The grapevine VvibZIPC22 transcription factor is involved in the regulation of flavonoid biosynthesis

Highlight Functional analysis of the grapevine VvibZIPC22 indicates that this factor is involved in the regulation of different branches of the flavonoid pathway, either alone or together with other factors.

Introduction coat of grapes, and fulfil several important functions. They (i) mediate the response to biotic and abiotic stresses (temperature, UV light, nutrition, water deficit); and (ii) provide pigmentation to flowers and fruits, thereby also influencing the quality and typicity of wines. Moreover, they are associated with health-promoting effects ascribed to grape-rich diets including wine (Flamini et al., 2013;Teixeira et al., 2013).
Due to this wide range of functions, the synthesis of these compounds is finely regulated, mainly at the transcriptional level. The expression of the flavonoid biosynthetic genes is often regulated at the organ-and stage-specific level during berry development, thus leading to the accumulation of specific flavonoids in the different organs and phases. Although flavonoid and especially anthocyanin biosynthesis has been extensively studied, also in grapevine, its regulation is not yet completely understood (Hichri et al., 2011a). The whole system is under the tight control of several transcription factors, which can mutually interact giving rise to regulatory specificity (Grotewold, 2006). In all species analysed to date, these factors mainly belong to the family of the R2R3-MYB domain-containing proteins (Hichri et al., 2011a;Li, 2014). However, there is evidence which indicates bZIP proteins as involved together with MYB and other factors in the transcriptional control of flavonoid biosynthesis depending on light (Hartmann et al., 2005;Czemmel et al., 2009;Stracke et al., 2010).
Plant bZIPs bind to ACGT-containing elements (Izawa et al., 1993) and usually regulate transcription by forming dimers (Strathmann et al., 2001;Ehlert et al., 2006) and interacting with non-bZIP proteins (Schutze et al., 2008), to gain flexibility in their regulatory capacity. In particular, bZIP heterodimers of the C/S1 groups of Arabidopsis have been implicated in stress response and development  and in sugar-regulated control of amino acid metabolism during the plant response to starvation (Dietrich et al., 2011).
Flavonoid biosynthesis in grapevine is also regulated by environmental stimuli (Teixeira et al., 2013) and is particularly affected by light (Fujita et al., 2006;Matus et al., 2009;Koyama et al., 2012). Koyama et al. (2012) described the influence of light quality on the regulation of flavonoids in young berry skins: visible light primarily induces proanthocyanidin (PA) biosynthesis, whereas UV light specifically induces flavonol biosynthesis. In Arabidopsis, the light regulation of flavonol biosynthesis is directed by MYB and bZIP factors which co-operatively bind to the light regulatory units (LRUs) present in the CHS and FLS promoter sequence (Hartmann et al., 2005). A well-studied example is the Arabidopsis HY5 gene, encoding a bZIP factor which regulates numerous genes, such as AtCHS, AtFLS, and AtMYB12, during photomorphogenesis (Lee et al., 2007;Stracke et al., 2010). The presence of LRUs in the promoters of VviFLS1 and VviMYBF1 and their light responsiveness led Czemmel et al. (2009) to propose them as targets of a grapevine HY5 homologue.
Recently, we have identified two grapevine bZIP genes, namely the predicted VvbZIP14 and VvbZIP22 (Liu et al., 2014), within two quantitative trait locus (QTL) regions associated with the fine tuning of flavonol content in mature grapes, on chromosomes 5 and 7, respectively. Notably, these bZIPs were also expressed at different levels in grapes of 'Syrah'×'Pinot Noir' progeny showing different contents of flavonols in the mature skin . The QTL on chromosome 7, containing VvbZIP22, was specifically related to kaempferol content, while the other, containing VvbZIP14, was also associated with anthocyanin content . Starting from this evidence, in this study, we decided to characterize further VvbZIP22, here renamed VvibZIPC22 according to the recently proposed grapevine genome nomenclature system (Grimplet et al., 2014).

Materials and methods
Grapevine material and sampling Inflorescences and clusters from a representative sample of V. vinifera cv. Pinot noir (clone ENTAV115) were collected at six developmental stages during the 2011 season at the Giaroni experimental field of FEM (Edmund Mach Foundation, San Michele a/Adige, Italy, 46°18'N, 11°13'E). To be more representative, different parts of each inflorescence/cluster were collected from shady and sunny sides. The six stages corresponded to flowering (50% opened flowers, E-L 23), pepper corn (E-L 29), pre-véraison (hard green berries, E-L 33), véraison (50% coloured berries, E-L 35), post-véraison (berries at 16° Brix, E-L 36), and maturity (berries at 18° Brix, E-L 38). During sampling, skins were separated from flesh and seeds directly in the field when feasible, precisely from véraison to maturity stages, and immediately frozen in liquid nitrogen.
In the 2012 season, vegetative organs (young leaf, mature leaf, bud, root, internode, and tendril), germinated seeds, green (E-L 20) and mature inflorescences (E-L 23) (divided into calyptra, stamen, and pistillum), and berries at different stages (E-L 29, 34, and 36) were sampled from the same Pinot Noir plants considered in 2011 and immediately frozen in liquid nitrogen. In the case of berries at stages E-L 34 and 36, seeds were removed during grinding. For each stage and organ of the panel, three different plants (biological replicates) were considered.
For light induction experiments, dormant hardwood cuttings of grapevine V. vinifera L. cv. Chardonnay were collected from a vineyard in Neustadt/W, Germany (49°22'9''S, 8°10'28''E). Cuttings were grown in sterile Perlite (Knauf, Sittingbourne, UK) under the following conditions: temperature, 25 °C; white light, ~100 µmol m −2 s −1 ; 9 h light cycle. In this hydroponic system, the apical buds burst after 10 d and roots developed after 4 weeks. After 19 d, the rooted plants were subjected to light treatment [4% UV-B and 30% UV-A light (18 W, 6500 K)] under the following conditions: temperature, 23 °C; 9 h light cycle. In the case of treated samples, the third and the fourth leaf from the shoot tip of five plants were collected at 0, 10, 24, 48, and 72 h after the onset of light exposure. Control samples were collected at the same time points, from five other plants grown under the same conditions, but without UV light. At each time point, leaves were pooled, immediately frozen in liquid nitrogen, and stored at -80 °C till use.
All the collected samples were finely ground by liquid nitrogen using frozen metal grinding jars mounted on a Mixer Mill MM 400 (Retsch, Haan, Germany). The powder was maintained at -80 °C maximum for 2 months to be used for both transcriptional and biochemical analyses.
pART7bZIPC22 was used both for the transient expression assays and as an intermediate to obtain pART27bZIPC22 for tobacco transformation. CaMV35S:VvibZIPC22 was therefore transferred into a NotI-digested pART27 binary vector to obtain pART27bZIPC22. Single colonies of Agrobacterium tumefaciens (strain C58C1) transformed with pART27bZIPC22 by electroporation were cultured in 50 ml of selective YEB medium and grown overnight (28 °C at 160 rpm). The cells were harvested by centrifugation at 4000 rpm for 30 min without break and resuspended in 50 ml of infiltration buffer (10 mM MES, pH 5.9 and 150 μM acetosyringone). The suspension was incubated at 28 °C and 50 rpm for 2 h. Thereafter the concentration was measured by spectroscopy (OD 600 nm ) and adjusted to a final concentration of OD 600 nm of 0.3 with H 2 O. Nicotiana tabacum cv. Samsung leaf sections (1 cm 2 ) were further incubated with the bacterial suspension right-side up for 30 min at room temperature. Leaf explants were transferred to a sterile wet filter paper and incubated in the dark for 2 d at room temperature. The explants were then transferred right-side down for 30 min to solid Murashige and Skoog (MS) medium plus timentin (400 mg l -1 ), supplemented with B5 vitamins, 3% (w/v) sucrose, 1 mg l -1 6-benzyladenine (BA), 0.1 mg l -1 α-naphthaleneacetic acid (NAA), 0.8% tissue culture agar, pH 5.6, and 100 mg l -1 kanamycin. Explants were then incubated in the dark for 2 weeks at 25 °C, and then transferred to 25 °C, 16 h light cycle. Finally, kanamycin-resistant plantlets were transferred to soil mix, acclimatized, and grown in the greenhouse. The presence of the transgene in transformed lines was checked by PCR with the primers 35SCaMVpART7 (5'-CAATCCCACTATCCTTCGCA-3') and bZIP-KpnI. Stamens and petal limbs at two developmental stages (open flower without pollen and mature flower) were collected from a representative sample of each transgenic line and wild-type plant and stored at -80 °C until biochemical and gene expression analyses.

Grapevine transient expression experiments
A transient expression system using Chardonnay liquid cell cultures was established as described previously (Bogs et al., 2007). Cells were bombarded with 1.6 µm gold particles using a PDS-1000/ He Biolistic Particle Delivery System from Bio-Rad (München, Germany) with 4481 kPa helium pressure, a vacuum of 86 kPa, and a distance of 9.5 cm. The luciferase assay was performed as set up by Czemmel et al. (2009). Briefly, gold particles were coated with 500 ng of each respective plasmid, giving a total amount of 2 µg of plasmid per transformation for reporter/effector bombardments. Additionally, each bombardment contained a positive control of 100 ng of the Renilla luciferase plasmid pRluc (Horstmann et al., 2004). After incubation in the dark at 27 °C for 48 h, the harvested cells were assayed for luciferase activities using the dual-luciferase reporter assay system (Promega) measured with a Lumat LB 9507 Luminometer (Berthold Technologies, Bad Wildbad, Germany). The relative luciferase activity was calculated as the ratio between the firefly and the Renilla (control) luciferase activities after subtraction of the cell background (ground cells but not bombarded). All transfection experiments were carried out in triplicate, and each promoter experiment was repeated at least three times.

Biochemical analyses of flavonoids in grapevine and tobacco samples
Extraction of anthocyanins, flavonol aglycons, and flavan-3-ol monomers and their analysis by HPLC-DAD (diode array detection) from grapevine samples were performed as described in Mattivi et al. (2006Mattivi et al. ( , 2009) using a Waters 2690 HPLC system equipped with Waters 996 DAD and Empower software (Waters Corporation, Milford, MA, USA). Some modifications were applied to the extraction of flavonols from light-treated and control leaf samples. Briefly, three separate 200 mg aliquots (technical replicates) of frozen powder were extracted with 1.5 ml of methanol by rotation (30 rpm for 20 min). After centrifugation (13 000 rpm for 5 min), 1.5 ml of the supernatant was transferred to a 50 ml flask. The same procedure was adopted in a second extraction with 1.5 ml of methanol. The flask with a total volume of 10 ml (3 ml of methanolic extract+2 ml of methanol+5 ml of trifluoroacetic acid 2 M in water) was used for the following acid hydrolysis as in Mattivi et al. (2006).
Extraction of anthocyanins and flavonols from tobacco stamens was performed starting from three separate 150 mg aliquots (technical replicates) of frozen powder. After the addition of 1.5 ml of methanol, each sample was shaken at room temperature for 20 min. After centrifugation (14 000 rpm for 5 min), each sample was then filtered with 0.22 µm filters into LC vials and subjected to HPLC-DAD analysis.
Each anthocyanin, flavonol, and flavan-3-ol compound was identified by comparison of its retention time and UV spectra at 520, 370, and 280 nm. The concentration (µg g FW −1 ) of each compound was determined using the external standard method, specific for each compound. Values under the limit of detection (LOD) and of quantification (LOQ) were given a value equal to zero in the final quantification.

Quantitative gene expression analysis in grapevine and tobacco samples
Total RNA was isolated from grapevine and tobacco samples with the Spectrum Plant Total RNA Kit (Sigma Aldrich) according to the manufacturer's instructions. RNA was quantified with an ND-8000 nanodrop spectrophotometer and checked for integrity with a 2001-Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). For reverse transcription-PCR and real-time RT-PCR (qRT-PCR) analyses, cDNAs were synthesized using the SuperscriptVILO™ cDNA Synthesis Kit from 1.5 μg of DNAse I-treated RNA (Thermo Fisher Scientific), following the manufacturer's instructions. qRT-PCR analyses were carried out using the Platinum SYBR Green qPCR SuperMix-UDG in a ViiA™ 7 thermocycler (Thermo Fisher Scientific). The 384-well plates were set up according to the sample maximization strategy proposed in Hellemans et al. (2007), and each reaction was run in triplicate. Reaction conditions and the protocol of analysis were the same as adopted in . Grapevine and tobacco housekeeping genes were tested for their stability in each experiment using the GeNorm software (Vandesompele et al., 2002). Normalized relative quantities (NRQs) were then calculated by dividing RQs by a normalization factor, based on the expression of the most stable reference genes (details are given in the figure legends) (Reid et al., 2006). Primer sequences used for the analysis are listed in Supplementary Table S1 at JXB online, as designed in this work or taken from the literature.

Bioinformatics
For the research of plant cis-regulatory elements, the first 1000 bp from the putative transcriptional start of the promoter sequences were screened manually and by the use of PLACE (http://www.dna. affrc.go.jp/PLACE/; Higo et al., 1999).
For phylogenetic studies, the complete CDS of VvibZIP genes belonging to clade B and C (Liu et al., 2014) and of their putative orthologues in different plant species were retrieved from public databases (http://genomes.cribi.unipd.it/DATA/V1 and http:// www.ncbi.nlm.nih.gov/, last accessed 11 December 2015). Codonbased alignments of bZIP CDS sequences were made using the Macse program (Ranwez et al., 2011). Automatically produced alignments were loaded into the Seaview alignment editor and manually edited in amino acid mode to discard misaligned regions. The final selection of alignment columns was saved to produce (i) a 1335 positions long curated alignment of coding clade B VvibZIP sequences and their putative orthologues (Supplementary Dataset S1) and (ii) a 297 positions long curated alignment of coding clade C VvibZIP sequences and their putative orthologues (Supplementary Dataset S2). Bayesian analyses were performed employing the MPI version of Phylobayes (version 1.4) and a general specification of the CAT+GTR+G4 model. Two chains were run on each data set under each model and until 5000 cycles were sampled. Trees were built for each data set based on the two chains, discarding the first 2500 cycles as burn in, which was sufficient for ML (maximum likelihood) parameter values to maximize in all analyses. Parameter values were then sampled every cycle thereafter till the 5000th cycle.
Co-expression analyses were performed by means of the Vitis Expression Studies Platform Using COLOMBOS Compendia Instances (VESPUCCI) (M. Moretto et al., 2016b), a grapevine gene expression compendium (http://vespucci.colombos.fmach.it/) based on COLOMBOS v3.0 technology (Moretto et al., 2016a). The analyses by COLOMBOS were essentially based on contrast relevance and gene similarity scores as thoroughly described in Text S1 from Engelen et al. (2011) and the updates made with the v3.0 release.

VvibZIPC22 belongs to one group of clade C in the phylogenetic tree of VvibZIP factors
The recent phylogenetic analysis of the grapevine bZIP gene family assigned VvbZIP22 to clade C, together with another nine VvbZIP genes. In particular, VvbZIP22, together with VvbZIP14 and VvbZIP37, forms a homologous triplet with collinearity along chromosomes 5, 7, and 14 (Liu et al., 2014). According to the recently suggested grapevine genome nomenclature system (Grimplet et al., 2014), we renamed the VvbZIP factors indicating the clade to which they belong, with the exception of the unresolved members.
To better define the structure of clade C and to identify sequences closely related to VvbZIP22 from other species, the CDS sequences of VvbZIP factors belonging to this clade were aligned and compared with those of bZIP factors from other important plants and crops with significant identity (>50%) to VvbZIP22 (Supplementary Dataset S2). Phylogenetic analysis revealed that clade C can be divided into three main groups, designated C1, C2, and C3 ( Fig. 1), similarly to clade S in Arabidopsis . In particular, VvibZIPC22 is part of group C1 together with VvibZIPC14 and VvibZIPC37 and with previously characterized bZIP genes from flowering species such as N. tabacum (NtBZI-3 and NtBZI-4; Strathmann et al., 2001), Petroselinum crispum (PcCPRF6; Rugner et al., 2001), Arabidopsis thaliana (AtbZIP53; Dietrich et al., 2011), and Solanum lycopersicum (SlbZIP04; named LebZIP2 in Seong et al. 2008). Moreover, other orthologues can be considered: SiOCS1 from Sesamum indicum, TcbZIP53 from Theobroma cacao, PtOCS1 from Populus trichocarpa, PxbOCS1 from Pyrus×bretschneideri, MdbZIP3 from Malus domestica, GmbZIP33 from Glycine max, CsOCS1 and CmOCS1 from Cucumis sativus and Cucumis melo, and CcOCS1 from Citrus clementina. In our analysis, VvibZIP07, VvibZIP13, and VvibZIP47, which in the tree of Liu et al. (2014) belong to clade C, were unresolved with respect to group C1 together with PcCPRF7 from P. crispum and NtBZI-2 from N. tabacum. VvibZIPC02 and VvibZIPC39 fall into group C2 together with members of group S2 in A. thaliana (AtbZIP3,AtbZIP8,AtbZIP42,AtbZIP43,AtbZIP48,AtbZIP58,AtbZIP70,and AtbZIP75), while VvibZIPC44 and VvibZIPC55 form group C3 together with members of group S3 in A. thaliana (AtbZIP4, AtbZIP5, AtbZIP6, and AtbZIP7) (Fig. 1).

VvibZIPC22 expression is flower specific and correlates with kaempferol and quercetin content during berry development
To determine if a significant correlation exists between VvibZIPC22 expression and the content of different flavonoids, we looked at their profiles at six time points of Pinot Noir berry development ( Fig. 2A). VvibZIPC22 expression as well as kaempferol and quercetin aglycons showed similar profiles: they peaked at flowering, reached a minimum between pre-véraison and véraison, and increased slightly towards maturity. It is noteworthy that VvibZIPC22 expression was highest at flowering, when a peak of expression was also reported for VviMYBF1 (Czemmel et al., 2009) and VviFLS1 (Downey et al., 2003) in developing Shiraz berries. The same analysis highlighted that delphinidin-like flavonols had an opposite trend, with maximum accumulation in mature berries ( Fig. 2A). Early and late peaks in the biosynthesis of different flavonols have already been described in the V. vinifera cv. Pinot Noir (Sternad Lemut et al., 2013). On the other hand, we did not find a significant correlation between VvibZIPC22 expression and the synthesis of either total anthocyanin 3-monoglucosides or total flavan-3-ol monomers ( Supplementary Fig. S1).
The expression analysis extended to different grape organs showed that VvibZIPC22 is mostly expressed in the flower, as demonstrated by its level in calyptras and stamens, 13 times higher than in berries at stage E-L 36 (16° Brix) and almost three times the expression observed in the other organs (Fig. 2B).

Induction of VvibZIPC22 expression by light is related to flavonol accumulation
Light-induced flavonol biosynthesis (Downey et al., 2004) is the result of an increase in VviMYBF1 and VviFLS1 expression mediated by light regulatory elements present in both promoters and recognized by MYB and bZIP factors (Czemmel et al., 2009). For this reason, we monitored VvibZIPC22 as well as VviMYBF1 and VviFLS1 gene expression over a 3 d time period together with flavonol accumulation in Chardonnay plants exposed to UV light. Within 10 h of light exposure, the transcript level of VvibZIPC22, VviFLS1, and VviMYBF1 increased ~2.5-, 7.5-, and 22-fold in treated compared with control leaves. VvibZIPC22 and VviFLS1 decreased by almost half at 24 h and 72 h, and VviMYBF1 was barely detectable at these times (Fig. 3A). The peak of induction of the three genes at 10 h of treatment was followed by the accumulation of flavonols starting from 24 h and reaching a maximum at 72 h of light induction (Fig. 3B).
The presence of light response cis-acting elements in the VvibZIPC22 promoter sequence in both Pinot Noir and Chardonnay was a further indication of its light responsiveness. Hartmann et al. (2005) showed that LRUs formed by MYB recognition elements (MREs) and ACGT-containing elements (ACEs) mediate light responsiveness of the AtCHS promoter. Similar elements, designated as ACS (ACGT-containing Fig. 1. Phylogenetic tree of clade C VvibZIP factors and of their putative orthologues. C1, C2, and C3 correspond to the identified groups within clade C, similarly to those identified in clade S in Arabidopsis . Bootstrap values are shown next to the nodes. Gene names with their accession numbers are the following: sequence similar to ACE) and MRS (MYB recognition sequence similar to MRE), were also identified in the promoter region of VviMYBF1 and VviFLS1 (Czemmel et al., 2009). In this study, we investigated 1000 bp of the sequence of Pinot Noir and Chardonnay bZIPC22 upstream of the putative transcriptional start site: they were identical, with the exception of a polymorphism at position -303. This was not unexpected since Pinot Noir is the male parent of Chardonnay. Moreover, we found that both promoter sequences contain an ACS element at a position (-137 bp) fairly well conserved with the ACE site in AtCHS and AtFLS, and with the ACS site in VviFLS1 and VviMYBF1. In addition, an MRS site (-583 bp), possibly recognized by MYB factors, and an IBOXCORE element (-193 bp), involved in binding of MYB factors in lightregulated genes in tomato (Rose et al., 1999), were localized upstream of the ACS site (Fig. 3B), as found in the VviFLS1 promoter (Czemmel et al., 2009).

VvibZIPC22 activates promoters of general and specific flavonoid pathway genes
In order to identify which flavonoid pathway genes could be the potential targets of VvibZIPC22, transient expression experiments were carried out by bombardment of Chardonnay liquid cell cultures and dual-luciferase assays.
Five different promoters were tested: the VviCHS and VviCHI promoters, as genes controlling the general flavonoid pathway, and the VviFLS1, VviUFGT, and VviANR promoters, as genes specifically involved in the synthesis of flavonols, anthocyanins, and PAs, respectively. VvibZIPC22 was tested alone and in combination with VviMYBF1, VviMYC1-VviMYBPA1, and VviMYBA2 or VviMYC1-VviMYBA2 factors to test the transactivation potential on the VviCHI and VviFLS1 promoters in the first case, on the VviCHI and VviANR promoters in the second case, and on the VviUFGT promoter in the third case (Fig. 4).
VvibZIPC22 was able to activate the promoter of VviCHS3, the isoform expressed in red-coloured berry skin during ripening (Goto-Yamamoto et al., 2002), approximately three times more than in the control without VvibZIPC22, suggesting a possible role in the general regulation of the flavonoid pathway (Fig. 4A). VvibZIPC22 also induced the promoter of VviFLS1 to an extent similar to the positive control VviMYBF1 (~3.5-fold) (Fig. 4C). Interestingly, VvibZIPC22 increased the transactivation potential of VviMYBF1 2-fold by acting synergistically on the promoters of both VviCHI (Fig. 4B) and VviFLS1 (Fig. 4C). As published before (Hichri et al., 2010), VviMYBPA1 in combination with the basic helix-loop-helix (bHLH) factor VviMYC1 induced the VviANR promoter (~20-fold), but we observed that its effect  Table S1), calibration against the mean expression value in all the stages, and normalization against VviGADPH and VviACTIN relative expression. Flavonol content was determined by HPLC-DAD analysis. Each value corresponds to the mean and SE of three different biological replicates. F, flowering (50% opened flowers, E-L 23); P, pepper corn (E-L 29); PV, pre-véraison (hard green berries, E-L 33); V, véraison (50% coloured berries, E-L 35); 16°, post-véraison (berries at 16° Brix, E-L 36); 18°, maturity (berries at 18° Brix, E-L 38); NRQ, normalized relative quantity; FW, fresh weight.
is amplified 2-fold by the action of VvibZIPC22, suggesting this bZIP might also play a role in the regulation of PA biosynthesis (Fig. 4D). Conversely, VvibZIPC22 did not display any significant positive effect on the VviUFGT promoter either alone or in combination with the complex VviMYC1-VviMYBA2 and with VviMYBA2 (Fig. 4E).
We also noted that the promoters of VviCHS3, VviCHI, VviFLS1, and VviANR have an ACGT-containing element in a rather conserved position, between 100 bp and 150 bp upstream of the transcriptional start site. In contrast, ACGTcontaining elements are present only in a more distal position in the promoter of VviCHS1 and completely absent in those of VviCHS2 and VviUFGT (Supplementary Table S2).

VvibZIPC22 overexpression in tobacco leads to flavonoid biosynthesis in different flower organs
The role of VvibZIPC22 in vivo was investigated by constitutive overexpression in tobacco plants. Several transgenic lines were generated, propagated, and grown in the greenhouse together with wild-type (WT) plants used for comparison.
We then carried out the following analyses on different flower organs of three WT plants and of six different lines sampled at three development stages: (i) anthocyanin and flavonol content by HPLC-DAD; (ii) PA content by reaction with DMACA; and (iii) gene expression of VvibZIPC22 and some tobacco flavonoid pathway genes by qRT-PCR.
Overexpression of VvibZIPC22 caused line-specific abnormalities, both in the vegetative part (reduced height of the plant, internodes, and size of the leaves) and in the reproductive system (corollas showed visibly increased colour and rounded lobes, and stamens appeared reddish and reduced in size) (Fig. 5A).
As expected, the six lines displayed overexpression of VvibZIPC22 and with a rather limited variability (±25%) (Fig. 5B). As far as secondary metabolites are concerned (Fig. 5C), striking differences, compared with the WT plants, were obtained looking at the flavonol and anthocyanin content in the stamens at both stages and at the PA content in the petal limbs at the second stage. Interestingly, the increase in flavonoids was paralleled by a significant increase in the expression of several structural genes (Fig. 5D). An induction  (underlined) and in the promoters of VviMYBF1 and VviFLS1 (Czemmel et al., 2009). These cis-elements were aligned with the corresponding Arabidopsis elements taken from the PLACE database (Higo et al., 1999) using GeneDoc software (v. 1.6). Nucleotides are labelled from black to light grey according to their conservation. Numbers in front of the alignment indicate the relative distance of each element from the putative transcriptional start site (+1). CHS, chalcone synthase; FLS, flavonol synthase; PN, Pinot Noir; Ch, Chardonnay; ACS, ACGT-containing sequence similar to ACE in the Arabidopsis CHS promoter (accession no. S000355); MRS, MYB recognition sequence similar to MRE in the Arabidopsis CHS promoter (S000356); MYBCORE (S000176), IBOXCORE element (S000199); RRS, R response sequence similar to RRE in the Arabidopsis CHS promoter (S000407).
of NtCHS, NtDFR, NtANS, and NtUFGT was observed in the stamens at the first stage together with a significant increase in the content of cyanidin 3-rutinoside (up to 4-fold more than WT plants in both stages), and of NtFLS together with an increase of kaempferol 3-rutinoside and quercetin 3-rutinoside in the stamens at the second stage (up to 2.6-and 2-fold more than in WT plants, respectively). In some lines, these results are in agreement with the visibly red colour of the stamens (Fig. 5A). Compared with WT petal limbs, the lines also displayed a higher expression of NtPAL and NtANR1 together with a higher content of PAs at the second stage.

In silico approaches provide new preliminary insights into the VvibZIPC22 regulatory network
A specific heterodimerization network of bZIP proteins from clade S (clade C in V. vinifera) and clade C (clade B in V. vinifera) was previously described in Arabidopsis , which is conserved in tobacco (Strathmann et al., 2001) and parsley (Rugner et al., 2001). We thus looked for potential interacting partners of VvibZIPC22 belonging to clade B, by a combination of phylogenetic and co-expression analyses. Our phylogenetic reconstruction ( Supplementary  Fig. S2) of VvibZIP genes from clade B and closely related sequences from other crop species indicated that VvibZIPB38 and VvibZIPB09 are more related to members of Rosaceae and Solanaceae, respectively, as is the case for VvibZIPC22 which shares group C1 with members from the same species.
In silico co-expression analyses against a grapevine gene expression compendium using VvibZIPC22 (VIT_07s0005g01450) as the query gene identified a set of 210 positively co-expressed genes (correlation cut-off=0.68) in 140 contrasts, conditions where VvibZIPC22 showed the highest modulation (Supplementary Table S3). Of the 210 co-expressed  A and B) correspond to the regulators and the promoters of the general flavonoid pathway genes, and plots on the right (C and E) to those of the specific flavonoid pathway genes. Each column represents the fold induction of each specific promoter, plus one or more factors, relative to the respective control (with empty pART7). Mean values and error bars indicating the SE of at least three independent experiments are shown. Each transfection contained the Renilla luciferase plasmid pRluc for normalization (Horstmann et al., 2004). CHS, chalcone synthase; CHI, chalcone isomerase; FLS, flavonol synthase; ANR, anthocyanidin reductase; UFGT, UDP-Glc:flavonoid-3-O-glucosyltransferase.
genes, 18 coded for transcription factors including the two bZIP factors, VvibZIPB38 (VIT_14s0030g02200, r=0.68) and VvibZIPC14 (VIT_05s0077g01140, r=0.69), belonging to clade B and clade C (group C1), respectively. Taken together, the two approaches suggest VvibZIPB38 as a putative heterodimerization partner of VvibZIPC22, and VvibZIPC14 and VvibZIPC22 as mutual interacting partners with VvibZIP38.
In addition, the co-expression analysis provided further preliminary insights into the VvibZIPC22 gene regulatory network. The list of co-expressed genes was enriched in genes belonging to carbohydrate and amino acid metabolism, to ABA and ethylene signalling, and flower development. In particular, it is worth noting that among the genes with the highest correlation to the VvibZIPC22 profile, eight code for starch and sucrose metabolism enzymes (one coding for a trehalose phosphatase, one for a trehalose synthase, two for trehalose 6-phosphate synthases, one for an α-amylase precursor, one for a β-galactosidase, one for a glycosyl transferase, and one for a β-fructofuranosidase, r from 0.79-0.87), one for a ferulate-5-hydroxylase (r=0.80), and one for a leucoanthocyanidin dioxygenase (r=0.77), involved in the upper part of the phenylpropanoid pathway and in flavonoid biosynthesis, respectively, and one for an amine oxidase (r=0.85), an oxidoreductase that can also participate in phenylalanine metabolism ( Fig. 6; Supplementary Table S3).

Discussion
bZIP factors belongs to a large and diverse gene family identified in several plant genomes. Recently, the bZIP family has also been characterized in grapevine in terms of identification of members and phylogenesis (Liu et al., 2014). Plant bZIPs are involved in the regulation of many different plant processes, including the transcriptional control of amino acid (Hanson et al., 2008;Dietrich et al., 2011) and phenylpropanoid biosynthesis (Heinekamp et al., 2002;Hartmann et al., 2005).
During development, grape flowers and berries accumulate different flavonoids at the organ-and stage-specific level as a consequence of the timely expression of genes necessary for their synthesis. Although flavonoid biosynthesis has been investigated in depth also in grapevine, its regulation is not completely elucidated (Hichri et al., 2011a).
R2R3-MYB factors are known to be the key regulators of this transcriptional network; however, there are examples supporting the involvement of bZIP proteins, acting synergistically with MYB and other factors, in the control of light-induced flavonoid biosynthesis (Hartmann et al., 2005). The most studied is ELONGATED HYPOCOTYL5 (HY5), a bZIP associated with AtCHS and PFG1/MYB12 activation and accumulation of flavonoids upon light treatment (Stracke et al., 2010).
Interestingly, we have recently identified a gene encoding a bZIP factor, namely VvbZIP22, located on chromosome 7 within a QTL region specific for kaemperol content in mature berries , and therefore a good candidate as a regulator of the flavonol pathway. Starting from this evidence, in this study we further characterized the involvement of VvbZIP22, here renamed VvibZIP22C, in the fine regulation of flavonol biosynthesis.
By phylogenetic reconstruction, we showed that VvibZIPC22, together with VvibZIPC14 and VvibZIPC37, forms a group of the tree and is the potential orthologue of previously characterized bZIPs from flowering species such as N. tabacum (NtBZI-3 and NtBZI-4), P. crispum (PcCPRF6), A. thaliana (AtbZIP53), and S. lycopersicum (SlbZIP04, also known as LebZIP2). These genes are associated with flower development (Strathmann et al., 2001), are induced by salt and light treatments (Seong et al., 2008;Li et al., 2015), mediate the sugar-dependent level of specific amino acids during starvation (Dietrich et al., 2011), and are involved in the regulation of phenylpropanoid pathway genes by binding to specific cis-regulatory elements (Rugner et al., 2001;Heinekamp et al., 2002).
VvibZIPC22 expression was particularly high in the flower, the same organ in which kaempferol and quercetin accumulate the most, and its profile during Pinot Noir berry development resembled that of VviMYBF1 (Czemmel et al., 2009) and VviFLS1 (Downey et al., 2003), encoding the key regulators of flavonol biosynthesis in grapes. This result was a further indication that VvibZIPC22 is involved in the fine regulation of flavonol biosynthesis.
Flavonols, among flavonoids, are known to be modulated by UV light and function as UV protectants (Teixeira et al.,  2013). The light-induced flavonol biosynthesis is guided by the expression of VviMYBF1 and VviFLS1, genes containing light regulatory cis-acting elements in their promoters, recognized by MYB and bZIP factors (Czemmel et al., 2009). Based on this knowledge, we examined whether VvibZIPC22 is also involved in UV light-induced flavonol biosynthesis. Indeed, VvibZIPC22 appeared to be a light-inducible gene, being up-regulated within 10 h of UV light treatment of Chardonnay leaves, together with VviMYBF1 and VviFLS1. A similar result was also previously observed for VviMYBF1 and VviFLS1 in Chardonnay cell cultures (Czemmel et al., 2009). The induction of these genes was accompanied by flavonol accumulation starting at 24 h and peaking at 72 h after treatment. The analysis of the promoter region of VvibZIPC22 indicated that its light-induced expression might be due to the presence of MRS and ACS light regulatory cisacting elements, the same as found in the promoter sequence of VviMYBF1 and VviFLS1 (Czemmel et al., 2009) and conferring light responsiveness to AtCHS and AtFLS (Hartmann et al., 2005). Whether VvibZIPC22 light induction is mediated by the grapevine HY5 orthologue and by MYB factors, however, remains to be elucidated.
Overexpression of VvibZIPC22 in tobacco and transient promoter assays in Chardonnay cell cultures were then adopted as complementary approaches to investigate the role VvibZIPC22 further in vivo, demonstrate a causal link between VvibZIPC22 and flavonol biosynthesis, and gain insights into its putative regulatory network.
By these experiments, we provided evidence that VvibZIPC22 impacts the entire flavonoid pathway. Indeed, the tobacco overexpressing lines displayed a significantly higher content of the flavonols kaempferol and quercetin 3-rutinoside, the anthocyanin cyanidin 3-rutinoside, and the PAs in flowers, at the organ-and stage-specific level. However, due to the type of assay, we cannot completely exclude that the monitored increase in PAs corresponded to a change of the mean degree of polymerization (mDP), rather than a net increase in the absolute amount of PAs. These biochemical changes were accompanied by a significant induction of several flavonoid structural genes. Noteworthy is the ectopic expression of VvibZIPC22 that caused in most lines a visibly stronger red colouring of the stamens and corolla but not of the vegetative organs, as observed in tobacco flowers overexpressing VviMYB5a and VviMYB5b (Deluc et al., 2006(Deluc et al., , 2008. Conversely, the ectopic expression of VlMYBA1-2 triggered de novo production and storage of anthocyanins in all vegetative organs of the grapevine transgenic lines, attesting to a major effect of the gene on anthocyanin biosynthesis (Cutanda-Perez et al., 2009). As proposed in the model derived from the functional characterization of VviMYB5b (Hichri et al., 2011b), we can speculate that similarly VvibZIPC22 interacts with endogenous partners in the flower, where the anthocyanin pathway is already active (corolla), or induces de novo anthocyanin biosynthesis by interacting with other endogenous partners of the organs where the pathway is not yet active (stamens).
The transient promoter assays confirmed that VvibZIPC22 acts on different points of the flavonoid biosynthetic pathway, working either alone or in combination with other factors. VvibZIPC22 was capable of directly activating the flavonol-specific gene VviFLS1 and the flavan-3-ol-specific gene VviANR, as well as the genes of the general flavonoid pathway, VviCHS3 and VviCHI, by putatively binding to an ACGT-containing element identified in a conserved position in their promoters. Interestingly, VvibZIPC22 also increased the transactivation potential of known regulators of the general and specific flavonoid pathway genes, with the exception of the VviMYBA2-MYC1 complex which is specific for the regulation of the anthocyanin gene VviUFGT (Hichri et al., 2010). This result, contrasting with findings in tobacco, might be explained by a limit in the transactivation capacity of the VviUFGT promoter in this assay. Here, the availability of specific regulators is much higher than in reality and therefore there is no possibility of further activation by other factors such as VvibZIPC22.
Heterodimerization is a typical mechanism for regulating bZIP factor activity, and the formation of specific bZIP pairs is essential to exert their function (Llorca et al., 2014). In particular, in Arabidopsis , bZIPs from clade S (clade C in V. vinifera, to which VvibZIPC22 belongs) form specific heterodimers with bZIPs from clade C (clade B in V. vinifera). This heterodimeric complex is also conserved in tobacco (Strathmann et al., 2001) and parsley (Rugner et al., 2001). Accordingly, the putative heterodimerization partners of VvibZIPC22 were searched for within the bZIPs of clade B (Liu et al., 2014). From phylogenetic and co-expression results, we obtained an indication of VvibZIPB38 as a putative heterodimerization partner of VvibZIPC22, to be experimentally validated.
The co-expression analysis also gave further insights into the VvibZIPC22 putative regulatory network. Remarkably, 40% of the 210 genes strongly co-expressed with VvibZIPC22 fell within the QTL regions recently associated with the fine regulation of anthocyanin and flavonol content in mature berries . This percentage represents a significant enrichment over a random distribution of these genes along the genome, supporting their involvement in flavonoid metabolism. In addition, we noted that several genes with the highest correlation to the VvibZIPC22 profile coded for enzymes related to glucose metabolism. Interestingly, in Arabidopsis the expression of bZIP genes from class S1 was shown to be regulated by sugars and therefore to participate in a sugar-dependent control of target genes (Hanson et al., 2008).
Overall the results of our study point to a role for VvibZIPC22 in the regulation of different branches of the flavonoid pathway. In our hypothetical model, UV light and perhaps the glucose level promote the transcription of VvibZIPC22, which then regulates the expression of several genes of the flavonoid pathway, either alone or in combination with other factors (Fig. 7).

Conclusions
We have functionally characterized VvibZIPC22, a grapevine bZIP factor belonging to clade C in the phylogenetic reconstruction of the grapevine bZIP gene family. Our results indicated that, upon UV light stimulus, VvibZIPC22 regulates the content of different flavonoids, acting on promoters of general and specific genes of the pathway, alone or in combination with other factors. Altogether, our data provide new insights into the transcriptional control of the grapevine flavonoid pathway.

Supplementary data
Supplementary data are available at JXB online. Table S1. Primers used in qRT-PCR analyses of grapevine and tobacco samples. Table S2. Putative ACGT-containing elements in the promoters of the grapevine flavonoid pathway genes tested in transient reporter assays .  Table S3. Results of co-expression analysis obtained with VvibZIPC22 as input query. Figure S1. Profiles of VvibZIPC22 relative expression and of anthocyanin 3-monoglucosides and flavan-3-ol monomers during Pinot Noir berry development. Figure S2. Phylogenetic tree of clade B VvibZIP factors and of their putative orthologues.
Dataset S1. Codon-based alignments of CDS sequences of clade B VvibZIP genes and of their putative orthologues.
Dataset S2. Codon-based alignments of CDS sequences of clade C VvibZIP genes and of their putative orthologues.