Abstract
Vitis amurensis grape cultivars and hybrids are mainly used to make wines in Northeast Asia with a cold winter. Anthocyanidin glucosylation at 5-O position catalyzed by 5-O-glucosyltransferase (5GT) in grape skins is crucial for color stability of red wines. To study 5GT functions in anthocyanidin diglucosides synthesis of V. amurensis, 20 5GTs were preliminarily identified from a genome-wide characterization of the UDP-glycosyltransferase family according to the reported 5GTs, which were also performed phylogenetic and bioinformatics analysis. Two important 5GTs, Vitvi0900582.t01 and Vitvi05g01269.t01, were screened through analyses of anthocyanidin diglucosides accumulation and gene expression in berry skins of three representative grape cultivars with significant differences in anthocyanin glycosylation. Fourteen alleles of each of the 5GTs were cloned from 14 V. amurensis cultivars and hybrids as well as from V. vinifera ‘Cabernet Sauvignon’, and sequence analysis and functional prediction were performed. From three perspectives of phenotype, transcriptional level, and genotype, it has been found that the functional allele at the Vitvi0900582.t01 locus of Chr 9 played a decisive role in the synthesis of abundant anthocyanidin diglucosides in V. amurensis. In addition, the trace anthocyanidin diglucosides detected in V. vinifera ‘Cabernet Sauvignon’ were led by the functional allele at the Vitvi05g01269.t01 locus of Chr 5. This study provides preliminary data for further research on the regulatory mechanism of anthocyanidin diglucosides in the grapes with the V. amurensis pedigree to improve their wine quality in future breeding efforts.
Vitis amurensis, which has high cold resistance, is the major grape variety cultivated in Northeast Asian, which has a long cold winter (Wang et al. 2021). The berries of V. amurensis are rich in anthocyanins and are mostly used for winemaking. The content and composition of anthocyanins in grape berries have important effects on the color, hue, and stability of wine. There are two main types of glucosylated anthocyanins in grapes, anthocyanidin monoglucoside and diglucosides. In contrast to anthocyanidin monoglucoside, the anthocyanidin diglucosides in wine do not readily react with other molecules, such as pyruvic acid, acetaldehyde, and flavanols, to form polymeric pigments during fermentation and ageing (Yuan et al. 2022). Such polymeric pigments are more stable than the monomeric anthocyanins from grape skin, which is of crucial importance in the color stability of aged V. vinifera wines (Burtch et al. 2017). During the aging process of wine, anthocyanidin diglucosides are undesirable because they cannot easily form more stable derivative pigments with other compounds, so they are prone to oxidative browning, losing the color of red wine and affecting the quality (Yang et al. 2014). However, anthocyanidin diglucosides have been found to account for more than 90% of total anthocyanin content in V. amurensis skins (Zhu et al. 2021), which is a disadvantage for the ageing potential of wines.
Glycosyltransferases (GTs) play a key role in the synthesis of glycosylated anthocyanins in plants and nearly all other organisms (Espley et al. 2019; Yue et al. 2021). The GTs are directly involved in the biosynthesis of disaccharides, monosaccharides, oligosaccharides, alkyl glucosides, and polysaccharide metabolites, which comprise large family (Mora-Montes et al. 2022). The key enzymes in the anthocyanidin diglucosides synthetic pathway are UDP-glucose:flavonoid 3-O-glucosyltransferase (3GT or UFGT) and UDP-glucose:flavonoid 5-O-glucosyltransferase (5GT), which belong to different groups of the UDP-X glucosyltransferase (UGT or GT1) family (Sui et al. 2018). At first, 3GT is responsible for converting colorless anthocyanidin into stable anthocyanidin 3-O-monoglucoside, which causes berry coloring from veraison (Ford et al. 1998). Many previous studies have focused on the molecular mechanism of coloring in grape berries (Cheng et al. 2021; Kobayashi et al. 2004, 2005). Then 5-O-glucosylation is the further addition of the second glucoside to form anthocyanidin 3,5-O-diglucosides via 5GT (Neumann et al. 2006) and generally occurs in non–V. vinifera grapes (Jánváry et al. 2009; Yang et al. 2014). However, there are few studies related to the synthesis and regulation of anthocyanidin diglucosides, especially in V. amurensis.
In this study, we aimed to reveal, comprehensively and systematically, the potential roles of 5GTs in anthocyanidin diglucosides accumulation in the berry skins of grapes with the V. amurensis pedigree. First, 5GTs in grape were identified from the UGT family according to reported 5GTs in previous researches. In addition, a set of bioinformatics analyses were performed to further our understanding of the 5GT family. During fruit development and ripening, quantitative reverse transcription-polymerase chain reaction (qRT-PCR) was carried out to investigate expression patterns of 5GT genes that were prescreened from the RNA-seq data of three representative grape berry skins with marked differences in anthocyanin glycosylation. Moreover, the allelic variations of the important 5GTs were surveyed in 15 grape cultivars. This work will provide basic information for further study of the synthesis and regulatory mechanism of anthocyanidin diglucosides, and for the breeding of V. amurensis grapes with improved wine color stability in the future.
Materials and Methods
Plant materials.
In 2018, all the samples of V. amurensis cultivars and hybrids were collected in the V. amurensis germplasm repository of the Chinese Academy of Agricultural Sciences, and the samples of V. vinifera ‘Cabernet Sauvignon’ (CS) were collected in the experimental vineyards of China Agricultural University. On the basis of our previous study (Zhu et al. 2021), three representative grape cultivars with significant differences in anthocyanin glycosylation, V. amurensis ‘Zuoshanyi’ (ZS-1), V. amurensis × V. vinifera ‘Zuohongyi’ (ZH-1) and V. vinifera CS, were selected to analysis the anthocyanins and gene expression. There were five collection stages of the three grape berries, including green-berry stage (GBS; 6 weeks after flowering), early veraison (EV, 5%–10% berries turned red per cluster at 9 weeks after flowering), middle veraison (MV, 50% berries turned red per cluster at 10 weeks after flowering), late veraison (LV, 100% berries turned red per cluster at 11 weeks after flowering), and mature stages (MS, 16 weeks after flowering for V. amurensis and 20 weeks after flowering for V. vinifera). The berry samples were plucked from the six selected grapevines for each cultivar. At least six fresh clusters were collected for each sample; they were then placed on ice packs and immediately taken to the laboratory. Three 50-berry batches randomly selected from the top, middle and bottom portions of the clusters. The skins were manually separated from the berry flesh, then rapidly frozen in liquid nitrogen. For the gene cloning, the fully opened fifth leaf uninfected with pests and diseases from growing point of annual shoot was picked with a tweeze in a cryotube and rapidly frozen in liquid nitrogen. All experiments were performed in three biological replicates.
Identification and phylogenetic analysis of 5GT genes in grape.
The UGT proteins in grape were identified based on the latest Grape Genome Database (12X.2 version, https://urgi.versailles.inra.fr/Species/Vitis/Annotations) using with Hidden Markov Model, BLASTP program, SMART, and Pfam. Phylogenetic analysis was conducted using the Neighbor–Joining method with a 1000 bootstrap value in the MEGA X software. According to the phylogenetic analysis of UGT family (Supplemental Fig. 1), a phylogenetic tree was constructed with UGTs of L group and several reported 5GTs. According to clustering branch and kinship, the putative 5GT members were identified in grape.
Chromosome localization, gene structure, and motif feature analysis of 5GT genes in grape.
Information on grape 5GTs, including open reading frames, description, and chromosome distribution, were obtained from Grape Genome Database. The distribution of UGT genes on the chromosomes was visualized with URGI website. Gene structure view of the TBtools software was used to integrate the motif and gene structure. The online program MEME Suite was applied to identify the conserved motifs.
Extraction and analyses of anthocyanins.
Extraction of anthocyanin compounds in grape skins was carried out according to the method described in our previous research (Zhu et al. 2021). Qualitative and quantitative analyses of phenolic compounds were performed on an Ultimate 3000 UPLC/Q-Exactive orbitrap MS (Thermo Fisher Scientific Inc., Waltham, MA, USA) with a Thermo GOLD Hypersil column (C18, 50 mm × 2.1 mm, 1.9 µm) as described previously (Zhu et al. 2021). The statistical significance of differences in the contents of anthocyanins among samples was calculated by SPSSAU using an online tool (The SPSSAU Project 2019).
RNA extraction and qRT-PCR.
Total RNA of grape skins was isolated and purified using MJzol reagent, then subjected to agarose gel electrophoresis to examine the RNA integrity. The RNA concentration and quality were assessed by OD260/280 and OD260/OD230 readings using a ultraviolet spectrophotometer. The total RNA (1 μg) was used for cDNA synthesis, which was carried out in a 20-μL reaction system using cDNA synthesis kit. VvUbiquitin1 (Vitvi16g01364.t01) was selected as the internal reference gene. The primers of the target 5GT genes and reference gene are provided in Supplemental Table 1. qRT-PCR was conducted as follows: 95 °C for 5 min, 40 cycles of 94 °C for 5 s, 55 °C for 30 s, and 72 °C for 30 s. Three replicates were performed for each biological sample. Quantitative data were calculated using the 2-ΔΔCT method. The SPSS 26.0 software was used to analyze the variance of gene expression among samples and the correlation between gene expression and anthocyanidin diglucosides accumulation.
Allelic cloning and sequence analysis of important 5GT family genes.
There were 15 grape cultivars with V. amurensis pedigree and V. vinifera CS selected as raw materials for allelic clone of the important 5GTs (Table 2). The genomic DNA (gDNA) was extracted from frozen leaf samples by a modified CTAB extraction protocol (Porebski et al. 1997; Kobayashi 1998). Two pairs of primers were used to amplify the entire coding sequence of Vitvi0900582.t01 and Vitvi05g01269.t01 by PCR from grape leaf (Supplemental Table 2). The PCR amplification was conducted in a 50-μL reaction mixture containing 32 μL double-distilled water, 1.5 μL gDNA, 1 μL dNTP mix, 10 μL 5× Phusion HF buffer, 2.5 μL forward primer (10 μM), 2.5 μL reverse primer (10 μM), and 0.5 μL Phusion High-Fidelity Polymerase (New England Biolabs, Ipswich, MA, USA). The PCR conditions were as follows: 98 °C for 30 s, 30 to 35 cycles at 98 °C for 10 s, 55 to 62°C for 30 to 60 s, 72 °C for 30 s, and 72 °C for 10 min. Agarose gel electrophoresis was used to verified the PCR products. The target fragments were purified and recycled with the QIAquick Gel Extraction Kit (Bao Bioscience and Technology Company, Shanghai, China). The purified target gene was then inserted into the multiple cloning sites of the pLB vector (Tiangen Rapid DNA Ligation Kit, Beijing, China). Positive Escherichia coli was used to obtain the full cDNA sequence of 5GT identified by PCR, enzyme digestion, and sequencing (Biomed Biotech Co., Ltd., Beijing, China). Seqman software in the DNASTAR Lasergene 7.1 package (DNAstar, Madison, WI, USA) was used for sequence splice and assembly. Sequence calibration, deduced amino acid sequence, and encoded open reading frames were analyzed by Editseq software (DNAStar Inc). Multiple sequence alignments were used with DNAMAN8.0 (Lynnon, Quebec Canada). The Cha5GT (Jánváry et al. 2009) and Vv5GT3 (Xing et al. 2015) were used as the reference sequence for Vitvi0900582.t01 and Vitvi05g01269.t01, respectively.
Results
Genome-wide identification and bioinformatic analysis of the 5GT family in grape
Identification and phylogenetic analysis of grape 5GTs.
To identify grape 5GTs, we conducted the phylogenetic analysis on grape UGT gene family. A total of 230 UGT genes were screened from the online database and divided into 16 groups. As shown in Supplemental Fig. 1, the 5GTs were clustered in Group L, which contained 39 members, in the entire UGT family of grape. Subsequently, a phylogenetic tree was constructed using the proteins encoded by 5GTs in Vitis and other species (Torenia hybrid, Perilla frutescens, Glandularia hybrida, Gentiana triflora, Eustoma grandiflorum, Solonum melongena, Petunia hybrida, Arabidopsis thaliana, and Iris hollandica) reported in previous studies together with the UGTs of the Group L (Fig. 1A). According to the phylogenetic relationship, 20 UGTs were found to be clustered together with 5GTs in grape reported previously, including the proteins encoded by 5GT-Cha and 5GT-Dia (Jánváry et al. 2009), Vv5GT1∼5 (Xing et al. 2015), Va5GT (He et al. 2015), VaSF5GT, Vr5GT, and VIOGT1∼3 (Chen and Liu 2019). The physical and chemical properties of the 20 5GTs are shown in Supplemental Table 3. On the basis of the phylogenetic tree, only 1 gene, Vitvi09g00582.t01, from the grape genomic sequence was clustered with the reported genes Cha5GT, Dia5GT, Vr5GT, Va5GT, and Vv5GT1 in the Class I subgroup. Vitvi01g00822.t01 and Vitvi17g00753.t01 were clustered with Vv5GT2 in the Class II subgroup. Vitvi05g01301.t01, Vitvi05g01269.t01, Vitvi05g01296.t01, Vitvi05g01300.t01, Vitvi05g02103.t01, Vitvi05g01271.t01, Vitvi05g01270.t01, and Vitvi05g01269.t01 and the functionally identified Vv5GT3 were clustered in the Class III subgroup. Vitvi05g01291.t01 and Vitvi05g01294.t01 were clustered together with Vv5GT4 in the Class IV subgroup. Vitvi05g01288.t01, Vitvi05g01278.t01, Vitvi05g01276.t01, Vitvi05g01289.t01, Vitvi05g01279.t01, and Vitvi05g01274.t01, together with VIOGT3 and Vv5GT5, were clustered in the Class V subgroup. In addition, Vitvi05g02116.t01 and Vitvi05g01304.t01 were clustered with the reported VIOGT2, VaSF5GT, and VIOGT1 genes in the Class VI subgroup. Therefore, the 20 UGTs were preliminarily identified as putative members of the 5GT family in grape based on their relatedness. Class IV showed relatively high similarity with other species, and the homology of Vitvi05g01294.t01 in Class IV with Ph5GT reached 60.00%. Class III showed relatively low similarity with other species, and the homology of Vitvi05g01271.t01 in Class III with Ph5GT reached 22.03%.
Phylogenetic and bioinformatics analysis of grape 5-O-glucosyltransferase (5GT) family. (A) Phylogenetic tree of 5GTs in the L group of grape UGT family (Supplemental Fig. 1) with reported 5GTs using the neighbor-joining method with a bootstrap value of 1000. The putative 20 5GTs and the reported 5GTs in grape were divided into six clades (I–VI) and identified with different colors. The prefixes “Th,” “Pf,” “Gh,” “Gt,” “Eg,” “Sm,” “Ph,” “At,” and “Ih” stand for Torenia hybrid, Perilla frutescens, Glandularia hybrida, Gentiana triflora, Eustoma grandiflorum, Solonum melongena, Petunia hybrida, Arabidopsis thaliana, and Iris hollandica, respectively. (B) Chromosome distribution map of 20 putative 5GT genes in grape. (C) Gene structures of the 5GT genes. Green boxes represent the coding sequence (CDS) whereas yellow boxes represent untranslated region (UTR). (D) The motif composition of 5GT proteins. Motifs in the 5GT proteins (1–9) represented by colored boxes. Black lines indicate relative protein lengths. (E) Sequence logo of motif 1.
Citation: J. Amer. Soc. Hort. Sci. 149, 4; 10.21273/JASHS05403-24
Chromosome distribution of grape 5GT genes.
To provide an overview of the location of grape 5GT genes, each 5GT was located on the chromosome of genome. Notably, the 20 putative 5GT genes were unevenly distributed on four of 19 chromosomes in grape (Fig. 1B). Vitvi0900582.t01 of the Class I subgroup was on the upper part of Chr 9. Vitvi01g00822.t01 and Vitvi17g00753.t01 in Class II were located in upper Chr 1 and middle Chr 17, respectively. All the genes in Class III, IV, V, and VI were located on the lower part of Chr 5.
Gene structure of grape 5GTs.
To explore the structural diversity of the grape 5GT genes, intron or exon arrangements were constructed based on the phylogenetic tree we generated (Fig. 1C). We discovered that the lengths of most grape 5GTs were similar (1077-1494 bp), except for in the case of Vitvi05g01274.t01 (1905 bp) in Class V. The 5GT exon numbers contained 1, 2, 3, and 5, and the 5GT intron numbers varied from 0 to 3. It was found that the genes on the same branch of the evolutionary tree had similar numbers of introns and exons as well as similar gene structures (Fig. 1C), which may be related to the functional differentiation and structural diversity of the 5GT gene family. Moreover, 15 of 20 genes had only one exon and no introns. The remaining five genes in Classes II or III had one to three introns and two, three, or five exons, respectively. Overall, the differences in the numbers of introns and exons among the 5GTs reflected their diverse structures and functions.
Motif composition of grape 5GTs.
To explore the functional domains of grape 5GT proteins, the MEME motif search tool was used to identify the motifs. A total of nine conserved motifs were characterized (motifs 1–9, Fig. 1D). Among the nine motifs, motif 1 was a conserve sequence consisting of 44 amino acid residues at the C-terminal (Fig. 1E), which is called PSPG-box. Twelve of the 20 members contained all nine motifs. The 5GT encoded by Vitvi05g01270.t01 in Class III contained the fewest motifs (four), and the 5GT encoded by Vitvi09g00582.t01 in Class I together with the other 5GT proteins also had fewer motifs (six). Among the nine motifs, motif 1 and motif 2 were found in all the 5GTs. There were 18 5GTs starting from motif 4 in the N-terminus, and 15 5GTs ending with motif 8 in the C-terminus. Motif 3, motif 6, motif 7, motif 5, and motif 9 were found in 18, 18, 17, 16, and 15 5GTs, respectively. The motif composition and distribution of the 5GTs belonging to Classes IV and V, except for the protein encoded by Vitvi05g01274.t01, were the same.
Anthocyanidin diglucoside accumulation and temporal expression of 5GTs in grape skins during fruit ripening
Anthocyanins are synthesized starting at veraison and accumulate continuously as grape berries ripen. In the present study, V. amurensis ZS-1 had the highest total anthocyanin levels during ripening, followed by the V. amurensis and V. vinifera hybrid ZH-1 (Fig. 2; Supplemental Table 4). The lowest levels were found in V. vinifera CS. At MS, the total anthocyanin accumulation in ZS-1 (32.35 mg/g DM) and ZH-1 (10.08 mg/g DM) was more than 23.12 and 7.2 times greater than that in CS (1.4 mg/g DM). Similar to total anthocyanins, the contents of anthocyanidin diglucosides increased from 0.16 mg/g DM and 0.04 mg/g DM at EV to 31.37 mg/g DM and 3.32 mg/g DM at MS in ZS-1 and ZH-1, respectively. Moreover, anthocyanidin diglucosides accumulation in ZS-1 was significantly greater than that in ZH-1 (P < 0.05). However, only traces of anthocyanidin diglucosides were detected in V. vinifera CS, which was consistent with the findings of Xing et al. (2015). In addition, there were differences in the proportions of anthocyanidin diglucosides between ZS-1 and ZH-1. From EV to MS, the proportion of anthocyanidin diglucosides in ZS-1 clearly increased, reaching 96.80% at MS. However, the proportion in ZH-1 decreased from 40.74% at EV to 32.85% at MS (Fig. 2). These results confirmed that V. amurensis stood out in a marked way of anthocyanidin diglucosides accumulation.
Anthocyanin accumulation in the three grape skins during berry ripening using UPLC-QE-MS2 (n = 3). ZS-1, ZH-1 and CS represent ‘Zuoshanyi’, V. amurensis × V. vinifera ‘Zuohongyi’, and V. vinifera ‘Cabernet Sauvignon’, respectively. EV, MV, LV, and MS respectively represent early veraison (5%–10% berries turned red per cluster at 9 weeks after flowering), middle veraison (50% berries turned red per cluster at 10 weeks after flowering), late veraison (100% berries turned red per cluster at 11 weeks after flowering), and mature stages (16 weeks after flowering for Vitis amurensis and 20 weeks after flowering for Vitis vinifera). Red letters represent the analysis of variance (ANOVA) result of anthocyanins contents. Black letters represent the ANOVA results of anthocyanidin diglycoside contents. Different uppercase letters indicate significant differences among different grapes at the same stage (P < 0.05). Different lowercase letters indicate significant differences among different stages for the same grape (P < 0.05).
Citation: J. Amer. Soc. Hort. Sci. 149, 4; 10.21273/JASHS05403-24
After removing the genes with very low transcript levels based on the RNA-seq data (Supplemental Table 5), eight 5GT genes were selected for berry skin qRT-PCR analysis during grape development (Fig. 3). The expression of Vitvi09g00582.t01 (Class I, 9 Chr) in the three cultivars tended to increase overall with fruit maturity. The expression of these genes decreased markedly in the order of ZS-1 > ZH-1 > CS. At MS, its expression in ZS-1 was nearly 69- and 11-fold higher than that in CS and ZH-1, respectively. The variation in Vitvi09g00582.t01 expression was consistent with the accumulation of anthocyanidin diglucosides. In addition, the expression of Vitvi05g01269.t01 (Class III, 5 Chr) in CS was significantly greater than that in ZS-1 and ZH-1 (P < 0.05). Its expression in CS was upregulated 968- and 45-fold compared with that in ZS-1 and ZH-1 at MS, respectively. In general, its expression increased obviously during berry development in CS, which aligned with the trend of anthocyanidin diglucosides. However, its expressions in ZS-1 and ZH-1 decreased as berry development. The other six 5GTs exhibited low expression in all three cultivars.
Relative expression levels of 5GTs in grape skin during berry development of ZS-1, ZH-1 and CS using quantitative reverse transcription-polymerase chain reaction (n = 3). ZS-1, ZH-1, and CS represent ‘Zuoshanyi’, Vitis amurensis × Vitis vinifera ‘Zuohongyi’, and Vitis vinifera ‘Cabernet Sauvignon’, respectively. Different uppercase letters indicate significant differences among different grapes at the same stage (P < 0.05). Different lowercase letters indicate significant differences among different stages for the same grape (P < 0.05).
Citation: J. Amer. Soc. Hort. Sci. 149, 4; 10.21273/JASHS05403-24
To clarify the relationship between 5GT expression and anthocyanidin diglucosides accumulation in grape skins, a correlation analysis was carried out (Fig. 4). As expected, Vitvi0900582.t01 had the greatest positive correlation (>0.8) among the 8 5GTs in all three cultivars, which were significant except for the proportion of anthocyanidin diglucosides in ZS-1 and ZH-1 (P < 0.05). Therefore, we speculated that 5GT plays a decisive role in the synthesis of anthocyanidin diglucosides in grape skin, especially in V. amurensis. Vitvi05g01269.t01 was negatively correlated with ZS-1 and ZH-1, but a significant positive correlation was detected in CS (P < 0.05). The relative expression of Vitvi0900582.t01 in CS was low (<1) at all five stages. These results indicate that Vitvi05g01269.t01 might play an important role in the detection of trace anthocyanidin diglucosides in CS. Except for Vitvi0900582.t01 and Vitvi05g01269.t01, the correlation coefficients of the remaining 6 5GTs were negative for the three cultivars. Therefore, Vitvi0900582.t01 and Vitvi05g01269.t01 were selected for subsequent allele cloning and analysis.
Correlation analysis between the relative expression levels and anthocyanidin diglycosides of three representative cultivars. *, ** P < 0.05 and < 0.01, respectively.
Citation: J. Amer. Soc. Hort. Sci. 149, 4; 10.21273/JASHS05403-24
Allele cloning and sequence analysis of important 5GT family genes
Va5GT alleles in V. amurensis cultivars and hybrids.
Jánváry et al. (2009) first cloned a functional Cha5GT and a nonfunctional Dia5GT from the interspecific hybrid Vitis Regent cultivar with the V. vinifera pedigree and identified two important loss-of-function sites in Dia5GT by the enzyme activity of the recombinant enzymes. He et al. (2015) cloned Va5GT from V. amurensis ZS-1 and verified its function via heterologous expression. Vitvi09g00582.t01 identified from the reference genomic sequence of V. vinifera ‘Pinot’ together with the reported 5GTs was categorized into the Class I subgroup. BLAST comparison revealed that the nucleotide sequence homology of Vitvi09g00582.t01 with Dia5GT, Cha5GT, Va5GT, Vr5GT, and Vv5GT1 was 100%, 88.15%, 87.93%, 87.50%, and 73.12%, respectively. These 5GTs are located at the same locus on Chr 9. These findings indicate that these genes are likely alleles of one 5GT gene from different grape species or cultivars. As a result, we selected the functional Va5GT as the reference sequence to clone and analyze the Vitvi09g00582.t01 alleles in V. amurensis cultivars and hybrids.
A total of 14 alleles, named 5GT1–5GT14, in the present study we were cloned among 15 cultivars in this study. To determine the relative functional importance and distribution patterns of allelic mutations, we aligned the deduced amino acid sequences of the 5GT genes from Va5GT (Fig. 5A). According to the mutation characteristics, the Va5GT alleles were divided into three types, types I, II, and III (Table 1). The type I 5GT alleles were characterized by premature stop codons and/or frameshift mutations, and included 5GT1, 5GT2, 5GT4, 5GT8, and 5GT12. For example, 5GT1 and 5GT2 contained a base pair deletion at the 902 nucleotide position that causes a frameshift [G301#: frameshift (#) to replace the amino acid G at amino acid 301]. In addition, 5GT1 and 5GT2 also contained a premature stop codon at position 328 [E328*: a stop codon (*) replacing E at amino acid position 328]. Compared with 5GT1, 5GT2 lacks 1182 nd and 1183 rd AG bases (AG1182-1183△), which is relatively rare (8%) in V. vinifera grapes investigated in the research of Yang et al. (2014). Thus, the Type I alleles should lose 5GT enzyme function due to frameshift mutation or premature codon production, which resulted in large abnormal protein changes or peptide chain truncation. Type II mutations were characterized by only amino acid substitutions, including 5GT3, 5GT5, 5GT7, 5GT9-5GT11, 5GT13, and 5GT14. These mutations might exert normal 5GT enzyme functions because of relatively complete peptide chains and small amino acid sequence mutations. The type III mutation, represented only by 5GT6 from ‘Hasang’, was characterized by in-frame amino acid deletion mutations or insertion mutations and no frameshift mutations or premature stop codons. The main feature was the in-frame deletion mutation (NKEE317-320△) caused by consecutive deletion of 12 bases, indicating that 5GT6 might lose the normal function of the 5GT enzyme. In summary, we speculated that the proteins encoded by 8 alleles (5GT3, 5GT5, 5GT7, 5GT9, 5GT10-5GT11, 5GT13, and 5GT14) could have a normal 5GT function, while the proteins produced by the remaining 6 alleles (5GT1, 5GT2, 5GT4, 5GT6, 5GT8, and 5GT12) lost their 5GT function.
Distribution and alignment of observed amino acid changes in grape 5GTs with deduced amino acid sequences of Va5GT alleles (A) and Vv5GT3 alleles (B). The functional reference sequences of Va5GT and Vv5GT3 are from He et al. (2015) and Xing et al. (2015), respectively. 5GT3, 5GT12, 5GT15, and 5GT27 were the representative alleles obtained in this study. Pink and blue shading indicate identical and similar amino acids, respectively. Dashes represents alignment gaps among sequences. Number represents amino acid position in allele. Letter below the alignment indicates amino acid substitution. * Indicates a stop codon, △ indicates an amino acid deletion, and # indicates a frameshift.
Citation: J. Amer. Soc. Hort. Sci. 149, 4; 10.21273/JASHS05403-24
Key mutation sites and distribution of important 5-O-glucosyltransferase (5GT) alleles in grapes with Vitis amurensis pedigree and Vitis vinifera ‘Cabernet Sauvignon’ (CS).
Vv5GT3 alleles in V. amurensis cultivars and hybrids.
Xing et al. (2015) confirmed that the recombinant protein Vv5GT3 cloned from V. vinifera CS could glucosylate the 3-O- and 5-O-positions of anthocyanidins and flavonols via heterologous expression. Vitvi05g01269.t01 from the reference genomic sequence together with Vv5GT3 was categorized into the Class III subgroup. Their nucleotide sequence homology was 97.79%. These findings indicate that there should be different alleles of one 5GT gene. Therefore, Vv5GT3 was used as the reference sequence for the functional prediction of alleles cloned in this study. A total of 14 alleles of Vv5GT3 (5GT15-5GT28) were obtained from 15 grape cultivars and hybrids (Fig. 5B). These mutations were divided into type I and II according to their characteristics (Table 2). The 5GT alleles of type I, including 5GT18, 5GT19, 5GT20, 5GT22, 5GT23, and 5GT27, were characterized by early termination codon or frameshift mutations, indicating that they might have lost 5GT enzyme function. For example, the key feature of 5GT18 was the presence of a premature stop codon at nucleotide position 80 (S27*). 5GT27 mainly contained a frameshift mutation (W162#) in the amino acid sequence starting at position 162 and producing an early termination codon (L230*) at position 230. The remaining alleles belonged to type II mutations and characterized by amino acid replacement only. Their encoded enzymes were speculated to have the ability to catalyze the synthesis of anthocyanidin diglucosides due to the relatively complete peptide chain and limited extent of mutations in the amino acid sequence. In general, we speculated that the proteins encoded by 8 alleles (5GT15-5GT17, 5GT21, 5GT24-5GT26, and 5GT28) could perform normal functions, whereas the remaining six alleles (5GT18, 5GT19, 5GT20, 5GT22, 5GT23, and 5GT27) were unfunctional 5GT genes.
5GT genotypes and contents and proportions of anthocyanidin diglucosides of the grape cultivars included in the present study.
Discussion
V. amurensis, which has the highest cold resistance among all grape species, originates from and is mainly distributed in Northeast Asia, which has a long, cold winter (Bak et al. 2016). Red wines are usually made from grapes with V. amurensis pedigrees in Northeast China. However, the undesirable color stability of red wines has an unfavorable effect on aging potential because of the abundant anthocyanidin diglucosides in V. amurensis berry skin. Although several previous studies have revealed that the gene encoding the key enzyme 5GT plays an important role in the synthesis of anthocyanidin diglucosides in grape berry skin (He et al. 2015; Jánváry et al. 2009; Xing et al. 2015; Yang et al. 2014), the reason for the large variation in the composition of glucosylated anthocyanin among different grape species or cultivars is not yet known. Moreover, few studies have focused on the mechanism of anthocyanin diglucosides synthesis in V. amurensis. To address this research gap, after the identification and bioinformatics analysis of the 5GT family, we studied the functions of the 5GT family genes involved in the accumulation of anthocyanidin diglucosides in grapes with the V. amurensis pedigree and CS as the V. vinifera control cultivar.
At first, we aimed to conduct a comprehensive search of the 5GTs that may play a role in the synthesis of anthocyanin diglucosides. On the basis of whole-genome identification and phylogenetic analysis of the UGT family, a total of 20 5GT family members combined with reported 5GTs were identified and partitioned into six groups. The bioinformatic analysis results, including those related to physicochemical properties, chromosome distribution, gene structure, protein domain, and motif composition, suggested diversity in the sequential structure and function of the 5GTs. More notably, 5GT genes with similar exon or intron structures were more likely to cluster together (Fig. 1C), which was consistent with the results of gene UGT family analysis in most plants, such as Zea mays (Han and Luthe 2021), Solanum tuberosum (Mo et al. 2022), and Citrus clementina (Liu et al. 2022). According to motif composition, all 20 5GTs identified contained motif 1, which was consistent with the PSPG box. The PSPG box is a highly conserved domain in UGTs, consisting of 44 amino acid residues at the C-terminus (Vogt and Jones 2000). In this study, some amino acids in motif 1 were the same among all 20 5GTs, such as 1 and 2 (WC), 4 (Q), 8 (L), 14 through 16 (GCF), 19 (H), 21 through 23 (GWN), 27 (E), and 33 (V), whereas others exhibited a certain degree of variation (Fig. 1E). This region is considered the site where glycosyltransferase recognizes and binds to the donor molecule. Analysis of the three-dimensional model of MtUGT71G1 revealed a direct interaction between the uracil part of UDP-glucose and the highly conserved HCGWNS residue (Shao et al. 2005). In addition, the last two amino acids of the PSPG region are considered the key amino acid residues for glycosyl recognition, such as Glu381 and Gln382 in MtUGT71G1 and Asp374 and Gln375 in VvGT1 (Ono et al. 2010; Shao et al. 2005). Therefore, motif 1 might play important roles in the evolution and functional differences of 5GTs in grape.
We selected three representative cultivars with significant differences in anthocyanin glycosylation—V. amurensis ZS-1, V. amurensis × V. vinifera ZH-1, and V. vinifera CS—to analyze anthocyanin accumulation and 5GT expression in grape berry skin during berry development. On the basis of the grape skin transcriptome data (Supplemental Table 5), 8 of 20 5GTs were subjected to qRT-PCR after removing genes with low expression levels (Fig. 3). Among the eight 5GTs, Vitvi0900582.t01 (Class I, 9 Chr) exhibited a dramatically greater expression level than did the other genes in ZS-1 and ZH-1, whereas Vitvi05g01269.t01 (Class III, 5 Chr) exhibited the greatest expression level in CS. Correlation analysis revealed that Vitvi0900582.t01 was significantly and positively correlated with the accumulation of anthocyanidin diglucosides (P < 0.05). Vitvi05g01269.t01 was positively correlated with the content and proportion of anthocyanidin diglucosides only in CS (P < 0.05), in contrast to the results of Xing et al. (2015). On the basis of the phylogenetic analysis (Fig. 1A) and nucleotide sequence homology, Vitvi09g00582.t01 and the reported 5GTs, such as Dia5GT (Jánváry et al. 2009), Cha5GT (Jánváry et al. 2009), Va5GT (He et al. 2015), Vr5GT (Chen et al. 2019) and Vv5GT1 (Xing et al. 2015) should be the same gene, and Vitvi05g01269.t01 and Vv5GT3 (Xing et al. 2015) should also be one 5GT. In addition, Vitvi05g01291.t01 and Vitvi05g01269.t01 showed significant negative correlations with the proportion of anthocyanidin diglucosides in ZS-1 (P < 0.05). However, their expression levels were relatively low. We suspect that these two genes might be involved in the glycosyl transfer of other phenolic compounds to negatively regulate anthocyanidin diglucosides accumulation in the flavonoid synthesis pathway, but further research is needed.
We subsequently cloned the alleles of Vitvi0900582.t01 and Vitvi05g01269.t01 and analyzed their function and genotype in 14 V. amurensis cultivars and hybrids and in V. vinifera CS using Va5GT and Vv5GT3, respectively, as functional reference sequences. Fourteen Va5GT alleles (5GT1-5GT14; Fig. 5A) and 14 Vv5GT3 alleles (5GT15-5GT28; Fig. 5B) were cloned. Although gene mutation is a random process, certain mutation patterns can still be observed. The Va5GT alleles were divided into three types: premature stop codon and frameshift mutations (type I), amino acid substitutions (type II), and in-frame amino acid deletion or insertion mutations (type III). Overall, frameshift indel mutations and introduced stop codons were involved in generating diverse Va5GT alleles. According to sequence analysis and functional prediction, both type I and type III mutations might inactivate 5GT enzyme activity, whereas type II mutations are more likely not to affect the normal function of the 5GT enzyme. For example, the E76D and V208A mutations were present in multiple grape cultivars, and the R412K mutation was found in all the Va5GT alleles, indicating that these three amino acid differences could not affect the possession or losing of 5GT enzyme function (Tables 1 and 2). There were six Va5GT alleles cloned in this study; in agreement with the alleles from Yang et al. (2014), 5GT1 and 5GT2 in CS belong to A1 and A2 of type A from V. vinifera, in which frameshift mutations started at amino acid 301 (G301#). 5GT4 is B1 of type B from V. vinifera, which was characterized by a frameshift mutation (V236#) starting from the 234th amino acid. 5GT3, 5GT5, and 5GT6 are W5, W19, and W23, respectively, of type W with no frameshift or premature stop mutations. Mutations of 5GT3 and 5GT6 were detected in V. amurensis and V. cinerea, respectively, and 5GT5 was detected in V. cinerea, V. labrusca, V. aestivalis, and some interspecific hybrids (Yang et al. 2014). These results demonstrate that some gene mutations could exist in different grape species. Despite variations in nucleotide sequences among some alleles, their amino acid sequences remained identical due to the mutation preserving the open reading frame. For example, 5GT19, 5GT22, and 5GT23 had different nucleotide sequences and gene sources, but the translated protein sequences were identical. Different alleles from different grape cultivars had the same mutation at the same amino acid position, indicating that the mutation location was only conserved to a limited extent and constituted a mutation hotspot, reflecting the genetic diversity of grape in the evolutionary process. For example, 5GT18 from ‘Hasang’ and 5GT20 from ‘Beimei’ have different nucleotide sequences, but the translated protein sequences are exactly the same.
Combining the results of the present study with those of our previous research (Zhu et al. 2021), we summarized the genetic composition of both 5GT genes and the total content and proportion of anthocyanidin diglucosides in the mature berry skin of each grape cultivar (Table 2). All the grapes with the V. amurensis pedigree contained at least one functional allele of Va5GT on Chr 9 and certain proportions of anthocyanidin diglucosides in their skins (2.83–42.92 mg/g, Zhu et al. 2021), whereas V. vinifera CS possessed two nonfunctional alleles of Va5GT on Chr 9 and two functional alleles of Vv5GT3 on Chr 5, as well as trace anthocyanidin diglucosides in skin (0.0032 mg/g). These differences between V. amurensis and V. vinifera indicated that the genotype of Va5GT plays a decisive role in the substantial accumulation of anthocyanidin diglucosides, whereas the functional alleles of Vv5GT3 can only lead to trace anthocyanidin diglucosides. Apparently, the synthesis mechanism of anthocyanidin diglucosides is different from that of anthocyanidin monoglucoside in grape skin. The synthesis of anthocyanidin monoglucoside depends on the expression of 3GT genes (Ford et al. 1998; Offen et al. 2006), which are mainly regulated by the MYB-bHLH-WDR (MBW) ternary complex formed by the transcription factors myeloblastosis (MYB), basic helix–loop-helix (bHLH) and WD-repeat (WD) (Ding et al. 2021). Furthermore, the genotype of the MYB gene has a crucial effect on the synthesis of the monoglucoside anthocyanin (Kobayashi et al. 2001). For example, white grape berries lack functional alleles of mybA1, leading to loss of pigmentation in their skins (Kobayashi et al. 2004, 2005).
Among the V. amurensis cultivars and hybrids, ZS-1 and Beichun had the same Va5GT and Vv5GT3 genotypes, whereas the content and proportion of anthocyanidin diglucosides in ZS-1 were much greater than those in ‘Beichun’. For the Vv5GT3 genotype, a homozygote of 5GT3 were found in ‘Zuoshan-2’, ‘Tonghua-3’, and ‘Changbai-5’, but there were significant differences in the content and proportion of anthocyanidin diglucosides among these grape skins (P < 0.05). Furthermore, there was no obvious correlation between the number of functional alleles of Va5GT (1 or 2) and anthocyanidin diglucosides accumulation in the grape skins of V. amurensis cultivars and hybrids. These phenomena remained unexplained by the genotypes of both important 5GTs. The large variation in the content and proportion of anthocyanidin diglucosides among different V. amurensis cultivars and hybrids may be related to the catalytic activities of enzymes led by different Va5GT allele mutations. Alternatively, some transcription factors may regulate the synthesis of anthocyanidin diglucosides, similar to the regulation of 3GT transcription by the MBW complex (Hichri et al. 2011). This hypothesis needs further study and verification.
Conclusions
In this study, a total of 20 5GT family members were identified for the first time in grape. The 5GTs were divided into six groups by phylogenetic analysis. Bioinformatics analysis revealed diverse gene chromosome distributions, sequential structures, and functions of 5GTs. Two important 5GTs, Vitvi0900582.t01 and Vitvi05g01269.t01, were screened through qRT–PCR and anthocyanin analysis. Next, 14 alleles of each important 5GT were cloned, and their functions were predicted using the functional gene sequences of Va5GT and Vv5GT3. Analyses of the anthocyanidin diglucosides phenotype, 5GT gene expression, and genotype of two important 5GTs indicated that grapes with the V. amurensis pedigree could synthesize a certain proportion of anthocyanidin diglucosides if at least one functional 5GT allele of class I was present at the Va5GT locus of Chr 9. The functional 5GT allele in Class III at the Vv5GT3 locus of Chr 5 was the reason for the trace anthocyanidin diglucosides detected in V. vinifera CS berry skin. This study provides systematic and fundamental information for further research on regulatory mechanisms of anthocyanidin diglucosides and provide new insights for future breeding to improve the color stability of the wines containing the V. amurensis pedigree.
References Cited
Bak MJ, Truong VL, Ko SY, Nguyen XNG, Ingkasupart P, Jun M, Shin JY, Jeong WS. 2016. Antioxidant and hepatoprotective effects of procyanidins from wild grape (Vitis amurensis) seeds in ethanol-induced cells and rats. Int J Mol Sci. 17(5):758. https://doi.org/10.3390/ijms17050758.
Burtch CE., Mansfield AK, Manns DC. 2017. Reaction kinetics of monomeric anthocyanin conversion to polymeric pigments and their significance to color in interspecific hybrid wines. J Agric Food Chem. 65(31):6379–6386. https://doi.org/10.1021/acs.jafc.6b05331.
Chen M, Liu H. 2019. Cloning and expression and genetic transformation analysis of Vitis amurensis flavonoid-5-O-glucosyltransferase gene (5GT). J China Agric Univ. 24(5):73–81.
Cheng J, Yu K, Shi Y, Wang J, Duan C. 2021. Transcription factor VviMYB86 oppositely regulates proanthocyanidin and anthocyanin biosynthesis in grape berries. Front Plant Sci. 11:613677. https://doi.org/10.3389/fpls.2020.613677.
Ding R, Che X, Shen Z, Zhang Y. 2021. Metabolome and transcriptome profiling provide insights into green apple peel reveals light- and UV-B-responsive pathway in anthocyanins accumulation. BMC Plant Bio. 21:1–15. https://doi.org/10.1186/s12870-021-03121-3.
Espley RV, Leif D, Plunkett B, McGhie T, Henry-Kirk R, Hall M, Johnston JW, Punter MP, Boldingh H, Nardozza S, Volz RK, Donnell SO, Allan AC. 2019. Red to brown: An elevated anthocyanic response in apple drives ethylene to advance maturity and fruit flesh browning. Front Plant Sci. 10:1248. https://doi.org/10.3389/fpls.2019.01248.
Ford CM, Boss PK, Høj PB. 1998. Cloning and characterization of Vitis vinifera UDP-glucose: flavonoid 3-O-glucosyltransferase, a homologue of the enzyme encoded by the maize bronze-1 locus that may primarily serve to glucosylate anthocyanidins in vivo. J Biol Chem. 273(15):9224–9233. https://doi.org/10.1074/jbc.273.15.9224.
Han Y, Luthe D. 2021. Identification and evolution analysis of the JAZ gene family in maize. BMC Genomics. 22:1–21. https://doi.org/10.1186/s12864-021-07522-4.
He F, Chen WK, Yu KJ, Ji XN, Duan CQ, Reeves MJ, Wang J. 2015. Molecular and biochemical characterization of the UDP-glucose: Anthocyanin 5-O-glucosyltransferase from Vitis amurensis. Phytochemistry. 117:363–372. https://doi.org/10.1016/j.phytochem.2015.06.023.
Hichri I, Barrieu F, Bogs J, Kappel C, Delrot S, Lauvergeat V. 2011. Recent advances in the transcriptional regulation of the flavonoid biosynthetic pathway. J Exp Bot. 62(8):2465–2483. https://doi.org/10.1093/jxb/erq442.
Jánváry L, Hoffmann T, Pfeiffer J, Hausmann L, Töpfer R, Fischer TC, Schwab W. 2009. A double mutation in the anthocyanin 5-O-glucosyltransferase gene disrupts enzymatic activity in Vitis vinifera L. J Agricultural Food Chem. 57(9):3512–3518. https://doi.org/10.1021/jf900146a.
Kobayashi N. 1998. A simple and efficient DNA extraction method for plants, especially woody plants. Plant Tissue Cult Biotech. 4:76–80.
Kobayashi S, Ishimaru M, Ding C K, Yakushiji H, Goto N. 2001. Comparison of UDP-glucose: flavonoid 3-O-glucosyltransferase (UFGT) gene sequences between white grapes (Vitis vinifera) and their sports with red skin. Plant Sci. 160(3):543–550. https://doi.org/10.1016/s0168-9452(00)00425-8
Kobayashi S, Goto-Yamamoto N, Hirochika H. 2004. Retrotransposon-induced mutations in grape skin color. Science. 304(5673):982–982. https://doi.org/10.1126/science.1095011.
Kobayashi S, Goto-Yamamoto N, Hirochika H. 2005. Association of VvmybA1 gene expression with anthocyanin production in grape (Vitis vinifera) skin-color mutants. J Jpn Soc Hortic Sci. 74(3):196–203. https://doi.org/10.2503/jjshs.74.196.
Liu S, Li Y, Fang H, Huang B, Zhao C, Sun C, Li S, Chen K. 2022. Genome-wide identification and expression analysis of MATE gene family in citrus fruit (Citrus clementina). Genomics. 114(5):110446. https://doi.org/10.1016/j.ygeno.2022.110446.
Mo F, Li L, Zhang C, Yang C, Chen G, Niu Y, Si J, Liu T, Sun X, Wang S, Wang, Chen Q, Chen Y. 2022. Genome-wide analysis and expression profiling of the phenylalanine ammonia-lyase gene family in Solanum tuberosum. Int J Mol Sci. 23(12):6833. https://doi.org/10.3390/ijms23126833.
Mora-Montes HM, García-Gutiérrez K, García-Carnero LC, Lozoya-Pérez NE, Ramirez-Prado JH. 2022. The search for cryptic L-rhamnosyltransferases on the Sporothrix schenckii genome. J Fungi. 8(5):529. https://doi.org/10.3390/jof8050529.
Neumann K, Eibach R, Zyprian E, Hausmann L, Töpfer R. 2006. Development of a molecular marker for an anthocyanin 5-O-glucosyltransferase homologous gene of Vitis ssp. correlated with anthocyanin 3, 5-diglucoside formation in berry skin. Acta Hortic. 827:457–460. https://doi.org/10.17660/ActaHortic.2009.827.78.
Offen W, Martinez-Fleites C, Yang M, Kiat‐Lim E, Davis BG, Tarling CA, Ford CM, Bowles DJ, Davies GJ. 2006. Structure of a flavonoid glucosyltransferase reveals the basis for plant natural product modification. EMBO J. 25(6):1396–1405. https://doi.org/10.1038/sj.emboj.7600970.
Ono E, Homma Y, Horikawa M, Kunikane-Doi S, Imai H, Takahashi S, Kawai Y, Ishiguro M, Fukui Y, Nakayama T. 2010. Functional differentiation of the glycosyltransferases that contribute to the chemical diversity of bioactive flavonol glycosides in grapevines (Vitis vinifera). Plant Cell. 22(8):2856–2871. https://doi.org/10.1105/tpc.110.074625.
Porebski S, Bailey LG, Baum BR. 1997. Modification of a CTAB DNA extraction protocol for plants containing high polysaccharide and polyphenol components. Plant Mol Bio Rep. 15:8–15. https://doi.org/10.1007/BF02772108.
Shao H, He X, Achnine L, Blount JW, Dixon RA, Wang X. 2005. Crystal structures of a multifunctional triterpene/flavonoid glycosyltransferase from Medicago truncatula. Plant Cell. 17(11):3141–3154. https://doi.org/10.1105/tpc.105.035055.
Sui X, Zhao M, Xu Z, Zhao L, Han X. 2018. RrGT2, a key gene associated with anthocyanin biosynthesis in Rosa rugosa, was identified via virus-induced gene silencing and overexpression. Int J Mol Sci. 19(12):4057. https://doi.org/10.3390/ijms19124057.
The SPSSAU Project. 2019. SPSSAU. Version 26.0 [online software]. https://www.spssau.com. [accessed on 5 Oct 2023].
Vogt T, Jones P. 2000. Glycosyltransferases in plant natural product synthesis: characterization of a supergene family. Trends Plant Sci. 5(9): 380–386. https://doi.org/10.1016/s1360-1385(00)01720-9.
Wang Y, Xin H, Fan P, Zhang J, Liu YB, Dong Y, Wang ZM, Yang YZ, Zhang Q, Ming R, Zhong GY, Li SH, Liang ZC. 2021. The genome of Shanputao (Vitis amurensis) provides a new insight into cold tolerance of grapevine. Plant J. 105(6):1495–1506. https://doi.org/10.1111/tpj.15127.
Xing RR, Li SY, He F, Yang Z, Duan CQ, Li Z, Wang J, Pan QH. 2015. Mass spectrometric and enzymatic evidence confirm the existence of anthocyanidin 3, 5-O-diglucosides in cabernet sauvignon (Vitis vinifera L.) grape berries. J Agric Food Chem. 63(12):3251–3260. https://doi.org/10.1021/acs.jafc.5b00053.
Yang Y, Labate JA, Liang Z, Cousins P, Prins B, Preece JE, Aradhya M, Zhong GY. 2014. Multiple loss-of-function 5-O-glucosyltransferase alleles revealed in Vitis vinifera, but not in other Vitis species. Theor Appl Genet. 127:2433–2451. https://doi.org/10.1007/s00122-014-2388-6.
Yue XF, Jing SS, Ni XF, Zhang KK, Fang YL, Zhang ZW, Ju YL. 2021. Anthocyanin and phenolic acids contents influence the color stability and antioxidant capacity of wine treated with mannoprotein. Front Nutr. 8:691784. https://doi.org/10.3389/fnut.2021.691784.
Yuan JF, Lai YT, Chen ZY, Song HX, Zhang J, Wang DH, Gong MG, Sun JR. 2022 Microwave irradiation: Effects on the change of colour characteristics and main phenolic compounds of Cabernet Gernischt dry red wine during storage. Foods. 11:1778. https://doi.org/10.3390/foods11121778.
Zhu L, Li X, Hu X, Wu X, Liu Y, Yang Y, Zang Y, Tang H, Wang C, Xu J. 2021. Quality characteristics and anthocyanin profiles of different Vitis amurensis grape cultivars and hybrids from Chinese germplasm. Molecules. 26(21):6696. https://doi.org/10.3390/molecules26216696.