The microbiota of ripening cheese is dominated by lactic acid bacteria, which are either added as starters and adjunct cultures or originate from the production and processing environments (nonstarter or NSLAB). After curd formation and pressing, starters reach high numbers, but their viability then decreases due to lactose depletion, salt addition, and low pH and temperature. Starter autolysis releases cellular contents, including nutrients and enzymes, into the cheese matrix. During ripening, NSLAB may attain cell densities up to 8 log cfu per g after 3 to 9 mo. Depending on the species and strain, their metabolic activity may contribute to defects or inconsistency in cheese quality and to the development of typical cheese flavor. The availability of gene and genome sequences has enabled targeted detection of specific cheese microbes and their gene expression over the ripening period. Integrated systems biology is needed to combine the multiple perspectives of post-genomics technologies to elucidate the metabolic interactions among microorganisms. Future research should delve into the variation in cell physiology within the microbial populations, because spatial distribution within the cheese matrix will lead to microenvironments that could affect localized interactions of starters and NSLAB. Microbial community modeling can contribute to improving the efficiency and reduce the cost of food processes such as cheese ripening.

The microbial community of industrially produced Canadian Cheddar cheese was examined from curd to ripened cheese at 30–32 months using a combination of viable plate counts of SLAB (GM17) and NSLAB (MRSv), qPCR and 16S rRNA gene amplicon sequencing. Cell treatment with propidium monoazide excluded DNA of permeable cells from amplification. The proportion of permeable cells of both Lactococcus spp. and Lacticaseibacillus spp. was highest at 3–6 months. While most remaining Lacticaseibacillus spp. cells were intact during later ripening stages, a consistent population of permeable Lactococcus spp. cells was maintained over the 32-month period. While Lactococcus sequence variants were significant biomarkers for viable cheese curd communities at 0–1 m, Lacticaseibacillus was identified as a distinctive biomarker for cheeses from 7 to 20 months. From 24 to 32 months, Lacticaseibacillus was replaced in significance by four genera (Pediococcus and Latilactobacillus at 24 m and at 30–32 m, Secundilactobacillus and Paucilactobacillus). These results underscore the importance of monitoring potential defects in cheeses aged over 24 months, which could be diagnosed early through microbial DNA profiling to minimize potential waste of product. Future perspectives include correlating volatile flavor compounds with microbial community composition as well as the investigation of intra-species diversity.

Shotgun metagenomic sequencing was used to investigate the diversity of the microbial community of Cheddar cheese ripened over 32 months. The changes in taxa abundance were compared from assembly-based, non-assembly-based, and mOTUs2 sequencing pipelines to delineate the community profile for each age group. Metagenomic assembled genomes (MAGs) passing the quality threshold were obtained for 11 species from 58 samples. Although Lactococcus cremoris and Lacticaseibacillus paracasei were dominant across the shotgun samples, other species were identified using MG-RAST. NMDS analysis of the beta diversity of the microbial community revealed the similarity of the cheeses in older age groups (7 months to 32 months). As expected, the abundance of Lactococcus cremoris consistently decreased over ripening, while the proportion of permeable cells increased. Over the ripening period, the relative abundance of viable Lacticaseibacillus paracasei progressively increased, but at a variable rate among trials. Reads attributed to Siphoviridae and Ascomycota remained below 1% relative abundance. The functional profiles of PMA-treated cheeses differed from those of non-PMA-treated cheeses. Starter rotation was reflected in the single nucleotide variant profiles of Lactococcus cremoris (SNVs of this species using mOTUs2), while the incoming milk was the leading factor in discriminating Lacticaseibacillus paracasei/casei SNV profiles. The relative abundance estimates from Kraken2, non-assembly-based (MG-RAST) and marker gene clusters (mOTUs2) were consistent across age groups for the two dominant taxa. Metagenomics enabled sequence variant analysis below the bacterial species level and functional profiling that may affect the metabolic interactions between subpopulations in cheese during ripening, which could help explain the overall flavour development of cheese. Future work will integrate microbial variants with volatile profiles to associate the development of compounds related to cheese flavour at each ripening stage.