In this study, the ¹H HRMAS-NMR (High-resolution Magic Angle Spinning-Nuclear Magnetic Resonance) spectra of 52 cheese samples obtained from the South Korean dairy farms were evaluated for their metabolic profiling and intensities associating with the sensory qualities. The NMR profiles displayed a broad range of compounds comprising amino acids, carbohydrates, organic acids, and phospholipids. Afterwards, the cheese samples were categorized into three groups (more likeness - G1, moderate likeness - G2, less likeness - G3), in relating to their sensory scores. The NMR data of the samples were later investigated through multivariate statistical tools to define the variations in metabolic fingerprints of every cheese sample and their intensities hailing in individual sensory groups. The unsupervised PCA employing all cheese samples unveiled the uniqueness in metabolite profiles of the brown and cheddar type cheeses (outliers). Moreover, Gouda and other types of cheeses displayed samples positioning in respective of their metabolite profiles. The pairwise comparison of sensory groups in the supervised models perceived better separation in OPLS-DA than PLS-DA. The corresponding VIP (PLS-DA) and loading (OPLS-DA) plots revealed amino acids and organic acids (lactate, citrate) as significant variables. The discrimination of G 1 Gouda type of cheeses against G 2 and G 3 was highly associated with their citrate levels. Further investigation using heatmaps displayed clear differentiation between each sensory group in terms of the levels of amino acids, lactate, citrate, phospholipids, and glycerol, conveying these variations are likely due to proteolytic and metabolic processes in cheese ripening. This study concluded that ¹H HRMAS-NMR metabolite profile of the Korean cheeses is consistence with their sensory qualities. Further, the candidate metabolites identified in this study confers their potential application as biomarkers in cheese industries for faster and effective validation of sensory characteristics.

在这项研究中，¹从韩国奶牛场获得的52个奶酪样品的H HRMAS-NMR（高分辨率魔角旋转核磁共振）光谱进行了代谢谱分析以及与感官品质相关的强度评估。NMR谱图显示了范围广泛的化合物，包括氨基酸，碳水化合物，有机酸和磷脂。然后，根据其感官评分，将奶酪样品分为三组（较高的相似度-G1，中等相似度-G2，较少的相似度-G3）。后来，通过多变量统计工具对样品的NMR数据进行了研究，以定义每个奶酪样品的代谢指纹图谱的变化及其在各个感官组中的强度。采用所有奶酪样品的无监督PCA揭示了棕色和切达干酪（离群值）的代谢产物特征的独特性。此外，高达奶酪和其他类型的奶酪还展示了分别位于其代谢产物谱中的样品。在监督模型中感觉组的成对比较认为，OPLS-DA中的感觉分离比PLS-DA更好。相应的VIP（PLS-DA）和装载量（OPLS-DA）图显示氨基酸和有机酸（乳酸，柠檬酸）为重要变量。G 1干酪型奶酪对G 2和G 3的区分与柠檬酸水平高度相关。使用热图的进一步研究表明，每个感觉类别之间在氨基酸，乳酸，柠檬酸盐，磷脂和甘油的水平上都有明显的区别，奶酪成熟过程中的蛋白水解和代谢过程可能传达了这些差异。这项研究得出的结论是韩国奶酪的¹ H HRMAS-NMR代谢产物谱与其感官品质一致。此外，本研究中鉴定出的候选代谢物具有潜在的应用价值，可作为奶酪行业中的生物标记物，以更快，更有效地验证感官特征。