gonzalez et al., 2008). In grapevines, at the onset of fruit maturation, VvMYBA1 and VvMYBA2 directly switch on the expression of the VvUFGT gene (Rinaldo et al., 2015; Walker et al., 2007). Our findings showed that VvMYBA2r and VvMYBA2w contained an N‐terminal R2R3 repeat that corresponded to the DNA‐binding domain of MYB TFs (Figure 1). Similar to that in more than 100 members of the MYB TF family in Arabidopsis as well as in VvMYBA1, VlmybA1‐2 and VlmybA2 from grapevine, the R2R3 domain within VvMYBA2r and VvMYBA2w was highly conserved and contained the motif [D/E]Lx2[R/K]x3Lx6Lx3R for the interaction with bHLH proteins (Zimmermann et al., 2004), whereas their C‐terminal region showed little homology with other MYB TFs (Stracke et al., 2001). Regardless of the C‐terminal diversity, the MYB TF families in rice and Arabidopsis were classified into subgroups on the basis of conserved motifs located in the C‐terminus of the MYB TFs (Jiang et al., 2004). Sequence alignment showed that VvMYBA2r and VvMYBA2w possessed the characteristic motif 6 (KPRPR[S/F]F), which is often found in MYB TFs regulating anthocyanin biosynthesis. Sequence similarity among MYB TFs is generally restricted to the R2R3 domain, but some MYB TFs share conserved motifs in their C‐terminus, suggesting that they might have similar functions (Stracke et al., 2001). Our analyses clearly supported the classification of VvMYBA2r and VvMYBA2w and other putative R2R3‐MYB TFs regulating anthocyanin biosynthesis from different plant species. Taken together, the VvMYBA2r and VvMYBA2w protein sequences show the typical characteristics of plant R2R3‐MYB TFs.

VvMYBA2r specifically activates anthocyanin biosynthetic genes in grapevine

This study showed that the constitutive overexpression of VvMYBA2r, an R2R3‐MYB TF from grapevine berry, resulted in obvious phenotypic changes in tobacco flower, especially in the corollas and stamens. This may be due to the fact that anthocyanin biosynthesis sub‐pathway was affected with a significant increase of anthocyanin accumulation in transgenic plants overexpressing VvMYBA2r. Indeed, our findings indicated that the overexpression of VvMYBA2r specifically induced the upregulation of several structural genes involved in anthocyanin biosynthesis, such as UFGT, CHS, CHI and OMT. In Arabidopsis, the ectopic expression of the R2R3‐MYB TF AtPAP1 resulted in the upregulation of CHS, PAL and DFR expression, which was sufficient to improve the production of anthocyanin compounds (Borevitz et al., 2000). However, unlike AtPAP1 in Arabidopsis, which appeared to be active evenly in all reproductive and vegetative organs when overexpressed (Borevitz et al., 2000), the phenotypic modification of VvMYBA2r‐overexpression mainly occurred in the reproductive organs of transgenic tobacco. This may be attributed to the difference in capacity of the two R2R3‐MYB regulators in activating anthocyanin biosynthesis. Walker et al. (2007), as well as Kobayashi et al. (2002, 2004), have indicated that VvMYBA1 specifically regulates UFGT gene expression in ripening grapevine berries. Our dual‐luciferase assay results also showed that VvMYBA2r could activate the promoter of VvUFGT in the absence or the presence of a bHLH partner. Data presented in this study depicted that the spatio‐temporal expression of VvMYBA2r, combined with the action of specific regulators such as VvMYBA1, VvMYB5a, VvMYB5b, VvMYBPA2, VvMYBPA1 and VvMYBPAR, controls the biosynthesis of both anthocyanin and proanthocyanidin (PA) throughout grapevine berry development (Figure 8). Moreover, considering the findings of Walker et al. (2007), our results suggested that VvMYBA2r encoded functional proteins and played an important role in regulating anthocyanin biosynthesis in grape berry. In contrast, VvMYBA2w lost its ability to act as a regulator to switch on anthocyanin biosynthesis owing to two non‐synonymous mutations, one of which leads to an R44L substitution and another to a frame shift, resulting in a smaller protein.

Figure 8

Summary of the possible implication of R2R3‐MYB transcription factors in the regulatory mechanisms of the flavonoid pathway during grape development. The roles of VvMYBA2r, VvMYB5a, VvMYB5b, VvMYBPA1, VvMYBPA2, VvMYBPAR and VvMYBA1 were assigned according to gene expression levels during berry development together with functional characterization data presented in this article for VvMYBA2r, in Deluc et al. (2006) for VvMYB5a, in Deluc et al. (2008) for VvMYB5b, in Bogs et al. (2007) for VvMYBPA1, in Terrier et al. (2009) for VvMYBPA2, in Koyama et al. (2014) for VvMYBPAR and in Walker et al. (2007) for VvMYBA1. According to several previous studies, VvMYB5a, VvMYB5b, VvMYBPA1 and VvMYBPA2 appear to be particularly involved in the regulation of PA synthesis before veraison in seed and skin tissues. At veraison, PA synthesis is complete, and anthocyanin synthesis begins in pericarp cells where UFGT gene expression is specifically regulated by VvMYBA1 and VvMYBA2r, whereas genes encoding enzymes of the general flavonoid pathway required for anthocyanidin synthesis appear to be regulated by VvMYB5b. Abbreviations are as follows: PAL, phenylalanine ammonia‐lyase; C4H, Cinnamate 4‐hydroxylase; 4CL, 4 coumarate: CoA ligase; CHS, chalcone synthase; CHI, chalcone isomerase; F3H, flavanone‐3‐beta‐hydroxylase; F3’H, flavonoid 3′‐hydroxylase; F3’,5’H, flavonoid 3′5′‐hydroxylase; FLS, flavonol synthase; DFR, dihydroflavonol 4‐reductase; LAR, leucoanthocyanidin reductase; ANR, anthocyanidin reductase; LDOX, leucoanthocyanidin dioxygenase; UFGT, UDP glucose‐flavonoid 3‐O‐glucosyltransferase; OMT, O‐methyltransferase; SK, shikimate kinase; and DHD/SDH, 3‐dehydroquinate dehydratase/shikimate 5‐dehydrogenase. The values, calculated as log2 gene expression levels throughout berry development, are represented as blue, green, yellow and red. The berry samples were harvested at 15, 30, 45, 60, 75 and 95 days after anthesis (DAA).

R44L substitution and CA dinucleotide deletion both seriously impair the ability of VvMYBA2w to regulate anthocyanin biosynthesis

Transcriptional activation assay showed that CA dinucleotide deletion in VvMYBA2w led to a frame shift that shortened the C‐terminal transcriptional activation region and disrupted the transcriptional activation activity of VvMYBA2w (Figure 4). In addition, yeast cells transformed with VvMYBA2rM could grow on the SD/‐His‐Ade medium and turn blue in the presence of X‐α‐Gal as well as could strongly activate the transcription of the LacZ gene, indicating that it still possessed strong transactivation activities like VvMYBA2r. However, a previous study showed that R69L substitution in VvMYB5b markedly reduced its transactivation capacity (Hichri et al., 2011). This phenomenon can be attributed to the difference in the aa substitutions at positions 44 and 66. Our findings showed that R44L substitution modifies the interaction of VvMYBA2w with VvMYCA1 protein, which subsequently impacts on the regulation of VvUFGT gene expression. Likewise, the R69L substitution within the R2 domain of grapevine VvMYB5b has a severe impact on the cooperative effect of VvMYB5b with its bHLH partner (Hichri et al., 2011). In maize, aa substitutions within the DNA‐binding domain of the MYB transcription factor ZmP1 also strongly influenced the cooperative effect of ZmP1 with its partners (Hernandez et al., 2004). Hichri et al. (2011) reported that an R69L substitution disrupted the ability of VvMYB5b to cause transcriptional activation of the VvCHI promoter in the presence of two interacting proteins AtEGL3 and AtTTG1 from Arabidopsis. Our data indicated that the R44L substitution and CA dinucleotide deletion both seriously impaired the ability of VvMYBA2r to activate the transcription of VvUFGT in the presence or absence of the bHLH partner, although neither had such a severe effect as VvMYBA2r or VvMYBA2w, indicating that two alterations might be additive in their effects. Furthermore, the ectopic expression of VvMYBA2r, VvMYBA2r, VvMYBA2r, VvMYBA2r and VvMYBA2w further supported the above conclusion. Therefore, investigation of the differences in VvMYBA2r and VvMYBA2w expression and/or transcriptional activity in different grapevine cultivars might allow the development of molecular markers for the breeding of grapevines with optimized anthocyanin composition and content.

VvMYBA2r positively regulates anthocyanin biosynthesis by forming the VvMYBA2r‐VvMYCA1‐VvWDR1 transcription complex in grapevine

The MBW transcription complex participates in the regulation of anthocyanin biosynthesis pathways, whereas MYB TFs are the vital factors regulating the anthocyanin biosynthesis, which could be precisely bound to a specific DNA sequence (Dubos et al., 2010; Jia et al., 2016). Our findings showed that VvMYBA2r and VvWDR1 both interacted with VvMYCA1 protein, implying that the MBW transcription complex was formed by VvMYBA2r, VvMYCA1 and VvWDR1 in grapevine (Figure 5). In the same species, the MBW complex regulates different synthetic pathways of flavonoids owing to the differences in the composition of members. For example, in Arabidopsis, the PAP1/PAP2‐TT8‐TTG1 complex regulates the synthesis of anthocyanins, whereas the TT2‐TT8‐TTG1 complex regulates the synthesis of procyanidins by activating BAN expression (Baudry et al., 2004; Gonzalez et al., 2008). In Petunia, the PhPH4‐PhAN1‐PhAN11 complex mainly participates in the regulation of cellular vacuolar acidity, whereas the PhAN2‐PhAN1‐PhAN11 complex is involved in the regulation of anthocyanin biosynthesis (Quattrocchio et al., 2006; Spelt et al., 2002). Similar to that in Arabidopsis and Petunia, the VvMYBA2r‐VvMYCA1‐VvWDR1 complex participates in the regulation of anthocyanin biosynthesis in grapevine, whereas the VvMYBA2‐VvMYC1‐VvWD40 complex is involved in the regulation of PA synthesis (Hichri et al., 2010). In Chinese bayberry, MrWD40‐1 interacted with each of MrMYB1 and MrbHLH1, whereas no interaction was noted between MrMYB1 and MrbHLH1(Liu et al., 2013). However, in this study, VvMYCA1 interacted with each of VvMYBA2r and VvWDR1 in grapevine, similar to that in apple (An et al., 2012). This difference in interactions may be attributed to the difference in species used in these studies. The tobacco transient expression system used in this study also showed that the accumulation of anthocyanins induced by the VvMYBA2r‐VvMYCA1‐VvWDR1 complex was significantly higher than that of induced by VvMYBA2r‐VvMYCA1. Therefore, the results indicated that the VvMYBA2r‐VvMYCA1‐VvWDR1 complex positively regulates anthocyanin biosynthesis, and VvWDR1 could further enhance the regulation of anthocyanin biosynthesis by interacting with the VvMYBA2r‐VvMYCA1 complex, which is consistent with the findings of Baudry et al. (2004) and Hichri et al. (2011).Thus, in grapevine, VvMYBA2r positively regulates anthocyanin biosynthesis by forming an MBW transcription complex with VvMYCA1 and VvWDR1 (Figure 9).

Figure 9

The mechanism model of VvMYBA2r and VvMYBA2w in regulating anthocyanin biosynthesis. In the presence of developmental signal, VvMYBA2r transcript was up‐regulated under the control of MYB transcriptional activator. Subsequently, VvMYBA2r along with VvMYCA1 and VvWDR1 formed the MBW transcriptional complex, which could activate the promoter of VvUFGT gene, which is specifically involved in anthocyanin synthesis, resulting in anthocyanin accumulation of berry skin. However, VvMYBA2w lost the capacity of the regulator to switch on anthocyanin biosynthesis owing to two non‐synonymous mutations (R44L substitution and CA dinucleotide deletion). The R44L substitution abolishes the interaction of VvMYBA2w with VvMYCA1 partner and CA dinucleotide deletion disrupts the transcriptional activation activity of VvMYBA2w. The semi‐elliptical arch with broken lines indicates R44L substitution leading to a frame shift resulting in an untranslated region (UTR). Grey rectangle indicates the promoter of grapevine VvUFGT gene. Amaranthine star indicates the R44L substitution within VvMYBA2w. Blue star indicates arginine residue at position 44 (R44) of VvMYBA2r.

Possible biological significance of the interactions of VvUBE2A and VvSAP5 with VvMYBA2w

The biological function of VvMYBA2w was further analysed by screening the possible proteins interacting with VvMYBA2w using Y2H and BiFC assay. VvMYBA2w was found to interact with VvUBE2A and VvSAP5 proteins, whereas VvMYBA2r did not, suggesting that both VvUBE2A and VvSAP5 are specific to the VvMYBA2w protein. This difference in interacting partners may be attributed to the presence of the two mutations that lead to changes in the spatial conformation of VvMYBA2w, unlike that in VvMYBA2r protein. We postulated that VvMYBA2w may be involved in the ubiquitination pathway mediated by the VvSAP5 and VvUBE2A proteins, implying that VvMYBA2w potentially played an important role in stress tolerance by regulating the protein degradation pathway. Anthocyanin accumulation is an important biological response of plants against stress, which is resulted from anthocyanin having light‐absorbing property and scavenging free radicals, so as to protect plants from being damaged by photoinhibition under strong light conditions and the increase of free radicals under adverse conditions (Dixon and Steele, 1999; Espley et al., 2007; Gould et al., 2010). Many studies have shown that SAP5 possesses E3 ubiquitin ligase activity that could hasten substrate protein degradation via the 26S protein complex by linking the ubiquitin proteins to the substrate proteins (Kang et al., 2011; Pilarski, 2008). By degrading related negative regulatory proteins, SAP5 indirectly activates relevant stress‐resistance pathways in vivo, such as the ABA‐signalling pathway, thereby protecting plants from abiotic stresses such as drought, osmotic stress, cold injury, salt stress and water deficit. In AtSAP5‐overexpressing Arabidopsis plants, AtSAP5 is highly co‐expressed with MYB75/PAP1, MYB90/PAP2 and DFR. Therefore, SAP5 is closely related with the anthocyanin synthesis pathway. Thus, VvMYBA2w, which interacts with VvSAP5 and VvUBE2A in grapevine, possibly participates in abiotic stress regulation by regulating anthocyanin synthesis or other stress‐related pathways. In addition, the yeast library screening showed that VvMYBA2w may interact with the protein kinase VvTMK4, suggesting that it may be regulated by the protein phosphorylation pathway. Moreover, the Y2H assay showed that VvMYBA2w could interact with polyphenol oxidase VvPPO; however, the related biological functions need to be investigated in the future.

Investigation of the relationship between haplotype composition and skin colour

In this study, several types of haplotypes in the loci controlling berry colour diversity were identified in the various Vitis cultivars. In all, 18 haplotype compositions were identified in the 213 grape germplasms. HapE was only detected in V. labrusca and its hybrids, suggesting that it originated from V. labrusca. Grape berries with HapC‐Rs were found to be mainly red or purplish‐red, whereas those with HapE2 or HapC‐N tended to be black. These results are consistent with the findings of Fournier‐Level et al. (2009, 2010) and Carrasco et al. (2015). The discrepancy in colour between HapC‐N and HapC‐Rs grapes could be explained by the numbers of functional MYB genes presenting in each haplotype. In HapE2 (VlmybA2 and VlmybA1‐3) and HapC‐N (VvmybA1c and VvMYBA2r), two functional MYB genes are present, whereas HapC‐Rs only contained one functional MYB gene (VvmybA1c). Thus, grape berries with two functional alleles (HapC/HapE) were found to show deeper skin colour than those with only a single functional allele (e.g. HapA/HapE or HapA/HapC; Figure 7). These findings indicate that the type and number of functional haplotypes at the colour locus are the main genetic determinants of skin colour variation.

V. vinifera has been ecogeographically categorized into three groups: orientalis, occidentalis and pontica (Negrul, 1938). The microsatellite marker analysis of the orientalis and occidentalis cultivars depicted a clear separation in a dendrogram on the basis of phenetic distance (Goto‐Yamamoto et al., 2006). HapF was found to be mostly present in the orientalis cultivars, including ‘Hetianhong’, ‘Lizixiang’, ‘Longyan’, ‘Lvnai’, ‘Manai’, ‘Manao’, ‘Mudanhong’, ‘Niuxin’ and ‘Pinger’, indicating that it might have originated from the orientalis cultivars. Whether VvmybA1 regulates anthocyanin biosynthesis has not yet been confirmed, because it has been detected in both white‐skinned (‘Lvnai’, ‘Manai’, ‘Manao’, and ‘Niuxin’) and red‐skinned varieties (‘Hetianhong’, ‘Lizixiang’, ‘Longyan’, ‘Pinger’ and ‘Mudanhong’). The VvmybA1 sequences in the above nine orientalis cultivars were identical, suggesting that the diversity of grape skin colour could not be explained by sequence variation in the coding region. These results indicated the existence of either other regulatory loci that control grape berry skin colour or a transposon insertion in the upstream promoter region of VvmybA1. Therefore, it is necessary to confirm whether the structural genes of VvmybA1 function in anthocyanin accumulation and to compare the uncovered upstream promoter region of VvmybA1 in the four white‐skinned varieties with that in the five red‐skinned varieties. In addition to the VvmybA and VlmybA alleles, some paralogous genes might exist at positions adjacent to the VvmybA1 locus and might regulate anthocyanin biosynthesis. Walker et al. (2006) speculated that there might be many more types of functional MYB alleles than non‐functional MYB alleles at the berry colour locus. Furthermore, Ban et al. (2014) reported that two novel loci (A8‐QTL2 and A14‐QTL) could affect anthocyanin content in berry skin. Therefore, Vitis species may have undiscovered functional alleles and haplotypes. Further studies with a wider range of germplasms, including East Asian wild and native American grape germplasms, should be unveiled to obtain a more comprehensive understanding of these alleles and haplotypes. The detailed analysis of the diversity of alleles and haplotypes at the berry colour locus will contribute to elucidating the origin and evolution of Vitis species.

Origin and evolution of white‐skinned grapes

The reasons for colour differences between the red‐ and white‐skinned grapes have been elucidated; hence, we could suggest how white‐skinned grapes were derived from red‐skinned grapes. In the present study, the most ancestral haplotype (HapC‐N) was found in several wine grape cultivars, for instance, ‘Cabernet Gernischet’ and ‘Heijianiang’, which is consistent with results of Carrasco et al. (2015). Most of the white‐skinned grapes surveyed in this study included VvmybA1a and VvMYBA2w, which were inactive genes because of the mutations in VvMYBA1 and VvMYBA2 loci, implying the presence of a common ancestor. The mutations of VvMYBA2 and retrotransposon insertion of VvMYBA1 leading to the inactivation of the two genes were the first steps to the formation of white‐skinned berries in grapevines, although, at this stage, there is no evidence to indicate which event occurred first. In this study, the homozygotes of HapC‐Rs and HapG, which inactivated one VvMYBA locus, were identified. The ancestral grapevine might be considered to be heterozygous containing both HapA and HapC‐N. The segregation of the functional haplotype (HapC‐N) and non‐functional haplotype (HapA) during sexual reproduction might have resulted in white‐fruited seedlings, the antecedents of modern white cultivars (Figure 10). The sequencing results of VvMYBA2 locus from 51 white‐skinned grapevines were identical, implying a common origin – an offspring from the heterozygous red‐skinned grapevine. In addition, the genotype analyses showed that 202 of the 213 grapevine germplasms identified in this study contained the VvMYBA2w allele, instead of the VvMYBA2r allele, further indicating that all extant white‐skinned grapevine varieties might have a common ancestor. The unlikely occurrence of unlinked mutational events in the adjoining MYB loci (VvMYBA1 and VvMYBA2) might have been a lucky coincidence for some white wine drinkers and the majority of consumers of white‐skinned table grapes worldwide. Thus, in‐depth analysis of these MYB gene functions will contribute to provide the unique opportunity of utilizing them to generate exciting and interesting new grapevine varieties.

Figure 10

Model for the origin and evolution of white‐skinned grape genes. After the duplication of the VvMYBA genes, VvMYBA2r and VvmybA1c were both functional genes resulting in a red grapevine. The presentation of two mutations in VvMYBA2w and the insertion of a retrotransposon in the promoter of VvmybA1c resulted in the inactivation of two genes; however, the colour of the berries would still be red until sexual reproduction, resulting in a homozygous genotype when white grapevines would finally be observed. Black star indicates the R44L substitution mutation site within VvMYBA2w. White star indicates the CA dinucleotide deletion mutation site within VvMYBA2w. Gret1: grapevine retrotransposon1.

Experimental procedures

Plant materials and growth conditions

Young leaves of 213 grapevine germplasms were collected from the Zhengzhou National Grape Germplasm Resources Garden of CAAS, Zhengzhou, China. A total of 69 four‐year‐old progenies of ‘Muscat Humburg’ (V. vinifera; red‐skinned) × ‘Crimson seedless’ (V. vinifera; red‐skinned) cross and 63 four‐year‐old progenies of ‘Qiuhongbao’ (V. vinifera; red‐skinned) × ‘Cuibao seedless’ (V. vinifera; white‐skinned) cross were collected from the Taigu National Grape Germplasm Resources Garden of Shanxi Academy of Agricultural Sciences (SAAS), Taigu, China. The skin colour of the progenies of the two interbreeding populations was visually assessed at the time of harvest.

The grapevine varieties (‘Heimeiren’ and ‘Shine Muscat’) were used as experimental material; they were grown under standard field conditions at the Nanjing Agricultural University Farm, Nanjing, China. Veraison was determined as the time when the berry obviously changed colour and began to soften. The maturity date was determined on the basis of the seed colour change to dark brown without senescence of berry tissue and previous maturity date records. The fruit samples were harvested at 15 (young fruit stage), 30 (fruit core‐hardening stage), 45 (before berry veraison), 60 (berry veraison), 75 (berry veraison later) and 95 (berry maturation) DAA, respectively. Three biological replicates each consisting of three clusters were sampled at each sampling date. Berry development phases were defined according to the criteria considering soluble sugar titres, size, colour and softening (Boss et al., 1996). Flesh and skin samples were obtained by peeling the grape berries. All samples were rapidly frozen in liquid nitrogen and kept at −80 °C until further analysis.

The tobacco (N. tabacum L. cv. ‘Xanthii’) seeds were sterilized in 70% ethanol for 2 min, incubated in 2.5% potassium hypochlorite solution for 10 min, and finally washed three times with sterile water. Subsequently, tobacco seeds were germinated under tissue culture conditions at 23 °C with an 8‐h‐dark/16‐h‐light cycle on Murashige and Skoog (MS) medium (Murashige and Skoog, 1962). Tobacco leaves were collected and used for transformation and regeneration, according to Burrow et al. (1990). N. tabacum and N. benthamiana were grown in a greenhouse at 25 °C with an 8‐h‐dark/16‐h‐light cycle.

Isolation of VvMYBA2r and VvMYBA2w and bioinformatics analysis

Based on the sequence from National Center for Biotechnological Information (NCBI) database, we amplified the full‐length ORFs of VvMYBA2r and VvMYBA2w using PCR with gene‐specific primers as described in Table S10. PCR was performed under the following conditions: 94 °C for 5 min, followed by 35 cycles of 94 °C for 30 s, 58 °C for 30 s, 72 °C for 90 s, with a final extension of 72 °C for 10 min. The PCR product was purified, subcloned into the pMD19‐T vector (Takara, China), and subsequently transformed into Escherichia coli DH5a. Positive colonies were selected and sequenced at Tsingke Biological Technology (Nanjing, China). The alignment and analysis of the VvMYBA2w and VvMYBA2r sequences were performed using the BLAST tool (Altschul et al., 1990; Jiu et al., 2018) at the NCBI (http://www.ncbi.nlm.nih.gov). Multiple alignments of the aa sequences in different species were completed using the ClustalW program (Ramu et al., 2003) and GeneDoc software (version 1.6). Phylogenetic tree was constructed using MEGA 7.0 software with the neighbour‐joining method and bootstrap analysis (1,000 replicates; Tamura et al., 2011). The Mw and pI were predicted using the Expert Protein Analysis System (http://web.expasy.org/compute_pi/). The structure of the R2R3 domain in VvMYBA2r/2w was modelled using the Phyre2 website (http://www.sbg.bio.ic.ac.uk/phyre2/html/page.cgi?id=index) using the Arabidopsis R2R3‐MYB TF as the template model (PDB code: 6kks).

DNA and RNA extraction

Total RNA was isolated from grapevine berries and tobacco corollas using the RNA extraction kit (Promega, Madison, WI), following manufacturer’s protocols (Jia et al., 2016). The RNA concentration was quantified using the NanoDropND‐1000 spectrophotometer (Thermo Scientific) and the quality was estimated by electrophoresis on an agarose gel (Jiu et al., 2019). To avoid genomic DNA contamination, RNA samples were digested with DNase I (TaKaRa, Japan). Subsequently, cDNA libraries were constructed using 1.0 μg of RNA samples according to manufacturer’s instructions (Revert Aid First Strand cDNA synthesis kit; Thermo). Genomic DNA was extracted from grapevine and tobacco leaves using the cetyltrimethylammonium bromide (CTAB) method (Murray and Thompson, 1980).

qRT‐PCR

qRT‐PCR analysis was performed on an ABI 7500 Real‐Time System (Applied Biosystems, Foster City, CA). The PCR mixture (20 µL) contained 10 µL of 2 × SYBR Green Premix Ex Taq, 2 µL cDNA template, and 0.4 µL each of 10 µM reverse specific primer and 10 µM forward specific primer. qRT‐PCR operating parameter was determined according to Jia et al. (2016). Each sample was analysed in three technical replicates, and the 2–ΔΔCT method described by Livak and Schmittgen (2001) was applied to calculate the relative expression levels of each gene. KyActin1 and NtTubA1 were analysed in parallel as internal reference controls for grapes and tobacco, respectively, to normalize the expression levels. The primers used for qRT‐PCR are listed in Table S11.

Subcellular localization of VvMYBA2r and VvMYBA2w

The full‐length cDNAs of VvMYBA2r and VvMYBA2w for subcellular localization were cloned using the primers (Table S10) containing BglII and SpeI restriction sites and inserted into a binary vector pCAMBIA1302 to generate two fusion constructs (35S:VvMYBA2r‐GFP and 35S:VvMYBA2w‐GFP, respectively). After the sequence was identified, two fusion constructs and the control vector (pCAMBIA1302) were transferred into A. tumefaciens strain GV3101 using heat shock. The abaxial surfaces of five‐week‐old N. benthamiana leaves were agroinfiltrated (Jiu et al., 2020; Kumar and Kirti, 2010) with the bacterial suspension (OD600 = 0.5) and then kept in an incubator for 72 h, followed by live‐cell imaging using an inverted confocal laser‐scanning microscope (Zeiss LSM 780, Germany).

Plant transformation

To generate the VvMYBA2r‐ and VvMYBA2w‐overexpression constructs, the full‐length coding sequence of VvMYBA2r and VvMYBA2w was amplified using the reverse transcript cDNA as the template and gene‐specific primers (Table S10) containing BglII and SpeI restriction sites. Another three mutant fragments (VvMYBA2r, VvMYBA2r and VvMYBA2r) were chemically synthesized by General Biosystems (Nanjing, China) based on the full‐length coding sequence of VvMYBA2r along with BglII and SpeI restriction sites. The first mutant construct VvMYBA2r generated by mutating a nucleotide of VvMYBA2r led to an aa change in arginine residue at position 44 altered to leucine (R44L) within the first MYB repeat. The second mutant construct VvMYBA2r generated by deleting the CA dinucleotide of VvMYBA2r altered the reading frame at aa 258 (Figure 3). The VvMYBA2r construct was also dual‐mutated to produce a third mutant construct VvMYBA2r containing two non‐synonymous SNP mutations as those in VvMYBA2w. Five fragments (VvMYBA2r, VvMYBA2r, VvMYBA2r, VvMYBA2r and VvMYBA2w) were ligated to the binary expression vector pCAMBIA1302, generating the pCAMBIA1302‐VvMYBA2r, pCAMBIA1302‐VvMYBA2r, pCAMBIA1302‐VvMYBA2r, pCAMBIA1302‐VvMYBA2r and pCAMBIA1302‐VvMYBA2w fusion plasmids. The constructs were transferred into the A. tumefaciens strain EHA105 using the freeze–thaw method (Jiu et al., 2016). Next, the A. tumefaciens‐positive clones were transformed into leaf discs of N. tabacum, and then the plantlets were regenerated, as described by Horsch et al. (1985). The transgenic tobacco plants were screened by PCR analysis with the primers listed in Table S10. The genomic DNA, as PCR templates, was extracted from putative transformants, using the WT as the control. For each construct, 15 transgenic lines were obtained for further analysis.

Tobacco transient expression assay

The plasmids used in the transient expression assay were constructed by ligating full‐length VvMYBA2r, VvMYBA2w, VvMYCA1 and VvWDR1 to pEAQ‐HT vector, which is usually used for transient transformation (Norkunas et al., 2018; Sainsbury and Lomonossoff, 2008). The primers used to amplify the coding region are listed in Table S10. The product was recombined with the linearized vector pEAQ‐HT. All constructs were transferred into A. tumefaciens GV3101 and then infiltrated into the abaxial leaf surface of N. tabacum together, either ternary or binary, or alone, for the induction of anthocyanins, as described by Sainsbury et al. (2009). In addition, the control was infiltrated with an empty vector (pEAQ‐HT). Digital photographs of anthocyanin development in the leaf patches were obtained at 2, 4, 6, 8 and 12 days after infiltration.

Transactivation activity assay in yeast

The coding regions of VvMYBA2w, VvMYBA2r and two truncated peptides (VvMYBA2rΔNterm and VvMYBA2rΔCterm), and three mutations (VvMYBA2r, VvMYBA2r, and VvMYBA2r) of VvMYBA2r were amplified using the primers (Table S10) containing BamHI and EcoRI restriction sites. The resulting fragments were ligated into the yeast expression vector pGBKT7 (Clontech, Mountain View, CA) to generate BD‐VvMYBA2w, BD‐VvMYBA2r, BD‐VvMYBA2rΔNterm, BD‐VvMYBA2rΔCterm, BD‐VvMYBA2rM, BD‐VvMYBA2rD and BD‐VvMYBA2rM+D fusion plasmids, for transactivation activity assay, which was performed following Clontech’s protocols. Either these constructs, pGBKT7 (as negative control) or pCL1 (as positive control) was transformed into Saccharomyces cereviseae strain Y2HGold (Clontech) according to manufacturer’s protocols. The transformants carrying either BD‐VvMYBA2w/2r/2rΔNterm/2rΔCterm/ 2rM/2rD/2rM+D or pGBKT7 were selected on the SD/‐Trp medium, whereas pCL1 transformants were selected on the SD/‐Leu medium (Chen et al., 2013; Xing et al., 2019; Zhang et al., 2019). After three days, single clones were transferred to SD/‐His‐Ade plates containing either 0 or 20 mm X‐α‐Gal (Zhang et al., 2019). The GAL4‐BD region can regulate the expression of His3 (Gao et al., 2012; Xu et al., 2007); hence, if VvMYBA2r/2w possessed activation ability, the BD‐VvMYBA2r/2w constructs would be expected to bind to the GAL4 BD upstream promoter sequence of His3, thereby activating its expression, and allowing the transformed cells to grow on the SD/‐His‐Ade medium.

Protein–protein interaction analysis and β‐galactosidase assay

In addition to BD‐VvMYBA2w, BD‐VvMYBA2r, BD‐VvMYBA2rΔNterm, BD‐VvMYBA2rΔCterm, BD‐VvMYBA2rM, BD‐VvMYBA2rD and BD‐VvMYBA2rM+D plasmids for transactivation activity assay, the full‐length cDNAs of VvWDR1 were also inserted into pGBKT7 to generate the BD‐VvWDR1 fusion plasmid, and the full‐length cDNAs of VvMYCA1 were ligated into pGADT7 to generate AD‐VvMYCA1. The pGBKT7‐53 plus pGADT7‐T and pGADT7‐lam plus pGADT7‐T plasmids were used as positive and negative controls, respectively. All constructs were transformed into Y2HGold using the PEG/LiAc method (Li et al., 2012). Yeast cells were cultured on the SD/‐Leu‐Trp (DDO) medium, following Clontech’s instructions. Transformed colonies were plated on the SD/‐Leu‐Trp‐His‐Ade (QDO) medium containing 125 ng/mL AbA with or without X‐α‐Gal. The β‐galactosidase assays with the ONPG substrate were then conducted following manufacturer’s protocols (BD MatchmakerTM Library Construction & Screening Kit; Clontech, BD Bioscience). Relative β‐galactosidase activity was obtained after normalization with the optical density at 600 nm.

Y2H screening of VvMYBA2w‐interacting proteins

Y2H assay was performed using the Matchmaker GAL4 two‐hybrid system, according to the Clontech’s protocols. In brief, full‐length cDNAs of VvMYBA2w with EcoRI and BamHI sites were ligated into pGBKT7, which was used as a bait to screen the grape cDNA library. The mated yeast cells were screened on the SD/‐Leu‐Trp‐His‐Ade (QDO) plates, and putative positive clones were grown on the QDO plates. The interaction between individual clones and VvMYBA2w was then confirmed by retransforming them back into yeast strain Y2HGold, selected as described above, and subjected to an X‐α‐Gal test for further confirming the protein–protein interaction. In addition, the interaction of two VvMYBA2w‐interacting proteins (VvUBE2A and VvSAP5) and VvMYBA2r was also detected in this study.

BiFC

The fragments of YNE‐VvMYBA2r, YNE‐VvMYBA2rM, YNE‐VvMYBA2rD, YNE‐VvMYBA2rM+D, YNE‐VvMYBA2w, YNE‐VvWDR1, YCE‐VvMYCA1, YCE‐VvSAP5 and YCE‐VvUBE2A were amplified for BiFC assay using the primers listed in Table S10. The fragments of YNE‐VvMYBA2r, YNE‐VvMYBA2rM, YNE‐VvMYBA2rD, YNE‐VvMYBA2rM+D, YNE‐VvMYBA2w and YNE‐VvWDR1 were inserted into pSPYNE‐35S (YNE) vector, whereas those of YCE‐VvMYCA1, YCE‐VvSAP5 and YCE‐VvUBE2A were subcloned into pSPYCE‐35S (YCE) vector, respectively. After the sequences were identified, the constructs and control vector (pSPYNE‐35S and pSPYCE‐35S) were transformed into A. tumefaciens strain GV3101 using heat shock. The abaxial surfaces of five‐week‐old N. benthamiana leaves were transiently co‐transformed using an A. tumefaciens infection method using different combinations of these constructs, as described by Walter et al. (2004). YFP fluorescence was observed 72 h after transfection using a confocal laser‐scanning microscope (Zeiss LSM 780, Germany).

Dual‐luciferase assay

For the dual‐luciferase assay, the promoter sequence of VvUFGT was amplified from grapevine genomic DNA using PCR and ligated into upstream of the LUC CDS of pGreen 0800‐LUC vector to generate the ProVvUFGT‐LUC reporter vector. The CDS of EGL3, VvMYCA1, VvMYBA2w and VvMYBA2r as well as of three mutations (VvMYBA2r, VvMYBA2r and VvMYBA2r) of VvMYBA2r were inserted into the pEAQ‐HT transient expression vector using the Gateway Cloning System (Invitrogen) to produce some overexpression constructs named pE‐EGL3, pE‐VvMYCA1, pE‐VvMYBA2w, pE‐VvMYBA2r, pE‐VvMYBA2r, pE‐VvMYBA2r and pE‐VvMYBA2r, respectively. The empty vector pEAQ‐HT was used as negative control (pE). All constructs were transformed into the A. tumefaciens strain EHA105 using the heat shock method. Transient expression in N. benthamiana was carried out according to the method described by Hellens et al. (2005). Luciferase activities were detected using the Dual‐Luciferase Reporter Assay System (Promega) with an Infinite200 Pro microplate reader (Tecan). The normalized luciferase activity was calculated as the ratio between the LUC and REN activity (Horstmann et al., 2004).

PCR analysis of the MYB‐related genes of the MYBA1 and MYBA2 loci

Genomic DNA was isolated from young leaves of the 213 grape germplasms, as well as interbreeding and self‐fertilization populations, and was used as the template for PCR. The primers for VvmybA1a, VvmybA1b, VvmybA1c, VvmybA1, VvmybA1, VvMYBA2r, VvMYBA2w, VlmybA1‐2, VlmybA1‐3 and VlmybA2 are shown in Figure [Link], [Link] and Table S12. The PCRs were performed in a total volume of 25 µL containing 50 ng of genomic DNA, 200 μM of each dNTP, 2.5 µL of 10× reaction buffer, 20 ng of reverse and forward primers and 0.5 U of ExTaq polymerase (Takara, Japan). To ensure the reliability of genotype testing results, the PCR amplification was performed twice for all grape samples. In the segregation analysis of the interbreeding populations, chi‐square tests were used to ascertain the consistency between the predicted haplotype ratios and the observed pericarp colour.

Construction and analysis of self‐fertilization populations

The inflorescences of ‘Olarra Queen’ and ‘Yuanruihei’ were bagged with white paper bags 2–3 days before anthesis to prevent exogenetic pollen contamination. At least 300 seeds were collected after fruit maturation and hidden with wet sand. The seeds were subjected to germination treatment in an incubator, and then sown in the plug tray. The seedlings were transplanted into plastic pots (15 cm × 15 cm) and grown in a greenhouse under the standard growth conditions by irrigating them daily.

Fruit surface colour measurement

Fruit surface colour measurements were performed at the maturation phase using a reflectance spectrophotometer (Hunter Lab Mini Scan XE Plus) and calculated in accordance with the colour index of red grapes (CIRG) as described by Zhang et al. (2008). According to this method, L*, a* and b* values of fruits were recorded and the colour index CIRG = (180 − H)/ (L* + C), where H = arc tan b*/a* and C = [(a*)2 + (b*)2] *0.5. Hue angle can be distributed in the four quadrants of the a *b* plane, and chroma will be higher the further it is from the origin of the coordinates. Measurements were performed on three fruits; for each germplasm, four measurements were performed on each fruit at equatorial positions around the fruit.

Determination of total anthocyanin contents

Total anthocyanin was extracted using a methanol–HCl method (Lee and Wicker, 1991). Samples (0.1 g) were soaked and incubated overnight in 5 mL methanol containing 0.1% (v/v) HCl in the dark at room temperature. Anthocyanin contents in samples were measured using the PH differential method as described in Zhang et al. (2008). The absorbance of sample extracts at 520 and 700 nm was measured using a UV‐2550 spectrophotometer (Shimadzu). Results were displayed as milligram cyanidin‐3‐glucoside equivalents/100 g FW using a molar extinction coefficient of 29 600.

Statistical analysis

The experiment was arranged in a completely randomized design (CRD) with three biological replicates. The data were statistically processed using the SAS software package (SAS Institute). Statistical difference was performed using one‐way ANOVA at the significance level of P < 0.05. For qRT‐PCR, anthocyanin content and CIRG analyses, data are shown as means ± standard error (SE) of three biological replicates. For the analysis of dual‐luciferase assay, data are presented as means ± SE of three independent experiments with at least four replicate reactions. For the analysis of β‐galactosidase activity, data are shown as means ± SE of three measurements of three independent yeast clones.

Conflict of interest

The authors declare that they have no competing interests.

Author contributions

J.F. and S.J. conceived and designed the experiments; J.F. supervised the experiments; S.J. performed most of the experiments, analysed the data and drafted the manuscript; X.T., Z.D. and C.Z. provided technical assistance; L.G., M.A. and H.U.J. revised the manuscript; K.Z., T.Z., X.Z. and M.G. collected the samples and worked on the phenotyping; X.L., X.Y., L.S. and J.H. performed the bioinformatics analyses; and H.J., C.W. and M.S.H. performed the statistical analysis. All authors provided final approval for publication.

Figure S1 Grape berry haplotypes at the color locus. The solid line is the haplotype that has been discovered, and the dashed line shows the possible haplotypes (HapG and HapH).

Click here for additional data file.

Figure S2 Pigmentation changes (A) and temporal gene expression patterns of VvMYBA2r and flavonoid‐related structural genes (B) at different development stages of grape fruits.

Click here for additional data file.

Figure S3 The expression level of VvMYBA2w in tissues of developing berries from the white ‘Shine Muscat’ cultivar.

Click here for additional data file.

Figure S4 Expression analysis of anthocyanin biosynthesis genes in the corollas of transgenic tobaccos overexpressing VvMYBA2r, VvMYBA2r, VvMYBA2r, VvMYBA2r, and VvMYBA2w.

Click here for additional data file.

Figure S5 Anthocyanin accumulation activated by VvMYBA2r and VvMYBA2w alone or a binary group of VvMYBA2r/2w and VvMYCA1 or a ternary group of VvMYBA2r/2w, VvMYCA1, and VvWDR1 in tobacco leaves.

Click here for additional data file.

Figure S6 VvMYBA2w interacts with VvSAP5 and VvUBE2A in vitro and in vivo.

Click here for additional data file.

Figure S7 VvMYBA2r could not interact with VvSAP5 and VvUBE2A in vitro and in vivo.

Click here for additional data file.

Figure S8 Seven kinds of grape berry skin colors in mature stage. 1. Greenish‐yellow ‘Lvzao’; 2. Yellowish‐green ‘Queen of Vineyard’; 3. Pink ‘Benizuihō’; 4. Red ‘Rosario rosso’; 5. Purplish‐red ‘Hongwuzilu’; 6. Purple‐black ‘Kyohō’; 7. Blue‐black ‘Venus Seedless’.

Click here for additional data file.

Figure S9 The genotype segregation and progeny phenotypic characteristics of the MYB genes from self‐intersection for the ‘phased phase’ and ‘repulsive phase’ cultivars.

Click here for additional data file.

Figure S10 The genotypes of MYBA1 and MYBA2 loci from self‐fertilization seedlings of ‘Yuanruihei’ grapevine.

Click here for additional data file.

Figure S11 The genotypes of MYBA1 and MYBA2 loci from self‐fertilization seedlings of ‘Olarra Queen’ grapevine.

Click here for additional data file.

Figure S12 The band patterns of PCR amplified products of MYBA1 and MYBA2 loci of 69 plantlets from the cross of ‘Muscat Hamburg’× ‘Crimson seedless’.

Click here for additional data file.

Figure S13 The band patterns of PCR amplified products of MYBA1 and MYBA2 loci of 63 hybrid progenies from the cross of ‘Cuibao seedless’× ‘Qiuhongbao’.

Click here for additional data file.

Figure S14 The berry colors of field survey of 69 hybrid progenies from the cross of ‘Muscat Hamburg’ × ‘Crimson seedless’.

Click here for additional data file.

Figure S15 The berry colors of field survey of 63 hybrid progenies from the cross of ‘Cuibao seedless’ × ‘Qiuhongbao’.

Click here for additional data file.

Figure S16 Structures of MYBA1 and MYBA2 alleles. Primer positions are indicated below the maps.

Click here for additional data file.

Table S1 Correlation analysis of the relative expression of genes correlated with flavonoid biosynthetic during grapevine berry development.

Click here for additional data file.

Table S2 The selected candidates for interaction protein of VvMYBA2w.

Click here for additional data file.

Table S3 Genotypes and haplotypes of MYBA1 and MYBA2 loci from 33 self‐fertilization seedlings of ‘Yuanruihei’ grapevine.

Click here for additional data file.

Table S4 Genotypes and haplotypes of MYBA1 and MYBA2 loci from 25 self‐fertilization seedlings of ‘Olarra Queen’ grapevine.

Click here for additional data file.

Table S5 Identification of genotype and haplotype compositions of 213 grape germplasms at the region of berry color locus.

Click here for additional data file.

Table S6 Relationship between skin coloration and haplotype compositions of MYBA1 and MYBA2 loci in the 213 grape germplasms.

Click here for additional data file.

Table S7 Haplotype combination and their number for V. vinifera and hybrids of V. vinifera and V. labrusca.

Click here for additional data file.

Table S8 PCR analysis of MYB‐related genes in the hybrid progenies of the ‘Muscat Hamburg’ × ‘Crimson seedless’ cross.

Click here for additional data file.

Table S9 PCR analysis of MYB‐related genes in the hybrid progenies of the ‘Cuibao seedless’ × ‘Qiuhongbao’ cross.

Click here for additional data file.

Table S10 The primer sequences used for transient and stable genetic transformation, subcellular localization, Y2H, BiFC and dual‐luciferase assay.

Click here for additional data file.

Table S11 Sequence of primers used for qRT‐PCR.

Click here for additional data file.

Table S12 The primers used for PCR amplification of allele from MYBA1 and MYBA2 gene loci.

Click here for additional data file.