Comparative Profiling of Aroma Volatiles and Flavonoid Biosynthesis in Crimson Seedless, Red Globe, and Shine Muscat Table Grapes Using GC-IMS and RNA-Seq

TL;DR

This study compares the aroma and flavonoid content of three grape varieties using advanced techniques to identify genes linked to quality traits.

Contribution

The study identifies specific genes correlated with flavonoid biosynthesis and aroma compounds in table grapes.

Findings

Shine Muscat (SM) grapes showed the highest quality traits, including firmness and flavonoid content.

Transcriptome analysis revealed 20 differentially expressed genes in the flavonoid pathway, with ten upregulated in SM.

Eight genes showed strong positive correlations with flavonoid content, offering targets for grape breeding.

Abstract

Table grapes are widely consumed fruits, and their quality is largely determined by aroma, texture, and the accumulation of health-promoting compounds such as flavonoids. In this study, we systematically compared the edible quality and molecular basis of flavonoid biosynthesis among three major table grape cultivars: Crimson Seedless (CRS), Red Globe (RG), and Shine Muscat (SM). An integrated approach combining physicochemical assays, Gas chromatography–ion mobility spectrometry (GC-IMS)-based volatile profiling, and transcriptome sequencing (RNA-seq) was employed. Among the three cultivars, SM exhibited superior fruit quality, characterized by the highest firmness, total soluble solids, ascorbic acid, total phenolic content, and flavonoid content, whereas CRS showed the highest titratable acidity. Volatile compound analysis revealed distinct aroma profiles among the cultivars. SM was enriched in ethanol and 2-methylbutanal, contributing sweet and fruity notes, while CRS contained higher levels of C6 aldehydes, such as hexanal and (E)-2-hexenal, which are associated with green and leafy aromas. RG exhibited a relatively simpler volatile profile with lower overall abundance. Transcriptome analysis identified 20 differentially expressed genes involved in the flavonoid biosynthetic pathway, including PAL, C4H, 4CL, CHI, F3H, F3′H, and DFR, of which ten were significantly upregulated in SM. Eight of these genes showed strong positive correlations (r > 0.8) with flavonoid content. These results provide a comprehensive understanding of the metabolic and genetic mechanisms underlying cultivar-specific quality traits and identify promising candidate genes for breeding table grapes with enhanced flavor and health-promoting properties.

Figures

Full text

1. Introduction

Vitis vinifera L. (grape), a member of the family Vitaceae and genus Vitis, is among the most widely consumed fruit crops worldwide [1]. China is one of the world’s leading grape producers, ranking second globally in cultivation area, with table grapes accounting for approximately 80% of total production [2,3]. Yunnan Province serves as a significant hub for table grape production in China. In particular, Binchuan County in Dali Prefecture benefits from a favorable geographical location and climatic conditions, making it a prominent region for high-quality grape cultivation.

Three major table grape cultivars are predominantly grown in Binchuan County. ‘Red Globe’ is a late-maturing Eurasian variety known for its large, deep-red berries, firm texture, high yield, and substantial economic and nutritional value [4,5]. ‘Crimson Seedless’, also a late-ripening Eurasian variety, is valued for its bright red color, crisp texture, seedlessness, and favorable sugar-acid ratio [6,7,8]. ‘Shine Muscat’ is a mid- to late-season hybrid variety distinguished by its large berry size, high sugar content, firm and crisp flesh, distinct muscat aroma, and strong market demand [9,10]. The pronounced phenotypic differences among these cultivars, particularly in aroma and taste, indicate substantial underlying genetic and metabolic variation. Therefore, a comparative evaluation of these cultivars is essential to elucidate the quality traits and the molecular mechanisms responsible for their superior characteristics.

Aroma is a critical determinant of grape berry quality, and its composition varies significantly among cultivars. Grape aroma is primarily composed of alcohols, aldehydes, esters, ketones, and terpenoids, which may exist in both free and glycosidically bound forms [11]. The composition and accumulation of these volatile compounds are influenced by multiple factors, including fruit maturity, genetic background, and postharvest conditions. Gas chromatography–ion mobility spectrometry (GC–IMS) has emerged as a powerful analytical technique for the rapid, sensitive, and high-resolution detection of volatile compounds, requiring minimal sample preparation [12,13]. Despite these advances, systematic investigations integrating aroma composition with underlying molecular regulation across different table grape cultivars remain limited.

Beyond aroma, grape quality is also shaped by flavonoids, a class of bioactive secondary metabolites that contribute to both functional properties and antioxidant capacity [14,15]. The flavonoid content varies substantially across different grape cultivars, resulting in differences in nutritional value and potential health benefits [2]. Although previous studies have reported cultivar-dependent variations in flavonoid content [16,17], the transcriptomic mechanisms governing these differences, particularly in relation to aroma-related metabolic diversity, remain poorly understood. Ribonucleic acid sequencing (RNA-seq) offers a powerful means of comprehensive transcriptome profiling, allowing the identification of key genes and regulatory pathways underlying flavonoid metabolism and quality differentiation [18]. However, systematic comparative transcriptomic analyses among table grape cultivars remain limited.

Therefore, this study aimed to comprehensively evaluate three major table grape cultivars—‘Shine Muscat’, ‘Crimson Seedless’, and ‘Red Globe’—grown in the Binchuan region by integrating fruit quality assessment, volatile compound profiling, and transcriptomic analysis. The specific objectives were to: (1) identify cultivar-specific aroma markers to support quality evaluation and origin traceability; (2) characterize differentially expressed genes (DEGs) and associated metabolic pathways among the cultivars; and (3) elucidate key regulatory pathways, particularly those involved in phenolic, flavonoid, and anthocyanidin biosynthesis, through Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. The results of this study are expected to enhance our understanding of the genetic mechanisms governing quality traits in table grapes and provide a theoretical basis for breeding cultivars with improved functional and nutritional attributes. This study represents the first integrated application of GC-IMS and RNA-seq to simultaneously compare volatile profiles and flavonoid biosynthesis across multiple major table grape cultivars grown under identical environmental conditions, providing a cultivar-specific resource of aroma markers and flavonoid-related candidate genes for grape quality improvement.

2. Materials and Methods

2.1. Plant Materials

Freshly harvested bunches of three table grape cultivars—Crimson Seedless (CRS), Red Globe (RG), and Shine Muscat (SM)—were collected from a commercial vineyard in Binchuan County (25°49′ N, 100°34′ E; altitude ≈ 1400 m), Dali Prefecture, Yunnan Province, China. Grapes were harvested between 15 July and 10 August 2024, at commercial maturity, defined as a soluble solids content ≥ 16.0 °Brix, titratable acidity ≤ 0.65%, and full color development according to local production standards.

For each cultivar, berries were sampled from at least 50 vines uniformly distributed across the vineyard. Three independent biological replicates were collected on different harvest dates, each from a distinct set of 50 vines. Visually uniform and defect-free bunches were selected. Approximately 20 kg of bunches per replicate were hand-picked, placed into sterile polyethylene bags (50 μm thickness), packed with ice, and transported to the laboratory at the Yunnan Academy of Agricultural Sciences within 6 h, with the internal temperature maintained at 0–4 °C.

Upon arrival, berries were inspected, gently rinsed with deionized water, and air-dried. For each replicate, approximately 100 intact berries from the middle portions of bunches were destemmed, flash-frozen in liquid nitrogen, and stored at −80 °C for GC-IMS and RNA-seq analyses. An additional 100 berries per replicate were stored at 4 °C in sealed containers and used for quality and physicochemical analyses within 24 h.

2.2. Determination of Quality Indicators

2.2.1. Texture Profile Analysis

Berry firmness was evaluated using a texture analyzer (TMS-Touch, FTC, Midland, MI, USA) equipped with a P/75 cylindrical aluminum probe (75 mm diameter). For each cultivar, ten berries from different bunches were randomly selected as biological replicates. Individual berries were subjected to a two-cycle compression test under the following conditions: trigger force of 0.1 N, target strain equivalent to 80% of the original berry height, pre-test speed of 1.0 mm s^−1^, test speed of 1.0 mm s^−1^, and post-test speed of 1.0 mm s^−1^ [19]. Firmness (N) was calculated automatically by the texture analysis software (TL-Pro, Version 1.18-408), and results were expressed as the mean value of ten berries.

2.2.2. Soluble Solids Content

Soluble solids (SS) content was determined using a digital handheld refractometer (PB-10, Sartorius, Göttingen, Germany). Grape juice was obtained by manual squeezing of berries. The instrument was calibrated with distilled water (°Brix = 0) prior to use. A drop of clarified juice was placed on the prism, and °Brix values were recorded at 25 °C. Measurements were performed in triplicate for each biological replicate.

2.2.3. Titratable Acidity

Titratable acidity (TA) was measured by potentiometric titration using the method reported by Kong et al. (2025) [20]. Briefly, 5 g of homogenized grape flesh was diluted with 50 mL of distilled water and titrated with 0.01 M sodium hydroxide (NaOH) (Tianjin Fengchuan Chemical Reagent Technology Co., Ltd., Tianjin, China) to a final pH of 8.3 using a calibrated pH meter (FE28, Mettler-Toledo Instruments, Shanghai, China). The results were expressed as grams of tartaric acid equivalents per kilogram of fresh weight (g kg^−1^), calculated based on the volume of NaOH consumed.

2.2.4. Ascorbic Acid Content

The ascorbic acid (AsA) content was quantified spectrophotometrically following the method of Kampfenkel et al. (1995) [21], with minor modifications. Briefly, 10 g of grape tissue was homogenized in 50 g L^−1^ trichloroacetic acid (TCA) and adjusted to a final volume of 50 mL. After standing for 10 min, the homogenate was filtered. Subsequently, 1 mL of filtrate was mixed sequentially with 1 mL TCA, 1 mL anhydrous ethanol, 0.5 mL phosphoric acid-ethanol solution (0.4%, v:v), 1 mL BP-ethanol (5 g L^−1^), and 0.5 mL FeCl_3_-ethanol (0.3 g L^−1^). The reaction mixture was incubated at 30 °C for 60 min, and absorbance was measured at 534 nm. The AsA content was calculated using the ascorbic acid standard curve and expressed as g kg^−1^ fresh weight.

2.2.5. Total Flavonoid Content

Total flavonoid (TF) content was determined using the sodium nitrite-aluminum nitrate-sodium hydroxide colorimetric method with slight modifications [22]. Briefly, 3 g of grape tissue ground in liquid nitrogen was extracted with 15 mL of 60% ethanol and centrifuged at 10,000× g for 15 min at 4 °C (HR26M, Hunan Hesi, Changsha, China). In a 10 mL reaction system, 2 mL of the collected supernatant was mixed with 0.4 mL of 5% NaNO_2_ and incubated for 6 min, followed by the addition of 0.4 mL of 10% Al(NO_3_)3 for an additional 6 min. Subsequently, 4 mL of 4% NaOH was added, and the volume was adjusted to 10 mL with the extraction solution. After standing for 15 min, absorbance was measured at 510 nm (UV-1100, Shanghai Mapuda, Shanghai, China). TF content was expressed as g kg^−1^ rutin equivalents on a fresh weight basis, as determined from a standard curve prepared with rutin (Beijing Solarbio, Beijing, China, ≥98%) at concentrations of 0−100 mg L^−1^.

2.2.6. Total Phenolic Content

Total phenolic (TP) content was quantified using the Folin–Ciocalteu method [23]. Briefly, 0.5 g of grape tissue was ground and extracted with 15 mL of 55% ethanol, followed by centrifugation at 10,000× g for 15 min at 4 °C. An aliquot of 1.0 mL of the supernatant was mixed with 2.5 mL of Folin–Ciocalteu reagent (Beijing Solarbio Science & Technology Co., Ltd., Beijing, China). Subsequently, 5.0 mL of 7% (w/v) sodium carbonate solution was added. The mixture was vortexed and incubated at 70 °C for 40 min. Absorbance was measured at 765 nm. Gallic acid (Beijing Solarbio Science & Technology Co., Ltd., Beijing, China, ≥98%) was used to generate a standard curve (0–200 mg L^−1^). TP content was expressed as mg gallic acid equivalents (GAE) per kg fresh weight.

2.3. Analysis of Volatile Compounds by GC-IMS

Volatile organic compounds (VOCs) were profiled using a FlavourSpec^®^ flavor analyzer (G.A.S. GmbH, Dortmund, Germany) equipped with an automatic headspace sampler, gas chromatograph, and ion mobility spectrometer with a tritium (^3^H) ionization source.

2.3.1. Sample Preparation

For each replicate, 2.00 ± 0.01 g of frozen grape powder was placed into a 20 mL headspace vial, sealed with a PTFE/silicone septum, and stored at 4 °C for no more than 24 h prior to analysis.

2.3.2. Headspace Sampling

Samples were incubated at 40 °C for 20 min with agitation at 500 rpm to achieve headspace equilibrium. Subsequently, 500 µL of headspace gas was injected in splitless mode (needle temperature: 85 °C).

2.3.3. GC-IMS Conditions

Separation was performed using an MXT^®^-WAX column (30 m × 0.53 mm, 1.0 µm film thickness) operated at 60 °C. High-purity nitrogen (N_2_) was used as both the carrier and drift gas. The carrier gas flow was programmed as follows: 2 mL min^−1^ for 2 min, ramped to 10 mL min^−1^ over 8 min, then increased to 50 mL min^−1^ over 10 min and held for 10 min, resulting in a total run time of 30 min. This flow program was optimized for the separation of a broad range of volatile compounds.

The column effluent was ionized and transferred to the IMS drift tube, which was maintained at 45 °C. The drift gas flow rate was set to 150 mL min^−1^. IMS operating parameters included a drift tube length of 9.8 cm, an electric field strength of 400 V cm^−1^, and an ion gate pulse width of 150 µs.

2.3.4. Data Acquisition, Processing, and Identification

Each sample was analyzed in duplicate. Data acquisition and processing were performed using FlavourSpec Control Software (v1.4, G.A.S.). Compounds were tentatively identified by matching the calculated retention indices (calibrated with C4–C9 ketones) and normalized drift times against the NIST 2023 Mass Spectral Library and the G.A.S. FlavourSpec IMS library (v2.2). Only compounds with match scores above 80% were retained. The relative abundance of each compound was calculated as normalized peak area (compound peak area/total peak area × 100%). Data from duplicate injections were averaged, and results were expressed as mean ± standard deviation (n = 3).

2.4. RNA-Seq Analyses

Transcriptome sequencing was performed to elucidate the molecular mechanisms underlying quality differences among the three cultivars. For each cultivar, three independent biological replicates were analyzed (CRS-1 to CRS-3, RG-1 to RG-3, and SM-1 to SM-3). Each replicate consisted of pooled berries collected from at least five vines. Samples were immediately flash-frozen in liquid nitrogen and stored at –80 °C for further analysis.

Total RNA was extracted from approximately 100 mg of frozen tissue powder using the Omega Plant RNA Kit (Omega, Norwalk, CT, USA) according to the manufacturer’s instructions, with elution in DEPC-treated water. RNA integrity was verified using an Agilent 2100 Bioanalyzer (RIN > 7.5). Libraries were constructed using the NEBNext^®^ Ultra™ RNA Library Prep Kit and sequenced in paired-end 150 bp mode on an Illumina NovaSeq 6000 platform (Beijing Biomarker Technologies Co., Ltd., Beijing, China).

Raw sequencing reads were processed using fastp [24] for quality control with the following parameters: --cut_front --cut_tail --average_qual 20 --length_required 50. The resulting high-quality clean reads were de novo assembled using Trinity (v2.15.1) with default settings [25]. Assembly quality was evaluated based on the read-mapping-back rate.

For functional annotation, assembled unigenes were aligned against the National Center for Biotechnology Information Non-Redundant (NCBI NR), Swiss-Prot, eukaryoticortholog groups (KOG), and KEGG databases using DIAMOND [26] with an e-value threshold of ≤1 × 10^−5^. Gene Ontology (GO) terms were assigned based on the functional annotations derived from the NR database.

Gene expression levels were quantified as fragments per kilobase of transcript per million mapped reads (FPKM) by aligning clean reads to the assembled transcriptome using Bowtie2 [27] and estimating transcript abundance with RSEM [28]. Differentially expressed genes (DEGs) were identified using DESeq2 [29], applying thresholds of |log_2_(fold change)| > 1 and a false discovery rate (FDR)-adjusted p-value < 0.05. Functional enrichment analyses of DEGs for GO terms and KEGG pathways were performed using clusterProfiler [30], with significance set at FDR < 0.05.

2.5. Statistical Analyses

All values are presented as the mean of three independent measurements. Statistical significance among data groups was evaluated by one-way analysis of variance (ANOVA) using SPSS 26.0 (IBM, Armonk, NY, USA). Relationships between several indicators were analyzed using Pearson’s correlation analysis. Differences were considered statistically significant at p < 0.05 and highly significant at p < 0.01. Graphical representations were generated with Origin 2021 (OriginLab, Northampton, MA, USA).

3. Results

3.1. Quality Indicators

Texture profile analysis revealed significant differences in berry firmness among the three grape varieties (Table 1). SM exhibited the highest firmness (35 N), approximately three times that of CRS, resulting in the order: SM > RG > CRS.

Total soluble solids (TSS), primarily consisting of soluble sugars, also varied significantly among cultivars. CRS and SM showed comparable TSS values, both of which were significantly higher (p < 0.05) than RG. In contrast, titratable acidity (TA) differed significantly (p < 0.05) among the varieties, with CRS exhibiting the highest TA, followed by RG and SM.

Ascorbic acid (AsA), an important antioxidant, was highest in SM (131.81 mg kg^−1^), approximately twice that of CRS and thrice that of RG. Similarly, polyphenols and flavonoids, key antioxidants associated with health benefits, also showed significant cultivar-dependent variation. Total polyphenol content was significantly higher (p < 0.05) in CRS (10.64 mg kg^−1^) and SM (10.12 mg kg^−1^) compared to RG (7.47 mg kg^−1^). Notably, SM contained the highest flavonoid content, significantly exceeding both CRS and RG (p < 0.05), indicating its superior potential antioxidant capacity.

3.2. Analysis of Volatile Organic Compounds (VOCs)

A total of 62 VOCs were identified across the three cultivars using GC-IMS, with 32 qualitatively characterized (Table 2), including 8 esters, 8 alcohols, 13 aldehydes, 2 ketones, and 1 furan. While the overall spectral patterns were similar, quantitative differences in signal intensities were evident between cultivars (Figure 1). The compounds within the region highlighted by the yellow box showed significantly higher signal intensities in CRS, whereas the volatile compounds clustered within the red box region were more abundant in SM.

Flavor fingerprint maps (Figure 2) confirmed that all three varieties shared common volatile compounds in regions A and D but at varying intensities. Cultivar-specific volatile markers were identified: compounds in region B were exclusively detected in CRS, whereas those in region E were unique to SM. Propan-2-ol, located in region C, was a shared compound present in both CRS and RG. Overall, RG displayed a simpler volatile profile, characterized by fewer detectable compounds and lower signal intensities than CRS and SM.

Principal component analysis (PCA) and partial least squares-discriminant analysis (PLS-DA) were performed to further elucidate cultivar differences based on the relative contents of the 32 volatile compounds identified. The PCA score plot (Figure 3A) showed clear separation among CRS, RG, and SM, indicating distinct volatile compositions. The first two principal components explained 74.3% (PC1) and 19.1% (PC2) of the total variance, accounting for 93.4% of the cumulative variance and effectively capturing the major variability within the dataset. The PLS-DA model (Figure 3B) demonstrated strong explanatory power for the X-block variables, with R^2^X_1_ and R^2^X_2_ values of 0.841 and 0.135, respectively, confirming pronounced differences in volatile profiles among the three cultivars. Model robustness and predictive reliability were evaluated by a permutation test (n = 200) (Figure 3C). The resulting R^2^ and Q^2^ intercepts were 0.116 and −0.283, respectively. The negative Q^2^ intercept indicates that the model was not overfit and possessed good predictive capability.

Variable importance in projection (VIP) scores obtained from the PLS-DA model were analyzed to identify the volatile compounds contributing most to cultivar discrimination. Seven volatile metabolites met the selection criteria of VIP > 1.0 and p < 0.05 and were thus identified as discriminatory markers (Figure 4A): ethanol, 2-methylbutanal, 2-methylpropanal, 1-hexanol (monomer), hexanal (dimer), (E)-2-hexenal (dimer), and hexanal (monomer). The relative abundance patterns of these markers are visualized in the heatmap (Figure 4B). Notably, 1-hexanol (monomer), hexanal (dimer), (E)-2-hexenal (dimer), and hexanal (monomer) were significantly more abundant in CRS, whereas ethanol, 2-methylbutanal, and 2-methylpropanal were predominantly accumulated in SM. These distinct accumulation patterns are consistent with the cultivar separation observed in the multivariate analyses (Figure 3) and the flavor fingerprint maps (Figure 2).

3.3. Transcriptome Profiling and DEG Analysis

Transcriptome sequencing of the three grape varieties generated a total of 58.51 Gb of high-quality clean data. Each sample yielded at least 5.75 Gb of clean data, with Q30 scores exceeding 93.18% and an average GC content of approximately 47.26% (Table S1), indicating robust sequencing quality suitable for downstream analyses.

Clean reads from each sample were aligned to the reference genome, with mapping rates ranging from 79.05% to 89.27% (Table S2). In total, 31,062 unigenes were assembled, of which 29,851 (96.10%) were successfully annotated using the NR, Swiss-Prot, GO, COG, KOG, Pfam, KEGG, and eggNOG databases (Table S3), reflecting a high level of annotation completeness.

Differentially expressed genes (DEGs) were identified through pairwise comparisons between varieties using a threshold of |Fold Change| ≥ 2 and FDR < 0.01. An overview of the global transcriptomic characteristics is presented in Figure 5, including FPKM distribution (Figure 5A), PCA plot (Figure 5B), DEG statistics (Figure 5C), and a Venn diagram illustrating shared and unique DEGs (Figure 5D). The comparison between SM and RG yielded the highest number of DEGs (2452), whereas the fewest DEGs were detected between CRS and RG (1577) (Table S4). A total of 194 DEGs were common to all three cultivars.

Gene Ontology (GO) functional classification revealed that DEGs from the three pairwise comparisons were distributed across 50 functional subcategories within the three main GO categories: biological process, cellular component, and molecular function (Figure 6A–C). Within the biological process category, encompassing 20 subcategories, metabolic process, cellular process, and single-organism process were the most enriched. In the cellular component category (15 subcategories), the majority of DEGs were assigned to cell, cell part, and membrane. Similarly, under molecular function (15 subcategories), binding and catalytic activity predominated. Notably, the SM vs. RG comparison showed a higher number of DEGs associated with the extracellular region part compared to the other two comparison groups (Figure 6B). In contrast, the CRS vs. RG group exhibited a higher abundance of DEGs associated with cell killing (Figure 6C).

KEGG pathway annotation identified 842 unigenes involved in 19 secondary metabolite biosynthesis pathways, including phenylpropanoid, flavonoid, and terpenoid metabolism (Table 3). Among these, the phenylpropanoid biosynthesis pathway was the most represented, comprising 231 unigenes. Antioxidant-related pathways such as flavonoid biosynthesis and flavone/flavonol biosynthesis contained 102 and 31 unigenes, respectively, whereas only one unigene was annotated for each of the isoflavonoid and anthocyanin biosynthesis pathways.

Pairwise DEG enrichment revealed cultivar-specific metabolic differences. In the SM vs. CRS comparison, DEGs were significantly enriched in pathways such as “fructose and mannose metabolism”, “plant–pathogen interaction”, “flavonoid biosynthesis”, and “carotenoid biosynthesis” (Figure 7A). For SM vs. RG, DEGs were primarily associated with “peroxisome”, “phenylalanine, tyrosine and tryptophan biosynthesis”, and “terpenoid backbone biosynthesis”, with additional substantial enrichment in “flavonoid biosynthesis” and “flavone and flavonol biosynthesis” (Figure 7B). In contrast, fewer enriched pathways were detected in the CRS vs. RG comparison, predominantly “ubiquinone and other terpenoid-quinone biosynthesis” and “alpha-linolenic acid metabolism” (Figure 7C). Notably, the “flavonoid biosynthesis” pathway was significantly more enriched in SM than in CRS and RG, suggesting enhanced flavonoid metabolic activity in the SM cultivar (Table S5).

3.4. Genes Involved in Flavonoid Biosynthesis

Transcriptomic analysis across the three grape varieties identified 20 DEGs encoding key enzymes in the phenylpropanoid, flavonoid, and anthocyanin biosynthesis pathways (Figure 8). In SM, ten genes showed significantly elevated expression levels. These included one phenylalanine ammonia-lyase (PAL) gene (VIT_08s0040g01710), two cinnamate-4-hydroxylase (C4H) genes (VIT_11s0065g00350, VIT_06s0004g08150), two 4-coumarate-CoA ligase (4CL) genes (VIT_06s0061g00450, VIT_16s0039g02040), one chalcone isomerase (CHI) gene (VIT_13s0067g03820), two flavanone 3-hydroxylase (F3H) genes (VIT_18s0001g14310, VIT_04s0023g03370), one flavonoid 3′-hydroxylase (F3′H) gene (VIT_17s0000g07200), and one dihydroflavonol 4-reductase (DFR) gene (VIT_18s0001g12800).

CRS exhibited significant upregulation of five genes, including three PAL-related genes (VIT_13s0019g04460, VIT_16s0039g01360, VIT_16s0039g01300), one chalcone synthase (CHS) gene (VIT_05s0136g00260), and one UDP-glucose: flavonoid glycosyltransferase (UFGT) gene (VIT_16s0039g02230).

In contrast, RG showed higher expression of four genes, comprising two CHS-related genes (VIT_16s0100g00830, VIT_16s0100g00750), one F3H gene (VIT_newGene_1858), and one anthocyanidin synthase/leucoanthocyanidin dioxygenase gene (ANS/LDOX) (VIT_16s0039g02230).

To further explore the relationship between gene expression and metabolite accumulation, correlation analysis was performed between the expression levels of the 20 DEGs and flavonoid content across the three grape varieties (Table 4). Eight genes exhibited a significant positive correlation with flavonoid content (r > 0.8), and all showed markedly higher transcript abundance in SM. Notably, the PAL gene VIT_08s0040g01710 showed a perfect positive correlation with flavonoid content (r = 1.00). In addition, F3′H (VIT_17s0000g07200) and DFR (VIT_18s0001g12800) showed very strong positive correlations (r > 0.95).

Conversely, four genes were significantly and negatively correlated with flavonoid content, including one PAL-related gene (VIT_16s0039g01300, r = −0.96), two CHS-related genes (VIT_16s0100g00870, r = −1.00; VIT_16s0100g00830, r = −0.86), and one (ANS/LDOX)-related gene (VIT_16s0039g02230, r = −0.98). These genes exhibited considerably lower expression levels in SM compared with CRS and RG, suggesting cultivar-specific regulatory differences within the flavonoid metabolic network.

Overall, these findings reveal distinct regulatory patterns of flavonoid biosynthesis among the three varieties, with SM displaying the most pronounced upregulation of genes across multiple branches of the pathway. Notably, the eight genes showing strong positive correlations with flavonoid content (r > 0.8, Table 4 and Figure 8) are all significantly upregulated in SM (Figure 9), consistent with the quality characteristics of higher flavonoid content in SM.

4. Discussion

Our study revealed notable differences in texture, flavor, and basic nutritional composition among Crimson Seedless (CRS), Red Globe (RG), and Shine Muscat (SM). These variations stem from their distinct genotypes, which influence both their market positioning and consumer acceptance.

4.1. Comparison of Edible Quality

The fruit firmness of SM was significantly higher than that of CRS and RG, representing a distinctive sensory attribute that contributes to its crispness and refreshing texture. These findings align with previous descriptions of Muscat grape varieties [9,10]. Grapes with higher firmness generally exhibit enhanced storage and transportability [31]. Firmness is closely linked to cell wall structure, pectin content, and the degree of pectin esterification. In SM, the expression of genes associated with pectin degradation was lower than in the other two varieties. The suppression of cell wall metabolic processes during storage helps maintain the superior firmness of SM grapes [20,32].

The sugar-acid ratio is a critical determinant of acceptability in table grapes [33]. Although both SM and CRS exhibited relatively high total soluble solids (TSS), CRS showed significantly higher titratable acidity (TA). Consequently, SM is characterized by a notably sweet flavor profile [34], whereas CRS presents a more pronounced acidic taste [7]. RG displayed moderate levels of both TSS and TA, resulting in a milder, more neutral flavor [5,33]. These distinct sensory characteristics underpin the cultivar-specific appeal in the consumer market.

Aroma is another key sensory attribute influencing grape quality. The GC-IMS results not only discriminated among the three cultivars but also revealed the chemical basis of their distinct aroma profiles. Twelve volatile compounds were commonly detected across all three cultivars, suggesting a shared aromatic foundation, consistent with previous studies in other grape varieties [35,36]. These shared compounds—including diethyl acetal, butan-2-one, propanal, acetone, hexanal (monomer and dimer), ethanol, (E)-2-hexenal (monomer and dimer), 1-propanol, and ethyl acetate (monomer and dimer)—likely constitute the backbone of the fundamental aroma profile.

SM was enriched in ethanol and 2-methylbutanal, associated with sweet, fruity, and wine-like notes [37], reflecting its characteristic muscat aroma. In contrast, CRS contained higher levels of hexanal and (E)-2-hexenal (C6 aldehydes), imparting green and leafy notes typical of many grape cultivars [38,39].

4.2. Genes Influencing Flavonoid Biosynthesis

Grapes are a natural source of diverse bioactive compounds, with antioxidant activity being among their most notable biological functions [40]. Flavonoids, as key antioxidant constituents, exhibit substantial variation in content among grape cultivars [41]. Flavonoid biosynthesis is closely associated with the phenylpropanoid and anthocyanin metabolic pathways [18,42]. In this study, SM exhibited the highest levels of total phenolics, flavonoids, and ascorbic acid (AsA), indicating superior antioxidant potential, consistent with prior reports [10,37].

Phenylpropanoid metabolism channels carbon toward either flavonoid or lignin biosynthesis, and the partitioning of this flux is tightly regulated at the transcriptional level. The PAL gene family, which catalyzes the first committed step of the pathway, plays a pivotal role in this allocation [15,42,43]. The PAL-related gene VIT_08s0040g01710 was significantly upregulated in SM and exhibited a strong positive correlation with flavonoid content (r = 1.00), whereas another PAL gene (VIT_16s0039g01300) showed a negative correlation with flavonoid content (r = −0.96). This opposing expression pattern suggests functional divergence among grape PAL paralogs, with some isoenzymes directing carbon toward flavonoids and others toward competing branches such as lignification. A direct precedent exists in soybean, where under shade stress, GmPAL expression is modulated to favor anthocyanin and isoflavonoid accumulation at the expense of lignin [44]. We propose that a similar carbon-flux reallocation operates in grape, and that the high PAL-4/low PAL-3 expression signature in SM shifts metabolic flow toward flavonoid biosynthesis, contributing to its elevated phenolic content, it should be further tested through functional validation via genetic manipulation and metabolic flux analysis.

Furthermore, while chalcone synthase (CHS) catalyzes the first committed step in flavonoid biosynthesis, its transcript levels did not show a positive correlation with flavonoid accumulation in this study. Notably, CHS-3 (VIT_05s0136g00260) was consistently and highly expressed across all three cultivars yet showed no correlation with flavonoid content (r = −0.02), indicating that transcript abundance alone does not linearly reflect metabolic flux. In contrast, chalcone isomerase (CHI2), which catalyzes the subsequent stabilization of the chalcone backbone, showed a strong positive correlation (r = 0.92) and high expression in SM. This indicates that downstream steps, particularly the CHI-mediated reaction, may be more critical in directing metabolic flux toward flavonoid end-products under these specific conditions [45,46].

Further downstream, two critical genes, F3′H (VIT_17s0000g07200) and DFR (VIT_18s0001g12800), were significantly upregulated in SM and showed strong positive correlations with flavonoid content (r > 0.95), highlighting their pivotal roles in promoting flavonoid accumulation. F3′H (flavonoid 3′-hydroxylase) catalyzes the hydroxylation of the B-ring of flavonoids, a key step that dictates the diversification of flavonoid subgroups [47]. Upregulation of F3′H transcription has been shown to positively regulate anthocyanin accumulation induced by phosphorus starvation in Arabidopsis thaliana [48]. DFR (dihydroflavonol 4-reductase), which catalyzes the first committed step in anthocyanin biosynthesis by reducing dihydroflavonols to leucoanthocyanidins, is also a key determinant of anthocyanin production. Overexpression of the wheat TaDFR gene increases anthocyanin accumulation in an Arabidopsis dfr mutant [49]. In pepper, nine members of the DFR gene family were identified, among which the expression level of CaDFR5 showed a close correlation with the patterns of anthocyanin accumulation [50].

In the present study, coordinated upregulation of PAL-4, CHI2, F3′H, and DFR in SM provides a molecular explanation for the enhanced accumulation of flavonoids and the superior antioxidant capacity observed in this cultivar. Collectively, these data reveal that SM’s high flavonoid phenotype is not attributable to a single “master” regulator, but rather to a concerted, pathway-wide shift in transcriptional output—from upstream carbon allocation (PAL) through intermediate flux control (CHI) to terminal anthocyanin biosynthesis (F3′H, DFR) (Figure 9).

5. Conclusions

This study comprehensively compared edible quality and the expression of flavonoid biosynthesis-related genes in three grape cultivars—Crimson Seedless (CRS), Red Globe (RG), and Shine Muscat (SM)—using integrated physicochemical analyses, volatile profiling, and transcriptome sequencing. Although the three cultivars shared a core set of volatile compounds, each displayed a distinct volatile profile, allowing clear discrimination. Nutritionally, SM possessed significantly higher levels of total soluble solids, ascorbic acid, total phenolics, and flavonoids.

At the molecular level, transcriptome analysis revealed significant activation of the flavonoid biosynthetic pathway in SM. Ten key genes encoding enzymes such as PAL, C4H, F3′H, and DFR were markedly upregulated relative to the other cultivars, with eight showing strong positive correlations with flavonoid content. This provides a genetic explanation for the enhanced flavonoid accumulation and antioxidant capacity observed in SM.

Collectively, these results elucidate the metabolic and molecular underpinnings of SM’s superior quality and identify promising candidate genes for improving flavor and nutritional traits in table grape breeding. In particular, the eight flavonoid-correlated genes—including PAL-4, CHI2, F3′H, and DFR—represent valuable molecular targets for marker-assisted selection and genetic improvement aimed at enhancing flavonoid content and antioxidant capacity.

Nevertheless, this study has several limitations. First, our analyses focused on transcriptional regulation of flavonoid biosynthesis; the molecular mechanisms underlying cultivar-specific aroma profiles, while characterized at the volatile level, were not explored transcriptomically. Second, all samples were collected from a single growing season and geographical location, precluding separation of environmental effects from genetic influences. Third, the observed correlations between gene expression and flavonoid content, though strongly suggestive, do not establish causality.

Future studies should include functional validation of candidate genes (e.g., PAL-3, PAL-4, F3′H, and DFR) through genetic manipulation to confirm their causal roles in flavonoid accumulation. Sampling should be expanded across multiple developmental stages and growing seasons to capture the dynamic regulation of quality traits under varying environmental conditions. Finally, integrating metabolomic and transcriptomic analyses—particularly for volatile organic compounds—will be essential to fully elucidate the genetic basis of both flavor and nutritional quality in these cultivars.