Yanxin Lin and Misha Kwasniewski*
*Food Science Department, Penn State University, Penn State University, Rodney A. Erickson Food Science Building, State College, PA, 16803 (mtk5407@psu.edu)

Procyanidins (PCs), or condensed tannins, significantly influence the mouthfeel and stability of red wine and polyphenol-rich foods. However, current analytical methods sometimes fail to correlate perception of astringency or predict tannin retention properly. This may be partly because common analytical methods oversimplify phenolics or tannins into one value (e.g., total concentration), not accounting for their diverse structures and sizes. Additionally, precipitation by methylcellulose (MC) or bovine serum albumin (BSA) employs reagents that only partially resemble endogenous grape proteins or human saliva (HS). In relation to perception, HS is more appropriate for studying complexation with tannins, as it contains the actual proteins and quantities inducing in-mouth precipitation, rather than analogs. A rapid ultra-performance liquid chromatography-tandem mass spectrometry-based method, coupled with in-source fragmentation, was used to quantitatively analyze and fingerprint a wide array of PC structures in wine after precipitation with MC, BSA, and HS. PCs were first fragmented using three different cone voltages (CVs) in the ESI interface, then further fragmented in the collision cell to enable their selective detection using optimized MRMs, which are finally used in a model developed for sample characterization and creation of a tannin “fingerprint”. The “fingerprint” is composed of a variety of MRM transitions and their corresponding ratios for PCs across CVs, resulting in high-dimensional vectors for each PC rather than a single value, as seen in previous methods. Comparing “fingerprints” of PC standards to samples allows accurate identification and quantification of unknown PCs. Accuracy of the model was expressed as root mean squared error ranging from 0.0001 to 0.1475. Precision, expressed as relative standard deviation, was <0.82%. Within a one to five DP range of the “fingerprint”, HS removes more small oligomers and monomers than BSA and MC, suggesting potential underestimation of low molecular weight tannin effect on astringency by BSA and MC.

Funding Support: None