Scientists Trace the Sour and Astringent Taste of Corni Fructus to Specific Acids and Tannins

The dried fruit of Cornus officinalis, known in Chinese as shanzhuyu and in pharmacopoeias as Corni Fructus, has been a fixture of traditional medicine and home cooking in central China for centuries. It flavors soups, wines, juices, and preserved foods, and it anchors classic herbal formulas such as Liuwei Dihuang Wan. In 2023, Chinese regulators officially approved it as a dual-use medicinal and edible material, opening the door to wider food development. Yet one deceptively simple question has lingered: which molecules actually make this fruit taste so sharply sour and mouth-puckeringly astringent? A new study published in Food Chemistry: X set out to answer it with an unusually comprehensive toolkit, combining high-resolution mass spectrometry, robotic taste sensing, human taste panels, and computational docking of molecules against taste receptors.

The research team, led by Yadi Hou and colleagues at Henan University of Chinese Medicine, assembled 27 batches of mature dried fruit pulp from four major producing regions: Xixia in Henan, Foping in Shaanxi, Chun’an in Zhejiang, and Yangcheng in Shanxi. All samples were harvested between September and October 2025, blanched in boiling water, deseeded, and air-dried, ensuring that differences in processing would not confound the comparison. The samples were grouped by origin, with ten batches from Henan, seven from Shaanxi, six from Zhejiang, and four from Shanxi, providing a geographically diverse but controlled dataset for chemical and sensory analysis.

At the analytical core of the study was ultra-high-performance liquid chromatography coupled to tandem mass spectrometry. Using an Orbitrap mass spectrometer operating in negative ion mode at resolutions of 120,000 for full scans and 30,000 for fragment spectra, the researchers profiled each extract and identified 82 constituents: 52 organic acids, 28 tannins, and two unclassified features. Twelve compounds, including malic acid, tartaric acid, citric acid, gallic acid, caftaric acid, caffeic acid, shikimic acid, and p-coumaric acid, were confirmed against authentic reference standards, while the rest were annotated by matching accurate masses, fragmentation patterns, and literature data. The tannin fraction proved dominated by galloylated derivatives of glucose, quinic acid, and sedoheptulose, including compounds such as 2,3-di-O-galloyl-d-glucose, erythrulose gallate, and a tentatively proposed methylol tannin-type structure carrying an extra terminal galloyl group.

Multivariate statistics revealed that geography leaves a clear chemical fingerprint. Principal component analysis showed that the first two components explained 54.7 percent of the total variance, with samples from Henan and Shaanxi clustering together while Zhejiang and Shanxi samples separated more distinctly. A supervised PLS-DA model sharpened these separations, achieving an R-squared of 0.89 and a Q-squared of 0.67, and passed 100 permutation tests without signs of overfitting. Quantification of the 12 reference-confirmed compounds showed that Zhejiang samples contained substantially higher levels of tartaric acid, citric acid, and shikimic acid, along with markedly higher total organic acid content, while caffeic acid, caftaric acid, and p-coumaric acid were significantly lower in that group.

To connect chemistry with perception, the team ran two parallel taste assessments. Twenty trained human panelists, aged 21 to 35 and screened for normal taste function, scored room-temperature infusions for sourness, astringency, bitterness, and taste harmony on 10-point scales in blinded, randomized triplicate sessions. The panel found that Zhejiang samples carried the highest sourness at 8.02 plus or minus 0.60, while Henan and Shaanxi samples showed the strongest astringency at roughly 7.0. An SA402B electronic tongue equipped with lipid-membrane sensors for bitterness, astringency, sourness, saltiness, and umami largely agreed with the human panel on sourness, with correlations reaching statistical significance at p below 0.001, though its agreement on astringency and bitterness was only moderate, a reminder that machines still struggle to fully replicate the human mouth.

The integration of chemical profiles with instrumental taste responses produced the study’s most concrete leads. Pearson correlation analysis showed that sourness tracked positively with malic acid and its derivatives, lactic acid, and quinic acid glucoside, while astringency correlated positively with 3-O-galloyl-d-glucose and p-coumaric acid but negatively with 1-O,7-O-digalloyl-d-glucose, suggesting that the degree of astringency depends on the specific architecture of gallic acid polymers rather than their total abundance. Bitterness and its lingering aftertaste correlated strongly with multiple galloylated compounds, including 4-O-galloylquinic acid, erythrulose gallate, and digalloyl-sedoheptulose derivatives. Partial least squares regression models quantified these relationships, with the sourness model performing impressively at R-squared 0.945, bitterness and aftertaste-B models at 0.820 and 0.826, while the astringency model lagged at 0.597.

That weaker astringency model is scientifically telling rather than disappointing. Astringency is not a taste in the strict sense but a tactile sensation arising when polyphenols bind and precipitate salivary proteins, reducing oral lubrication and producing the characteristic drying, puckering feel. Its intensity depends on tannin molecular size, polymerization degree, individual salivary protein composition, and aftertaste dynamics, factors that a single methanol extract cannot fully capture. The authors are careful to frame their galloylated tannins and phenolic acids as candidate astringency-associated markers rather than confirmed causal contributors, and they call for sensory recombination, omission experiments, and saliva-binding assays to establish causality.

To probe mechanism, the researchers turned to molecular docking, simulating how 82 candidate compounds interact with sour taste receptors and a protein model of astringency. For sourness, they docked ligands against OTOP1, PKD1L2, PKD1L3, and PKD2L1, the proton channels and ion channels known to mediate sour taste transduction, retrieving receptor structures from UniProt. For astringency, they built a homology model of human saliva-derived beta-casein using the I-TASSER platform, since polyphenol binding to casein-like proteins underlies the puckering sensation. Compounds such as trans-aconitic acid, malic acid glucoside, and malic acid diesters showed favorable binding energies with the sour receptors, while galloylated tannins including 7-O-galloyl-D-sedoheptulose and erythrulose gallate bound favorably to both OTOP1 and beta-casein, lending preliminary mechanistic support to the statistical associations.

The team also confronted a subtle methodological question: does the extraction solvent change the picture? By preparing parallel extracts in 70 percent methanol and in water from the same 27 batches and correlating the peak areas of 79 matched analytes, they found a significant but modest overall concordance of r equal to 0.308, with class-level correlations of 0.343 for organic acids and 0.266 for tannins. Encouragingly, key compounds including malic acid, fumaric acid, tartaric acid, gallic acid, caffeic acid, and erythrulose gallate showed much stronger agreement, with correlations between 0.683 and 0.929 that survived false-discovery-rate correction. This analyte-dependent pattern means the methanol-based screening results are best treated as hypothesis-generating, matrix-dependent associations that now require validation in matrix-matched sensory systems.

The broader significance of the work extends well beyond one fruit. By demonstrating that a layered strategy of targeted metabolomics, electronic tongue fingerprinting, human sensory panels, and receptor docking can pinpoint candidate taste markers in a complex plant matrix, the study offers a template for flavor-oriented quality control across medicinal and edible plants. For Corni Fructus specifically, the identified markers, malic acid-related constituents for sourness and specific galloylated tannins for astringency and bitterness, could underpin standardized gustatory fingerprints that distinguish geographic origins and guide breeding, harvesting, and processing. As the medicinal-edible food sector grows, the molecules behind an ancient fruit’s characteristic bite are finally coming into chemical focus.

Subject of Research: Chemical basis of sourness and astringency in Corni Fructus

Article Title: Candidate compounds associated with sourness and astringency in Corni Fructus: an integrated analysis of chemical profiles, sensory evaluation, and molecular docking

Article References: Hou, Y., Guo, H., Liu, Y., Li, Y., Chen, M., Du, H., Chi, J., Wang, Z., & Dai, L. (2026). Candidate compounds associated with sourness and astringency in Corni Fructus: an integrated analysis of chemical profiles, sensory evaluation, and molecular docking. Food Chemistry: X, Article 104564. https://doi.org/10.1016/j.fochx.2026.104564

Image Credits: AI Generated

DOI: 10.1016/j.fochx.2026.104564

Keywords: Corni Fructus, Cornus officinalis, organic acids, tannins, sourness, astringency, UHPLC-MS/MS, electronic tongue, molecular docking, sensory evaluation, food chemistry, taste markers