Raw milk microbiota are complex communities with a significant impact on the hygienic, sensory and technological quality of milk products. However, there is a lack of knowledge on factors determining their composition. In the present study, four bulk tank milk samples of two farms at two different time points were analyzed in detail for their microbiota using cultivation and 16S rRNA amplicon sequencing. Diversity in samples from the first time point was assessed via cultivation of 500 aerobic mesophilic bacterial isolates in each sample. A high biodiversity of 70 and 110 species per sample was determined, of which 25–28% corresponded to yet unknown taxa. The isolates were dominated by Gram-positive members of the genera Staphylococcus, Corynebacterium, Streptococcus, or Janibacter, whilst Chryseobacterium and Acinetobacter were most abundant among the Gram-negative taxa. At the second time point, samples of the same farms were analyzed via both cultivation (1,500 individual colonies each) and high-throughput 16S rRNA gene amplicon sequencing. The latter revealed a threefold higher biodiversity at the genus level, as anaerobic or fastidious species were also detected. However, cultivation identified genera not captured by sequencing, indicating that both approaches are complementary. Using amplicon sequencing, the relative abundance of a few genera was distorted, which seems to be an artifact of sample preparation. Therefore, attention needs to be paid to the library preparation procedure with special emphasis on cell lysis and PCR.

Eight facultatively anaerobic rod-shaped bacteria were isolated from raw milk and two other dairy products. Results of phylogenetic analyses based on 16S rRNA gene sequences showed that the isolates are placed in a distinct lineage within the familyPropionibacteriaceaewithPropioniciclava sinopodophylliandPropioniciclava tardaas the closest relatives (94.6 and 93.5 % similarity, respectively). The cell-wall peptidoglycan containedmeso-diaminopimelic acid, alanine and glutamic acid and was of the A1γ type (meso-DAP-direct). The major cellular fatty acid was anteiso-C_15 : 0and the major polar lipids were diphosphatidylglycerol, phosphatidyglycerol and three unidentified glycolipids. The quinone system contained predominantly menaquinone MK-9(H₄). The G+C content of the genomic DNA of strain VG341^Twas 67.7 mol%. The whole-cell sugar pattern contained ribose, rhamnose, arabinose and galactose. On the basis of phenotypic and genetic data, eight strains (VG341^T, WS4684, WS4769, WS 4882, WS4883, WS4901, WS4902 and WS4904) are proposed to be classified as members of a novel species in a new genus of the familyPropionibacteriaceae, for which the nameBrevilactibacter flavusgen. nov., sp. nov. is proposed. The type strain is VG341^T(=WS4900^T=DSM 100885^T=LMG 29089^T) and seven additional strains are WS4684, WS4769, WS4882, WS4883, WS4901, WS4902 and WS4904. Furthermore, we propose the reclassification ofP. sinopodophylliasBrevilactibacter sinopodophyllicomb. nov.

Mass spectrometry-based proteomics has revolutionized bacterial identification and elucidated many molecular mechanisms underlying bacterial growth, community formation, and drug resistance. However, most research has been focused on a few model bacteria, overlooking bacterial diversity. In this study, we present the most extensive bacterial proteomic resource to date, covering 303 species, 119 genera, and five phyla with over 636,000 unique expressed proteins, confirming the existence of over 38,700 hypothetical proteins. Accessible via the public resource ProteomicsDB, this dataset enables quantitative exploration of proteins within and across species. Additionally, we developed MS2Bac, a bacterial identification algorithm that queries NCBI's bacterial proteome space in two iterations. MS2Bac achieved over 99% species-level and 89% strain-level accuracy, surpassing methods like MALDI-TOF and FTIR, as demonstrated with food-derived bacterial isolates. MS2Bac also effectively identified bacteria in clinical samples, highlighting the potential of MS-based proteomics as a routine diagnostic tool.

Full-length SSU rRNA gene sequencing allows species-level identification of the microorganisms present in milk samples. Here, we used bulk-tank raw milk samples of two German dairies and detected, using this method, a great diversity of bacteria, archaea, and yeasts within the samples. Moreover, the species-level classification was improved in comparison to short amplicon sequencing. Therefore, we anticipate that this approach might be useful for the detection of possible mastitis-causing species, as well as for the control of spoilage-associated microorganisms. In a proof of concept, we showed that we were able to identify several putative mastitis-causing or mastitis-associated species such as Streptococcusuberis, Streptococcusagalactiae, Streptococcusdysgalactiae, Escherichiacoli and Staphylococcusaureus, as well as several Candida species. Overall, the presented full-length approach for the sequencing of SSU rRNA is easy to conduct, able to be standardized, and allows the screening of microorganisms in labs with Illumina sequencing machines.

The highly complex raw milk matrix challenges the sample preparation for amplicon-sequencing due to low bacterial counts and high amounts of eukaryotic DNA originating from the cow. In this study, we optimized the extraction of bacterial DNA from raw milk for microbiome analysis and evaluated the impact of cycle numbers in the library-PCR. The selective lysis of eukaryotic cells by proteinase K and digestion of released DNA before bacterial lysis resulted in a high reduction of mostly eukaryotic DNA and increased the proportion of bacterial DNA. Comparative microbiome analysis showed that a combined enzymatic and mechanical lysis procedure using the DNeasy^® PowerFood^® Microbial Kit with a modified protocol was best suitable to achieve high DNA quantities after library-PCR and broad coverage of detected bacterial biodiversity. Increasing cycle numbers during library-PCR systematically altered results for species and beta-diversity with a tendency to overrepresentation or underrepresentation of particular taxa. To limit PCR bias, high cycle numbers should thus be avoided. An optimized DNA extraction yielding sufficient bacterial DNA and enabling higher PCR efficiency is fundamental for successful library preparation. We suggest that a protocol using ethylenediaminetetraacetic acid (EDTA) to resolve casein micelles, selective lysis of somatic cells, extraction of bacterial DNA with a combination of mechanical and enzymatic lysis, and restriction of PCR cycles for analysis of raw milk microbiomes is optimal even for samples with low bacterial numbers. Key points: • Sample preparation for high-throughput 16S rRNA gene sequencing of raw milk microbiota. • Reduction of eukaryotic DNA by enzymatic digestion. • Shift of detected microbiome caused by high cycle numbers in library-PCR.