Abstract
The objectives of this study were to targetedly investigate metabolic profiles of the serum in dairy cows (both healthy controls (CON) and cows that developed subclinical mastitis (SCM) at early lactation) up to -8 wks prepartum, and evaluate the performance of new biomarkers for both early onset and progression of the disease. In this research, DI/LC-MS/MS based metabolomics was employed to identify and quantify biochemical signatures in the serum of dairy cows at -8 wks, -4 wks, disease diagnosis, +4 wks, and +8 wks relative to parturition. One hundred and twenty-eight metabolites including amino acids (21), acylcarnitines (7), biogenic amines (8), glycerophospholipids (77), sphingolipids (14), and hexose (1) were quantified in all sera. Univariate analysis showed that several metabolites (e.g., phosphatidylcholine (PC aa C30:2), hydroxysphingomyelin (SM (OH) C22:2), isoleucine, leucine, and lysine; P < 0.01) were consistently elevated in the serum of cows with SCM at five tested time points. In the supervised multivariate analysis (i.e., PLS-DA; permutation test: P < 0.05), a variable importance in projection (VIP) plot was used to rank the most significant discriminators between SCM and CON cows. Two predictive biomarker models and one diagnostic biomarker model for SCM were developed by combination of amino acids, sphingomyelin, phosphatidylcholine, and kynurenine. AUC values of three ROC curve of biomarker models were all greater than 0.995. The predictive and diagnostic models for SCM based on metabolites in serum are practical biomarker alternatives as compared to the current milk somatic cell count (SCC) assays. The data demonstrate that metabolomics may be effective for screening cows during the dry off for susceptibility to SCM and evaluating the health status of udder in dairy cows in the future. Results also help in better understanding the metabolic status and pathomechanisms of the disease.