Abstract
A targeted UPLC-MS/MS method was developed for the comprehensive characterization of 58 anthocyanins and their derivatives (A&D) in Marselan wines. Significant differences among vintages were observed for 12 compounds, with acetylated anthocyanins and pyranoanthocyanins emerging as potential vintage markers. Multi-vintage analysis revealed a pronounced decline in total A&D during aging, accompanied by an increasing proportion of pyranoanthocyanins. Geographical comparison indicated that the A&D profiles clearly discriminated wines from distinct regions (Ning xia, Hebei, Xinjiang), whereas subregional distinction within Ningxia was limited. Multivariate models, particularly SVM, achieved high accuracy at the regional level (> 96%), confirming the potential of A&D fingerprints as robust markers for vintage differentiation and regional traceability of Marselan wines.
Keywords: Marselan wine, Anthocyanins and their derivatives, UPLC-MS/MS, Vintages, Origin
Graphical abstract
Highlights
-
•
UPLC-MS/MS method development for the analysis of 58 anthocyanins and derivatives.
-
•
Comprehensive A&D fingerprinting of Marselan wines across vintages and origins.
-
•
Acetylated anthocyanins and pyranoanthocyanins serve as potential markers for wine vintage
-
•
A&D effectively traces regions but shows limited sub-region discrimination.
1. Introduction
Anthocyanins are a class of phenolic compounds present in the grape skin that are released into wine during maceration and fermentation (Freitas & Mateus, 2011). The structural and compositional diversity of anthocyanins and their derivatives (A&D), together with their intrinsic instability arising from the characteristic 2-phenyl-benzylpyrilium structure, strongly affect wine color development and quality during storage (Dangles & Fenger, 2018; Zhao et al., 2022), and additionally provide information regarding grape variety, geographical origin, vintage, and winemaking practices. Therefore, comprehensive characterization of the structures and distribution profiles of A&D in wine is of great significance, particularly for authenticity verification and quality rating.
The A&D present in red wines can be broadly classified into basic anthocyanins, acylated anthocyanins, pyranoanthocyanins, and polymeric anthocyanins (Han et al., 2019). As the most common form, basic anthocyanins consist of six aglycones conjugated to mono- or disaccharides via glycosidic bonds, including pelargonidin-3-O-glucoside, cyanidin-3-O-glucoside, delphinidin-3-O-glucoside, peonidin-3-O-glucoside, petunidin-3-O-glucoside and malvidin-3-O-glucoside, the latter being the most abundant. (He et al., 2012a; Liu et al., 2024). Acylated anthocyanins, formed through the esterification of anthocyanins with aromatic or aliphatic acids, generally exhibit enhanced stability compared to non-acylated forms (Leonarski et al., 2023; Pascual-Teresa & Sanchez-Ballesta, 2008). Pyranoanthocyanins are formed during alcoholic fermentation through reactions between native anthocyanins and fermentation-derived compounds (e.g., acetaldehyde, pyruvic acid, and vinylphenols), which enhances their resistance to oxidation (He et al., 2010). They can be further divided into several structural types, including vitisin-, methyl-, hydroxyphenyl-, flavanols-, portisins-, dimeric-, and oxovitisin-type pyranoanthocyanins (Rentzsch et al., 2007), and tend to accumulate in aged wines, contributing to orange or yellow hues (He et al., 2010). Polymeric anthocyanins, which remain relatively less studied, are formed through covalent bonding between anthocyanins and flavan-3-ols, and exhibit greater stability and more yellowish coloration relative to monomeric anthocyanins (He et al., 2012b). A comprehensive understanding of the types, structures, and biochemical transformations of A&D is critical for explaining color variation in red wines across varieties, vintages, geographical origins, and enological processes.
The structural complexity of A&D in wine is not only closely associated with grape varietal differences but also dynamically influenced by interactions with other constituents during fermentation and aging, which renders their separation, structural identification, and quantitative determination particularly challenging. Various techniques have been widely applied for identifying A&D, including vibrational spectroscopy (Ferrer-Gallego et al., 2022), nuclear magnetic resonance spectroscopy (NMR), and mass spectrometry (MS) allowing for their qualitative and quantitative determination in wine (Ongkowijoyo et al., 2018; Zhang et al., 2021). However, each of these analytical techniques has its own advantages and limitations. Vibrational spectroscopy, including Near-infrared (NIR), Fourier transform infrared (FTIR), and Raman spectroscopy, is generally employed to construct predictive models for the total anthocyanin content, rather than to identify individual A&D. However, variations in spectral ranges and modeling approaches markedly influence the final outcome, resulting in limited comparability across studies (Cozzolino, 2025). NMR spectroscopy is primarily applied for the structural elucidation of pure compounds, whereas its application to the qualitative and quantitative analysis of A&D in complex wine matrices remains challenging (Mateus et al., 2002). Liu et al. (2024) investigated the influence of additives on wine color by utilizing high-performance liquid chromatography (HPLC) with diode-array detection (DAD) to identify monomeric anthocyanins. However, polymeric anthocyanins and isomers cannot be analyzed by HPLC-DAD. HPLC-MS is a complementary technique that overcomes this limitation and has consequently become the method of choice for analyzing anthocyanin profiles (García-Beneytez et al., 2003). More recently, a targeted HPLC coupled with tandem mass spectrometry (MS/MS) method capable of quantifying 37 anthocyanin derivatives was developed to discriminate wines at different aging stages (Zhang et al., 2020). Furthermore, Ultra-performance liquid chromatograph (UPLC) coupled with MS/MS offers improved selectivity for characteristic fragments and enables the identification of dimers in complex matrices (Deshaies et al., 2022). However, some low-abundance, highly polymerized, and structurally complex anthocyanin derivatives in wine still pose challenges in detection, recognition, and structural identification. Accurate determination of A&D is crucial not only for characterizing varietal wine attributes, verifying authenticity, and elucidating aging-related transformations, but also for elucidating the mechanisms underlying wine color stability and antioxidant activity. Consequently, UPLC-MS/MS based analysis of A&D in wine offers considerable potential for wine quality evaluation and authentication applications.
The eastern foothills of the Helan Mountain in NX are represent the second-largest wine-producing region in China. Geographically, the area is situated between the alluvial plain of the Yellow River and the alluvial fan of the Helan Mountain (latitude 34°-39°, longitude 104°-107°), characterized by a temperate climate with pronounced diurnal temperature variations, abundant sunshine, and a semi-arid environment, all of which provide favorable conditions for high-quality wine production (Yao et al., 2024; Zhao et al., 2023). In recent years, Marselan has emerged as the third most widely cultivated grape variety in this region, following Cabernet Sauvignon and Merlot. Marselan wines, a hybrid of Cabernet Sauvignon and Grenache, exhibit a deep purple hue and a rich diversity and abundance of A&D. However, research on the varietal, vintage, and regional characteristics of Marselan in China remains limited, and studies focusing on the types, structures, and concentrations of A&D are particularly scarce. Therefore, it is essential to establish a simple and efficient analytical strategy for the characterization of these compounds in wine. This study aims to develop a targeted UPLC-MS/MS method and database for the determination of A&D in wines, reveal the vintage-dependent evolution of anthocyanin profiles, and distinguish wines from different geographical regions using chemometric analysis based on their anthocyanin composition, thereby elucidating the geographical and vintage specificities of Marselan wines.
2. Materials and methods
2.1. Wine samples
A total of 64 red wines produced from Marselan grapes (Vitis vinifera L.) were analyzed in this study. Among them, 52 samples were collected from five sub-regions of the eastern foothills of the Helan Mountain across six vintages (2017–2022), including Helan (HL, n = 9), Yinchuan (YC, n = 6), Yongning (YN, n = 14), Qingtongxia (QTX, n = 19), and Hongsibu (HSB, n = 4). Additional samples were obtained from Huailai, Hebei (HuL, n = 5; vintages 2017–2020) and the Yili River valley, Xinjiang (YL, n = 7; vintage 2021). To further investigate the varietal differences of A&D, wines produced from Cabernet Sauvignon (n = 2), Yan 73 (n = 2), Tintorera (n = 2), Vitis davidii (n = 3), and Vitis amurensis (n = 3) were also collected. All physicochemical parameters of the wine samples complied with the National Standard of the People's Republic of China (GB/T 15037-2006, Wine, 2006).
2.2. Chemicals and standards
Methanol, formic acid, and acetonitrile (HPLC-grade) were and purchased from Dikma Technologies Co., Ltd. (Beijing, China). Twenty-six anthocyanin standards (purity >95%) were used, including delphinidin (Dp), cyanidin (Cy), pelargonidin (Pg), peonidin (Pn), petunidin (Pt), and malvidin (Mv), along with their derivatives: Dp-3-O-glucoside (Dp-3-O-glc), Dp-3,5-O-diglucoside (Dp-3,5-O-diglc), Dp-3-O-rutinoside (Dp-3-O-rutin), Dp-3-O-galactoside (Dp-3-O-galact); Cy-3-O-glucoside (Cy-3-O-glc), Cy-3,5-O-diglucoside (Cy-3,5-O-diglc), Cy-3-O-rutinoside (Cy-3-O-rutin), Cy-3-O-galactoside (Cy-3-O-galact); Pg-3-O-glucoside (Pg-3-O-glc), Pg-3,5-O-diglucoside (Pg-3,5-O-diglc), Pg-3-O-rutinoside (Pg-3-O-rutin); Pn-3-O-glucoside (Pn-3-O-glc), Pn-3,5-O-diglucoside (Pn-3,5-O-diglc), Pn-3-O-rutinoside (Pn-3-O-rutin), Pn-3-O-galactoside (Pn-3-O-galact); Pt-3-O-glucoside (Pt-3-O-glc), Pt-3,5-O-diglucoside (Pt-3,5-O-diglc); and Mv-3-O-glucoside (Mv-3-O-glc), Mv-3,5-O-diglucoside (Mv-3,5-O-diglc), Mv-3-O-galactoside (Mv-3-O-galact). Detailed information for all standards is provided in Table S1.
2.3. UPLC-MS/MS analysis
2.3.1. Apparatus
Analysis was performed using a Shimadzu 30 A ultra-high-performance liquid chromatography (UPLC, Japan) coupled with a SCIEX QTRAP 4500 triple quadrupole mass spectrometer (AB Sciex, USA). Ultra-pure water was prepared using a Millipore IQ7000 system (Millipore, USA).
2.3.2. Sample preparation
A 1 mL aliquot of sample was diluted 20-fold with the initial mobile phase and vortexed thoroughly. It was transferred to a chromatographic vial for direct injection and analysis. To minimize the loss of target compounds, membrane filtration was omitted.
2.3.3. Standard solution preparation
Approximately 1 mg of each standard was accurately weighed into a 10 mL volumetric flask, dissolved in methanol containing 1% formic acid, and diluted to prepare stock solutions with a concentration of 100 mg/L. All stock solutions were transferred into amber vials and stored at −20 °C until analysis.
2.3.4. Matrix solution preparation
An aliquot of 500 μL of each standard solution was transferred into a 10 mL volumetric flask and diluted to volume with methanol to obtain 5 mg/L intermediate standard for the MS parameter optimization. According to experimental requirements, the intermediate standard solutions were further diluted with 1% aqueous formic acid containing 5% methanol to prepare mixed working standard solutions of graded concentrations, which were freshly prepared before use.
2.3.5. UPLC condition
Chromatographic separation was performed on an ACQUITY UPLC BEH C18 column (2.1 mm × 100 mm, 1.7 μm; Waters, USA). The column temperature was maintained at 30 °C, the injection volume was 5 μL, and the flow rate was set to 0.25 mL/min. The mobile phase consisted of (A) ultrapure water containing 8% formic acid and (B) acetonitrile containing 8% formic acid. Two gradient elution programs were employed. Gradient program I: 0–3 min, 0%–0.5% B; 3–10 min, 0.5%–5% B; 10–20 min, 5%–25% B; 20–21 min, 25%–0.5% B; 21–25 min, 0.5%–0% B. Gradient program II: 0–1.5 min, 0%–3% B; 1.5–10 min, 3%–35% B; 10–11 min, 35%–100% B; 11–12 min, 100%–3% B; 12–15 min, 3%–0% B.
2.3.6. MS conditions
The SCIEX QTRAP 4500 triple quadrupole mass spectrometer, equipped with an electrospray ionization (ESI) source, Turbo V ion source, and Analyst 1.5 data processing system, was employed for analysis. The ionization mode was set to positive electrospray ionization (ESI+), and two detection modes were utilized: multiple reaction monitoring (MRM) for QQQ scanning and neutral loss scanning. For neutral loss (NL) scanning, the mass range of m/z 400–900 was monitored for neutral fragment losses of m/z 162, 204, 308, 324, and 338, with a collision energy of 25 V. Key parameters included an ion source temperature of 550 °C, an electrospray voltage of 5500 V, a curtain gas (CUR) pressure of 50 psi, a nebulizer gas (GS1) pressure of 50 psi, and an auxiliary gas (GS2) pressure of 60 psi. The declustering potential (DP) and collision energy (CE) were optimized for each MRM ion pair.
2.4. Data analysis
Significant differences among groups were assessed using t-tests and one-way analysis of variance (ANOVA), followed by Duncan's multiple range test (p < 0.05) with SPSS 24.0 (IBM, Armonk, NY, USA). Principal component analysis (PCA) and orthogonal partial least squares-discriminant analysis (OPLS-DA) were conducted with SIMCA 14.1 (Umetrics, Umeå, Sweden) to visualize the variance structure and identify discriminant variables. Support vector machine (SVM) models were constructed in Python (Version 3.6) using the sklearn package for statistical modeling to distinguish vintages and geographical origins.
To minimize the potential impact of unequal sample sizes among sub-regions and vintages, several strategies were implemented. All variables were autoscaled (mean-centered and unit variance scaled) prior to multivariate analysis to ensure comparability. Unsupervised principal component analysis (PCA) was first performed to explore the intrinsic data structure without class bias. Supervised models, including orthogonal partial least squares discriminant analysis (OPLS-DA) and support vector machine (SVM), were applied to evaluate classification performance. The dataset was randomly split into a training set (70%) and a test set (30%) for model training. For OPLS-DA, 7-fold cross-validation was used to evaluate model stability and avoid overfitting. In addition, permutation tests (n = 200) were conducted to assess model reliability. The SVM model was constructed using a RBF kernel, and k-fold cross-validation was employed to assess model stability and generalization ability.
3. Results and discussions
3.1. Analytical method development for a&D
UPLC-MS/MS was employed to analyze 26 A&D standards, including 6 anthocyanidin monomers, 14 anthocyanidin monoglucosides, and 6 anthocyanidin diglucosides. Analyses were performed in ESI+ mode due to the preferential formation of positive ions. The M+ ion was selected as the precursor ion, and MS/MS parameters were optimized to obtain characteristic product ions. According to the criteria outlined in Commission Decision 2002/657/EC, two ion pairs with high signal intensity and structurally meaningful fragmentation patterns were selected for each analyte, including one for quantification and one for confirmation. The MS1 and MS2 spectra are shown in Figs. S1 and S2, respectively. The optimized MS parameters for 26 anthocyanin standards are summarized in Table 1.
Table 1.
Mass spectrometric parameters of 26 anthocyanin standards.
| Number | Compound | Precursor ion (m/z) | Product ion (m/z) |
DP/V | CE/V | Retention time/min |
|---|---|---|---|---|---|---|
| 1I | Delphinidin-3,5-O-diglucoside | 627.2* | 302.8* | 161 | 37 | 11.15 |
| 627.2 | 465.2 | 161 | 25 | |||
| 2I | Cyanidin-3,5-O-diglucoside | 611.3* | 287.0* | 116 | 39 | 12.86 |
| 611.3 | 449.0 | 116 | 27 | |||
| 3I | Petunidin-3,5-O-diglucoside | 641.3* | 478.8* | 120 | 27 | 13.49 |
| 641.3 | 316.9 | 120 | 39 | |||
| 4I | Pelargonidin-3,5-O-diglucoside | 595.1* | 270.9* | 126 | 39 | 13.72 |
| 595.1 | 433.1 | 126 | 25 | |||
| 5I | Peonidin-3,5-O-diglucoside | 625.2* | 301.0* | 126 | 41 | 14.26 |
| 625.2 | 463.2 | 126 | 27 | |||
| 6I | Malvidin-3,5-O-diglucoside | 655.2* | 331.0* | 151 | 41 | 14.53 |
| 655.2 | 493.4 | 151 | 27 | |||
| 7I | Delphinidin-3-O-galactoside | 465.2* | 303.0* | 100 | 29 | 12.13 |
| 465.2 | 229.0 | 100 | 75 | |||
| 8I | Cyanidin-3-O-galactoside | 449.1* | 286.9* | 90 | 26 | 13.41 |
| 449.1 | 213.3 | 90 | 71 | |||
| 9I | Peonidin-3-O-galactoside | 463.1* | 301.0* | 100 | 26 | 14.99 |
| 463.1 | 286.1 | 100 | 55 | |||
| 10I | Malvidin-3-O-galactoside | 493.0* | 330.9* | 111 | 27 | 15.42 |
| 493.0 | 315.0 | 111 | 61 | |||
| 11I | Delphinidin | 303.0* | 229.0* | 166 | 43 | 12.94 |
| 303.0 | 256.9 | 166 | 37 | |||
| 12I | Cyanidin | 287.0* | 136.9* | 171 | 42 | 13.98 |
| 287.0 | 213.1 | 171 | 42 | |||
| 13I | Petunidin | 317.1* | 261.0* | 108 | 27 | 14.60 |
| 317.1 | 302.0 | 160 | 35 | |||
| 14I | Pelargonidin | 271.0* | 120.9* | 141 | 39 | 14.85 |
| 271.0 | 197.0 | 141 | 37 | |||
| 15I | Peonidin | 301.1* | 285.8* | 91 | 33 | 15.41 |
| 301.1 | 258.1 | 91 | 39 | |||
| 16I | Malvidin | 331.1* | 298.9* | 110 | 23 | 15.44 |
| 331.1 | 215.2 | 110 | 33 | |||
| 17I | Delphinidin-3-O-glucoside | 465.2* | 303.0* | 101 | 25 | 12.95 |
| 465.2 | 229.1 | 101 | 73 | |||
| 18I | Cyanidin-3-O-glucoside | 449.0* | 287.1* | 96 | 25 | 13.99 |
| 449.0 | 213.1 | 96 | 70 | |||
| 19I | Petunidin-3-O-glucoside | 479.2* | 317.1* | 91 | 23 | 14.61 |
| 479.2 | 302.0 | 91 | 55 | |||
| 20I | Pelargonidin-3-O-glucoside | 433.0* | 271.2* | 96 | 25 | 14.86 |
| 433.0 | 121.0 | 96 | 75 | |||
| 21I | Peonidin-3-O-glucoside | 463.1* | 301.0* | 96 | 25 | 15.42 |
| 463.1 | 286.1 | 96 | 53 | |||
| 22I | Malvidin-3-O-glucoside | 493.0* | 331.1* | 111 | 27 | 15.78 |
| 493.0 | 315.0 | 111 | 61 | |||
| 23I | Delphinidin-3-O-rutinoside | 611.3* | 302.8* | 113 | 36 | 13.52 |
| 611.3 | 465.2 | 113 | 28 | |||
| 24I | Cyanidin-3-O-rutinoside | 595.1* | 287.1* | 130 | 35 | 14.42 |
| 595.1 | 449.0 | 130 | 28 | |||
| 25I | Pelargonidin-3-O-rutinoside | 579.2* | 271.1* | 100 | 35 | 15.20 |
| 579.2 | 301.1 | 80 | 18 | |||
| 26I | Peonidin-3-O-rutinoside | 609.3* | 301.1* | 130 | 36 | 15.62 |
| 609.3 | 463.2 | 130 | 28 |
To further expand the mass spectrometric information of A&D in wine, particularly in the absence of reference standards, an MS method was established based on ion pair data reported in the literature together with the spectral information of the 26 standards described above (He et al., 2012a; He et al., 2012b). Wine samples diluted 20-fold were analyzed by UPLC-MS/MS under gradient program I using MRM and NL scanning modes. For NL scans of m/z 162, 204, 308, 324, and 338, the scan range was set to m/z 400–900 with a collision energy of 25 V as the initial acquisition condition. DP and CE were further optimized according to the actual instrumental responses to improve analyte separation while maintaining satisfactory peak shape. For quantification, semi-quantification was achieved using calibration curves established from the corresponding monoglucoside standard of each anthocyanidin class.
Finally, 49 compounds were successfully identified under gradient program I, and effective separation was achieved for the following isomeric pairs of anthocyanins: Cy-3-O-glu-4-vinylcatechol and Dp-3-O-glu-4-vinylphenol, Peonidin-3-O-glucoside and Peonidin-3-O-galactoside, Malvidin-3-O-glucoside and Malvidin-3-O-galactoside, Delphinidin-3-O-glucoside and Delphinidin-3-O-galactoside, Pn-3-O-glu-4-vinylcatechol, Pt-3-O-glu-4-vinylphenol, and Cy-3-O-glu-4-vinylguaiacol (Fig. S3). Furthermore, gradient program II was applied as a supplementary approach to improve the accuracy of pyranoanthocyanin identification, under which these compounds were specifically analyzed. Consequently, a total of 58 A&D were identified in 64 Marselan dry red wine samples, and the corresponding optimized mass spectrometric parameters are summarized in Table 2. Notably, for certain compounds, only one ion pair could be assigned from the fragmentation pattern, which necessitated further confirmation.
Table 2.
Mass spectrometric parameters of A&D identified in Marselan dry red wines.
| Number | Compound | Precursor ion (m/z) | Product ion (m/z) |
DP/V | CE/V | Retention time/min |
|---|---|---|---|---|---|---|
| 2I | Petunidin-3,5-O-diglucoside | 641.3* | 478.8* | 120 | 27 | 13.49 |
| 641.3 | 316.9 | 120 | 39 | |||
| 5I | Peonidin-3,5-O-diglucoside | 625.2* | 301.0* | 126 | 41 | 14.26 |
| 625.2 | 463.2 | 126 | 27 | |||
| 6I | Malvidin-3,5-O-diglucoside | 655.2* | 331.0* | 151 | 41 | 14.53 |
| 655.2 | 493.4 | 151 | 27 | |||
| 11I | Delphinidin | 303.0* | 229.0* | 166 | 43 | 12.94 |
| 303.0 | 256.9 | 166 | 37 | |||
| 12I | Cyanidin | 287.0* | 136.9* | 171 | 42 | 13.98 |
| 287.0 | 213.1 | 171 | 42 | |||
| 13I | Petunidin | 317.1* | 261.0* | 108 | 27 | 14.60 |
| 317.1 | 302.0 | 160 | 35 | |||
| 15I | Peonidin | 301.1* | 285.8* | 91 | 33 | 15.41 |
| 301.1 | 258.1 | 91 | 39 | |||
| 16I | Malvidin | 331.1* | 298.9* | 110 | 23 | 15.44 |
| 331.1 | 215.2 | 110 | 33 | |||
| 17I | Delphinidin-3-O-glucoside | 465.2* | 303.0* | 101 | 25 | 12.95 |
| 465.2 | 229.1 | 101 | 73 | |||
| 18I | Cyanidin-3-O-glucoside | 449.0* | 287.1* | 96 | 25 | 13.99 |
| 449.0 | 213.1 | 96 | 70 | |||
| 19I | Petunidin-3-O-glucoside | 479.2* | 317.1* | 91 | 23 | 14.61 |
| 479.2 | 302.0 | 91 | 55 | |||
| 20I | Pelargonidin-3-O-glucoside | 433.0* | 271.2* | 96 | 25 | 14.86 |
| 433.0 | 121.0 | 96 | 75 | |||
| 21I | Peonidin-3-O-glucoside | 463.1* | 301.0* | 96 | 25 | 15.42 |
| 463.1 | 286.1 | 96 | 53 | |||
| 22I | Malvidin-3-O-glucoside | 493.0* | 331.1* | 111 | 27 | 15.78 |
| 493.0 | 315.0 | 111 | 61 | |||
| 27I | Dp-3-O-acetylglc | 507.0 | 303.0 | 85 | 40 | 16.12 |
| 507.0 | 229.0 | 85 | 60 | |||
| 28I | Cy-3-O-acetylglc | 491.0 | 287.0 | 145 | 40 | 16.94 |
| 491.0 | 137.0 | 145 | 70 | |||
| 29I | Pt-3-O-acetylglc | 521.0 | 317.0 | 103 | 37 | 17.26 |
| 521.0 | 301.8 | 103 | 40 | |||
| 30I | Pg-3-O-acetylglc | 475.0 | 271.0 | 120 | 35 | 17.71 |
| 475.0 | 121.0 | 120 | 70 | |||
| 31I | Pn-3-O-acetylglc | 505.0 | 301.0 | 115 | 24 | 18.10 |
| 505.0 | 286.0 | 115 | 50 | |||
| 32I | Mv-3-O-acetylglc | 535.0 | 331.0 | 115 | 33 | 18.30 |
| 535.0 | 287.0 | 115 | 46 | |||
| 33I | Dp-3-O-cafglc | 627.0 | 465.0 | 105 | 23 | 16.76 |
| 627.0 | 303.0 | 105 | 45 | |||
| 34I | Pt-3-O-cafglc | 641.0 | 317.0 | 105 | 45 | 17.68 |
| 641.0 | 301.8 | 105 | 55 | |||
| 35I | Pn-3-O-cafglc | 625.0 | 301.0 | 125 | 38 | 18.49 |
| 625.0 | 286.0 | 125 | 22 | |||
| 36I | Mv-3-O-cafglc | 655.1 | 331.0 | 115 | 50 | 18.62 |
| 655.1 | 287.0 | 115 | 60 | |||
| 37I | Dp-3-O-coumglc (cis) | 611.0 | 303.0 | 105 | 53 | 17.48 |
| 611.0 | 229.0 | 105 | 75 | |||
| 38I | Dp-3-O-coumglc (trans) | 611.0 | 303.0 | 105 | 53 | 17.88 |
| 611.0 | 229.0 | 105 | 75 | |||
| 39I | Cy-3-O-coumglc (cis) | 595.0 | 287.0 | 103 | 47 | 18.24 |
| 595.0 | 137.0 | 103 | 69 | |||
| 40I | Cy-3-O-coumglc (trans) | 595.0 | 287.0 | 103 | 47 | 18.63 |
| 595.0 | 137.0 | 103 | 69 | |||
| 41I | Pt-3-O-coumglc (cis) | 625.0 | 317.0 | 105 | 42 | 18.42 |
| 625.0 | 301.8 | 105 | 45 | |||
| 42I | Pt-3-O-coumglc (trans) | 625.0 | 317.0 | 105 | 42 | 18.84 |
| 625.0 | 301.8 | 105 | 45 | |||
| 43I | Pg-3-O-coumglc (trans) | 579.0 | 271.0 | 120 | 35 | 19.34 |
| 579.0 | 121.0 | 120 | 52 | |||
| 44I | Pn-3-O-coumglc (cis) | 609.0 | 301.0 | 115 | 48 | 19.25 |
| 609.0 | 286.0 | 115 | 50 | |||
| 45I | Pn-3-O-coumglc (trans) | 609.0 | 301.0 | 115 | 48 | 19.67 |
| 609.0 | 286.0 | 115 | 50 | |||
| 46I | Mv-3-O-coumglc (cis) | 639.0 | 331.0 | 135 | 29 | 19.40 |
| 639.0 | 287.0 | 135 | 35 | |||
| 47I | Mv-3-O-coumglc (trans) | 639.0 | 331.0 | 135 | 29 | 19.80 |
| 639.0 | 287.0 | 135 | 35 | |||
| 48I | Mv-3-O-feruloylglc | 669.0 | 331.0 | 120 | 45 | 19.86 |
| 669.0 | 287.0 | 120 | 58 | |||
| 49I | Dp-3-O-glu-pyruvic acid | 533.0 | 371.0 | 110 | 20 | 13.35 |
| 50I | Pt-3-O-glu-pyruvic acid | 547.0 | 385.0 | 100 | 20 | 15.04 |
| 51I | Pn-3-O-glu-pyruvic acid | 531.0 | 369.0 | 110 | 25 | 16.04 |
| 52I | Mv-3-O-glu-pyruvic acid (Vitisin A) |
561.0 | 399.0 | 95 | 30 | 16.34 |
| 53I | Dp-3-O-glu-acetaldehyde | 489.0 | 327.0 | 125 | 30 | 13.94 |
| 54I | Cy-3-O-glu-acetaldehyde | 473.0 | 311.0 | 115 | 35 | 14.38 |
| 55I | Pt-3-O-glu-acetaldehyde | 503.0 | 341.0 | 115 | 32 | 14.75 |
| 56I | Pn-3-O-glu-acetaldehyde | 487.0 | 325.0 | 120 | 34 | 16.42 |
| 57I | Mv-3-O-glu-acetaldehyde (Vitisin B) |
517.0 | 355.0 | 100 | 28 | 16.79 |
| 58I | Dp-3-O-glu-acetone | 503.1 | 341.0 | 120 | 25 | 14.76 |
| 59I | Pt-3-O-glu-acetone | 517.2 | 355.0 | 120 | 26 | 16.79 |
| 60I | Pn-3-O-glu-acetone | 501.1 | 339.0 | 120 | 30 | 17.37 |
| 61I | Mv-3-O-glu-acetone | 531.1 | 369.0 | 120 | 28 | 17.67 |
| 1II | Cy-3-O-glu-4-vinylcatechol | 581.0 | 419.0 | 115 | 22 | 8.28 |
| 2II | Pt-3-O-glu-4-vinylcatechol | 611.0 | 449.0 | 100 | 26 | 9.12 |
| 3II | Pn-3-O-glu-4-vinylcatechol | 595.0 | 433.0 | 110 | 23 | 9.35 |
| 4II | Mv-3-O-glu-4-vinylcatechol | 625.0 | 463.0 | 95 | 25 | 9.85 |
| 5II | Dp-3-O-glu-4-vinylphenol | 581.0 | 419.0 | 125 | 22 | 8.39 |
| 6II | Pt-3-O-glu-4-vinylphenol | 595.0 | 433.0 | 115 | 23 | 9.46 |
| 7II | Mv-3-O-glu-4-vinylphenol | 609.0 | 447.0 | 100 | 26 | 10.39 |
| 8II | Cy-3-O-glu-4-vinylguaiacol | 595.0 | 433.0 | 120 | 23 | 9.67 |
| 9II | Mv-3-O-glu-4-vinylguaiacol | 639.0 | 477.0 | 120 | 28 | 10.61 |
Notes: The first 14 compounds were absolutely quantified using authentic standards. The remaining derivatives were semi-quantified based on calibration curves of structurally corresponding anthocyanin glucosides (Table S2).
With the optimized separation and MS/MS conditions in hand, method validation was performed as follows. Mixed standard solutions of anthocyanins at eight different concentrations (1, 5, 10, 20, 50, 100, 200, and 500 μg/L) were analyzed to establish calibration curves and correlation coefficients (R2) (Table S2). Limits of detection (LOD) were determined at a signal-to-noise ratio (S/N) of 3, ranging from 0.17 to 3.81 μg/L, while limits of quantification (LOQ) were determined at S/N = 10, ranging from 0.56 to 12.57 μg/L. Method accuracy was evaluated by spiking blank wine samples at three concentration levels (20, 40, and 200 μg/L), yielding recoveries between 90.80% and 112.50%, and intra-day coefficients of variation ranging from 0.81% to 4.87%, demonstrating good stability of the method (Table S3). Analytical methods for A&D in wines are diverse, with HPLC or UPLC combined with MS techniques enabling comprehensive molecular structure identification and accurate quantification (Custodio-Mendoza et al., 2024). Zhang et al. (2020) established an HPLC-QQQ-MS/MS method for the determination of 37 malvidin derivatives, achieving good linearity and sensitivity that enabled the discrimination of various vintage wines. The present study employed an optimized UPLC-MS/MS approach to improve both resolution and detection efficiency, enabling the comprehensive identification and quantification of 58 anthocyanins and their derivatives. Notably, this study extends the qualitative characterization of anthocyanins in the absence of standards, particularly for pyranoanthocyanins, which are rarely systematically reported.
3.2. Anthocyanin and derivative profiles
A total of 58 A&D were identified in 64 Marselan wines, including five anthocyanidins, six monoglucoside anthocyanins, three diglucoside anthocyanins, 22 acylated anthocyanins, and 22 pyranoanthocyanins. Among them, acylated anthocyanins and pyranoanthocyanins were semiquantified, while the remaining A&D were fully quantified. The results showed that the total anthocyanin content ranged from 3.52 to 626.53 mg/L, with an average concentration of 254.75 mg/L. The anthocyanin content in red wines can reach levels of up to 2000 mg/L, depending on grape variety, climatic conditions, and winemaking techniques (Han et al., 2019). Fig. 1 shows the concentrations and relative proportions of A&D in the 64 wine samples. Anthocyanidins, monoglucoside anthocyanins, and diglucoside anthocyanins accounted for more than 70% of the total on average, among which Mv, Mv-3-O-glucoside, and Mv-3,5-O-diglucoside were dominant. Malvidin is the most abundant anthocyanin in wine, and malvidin-3-O-glucoside is typically present at concentrations of approximately 300–500 mg/L in young wines (Torres-Rochera et al., 2023). The total concentrations of anthocyanidins, monoglucoside anthocyanins, and diglucoside anthocyanins were in ranges of 0.72–225.15 mg/L, 0.75–183.89 mg/L, and 0.03–103.81 mg/L, respectively. Notably, Pg was not detected in free form, but its glycosylated derivatives were present in the wines. Among acylated A&D, all six anthocyanidins were detected in acetylated forms. For p-coumaroylated derivatives, Pg-3-O-coumglc was exclusively present in the trans configuration, whereas both cis and trans isomers of the other five anthocyanidins were detected. The concentration of acylated anthocyanins decreased as follows: acetylated > p-coumaroylated > caffeoylated > feruloylated, with malvidin-based derivatives being the most abundant. The total concentration of acylated anthocyanins ranged from 0.55 to 174.57 mg/L. Wang et al. (2023) reported that non-acylated anthocyanins primarily contribute to the red-violet coloration of wines, whereas acylated anthocyanins are more closely associated with wine aging Therefore, the shelf life of anthocyanins can be extended either by controlling environmental factors or by modifying their chemical structure (Cai et al., 2022; MohdMaidin et al., 2019). Pyranoanthocyanins in aged red wines exhibit greater stability and bioactivity than their parent anthocyanins, and are affected by grape variety, yeast strain, and aging conditions such as temperature, closure type, and micro‑oxygenation (Huang et al., 2024; Quaglieri et al., 2017). They occur in relatively low concentrations, primarily as A-type and B-type vitisins, with total levels in the range of 1.44–43.97 mg/L. Pyranoanthocyanins, influenced by grape variety, yeast strain, and storage and aging conditions (e.g., temperature, closure type, and micro‑oxygenation), generally increase during aging, particularly under enhanced micro‑oxygenation (Quaglieri et al., 2017).
Fig. 1.
Proportional distribution and mean concentrations of various classes of anthocyanins and their derivatives (A&D) in Marselan wines, including total anthocyanidins (A), anthocyanidins (B), monoglucoside anthocyanins (C), diglucoside anthocyanins (D), acylated anthocyanins (E), and pyranoanthocyanins (F).
3.3. Vintage variability of Marselan wines
3.3.1. Variability between two vintages
To investigate the inter-annual variation of A&D, Marselan wines from two consecutive vintages, 2021 (n = 23) and 2022 (n = 23), were collected and analyzed (Table S4). Independent-sample t-tests were performed on 58 A&D compounds, and only 12 exhibited significant differences between the two vintages (p < 0.05). These included one anthocyanidin (Dp), three monoglucoside anthocyanins (Mv-3-O-glucoside, Pg-3-O-glucoside, Pn-3-O-glucoside), four acylated anthocyanins (Cy-3-O-acetylglc, Pg-3-O-acetylglc, Pn-3-O-acetylglc, Mv-3-O-feruloylglc), and four pyranoanthocyanins (A-type vitisin of Pt-3-glucoside, A-type vitisin of Pn-3-glucoside, A-type vitisin of Mv-3-glucoside, Cy-3-O-glucoside-4-vinylcatechol). These results indicate that the majority of A&D compounds exhibited inter-annual stability. Thus, although vineyard soil and harvest timing can significantly influence A&D levels, certain characteristics are retained during wine aging (Lu et al., 2023).
To further explore the 12 compounds that exhibited significant differences, OPLS-DA was employed by considering both inter-variable correlations and the maximal separation between the two groups. As shown in the score plot (Fig. 2A), the red dots representing the 2021 samples and blue dots representing the 2022 samples are clearly separated along predictive component 1, indicating that these 12 A&D display potential vintage discriminant characteristics. The corresponding loading plot illustrates the contributions of different A&D to sample discrimination, with anthocyanidins, monoglucoside anthocyanins, and pyranoanthocyanins clustering more closely together and contributing similarly to group classification (Fig. 2B). The association between samples and variables is further illustrated in Fig. 2C, which shows that Cy-3-O-glucoside-4-vinylcatechol contributed predominantly to the separation of the 2021 samples, while two acylated anthocyanins (Cy-3-O-acetylglc, and Mv-3-O-feruloylglc), contributed more strongly to the separation of the 2022 samples. Variables with variable importance in the projection (VIP) values greater than 1, including Pn-3-O-acetylglc, Cy-3-O-glucoside-4-vinylcatechol, Mv-3-O-feruloylglc, Mv-3-O-glucoside, and Pg-3-O-glucoside, can thus be regarded as potential biomarkers for vintage differentiation (Fig. 2D).
Fig. 2.
OPLS-DA analysis of A&D in Marselan wines. (A) Score plot; (B) Loading plot; (C) Biplot; (D) Variable importance in the projection (VIP) plot.
3.3.2. Multi-vintage evolution patterns
The analysis of variance (ANOVA) with Duncan's test at p < 0.05, indicated that a total of 14 A&D exhibited significant differences (Table S5). As the samples were analyzed in 2023, they were classified according to aging time from 1 (2022) to 6 (2017) years. The relative decrease in A&D concentration was calculated based on sequential measurements [(Ci-Ci+1)/Ci]. All five anthocyanidins showed significant differences, with the most pronounced decline occurring between 4 and 5 years (approximately 6.8-fold), followed by that between 2 and 3 years (approximately 2-fold). Overall, the average decrease rate reached up to 95% over the six years period. In wines produced from international grape varieties (Cabernet Sauvignon, Syrah, Merlot, and Cabernet Franc), monomeric anthocyanins, anthocyanin-flavan-3-ol condensation products, and certain anthocyanin derivatives similarly exhibit a rapid decline over 5–10 years of aging (Zhang et al., 2021). This phenomenon can be explained by the dynamic changes of A&D in wine resulting from several factors. Under light exposure and elevated temperatures (25–30 °C), a decrease in A&D concentrations has been observed (Medina-Plaza et al., 2024). At pH 3–4, anthocyanin solubility decreases with increasing pH (Forino et al., 2020). In the presence of pyruvic acid, low temperatures (10–15 °C) accelerate anthocyanin degradation, whereas the opposite effect occurs at higher temperatures (20–32 °C) (Romero & Bakker, 2000). Among monoglucoside anthocyanins, all except Pg-3-O-glucoside showed interannual variation. Three acetylated anthocyanins (Dp-3-O-acetylglc, Mv-3-O-acetylglc, and Pt-3-O-acetylglc) and one feruloylated anthocyanin (Mv-3-O-feruloylglc) exhibited significant differences. However, diglucoside and pyranoanthocyanins showed no significant variation over the six consecutive years.
Fig. 3 illustrates interannual variability of A&D. As expected, the average A&D concentration was highest in 2022 (approximately 370.5 mg/L) and decreased sharply with aging, reaching less than 20 mg/L in 2017 (Fig. 3A). The more stable pyranoanthocyanins increased in concentration from 3% to 23% (Fig. 3B). For comparison, in aged Primitivo red wines from southern Italy, pyranoanthocyanin derivatives formed from malvidin, peonidin, and petunidin accounted for approximately 49% of the total anthocyanin-derived pigments (Dipalmo et al., 2016), highlighting the accumulation of these stable pigments during wine aging. Notably, diglucoside anthocyanins were undetected in three vintages, and the proportion of acylated A&D markedly increased in 2018. Fig. 3C-E show the evolution of 13 A&D subclasses, indicating consistent declining trends for anthocyanidins, monoglucosides, diglucosides, and acylated anthocyanins, whereas pyranoanthocyanins became relatively more abundant in older wines. Contrary to the exponential increase in A-type vitisin concentration reported by Schwarz et al. (2004), only minor fluctuations with no evident exponential pattern were observed in this study during 1–4 years. In addition, a slow upward trend was observed in the content of vinylphenol derivatives. Fig. 3F-H display the interannual variability of the 14 A&D compounds. Although variation existed among compounds, the fastest decline was observed during 2018–2019 and 2020–2021. Except for a few outliers, most compounds exhibited similar distribution patterns of relative decrease within the same time interval, indicating consistent degradation behavior during storage (Fig. 3F and G). The maximum cumulative loss rate reached 98%, while acetylated as well as feruloylated anthocyanins showed a transient increase after one year, followed by continuous annual losses (Fig. 3H).
Fig. 3.
Inter-annual variations in A&D in wines from 2017 to 2022. (A) Total concentrations; (B) Percentage of A&D; (C–E) Concentrations of various classes of A&D; (F, G) Relative decrease in A&D across vintages; (H) Cumulative decrease in A&D across vintages.
3.4. Geographical variability of Marselan wines
3.4.1. Variability among distinct regions and subregions
Samples were classified according to regions and subregions, with the regions including NX, HB, and XJ, and the sub-regions comprising five areas within the Ningxia wine region: HL, HSB, QTX, YC, and YN. The mean and standard deviation (SD) values of 58 anthocyanins and derivatives (A&D), along with their significant differences (p < 0.05) among regions and sub-regions, are listed in Table S6 and Table S7. Among the three regions, 32 A&D showed significant differences, including five anthocyanidins, five monoglucoside anthocyanins, 16 acylated anthocyanins, and six pyranoanthocyanins. In the five sub-regions, 25 A&D exhibited significant differences, including three anthocyanidins, three monoglucoside anthocyanins, 14 acylated anthocyanins, and five pyranoanthocyanins. The overall distribution of A&D concentrations indicated that the HB samples contained the lowest, whereas XJ and NX samples contained approximately five-fold higher concentration than HB, with those in XJ being slightly higher than in NX (Fig. 4A). Further analysis using stacked percentage bar plots highlighted the characteristic composition of A&D types across regions. In XJ, acylated anthocyanins accounted for the highest proportion (37%), NX samples showed the highest proportion of anthocyanidins (35%), and HB samples had the largest proportions of monoglucoside anthocyanins (37%) and pyranoanthocyanins (15%), while diglucoside anthocyanins were only present in NX samples at 1% (Fig. 4C). Within the five NX subregions, A&D concentrations followed the order HL > HSB > YC > QTX > YN (Fig. 4B). Fig. 4D shows that HL had the highest proportion of diglucoside anthocyanins (3%), whereas YN showed the highest proportions of monoglucoside (37%) and pyranoanthocyanins (5%). Overall, the variation in A&D composition among the five sub-regions was smaller than that observed among the three regions. The relationships between regions and A&D were further visualized using chord diagrams (Fig. 4E and F).
Fig. 4.
Distribution and compositional patterns of A&D in wines from distinct regions (A, B, C) and subregions (D, E, F). (A, D) Total concentrations; (B, E) Percentage distributions; (C, F) Chord diagrams.
3.4.2. Multivariate analysis
The A&D dataset with significant differences was combined with multivariate statistical methods (PCA, OPLS-DA, and SVM) to investigate the geographical specificity of Marselan wines from the three regions and five sub-regions. In the dataset of the three regions, the PCA score plot revealed clear distinctions among the regions (Fig. 5A), with a total explained variance of 67.04%, where PC1 accounted for 56.41% and PC2 accounted for 10.63% of the variability. The XJ samples were relatively clustered, whereas the NX samples exhibited both clustering and dispersion, with partial overlap with the HB and XJ samples. OPLS-DA has been applied to volatile compounds to discriminate the origins of Marselan wines (Cassago et al., 2025). In this study, OPLS-DA based on A&D data was employed to differentiate the three regions. The model exhibited satisfactory goodness-of-fit (R2Y = 0.812) and the predictive ability (Q2 = 0.613) based on seven-fold cross-validation, and a permutation test with 200 iterations absence of overfitting. As shown in Fig. 5B, the regions were distinctly separated, and 17 variables with VIP > 1 were identified as the major contributors to interregional discrimination. These included three anthocyanidins (Pn, Pt, and Mv), four monoglucoside anthocyanins (Mv-3-O-glucoside, Pt-3-O-glucoside, Pn-3-O-glucoside, and Dp-3-O-glucoside), seven p-coumaroylated anthocyanins [Pt-3-O-coumglc(cis), Pn-3-O-coumglc(cis), Pn-3-O-coumglc(trans), Cy-3-O-coumglc(trans), Dp-3-O-coumglc(cis), Pt-3-O-coumglc(trans), and Dp-3-O-coumglc(trans)], and three pyranoanthocyanins (B-type vitisin of Cy-3-glc, acetone derivative of Dp-3-glc, and Mv-3-O-glc-4-vinylcatechol).
Fig. 5.
PCA and OPLS-DA score plots of A&D in wines from (A, B) three distinct regions; (C, D) five distinct sub-regions.
Previous studies identified Mv-3-glucoside, Mv-3-O-coumglc, and Pn-3-O-coumglc as characteristic markers for differentiating vintages, vineyard plots, and aging in different French oak barrels brands in Brazilian ‘Syrah’ wines (Cassago et al., 2025). In the dataset of the five subregions, the total explained variance was 67.13%, with PC1 accounting for 53.99% and PC2 for 12.14% of the variability. The separation among the five sub-region groups was relatively small, with QTX and YN samples exhibiting greater within-group dispersion (Fig. 5C). The subregion OPLS-DA model not exhibited satisfactory goodness-of-fit (R2Y = 0.133) and the predictive ability (Q2 = 0.0382). Fig. 5D failed to show clear separation among subregions. Support vector machines (SVMs) have been shown to effectively classify wine samples from relatively small regions with high accuracy (Yu et al., 2025). As shown in Fig. S4, the confusion matrix of wines from the three regions yielded accuracies of 97.92% for the training set and 96.3% for the test set. However, when applying the same SVM algorithm at the sub-regional level, the accuracies for both the training and prediction sets were relatively low, indicating that A&D exhibited limited discriminative power among the subregions.
4. Conclusions
This study aimed to comprehensively characterize the anthocyanins and their derivatives (A&D) in Marselan wines and to evaluate their potential as chemical markers for vintage and geographical differentiation. Using an optimized UPLC-MS/MS method, 58 A&D were identified and quantified in 64 wines from distinct vintages and regions. Results showed that anthocyanidin monomers, monoglucoside anthocyanins, and diglucoside anthocyanins accounted for over 70% of the total anthocyanins, with malvidin derivatives being predominant; this trend was generally consistent across regions and was clearly observed in wines from the most recent four vintages (2022–2019). Vintage variability was mainly characterized by significant differences in acetylated anthocyanins and pyranoanthocyanins, which can serve as potential vintage markers. Multi-vintage analysis further demonstrated a sharp decrease in total A&D concentrations among the six vintages, accompanied by an increasing proportion of pyranoanthocyanins, reflecting their higher stability. Geographical variability was more pronounced among regions than sub-regions, with 32 differential A&D across NX, HB, and XJ. Multivariate analyses (PCA, OPLS-DA, and SVM) confirmed that A&D profiles provide effective discrimination at the regional level, although their traceability power at the sub-regional scale was limited. Future research should integrate A&D fingerprints with complementary data sources, such as volatile compounds, elemental composition, or isotopic signatures, to enhance the accuracy of wine origin authentication. These findings provide important insights into the vintage evolution and geographical specificity of A&D in Marselan wines, which are crucial for understanding their dynamics, identifying reliable markers of wine origin, aging, and pigment stability, and guiding winemaking practices.
CRediT authorship contribution statement
Xianhua Zhao: Writing – original draft, Project administration, Methodology, Funding acquisition. Yingyue Su: Writing – original draft, Validation, Resources, Methodology. Lulu Huang: Writing – review & editing, Methodology, Investigation. Zhiwei Yang: Methodology, Investigation. Pengyu Qi: Writing – review & editing. Yefeng Li: Writing – review & editing, Investigation. Pengbao Shi: Writing – review & editing, Supervision, Resources, Project administration, Investigation, Funding acquisition. Ang Zhang: Writing – review & editing, Supervision, Resources, Methodology.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
The authors also wishes to acknowledge the Science and Technology Development Plan of Taian (Grant Numbers 2024NS007 and 2023NS096) and the Qinhuangdao Science and Technology Research and Development Program (Project No. 202201B019) for their financial support.
Footnotes
This article is part of a Special issue entitled: ‘Wine Chemistry’ published in Food Chemistry: X.
Supplementary data to this article can be found online at https://doi.org/10.1016/j.fochx.2026.104087.
Contributor Information
Pengbao Shi, Email: pengbaoshi@163.com.
Ang Zhang, Email: angzhangwine@163.com.
Appendix A. Supplementary data
Supplementary data associated with this article can be found in Table S1–Table S7 and Fig. S1–Fig. S4.
Figure S1 MS¹ spectra of the 26 anthocyanin standards
Figure S2 MS² spectra and chemical structures of the 26 anthocyanin standards
Chromatographic separation of A&D isomers achieved by the optimized elution program (A) Cy-3-O-glu-4-vinylcatechol and Dp-3-O-glu-4-vinylphenol; (B) Peonidin-3-O-glucoside and Peonidin-3-O-galactoside; (C) Malvidin-3-O-glucoside and Malvidin-3-O-galactoside; (D) Delphinidin-3-O-glucoside and Delphinidin-3-O-galactoside; (E) Pn-3-O-glu-4-vinylcatechol, Pt-3-O-glu-4-vinylphenol, and Cy-3-O-glu-4-vinylguaiacol
Figure S4 Confusion matrices obtained using SVM based on A&D for wines from different regions (A, B) and sub-regions (C, D).
Reagents and standards
Standard calibration curves and correlation coefficients (R2) of 26 anthocyanins and their derivatives
Recovery and relative standard deviation (RSD) of anthocyanins and their derivatives (A&D)
Mean values and standard deviations of anthocyanins and their derivatives in wines from the 2021 and 2022 vintages
Mean values and standard deviations of anthocyanins and their derivatives in wines from the 2017 to 2022 vintages
Mean values and standard deviations of anthocyanins and their derivatives in wines from different regions
Mean values and standard deviations of anthocyanins and their derivatives in wines from different sub-regions
Data availability
Data will be made available on request.
References
- Cai D., Li X., Chen J., et al. A comprehensive review on innovative and advanced stabilization approaches of anthocyanin by modifying structure and controlling environmental factors. Food Chemistry. 2022;366 doi: 10.1016/j.foodchem.2021.130611. [DOI] [PubMed] [Google Scholar]
- Cassago A.L.L., Artencio M.M., Zocolo G.J., et al. Chemical evolution of wine from a new Brazilian terroir: Metabolomic insights into vinification and aging processes. Food Research International. 2025;221(1):117272. doi: 10.1016/j.foodres.2025.117272. [DOI] [PubMed] [Google Scholar]
- Cozzolino D. Phenolics and spectroscopy: Challenges and successful stories in the grape and wine industry. Journal of the Science of Food and Agriculture. 2025;105(3):1408–1412. doi: 10.1002/jsfa.13173. [DOI] [PubMed] [Google Scholar]
- Custodio-Mendoza J.A., Aktaş H., Zalewska M., et al. A review of quantitative and topical analysis of anthocyanins in food. Molecules. 2024;29(8) doi: 10.3390/molecules29081735. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dangles O., Fenger J.A. The chemical reactivity of anthocyanins and its consequences in food science and nutrition. Molecules. 2018;23(8):1970. doi: 10.3390/molecules23081970. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Deshaies S., Sommerer N., Garcia F., et al. UHPLC-Q-Orbitrap/MS2 identification of (+)-Catechin oxidation reaction dimeric products in red wines and grape seed extracts. Food Chemistry. 2022;382 doi: 10.1016/j.foodchem.2022.132505. [DOI] [PubMed] [Google Scholar]
- Dipalmo T., Crupi P., Pati S., et al. Studying the evolution of anthocyanin-derived pigments in a typical red wine of southern Italy to assess its resistance to aging. LWT-Food Science and Technology. 2016;71:1–9. [Google Scholar]
- Ferrer-Gallego R., Rodriguez-Pulido F.J., Toci A.T., et al. Phenolic composition, quality and authenticity of grapes and wines by vibrational spectroscopy. Food Reviews International. 2022;38(5):884–912. [Google Scholar]
- Forino M., Picariello L., Rinaldi A., et al. How must pH affects the level of red wine phenols. LWT. 2020;129 [Google Scholar]
- Freitas V., Mateus N. Formation of pyranoanthocyanins in red wines: A new and diverse class of anthocyanin derivatives. Analytical and Bioanalytical Chemistry. 2011;401(5):1463–1473. doi: 10.1007/s00216-010-4479-9. [DOI] [PubMed] [Google Scholar]
- García-Beneytez E., Cabello F., Revilla E. Analysis of grape and wine anthocyanins by HPLC-MS. Journal of Agricultural and Food Chemistry. 2003;51(19):5622–5629. doi: 10.1021/jf0302207. [DOI] [PubMed] [Google Scholar]
- Han F., Yang P., Wang H., et al. Digestion and absorption of red grape and wine anthocyanins through the gastrointestinal tract. Trends in Food Science & Technology. 2019;83:211–224. [Google Scholar]
- He F., Liang N.N., Mu L., et al. Anthocyanins and their variation in red wines I. Monomeric anthocyanins and their color expression. Molecules. 2012;17(2):1571–1601. doi: 10.3390/molecules17021571. [DOI] [PMC free article] [PubMed] [Google Scholar]
- He F., Liang N.N., Mu L., et al. Anthocyanins and their variation in red wines II. Anthocyanin derived pigments and their color evolution. Molecules. 2012;17(2):1483–1519. doi: 10.3390/molecules17021483. [DOI] [PMC free article] [PubMed] [Google Scholar]
- He J., Carvalho A.R.F., Mateus N., et al. Spectral features and stability of oligomeric pyranoanthocyanin-flavanol pigments isolated from red wines. Journal of Agricultural and Food Chemistry. 2010;58(16):9249–9258. doi: 10.1021/jf102085e. [DOI] [PubMed] [Google Scholar]
- Huang K., Hu J., Li X., et al. Advancements in the promotion of pyranoanthocyanins formation in wine: A review of current research. Food Chemistry. 2024;438 doi: 10.1016/j.foodchem.2023.137990. [DOI] [PubMed] [Google Scholar]
- Leonarski E., Cesca K., de Oliveira D., et al. A review on enzymatic acylation as a promising opportunity to stabilizing anthocyanins. Critical Reviews in Food Science and Nutrition. 2023;63(24):6777–6796. doi: 10.1080/10408398.2022.2041541. [DOI] [PubMed] [Google Scholar]
- Liu C., Wu L., Fan S., et al. The protective effect of cyclodextrin on the color quality and stability of cabernet sauvignon red wine. Journal of Integrative Agriculture. 2024;23(1):310–323. [Google Scholar]
- Lu H.C., Tian M.B., Han X., et al. The key role of vineyard parcel in shaping flavonoid profiles and color characteristics of Cabernet Sauvignon wines combined with the influence of harvest ripeness, vintage and bottle aging. Food Chemistry: X. 2023;19 doi: 10.1016/j.fochx.2023.100772. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mateus N., Silva A.M.S., Santos-Buelga C., et al. Identification of anthocyanin-flavanol pigments in red wines by NMR and mass spectrometry. Journal of Agricultural and Food Chemistry. 2002;50(7):2110–2116. doi: 10.1021/jf0111561. [DOI] [PubMed] [Google Scholar]
- Medina-Plaza C., DuBois A., Tomasino E., et al. Effect of storing conditions (lighting, temperature and bottle color) on rosé wine attributes. Food Chemistry. 2024;439 doi: 10.1016/j.foodchem.2023.138032. [DOI] [PubMed] [Google Scholar]
- MohdMaidin N., Oruna-Concha M.J., Jauregi P. Surfactant TWEEN20 provides stabilisation effect on anthocyanins extracted from red grape pomace. Food Chemistry. 2019;271:224–231. doi: 10.1016/j.foodchem.2018.07.083. [DOI] [PubMed] [Google Scholar]
- Ongkowijoyo P., Luna-Vital D.A., de Mejia E.G. Extraction techniques and analysis of anthocyanins from food sources by mass spectrometry: An update. Food Chemistry. 2018;250:113–126. doi: 10.1016/j.foodchem.2018.01.055. [DOI] [PubMed] [Google Scholar]
- Pascual-Teresa S., Sanchez-Ballesta M.T. Anthocyanins: From plant to health. Phytochemistry Reviews. 2008;7(2):281–299. [Google Scholar]
- Quaglieri C., Jourdes M., Waffo-Teguo P., et al. Updated knowledge about pyranoanthocyanins: Impact of oxygen on their contents, and contribution in the winemaking process to overall wine color. Trends in Food Science & Technology. 2017;67:139–149. [Google Scholar]
- Rentzsch M., Schwarz M., Winterhalter P. Pyranoanthocyanins–an overview on structures, occurrence, and pathways of formation. Trends in Food Science & Technology. 2007;18(10):526–534. [Google Scholar]
- Romero C., Bakker J. Effect of storage temperature and pyruvate on kinetics of anthocyanin degradation, vitisin A derivative formation, and color characteristics of model solutions. Journal of Agricultural and Food Chemistry. 2000;48(6):2135–2141. doi: 10.1021/jf990998l. [DOI] [PubMed] [Google Scholar]
- Schwarz M., Hofmann G., Winterhalter P. Investigations on anthocyanins in wines from Vitis vinifera cv. Pinotage: Factors influencing the formation of pinotin A and its correlation with wine age. Journal of Agricultural and Food Chemistry. 2004;52(3):498–504. doi: 10.1021/jf035034f. [DOI] [PubMed] [Google Scholar]
- Torres-Rochera B., Manjón E., Brás N.F., et al. Supramolecular study of the interactions between malvidin-3-O-glucoside and wine phenolic compounds: Influence on color. Journal of Agricultural and Food Chemistry. 2023;72(4):1894–1901. doi: 10.1021/acs.jafc.2c08502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang Z., Zhang L., Li Y., et al. Non-acylated and acylated anthocynins in red wines of different ages: Color contribution and evaluation. Journal of Food Composition and Analysis. 2023;115 [Google Scholar]
- Yao X.C., Zhang H.L., Ma X.R., et al. Leaching and evolution of anthocyanins and aroma compounds during Cabernet Sauvignon wine fermentation with whole-process skin-seed contact. Food Chemistry. 2024;436 doi: 10.1016/j.foodchem.2023.137727. [DOI] [PubMed] [Google Scholar]
- Yu H., Chu Y., Bian X., et al. Identifying the vintage of French wine using stable isotopes, elemental fingerprints, and a data-driven but explainable approach. Food Chemistry. 2025;464 doi: 10.1016/j.foodchem.2024.141907. [DOI] [PubMed] [Google Scholar]
- Zhang X.K., Lan Y.B., Huang Y., et al. Targeted metabolomics of anthocyanin derivatives during prolonged wine aging: Evolution, color contribution and aging prediction. Food Chemistry. 2021;339 doi: 10.1016/j.foodchem.2020.127795. [DOI] [PubMed] [Google Scholar]
- Zhang X.K., Li S.Y., Zhao X., et al. HPLC-MS/MS-based targeted metabolomic method for profiling of malvidin derivatives in dry red wines. Food Research International. 2020;134 doi: 10.1016/j.foodres.2020.109226. [DOI] [PubMed] [Google Scholar]
- Zhao B.Y., Zhang X.K., Lan Y.B., et al. Sub-regional variation and characteristics of cabernet sauvignon wines in the Eastern Foothills of the Helan Mountain: A perspective from phenolics, visual properties and mouthfeel. Foods. 2023;12(5) doi: 10.3390/foods12051081. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhao X., He F., Zhang X.K., et al. Impact of three phenolic copigments on the stability and color evolution of five basic anthocyanins in model wine systems. Food Chemistry. 2022;375 doi: 10.1016/j.foodchem.2021.131670. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1 MS¹ spectra of the 26 anthocyanin standards
Figure S2 MS² spectra and chemical structures of the 26 anthocyanin standards
Chromatographic separation of A&D isomers achieved by the optimized elution program (A) Cy-3-O-glu-4-vinylcatechol and Dp-3-O-glu-4-vinylphenol; (B) Peonidin-3-O-glucoside and Peonidin-3-O-galactoside; (C) Malvidin-3-O-glucoside and Malvidin-3-O-galactoside; (D) Delphinidin-3-O-glucoside and Delphinidin-3-O-galactoside; (E) Pn-3-O-glu-4-vinylcatechol, Pt-3-O-glu-4-vinylphenol, and Cy-3-O-glu-4-vinylguaiacol
Figure S4 Confusion matrices obtained using SVM based on A&D for wines from different regions (A, B) and sub-regions (C, D).
Reagents and standards
Standard calibration curves and correlation coefficients (R2) of 26 anthocyanins and their derivatives
Recovery and relative standard deviation (RSD) of anthocyanins and their derivatives (A&D)
Mean values and standard deviations of anthocyanins and their derivatives in wines from the 2021 and 2022 vintages
Mean values and standard deviations of anthocyanins and their derivatives in wines from the 2017 to 2022 vintages
Mean values and standard deviations of anthocyanins and their derivatives in wines from different regions
Mean values and standard deviations of anthocyanins and their derivatives in wines from different sub-regions
Data Availability Statement
Data will be made available on request.