The Biases in Flavonoids and Anthocyanin Biosynthesis of a Red Flesh Table Grape Revealed by Metabolome and Transcriptome Co-analysis

Red esh is a welcomed fruit trait, yet the regulation of red esh formation in grape is not well understood. ‘Mio Red’ is a seedless table grape variety with light red esh and blue-purple skin, the esh color developed in the late stage of berry ripening, remarkably later than the skin coloring at veraison. The esh and skin avonoids metabolome and the transcriptome were analyzed. A total of 173 avonoids including 17 anthocyanins were identied, 68 were found signicantly different (Fold change ≥ 2 or ≤ 0. 5, VIP ≥ 1). Quercetin 3-O-glucoside, epicatechin-epiafzelechin, apigenin 6,8-C-diglucoside and hesperetin 5-O-glucoside were of higher content in the esh, while the rest avonoids were of higher content in the skin. The main anthocyanin in the esh was pelargonidin derivatives in contrast to peonidin derivatives in the skin. Transcriptome comparison recruited 3970 differentially expressed genes (DEGs, log2Fold change > = 1, FDR < 0.05, FPKM ≥ 1), among them 57 were structural genes of avonoid metabolism pathway. Two anthocyanin synthase (ANS) DEGs were annotated, ANS1 (Vitvi11g00565) and ANS2 (Vitvi02g00435) led the expression in the esh and skin respectively. In the esh, anthocyanin biosynthesis structural gene UFGT, positive regulators MYBA1/2/3, and anthocyanin transporters GST14 and MATE5 were of signicantly lower expression, while negative regulators MYBC2-L1 and MYB3 were of higher transcription. The results of this study provide new information in the coloring mechanism of red esh grape and assisting breeding of future table grapes having higher content of phytonutrient providing the health benet as red wines. and skin of a new red esh table grape variety.


Introduction
Grape is one of the most important fruits in the world, has been recorded several thousands of varieties.
In 2019, the planting area was 69.3 hectares (ha), and the production reached 77.14 million tons (http://www.fao.org/faostat/zh/ # data/QC), China is the largest table grape producer and consumer. Grape berry development follows a typical double sigmoid growth curve. Veraison, the start of stage III, is marked by berry softening and skin color change (Coombe and Bishop, 1980). According to the skin color, grape varieties can be divided into two major categories, red grapes which are of purple, red or pink appearance, and white grapes which are of light green to golden color at ripening.
Flavonoids are a kind of secondary metabolites including avonoids, avanones, proanthocyanidins and anthocyanidins, they are reactive oxygen species scavengers in plants, and have high value in human health (Agati et al., 2012). Anthocyanin is the most important avonoids in grape berries determining berry color. Six anthocyanins are commonly identi ed in grape berries, namely cyanidin, peonidin, delphinin, malvidin, pelargonidin and petunidin (Fuleki and Ricardo-da-Silva, 1997). The dominant anthocyanin is different among grape varieties, and pelargonidin derivatives can hardly be detected in most of Vitis vinifera varieties (Ageorges et al., 2006).
Anthocyanin biosynthesis is regulated by the MYB-bHLH-WD40 transcription complexes (Jaakola, 2013). Among them, subgroup 5 of MYB transcription factor family is related to proanthocyanidin synthesis, subgroup 7 regulates avonol biosynthesis, and the subgroup 5 and 6 are regarded as inhibitors and activators of anthocyanin biosynthesis respectively (Dubos et al., 2010). VvMYBA1 and VvMYBA2 were identi ed as the main transcription factors regulating grape skin coloring, they can bind to ciselements in the promoter of UFGT and activate its transcription. In white grapes, a retrotransposon gret1 was inserted into the promoter region of VvMYBA1, blocked VvMYBA1 transcription (Kobayashi, 2004), meanwhile VvMYBA2 mutated into a non-functional type, resulting in no anthocyanin accumulation (Azuma et al., 2008). Subgroup III of bHLH have also been reported to be involved in anthocyanin biosynthesis Xi et al., 2021), VvMYC1 interacted with MYBs in grapes and participated in anthocyanin and proanthocyanin biosynthesis regulation (Hichri et al., 2010). Some ERF transcription factors were also found regulating anthocyanin accumulation in Arabidopsis thaliana (Koyama and Sato, 2018), apple  and other fruits.
Flavonoids are synthesized in cytoplasm and then transported to vacuoles for storage. At present, the transport mechanism of anthocyanin after synthesis is still unclear. There are four types of anthocyanin transporters reported in grapes: glutathione S-transferase (GST), ATP-binding cassette (ABC) transporters, multidrug and toxic extrusion compound transporters (MATE), and bilitranslocase (BTL) (Petrussa et al., 2013). A total of 161 transporters were identi ed in the study of grape vacuolar proteome in our laboratory, among which 13 ABC transporters were more abundant at fruit maturity (Kuang et al., 2019).
For most red grape varieties, anthocyanin only exists in vacuoles in the 3-4 layers of cells of skin, and have white esh and juice, provides the possibility of making white wine from red grapes. Red esh is very rare among several thousands of grape varieties. Alicante Bouschet is the most well-known red esh variety, it has rose-red esh, and is used as a teinturier variety to enhance the color of red wine. Yan73 is a Chinese teinturier variety resulted from a cross between Alicante Bouschet and Muscat Hamburg (Xi et al., 2013). Anthocyanin accumulation happened in the esh remarkably before veraison, compared with its dark-skin and white-esh parent Muscat Hamburg, the key enzymes of anthocyanin biosynthesis, such as CHS3, UFGT, F3 '5' H, F3H1 and LDOX, and transcription factor MYBA1 were found highly expressed in the esh of Yan73 (Xie et al., 2015).
Red esh is a phenotype that exists in many fruits, such as orange, kiwi, apple, plum and others. With higher antioxidant capacity and attractive color, red esh fruits are welcomed by the market. Our laboratory bred a red-esh table grape variety 'Mio Red'. Besides fresh consumption, 'Mio Red' is also suitable for making juice. Comparing and analyzing the divergent coloring in esh and skin of 'Mio Red' berry could provide us further understanding of the coloring regulation of berry esh, which could bene t breeding of a grape as a healthy snack in the future. (middle of stage III) and 86 DAA (late stage III) respectively. Three biological replicates were made at each sampling time with 40 berries per replicate. Samples were taken randomly from clusters in different orientations, transported to the laboratory in ice box. The esh tissues and skin were carefully separated with a scalpel and then quickly frozen in liquid nitrogen. All samples were stored at −80°C for the subsequent analysis.

Determination of total anthocyanin
The method of Ni et al. (2020b) was used for the determination of total anthocyanin, and some modi cations were made. In brief, 0.2 g grape tissue was ground in liquid nitrogen, then mixed in 1 ml methanol: acetic acid (99 : 1, v/v) solution, and placed in refrigerator at 4 ℃ for 24 h. The absorbance of the supernatant was measured by ultraviolet spectrophotometer (Beckman Coulter, Brea, CA, USA) at wavelengths of 530, 620 and 650 nm. The relative anthocyanin content was calculated as follows: [(A530-A650)-0.2 × (A650-A620)]/0.1.

Determination of total content of avonoids
The content of avonoids was determined as Robinson et al. (2019). After the sample was thoroughly grounded in liquid nitrogen, 0.2 mg was mixed with 1 ml precooled 80% ethanol, and extracted at 4 ℃ for 24 h. After centrifuging at 12,000 rpm at 4 ℃ for 20 min, 0.5 mL supernatant was put into a 10 mL centrifuge tube, then 0.3 mL 8% NaNO 2 , 0.3 mL 10% Al (NO 3 ) 3 solution, 2 mL 2M NaOH solution and 4.9 mL ethanol were added in sequence. After standing for 10 min, the absorbance of the reaction solution was determined at 510 nm. Rutin was used as the standard, and the nal avonoid content was calculated as mg rutin/g.FW.

Flavonoids metabolome analysis
About 1g of the frozen-stored grape esh/ skin samples was freezing dried in vacuum, then grounded to powder with a mixer mill (MM 400, Retsch). For each extraction, 100 mg powder was dissolved in 1ml 70% methanol, standing at 4 ℃ for 24h, then 10,000 g centrifuged for 10min, the supernatant was ltered with a microporous membrane (0.22 µm) for LC-MS/MS analysis. Each sample had three biological replicates.
Metabolites were identi ed using Analyst 1.6.3 (AB SCIEX, Ontario, Canada) and quanti ed by MRM. The metabolites were then submitted to Principal Component Analysis (PCA) and Orthogonal partial leastsquares discrimination analysis (OPLS). Thresholds of VIP (variable impact in project) ≥ 1 and fold change ≥ 2 or ≤ 0. 5 were set as criteria for metabolites with signi cant differences.

transcriptome analysis
RNA was extracted from berry esh and skin using a modi ed CTAB method  with three biological replicates per sample at each sampling point. RNA quality was checked by 1% agarose gel electrophoresis, NanoPhotometer spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA), Qubit 2.0 uorometer (Qubit 2.0, Life Technologies, ThermoFisher Scienti c, USA), and Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA) before constructing the libraries. The mRNA enrichment and cDNA library establishment were following the standard protocol. The libraries were sequenced by an Illumina HiSeq platform after quali ed quality inspection. After removing the sequencing connectors and low-quality reads, the clean reads were annotated against the grape reference genome (Canaguier et al., 2017), the libraries were compared using HISAT2 (Kim, 2015). Fragments per kilobase of exon per million fragments mapped (FPKM) was used to measure the expression level of transcripts. DESEQ2 (Varet et al., 2016) was used to analyze the differential expression among sample groups. The differentially expressed genes were screened with |log2Fold Change| >= 1 and FDR < 0.05.

Quantitative Real-Time PCR (qRT-PCR)
According to transcriptome data, 11 structural genes in avonoid synthesis pathway were selected, and their expression levels at ve sampling time points of berry development were validated by qRT-PCR. The primers were listed as Table S1. The PCR reaction was carried out with ABI QuantStudio 6 Q6 (Applied Biosystems) using ChamQ Universal SYBR qPCR Master Mix (Vazyme, China). The 10 µL ampli cation system was consisted of 5 µL SYBR qPCR Master Mix, 0.2 µL upstream and downstream primer respectively (10 µM), 1 µL template, and 3.6 µL distilled water. The ampli cation procedure was 95 ℃ 30 sec; 95 ℃ 10 sec, 60 ℃ 30 sec, 40 cycles; 95 ℃ 15 sec, 60 ℃ 60 sec, 95 ℃ 15 sec. All samples were detected by at least three technical repetitions. β-actin of grapes was used as the reference gene, and the data were quantitatively analyzed by 2 −ΔΔCT method.

Flavonoids in the skin and esh of 'Mio Red' at ripening
The skin and esh of 'Mio Red' berry both showed coloration at ripening. The skin was uniformly colored, and the anthocyanin in the esh were mainly concentrated around the style and near the skin (Fig. 1A).
Microscopic observation revealed esh coloration was an independent event other than diffusion of anthocyanin from the skin (Fig. 1B). The anthocyanin content of 'Mio Red' esh was about 6.2 times lower than that of the skin. The total avonoids content demonstrated the same trend as anthocyanin, that in the esh was 14.25 times lower than in the skin (Fig. 1C).
A total of 173 avonoids were identi ed in colored skin and esh of 'Mio Red', which could be further divided into 8 groups, including 63 avones, 34 avonols, 18 avanones, 17 polyphenols, 17 anthocyanins, 16 avonoids, 5 iso avones and 3 proanthocyanins (Table S2). Clustering analysis of the avonoids data revealed obvious differences between and within the two samples ( Fig. S1A). Principal component analysis showed a clear trend of separation between the sample groups, PC1 and PC2 contributed 84.23% and 5.89% of the differences respectively ( Fig. S1B), indicating that there was a signi cant difference between the metabolites of the skin and the esh of 'Mio red' grape. Hierarchical heatmap clustered all biological replicates together, and metabolites of the same type together (Fig. S1C), which shows that the results of the avonoid metabolome were of high reliability.
Sixty-eight metabolites were recruited as of different content between 'Mio Red' esh and skin by VIP ≥ 1 and fold change ≥ 2 or ≤ 0.5 (Table S3). Compared with the skin, 64 avonoids were of lower content, and four were of signi cant higher content in the esh (Fig. S2A), which were quercetin 3-O-glucoside ( avonol), epicatechin-epiafzelechin (polyphenol), apigenin 6,8-C-diglucoside ( avonoids) and hesperetin 5-O-glucoside ( avanone). Among them, quercetin 3-O-glucoside, epicatechin-epiafzelechin and apigenin 6,8-C-diglucoside were hardly detected in the skin (Table 1). Four avonoid biosynthesis pathways with signi cant differences were found by locating the differential metabolites in KEGG pathway, among which the avonoid and avonol biosynthesis pathway had the largest number of differential metabolites, followed by avonoid biosynthesis pathway (Fig. S2B). The six common anthocyanins reported in grapes were detected in both the skin and the esh of 'Mio Red', but their contents were of signi cant difference in the two tissues. Peonidin derivatives were the main anthocyanin in the skin, while the esh was dominated by pelargonidin derivatives (Table S4).

Transcriptome differences of 'Mio Red' skin and esh
Transcriptome sequencing of skin and esh of ripe 'Mio Red' berries (3 biological replicates in each sample), resulted a total of 307,871,952 clean reads, of which 93.8% reached Q30 (Table S5), 78.01% -83.80% of the total clean reads were uniquely matched with grape reference genome (map quality ≥ 30) (Table S6). There were 20,929 genes expressed in both the skin and esh, 1,326 and 1,531 genes were speci cally expressed in the skin and esh respectively ( Fig. 2A).

Expression of structural genes of avonoid and anthocyanin biosynthesis pathway
The structural genes which were identi ed as DEGs in avonoid biosynthesis included CHS, CHI, F3'H, F3'5'H and F3H. The expression of these genes was higher in the skin than the esh of 'Mio Red' berry. The low expression of CHS and CHI in the esh was in consistent with the low content of total avonoids in the tissue. Among F3H, F3'H and F3'5'H, F3H had the highest FPKM value, followed by F3'H, F3'5'H was with the lowest FPKM, and one F3'5'H (Vitvi06g01895) exhibited the highest fold change in transcript abundance, the FPKM value in the skin was about 12 times higher than that in the esh. The signi cant difference in F3'5'H expression could eventually lead to the difference in anthocyanin ratio between delphinidin 3-O-glucoside and petunidin 3, 5-diglucoside (Fig. 3).
Only one DFR gene was found differentially expressed. DFR, acted as the gatekeeper controlling upstream substrate owing to anthocyanin biosynthesis pathway, the gene was highly expressed in both skin and esh, speci cally in the esh the FPKM value of DFR was 147. Two ANS DEGs were identi ed, ANS1 (Vitvi02g00435) and ANS2 (Vitvi11g00565) were found highly expressed in the skin and esh, respectively. The FPKM value of ANS2 reached 201 in the esh, which was 5 times of ANS2' expression in the skin. Only one DEG was revealed with UFGT, the expression was about 8 times higher in the skin than in the esh (Fig. 3).
The transcript abundance of genes in avonol and proanthocyanidin biosynthesis was signi cantly lower than that of the anthocyanin biosynthesis branch. It is worth noting that the expression level of FLS (Vitvi18g02538) in the esh was 8.2 times higher than that in the skin, which is in agreement with the metabolome result that the esh of 'Mio Red' tends to accumulate avonols compared with the skin. DEGs of LAR and ANR, which were found extremely low expression in both the skin and esh, were in consistent with the detected low proanthocyanidin content.

Flavonoid biosynthesis gene expression during berry development
The expression of avonoid biosynthesis structure genes was found in well agreement with the speci c avonoid and anthocyanin accumulation character between the esh and skin. In the phenylpropanoid pathway, CHS and CHI demonstrated overall increased expression in the early stage of berry development and decreased from veraison to berry ripening. Overall higher expression was found in the skin than in the esh, which was in line with the higher avonoid and anthocyanin content in the skin. The high expression of CHS and CHI in the esh were found at 72 and 65 DAA respectively (Fig. 4).
In the iso avone pathway, F3'5'H, F3'H, F3H, ANS, UFGT and LAR showed overall higher expression in the skin than in the esh, especially that of UFGT demonstrated the largest expression difference. F3'H and F3H revealed higher expression in the esh than in the skin at 86 DAA which supported the late anthocyanin accumulation feature of 'Mio Red' esh. Moreover, the expression of FLS and ANR, key genes for the biosynthesis of avonols and proanthocyanins, had generally higher expression in esh than the skin (Fig. 4), supported the assayed high apigenin 6,8-C-diglucoside, epicatechin-epiafzelechin, hesperetin 5-O-glucoside and quercetin 3-O-glucoside content found in the esh.

Identi cation and correlation analysis of transcription factors related to avonoid biosynthesis
The important transcription factor family associated with fruit growth and development identi ed in the transcription of skin and esh of 'Mio Red' included 45 AP2/ERF, 39 MYB (including MYB-related), 32 bHLH, 27 WRKY, 23 NAC and 3 WD40. Most of the of NAC and WD40 were up-regulated in the esh. The number of up-regulated and down-regulated genes of bHLH was the same. Members of the rest transcription factor families were mostly down-regulated in the esh (Fig. 5A).
The ERF, MYB (subgroup 4/5/6/7) and bHLH (subgroup ) transcription factors annotated in the transcriptome of 'Mio Red' related to the biosynthesis of anthocyanin were separately constructed into phylogenetic trees to screen TFs that might be involved in berry esh coloration regulation. A total of 12 ERF, 9 MYB and 3 bHLH genes were screened. ERF Vitvi18g01617, ERFBP-like1, Vitvi18g02398 were highly expressed in both skin and esh. They were clustered with PbERF3 of pear, JcERF035 of physic nut (Jatropha curcas L.) and MdERF38 of apple, respectively (Fig. 5B).
Among the MYB transcription factors, MYBA1/2/3 clustered in subgroup 6, and was highly expressed in both skin and esh, which may directly regulate fruit coloring. MYBC2-L3 and MYB3 were clustered in subgroup 4, which may play the role of feedback regulation and inhibit the excessive accumulation of anthocyanin. MYBPA7 was clustered in subgroup 5, its expression level was low, which was consistent with the low content of proanthocyanin in 'Mio Red' fruit. MYBPA1 was clustered in subgroup 7, homologous to VvMYBPA1 and VvMYBF1 (Fig. 5C). Among bHLH transcription factors, MYC2 with high expression level and cluster with AtMYC2/4/5 of Arabidopsis thaliana. bHLH93 and bHLH116, which were clustered together with AtbHLH33, were highly expressed in the skin and esh, respectively (Fig. 5D).
There were only 3 differentially expressed WD40 found in 'Mio Red', and their expression level was low. NAC and bZIP TFs family were screened with FPKM ≥ 10 in the skin or esh, 6 and 2 transcription factors were obtained respectively. Three NACs namely NAC100, NAC71 and NAC83, and 2 bZIPs i.e. BZIP41 and BZIP9, were signi cantly upregulated in the esh than the skin (Fig. 6A).

Integrated analysis of DEG and DAM related to avonoid biosynthesis and transport
The genes and metabolites with high abundance in avonoid pathway and Pearson correlation coe cient greater than 0.8 were selected for correlation analysis. In Figure 7A, genes and metabolites were negatively correlated in quadrants 1 and 9, positively correlated in quadrants 3 and 7. The results showed that the number of genes and metabolites in quadrants 7 and 9 was much higher than that in quadrants 3 and 1, which indicated that the genes and metabolites related to avonoids in the esh were signi cantly down-regulated compared with the skin, regardless whether genes were positively or negatively regulated. Combined DEGs and DAMs KEGG enrichment analysis showed that only avonoid biosynthesis pathway was signi cantly enriched (p < 0.05) (Fig. 7B).
The correlation network was constructed with genes and metabolites enriched in avonoids, avone and avonol, and anthocyanin biosynthesis, the coe cient of correlation was higher than 0.8. Flavonoids, avone and avonol biosynthesis were associated with most genes. Three UFGT genes were screened correlating with anthocyanin biosynthesis, which could play a key role in the differential coloring of esh and skin (Fig. S3).
In this study, 23 putative avonoid-related transporters were identi ed in 3 types: 15 GST, 5 MATE and 3 ABC transporter. They may implement the transporting of anthocyanins, proanthocyanins and avonols. The results of transcriptome showed that all the transporter genes were higher expressed in the skin than the esh. Among the transporters GST14 and MATE4 were of high FPKM in both the skin and esh.

Pattern of anthocyanin accumulation in the esh of different red esh grape varieties
Most of the colored grapes only accumulate anthocyanin in the skin at berry ripening, veraison is the common start point of skin anthocyanin accumulation. However, esh anthocyanin development time is of remarkable diversity among the small population of red esh varieties. The esh coloring of the teinturier wine variety Yan73 occurred in the stage II of the double sigmoid curve and was obviously earlier than its skin coloring (He et al., 2010). 'Summer Black' is a table grape variety with a V. vinifera and V. labrusca hybrid background. The esh and skin of 'Summer Black' red-esh mutant start coloring almost at the same time at berry ripening (Zhang et al., 2018b). In the case of 'Mio red', the coloration of the esh was signi cantly later than that of the skins. Since esh coloration occurs only when the skin was dark, the coloration of 'Mio Red' esh may have a non-light-dependent character.
There was small difference in anthocyanin content between the skin and esh of Yan73, and the proportion of different anthocyanin derivatives was similar, with the highest content of malvidin and peonidin derivatives and the lowest content of pelargonidin (Chen et al., 2018). The anthocyanin pro le in the esh and skin of 'Summer Black' red esh mutant was also similar, the appearance of the red esh trait may relate to the enhanced anthocyanin biosynthesis in the whole berry. As the color of the skin gradually increasing, the shading effect prevents the synthesis of anthocyanin in the esh, which resulted the anthocyanin level of the esh far lower than that of the skin (Zhang et al., 2018b). DAMs in the skin and esh of 'Mio red' grape include avonols, avonoids and anthocyanins, more than 1/3 of avonoids metabolites were of signi cant content different, which implied that the main avonoids synthesized in the skin and esh of 'Mio red' grape were different.
It was previously thought that grapes did not contain pelargonidin (Jeong et al., 2006), as the DFR of the studied varieties could not effectively reduce dihydrokaempferol to produce leucopelargonidin (Xie et al., 2004). Later on, pelargonidin was detected with extremely low content in V. vinifera varieties such as Cabernet Sauvignon and Pinot Noir. In Yan73 grape, DFR can complete the reduction of dihydrokaempferol, but it was more inclined to synthesize cyanidin and delphinin using dihydroquercetin and dihydromyricetin as substrates, so the content of pelargonidin derivatives in Yan73 was low (Xie et al., 2018). However, in 'Mio Red' grape, especially in the esh, pelargonidin accumulated to a certain extent.
Previous studies reported that the expression pattern of ANS varies within different grape varieties, ANS protein accumulates in various tissues of grapevine and exhibits different effects on secondary metabolites (Boss et al., 1996). In grape leaves, ANS was mainly involved in the biosynthesis of PAs, while in stem phloem, ANS responds to both anthocyanin and PAs (Wang et al., 2011). There were two main ANS in 'Mio Red' berry, ANS1 (Vitvi11g00565) and ANS2 (Vitvi02g00435), which were highly expressed in the skin and esh, respectively. The speci c spatiotemporal expression of the two ANS genes could be involved in the divergent coloration initiation and anthocyanin content in the skin and esh of 'Mio Red' grape.

Transcription factors on avonoid accumulation
At present, there are few studies on transcription factors that regulate the expression of structural genes that synthesize avonoids in grape esh. The study of Yan73 suggested that the accumulation of anthocyanin in esh tissue was probably due to speci c expression of the structural genes and TFs, which was coordinated and regulated by VvMYBA1 transcription activator and VvMYBC2-L1 transcription repressor (Xie et al., 2019). Comparing the metabolic and transcriptomic pro les of the skin and esh of three table grape varieties ('Kyoho', 'Wink', 'Italia'), it was found that besides anthocyanin, avonol and avanol were also different in a wide range, and MYB24, MADS5 and two ubiquitin proteins (RHA2) were suggested as promising candidates for regulating avonoids biosynthesis in grape (Lu et al., 2021).
In the present study, TFs related to avonoid biosynthesis were screened, and AP2/ERF took a high number. The direct relationship between ERFs and avonoid biosynthesis is still unclear. ERFs generally regulated anthocyanin accumulation by interacting with MYB or bHLH TFs. It was reported that in apple MdERF1B and MdERF3 promote anthocyanin and proanthocyanidin biosynthesis by interacting with MdMYBs (Zhang et al., 2018a;An et al., 2018). In pear, PpERF105 inhibits anthocyanin biosynthesis by inducing the expression of a transcription repressor PpMYB140 (Ni et al., 2019).
Among of the 11 MYBs screened from our transcriptome and phylogenetic tree analysis, the expression of MYBA1, MYBA2 and MYBA3 as activators in the skin was signi cantly higher than that in the esh.
MYBA1 and MYBA2 regulate anthocyanin biosynthesis by binding to the promoter of UFGT . As a homolog of MYBA1/A2, MYBA3 was also considered to play an activator role in anthocyanin biosynthesis (Leng et al., 2020). The expression patterns of these three MYBs in the esh of 'Mio Red' could be important to the esh coloration. MYBPA7, VvMYBPA2 and VvMYBPAR regulating proanthocyanin biosynthesis were clustered into subgroup 5 (Koyama et al., 2014). The low expression of MYBPA7, especially in the esh, could be the main reason for the low accumulation of proanthocyanins in 'Mio Red'. MYBPA1 and its homologous proteins VvMYBPA1, VvMYBF1 and AtMYB12 in subgroup 7 regulate avonol biosynthesis (Czemmel et al., 2009). Studies have shown that VvMYBPA1 activates LAR, ANR and promoters of several avonoid pathway structural genes including CHS, F3'5'H and ANS, participates the biosynthesis of procyanidins .
The existence of suppressors is essential, which can ne-tune the synthesis process when anthocyanin could accumulate too much, and maintain a homeostasis of fruits (LaFountain and Yuan, 2021). MYB inhibitors are generally divided into two types, MYB4 and MYBC2-L1 homologs. MYB4 inhibited small molecular weight phenolic compounds, while MYBC2-L1/L3 regulated the level of avonoids and balanced the action of activators (Cavallini et al., 2015). MYBC2-L1 and MYB3 were screened in this study as inhibitors, which could inhibit the accumulation of avonoids. In the correlation analysis of MBW complex in 'Mio Red', both MYB activators and inhibitors were positively correlated with VvbHLH93.
Among the 3 bHLH screened from the phylogenetic tree, bHLH 116 and bHLH93 were clustered together with AtbHLH116 (ICE1) and AtbHLH93 (ICE2), they were reported a role in low temperature stress (Fursova et al., 2009). MYC2 clustered with AtMYC2, AtbHLH3. In Arabidopsis thaliana, MYC2 promotes avonoid biosynthesis in the downstream of JA signaling by regulating the expression of positive regulators PAP1 and EGL3. MYC2 can also regulate anthocyanin biosynthesis in corn (Zea mays) (Dombrecht et al., 2007). MYC2 had a high expression in the skin and esh of 'Mio Red', which could promote the coloring.
NAC and BZIP family TFs participate in various life processes in plants, including the regulation of avonoid biosynthesis. Overexpression of PaNAC03 in spruce reduced the accumulation of avonoids (Dalman et al., 2017). Overexpression of VvibZIP22 in grapes signi cantly increased the content of avonoids (Malacarne et al., 2016). These two kinds of TFs may also contribute to the synthesis of avonoids in 'Mio Red' grape.

Other regulatory factors on avonoid accumulation
Flavonoids are synthesized in the cytoplasm and then transported to vacuoles for storage or to other destinations where they function as bioactive molecules. Different compartment of avonoids biosynthesis, storage and function requires effective transport mechanisms to realize their biological functions. In grapes, VvABCC1, VvGST4, AM1 and AM3 were con rmed roles in anthocyanin transport, and VvGST1 in proanthocyanidin transportation (Zhao, 2015). VvMATE1/2, which are homologs to AtTT12 in Arabidopsis thaliana, was located in vacuoles and Golgi complex respectively (Perez-Diaz et al., 2014). Stress-induced TaGSTL1 selectively recognized avonol in wheat (Dixon and Edwards, 2010). Our DEG and DAM integrated analysis suggested GST14 and MATE4 may play a role in the transport of avonoids, while GST8 could involve in the transportation of quercetin 3-O-glucoside, hesperetin 5-Oglucoside and epicatechin-epiafzelechin. Compared with the skin, the expression of transporter genes in the esh was lower, the ine cient transport after synthesis could be another reason for the less accumulation of avonol in the esh.
Structural genes and regulatory genes in avonoid biosynthesis pathway can be induced by light (Jaakola et al., 2010). Light-induced anthocyanin accumulation requires light-responsive elements, including Constitutively Photomorphogenic 1 (COP1) and Long Hypocotyl 5 (HY5) (Maier et al., 2013). In apples, MdMYB1 accumulated in light and degraded through an ubiquitin-dependent pathway in dark (Li et al., 2012). In addition, under low ultraviolet irradiation, the transcription levels of FLS, HY5 and MYB10 in apple fruits were down-regulated, and anthocyanin and avonol content decreased (Henry-Kirk et al., 2018). PpBBX16, an activator of light-induced anthocyanin accumulation, was found in pears (Bai et al., 2019). In nature, there are many red esh fruits whose esh coloration does not depend on light signals and in a different time line to their light dependent coloration skin, such as g (Ficus carica L.) (Cui et al., 2021). Anthocyanin accumulation in the esh of 'Mio Red' berry happens in a low light if not a light absent condition, elucidation of the mechanism could help us to improve the coloring of the historical fruit.

Conclusion
'Mio Red', a dark skin table grape develops color in the esh at ripening, the color was identi ed as anthocyanin accumulation. The esh color development pattern is of remarkable difference from the other reported red esh grape variety. Metabolome analysis revealed peonidin and pelargonidin derivatives as the leading anthocyanin in 'Mio Red' skin and esh respectively, exhibited bio-diversity in the trait of red esh formation. Transcriptome analysis uncovered that most of the catalyze enzymes in the anthocyanin biosynthesis pathway was coded by two genes, that showed different expression preference in the esh and skin. A set of transcription factors were co-expression analyzed and recruited for the future function validation. The results further our understanding of avonoids and anthocyanin accumulation in grape esh, and provide new clues for subsequent regulatory mechanism study. Moreover, this study could pave the way for breeding of future red-ash seedless table grapes having the health bene t of red wines.

Declarations Funding
The work was supported by 111 Project (B17043).

Declarations
There is no con ict of interest.

Author Contributions
HM and MS designed experiments, RL and MS prepared samples for RNA-seq, RL completed physiological experiments and qRT-PCR, RL and MS completed all data collation and analysis, RL, MS and HM, ZW, YZ, CH, AP wrote articles and made revisions. All authors have read and approved the manuscript for publication.  Expression analysis of structural genes in avonoid biosynthesis pathway during berry development.
DAA: days after anthesis. The expression levels were normalized to mean ± SE of 3 biological replicates Integral analysis of DAMs and DEGs. (A) Nine-quadrant diagram, DEGs and DAMs with Pearson correlation coe cients greater than 0.8 were divided into 1-9 quadrants from left to right and from top to bottom with black dotted lines. (B) KEGG enrichment histogram, the X-axis represents the metabolic pathways, the Y-axis represents the expression as -log (p-value). Green and red represent DEGs and DAMs, respectively. (C) Expression pro le of avonoids-associated transporters and their correlation heatmap with DAMs. Red presents positively correlated, blue presents negatively correlated, and the size of the circle indicates signi cance. The larger the circle, the higher the signi cance

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.