Aiming at the identification of the key bitter peptides in fermented foods, a new approach, coined "sensoproteomics", was developed and applied to fresh cheese samples differing in bitter taste intensity. By means of MPLC fractionation of the water-soluble cheese extracts in combination with taste dilution analysis, complex fractions with intense bitter taste were located and then screened by UPLC−MS/MS for the entire repertoire of ∼1600 candidate peptides, extracted from a literature meta-analysis on dairy products, by using a total of 120 selected reaction monitoring methods computed in silico. A total of 340 out of the 1600 peptides were found in the cheese samples, among which 17 peptides were identified as candidate bitter peptides by considering only peptides that were located in the bitter-tasting MPLC fractions (signal-to-noise ratio: ≥10) with a fold-change of ≥3 when comparing the less bitter to the more bitter cheese sample and that were validated by comparison with the synthetic reference peptides. While EIVPNS[phos]VEQK (α s1 -CN 70−78 ) and INTIASGEPT (κ-CN 122−131 ) did not exhibit any bitter taste up to 2000 μmol/L, 15 of the 17 target peptides showed bitter taste thresholds ranging from 30 (ARHPHPHLSFM, κ-CN 96−106 ) to 690 μmol/L (IQKEDVPS, α s1 -CN 81−88 ). Finally, quantitative peptide analysis followed by calculation of dose-overthreshold factors revealed a primary contribution of MAPKHKEMPFPKYPVEPF (β-CN 102−119 ) and ARHPHPHLSFM (κ-CN 96−106 ) to the perceived bitter taste of the fresh cheese samples. Finally, the evolution of the bitter peptides throughout two different fresh cheese manufacturing processes was quantitatively recorded.

No abstract

This is the first application of fully automated, preparative, two-dimensional HPLC combined with sensory analysis for taste compound discovery using a sweet and licorice-like bitter-tasting aniseed extract as an example. Compared to the traditional iterative fractionation of food extracts by sensory-guided sequential application of separation techniques, the fully automated 2D-HPLC allowed the comprehensive separation of the aniseed extract into 256 subfractions and reduced the fractionation time from about 1 week to <1day. Using a smart sensory strategy to locate high-impact fractions, e.g., by evaluating first-dimension fractions by reconstituting them from second-dimension subfractions, followed by straightforward application of the taste dilution analysis on the individual second-dimension subfractions revealed the sweet-tasting trans-anethole and the bitter-tasting trans-pseudoisoeugenol 2-methylbutyrate, showing recognition thresholds of 70 and 68 μmol/L, respectively, as the primary orosensory active compounds in aniseed. 2D-HPLC combined with smart sensory analysis seems to be a promising strategy to speed the discovery of the key players imparting the attractive taste of foods.

During the last few years, key taste-active compounds have been isolated and identified by means of a combination of a time-and lab-consuming successive fractionation and sensory characterization. Because the peptidome of fermented, protein-rich food is very complex, new strategies are necessary to accelerate the identification of taste-active peptides. In this study, two advanced mass spectrometric approaches were developed to comprehensively map the bitter tasting peptidome of fermented foods by data-independent acquisition (DIA) using sequential window acquisition of all theoretical fragment ion−mass spectrometry (SWATH−MS) and an in silico-assisted triple quadrupole (QQQ)-based targeted proteomics approach, separately. Application of both techniques on two fresh cheese samples as well as on crude medium-pressure liquid chromatography fractions exhibiting intense bitter taste, followed by filtering the hydrophobic target peptides (Q value of ≥1200 cal/mol) showing a signal-to-noise ratio of ≥10 and a fold change of ≥3 when comparing the less bitter to the more bitter cheese sample, revealed the candidate bitter peptides, which were then validated by means of synthetic reference peptides and human sensory evaluation. The bitter peptides were then quantitated in the fresh cheese samples as well as in a series of dairy products by means of QQQ−MS and SWATH−MS, separately. Although the QQQ−MS method showed 2−80-fold lower limits of quantitation (LOQ), the SWATH−MS method could be shown for the first time to enable the comprehensive quantitation of all sensorially relevant key bitter peptides with LOQs far below the bitter taste recognition concentration of each peptide.