Genome-Scale Metabolic Models Guided Improvement of Fermented Milk Quality and Flavor by Lacticaseibacillus paracasei subsp. paracasei 63

The quality and flavor of probiotic fermented milk are highly dependent on the strain composition of the starter culture and their metabolic interactions. Although constructing a multi-strain system is an effective strategy for enhancing product quality, traditional formulation methods rely heavily on empirical approaches and lack mechanistic guidance. Herein, this study utilized genome-scale metabolic models (GEMs) to rationally design a multi-strain co-fermentation system. The results demonstrated that the GEM-predicted optimal system, comprising Lacticaseibacillus paracasei subsp. paracasei 63 (L. paracasei subsp. paracasei 63) and Lactococcus cremoris 290 (Lc. cremoris 290), significantly reduced the curd time by approximately 44.0% and 71.0% compared to the L. paracasei subsp. paracasei 63 and Lc. cremoris 290 monocultures, respectively. Furthermore, the co-fermented milk exhibited a 4.3-fold increase in apparent viscosity relative to the 290 single-strain group and achieved a significantly higher diacetyl concentration (1.98 ± 0.09 mg/L), representing a 2.8-fold enhancement. Volatile flavor profiling and untargeted metabolomics provided suggestive evidence supporting the GEM-predicted cross-feeding mechanisms, particularly within the arginine and pyruvate metabolic pathways. This study offers a solid theoretical foundation and practical guidance for the rational design of synthetic microbial communities to develop high-quality fermented dairy products with optimized flavor and functional properties.