Differentiation of Geographical Origins for Cabernet Sauvignon Wines

Significance of the Topic

Wine adulteration and mislabeling pose significant health risks and undermine consumer trust, especially for premium products such as Cabernet Sauvignon. Reliable methods to verify geographic origin are essential for regulatory compliance, quality assurance, and brand protection. Non-targeted metabolomic profiling offers a comprehensive chemical fingerprint that reflects terroir-related variations in grape metabolites and wine processing.

Aims and Overview of the Study

This application note presents a workflow combining UHPLC-Q-TOF/MS with advanced chemometric analysis to differentiate Cabernet Sauvignon wines from two U.S. wineries and three regions in China. By extracting and statistically filtering metabolite features, the study aims to identify key markers that discriminate wines according to their geographic origin and to build predictive classification models.

Methodology and Instrumentation

• Sample Preparation: 113 Cabernet Sauvignon wines were sampled from V. Sattui and Robert Mondavi (Napa Valley, U.S.) and three Chinese regions (Zhangjiakou, Qinghuangdao, Shandong). Samples were centrifuged and injected (2 µL) directly.
• UHPLC Conditions: Agilent 1290 Infinity system with ZORBAX Eclipse Plus C18 column (2.1×100 mm, 1.8 µm) at 40 °C; mobile phases of 5 mmol/L ammonium acetate/0.1% formic acid (A) and methanol/water (95:5) with additives (B); 20 min gradient, 0.4 mL/min.
• MS Detection: Agilent 6530 Q-TOF with dual Jet Stream ESI in positive mode; TOF scan (100–1,100 m/z) and targeted MS/MS; accurate mass reference at 121.0509 and 922.0098 m/z.
• Data Processing: Feature extraction via Molecular Feature Extraction in MassHunter Qual or Profinder; alignment, filtration (occurrence, variability, ANOVA P≤0.01, fold change ≥3, correlation), and chemometrics in Mass Profiler Professional.

Main Results and Discussion

Over 3,000 features per sample were detected. After multistep filtering, 65 differential entities remained. PCA captured 84.8% of variance in the first three components, separating U.S. and Chinese wines. Hierarchical clustering and K-means revealed three compound clusters with distinct abundance patterns across regions. A PLS-DA model achieved 86.7% overall accuracy for five classes and 100% when grouping Chinese sources. Twenty-three metabolites were tentatively identified, including polyphenols (e.g., procyanidin dimers, flavonoid glycosides), organic acids, and small peptides, linking terroir-driven grape metabolism to wine chemical profiles.

Benefits and Practical Applications

• Delivers robust non-targeted chemical fingerprints for origin authentication and anti-fraud control.
• Captures a broad range of endogenous metabolites without prior compound selection.
• Chemometric models enable rapid classification and quality screening in regulatory and industrial laboratories.

Future Trends and Applications

• Validation of marker panels across larger and more diverse wine collections to ensure robustness.
• Extension of the approach to other grape varieties, vintages, and fermented beverages.
• Integration of expanded spectral libraries, machine learning algorithms, and high-throughput workflows.
• Combination with genomic and sensory data for multidimensional authenticity frameworks.

Conclusion

UHPLC-Q-TOF/MS coupled with comprehensive chemometric analysis provides a powerful platform for tracing wine geographic origin based on non-targeted metabolomic profiles. The identification of differential markers and predictive models establishes a foundation for reliable authentication tools to support quality assurance and combat wine fraud.

References

1. Liang N.N. et al., China Fermentation 33(12), 23–28 (2014).
2. Serrano-Lourido D. et al., Food Chemistry 135(3), 1425–1431 (2012).
3. Bellomarino S.A. et al., Talanta 80(2), 833–838 (2009).
4. Kallithraka S. et al., Food Chemistry 73(4), 501–504 (2001).
5. Vaclavik L. et al., Analytica Chimica Acta 45, 685 (2011).
6. Cuadros-Inostroza A. et al., Analytical Chemistry 82, 3573–3580 (2010).