A Grapevine Anthocyanin Acyltransferase, Transcriptionally Regulated by VvMYBA, Can Produce Most Acylated Anthocyanins Present in Grape Skins

Abstract

Anthocyanins are flavonoid compounds responsible for red/purple colors in the leaves, fruit, and flowers of many plant species. They are produced through a multistep pathway that is controlled by MYB transcription factors. VvMYBA1 and VvMYBA2 activate anthocyanin biosynthesis in grapevine (Vitis vinifera) and are nonfunctional in white grapevine cultivars. In this study, transgenic grapevines with altered VvMYBA gene expression were developed, and transcript analysis was carried out on berries using a microarray technique. The results showed that VvMYBA is a positive regulator of the later stages of anthocyanin synthesis, modification, and transport in cv Shiraz. One up-regulated gene, ANTHOCYANIN 3-O-GLUCOSIDE-6″-O-ACYLTRANSFERASE (Vv3AT), encodes a BAHD acyltransferase protein (named after the first letter of the first four characterized proteins: BEAT [for acetyl CoA:benzylalcohol acetyltransferase], AHCT [for anthocyanin O-hydroxycinnamoyltransferase], HCBT [for anthranilate N-hydroxycinnamoyl/benzoyltransferase], and DAT [for deacetylvindoline 4-O-acetyltransferase]), belonging to a clade separate from most anthocyanin acyltransferases. Functional studies (in planta and in vitro) show that Vv3AT has a broad anthocyanin substrate specificity and can also utilize both aliphatic and aromatic acyl donors, a novel activity for this enzyme family found in nature. In cv Pinot Noir, a red-berried grapevine mutant lacking acylated anthocyanins, Vv3AT contains a nonsense mutation encoding a truncated protein that lacks two motifs required for BAHD protein activity. Promoter activation assays confirm that Vv3AT transcription is activated by VvMYBA1, which adds to the current understanding of the regulation of the BAHD gene family. The flexibility of Vv3AT to use both classes of acyl donors will be useful in the engineering of anthocyanins in planta or in vitro.

Anthocyanins, a group of water-soluble flavonoid compounds, are produced by almost all vascular plants and have been shown to have a diverse range of biological functions. They are major contributors to the orange, red, purple, and blue colors seen in the leaves, fruit, and flowers of many plant species and, hence, have important roles in attracting pollinators and seed dispersers (Schaefer et al., 2004). It has been suggested that they also act as protection agents against UV light (Markham, 1988) and are involved in plant stress responses (Dixon and Paiva, 1995). Anthocyanins have potent antioxidant capacity, providing numerous health-promoting properties, including cardiovascular disease prevention and antiinflammatory, antimicrobial, and anticarcinogenic activities (He and Giusti, 2010).

Wine grapes (Vitis vinifera) contain both 3-O-monoglucoside and 3-O-acyl-monoglucoside anthocyanins derived from five main anthocyanidin aglycones: delphinidin, cyanidin, peonidin, petunidin, and malvidin. The structural genes involved in their production have all been isolated and characterized (Fig. 1; for review, see He et al., 2010). The regulation of the flavonoid pathway is through the action of transcriptional complexes comprising three transcription factor (TF) families: a basic helix-loop-helix protein (bHLH), a Trp-Asp repeat protein (WDR or WD40), and an R2R3-MYB protein. The MYB/bHLH/WDR complexes are thought to recognize and bind to responsive elements found in the promoters of biosynthesis genes in the pathway, usually resulting in activation of the expression of that gene (for review, see Matus et al., 2010). The MYB TFs determine the specificity of this complex and have been shown to bind directly to the structural gene promoters (Sainz et al., 1997).

Anthocyanin synthesis in grapes is specifically regulated by transcriptional complexes containing either VvMYBA1 or VvMYBA2 through the activation of URIDINE DIPHOSPHATE GLUCOSE-FLAVONOID 3-O-GLUCOSYLTRANSFERASE (VvUFGT) transcription (Walker et al., 2007). VvUFGT catalyzes the final step of anthocyanin synthesis, where anthocyanidins are glycosylated on the 3-hydroxyl group of the C ring of the flavylium molecule, to produce stable anthocyanins (Ford et al., 1998). The VvMYBA1 and VvMYBA2 coding sequences are almost identical in their R2R3 domains, but VvMYBA2 has a 93-amino acid repeat of the activation domain (Walker et al., 2007). White-fruited grapevine cultivars have arisen due to the lack of a functional VvMYBA protein, caused by a retrotransposon insertion in the promoter of VvMYBA1 (Kobayashi et al., 2004) and two nonconservative mutations in VvMYBA2 (Walker et al., 2007).

In grapevine, anthocyanins can be further modified by O-methyltransferases, which methylate hydroxyl groups at the 3′ and 5′ positions of the B ring (Fournier-Level et al., 2011), and acyltransferases, which produce 3-O-acetyl-, 3-O-coumaroyl-, and 3-O-caffeoyl-monoglucosides by attaching acyl groups to the C6″ position of the Glc moiety (Mazza, 1995; Nakayama et al., 2003). Acylated anthocyanins in various flowering species are more stable compared with their nonacylated counterparts, most likely due to increased intramolecular stacking (Yonekura-Sakakibara et al., 2008). Wines made from red-berried grapevine cultivars such as cv Ives or Veeport with high proportions of acylated anthocyanins can have greater color stability when exposed to light compared with those from red varieties with no acylated anthocyanins, such as cv Pinot Noir (Van Buren et al., 1968; Smart, 1992). Despite the importance of red color stability to wine quality and the correlation between poorly colored red wines and cultivars containing no acylated anthocyanins, anthocyanin acyltransferases from grapevine have not yet been identified. A recent study by Costantini et al. (2015) identified a number of quantitative trait loci (QTLs) associated with variation in acylated anthocyanin levels in F1 progeny from a cv Syrah × cv Pinot Noir cross. The strongest candidate genes within these QTLs included those belonging to the BAHD acyltransferase family (named after the first letter of the first four characterized proteins: BEAT [for acetyl CoA:benzylalcohol acetyltransferase], AHCT [for anthocyanin O-hydroxycinnamoyltransferase], HCBT [for anthranilate N-hydroxycinnamoyl/benzoyltransferase], and DAT [for deacetylvindoline 4-O-acetyltransferase]; St-Pierre and Luca, 2000) and the Serine carboxypeptidase-like acyltransferase family. Yet, no QTL was found to describe the presence/absence of acylation in berries, probably due to the genotype of the parents and the lack of segregation for this in the F1 population.

In a number of species, the expression of flavonoid pathway MYB TF transgenes, or the overexpression or silencing of endogenous genes, has enabled detailed studies of their regulatory roles in this pathway. This has also highlighted substantial differences in the flavonoid regulatory network between different species, particularly between monocots and dicots, and within reproductive organs (for review, see Petroni and Tonelli, 2011). Transcriptomic analyses of these transgenic plants have often led to the discovery of novel genes associated with anthocyanin synthesis, modification (including the identification of anthocyanin acyltransferases), or transport. For example, Tohge et al. (2005) used microarrays to identify putative acyltransferases, later characterized as anthocyanin acyltransferases by Luo et al. (2007), in an activation-tagged PAP1D line of Arabidopsis (Arabidopsis thaliana) in which PRODUCTION OF ANTHOCYANIN PIGMENT1 (AtPAP1) was overexpressed. Butelli et al. (2008) also found putative anthocyanin acyltransferases that were up-regulated in transgenic tomato (Solanum lycopersicum) expressing the MYB TF transgene ROSEA1 from Antirrhinum majus flowers. Cutanda-Perez et al. (2009) ectopically expressed the VlMYBA1 gene from Vitis labruscana in grapevine hairy root tissue and analyzed gene expression changes to the transcriptome of these roots. They concluded that VlMYBA1 regulated a narrow set of genes involved in anthocyanin biosynthesis and identified novel genes associated with anthocyanin transport. The regulatory effect of VvMYBA on the transcriptome in berries of grapevine, where it is naturally expressed, has not yet been studied due to the complicated and lengthy protocols required to produce stably transformed vines.

In this study, transgenic grapevines with altered VvMYBA gene expression were generated. Overexpression of VvMYBA1 in the white-berried cv Chardonnay resulted in pigmented fruit, while silencing the VvMYBA1 and VvMYBA2 genes in the pigmented cv Shiraz produced berries with lightly colored or white phenotypes. When microarray techniques were used to analyze the transcriptomes of the transgenic berries, transcription of a gene putatively encoding for a member of the BAHD protein family correlated with the expression of VvMYBA. Members of the BAHD gene family encode acyltransferases that utilize CoA thioesters as their donor substrate both in planta (Fujiwara et al., 1998) and/or in vitro (Yonekura-Sakakibara et al., 2000; Yabuya et al., 2001; Suzuki et al., 2004b; D’Auria et al., 2007). Here, the characterization of the function of this BAHD gene from grapevine, identified through the microarray analysis of the transcriptome of berries with altered VvMYBA expression, is presented. Gene expression analyses, stable plant transformations, and recombinant protein assays were used to demonstrate that the gene encodes an anthocyanin acyltransferase, ANTHOCYANIN 3-O-GLUCOSIDE-6″-O-ACYLTRANSFERASE (Vv3AT), which is capable of producing the common acylated anthocyanins found in grape berries. Vv3AT belongs to a BAHD protein clade responsible for a range of functions, most of which do not involve the modification of anthocyanins, and can use a broad range of acyl acceptor and donor substrates, which include both aliphatic and aromatic acyl-CoA thioesters.

RESULTS

Altering VvMYBA Gene Expression Alters Berry Color in Transgenic Grapevines

A VvMYBA1 gene construct, under the control of either its own promoter (VvMYBA1pr:VvMYBA) or a 35S constitutive promoter from the Cauliflower mosaic virus (35S:VvMYBA1), was used to stably transform embryogenic callus of the white-berried cv Chardonnay to produce mature transgenic vines. Compared with cv Chardonnay control berries (Fig. 2A), the transgenic berries expressing the VvMYBA1pr:VvMYBA construct had red/purple pigmentation in their skin (Fig. 2B) that developed at veraison (when anthocyanin accumulation normally begins as berries ripen). The total anthocyanin concentration in these berries was approximately 30% of that detected in purple/black berries of cv Shiraz (Supplemental Fig. S1). The cv Chardonnay berries constitutively expressing VvMYBA1 (35S:VvMYBA) had purple/black pigmentation in their skin (Fig. 2C) and purple flesh, which was present throughout development and ripening, and most other tissues of the plant were also pigmented. The cv Shiraz (Fig. 2D) was used to create transgenic vines containing a VvMYBA-silencing construct (VvMYBAsi) that silenced both the VvMYBA1 and VvMYBA2 genes. Transgenic cv Shiraz berries possessed either rose-colored skin pigmentation (rose cv Shiraz; Fig. 2E) or nonpigmented skin (white cv Shiraz; Fig. 2F).

Conserved Transcriptomic Changes in Transgenic Grapevines with Altered VvMYBA Gene Expression

A comparison of the transcriptomes of transgenic whole berries, close to ripeness, with altered VvMYBA gene expression and their controls was carried out by hybridizing RNA to a NimbleGen microarray containing probes that represent the 98.6% of the genes predicted in the 12× grapevine V1 genome (Microarray Gene Expression Omnibus database accession no. GSE56915; http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=cnmxwmmsfjctvgt&acc=GSE56915). There were 188 and 160 genes significantly up- and down-regulated, respectively, using a false discovery rate of 2.5% and a fold change (FC) of 2 or greater in the VvMYBA1-overexpressing red cv Chardonnay berries compared with nonpigmented controls (Supplemental Table S1). Using the same parameters, there were 48 and 319 genes significantly up- and down-regulated, respectively, in transgenic white cv Shiraz berries compared with nontransgenic red-berried controls (Supplemental Table S2). All the significantly modulated genes were automatically annotated against the V1 gene prediction version of the grapevine genome (http://genomes.cribi.unipd.it/grape/index.php) and manually improved where possible.

To gain information about the genes probably controlled by VvMYBA, these microarray data were analyzed for genes whose expression was consistently up- or down-regulated in response to the presence or absence of VvMYBA gene expression in both cv Chardonnay and cv Shiraz and their respective transgenic lines. Only 26 and one genes were up- and down-regulated, respectively (FC ≥ 2), in a consistent manner between the two sets of plants (Table I). Those with the greatest expression changes were characterized or putative flavonoid-related genes that were up-regulated in red berries compared with white. These included the anthocyanin transporter ANTHOCYANIN MULTIDRUG AND TOXIC EFFLUX TRANSPORTER2 (VvanthoMATE2; Gomez et al., 2011) and GLUTATHIONE-S-TRANSFERASE4 (VvGST4; Conn et al., 2008), which is involved in the transport of flavonoids either directly or due to their modification before their transport, and several anthocyanin biosynthesis genes, including nine members of the FLAVONOID 3′,5′-HYDROXYLASE (F3′5′H; Falginella et al., 2010) gene family, three putative CHALCONE SYNTHASE (CHS) genes, and VvUFGT. In addition, the list included one transcript corresponding to FLAVONOL AND ANTHOCYANIDIN-GLUCOSIDE 3′,5′-O-METHYLTRANSFERASE (VvFAOMT; Hugueney et al., 2009; Lücker et al., 2010), involved in flavonol and anthocyanin methylation, another putative anthocyanin methyltransferase, and a transcript belonging to the BAHD acyltransferase family, possibly involved in the acylation of anthocyanin. Only one gene, PHENYLALANINE AMMONIA LYASE2 (VvPAL2; Guillaumie et al., 2011), was related to general phenylpropanoid biosynthesis. There were no flavonoid biosynthetic genes that were consistently down-regulated in red berries in both grapevine cultivars. There were, however, some phenylpropanoid/flavonoid pathway genes that had modulated expression in only one of the cultivars (i.e. either cv Chardonnay or cv Shiraz transgenic berries), and these are discussed in the “Supplemental Data.”

Genes with altered transcription levels (FC ≥ 2) in response to VvMYBA gene expression in a consistent manner in both transgenic cv Chardonnay and cv Shiraz berries

To validate the microarray data, the expression of a subset of flavonoid biosynthesis genes was analyzed in the transgenic berries using quantitative PCR (qPCR; Supplemental Figs. S2 and S3).

Overall, the transcriptomic data confirm the specific role of VvMYBA in the control of anthocyanin-related genes. The core set of genes most up-regulated by overexpression of VvMYBA and most down-regulated by the silencing of this gene includes known targets of the TF as well as new putative players in the anthocyanin biosynthetic pathway. Among these, the uncharacterized gene annotated as belonging to the BAHD protein superfamily (VIT_03s0017g00870) was hypothesized to function as an anthocyanin acyltransferase and chosen for further characterization.

Vv3AT Belongs to the BAHD Superfamily of Acyltransferases within a Clade Distinct from Most Other Anthocyanin Acyltransferases

A full-length complementary DNA (cDNA) clone of the BAHD gene, identified from the microarrays, was isolated from cv Cabernet Sauvignon (Vv3AT-CS). The cv Cabernet Sauvignon gene consists of a single exon 1,284 bp in length. Two alleles were present in this cultivar that differ by two nucleotides: an A/G polymorphism at position 238 of the putative coding region does not alter the predicted amino acid sequence, and a C/T polymorphism at 784 bp results in the conversion of Arg-262 to Cys in the 428-amino acid protein sequence. Allele 1 encodes the same predicted protein as VIT_03s0017g00870, from the PN40024 grapevine genome sequencing project (Jaillon et al., 2007).

A comprehensive protein alignment and a phylogenetic tree were created using 72 BAHD protein sequences from 38 different plant species, including the predicted Vv3AT protein sequence. These proteins included the list of genetically or biochemically characterized BAHD proteins reported previously by D’Auria (2006) plus additional published BAHD proteins characterized since 2006 (Supplemental Table S3; Supplemental References S1). The tree separated the proteins into five major clades (Fig. 3), with Vv3AT falling into clade IIIa (Tuominen et al., 2011), most similar to an anthocyanin acyltransferase, ANTHOCYANIN 5-O-GLUCOSIDE-4‴-O-MALONYLTRANSFERASE (Ss5MaT2), from Salvia splendens (Suzuki et al., 2004b). A nucleotide BLAST in the grapevine nucleotide database of the National Center for Biotechnology Information server revealed 10 related sequences ranging from 66% to 95% identity over 68% to 100% coverage of the Vv3AT coding sequence. A translated nucleotide BLAST search in the grapevine protein database of the National Center for Biotechnology Information found 45 hypothetical proteins with 23% or greater identity to 54% or greater coverage of the Vv3AT sequence. None of these genes or proteins had been functionally characterized, so they were not included in this phylogenetic analysis.

Protein Sequence Changes and Gene Expression Patterns of Vv3AT Correlate with the Absence of Acylated Anthocyanins in cv Pinot Noir and Pale-Fruited cv Cabernet Sauvignon Mutants

The cv Pinot Noir, a red-berried grapevine cultivar, does not synthesize acylated anthocyanins (Van Buren et al., 1968) and, therefore, represents a mutant that can be compared with most other cultivars that do contain these compounds, such as cv Cabernet Sauvignon and cv Shiraz. Sequence analysis of cDNA of Vv3AT-PN from cv Pinot Noir showed two single-nucleotide polymorphisms compared with the cv Cabernet Sauvignon cDNA sequences. The first of these is a C/T polymorphism at position 349 bp of the coding region, which introduces a premature stop codon resulting in a truncated protein of 116 amino acids. An alignment of Vv3AT-CS and Vv3AT-PN against previously characterized BAHD proteins showed that the predicted Vv3AT-PN protein did not contain either of the functional motifs, HXXXDG and DFGWG, found in almost all members of the BAHD family (St-Pierre and Luca, 2000), most likely rendering it inactive (Fig. 4).

Transcript levels of VvMYBA1 and VvMYBA2 genes are difficult to determine in grapevine due to their similarity to, and the expression of, nonfunctional VvMYBA genes and alleles within the grapevine genome. VvUFGT gene expression has been correlated previously to anthocyanin accumulation (Boss et al., 1996a) and is known to be transcriptionally activated by VvMYBA1 and VvMYBA2 TFs; hence, it is a useful marker for the expression of these TFs (Kobayashi et al., 2002; Walker et al., 2007). The transcript level of Vv3AT and VvUFGT (as a marker for VvMYBA expression) was analyzed using qPCR over the development of whole berries from cv Cabernet Sauvignon and cv Pinot Noir starting from 2 weeks post flowering (wpf) through to harvest at 14 wpf (Fig. 5). VvUFGT transcripts began accumulating postveraison in both cv Cabernet Sauvignon and cv Pinot Noir, as shown previously (Boss et al., 1996b). Vv3AT transcript levels followed a very similar pattern to those of VvUFGT in cv Cabernet Sauvignon, suggesting a link between the function and regulation of these two genes. In comparison, Vv3AT transcript levels in cv Pinot Noir were low throughout development, with only a slight increase postveraison.

Two color sports of cv Cabernet Sauvignon (cv Malian and cv Shalistin; Supplemental Fig. S4) have arisen due to a deletion of the berry color locus carrying the VvMYBA genes (Walker et al., 2006). In cv Malian, this deletion occurred in the L2 cell layer, resulting in bronze/rose-colored berries, while in cv Shalistin, the deletion is also present in the L1 cell layer, resulting in white berries. To further investigate the link between Vv3AT gene expression and VvMYBA TFs, transcript levels of Vv3AT and VvUFGT were analyzed over early (whole berries) and late (skins only) development of cv Cabernet Sauvignon, cv Malian, and cv Shalistin berries (Fig. 6). The changes in transcript levels of Vv3AT were very similar to that of VvUFGT. In both cases, transcripts began accumulating after veraison in cv Cabernet Sauvignon and cv Malian, but levels were lower in cv Malian. No transcripts of either Vv3AT or VvUFGT were detected in cv Shalistin berries. It has been shown previously that VvUFGT transcription is activated by the VvMYBA1 and VvMYBA2 TFs (Walker et al., 2007). This would explain the expression pattern of VvUFGT seen here, as VvMYBA genes are only expressed in the L1 cell layer of the berry skin in cv Malian, while cv Shalistin does not contain functional VvMYBA genes at all (Walker et al., 2006). These results demonstrate a strong link between the presence of the VvMYBA TFs and Vv3AT gene expression.

The Grapevine Color Regulator VvMYBA1 Activates the Expression of Vv3AT in cv Chardonnay Suspension Cells

Transient promoter-binding assays using a luciferase reporter system were performed to determine if VvMYBA1 can activate expression from the Vv3AT promoter. Genomic DNA 711 bp upstream of the putative initiation codon of the Vv3AT gene was isolated from cv Cabernet Sauvignon and cv Pinot Noir. These promoter regions differed by six point mutations, which were all located more than 400 bp upstream of the predicted start codon. VvUFGT, Vv3AT-CS, and Vv3AT-PN promoters were ligated upstream to a luciferase reporter gene in a pLUC vector (Horstmann et al., 2004). These constructs, along with two constructs expressing VvMYBA1 and a bHLH TF (ENHANCER OF GLABRA3 [AtEGL3], previously shown to be required for TF complex formation in similar assays [Bogs et al., 2007], and VvMYC1 [Hichri et al., 2010]; Fig. 7), were introduced into grapevine cell suspension cultures using biolistic-mediated transfection. Luciferase activity in cells bombarded with the VvUFGTpr:LUC, VvMYBA1, and either AtEGL3 or VvMYC1 genes was over 60-fold higher than the activity in control cells (where an empty vector construct replaced that of VvMYBA1). The luciferase activity in the grape cells when the promoters of Vv3AT-CS and Vv3AT-PN were used was 25- to 40-fold higher than background controls, suggesting that VvMYBA1 can activate either promoter of Vv3AT in planta. When VvMYC1 was used as the bHLH TF, both Vv3AT-CS and Vv3AT-PN promoters were activated to similar levels (Fig. 7; Supplemental Table S4).

Constitutive Expression of Vv3AT in Tobacco Results in the Production of Acylated Anthocyanins

Vv3AT was constitutively expressed under the control of the cauliflower mosaic virus 35S promoter in tobacco (Nicotiana tabacum var Samsun) through Agrobacterium tumefaciens-mediated stable transformation. Six transgenic lines from independent transformation events were created that expressed Vv3AT in their flowers at varying levels (Supplemental Fig. S5). Some flowers from one transgenic line (Vv3AT-1; Fig. 8) expressing Vv3AT visually displayed a slightly more blue/purple hue compared with wild-type untransformed controls (Fig. 8A), while the others had similar phenotypes to the wild type. Anthocyanins were extracted from flowers of all six lines and separated using HPLC, and peaks were identified using liquid chromatography-tandem mass spectrometry (LC-MS/MS). Two anthocyanin species could be detected in wild-type tobacco flowers. The most abundant species was cyanidin-3-O-rutinoside (peak 1), with lesser amounts of pelargonidin-3-O-rutinoside also present (peak 2; Fig. 8B; Supplemental Table S5). Extracts from tobacco flowers constitutively expressing Vv3AT contained six extra peaks not present in the wild-type samples. These peaks were identified as cyanidin-3-O-acetylglucoside (peak 3), pelargonidin-3-O-acetylglucoside (peak 4), cyanidin-3-O-caffeoylglucoside (peak 5), pelargonidin-3-O-caffeoylglucoside (peak 6), cyanidin-3-O-coumaroylglucoside (peak 7), and pelargonidin-3-O-coumaroylglucoside (peak 8; Fig. 8B; Supplemental Table S5). Of the acylated anthocyanins, all of the transgenic tobacco lines expressing Vv3AT contained a higher concentration of coumaroylated anthocyanins than acetylated anthocyanins and had lowest amounts of caffeoylated anthocyanins (Fig. 8C).

Recombinant Vv3AT Can Acylate Anthocyanins in Vitro

In order to study the function of Vv3AT and its substrate preferences in vitro, recombinant His-tagged Vv3AT protein was generated and purified using affinity chromatography (Supplemental Figs. S6 and S7). The kinetic properties of this enzyme were determined using various acyl donor and acceptor substrates (Table II; Supplemental Figs. S8 and S9). Due to the inability to purify the recombinant protein to homogeneity, the concentration of the enzyme was estimated based on a His-tag ELISA and visualization on western blots. Hence, the kinetic parameters described are estimates only but are comparable between the various substrates tested. First, three acyl donors were assayed using malvidin-3-O-glucoside as the anthocyanin acyl acceptor, as this is the predominant anthocyanin found in the berry skins of grapevine (Mazza, 1995). As acetyl-, coumaroyl-, and caffeoyl-conjugated anthocyanins are present in grapevine, Vv3AT activity was tested using acetyl-CoA, coumaroyl-CoA, and caffeoyl-CoA donors. It was found that Vv3AT was capable of using all three of these acyl donors in our assay conditions. The lowest apparent K  _m was observed when coumaroyl-CoA was the acyl donor (1.25 ± 0.16 µm), while the enzyme had the lowest affinity for acetyl-CoA, with an apparent K  _m of 16.86 ± 6.43 µm. When acetyl-CoA was the acyl donor, an apparent catalytic rate constant (k  _cat) of 1.11 ± 0.61 s⁻¹ was observed, which was 8.5- and 3.8-fold higher than the apparent k  _cat data when coumaroyl- and caffeoyl-CoA were the acyl donors, respectively (0.13 ± 0.04 and 0.29 ± 0.09 s⁻¹). The apparent specificity constant, taking into account both the K  _m and k  _cat of the substrate, was similar for all three CoA donors: 0.08 ± 0.07 µm  ⁻¹ s⁻¹ with acetyl-CoA, 0.10 ± 0.03 µm  ⁻¹ s⁻¹ with coumaroyl-CoA, and 0.07 ± 0.02 µm  ⁻¹ s⁻¹ with caffeoyl-CoA. Malonyl-CoA was also tested to see if it could act as substrate for Vv3AT, as many other BAHD anthocyanin acyltransferases use this compound as an acyl donor, including Ss5MAT2 (D’Auria, 2006), the closest characterized homolog of Vv3AT. While Vv3AT could utilize malonyl-CoA as a substrate, the apparent K  _m was calculated at 588.77 ± 74.48 µm, which was much higher than the other acyl donors tested, and the specificity constant was approximately 100-fold lower at 0.0016 ± 0.0002 µm  ⁻¹ s⁻¹. Therefore, it was concluded that Vv3AT can catalyze the acylation of malvidin-3-O-glucoside with a range of CoA-conjugated acyl donors.

Kinetics of recombinant Vv3AT enzyme with various acyl donor and acceptor substrates

When coumaroyl-CoA and caffeoyl-CoA were used as acyl donors in concentrations above 100 µm, it was found that the production of the expected acylated anthocyanin was inhibited. No substrate inhibition was observed when using acetyl-CoA in concentrations up to 1 mm (Supplemental Fig. S10). For this reason, Vv3AT enzyme kinetics using various anthocyanin acyl acceptors was conducted using acetyl-CoA as the acyl donor.

The 3-O-glucosides of malvidin, cyanidin, delphinidin, peonidin, and petunidin are the five major anthocyanin species found in grapes (He et al., 2010). The activity of Vv3AT was assayed using all of these anthocyanins as acyl acceptors except petunidin-3-O-glucoside (not commercially available). The kinetic parameters of Vv3AT when malvidin- and peonidin-3-O-glucosides were used as substrates were similar, with apparent K  _m values of 98.66 ± 19.59 and 79.52 ± 8.37 µm, apparent k  _cat values of 13.92 ± 0.71 and 14.2 ± 1.97 s⁻¹, and apparent specificity constants of 0.15 ± 0.03 and 0.18 ± 0.03 µm  ⁻¹ s⁻¹, respectively. Apparent K  _m values calculated with cyanidin-3-O-glucoside and delphinidin-3-O-glucoside acyl acceptors were much higher at 506.22 ± 8.55 and 765.13 ± 51.97 µm, respectively, but the apparent k  _cat values were also higher at 77 ± 5.92 and 67.89 ± 13.01 s⁻¹, indicating that the specificity constants of 0.15 ± 0.01 and 0.09 ± 0.02 µm  ⁻¹ s⁻¹, respectively, were not that dissimilar to those of malvidin- and peonidin-3-O-glucosides. To determine if Vv3AT would also acylate anthocyanins with other glycosylation patterns, activity using cyanidin-3,5-O-diglucoside and cyanidin-3-O-rutinoside as acyl acceptors was also tested. Vv3AT was capable of acylating cyanidin-3,5-O-diglucoside, with an apparent K  _m of 64.17 ± 3.87 µm, a k  _cat of 0.18 ± 0.02 s⁻¹, and an apparent specificity constant of 0.003 ± 0.0004 µm  ⁻¹ s⁻¹, the latter two being much lower than those calculated when the monoglucoside anthocyanins were used as substrates. Recombinant Vv3AT showed no activity with cyanidin-3-O-rutinoside.

DISCUSSION

VvMYBA Regulates the Later Stages of Anthocyanin Synthesis, Modification, and Transport

The production of transgenic grapevines with altered expression of VvMYBA genes has enabled what is, to our knowledge, the first study of the effect of these TFs in berries. Expression or silencing of these genes resulted in altered berry pigment phenotypes due to the role of VvMYBA TFs in activating anthocyanin biosynthesis (Kobayashi et al., 2002; Walker et al., 2007; Cutanda-Perez et al., 2009). A comparison of the transcriptomic changes detected in the transgenic cv Shiraz and cv Chardonnay berries revealed altered expression of 27 genes with FC > 2 in both cultivars (Table I). Seventeen genes with FCs greater than 10 were up-regulated in red berries and have characterized roles in anthocyanin biosynthesis, while the only gene that was down-regulated does not have this function and the FCs were much lower. This suggests that VvMYBA is a positive regulator of anthocyanin biosynthesis. Transcriptome studies in other plants overexpressing regulators of anthocyanin synthesis, such as ANT1 (named for the activation-tagged insertion line ant1) in tomato and AtPAP1 in Arabidopsis, have shown similar results (Mathews et al., 2003; Tohge et al., 2005). The list of modulated genes within the individual cultivars with altered VvMYBA expression was much larger than the 27 common genes affected in both cultivars (Supplemental Tables S1 and S2). While the VvMYBA1 transgene was clearly expressed in the transgenic red cv Chardonnay berries, the combined transcript levels of all three endogenous VvMYBA genes (nonfunctional) were similar in the mature transgenic berries compared with the cv Chardonnay control (Supplemental Figs. S2 and S3). This is consistent with the microarray results, where nonfunctional MYBA gene expression was not altered significantly.

In grapevine, VvUFGT catalyzes the final step in producing stable anthocyanins by adding a Glc molecule to anthocyanidins (Ford et al., 1998). Anthocyanins can then be further modified by methyltransferases and acyltransferases before they are transported into the vacuole (Mazza, 1995; Fournier-Level et al., 2011). VvUFGT and genes involved in anthocyanin methylation (VvFAOMT; Hugueney et al., 2009; Lücker et al., 2010) and modification for transport and vacuolar transport (VvGST4 and VvanthoMATE2; Conn et al., 2008; Gomez et al., 2011), plus Vv3AT encoding the BAHD anthocyanin acyltransferase characterized in this study, were all strongly up-regulated in red berries. These results show that VvMYBA up-regulates the final steps of anthocyanin coloring of the berry and, hence, is an essential MYB TF required for the activation of these later stages of anthocyanin synthesis and transport. While this was shown previously by Cutanda-Perez et al. (2009) in an analysis of transgenic grapevine hairy root cultures, the up-regulation of Vv3AT was not detected. Tohge et al. (2005), Luo et al. (2007), and Butelli et al. (2008) also identified anthocyanin acyltransferase genes that were up-regulated by MYB TFs in Arabidopsis and tomato, suggesting that this could be a widespread pattern. However the direct regulation of acyltransferase genes was not demonstrated in their studies.

Many members of the VvF3′5′H family were up-regulated in red berries to similar levels as VvUFGT and Vv3AT. In grapes, anthocyanins can be either dihydroxylated or trihydroxylated at the 3′4′ (cyanidin and peonidin) or the 3′4′5′ (malvidin, delphinidin, and petunidin) positions of the B ring, with VvF3′5′H enzymes catalyzing the later reactions (Bogs et al., 2006), suggesting that VvMYBA preferentially up-regulates the synthesis of trihydroxylated anthocyanins in berries. Many other early flavonoid pathway genes were up-regulated in red berries expressing VvMYBA (see Supplemental Methods S1; Supplemental Figs. S2 and S3), albeit with much lower FCs compared with the aforementioned genes. The transcriptomic data show evidence that VvMYBA may also exert antagonistic effects on genes related to the general phenylpropanoid metabolism, such as several PAL genes, proanthocyanidin biosynthesis, such as VvMYBPA1 (Bogs et al., 2007), and some PA-related structural genes. These effects may be the result of a metabolic feedback and/or of transcriptional interplays between regulators.

Evidence of Convergent Evolution of Vv3AT to other BAHD Acyltransferases

The putative BAHD anthocyanin acyltransferase gene, Vv3AT, was identified as being up-regulated in red berries from the microarray experiments. This previously uncharacterized gene was not identified as a candidate involved in anthocyanin acylation variation in the QTL analysis recently published by Costantini et al. (2015). A comprehensive phylogenetic analysis of characterized BAHD plant acyltransferases, including Vv3AT, found that the proteins are grouped into five major clades that could be classified predominantly by the substrate specificities of the enzymes (Fig. 3; Supplemental Table S3), in agreement with previously published phylogenetic trees (D’Auria, 2006; Yu et al., 2009; Tuominen et al., 2011). Most of the anthocyanin and flavonoid acyltransferases clustered together in clade I, but Vv3AT was placed in clade IIIa. The majority of enzymes in clade I utilize malonyl-CoA as their major acyl donor, while those in clade III mostly utilize acetyl-CoA. Malonylated anthocyanins have not been detected in grapevine, while acetylated anthocyanins are common (for review, see He et al., 2010). However, another anthocyanin acyltransferase in this clade, from S. splendens (Ss5MAT2), has been shown to utilize malonyl-CoA to acylate anthocyanin-5-O-glucosides (Suzuki et al., 2004b). This enzyme and Vv3AT both lack the characteristic YFGNC motif common to the anthocyanin transferases present in clade I (Fig. 3). This suggests that Vv3AT, like Ss5MAT2, has evolved from a different branch of the BAHD family than the other anthocyanin acyltransferases designated as members of clade I.

Due to the absence of acylated anthocyanins in the grapevine cv Pinot Noir (Van Buren et al., 1968), a comparison of Vv3AT gene sequences and transcript levels in this cultivar with those of cv Cabernet Sauvignon, which does produce acylated anthocyanins, was carried out. Very low or undetectable levels of the Vv3AT transcript were present in cv Pinot Noir berries compared with fruit from cv Cabernet Sauvignon, which accumulated the transcript postveraison, peaking at 10 wpf (2 weeks after veraison; Fig. 5). Furthermore, any transcript that was expressed in cv Pinot Noir would contain a premature stop codon, due to a nonsense mutation resulting in a truncated protein lacking two biochemically important motifs, HXXXDG and DFGWG (St-Pierre and Luca, 2000; Fig. 4). Hence, cv Pinot Noir does not possess a functional Vv3AT protein, despite the presence of an annotated sequence of full length in the backcrossed PN40024 genome. This gene sequence, VIT_03s0017g00870, could have been contributed by another probable parent, Frankenthal, rather than cv Pinot Noir (Jaillon et al., 2007). The VvMYBA1/2 genes were analyzed by Walker et al. (2007) and shown to be similar to those of cv Shiraz and cv Cabernet Sauvignon. It is unlikely that MYBA plays a direct role in the lack of acylated anthocyanins in this cultivar, as anthocyanin biosynthesis is still prevalent. Taken together, the results from the protein assays and analysis of the cv Pinot Noir Vv3AT gene strongly indicate that Vv3AT is the gene responsible for the acylation of anthocyanins in grapevine.

Transcriptional Regulation of Vv3AT

Transient promoter-binding luciferase gene expression assays using grapevine suspension cells showed that VvMYBA can activate the expression of Vv3AT (Fig. 7). This result was supported by the pattern of Vv3AT transcription compared in developing berries of cv Cabernet Sauvignon, cv Malian, and cv Shalistin that differ in VvMYBA gene expression (Fig. 6). VvMYBA has been shown to activate VvUFGT transcription (Walker et al., 2007), and the expression of this gene followed almost identical patterns to that of Vv3AT in the three cultivars over berry development and ripening. There has been relatively little work done on the regulation of BAHD gene transcription, as studies have mostly focused on their enzymatic functions. A limited number of studies have used promoter fusion constructs to investigate the spatial and developmental expression of the BAHD acyltransferase genes: SUBERIN FERULOYL TRANSFERASE, tomato ACYLTRANSFERASE2, FATTY ALCOHOL:CAFFEOYL-COENZYME A CAFFEOYLTRANSFERASE, ALIPHATIC SUBERIN FERULOYL TRANSFERASE, DWARF AND ROUND LEAF1, DEFICIENT IN CUTIN FERULATE, and SPERMIDINE HYDROXYCINNAMOYL TRANSFERASE (Grienenberger et al., 2009; Molina et al., 2009; Kosma et al., 2012; Rautengarten et al., 2012; Schilmiller et al., 2012; Boher et al., 2013; Zhu et al., 2013). Several studies have also shown links between the expression of MYB TFs and BAHD acyltransferase genes. For example, Onkokesung et al. (2012) identified the putrescine and spermidine acyltransferases, NaAT1 and NaDH29, through microarray studies utilizing N. attenuata plants, where the NaMYB8 gene, encoding a TF known to regulate phenolamide biosynthesis, was silenced. Strong links between the expression of NaMYB8, NaAT1, and NaDH29 were demonstrated, suggesting that their expression was controlled by this TF. In another study, Luo et al. (2008) expressed the AtMYB12 gene in tomato, which activates the biosynthesis of chlorogenic acids in Arabidopsis, compounds derived from the phenylpropanoid pathway. The expression of a number of tomato genes involved in chlorogenic acid biosynthesis was increased in AtMYB12-expressing lines, including HYDROXYCINNAMOYL-COENZYME A QUINATE TRANSFERASE and HYDROXYCINNAMOYL-COENZYME A SHIKIMATE/QUINATE HYDROXYCINNAMOYL TRANSFERASE, two characterized BAHD acyltransferases. The research presented in this article provides further insight into BAHD gene regulation and presents additional evidence that MYB TFs may directly activate their transcription (Fig. 7).

The cv Cabernet Sauvignon and cv Pinot Noir Vv3AT gene promoters, which differed by only six point mutations, were activated to similar levels by VvMYBA in transient assays (Fig. 7), even though very low gene expression was detected in cv Pinot Noir berries postveraison compared with cv Cabernet Sauvignon (Fig. 5). Premature stop codons in mRNA transcripts can lead to the activation of the nonsense-mediated mRNA decay pathway resulting in the rapid degradation of nonsense transcripts (for review, see van Hoof and Green, 2006). There is evidence that nonsense-mediated mRNA decay acts on many plant gene mutants, including the flavonoid biosynthetic pathway gene CHS in petunia (Que et al., 1997). It is possible that the premature stop codon in Vv3AT-PN is detected by this pathway, resulting in rapid degradation of the transcript and, hence, the low levels of Vv3AT transcripts detected in cv Pinot Noir berries.

Vv3AT Functions as a BAHD Acyltransferase That Can Use Both Aliphatic and Aromatic Acyl Donors

The in planta function of Vv3AT was tested by constitutive expression of this gene in tobacco. One transgenic line displayed a slightly more blue/purple hue to the flowers compared with wild-type controls (Fig. 7), but this was not consistent in all lines, most of which displayed no significant difference from the controls. Biochemical analyses showed that wild-type flowers contained cyanidin- and pelargonidin-3-O-rutinosides, while the flowers expressing Vv3AT also contained cyanidin- and pelargonidin-3-O coumaroyl-, caffeoyl-, and acetyl-monoglucosides (Fig. 8). No acylated rutinoside anthocyanin conjugates were detected in the transgenic tobacco flowers, suggesting that Vv3AT is not able to acylate such anthocyanins. Rutinose is a disaccharide made up of Glc with Rha attached at the 6″ position, the same position where acylation occurs in grapevine. Luo et al. (2007), who expressed the Arabidopsis COUMAROYL-COENZYME A:ANTHOCYANIDIN 3-O-GLUCOSIDE-6″-O-COUMAROYLTRANSFERASE1/2 (At3AT1 and At3AT2) genes in tobacco, also reported no visible significant difference in transgenic flower color compared with control flowers, despite the accumulation of coumaroylated anthocyanins.

Substrate preferences and enzyme kinetics of Vv3AT were studied using in vitro assays with His-tagged recombinant Vv3AT protein. The activity of this enzyme was tested with four anthocyanin acyl acceptor molecules found in grapevine and the three most common acyl donors (acetyl-CoA, coumaroyl-CoA, and caffeoyl-CoA). Vv3AT catalyzed acylation using all of these substrates. A recent study showed that berries of 34 grapevine cultivars contained varying proportions of acylated anthocyanins, but the majority contained much less caffeoylated anthocyanins compared with their acetylated and coumaroylated counterparts (Ferrandino et al., 2012). Considering the similar kinetic properties of Vv3AT with all three acyl donors (Table II), other factors must impact anthocyanin acylation, the most likely being the availability of the CoA substrates and the involvement of other enzymes in anthocyanin acylation, such as the candidate genes identified by Costantini et al. (2015). The aromatic acyl donors, caffeoyl- and coumaroyl-CoA, were shown to inhibit the production of the corresponding acylated anthocyanin by recombinant Vv3AT at concentrations greater than 100 µm (Supplemental Fig. S10). This was not observed for the aliphatic acyl donor acetyl-CoA, suggesting that the aromatic ring of caffeoyl- and coumaroyl-CoA may interfere with Vv3AT enzyme function when present in high concentrations. This suggests that in planta substrate availability may have a 2-fold effect on the types of acylated anthocyanins found in a particular cultivar, as aromatic acyl donor concentrations, either too low or too high, will have a negative effect on the occurrence of that type of acylation event. An intriguing observation, reported also for the kinetic parameters of Arabidopsis BAHD family acyltransferase (D’Auria et al., 2007), was the approximately 10-fold discrepancy in k  _cat values of Vv3AT determined for the donor or acceptor substrate, with the other substrate at a fixed concentration above its K  _m. The Arabidopsis BAHD family acyltransferase At5MAT showed comparable k  _cat values (0.12 and 0.2 s⁻¹, respectively) for the acceptor substrate cyanidin 3-O-[2″-O-(xylosyl)-6″-O-(p-O-(glucosyl)-p-coumaroyl) glucoside]5-O-glucoside and the donor substrate malonyl-CoA but an approximately 10-fold difference between the k  _cat values for a second acceptor substrate, cyanidin 3-O-[2″-O-(2‴-O-(sinapoyl) xylosyl)6″-O-(p-O-(glucosyl)p-coumaroyl) glucoside]5-O-glucoside, and the donor substrate malonyl-CoA (0.00012 and 0.0013 s⁻¹, respectively). Vv3AT k  _cat values were 13.93 and 1.11 s⁻¹ for malvidin-3-glucoside and acetyl-CoA, respectively.

Vv3AT activity was also tested with donor substrates used by other BAHD protein family members (Fig. 3; Supplemental Table S3). Most characterized anthocyanin acyltransferases utilize malonyl-CoA as the major acyl donor. Vv3AT was able to utilize malonyl-CoA, but it had much lower affinity for this substrate compared with acetyl-, coumaroyl-, and caffeoyl-CoA (Table II). This may explain the lack of malonylated anthocyanins in grapevine, although the availability of this acyl donor may also play a role. Vv3AT could also acylate cyanidin-3,5-O-diglucoside, but at lower efficiency compared with the monoglucosides tested (Table II). This preference for a particular glycosylation pattern has been reported before. For example, the Arabidopsis anthocyanin acyltransferases, At3AT1 and At3AT2, had over 100-fold lower k  _cat values when diglucosides were used as substrates compared with monoglucosides (Luo et al., 2007), while anthocyanin malonyltransferases from chrysanthemum (Dendranthema × morifolium; Suzuki et al., 2004a) and Dahlia variabilis (Suzuki et al., 2002) could not acylate pelargonidin-3,5-O-diglucoside. Recombinant Vv3AT could not acylate cyanidin-3-O-rutinoside in vitro, in agreement with the observations of transgenic tobacco flowers constitutively expressing Vv3AT (Fig. 8). Vv3AT uses 3-O-glucoside substrates and acylates the sugar molecule at the 6″ position. As such, it has similar activity to a number of previously characterized BAHD enzymes with anthocyanin-3-O-glucoside-6″-O-acyltransferase activity. These include the aromatic acyltransferases Perilla frutescens HYDROXYCINNAMOYL COENZYME A:ANTHOCYANIN 3-O-GLUCOSIDE-6′′-O-ACYLTRANSFERASE as well as At3AT1 and At3AT2 and the aliphatic acyltransferases D. variabilis MALONYL COENZYME A:ANTHOCYANIN 3-O-GLUCOSIDE-6′′-O-MALONYLTRANSFERASE, Sc3MaT (from Senecio cruentus), and MALONYL COENZYME A:ANTHOCYANIN 3-O-GLUCOSIDE-6″-O-MALONYLTRANSFERASE1 (from D. morifolium; Yonekura-Sakakibara et al., 2000; Suzuki et al., 2002, 2003, 2004a; Luo et al., 2007). Yet, Vv3AT is set apart from these enzymes by its ability to utilize both aliphatic and aromatic acyl donor substrates with similar specificities, a novel activity for this enzyme family.

Suzuki et al. (2007) showed the versatility of these enzymes by engineering in vitro the aliphatic anthocyanin acyltransferase Ss5MaT1 so that it could use both aliphatic and aromatic acyl donors. They converted the Val-39-Arg-40-Arg-41 amino acids in Ss5MaT1, the two Arg residues having positive charge, to the uncharged amino acids Met-Leu-Gln to create the mutant Ss5AT306. This mutant enzyme could use p-coumaroyl-CoA and caffeoyl-CoA acyl donors in addition to malonyl-CoA, the major donor used by Ss5MaT1. Sequence comparison of Ss5MaT1 with Dm3MaT3, a malonyl-CoA:anthocyanin 3-O-glucoside-6″-O-malonyltransferase for which the crystal structure was solved by Unno et al. (2007), suggests that these three residues form part of the binding pocket for the acyl acceptor rather than the acyl donor. Suzuki et al. (2007) hypothesized that the amino acid substitutions in Ss5AT306 were not directly involved in acyl donor binding but rather resulted in a conformational change of the active site, which in turn allowed for the binding of aromatic acyl donors. The equivalent amino acids in Vv3AT are also uncharged (i.e. Ala-38-Ser-39-Asn-40), which may contribute to the flexibility of this enzyme toward acyl donor substrates. Homology modeling of the grapevine enzyme and Ss5AT306 indicated that the corresponding regions in both proteins share significant structural similarities and are in each case distal from the likely binding site for malonyl-CoA. Additionally, His-151, which corresponds to the strictly conserved His-170 in members of the BAHD family of enzymes, is located farther from malonyl-CoA in Vv3AT (Supplemental Methods, Results, and Discussion; Supplemental Figs. S11 and S12). The homology model shows also that the channel into which the acyl donor substrate is predicted to bind is slightly narrower in Vv3AT than in the other enzymes for which structural data are available (Supplemental Fig. S13). Taken together, these data provide good preliminary evidence to explain the substrate specificity of the grapevine enzyme.

Interestingly, the four amino acids in Vv3AT (Ile-Phe-Tyr-Tyr) following the Ala-38-Ser-39-Asn-40 residues discussed above are quite divergent from those seen in many other flavonoid acyltransferases, which were demonstrated to contain the conserved motif (Leu/Val)-X-Phe-Tyr by Suzuki et al. (2007; Fig. 4). Another key amino acid hypothesized to be involved in governing acyl donor preferences corresponds to the positively charged Arg-178 in Dm3MaT3. Suzuki et al. (2007) showed that aromatic anthocyanin acyltransferases share a Phe at this residue, which may dictate their coumaroyl-CoA preference. The equivalent amino acid residue in Vv3AT is an Ala, divergent from that in other anthocyanin acyltransferases capable of utilizing aromatic acyl donors.

CONCLUSION

In this study, transgenic grapevines with altered expression of VvMYBA genes were used in microarray analyses to show that VvMYBA is a positive regulator of the later stages of anthocyanin biosynthesis, including their glycosylation, methylation, acylation, and transport into the vacuole. Some genes controlling enzymatic steps upstream of VvUFGT were also affected. From this microarray analysis, we identified and subsequently characterized, to our knowledge, the first anthocyanin acyltransferase (Vv3AT) from grapevine and demonstrated this enzyme activity in planta and in vitro. It has a preference for anthocyanin monoglucoside molecules and can use both aromatic and aliphatic acyl-CoA thioesters as substrates, an activity that is novel for previously characterized naturally occurring BAHD acyltransferases, which show a strong preference for one type of thioester. Vv3AT is capable of synthesizing the common acylated anthocyanins identified in grapevine. This study also shows that transcription from the promoter of the Vv3AT gene is activated by VvMYBA1. This provides an example of direct TF-driven regulation of a BAHD gene family member and adds to the current understanding of the anthocyanin biosynthetic pathway and how it is regulated in plants. The cv Pinot Noir does not produce acylated anthocyanins, as it appears to contain a truncated version of Vv3AT. This research will assist breeding programs aimed at producing varieties with good potential for stable red wine color and provides valuable information to those wishing to engineer anthocyanin acyltransferases and/or anthocyanin structures.

MATERIALS AND METHODS

Plant Material

Transgenic grapevines (Vitis vinifera) plus controls were all glasshouse grown in potting mix (20 L of composted pine [Pinus spp.] bark, 10 L of river sand, 30 g of FeSO₄, 60 g of pH amendment, and 140 g of long-life Osmocote) in ambient light with a night break. Day and night temperatures were approximately 27°C and 22°C, respectively. For microarray experiments, whole berries were sampled from three independent transgenic lines of cv Chardonnay and cv Shiraz. Bunches were harvested close to ripeness based on average total soluble solids, which were between 20 and 24 degrees Brix (Supplemental Table S6). A sample consisted of a single bunch except when there were fewer than 100 berries, in which case bunches were pooled. Skin samples were first removed from fresh berries. All samples were frozen in liquid N₂ and stored at −80°C. Due to unsynchronized flowering of the glasshouse-grown vines, sampling occurred throughout the year.

Berry samples of cv Cabernet Sauvignon and cv Malian and cv Shalistin varieties were collected from grapevines grown in a commercial vineyard at Langhorne Creek, South Australia (35°17′30″ south, 139°2′33″ east), in the season of 2010/11. Samples were collected at 2, 4, 6, 8, 9, 10, 12, 15, and 18 wpf. Four to six bunches were randomly selected from the same 20 vines (per variety), and berries from these were pooled and frozen in liquid N₂. Veraison occurred at 9 wpf (when 50% of the cv Cabernet Sauvignon and cv Malian berries were colored).

Extraction of RNA from cv Cabernet Sauvignon and cv Pinot Noir berry samples was described by Dunlevy et al. (2013). Genomic DNA extractions were performed previously by Walker et al. (2007).

Stable Transformation of Grapevines

A pART7 vector containing the VvMYBA1 cDNA sequence between a cauliflower mosaic virus 35S promoter and an octopine synthase gene) transcriptional terminator (Gleave, 1992) had been constructed previously (Walker et al., 2007). This expression cassette was excised from pART7 using a NotI restriction enzyme and ligated to the plant expression vector pART27 (Gleave, 1992) to create the 35S:VvMYBA1 construct.

Primers A13pfxho and A13prasp (containing XhoI and Asp718 restriction sites, respectively; Supplemental Table S7) were used to amplify a 427-bp sense fragment from the 3′ end of the VvMYBA1 gene, and primers A13pfxba and A13prcla (containing XbaI and ClaI restriction sites, respectively; Supplemental Table S7) amplified the same fragment in the antisense direction by PCR (RED TAQ; Invitrogen). PCR cycling conditions were as follows: 94°C for 3 min; followed by 35 cycles of 94°C for 30 s, 58°C for 30 s, and 72°C for 2 min; then a final extension at 72°C for 10 min. Fragments were ligated to pDrive (Qiagen), from which the sense and antisense fragments were excised using XhoI/Asp718 and XbaI/ClaI restriction enzymes, respectively. These fragments were sequentially ligated to the pN6 vector (Wesley et al., 2001). The silencing cassette was excised from pN6 using a NotI restriction enzyme, and this fragment was ligated into pART27 to create the VvMYBA-silencing (VvMYBAsi) construct.

The transformation vector with VvMYBA1 expressed under its own promoter was generated by PCR ligation to join the promoter amplified from bacterial artificial chromosome CS63P1 with the cDNA sequence of the gene, both from cv Cabernet Sauvignon (Walker et al., 2007). PCR was performed with Pfu DNA polymerase (Stratagene) using primers A1prof1 and A1promr1 and the bacterial artificial chromosome template, while the cDNA was amplified with primers A1rxba and A1Ex2f with annealing temperature of 58°C. Products were extracted from a TAE agarose gel and used as the template in a further 10 rounds of PCR. A 3-kb fragment was then ligated to Zero Blunt (Invitrogen), sequenced, and then transferred to pART7NAPX, a promoterless derivative of pART7 (Goetz et al., 2006), using Asp718 and XbaI restriction sites. The construct was transferred to pART27 utilizing NotI restriction sites.

VvMYBA expression or silencing pART27 constructs were inserted into Agrobacterium tumefaciens strain EHA105, which was used to transform embryogenic callus utilizing the method of Iocco et al. (2001). Anthers of immature flowers of cv Shiraz (BVRC12) and cv Chardonnay (I10V1) from Coombe Vineyard, University of Adelaide, Waite Campus, were used to initiate calli on PIV and HT media, respectively. Embryogenic callus from both cultivars was maintained on C1 for 12 to 20 months prior to transformation. Transformed callus was selected using 100 µg mL⁻¹ kanamycin. Germinated embryos were cut to remove the roots and the tips, comprising the cotyledons supported on approximately 5 mm of hypocotyl, and transferred to shooting medium containing 10 µm benzylaminopurine (BAP) for shoot development. Rooted plantlets, whose leaves tested positive by PCR to the transgene, were deflasked into potting mix and hardened off over 3 weeks in the glasshouse.

RNA Extractions, cDNA Synthesis, Labeling, and Microarray Experiments

Frozen whole berry samples were ground to a fine powder under liquid nitrogen using a chilled grinding mill (IKA) and a mortar and pestle. Total RNA was extracted using a modified perchlorate method described previously by Boss et al. (2001). Genomic DNA was removed using DNase (RNase-free DNase; Qiagen) in conjunction with the RNeasy Mini kit (Qiagen). A spectrophotometer (NanoDrop 1000 V3.7.1; Thermo Fisher Scientific) and a bioanalyzer (Bioanalyser Chip RNA 7500 series II; Agilent) were used to determine RNA quantity and quality by ensuring that the absorbance ratios A  ₂₆₀/A  ₂₈₀ and A  ₂₆₀/A  ₂₃₀ were 1.8 or greater and the RNA integrity number was 1.7 or greater.

The cDNA synthesis, labeling, and chip hybridization were all carried out according to the NimbleGen Arrays User’s Guide: Gene Expression Analysis v3.2 protocols (Roche) using the NimbleGen microarray 090818 Vitis exp HX12 (Roche). The microarray was scanned (ScanArray 4000XL; Perkin-Elmer) at 532 nm (Cy-3 absorption peak) in conjunction with GenePix Pro-7 software (Molecular Devices) to produce high-resolution images. Images were then analyzed using NimbleScan version 2.5 software (Roche) which used a robust multichip average procedure to produce normalized expression data for each gene derived from the average of the signal intensities of the four probes on the microarray for that gene.

Normalized expression values were converted to log₂ values, and a Pearson correlation analysis was carried out to evaluate the robustness of the biological replicates in each sample. A gene was considered to be expressed if the normalized expression value for at least two of the three biological replicates was higher than the value obtained by averaging the fluorescence of negative controls present on the chip. A two-class unpaired significance analysis of microarrays was utilized using TMeV software (Saeed et al., 2003) to compare the expression values between transgenic lines and the respective controls. The false discovery rate was set to 2.5% for the cv Chardonnay and cv Shiraz data sets.

To find gene expression that was altered due to VvMYBA in a consistent manner in both cv Chardonnay and cv Shiraz berries, genes that were up-regulated in the red cv Chardonnay berries (positive red-white ratio) and conversely down-regulated in white cv Shiraz berries (negative white-red ratio), or vice versa, were of interest. Only genes with FC ≥ 2 in both the cv Chardonnay and cv Shiraz data sets were included.

Nucleic Acid Extractions and cDNA Synthesis for qPCR

RNA extraction and cDNA synthesis for cv Cabernet Sauvignon and cv Pinot Noir berry developmental series were conducted by Dunlevy et al. (2010).

All other RNA extractions were performed using the Spectrum Plant Total RNA Kit (Sigma-Aldrich) and On-column DNase I Digestion Kit (Sigma-Aldrich) for genomic DNA removal. cDNA was synthesized with a reverse transcriptase kit (Phusion RT-PCR Kit; Finnzymes).

DNA was extracted from young leaves of cv Cabernet Sauvignon and cv Pinot Noir vines previously by Walker et al. (2007) and from young transgenic tobacco (Nicotiana tabacum) leaves using the ISOLATE Plant DNA mini kit (Bioline).

Analysis of Gene Expression

Specific primers were designed to amplify 100- to 300-bp products from Vv3AT and reporter genes (Supplemental Table S7). The specificity of each primer pair was confirmed by PCR, sequencing, and detection of a single peak of fluorescence from melt curves during qPCR. cDNA was diluted 1:40 in sterile Nanopure water (Thermo Fisher Scientific) before use. qPCR experiments were conducted using a thermocycler (LightCycler 480 II instrument; Roche). Each sample was assayed in triplicate in a reaction volume of 15 µL made up of 5 µL of diluted cDNA and 0.5 µm of each primer (Supplemental Table S7) in 1× LightCycler 480 SYBR Green I Master Mix (Roche). Thermocycling conditions were as follows: initial activation at 95°C for 5 min; followed by 45 cycles of 95°C for 20 s, 58°C for 20 s, and 72°C for 20 s; then a final extension at 72°C for 5 min. Reactions were then heated to 95°C for 5 min, cooled to 50°C for 45 s, and then heated to 95°C at a 0.11°C s⁻¹ ramping rate to produce melt curves. For each gene, standard curves were produced from a linear dilution series of target DNA fragments created by PCR, from which sample transcript concentration was determined. These concentrations were normalized against an average value obtained from three housekeeping genes, UBIQUITIN, ACTIN2, and VvEF1α-2 (GenBank accession nos. CF406001, AF369525.1, and TC38276, respectively), and are reported as relative transcript levels.

Isolation of the Vv3AT Gene from Grapevine

Primers flanking the predicted start and stop codons of the VIT_03s0017g00870 gene (Supplemental Table S7) based on the 12× grapevine genome (V.1 version; http://genomes.cribi.unipd.it/grape/) were used to amplify cv Cabernet Sauvignon and cv Pinot Noir genomic DNA and cDNA clones using PCR (Platinum Taq DNA Polymerase; Invitrogen). Cycling conditions were as follows: 95°C for 5 min; followed by 35 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 2.5 min; then a final extension at 72°C for 10 min. Fragments were ligated to pDrive (Qiagen) and sequenced.

Sequence Analysis and Phylogenetic Tree Construction

All sequencing was carried out at the Australian Genome Research Facility in Adelaide using their purified DNA sequencing service. Nucleotide and protein sequence alignments were carried out using AlignX (a component of Vector NTI Advance 11.0; Invitrogen) except for the alignment used in the phylogenetic analysis, for which a ClustalW alignment was used. In this case, the final sequence alignment was generated by manually editing a ClustalW alignment (version 1.83) to select for conserved positions (Larkin et al., 2007). Phylogenies were constructed using the BEAST (Bayesian Evolutionary Analysis by Sampling Trees) version 1.7.5 package (Ayres et al., 2012; Drummond et al., 2012). The WAG (Whelan and Goldman) substitution model with gamma + invariant site heterogeneity was used with a strict molecular clock. A random starting tree was used with the Yule Process tree prior to determining speciation. A Markov chain Monte Carlo maximum chain length of 100 million was set, and the tree was run until completion. Tracer version 1.5 was used to ensure convergence by assessing the estimated sample size and the likelihood of the estimated parameters. TreeAnnotator version 1.7.5 was used to generate the maximum clade credibility tree with a burn in of 50,000. The tree was then annotated using FigTree (Rambaut, 2009). RAxML was used to confirm the structure of the BEAST tree, and the ZmGlossy2 and AtCER2 sequences were set as the outgroup (Stamatakis et al., 2008).

Production of Genetically Modified Tobacco Expressing the Vv3AT Coding Sequence

Primers were designed to the Vv3AT gene to include an XhoI restriction site immediately 5′ of the start codon and an Asp718 site immediately 3′ of the stop codon (Supplemental Table S7). The gene fragment was amplified by PCR, ligated to pDrive (Qiagen), and sequenced. The gene was then inserted into the multiple cloning site of the pART7 vector (Gleave, 1992). This expression cassette was excised from pART7 using NotI restriction enzyme and ligated to the plant transformation vector pART27uGFP (derived from pART27 containing GFP driven by the Arabidopsis [Arabidopsis thaliana] ubiquitin promoter) to create the 35S:Vv3AT construct.

A. tumefaciens strain LBA4404 containing the 35S:Vv3AT construct was used to transform tobacco var Samsun, based on the methods of Horsch et al. (1985). Bacteria were grown on Luria broth plates with appropriate antibiotics and 200 µm acetosyringone at 28°C for 4 d, then suspended in Murashige and Skoog (MS) medium (1× MS salts, 1× Gamborg’s vitamins, and 30 g L⁻¹ Suc) to an optical density at 600 nm of 1. Incisions were made on the abaxial side of tobacco leaves parallel to the midrib, submerged in the MS medium/A. tumefaciens mixture for 10 min, blotted onto sterile filter paper, and transferred (abaxial side up) to MS plates containing 1 µm each of α-naphthaleneacetic acid and BAP (Sigma) for 4 d at 20°C. Leaf pieces were washed in MS medium containing 500 µg mL⁻¹ cefotaxime, blotted on sterile filter paper, and transferred (abaxial side down) onto MS plates containing 1 µm each of α-naphthaleneacetic acid and BAP, 500 µg mL⁻¹ cefotaxime, and 100 µg mL⁻¹ kanamycin. These were kept at 27°C under a 16/8-h photoperiod and transferred to fresh medium every 2 weeks. Shoots about 1 cm in length were transferred onto MS plates containing 100 µg mL⁻¹ kanamycin. Six independent transformants, positive for the Vv3AT construct by PCR analysis, were transferred to potting mix and hardened off in the glasshouse under the same conditions as the transgenic grapevines. Control tobacco plants were grown from cuttings in tissue culture on MS medium and transferred to the glasshouse simultaneously. Vv3AT transcript levels in flowers of all six transformant lines were determined by qPCR as described above.

Vv3AT Protein Expression and Purification

Vv3AT was amplified using primers VvBAHDNotI_F1 and VvBAHDXhoI_R1 (Supplemental Table S7) and ligated to the pET30a(+) expression vector (Novagen, Merck), using the XhoI and NotI restriction sites, to generate an N-terminal His-Vv3AT fusion protein.

Escherichia coli [pBL21(DE3)] cells were cotransformed with pRIL (Stratagene) and either pET30a:His-Vv3AT or the empty pET30a. Cultures were used to produce recombinant His-Vv3AT protein according to an autoinduction, high-density culturing method described by Studier (2005). Cells were pelleted and lysed, and the lysate was clarified as described previously by Dunlevy et al. (2010). His-tagged protein was purified, although not to homogeneity, by His-tag affinity chromatography, and the final concentration of recombinant protein was estimated using a His-Tag Protein ELISA Kit (Cell Biolabs) according to the manufacturer’s instructions (Supplemental Figs. S6 and S7) and visualization of anti-His immunoreactive bands on western blots.

Enzyme Assays

Recombinant enzyme assays were conducted using recombinant His-Vv3AT protein from a single purification batch and repeated with two other batches grown and purified independently. All reactions were carried out in triplicate for the first batch and in duplicate thereafter; the same protein fraction from E. coli cells containing an empty pET30a vector was used as a negative control. Suitable assay parameters were determined by first testing various buffer, pH, and temperature conditions (Supplemental Fig. S8). Reactions were conducted in a volume of 50 µL using 0.1 m sodium phosphate buffer, pH 6.5, and containing 5 µL of CoA-conjugated acyl donor dissolved in 0.1 m sodium phosphate buffer, 1 µL of anthocyanin acyl acceptor dissolved in 100% methanol, and 0.5, 2, or 0.67 µL of concentrated protein fraction from purification batches 1, 2, and 3, respectively. When determining K  _m, acyl donor and acceptors were maintained at 200 µm, while the concentration of the other substrate was varied. Reactions were carried out at 30°C for 20 min, which was within the linear range of the reaction based on preliminary tests (Supplemental Fig. S9). The reaction was stopped by the addition of 50 µL of 100% methanol. The K  _m was calculated by creating a Lineweaver-Burk plot and using the following equation: 1/V = (1/V  _max) + ((K  _m/V  _max) × (1/[S])).

Extraction and Detection of Anthocyanins

Anthocyanins were extracted from 100-mg aliquots of ground, frozen tobacco flowers with 300 µL of 0.3% (v/v) formic acid in 70% (v/v) methanol, sonicated for 20 min in an ice bath, and centrifuged to pellet debris. A total of 25 µL of the tobacco flower extraction supernatants and recombinant enzyme assay reactions was used for anthocyanin separation and quantification.

Anthocyanins were separated using a Hewlett Packard 1100 HPLC system with a Wakosil C18 analytical column (3 μm, 150 mm × 4.6 mm; SGE) protected by a C18 guard column (SGE), following the method described by Downey and Rochfort (2008). Anthocyanin concentrations in tobacco extracts were determined by comparison of peak areas with a standard curve of cyanidin-3-O-rutinoside (0–37.8 ng µL⁻¹, r  ² = 0.998). Acylated anthocyanin product concentrations within the recombinant enzyme assays were determined by comparison with a malvidin-3-O-glucoside standard curve (0–144.1 ng µL⁻¹, r  ² = 0.998). Anthocyanin peaks were identified by their tandem mass spectrometry parent and major daughter ions utilizing the HPLC method as described above coupled to a 6410 triple quadrupole mass spectrometer (Agilent) as described by Downey and Rochfort (2008). Supplemental Table S5 summarizes the ions detected for each compound, which were similar to reported values (Luo et al., 2007; Downey and Rochfort, 2008).

Luciferase-Binding Assays

Promoter regions, 711 bp upstream of the putative start codon, were amplified from cv Cabernet Sauvignon and cv Pinot Noir DNA using primers VvBAHDPrF2_SacI and VvBAHDPrR1_BglII (Supplemental Table S7). Products were ligated to the firefly (Photinus pyralis) luciferase (LUC) plasmid pLUC (Horstmann et al., 2004) using the SacI and BglII restriction sites. VvMYC1 (NM001281253) was amplified using primers VvMyc1 F2 and R2 (Supplemental Table S7) and ripe cv Shiraz berry skin cDNA, sequenced, and inserted in pART7 (Gleave, 1992). Transient transfection of cv Chardonnay suspension cultures and luciferase assays were carried out as described by Harris et al. (2013). All transfections were done in at least triplicate and repeated twice. A statistical comparison of relative luciferase activity from the different promoters and bHLH TFs used in these experiments was conducted using a one-way ANOVA (P < 0.05) within the SPSS 16.0 statistical software package (SPSS).

Sequence data from this article can be found in the Microarray Gene Expression Database (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=cnmxwmmsfjctvgt&acc=GSE56915) under accession numbers Vv3AT-CS1, KM267559; Vv3AT-CS2, KM267560; and Vv3AT-PN, KM267561.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Total anthocyanin concentration in wild-type and transgenic cv Chardonnay and cv Shiraz whole berries with altered VvMYBA expression.

• Supplemental Figure S2. Transcript abundance, determined by qPCR, of VvMYBA transgene, VvMYBA, VvUFGT, and Vv3AT in wild-type and transgenic cv Chardonnay and cv Shiraz whole berries with altered VvMYBA expression using samples that were used in microarray experiments.

• Supplemental Figure S3. Transcript abundance, determined by qPCR, of VvCHS1, VvCHS2, VvCHS3, and DFR in wild-type and transgenic cv Chardonnay and cv Shiraz whole berries with altered VvMYBA expression using samples that were used in microarray experiments.

• Supplemental Figure S4. Photographs of cv Cabernet Sauvignon, cv Malian, and cv Shalistin berries.

• Supplemental Figure S5. Vv3AT transcript levels in flowers of transgenic tobacco lines expressing Vv3AT.

• Supplemental Figure S6. Protein gels from the His-tag affinity chromatography purification and quantification of Vv3AT.

• Supplemental Figure S7. Western blots from the His-tag affinity chromatography purification and quantification of Vv3AT.

• Supplemental Figure S8. Optimization of recombinant Vv3AT protein assay buffer, pH, and temperature conditions.

• Supplemental Figure S9. Determination of the linear range of recombinant Vv3AT protein assay reactions for a single purification.

• Supplemental Figure S10. Activity of Vv3AT protein with different acyl acceptor molecules at varying concentrations.

• Supplemental Figure S11. Structure superimposition of Ss5MaT1, Ss5AT306, and Vv3AT with Dm3MaT3.

• Supplemental Figure S12. The interaction potentials of key residues with malonyl-CoA ligand in Dm3MaT3, Ss5AT306, Ss5MaT1, and Vv3AT.

• Supplemental Figure S13. The overall hydrophobicity profile of the acyl donor-binding pocket and modeled interaction with a malonyl-CoA ligand for Dm3MaT3, Ss5AT306, Ss5MaT1, and Vv3AT.

• Supplemental Table S1. Differentially expressed genes in red cv Chardonnay 35S:VvMYBA1 berries by SAM analysis (provided as Excel file).

• Supplemental Table S2. Differentially expressed genes in white cv Shiraz VvMYBAsi berries by SAM analysis (provided as Excel file).

• Supplemental Table S3. Characterized BAHD proteins used in phylogenetic tree analysis.

• Supplemental Table S4. Absolute chemiluminescence values from luciferase reporter promoter activation assays in grapevine suspension cells.

• Supplemental Table S5. Mass spectrometry parent ion and mass spectrometry 2 major daughter ions detected by LC-MS/MS used to identify acylated anthocyanins in recombinant Vv3AT bioassays and transgenic tobacco flowers.

• Supplemental Table S6. Total soluble sugars in berries used in microarray experiments.

• Supplemental Table S7. Details of primers.

• Supplemental Methods S1. Additional methods, results, and discussion.

• Supplemental References S1. Additional references.

ACKNOWLEDGMENTS

We thank Mac Cleggett and Anne McLennan (Cleggett Wines) for providing samples of the cv Cabernet Sauvignon mutant grapes from the vineyard in Langhorne Creek; Christine Böttcher for expert advice on protein expression and purification techniques as well as article preparation; and Elizabeth Lee, Karin Sefton, Lorraine Carruthers, Sue Maffei, Emily Nicholson, Adelle Craig, and Bronwyn Smithies for technical assistance.

Glossary

 
• TF

  transcription factor

 
• QTL

  quantitative trait locus

 
• FC

  fold change

 
• qPCR

  quantitative PCR

 
• cDNA

  complementary DNA

 
• wpf

  weeks post flowering

 
• LC-MS/MS

  liquid chromatography-tandem mass spectrometry

 
• BAP

  benzylaminopurine

 
• MS

  Murashige and Skoog

LITERATURE CITED

Author notes