VvMYB14 participates in melatonin-induced proanthocyanidin biosynthesis by upregulating expression of VvMYBPA1 and VvMYBPA2 in grape seeds

Abstract This work demonstrated that melatonin increases continuously in seeds, particularly seed coats, during berry ripening. Exogenous melatonin treatments significantly increased the proanthocyanidin (PA) content, partially through ethylene signaling, in seed coats. VvMYB14 expression exhibited patterns similar to melatonin accumulation over time, which was largely induced by melatonin treatment in seed coats during berry ripening. Additionally, VvMYB14 bound to the MBS element of the VvMYBPA1 promoter to activate expression. VvMYB14 overexpression largely upregulated expression of VvMYBPA1, VvMYBPA2 and VvLAR1 and increased the PA content in grape seed-derived calli. Similar increases in AtTT2 and AtBAN expression and PA content were found in VvMYB14-overexpressing Arabidopsis seeds. It was also observed that VvMYB14 overexpression increased ethylene production and thereby induced expression of VvERF104, which bound to the ERF element of the VvMYBPA2 promoter and activated its expression. Additionally, VvERF104 suppression reduced the VvMYB14 overexpression-induced increases in expression of VvMYBPA2 and VvLAR1 and PA content. Further experiments revealed that melatonin-induced increases in the expression of VvMYBPA1, VvMYBPA2, VvERF104 and VvLAR1 and PA accumulation were significantly reduced in VvMYB14-suppressing grape calli and leaves. Collectively, VvMYB14 mediates melatonin-induced PA biosynthesis by directly transactivating VvMYBPA1 expression and indirectly upregulating VvMYBPA2 expression via VvERF104.


Introduction
Grape is one of the most important fruit crops worldwide. Approximately half of all grapes are used to produce wine, and onethird are consumed as fresh fruit; the remainder is used for other grape products. Grapes are among the richest sources of polyphenols, including proanthocyanidins (PAs). PAs as the second most abundant polyphenolic compounds and are widely present in many plant tissues including bark, leaves, fruits and seeds [1,2]. Approximately 30% of total PAs are stored in grape seeds and 15% in grape skin [3]. PAs not only greatly affect the astringent or rough sensation of red wine but also may act as strong antioxidants to provide benefits to human health.
Melatonin is an indoleamine that is synthesized from L-tryptophan metabolism via serotonin; it is not only a strong antioxidant [8] but also a multifunctional signaling molecule in plants [9]. It has been reported that melatonin plays a key role in promoting fruit ripening and delaying postharvest senescence in grape, banana and tomato [10][11][12]. Additionally, melatonin has been reported to increase accumulation of sucrose and sorbitol in pear fruits [13]. Postharvest treatment with melatonin has demonstrated its key role in increasing or maintaining the content of total anthocyanins, total f lavonoids, and total phenols in strawberry fruits [14]. In particular, our previous study elucidated that melatonin alters the profile of 27 secondary metabolites in grape berry skin [15]. Therefore, melatonin accelerates fruit ripening and affects metabolite accumulation; however, the pathways involved in melatonin sensing and signaling remain largely unknown. Several studies have shown that melatonin modulates gene expression related to metabolism of ABA, indole-3-acetic acid, cytokinins, gibberellins and ethylene [16], and our previous study revealed that melatonin functions partially through ethylene signaling in regulation of berry ripening [17].
Additionally, our previous studies revealed that VvMYB14 might participate in melatonin signaling during secondary metabolite metabolism regulation in Merlot berry skin; the role of VvMYB14 in Figure 1. Expression of VvMYB14 and accumulation of melatonin and PAs during berry ripening. Changes in TSS, titratable acid, and total anthocyanin contents were used to indicate the onset of ripening (A). In Panels B-H, the red and blue lines represent skin and seeds, respectively. Panels E-H share the same coordinate axis titles. The values represent the means ± SD of three replicates. * , Significant difference, P < 0.05; * * , highly significant difference, P < 0.01. ns, not significant at P < 0.05. regulating accumulation of secondary metabolites was elucidated by its overexpression in Merlot grape berry skin-derived calli [15]. It was also reported that MtMYB14 and MtMYB5 physically interact and synergistically activate expression of ANR and LAR and that MtMYB14 overexpression strongly induces PA accumulation in Medicago truncatula [18]. Therefore, a pathway by which VvMYB14 regulates PA accumulation in response to melatonin might exist. Nonetheless, such a pathway might not be a primary pathway involved in PA regulation in berry skin because the melatonin content declines sharply during ripening and is undetectable in the skin at late ripening stages [15]. Although the melatonin accumulation pattern in grape seeds remains unknown, because PAs are stored predominantly in grape seeds [3], we hypothesized that melatonin promotes PA biosynthesis primarily via a VvMYB14-mediated pathway in grape seeds.
To test the above hypothesis, accumulation patterns of melatonin, PA and VvMYB14 transcripts in seeds during berry ripening were determined. The function of melatonin in promoting PA biosynthesis was demonstrated in grape seed coats. The role of VvMYB14 in regulating PA biosynthesis by modulating expression of VvMYBPA1 and VvMYBPA2 or their homologous genes in Arabidopsis was revealed through its overexpression in Merlot grape berry seed-derived calli and Arabidopsis seeds. Moreover, the role of VvMYB14 in mediating melatonin-induced PA biosynthesis was determined using VvMYB14-suppressed grape calli and grape leaves. This research provides insight into the molecular mechanism underlying melatonin signaling in regulation of PA biosynthesis in grape seeds.

Changes in expression of VvMYB14 and contents of melatonin and PAs during berry ripening
Total soluble solids (TSS), titratable acid, and total anthocyanin contents were determined to evaluate the berry ripening process. TSS and anthocyanins began to accumulate, and titratable acid began to decrease at 70 days after full bloom (DAB; Fig. 1A), indicating the beginning of berry ripening. Expression of VvMYB14 continued to increase from 70 DAB and reached a maximum in the seeds of ripened berries; in contrast, expression of this gene decreased in berry skin after 80 DAB (Fig. 1B). The melatonin content showed very similar patterns to those of VvMYB14 in seeds and was undetectable in the skin during late ripening stages (Fig. 1C). Generally, PA contents in seeds and skin were high before the onset of ripening (Fig. D-H). Compared to values at 80 DAB, when melatonin began to accumulate, the contents of total PAs and four PA compounds (catechin, epicatechin, procyanidin B2 and procyanidin B3) were significantly increased at 90 and 110 DAB in seeds, suggesting the possible positive contribution of melatonin in inducing PA accumulation. However, the low level of PAs at 100 DAB suggested the existence of other negative regulators of PA accumulation. In contrast, the PA content generally showed a continuous decreasing trend in the skin (Fig. D-H). . The scale bar is one millimeter in Panels A and B. The total PA content in different parts of the seeds is shown in Panel C. Melatonin content and VvMYB14 expression were determined using the seed coats from the berries at different DAB (D, E). The berries were treated at 60 DAB, corresponding to 0 DAT (D-J). Panels F-J share the same y-axis titles. The values represent the means ± SD of three replicates. * , significant difference, P < 0.05; * * , highly significant difference, P < 0.01. ns, not significant at P < 0.05.

Exogenous melatonin treatment increases VvMYB14 expression and PA accumulation in seed coats
Histochemical staining and content determination showed that PAs primarily accumulated in seed coats and were also detected in the endosperm and embryo at 70 and 110 DAB ( Fig. 2A-C). Therefore, seed coats were used to further evaluate the effects of melatonin treatment on VvMYB14 expression and PA accumulation. Melatonin treatment significantly increased the content of melatonin and the expression level of VvMYB14 in seed coats on different days after treatment (DAT) (Fig. 2D, E). Additionally, melatonin treatment significantly increased the content of total PAs and four PA compounds at the three time points detected, particularly at 50 DAT ( Fig. 2F-J). Overall, increases in procyanidin B2 and procyanidin B3 reached 4.48-and 7.01-fold, respectively, in melatonin-treated seed coats compared with controls at 50 DAT (Fig. 2I, J). Moreover, expression of VvMYBPA1, VvMYBPA2, VvLAR1 and VvLAR2 was upregulated to varying extents at 10, 30 and 50 DAT (Supplementary Data Fig. S1). Melatonin treatment significantly increased the melatonin content and VvMYB14 expression but significantly reduced the content of PAs, accompanied by reduced expression of VvMYBPA1, VvMYBPA2, VvANR, VvLAR1 and VvLAR2, in berry skin at 10 and/or 30 DAT (Supplementary Data Fig. S2). Therefore, melatonin increased both VvMYB14 expression and PA accumulation in seed coats but not in berry skin.

Exogenous melatonin treatment increases PA accumulation partially via ethylene production in seed coats
Our previous study found that exogenous melatonin treatment increased ethylene production in grape berries [17]. In the present, we determined whether melatonin regulates PA synthesis via ethylene production in seed coats. Melatonin treatment significantly enhanced the ACC (precursor of ethylene synthesis) content and ethylene production rate at 10-30 DAT in seed coats, and the largest increases were observed at 20 DAT (Fig. 3A, B). Compared to the control, ethephon treatment significantly increased the contents of catechin, epicatechin, procyanidin B2 and procyanidin B3 at 10, 30 and 50 DAT; in contrast, application of the ethylene receptor inhibitor 1-MCP led to the opposite results. Therefore, ethylene plays a key role in regulating PA synthesis Figure 3. The levels of ethylene production and PA accumulation in the control seed coats and those treated with ethephon, 1-MCP, melatonin, and melatonin plus 1-MCP. The ethylene production level was evaluated by the determination of the ACC (precursor of ethylene) content (A) and ethylene production rate (B). The contents of catechin (C), epicatechin (D), procyanidin B2 (E) and procyanidin B3 (F) were determined in seed coats at 10, 30 and 50 DAT. The berries were treated at 60 DAB, corresponding to 0 DAT. The values represent the mean ± SD of three replicates. * , significant difference, P < 0.05; * * , highly significant difference, P < 0.01. The values indicated by the different lowercase letters are significant at P < 0.05.
( Fig. 3C-F). Melatonin also significantly increased the content of the four compounds, whereas the application of 1-MCP significantly inhibited the melatonin-induced increases in the content of the four compounds ( Fig. 3c-f). Therefore, melatonin regulates accumulation of PAs partially through ethylene production.

VvMYB14 and VvERF104 activate transcription of VvMYBPA1 and VvMYBPA2, respectively, two key transcription factors involved in regulating PA synthesis
VvMYBPA1 and VvMYBPA2 have been identified as regulators of PA synthesis [6,7]. Hence, the 1000 bp fragment upstream of the start codon of VvMYBPA1 and VvMYBPA2 was used to screen transcription factors via a yeast one-hybrid system (Y1H), and VvMYB14 and VvERF104 were identified as possible regulatory factors of VvMYBPA1 and VvMYBPA2, respectively.
Two possible MYB binding sites (MBS and MRE) are predicted in the VvMYBPA1 promoter (Supplementary Data Fig. S3). In contrast, MBS and MRE were not found in the promoter of VvMYBPA2 (Supplementary Data Fig. S3). Y1H and electrophoretic mobility shift assays (EMSAs) indicated binding of the VvMYB14 protein to the sequence containing the MBS element within the VvMYBPA1 promoter (Fig. 4A, B; Fig. 4A, C). In contrast, VvMYB14 did not bind to the fragment containing a mutated MBS element (Fig. 4A, C). Therefore, the VvMYB14 protein specifically binds to the MBS element of the VvMYBPA1 promoter. On the other hand, Y1H and EMSA results showed very weak binding between VvMYB14 and MRE (Supplementary Data Fig. S4). Collectively, MBS in VvMYBPA1 is the primary binding site for VvMYB14.
Additionally, Agrobacterium-mediated transient expression of the LUC and GUS reporter genes in tobacco leaves and grape calli was performed. Leaves cotransformed with MBS-35 s mini-LUC and 35S-MYB14 showed markedly increased luminescence intensity compared with those transformed with mMBS-35 s mini-LUC and 35S-MYB14 (Fig. 4D). Additionally, grape callus cotransformed with P MYBPA1 ::MYBPA1-GUS (VvMYBPA1-GUS fusion gene driven by the VvMYBPA1 promoter) and 35S::MYB14 showed a darker blue color and higher GUS activity and VvMYBPA1 expression than callus transformed with P MYBPA1 ::MYBPA1-GUS alone (Fig. 4E). Therefore, the VvMYB14 protein acts upstream of VvMYBPA1 to activate its expression.
Previous studies have shown that the bHLH protein TT8 and WD40 protein TTG1 interact with various R2R3-MYBs and form ternary protein complexes named MBW [19]. However, induction of MYBPA1 promoter activity by VvMYB14 was not dependent on the presence of TT8 and TTG1 (Fig. 4F, G).
Nevertheless, VvERF104 was demonstrated to bind to the ERF element in the VvMYBPA2 promoter and activate its expression using Y1H, EMSA, and transient expression assays in tobacco leaves and grape calli ( Fig. 5A-E).

VvMYB14 regulates PA accumulation by modifying expression of VvMYBPA1, VvMYBPA2 and AtTT2 in grape calli and Arabidopsis seeds
To identify the function of VvMYB14, transgenic grape seedderived calli with different levels of VvMYB14 expression were obtained, including three overexpression lines (M14OE1-3) and two suppression lines (M14SE1-2) (Fig. 6A-C). Overexpression of VvMYB14 significantly upregulated expression of VvERF104, VvMYBPA2 and particularly VvMYBPA1, whereas VvMYB14 suppression decreased expression of the above three genes (Fig. 6D-F). Additionally, expression of VvLAR1, the possible target gene of VvMYBPA1 and VvMYBPA2 [6,7], was largely increased by VvMYB14 overexpression. In contrast, two other possible target genes, VvLAR2 and VvANR [6,7], showed a small increase in gene expression compared to WT (Fig. 6G-I). Except for procyanidin B3, the remaining PAs were detected in grape calli. Contents of catechin and procyanidin B2 were largely increased in VvMYB14-overexpressing calli but significantly reduced in VvMYB14suppressed lines. In contrast, epicatechin was significantly changed by VvMYB14 only in M14OE2, M14OE3 and M14SE1 ( Fig. 6J-L).
Additionally, three lines of VvMYB14-overexpressing Arabidopsis plants (AtM14OE1, AtM14OE2 and AtM14OE3) were obtained (Fig. 6M, N) to evaluate their function of this gene in regulating PA biosynthesis in seeds. Expression levels of AtTT2 (the homologous gene of VvMYBPA1 and VvMYBPA2) [21] and AtBAN (the homologous gene of VvANR) [22] were largely upregulated, and the contents of procyanidin B3 and epicatechin were significantly increased in seeds from immature siliques of the three lines (Fig. 6N, O).
Moreover, M14OE2 calli were used to obtain grape calli with simultaneous VvMYB14 overexpression and VvERF104 suppression (M14OE2 + E104SE1 and M14OE2 + E104SE2). The VvMYB14 overexpression-induced increases in expression of VvMYBPA2 and VvLAR1 and the contents of catechin and procyanidin B2 were significantly reduced by suppression of VvERF104 (Fig. 6 P, Q). Therefore, VvMYB14 promotes VvMYBPA2 and VvLAR1 expression and thereby PA accumulation via VvERF104. Additionally, absence of MBS and MRE excludes the possibility of the interaction between VvMYB14 and the VvERF104 promoter (Supplementary Data Fig. S5), andVvMYB14 overexpression increased ethylene production in grape calli (Fig. 6R). These results suggest that VvMYB14 increases VvERF104 expression by promoting ethylene production. This inference was verified by determination of VvERF104 expression in VvMYB14-overexpressing calli treated with 1-MCP (Fig. 6S).
Taken together, these results suggest that VvMYB14 promotes PA synthesis by directly transactivating VvMYBPA1 expression or indirectly increasing VvMYBPA2 expression through enhanced ethylene production. In addition, it was observed that alterations in VvMYB14 expression affected callus growth (Fig. 6A,   Supplementary Data Fig. S6), suggesting the role of VvMYB14 in regulating growth.

VvMYB14 mediates melatonin-induced PA biosynthesis
To identify the role of VvMYB14 in mediating melatonin signaling, VvMYB14-suppressed grape calli were treated with melatonin. In WT grape calli, melatonin treatment largely induced expression of VvMYB14, VvMYBPA1, VvERF104, VvMYBPA2 and VvLAR1, whereas melatonin increased expression of VvLAR2 and VvANR to a small extent. These melatonin-induced increases in expression of VvMYB14, VvMYBPA1, VvERF104, VvMYBPA2 and VvLAR1 were largely reduced in VvMYB14-suppressed calli (Fig. 7A-H). Similarly, the contents of catechin, epicatechin and procyanidin B2 were significantly increased in response to melatonin; in contrast, VvMYB14 suppression reduced their increase under melatonin treatment (Fig. 7J-L). Additionally, two VvMYB14-suppressed  The data in Panels A to L were from WT and VvMYB14-suppressing grape calli (M14SE). The phenotypes of the WT and VvMYB14-suppressing grapevines (vM14SE1) are shown in Panel M. PCR identification of VvMYB14-suppressing grapevines using the specific primer pair for the 35S promoter and the VvMYB14 gene is shown in Panel N, in which 1, 2, 3 and 4 represent the empty vector, WT, vM14SE1 and vM14SE1, respectively. The data in Panels O-T are for WT and VvMYB14-suppressing grapevines. Mel represents melatonin. * , Significant difference, P < 0.05; * * , highly significant difference, P < 0.01. ns, not significant at P < 0.05. grapevines (vM14SE1 and vM14SE2) were obtained to further evaluate the role of VvMYB14 in mediating melatonin-induced PA accumulation (Fig. 7M-O). In grape leaves, VvMYB14 suppression significantly reduced the melatonin-induced increases in expression of VvMYB14, VvMYBPA1, VvERF104, VvMYBPA2 and VvLAR1 and in the contents of catechin, epicatechin, procyanidin B2 and procyanidin B3 (Fig. 7P-T).
Collectively, melatonin promotes expression of VvMYBPA1, VvERF104, VvMYBPA2 and VvLAR1 via VvMYB14 and therefore increases PA accumulation. In addition, VvMYB14-suppressed vines grew quickly and exhibited more white roots than WT ( Fig. 7M; Supplementary Data Fig. S7), suggesting a role for VvMYB14 in regulating growth.

Possible roles of melatonin and VvMYB14 in the PA biosynthesis pathway
The role of melatonin in increasing PA accumulation was indicated by exogenous treatment in seed coats (Fig. 2F-J). In contrast, a high PA content was detected in seeds at 60 DAB, a time when melatonin was undetectable (Fig. 1C, E-H). Although there was a continuous increase in melatonin content and VvMYB14 expression (Fig. 1B, C), a relatively low content of PAs was detected in seeds at 100 DAB (Fig. 1D-H), suggesting the existence of other negative regulators of PA biosynthesis. Therefore, it is suggested that melatonin acts as a modulator of PA biosynthesis rather than a trigger. Additionally, the contrasting effects of exogenous melatonin treatment on PA accumulation and expression of genes related to PA biosynthesis in seeds and skin (Fig. 2F-J, Supplementary Data Figs. S1 and S2) indicate that the effects of melatonin are complex and may largely depend on the plant tissue. This inference is also supported by other studies showing that melatonin treatment had opposite effects on postharvest ripening of tomato and banana [10,12].
Overexpression of VvMYB14 increased accumulation of PA compounds in seed-derived grape calli (Fig. 6J-L) and Arabidopsis seeds (Fig. 6O). Similar results were found for MtMYB14overexpressing hairy roots in M. truncatula [18]. A phylogenetic tree shows that VvMYB14, MtMYB14 and AtMYB14 belong to the same subgroup (Supplementary Data Fig. S8). Therefore, MYB14 might regulate PA biosynthesis in different species.
Notably, VvMYB14 overexpression led to different effects on PA compound accumulation in grape calli and Arabidopsis seeds ( Fig. 6J-L, O), which is probably related to the absence of the LAR gene in Arabidopsis genome, at least partially [21]. Therefore, the role of MYB14 in regulating PA biosynthesis is affected by its downstream genes. Additionally, VvMYB14 was strongly induced by melatonin in seed coats (Fig. 2E) and skin [15]. Suppression of VvMYB14 reduced the effects of melatonin on PA accumulation in grape calli (Fig. 7J-L) and leaves (Fig. 7 P-S). Thus, VvMYB14 mediates melatonin-induced PA biosynthesis. Notably, VvMYB14 overexpression largely increased the catechin content in seedderived grape calli (Fig. 6J); however, the opposite results were found in skin-derived calli [15]. These contrasting results suggest that additional seed-derived factors may be needed to increase PA biosynthesis in grape calli under VvMYB14 overexpression. The cell-specific function of VvMYB14 might be associated with the tissue-specific regulation of melatonin, as mentioned above.
VvMYBPA1 and VvLAR1 might be primary regulatory and structural genes, respectively, in the VvMYB14-mediated PA biosynthesis pathway VvMYBPA1 and VvMYBPA2 are two key transcription factors responsible for regulating PA biosynthesis through modification of VvLAR and VvANR expression [6,7]. A phylogenetic tree showed that VvMYBPA1 and AtMYB123 (TT2) cluster into the same subgroup ( Supplementary Data Fig. S8). In particular, VvMYBPA1 complements the PA-deficient seed phenotype of the Arabidopsis tt2 (Atmyb123) mutant, which shows almost complete loss of PA accumulation in the seed coat [6]. Therefore, VvMYBPA1 is a key transcription factor regulating PA biosynthesis. Additionally, VvMYB14 directly bound to the promoter of VvMYBPA1 and activated its expression (Fig. 3), and an extreme increase in expression of VvMYBPA1 and AtTT2 was detected in VvMYB14-overexpressing calli and Arabidopsis seeds, respectively (Fig. 6D, N). Accordingly, it is strongly suggested that VvMYBPA1 might be the key transcription factor in the VvMYB14-induced PA biosynthesis pathway. VvMYB14 also increased VvERF104 expression via ethylene signaling and thereby upregulated VvMYBPA2 expression (Fig. 6R, S). It has been reported that overexpression of VvMYBPA2 results in accumulation of VvMYBPA1 transcripts in grape hairy roots [7]. Collectively, VvMYB14 increases VvMYBPA1 expression via direct transactivation and indirectly via VvMYBPA2. Notably, VvMYB14 transactivated VvMYBPA1 in the absence of bHLH and WD40 (Fig. 4F, G). Similarly, it has been reported that VvMYB14, VvMYB15, and VvMYBF1 induce the promoter activity of target genes in a bHLH cofactor-independent manner [16]. Moreover, the protein sequence of VvMYB14 lacks the [D/E]Lx2[R/K]x3Lx6Lx3R motif (Supplementary Data Fig. S9), which is necessary for interaction with bHLHs [16,23]. These MYBs that lack the bHLH interaction amino acid motif may not require bHLH TFs for their activity.
LAR is predominantly responsible for producing (+)-catechin, and ANR is predominantly responsible for producing (−)epicatechin [5,24,25]. VvMYB14 overexpression led to much higher expression of VvLAR1 than that of VvLAR2 and VvANR (Fig. 6G-I), which corresponded to the larger increase in catechin than epicatechin (Fig. 6J, K). Additionally, overexpression of VvMYBPA1 or VvMYBPA2 did not result in significant induction of VvLAR2 [7]. Hence, it is suggested that LAR1 is the primary structural gene for PA biosynthesis in the VvMYB14-mediated pathway. It has also been reported that MtMYB14 strongly induces PA accumulation by activating the promoters of ANR and LAR in M. truncatula hairy roots [16]. Therefore, MYB14 might regulate PA biosynthesis by upregulating MYBPA1 and MYBPA2, thereby increasing expression of ANR and LAR, or by directly activating ANR and LAR.

VvMYB14 serves as a bridge between melatonin and ethylene, affecting PA biosynthesis via ethylene.
Ethylene is an important signaling molecule that plays a key role in regulating fruit maturation and senescence. Ethylene is the dominant regulator of anthocyanin and some other phenolic compound biosynthesis in grape berry skin [15,26]. However, the specific role of ethylene in modifying PA accumulation is poorly understood. Here, we elucidate the positive inf luence of ethylene on PA biosynthesis via ethephon and 1-MCP treatment in seed coats (Fig. 3). Melatonin treatment significantly increases ethylene production levels in different tissues of grapes, including Merlot grape seeds (Fig. 3A, B) and skin [15], whole Moldova berries [11], and the roots and leaves of Crimson seedless grapes [27]. Similar results have been found in tomato fruits [10]. Therefore, it is inferred that melatonin promotes PA biosynthesis through ethylene, at least partially. This inference was verified by treating seed coats with melatonin plus 1-MCP (Fig. 3C-F). However, the ethylene production rate and melatonin content showed different patterns in seeds during berry ripening ( Fig. 2D and Fig. 3B), suggesting that melatonin is not the sole signaling molecule that regulates ethylene production. For example, reciprocity between ABA and ethylene exists in grape fruits [28].
VvMYB14 is strongly induced by melatonin (Fig. 2D), and a 580-bp core region responding to melatonin has been identified [15]. VvMYB14 promotes ethylene production by transactivating VvACS1 expression by binding to its promoter [27]. Therefore, VvMYB14 mediates melatonin-induced ethylene production. In ethylene signaling, ERFs are downstream transcription factors that activate expression of ethylene-responsive genes and are involved in the regulation of numerous developmental processes and stress tolerance. In particular, MdERF1B is reported to regulate anthocyanin and PA biosynthesis in apple [29]. In the present study, VvERF104 was found to function downstream of VvMYB14 to increase PA biosynthesis by transactivating expression of VvMYBPA2 (Fig. 5). ERF104 may be involved in pathogen responses [30]. In fact, PAs, as astringent compounds, are considered to be involved in defense against herbivores and pathogens in leaves and unripe fruits [31]. These results suggest that ERF104 might regulate stress tolerance by increasing PA biosynthesis.
Collectively, melatonin increases ethylene production by inducing VvMYB14 expression, and the increase in ethylene promotes PA biosynthesis through VvERF104-activated VvMYPA2 expression.

MYB14 might possess multiple functions, including growth regulation
In this study, we also found that VvMYB14 overexpression inhibited callus growth and that its suppression promoted the growth of calli and vines and caused vines to produce more white roots ( Fig. 6A; Fig. 7M; Supplementary Data Figs. S6 and S7). It has also been reported that MYB14 is related to biosynthesis of stilbene in grapevine [16] and starch in the maize endosperm [32]. Some direct target genes or downstream genes regulated by MYB14 have been identified, including STSs in grape [16] and six starchsynthesizing genes and 12 genes involved in the general phenylpropanoid pathway in lotus [33]. Therefore, MYB14 might play a broad role in regulating metabolism and growth.

Figure 8. Model of the regulation of PAs by melatonin via the
VvMYB14-mediated pathway. The red lines and arrows indicate the pathway verified in grapevine in the present work; conversion of catechin/epicatechin into procyanidin B2/procyanidin B3 refers to previous studies [6,18]. The blue lines and arrows indicate pathways that have been previously reported in grapevines: 1 Xu et al., 2019 [24]; 2 Bogs et al., 2007 [6]; 3 Terrier et al., 2009 [7]; and 4 Liu et al., 2014 [18]. Cross marks indicate that VvMYB14 functions independently of VvTTG1 and VvTT8.
Regarding the growth inhibition caused by VvMYB14 overexpression, we inferred that it might be related to high-level PAs that excessively accumulated and produced toxic effects in VvMYB14-overexpressing cells. The toxic effect of PAs is also shown by results that large accumulation of PAs in the roots and leaves of transgenic grapevine or Arabidopsis leads to growth abnormalities and even death of the plants [6,34]. Alternatively, VvMYB14 overexpression increases the level of ethylene, which is a pleiotropic molecule with diverse functions in plants, including growth inhibition [35]. Additionally, ERF104 overexpression largely increases expression of four auxin-responsive proteins and one auxin eff lux carrier in Arabidopsis [30], suggesting that MYB14 affects auxin signaling and therefore regulates growth.
Taken together, these results indicate that VvMYB14 expression and melatonin accumulation continuously increase in seeds during berry ripening. VvMYB14 binds to the promoter of VvMYBPA1 and activates its expression. VvMYB14 also increases VvERF104 expression by promoting ethylene production and thereby enhances VvMYBPA2 expression. VvMYB14 overexpression increases PA accumulation in grape seed-derived calli and Arabidopsis seeds. Exogenous melatonin treatments largely increase VvMYB14 expression and PA accumulation in seed coats. Melatonin-induced increases in PA compounds were significantly reduced by VvMYB14 suppression in grape calli and leaves. Therefore, it is most likely that melatonin promotes PA biosynthesis by upregulating expression of VvMYB14 and thereby VvMYBPA1 and VvMYBPA2 in grape seeds (Fig. 8).

Plant materials and growth conditions
Skin and seeds from Merlot grape (Vitis vinifera) berries at different developmental stages were collected to determine gene expression and melatonin and PA contents. Preveraison Merlot grape berries at 60 days after bloom (DAB) were treated with 50 μM melatonin, which was verified to effectively increase the melatonin level in berries and affect berry ripening and secondary metabolite accumulation [15,17], 200 mg. l −1 ethephon, 10 μl. l −1 1-MCP, or 50 μM melatonin plus 10 μl. l −1 1-MCP. Clusters were soaked in the above solutions with 0.05% Triton X-100 and in 0.05% Triton X-100 alone (control). Each treatment consisted of three replications, and each replication consisted of 3 vines (approximately 24 clusters). Berries were randomly sampled from each cluster for subsequent experiments.

Determination of total anthocyanins, total soluble solids (TSS), and titratable acid contents
Total anthocyanins of berry skin were extracted and determined using the spectrophotometric pH differential method as described in a previous study [17]. Fresh pulp was ground and filtered, and the filtrate was used to determine TSS and titratable acid. The TSS content was determined using a digital display sugar meter, and titratable acid was measured by titration of the filtrate with 0.1 M NaOH to pH 8.3 [15].

Melatonin extraction and determination using UPLC-MS
Extraction and determination of melatonin were performed according to our previous study [17] with minor modifications. Brief ly, two grams of ground tissue was preliminarily extracted two times using methanol by sonicating for 20 min each time. The combined supernatants were centrifuged at 8000 rpm for 10 min and then evaporated to dryness at 30 • C. The residue was dissolved in methanol and then purified using a ProElut™ C 18 solid-phase extraction (SPE) cartridge (Dikma, China). The purified samples were separated and detected by UHPLC-MS (Waters, Milford, MA, USA) equipped with a BEH C 18 column and a QTof-micro mass spectrometer. The same UHPLC and MS conditions as in our previous study were applied in this study. Melatonin was quantified using the external calibration curve of a melatonin standard.

PA histochemistry, extraction and determination
For detection of PAs, seed cross sections were stained with DMACA as described by Feucht and Polster [36]. PA was extracted according to our previous study [37], and the process was similar to the melatonin extraction but with the following changes: methanol was replaced by acidified methanol (0.1% HCl, v/v). Samples without SPE purification were directly used for total PA determination. Purified samples were used for determination of PA compounds. The total PA content was measured using a vanillin assay [38]. Brief ly, 200 μl of extract was added to a reaction solution with 500 μl of 1% (w/v) vanillin in methanol and 500 μl of 25% (v/v) H 2 SO 4 in methanol; after 15 minutes at 30 • C, the A500 value was measured by a spectrophotometer. PA compounds were detected using a Dionex Ultimate 3000 UHPLC system and a ESI-triple quadrupole mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA). The UHPLC system was equipped with a reversedphase C 18 analytical column (100 mm × 2.1 mm, 1.9 μm). The detailed parameters and conditions of HPLC and MS were fully described in our previous study [37]. The amount of PA compounds was quantified using the external calibration curves of corresponding standards.

Determination of ACC content and ethylene production rate
Extraction and determination of ACC were performed based on a previously described method [39]. Ethylene production rate was determined using a Shimadzu GC-9A gas chromatograph (Kyoto, Japan) equipped with a f lame ionization detector and a GDX-502 column as described in a previous study [17].

Yeast one-hybrid (Y1H) assays and electrophoretic mobility shift assays (EMSAs)
For Y1H assays, sequences including the MBS or ERE elements from the promoter of VvMYBPA1 or VvMYBPA2 were cloned into the pHis2 vector. Mutant MBS or ERE elements were used as a negative control. The ORF of VvMYB14 or VvERF104 was cloned into the pGADT7 vector. The resulting plasmid was integrated into yeast strain Y1HGold. Y1H assay was conducted using a Matchmaker™ Gold Yeast One-Hybrid Library Screening kit from Clontech (Mountain View, CA, USA).
For EMSA, the recombinant VvMYB14-or VvERF104-His protein was expressed using a pET-32a vector and the soluble proteins were purified via a BeaverBeads IDA-Nickel kit (Beaver, BioBay, China). DNA probes containing an MBS or ERE element were synthesized and labeled with biotin. EMSAs were conducted according to the user manual of the LightShift Chemiluminescent EMSA kit (Thermo Fisher Scientific, Waltham, MA, USA). The primers used in Y1H assays and EMSA are listed in Supplementary Data Table S1.

Transient cotransformation in tobacco leaves and grape calli
The ORFs of VvMYB14 and VvERF104 were inserted into the pGreenII 62-SK vector, and the promoter fragments of VvMYBPA1 and VvMYBPA2 containing MBS and ERE elements, respectively, were cloned into the pGreenII 0800-LUC vector. The plasmids were transiently transformed into tobacco leaves using a previously described method [40].
The VvMYB14 ORF was inserted into the pRI101-AN vector to produce the construct 35S::MYB14. The 35S promoter within pRI101-GUS was replaced with the promoter of VvMYBPA1 or VvMYBPA2, 1000 bp upstream of the start codon, and the VvMYBPA1 or VvMYBPA2 ORF was inserted upstream of GUS, yielding the construct P MYBPA1/MYBPA2 ::MYBPA1/MYBPA2-GUS. The plasmids were transformed into nonembryogenic grape calli according to the method of Xu et al. [27] Histochemical staining and activity determination of GUS were conducted according to a previously reported method [41].

Transformation of VvMYB14 or VvERF104 into merlot seed-derived calli and grapevines and Arabidopsis plants
The construct 35S::MYB14 involving the pRI101-AN vector mentioned above was used for sense overexpression. The 3'-UTR sequence of VvMYB14 was inserted into the pRI101-AN vector, and the obtained construct was used for antisense suppression. Additionally, the 3'-UTR sequence of VvERF104 was inserted into the pHB vector with hygromycin resistance for cotransformation with VvMYB14. Transgenic grape calli were obtained using an Agrobacterium-mediated transformation method [42].
Additionally, the pRI101-AN vector containing the 3'-UTR sequence of VvMYB14 was transferred into grapevine embryogenic calli using the Agrobacterium (GV3101)-mediated method [20]. Brief ly, the calli were soaked in bacterial solution for 20 min with gentle shaking. Cocultivation in solid MS medium containing 15 mg. l −1 acetosyringone and 2% sucrose was performed for 3 days at 28 • C in the dark. The calli were screened on KBN delayed-screening medium containing 250 mg. l −1 cefotaxime for 4 weeks and then on X3 delayed-screening medium containing 250 mg. l −1 cefotaxime for 1 week. The transformed calli were cultured in resistance screening medium containing 250 mg. l −1 cefotaxime and 75 mg. l −1 kanamycin until germinated embryos developed. The embryos were then cultured in germination medium under light until true leaves appeared, and the plantlets were transferred to rooting medium containing 1 mg. l −1 IBA.
For Arabidopsis transformation, the pRI101-AN vector containing 35S::MYB14 was transferred into Columbia-O by the Agrobacterium-mediated f loral dip method [43]. All of the above transgenic calli or plants were verified by PCR identification using the specific primer pair for the 35S promoter and the VvMYB14 gene and determination of the VvMYB14 expression level.

Real-time quantitative PCR
Real-time quantitative PCR was performed using SYBR Green Master Mix (SYBR Premix EX TaqTM, Dalian, China) on an ABI7500 qRT-PCR instrument (ABI, MA, USA), and the primers used are listed in Supplementary Data Table S1.