Milk microbiome diversity and bacterial group prevalence in a comparison between healthy Holstein Friesian and Rendena cows

Abstract

Dry and early lactation periods represent the most critical phases for udder health in cattle, especially in highly productive breeds, such as the Holstein Friesian (HF). On the other hand, some autochthonous cattle breeds, such as the Rendena (REN), have a lower prevalence of mastitis and other transition-related diseases. In this study, milk microbiota of 6 HF and 3 REN cows, all raised on the same farm under the same conditions, was compared. A special focus was placed on the transition period to define bacterial groups' prevalence with a plausible effect on mammary gland health. Four time points (dry-off, 1 d, 7-10 d and 30 d after calving) were considered. Through 16S rRNA sequencing, we characterized the microbiota composition for 117 out of the 144 milk samples initially collected, keeping only the healthy quarters, in order to focus on physiological microbiome changes and avoid shifts due to suspected diseases. Microbial populations were very different in the two breeds along all the time points, with REN milk showing a significantly lower microbial biodiversity. The taxonomic profiles of both cosmopolitan and local breeds were dominated by Firmicutes, mostly represented by the Streptococcus genus, although in very different proportions (HF 27.5%, REN 68.6%). Large differences in HF and REN cows were, also, evident from the metabolic predictive analysis from microbiome data. Finally, only HF milk displayed significant changes in the microbial composition along the transition period, while REN maintained a more stable microbiota. In conclusion, in addition to the influence on the final characteristics of dairy products obtained from milk of the two breeds, differences in the milk microbiome might, also, have an impact on their mammary gland health.

Fig 1. Alpha and beta-diversity among HF (red) and REN (blue).

Rarefaction at 35,959 sequences per sample. Alpha-diversity average indexes (plus standard error bars) for phylogenetic diversity Chao1 (A) and observed species (B) are reported for HF and REN milk samples. Diversity among breeds is statistically significant in all the metrics (including Shannon index, not shown), p-value = 0.001. Beta-diversity analysis is represented by PCoA graphs of weighted UniFrac distance between HF and REN along the principal components (C-D). Each dot represents a single quarter milk sample, while the centroids represent their average value. Separation among the centroids is statistically significant (p-value < 0.001). Percent variance accounted for by the first, second and third principal component is shown along the axis.

Fig 2

Distribution of the sequence relative abundances summarized at phylum (A) and family (B) levels. Relative proportions of bacterial taxonomic groups that were present in at least 1% relative abundance in quarter milk samples at a rarefaction depth of 35,959 sequences. All bacterial taxa present at less than 1% relative abundance were grouped into the “Other” classification.

Fig 3. Bubble graph illustrating the groups significantly different between the two breeds (HF and REN) at genus level.

X-axis reports the log₂ ratio (REN/HF) of relative abundances; Y-axis depicts the–log₁₀(p-value) of the two-sided Mann-Whitney U-test for comparing bacterial groups; bubble dimension is related to the average relative abundance of sequences; color code is according to Cohen's size effect. Bacterial groups namely indicated are the ones with relative abundance > 1%, p-value < 0.05 and log₂(ratio) > 1.5.

Fig 4. PCoA of weighted UniFrac distances representing the differences in milk microbiota structure along time points.

Each dot represents a single quarter milk sample, while the centroid represents its average value. Percent variance accounted for by the second and third principal component is shown along the axis. (A) PCoA plots for HF; P-values are statistically significant (p < 0.02) for all pairwise comparison, except T2 vs. T3. (B) PCoA plots for REN; P-values are not statistically significant (p > 0.05).

Fig 5. Functional comparison among HF and REN milk microbiota.

(A) PCA of HF (red) and REN (blue) samples based on level 3 KEGG predicted pathways; the difference between breeds is highly significant (p = 0.001). Each dot represents a single quarter milk sample. Percent variance accounted for by the first and third principal component is shown. (B) Dot plot showing the specific level 3 KEGG predicted pathways that are enriched in REN and HF milk quarter samples. Most abundant gene categories for each breed were sorted out and the ratio between their averages was calculated. Only the first 20 significantly different gene categories between cow breeds (p-value < 0.05) are shown.