Microbiota and Resistome Analysis of Colostrum and Milk from Dairy Cows Treated with and without Dry Cow Therapies

Abstract

This study investigated the longitudinal impact of methods for the drying off of cows with and without dry cow therapy (DCT) on the microbiota and resistome profile in colostrum and milk samples from cows. Three groups of healthy dairy cows (n = 24) with different antibiotic treatments during DCT were studied. Colostrum and milk samples from Month 0 (M0), 2 (M2), 4 (M4) and 6 (M6) were analysed using whole-genome shotgun-sequencing. The microbial diversity from antibiotic-treated groups was different and higher than that of the non-antibiotic group. This difference was more evident in milk compared to colostrum, with increasing diversity seen only in antibiotic-treated groups. The microbiome of antibiotic-treated groups clustered separately from the non-antibiotic group at M2-, M4- and M6 milk samples, showing the effect of antibiotic treatment on between-group (beta) diversity. The non-antibiotic group did not show a high relative abundance of mastitis-causing pathogens during early lactation and was more associated with genera such as Psychrobacter, Serratia, Gordonibacter and Brevibacterium. A high relative abundance of antibiotic resistance genes (ARGs) was observed in the milk of antibiotic-treated groups with the Cephaguard group showing a significantly high abundance of genes conferring resistance to cephalosporin, aminoglycoside and penam classes. The data support the use of non-antibiotic alternatives for drying off in cows.

Keywords: cow milk microbiota; longitudinal; resistome; shotgun metagenomics.

Figure 1

(A) Alpha diversity of milk microbiota using Shannon index between all time-points for each group. The plot shows significant differences in diversity between milk obtained at M0, M2, M4 and M6 for both the antibiotic-treated groups (UBRO and CEF). (B) Alpha diversity using Shannon index between the three groups (NOAB, UBRO and CEF) at all time-points. Plot shows no significant difference between groups at M0, while significance between antibiotic and no-antibiotic group was seen at later time points. (C) PCoA plot using Bray–Curtis distance matrix between the three groups at all time-points, showing distinct clustering between groups. (D) PCoA plot using Bray–Curtis distances (a) showing distinct clustering of all three groups, with antibiotic groups clustering discretely compared to the no-antibiotic group; and plot (b) showing separate grouping of points between groups and time-points. * p-value ≤ 0.05; ** p-value ≤ 0.01; *** p-value ≤ 0.001.

Figure 2

(A) Phylum level distribution of taxa in all three groups across all time points. (B) Overall top 10 genera in all three groups (NOAB, UBRO and CEF) at all time-points. (C) Trend plots at genera level, showing relative abundance of top 10 genera from all three groups (NOAB, UBRO and CEF) over time with confidence interval of 95% (with different Y-axis scale for each taxa).

Figure 3

(A) Boxplots showing abundance of potential mastitis-causing pathogens in milk of all three groups (NOAB, UBRO and CEF) over time of lactation. Significance was determined using Wilcoxon test in R, and p.adj values < 0.05 are considered significant. The black dots correspond to the outliers. (B) Songbird differentials obtained using the formula C (Group, Treatment(‘NOAB’)) were sorted by ranks and the top 10 positive and negative features are plotted and depicted using bar charts. NOAB group is reference (negative here is more associated with reference group—NOAB here) and (a) and (b) are top 10 positive and negative associations with treatment group CEF and UBRO respectively.

Figure 4

(A) Boxplot showing log-transformed abundance of the different classes of antibiotics to which resistance was observed in this study. (B) Bar plot showing representation of ARGs in all three groups (NOAB, UBRO and CEF) over all time-points. The classes highlighted in blue boxes correspond to the class of antibiotics administered to the UBRO group, while those in red boxes correspond to the CEF group.

Figure 5

(A) Alpha diversity with Shannon index shows no difference between groups (NOAB, UBRO and CEF) at any time points for ARG abundance. (B) Beta diversity of ARGs using PCoA and Bray–Curtis distance. (C) Beta diversity of ARGs at each time point using PCoA plot with Bray–Curtis distance.

Figure 6

Heatmap showing log-transformed abundance (cpm) of various ARG classes per group at each timepoint. Significance was calculated with NOAB group as reference group against each antibiotic-treated group using Dunn test for multiple comparisons. P.adj values below 0.05 were considered significant and denoted with *.