Concurrent and long term associations between the endometrial microbiota and endometrial transcriptome in postpartum dairy cows

Moore et al BMC Genomics (2019) 20:405 https://doi.org/10.1186/s12864-019-5797-8 RESEARCH ARTICLE Open Access Concurrent and long-term associations between the endometrial microbiota and endometrial transcriptome in postpartum dairy cows Stephen G Moore1,2, Aaron C Ericsson3,4, Susanta K Behura1,5, William R Lamberson1, Timothy J Evans3, Matthew S McCabe6, Scott E Poock7 and Matthew C Lucy8* Abstract Background: Fertility in dairy cows depends on ovarian cyclicity and on uterine involution Ovarian cyclicity and uterine involution are delayed when there is uterine dysbiosis (overgrowth of pathogenic bacteria) Fertility in dairy cows may involve a mechanism through which the uterine microbiota affects ovarian cyclicity as well as the transcriptome of the endometrium within the involuting uterus The hypothesis was that the transcriptome of the endometrium in postpartum cows would be associated with the cyclicity status of the cow as well as the microbiota during uterine involution The endometrium of first lactation dairy cows was sampled at 1, 5, and weeks postpartum All cows were allowed to return to cyclicity without intervention until week and treated with an ovulation synchronization protocol so that sampling at week was on day 13 of the estrous cycle The endometrial microbiota was measured by 16S rRNA gene sequencing and principal component analysis The endometrial transcriptome was measured by mRNA sequencing, differential gene expression analysis, and Ingenuity Pathway Analysis Results: The endometrial microbiota changed from week to week but the week and week microbiota were similar The endometrial transcriptome differed for cows that were either cycling or not cycling at week and cyclicity status depended in part on the endometrial microbiota Compared with cows cycling at week 5, there were large changes in the transcriptome of cows that progressed from non-cycling at week to cycling at week There was evidence for concurrent and longer-term associations between the endometrial microbiota and transcriptome The week endometrial microbiota had the greatest effect on the subsequent endometrial transcriptome and this effect was greatest at week and diminished by week Conclusions: The cumulative response of the endometrial transcriptome to the microbiota represented the combination of past microbial exposure and current microbial exposure The endometrial transcriptome in postpartum cows, therefore, depended on the immediate and longer-term effects of the uterine microbiota that acted directly on the uterus There may also be an indirect mechanism through which the microbiome affects the transcriptome through the restoration of ovarian cyclicity postpartum Keywords: Uterus, Bacteria, Gene expression, Ovarian cyclicity, Progesterone, Microbiome, Principal component analysis, Limma, RNAseq, 16S rRNA * Correspondence: lucym@missouri.edu Division of Animal Sciences, 160 Animal Science Research Center, 920 East Campus Drive, University of Missouri, Columbia, MO 65211, USA Full list of author information is available at the end of the article © The Author(s) 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated Moore et al BMC Genomics (2019) 20:405 Background Female fertility in dairy and beef cattle is an important component of agricultural productivity and farm profitability worldwide [1] For cows after calving, conception rate after insemination progressively improves for the first 60 days postpartum The improvement in conception rate over time depends on the resumption of ovarian cyclicity and the involution of the uterus Uterine involution occurs coincident with the proliferation of bacteria in the uterus during the first week after calving The presence of bacteria in the uterus is normal in postpartum cows [2–4] and there are bacteria from the external environment and also bacteria remnant from the previous pregnancy that contribute to the microbiota postpartum [5, 6] Overgrowth of pathogenic bacteria (dysbiosis) may occur leading to metritis; a disease defined by a foul smelling uterine discharge, fever, and loss of appetite within the first week postpartum Ten to 20% of dairy cows fail to begin cycling by the start of the breeding period [7, 8] Once they begin cycling and are inseminated, these late-cycling cows will have decreased conception rates and increased pregnancy loss [7–9] Cycles before breeding, therefore, may condition the uterus and conceptus through a mechanism involving estradiol and progesterone (P4) from the cyclic ovary [10] The objective of this study was to explore the relationships between the endometrial microbiota and the endometrial transcriptome in cows that were or were not cycling by weeks postpartum We performed 16S rRNA gene sequencing of endometrial bacteria at 1, and weeks postpartum and tested for both concurrent and long-term associations of the microbiota with the endometrial transcriptome The hypothesis was that the endometrial microbiota would be associated with the differential gene expression within the transcriptome of the postpartum endometrium and that effects of cyclicity on the endometrium would also be identified There was a major effect of cyclicity status on the endometrial transcriptome We also found evidence for both a concurrent and long-term association of the endometrial microbiota with the endometrial transcriptome and also associations of the microbiome with the resumption of ovarian cyclicity This later observation may implicate an indirect mechanism through which the endometrial microbiota can mediate endometrial function through a pathway that involves restoration of ovarian cyclicity postpartum The endometrial transcriptome in postpartum cows, therefore, depends on the immediate and longer-term effects of the uterine microbiota that act either directly or through an indirect mechanism that involves the restoration of ovarian cyclicity postpartum Results Composition of the endometrial microbiota Average number of 16S rRNA sequence reads The average number of 16S rRNA sequence reads generated from Page of 20 the endometrial microbiota of the postpartum cows was greater for week compared with either week (P = 0.01) or week (P = 0.03) and was similar for week and week (P = 0.25) There was a week ovarian cyclicity status x week postpartum interaction (P = 0.01) for the average number of 16S rRNA sequence reads Compared with the week cycling (CycW5) cows, the week non-cycling (NoCycW5) cows had a greater number of 16S rRNA sequence reads on week [5302 (1926–14,598) vs 1134 (384–3351) reads; LSM with 95% CI; P = 0.04] The CycW5 and NoCycW5 cows had a similar number of reads on week [211 (77–582) vs 848 (287–2509) reads; P = 0.07] and week [496 (180–1366) vs 1449 (490– 4283) reads; P = 0.15] Relative abundance of operational taxonomic units (OTU) The relative abundance of each OTU is provided in Additional file Table S1 Only OTU with an average relative abundance of at least 1% were retained for analysis Eleven, 5, and OTU with a mean relative abundance of 56, 7, and 5% were unique to week 1, 5, and 9, respectively (Fig 1) Seventeen OTU were unique to week and 9, and their combined relative abundance increased from 10.9% on week to 20.0% on week (P = 0.004) Four OTU (Bacteroidales S24–7, Lachnospiraceae NK4A136, Clostridium sensu stricto 1, and Ruminococcaceae UCG-005) were present on each week and their combined relative abundance was greater on week (19.8%) and (20.3%) compared with week (8.8%; P = 0.004) When principal components (PC) were generated for the postpartum microbiota on week 1, 5, and 9, a plot of PC1 vs PC2 illustrated the separation of the week endometrial microbiota from both week and (Fig 2) Permanova analysis indicated that the Bray-Curtis similarity index (a measure of microbial similarity) was significantly different between week and week (P = 0.0001) and week and (P = 0.0001) The Permanova analysis did not detect a difference between week and microbiota (P = 0.45) The CycW5 and NoCycW5 cows differed with respect to the relative abundance of OTU at certain sampling times The relative abundance of Actinobacillus seminis on week was greater in the NoCycW5 cows compared with the CycW5 cows (29.9% vs 13.6%, Wilcoxon P = 0.049) The relative abundance of Gammaproteobacteria_B38 on week (0.2% vs 1.9%, Wilcoxon P = 0.004) and (0.7% vs 2.6%, Wilcoxon P = 0.005) was greater in NoCycW5 cows compared with the CycW5 cows The relative abundance of Aeromonas on week was greater in the NoCycW5 cows compared with the CycW5 cows (8.6% vs 0.6%, Wilcoxon P = 0.003) Subsequent analyses were performed on the microbiota for each individual week The PC1 and PC2 on each week were the focus of the current study because they explained most of the variation in the microbiota Moore et al BMC Genomics (2019) 20:405 Page of 20 Fig Venn diagram of the endometrial microbiota with a relative abundance greater than 1% at week 1, 5, and postpartum in lactating dairy cows Eleven, 5, and operational taxonomic units (OTU) with a mean relative abundance of 56, 7, and 5% were unique at week 1, 5, and 9, respectively Seventeen OTU were unique at week and 9, and their combined relative abundance increased from 10.9% on week to 20.0% on week (P = 0.004) Four OTU were present on each week and their combined relative abundance was greater on week (19.8%) and (20.3%) compared with week (8.8%; P = 0.004) The PC1 and PC2 for week (WK1_PC1 microbiota, WK1_PC2 microbiota), week (WK5_PC1 microbiota, W5_PC2 microbiota), and week (WK9_PC1 microbiota, and WK9_PC2 microbiota) accounted for 27.5, 16.9, 26.0, 11.4, 38.4, and 13.9% of the variation within each week, respectively The loading plot for each PC was examined to identify the most influential OTU (Figs 3, 4, 5) The top loadings for WK1_PC1 microbiota were Fusobacteriales bone C3G7 and Porphyromonas (Fig 3), and for WK1_PC2 microbiota were Actinobacillus seminis, Bacteroides, Fusobacteriales bone C3G7, and Helcococcus ovis (Fig 3) No predominant loadings were identified for the WK5_ PC1 microbiota, WK5_PC2 microbiota (Fig 4), and WK9_PC1 microbiota (Fig 5) The top loadings for WK9_PC2 microbiota were Aeromonas and Acinetobacter Ziziphus jujuba (Fig 5) Effect of ovarian cyclicity and the week and week microbiota on the endometrial transcriptome at weeks postpartum There were 809 genes differentially expressed between the CycW5 cows and the NoCycW5 cows on week (BH P ≤ 0.05; Fig 6; Additional file 1: Table S2) Based on the Ingenuity Pathway Analysis (IPA) of the differentially expressed genes (DEG), the CycW5 cows had a down-regulation of the canonical pathway EIF2 signaling (i.e., protein synthesis) and a down-regulation of the biological functions cell death, necrosis, astrocytosis, and liver lesion compared with the NoCycW5 cows (Table 1) There was an up-regulation in the CycW5 cows of the biological functions transport of molecule, transport of lipid, oxidation of lipid, quantity of carbohydrate, cell-to-cell contact, fusion of cells, formation of muscle, and head and neck cancer (Table 1) Target molecules of differentially expressed genes in the CycW5 cows were involved in increased molecule transport of glutamine (SLC38A1, BH P = 0.007), arginine, lysine, ornithine (SLC7A4, BH P = 0.009 and SLC25A9, BH P = 0.02), zinc (SLC30A5, BH P = 0.04 and SLC30A6, BH P = 0.04), copper (SLC31A2 BH P = 0.02), thiamine (SLC19A2, BH P = 0.005), and sialic acid (SLC17A5, BH P = 0.04) There were 516 genes differentially expressed (BH P ≤ 0.05; Fig 6; Additional file 1: Table S3) on week in response to the first PC derived from the microbiota present at week (WK1_PC1 microbiota; Fig 3) Moore et al BMC Genomics (2019) 20:405 Page of 20 Fig Plot of principal components PC1 vs PC2 generated from the endometrial microbiota at week 1, 5, and postpartum Separation of the week endometrial microbiota from both week and is evident Permanova analysis indicated that the Bray-Curtis similarity index (a measure of microbial similarity) was significantly different between week and (P = 0.0001) and between week and (P = 0.0001) and similar between week and (P = 0.45) Each 95% concentration ellipse estimates a region where 95% of population points are expected to fall These DEG represent the association between the microbiota present on week and the endometrial transcriptome present weeks later (week 5) Based on the IPA, the WK1_PC1 microbiota was associated with a down-regulation of the biological functions organization of cytoplasm, organization of cytoskeleton, proliferation of neuroblastoma cell lines, formation of muscle cells, and formation of muscle, and an up- regulation of the biological functions thoracic hypoplasia, cardiac lesion, and fibrosis of heart (Table 1) There were 151 genes differentially expressed on week in response to the second PC derived from the microbiota present at week [WK1_PC2 microbiota (Fig 3); BH P ≤ 0.05; Additional file 1: Table S4] The IPA did not predict significant activation of any pathways or biological processes Fig Plot of loadings from the endometrial microbiota principal components WK1_PC1 and WK1_PC2 Moore et al BMC Genomics (2019) 20:405 Fig Plot of loadings from the endometrial microbiota principal components WK5_PC1 and WK5_PC2 Fig Plot of loadings from the endometrial microbiota principal components WK9_PC1 and WK9_PC2 Page of 20 Moore et al BMC Genomics (2019) 20:405 Page of 20 Fig Numbers of differentially expressed genes in endometrium of lactating dairy cows at weeks and postpartum associated with ovarian cyclicity status (cycling at week 5, non-cycling at week 5), and the endometrial microbiota present at week (WK1_PC1, WK1_PC2), week (WK5_PC1, WK5_PC2), and week (WK9_PC1, WK9_PC2) There were 96 genes differentially expressed (BH P ≤ 0.05; Fig 6; Additional file 1: Table S5) in response to the first PC derived from the microbiota present at week (WK5_PC1 microbiota; Fig 6) These DEG represent the association between the microbiota present on week and endometrial transcriptome at the same time Based on the IPA, the WK5_PC1 microbiota was associated with down-regulation of the biological functions organismal death and thoracic hypoplasia on week There were 65 genes differentially expressed (BH P ≤ 0.05; Fig 6; Additional file 1: Table S6) in the week transcriptome in response to the second PC derived from the microbiota present at week (WK5_PC2 microbiota; Fig 4) but the IPA did not predict significant activation of any pathways or biological processes Changes in the endometrial transcriptome from week to week (day 13 of the estrous cycle) for cows that were either cycling or non-cycling at week Temporal changes in the endometrial transcriptome between week and week were dependent on the ovarian cyclicity status at week (Fig 7) Cows that were non-cycling at week (NoCycW5) were all cycling on day 13 of the estrous cycle when sampled at week There were 1489 genes differentially expressed between week and week in the NoCycW5 cows (BH P ≤ 0.05; Fig 8; Additional file 1: Table S7) The IPA predicted a downregulation of the canonical pathway EIF2 signaling, a down-regulation of biological functions recruitment of cells, branching of epithelial tissue, quantity of Ca2+, and accumulation of carbohydrate at week compared with week (Table 2) There was an up-regulation of the biological functions cell death of tumor cells, cancer, conversion of fatty acid, metabolism of membrane lipid derivative, production of reactive oxygen species, transport of molecule, and quantity of connective tissue at week compared with week in the NoCycW5 cows (Table 2) Compared with the previous comparison, cows that were cycling at week and week had numerically fewer DEG between week and (n = 145; BH P ≤ 0.05; Fig 7; Additional file 1: Table S8) The IPA predicted a down-regulation of the biological function apoptosis and up-regulation of the biological function inflammation of lung and synthesis of lipid on week compared with week for cows that were cycling at week in the CycW5 cows (Table 2) Of the 1489 and 145 genes differentially expressed between week and week in the NoCycW5 cows and the CycW5 cows, respectively, there was an overlap of 39 genes that represented and 27% of the respective gene lists The fold change in gene expression in response to cyclicity status was highly correlated (R2 = 0.79; Additional file 2: Figure S1) Effect of week ovarian cyclicity status and the week 1, week 5, and week microbiota on the endometrial transcriptome at weeks postpartum (day 13 of the estrous cycle) The endometrial transcriptome on week (d 13 of the estrous cycle) was affected by the cyclicity status of the cows at week (CycW5 versus NoCycW5) as well as the microbiota present at week (8 weeks earlier), week (4 weeks earlier), and week (present at the time of tissue collection) (Fig 6) There were 81 genes differentially expressed at week between the CycW5 cows and the NoCycW5 cows (BH P ≤ 0.05; Fig 6; Additional file 1: Table S9) These DEG Moore et al BMC Genomics (2019) 20:405 Page of 20 Table Ingenuity Pathway Analysis of endometrial genes differentially expressed at week Pathway or Biological Function Z-score P-value Cyclic vs non-cyclic (n = 809 DEG) Canonical Pathway - Decreased EIF2 signaling (i.e., protein synthesis) 6.16E-10 Biological Functions - Increased Transport of molecule 2.809 5.01E-05 Transport of lipid 2.753 7.24E-03 Oxidation of lipid 2.325 5.03E-03 Quantity of carbohydrate 2.057 1.72E-04 Cell-cell contact 2.295 5.25E-03 Fusion of cells 2.266 2.02E-03 Formation of muscle 2.209 3.63E-03 Head and neck cancer 2.4 6.69E-03 Cell death −2.193 2.11E-03 Necrosis − 2.493 2.45E-03 Astrocytosis −2.353 3.53E-03 Liver lesion −2.452 1.71E-04 Thoracic hypoplasia 2.44 1.73E-05 Cardiac lesion 2.423 8.65E-03 Fibrosis of heart 2.241 6.66E-03 Organization of cytoplasm −2.383 1.32E-03 Organization of cytoskeleton −2.383 1.41E-03 Proliferation of neuroblastoma cell lines −2.412 6.95E-03 Formation of muscle cells −2.607 4.07E-04 Formation of muscle −2.902 2.44E-04 Biological Functions - Decreased Week microbiota (WK1_PC1; n = 516 DEG) Biological functions - Increased Biological functions - Decreased Week microbiota (WK5_PC1; n = 96 DEG) Biological functions - Decreased Organismal death −2.35 3.58E-03 Thoracic hypoplasia −2 7.07E-03 represented an association between previous ovarian cyclicity status at week and the endometrial transcriptome at week when all cows were on d 13 of the estrous cycle The analysis differs from that presented in the previous section where gene expression fold changes from week to week were presented The IPA predicted a down-regulation of the biological function migration of mononuclear leukocytes and an up-regulation of the biological functions growth of tumor, neoplasia of epithelial tissue, and binding of protein site at week in the CycW5 cows compared with the NoCycW5 cows (Table 3) Of the 809 and 81 genes differentially expressed between the CycW5 cows and the NoCycW5 cows on week and week 9, respectively, there was an overlap of 21 genes that represented and 26% of the respective gene lists The fold change in gene expression in response to cyclicity status was highly correlated (R2 = 0.97; Additional file 2: Figure S2) There were 64 genes that were differentially expressed (BH P ≤ 0.05; Additional file 1: Table S10) in response to the WK1_PC2 microbiota (Fig 6) on week These DEG represent the association between the second PC of the week endometrial microbiota and the endometrial transcriptome on week The IPA predicted upregulation of biological functions inflammatory response, influx of myeloid cells, and quantity of leukocytes (Table 3) A variety of up-stream regulators for the observed gene expression signature associated with WK1_ PC2 microbiota were predicted based on the IPA (Table 4) These upstream regulators included response to bacterial products [lipopolysaccharide, endotoxin B, peptidoglycan, and 5-O-mycolyl-β-Araf-(1 → 2)-5-O-mycolyl-α-Araf-(1 → 1′)-glycerol (designated as Mma_DMAG)], interleukins (IL1, IL1A, IL1B, IL6, IL10RA, IL17A), tumor necrosis factor, and toll-like receptors (TLR3, TLR4, TLR5, TLR7, TLR9) There were 179 genes differentially expressed (BH P ≤ 0.05; Additional file 1: Table S11) in response to the WK9_PC2 microbiota (Fig 6) These DEG represent the association between the second PC of the week endometrial microbiota and the endometrial transcriptome on week The IPA predicted down-regulation of the canonical pathway cAMP-mediated signaling and the biological functions synthesis of terpenoid, angiogenesis, and cytostasis and up-regulation of the biological function malignant solid tumor (Table 3) There were 80, 158, 15, and 25 genes differentially expressed (BH P ≤ 0.05) on week in response to WK1_PC1 microbiota (Additional file 1: Table S12), WK5_PC1 microbiota (Additional file 1: Table S13), WK5_PC2 microbiota (Additional file 1: Table S14), and WK9_PC1 microbiota (Additional file 1: Table S15), respectively (Fig 6) The IPA did not predict significant activation of any pathways or biological processes for these PC Of the 516 and 80 genes differentially expressed on week and week in response to WK1_PC1, there was an overlap of 27 genes that represented and 34% of the respective gene lists The fold change in gene expression in response to WK1_PC1 microbiota at week for these overlapping genes was highly correlated (R2 = 0.97; Additional file 2: Figure S3) A similar analyses identified a minimal overlap (< genes) when the effect of WK1_ PC2 microbiota, WK5_PC1 microbiota, and WK5_PC2 microbiota on differential gene expression at week and week was assessed  ... tissue, and binding of protein site at week in the CycW5 cows compared with the NoCycW5 cows (Table 3) Of the 809 and 81 genes differentially expressed between the CycW5 cows and the NoCycW5 cows. .. tested for both concurrent and long- term associations of the microbiota with the endometrial transcriptome The hypothesis was that the endometrial microbiota would be associated with the differential... transcriptome in postpartum cows, therefore, depends on the immediate and longer -term effects of the uterine microbiota that act either directly or through an indirect mechanism that involves the restoration