Metritis and the uterine disease microbiome are associated with long-term changes in the endometrium of dairy cows†

Abstract

Cows with metritis (uterine disease) during the first 1 to 2 weeks postpartum have lower pregnancy rates when inseminated later postpartum (typically >10 weeks). We hypothesized that metritis and the disease-associated uterine microbiome have a long-term effect on endometrial gene expression. Changes in gene expression may inform a mechanism through which disease lowers pregnancy rates. A total of 20 cows were enrolled at 1 to 2 weeks postpartum to either metritis (clinical disease; n = 10) or healthy (control; n = 10) groups and randomly assigned to be slaughtered at approximately 80 and 165 dpp (mid-lactation). The microbiome of the reproductive tract was sampled to confirm the presence of pathogens that are typical of metritis. In addition to the original clinical diagnosis, study cows were retrospectively assigned to uterine-disease and control groups based on the composition of their microbiome. There was no effect of early postpartum uterine disease on the uterine microbiome at mid-lactation (time of slaughter). Nonetheless, early postpartum metritis and the disease microbiome were associated with a large number of differentially-expressed genes at mid-lactation primarily in the caruncular compared with the inter-caruncular endometrium. Gene enrichment analysis identified oxidative phosphorylation as the primary pathway increased in caruncular endometrium of diseased cows whereas growth factor signaling pathways were reduced. The current study demonstrated that metritis and a uterine disease microbiome leave a sustained imprint on gene expression in the caruncular endometrium that may explain lower fertility in cows with postpartum uterine disease.

Keywords: cow; endometrium; metritis; microbiome.

Graphical Abstract

Figure 1

Study timeline. Metritis cows were diagnosed and matched with contemporary control healthy cows at 7 to 10 dpp. Vaginal samples for microbiome analyses were collected at DD (week 1 to 2 postpartum) and during EL (weeks 4 to 5 postpartum). Two groups of cows were slaughtered at an interval that encompassed a typical breeding period (80 to 165 dpp). Samples were collected for uterine microbiome (Study 1) and uterine transcriptome (Study 2).

Figure 2

(A) Milk production, (B) BCS, and (C) plasma haptoglobin concentrations during the first 3 weeks post DD, for cows diagnosed with metritis at 7 to 10 dpp compared with healthy control. ^*P < 0.05 and ^#P < 0.10 for within week post-diagnosis comparison.

Figure 3

Changes in relative abundance within the vaginal microbiome for the most abundant genera found by metagenomic sequencing [(A) Porphyromonas, (B) Bacteroides, (C) Helcococcus, (D) Peptostreptococcus, (E) Gallicola, and (F) Clostridium]. Regardless of the genera, there was increased relative abundance in metritis cows when compared with control at DD (weeks 1 to 2 postpartum) and a decreased relative abundance over time but no differences in the vaginal microbiome at EL (weeks 4 to 5 postpartum).

Figure 4

Relative abundance within the vaginal microbiome for the genera Fusobacterium, Porphyromonas, Bacteroides, Helcococcus, Peptostreptococcus, Gallicola, and Clostridium, and all other genera for cows diagnosed with metritis compared with healthy controls at DD [weeks (A) 1 to (B) 2 postpartum] and EL [weeks (C) 4 to (D) 5 postpartum].

Figure 5

Analysis of abundance data for individual ASV using (A) Faith’s phylogenetic diversity (box plot; alpha diversity metric indicative of species richness) and (B) Jaccard PCoA plots (beta diversity metric demonstrating similarity or dissimilarity between samples) for samples collected from the external surface of the uterus, vagina, uterine body and both uterine horns (CL horn and non-CL horn) for cows following slaughter at mid-lactation (80 to 165 dpp). In A, samples collected from the outside of the tract (external) had greater (P < 0.001) alpha diversity when compared with all other sample locations. Vagina had greater alpha diversity when compared with non-CL horn (P < 0.017) but was similar to uterine body and CL horn.

Figure 6

Analysis of abundance data for individual ASV using (A) Faith’s phylogenetic diversity (box plot; alpha diversity metric indicative of species richness) and (B–D) Jaccard PCoA plots (beta diversity metric demonstrating similarity or dissimilarity between samples) for samples collected from the vagina, uterine body and both uterine horns (CL horn and non-CL horn) for cows that were initially diagnosed as clinically healthy or metritis within 8 ± 2 dpp and slaughtered at mid-lactation (80 to 165 dpp). For the PCoA plots, individual samples are classified by their location within the (B) reproductive tract, (C) by their day postpartum at collection, or (D) by the original DD.

Figure 7

Analysis of the association between microbiome PC at DD (weeks 1 to 2 postpartum) and DEG within the uterus at mid-lactation (80 to 165 dpp). The PC analysis identified PC associated with different (A) bacterial genera that were then used to separate cows into groups (diseased versus control) based on their (B) microbiome (each dot representing a different cow). The groups were then fit to a statistical model to determine DEG that were (C) increased in diseased cow caruncular endometrium or (D) decreased in diseased cow caruncular endometrium compared with control. For C and D, the ShinyGO program was used to determine the number of DEG, fold enrichment, and -log 10 of the P-value of the FDR. Data for intercaruncular endometrium are not shown because there were few DEG. See Supplementary Tables 8 and 9 for complete gene lists and results of the gene enrichment analysis.

Figure 8

Analysis of the association between microbiome PC at EL (weeks 4 to 5 postpartum) and DEG within the uterus at mid-lactation (80 to 165 dpp). The PC analysis identified PC associated with different (A) bacterial genera that were then used to separate cows into groups (diseased versus control) based on their (B) microbiome (each dot representing a different cow). The groups were then fit to a statistical model to determine DEG that were (C) increased in diseased cow caruncular endometrium or (D) decreased in diseased cow caruncular endometrium compared with control. For C and D, the ShinyGO program was used to determine the number of DEG, fold enrichment, and -log 10 of the P-value of the FDR. Data for intercaruncular endometrium are not shown because there were few DEG. See Supplementary Tables 10 and 11 for complete gene lists and results of the gene enrichment analysis.

Figure 9

Analysis of the microbiome PC at mid-lactation (80 to 165 dpp). The PC analysis identified PC associated with different (A) bacterial genera that could be plotted based on (B) PC 1 and PC 2 (each dot representing a different cow). The DEG analysis was not attempted because the PC plot (B) analysis failed to clearly separate study cows into two groups (indicating very similar microbiomes across individual animals).

Figure 10

Overlap of DEG sets from Moore et al. [41] and the present study (Silva et al.). For Moore et al. [41] the DEG were for the week 1 postpartum PC 1 and the week 9 postpartum uterine transcriptome. For Silva et al., the DEG were for DD (weeks 1 and 2 postpartum) PC 1 and mid-lactation (80 to 165 dpp) uterine (caruncular and inter-carcular endometrium combined) transcriptome. The overlap between the two genes sets included (A) 235 genes that were enriched for specific (B) KEGG pathways.

Figure 11

Reanalysis of DEG from the study of Moore et al. [64] using ShinyGO. The identified DEG were either (A) increased or (B) decreased in placenta from days 28 to 42 of pregnancy in dairy cows.