Multi-omics integration and immune profiling identify possible causal networks leading to uterine microbiome dysbiosis in dairy cows that develop metritis

Multi-omics integration and immune profiling identify possible causal networks leading to uterine microbiome dysbiosis in dairy cows that develop metritis | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Multi-omics integration and immune profiling identify possible causal networks leading to uterine microbiome dysbiosis in dairy cows that develop metritis S. Casaro, J. G. Prim, T. D. Gonzalez, F. Cunha, A. C. M. Silva, and 9 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4571697/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 09 Jan, 2025 Read the published version in Animal Microbiome → Version 1 posted 11 You are reading this latest preprint version Abstract Background Cows that develop metritis experience dysbiosis of their uterine microbiome, where opportunistic pathogens overtake uterine commensals. Whether opportunistic pathogens thrive and cause metritis may be determined by how efficiently the immune system responds to these pathogens. Nonetheless, periparturient cows experience immune dysregulation, which seems to be intensified by prepartum obesity and lipid mobilization Herein, Bayesian networks were applied to investigate the directional correlations between prepartum body weight (BW), BW loss, pre- and postpartum systemic immune profiling and plasma metabolome, and postpartum uterine metabolome and microbiome. Results According to the directionality network, greater prepartum BW led to greater BW loss, which led to an increase in plasma fatty acids at parturition, indicating that heavier cows were in lower energy balance than lighter cows. Greater prepartum BW also led to an increase in prepartum systemic cellular death, which led to an increase in systemic inflammation, immune activation, and metabolomic changes associated with oxidative stress prepartum and at parturition, which indicates a positive directional correlation between BW and systemic inflammation. These changes led to an increase in polymorphonuclear cell extravasation postpartum, which led to an increase in uterine metabolomic changes associated with tissue damage, suggesting that excessive polymorphonuclear cell migration to the uterus leads to excessive endometrial damage. These changes led to an increase in pathogenic bacteria in cows that developed metritis, suggesting that excessive tissue damage may disrupt physical barriers or increase substrate availability for bacterial growth. Conclusions This work provides robust mechanistic hypotheses for how prepartum body weight impacts peripartum immune and metabolic profiles, leading to uterine opportunistic pathogens overgrowth and metritis development. microbiome metabolome immune dysregulation multi-omics causal networks Figures Figure 1 Figure 2 Background Metritis affects 25% of dairy cows [ 1 ], affecting animal wellbeing [ 2 ] and performance, which leads to reduced dairy farm profitability [ 2 – 4 ]. The uterine microbiome from cows that develop metritis and those that remain healthy do not differ until 2 days postpartum [ 5 , 6 ], after which opportunistic pathogens, such as Fusobacterium , Porphyromonas , and Bacteroides overtake the uterine commensals [ 5 – 8 ]. It is not clear why these opportunistic pathogens thrive and cause metritis; however, the metabolic and immune changes associated with parturition and the onset of lactation seem to play an important role in metritis development [ 9 , 10 ]. Around parturition, cows that develop metritis experience a period of negative nutrient balance [ 4 ], which is more pronounced in overconditioned cows [ 11 ]. This period of negative nutrient balance leads to lesser blood calcium [ 12 ], lesser amino acids [ 13 ], and greater circulating fatty acids concentration [ 10 ]. These metabolic changes have been associated with persistent systemic inflammation, oxidative stress, cellular damage, and immune dysregulation [ 9 , 10 ]. Studies to date have only evaluated how a limited number of metabolites affect the systemic immune response in cows that develop metritis, which do not fully represent the wide variety of metabolic pathways involved in immune processes [ 14 ]. Furthermore, vascular degeneration that occurs shortly after parturition in dairy cows allows for the exchange of metabolites between blood and uterus [ 15 , 16 ], thereby raising the potential for blood metabolites to influence the uterine microbiome, or microbial-derived uterine metabolites to influence the systemic immune response. Either of these mechanisms could affect bacterial proliferation in the uterus. We have previously demonstrated that the peripartum plasma [ 13 ] and uterine [ 6 ] metabolome and immune response [ 10 ] differ between cows that develop metritis and those that do not; however, the interplay between plasma and uterine metabolome, systemic immune response, and the uterine microbiome has not been investigated. Therefore, the hypothesis of the current study was that plasma and/or uterine metabolites promote opportunistic pathogenic bacterial overgrowth in cows that develop metritis either directly or indirectly by regulating the systemic immune response. Thus, the objective of the current study was to apply directionality networks to infer possible causal relationships between the uterine and plasma metabolome, systemic immune response, and the uterine microbiome to advance the understanding of metritis development. Methods This case-control observational study was conducted at the University of Florida Dairy Unit from September 2019 to March 2020. The cows used for the current study were a subset of cows used in our previous studies [ 6 , 9 , 10 ]. All procedures involving cows were approved by the Institutional Animal Care and Use Committee of the University of Florida; protocol number 201910623. Briefly, cows were enrolled at 240 d of gestation and uterine discharge was evaluated using a Metricheck device (Metricheck, Simcro, New Zealand) at 3 ± 1, 7 ± 1, 10 ± 1 and 13 ± 1 days after parturition. Cows with a fetid red-brownish, watery discharge were classified as having metritis (n = 50) and paired with cows without metritis (n = 50). All cows had blood collected prepartum (14 ± 6 d before parturition), at parturition (first 24 hours after parturition), and at the day of metritis diagnosis (7 ± 2 d after parturition). All cows had uterine fluid collected at parturition and at diagnosis of metritis. Blood samples were used for flow cytometry, multiplex, and untargeted gas chromatography mass spectrometry. Flow cytometry and multiplex were performed to assess the systemic immune profile and activation status. Gas chromatography mass spectrometry was performed to assess the systemic metabolomic profile. Uterine fluid samples were used for untargeted gas chromatography mass spectrometry and 16S rRNA gene amplification by PCR. Gas chromatography mass spectrometry was performed to assess the uterine metabolomic profile and the 16S rRNA gene was amplified by PCR to assess the uterine microbiome. Prepartum body weight (ppBW) and prepartum daily body weight change (BWC) were evaluated. Full details can be found in SI Materials and Methods. Data Pre-processing Data pre-processing was performed using R [ 17 ]. Datasets included systemic immune data and plasma metabolome prepartum, at parturition, and on the day of metritis diagnosis from two of our previous studies [ 9 , 10 ], uterine microbial estimated counts and uterine metabolome at parturition and on the day of metritis diagnosis from our previous study [ 6 ]. Lastly, ppBW and BWC from our previous study [ 10 ]. Because Bayesian networks do not allow for variables with missing values, these were excluded from further analysis. For systemic immune data, only variables with a P ≤ 0.10 to the effect of metritis in our previous study [ 10 ] and without missing values were selected for further analysis. The variables used herein for prepartum were Live %, Monocytes %, PMN %, B-cell activation (CD62L negative), IL-8, and IFN-y. For parturition the variables were Live %, B-cell activation, Monocyte activation (increase in CD62L MFI), IL-1β, IL-8, and IFN-γ. For the day of diagnosis, the variables were Live %, Singlets %, Monocytes %, PMN %, Monocyte activation, Monocyte MHCII %, PMN activation (increase in CD62L MFI), B-cell activation, and T helper activation (reduction in CD62L MFI). To eliminate any possible effect of parity from the analyses, all variables from the datasets were divided by parity group (primiparous and multiparous), log transformed and auto scaled within parity group, and then merged back together. Latent variables identification and significance Latent variables identification and significance were performed using the mixOmics R package [ 18 ]. Because microbiome and metabolome data are composed of highly correlated variables, principal component analysis (PCA) was applied to microbiome and metabolome data for orthogonal latent variables identification. For the five metabolomics datasets, sparse PCA (sPCA) via regularized single value decomposition [ 19 ] was applied using the ‘spca’ function of the mixOmics package. Briefly, the sPCA from MixOmics is based on singular value decomposition, which is appropriate when dealing with large datasets where not all variables are likely to be equally important. The sparsity is achieved via LASSO penalization, such that latent variables are no longer a linear combination of all original variables but are a linear combination of a subset of variables that maximize the captured variance. Five latent variables were created for each dataset. For each latent variable, optimal number of variables were selected using 3-fold cross-validation with 10 repetitions. For the uterine microbiome at parturition and at the day of diagnosis a PCA was applied using the ‘pca’ function of the mixOmics package selecting the opportunistic pathogenic bacteria with the greatest relative abundance [ 6 , 8 ] and estimated counts [ 6 ] in cows with metritis compared with cows without metritis. These were Fusobacterium , Porphyromonas , and Bacteroides . One latent variable was created for parturition and one for the day of metritis diagnosis. The rationale behind this decision was to observe how different immune and metabolic variables affected the growth of these bacteria since parturition. Latent variables significance between cows that developed metritis and cows that did not develop metritis were assessed by t -tests using the t.test function of the stats R package [ 17 ]. Statistical significance was considered if P ≤ 0.10. Bayesian Networks Bayesian networks is a powerful statistical method that models conditional dependencies between variables, which can be interpreted as potential causal relationships among variables. Nonetheless, care must be taken in interpreting these directional relationships as causal because not all the assumptions for causality can be met in an observational study such as that there are no unobserved external variables affecting the variables in the network [ 20 , 21 ]. Bayesian network analysis was performed using the bnlearn R package [ 20 ]. Highly correlated variables (> 0.80 or < -0.80) were excluded to avoid multicollinearity. Prepartum B-cell activation was excluded because they were highly correlated with parturition (Pearson correlation coefficient r = 0.83) and diagnosis (r = 0.82) B-cell activation. A total of 20 immune variables, 14 metabolomic latent variables, 2 microbiome latent variables, and 2 metadata variables were used to study their probabilistic relationships. Structure learning was performed using hill climbing, tabu, and max–min hill climbing algorithms. A blacklist based on prior biological knowledge was added to the structure learning algorithms to establish a temporal logic model such that nodes from the day of diagnosis were not allowed to predict nodes at parturition or prepartum, nodes at parturition were not allowed to predict nodes prepartum, and prepartum nodes were not allowed to predict nodes at the day of diagnosis. Each algorithm was bootstrapped 1,000 times and mean and median Bayesian information criterion (BIC) scores were extracted. The algorithm with the mean and median larger BIC scores was chosen since bnlearn re-scales the BIC scores by − 2, meaning that the larger the BIC scores, the better the Bayesian network explains the data. The best algorithm was then bootstrapped 1,000 times to estimate the uncertainty of the edge’s strength and the direction of the network as previously described [ 21 , 22 ]. Edges showing presence in at least 60% (strength) among all the 1,000 models were kept in the Bayesian network through model averaging. Model estimates were extracted for each permutation to calculate a mean estimate with a 95% confidence interval for each edge to understand if the parent node was positively or negatively predicting the child node. Once the network was created, it was exported and edited using Metscape 2 [ 23 ] within the CytoScape 3.8 platform. Only edges and nodes that were part of the network connected to the latent variable composed of opportunistic pathogenic bacteria at the day of diagnosis were kept in the final network for ease of interpretation. Results Latent variables identification Within each metabolomic dataset, five principal components (PC) were generated as latent variables. These components were sequentially labeled from PC1 to PC5, representing distinct dimensions of variation captured from each dataset. For the microbiome datasets, one PC at parturition and one PC at metritis diagnosis were generated. Only PCs that remained in the Bayesian network will be described. Full details about each PC can be found in SI Results. Each dataset explained variation by each PC can be found in Supplemental Table S1 . Loadings for each plasma and uterine PCs can be found in Supplemental Table S2 and Supplemental Table S3 , respectively. Association between each PC and metritis can be found in Supplemental Table S4 . For plasma metabolome prepartum, PC1 and PC3 remained in the Bayesian network. Plasma prepartum PC1 explained 14.7% of the variation in the data (Supplemental Table S1 ) and was composed of 49 metabolites, of which 75.3% were amino acids (Supplemental Table S2 ; Supplemental Figure S1 A). Plasma prepartum PC3 explained 4.8% of the variation in the data (Supplemental Table S1 ) and was composed of 5 metabolites, of which 53.8% were carbohydrates (Supplemental Table S2 ; Supplemental Figure S1 C). Both plasma prepartum PC1 ( P = 0.01) and PC3 ( P < 0.01) were lesser in cows that developed metritis when compared with cows that did not develop metritis (Fig. 1 ; Supplemental Table S4 ). For plasma metabolome at parturition, PC1, PC2, PC3, and PC4 remained in the Bayesian network. Plasma PC1 at parturition explained 11.0% of the variation in the data (Supplemental Table S1 ) and was composed of 18 metabolites, of which 95.9% were amino acids (Supplemental Table S2 ; Supplemental Figure S2 A). Plasma PC2 at parturition explained 8.6% of the variation in the data (Supplemental Table S1 ) and was composed of 6 metabolites, of which 100.0% were lipids (Supplemental Table S2 ; Supplemental Figure S2 B). Plasma PC3 at parturition explained 6.9% of the variation in the data (Supplemental Table S1 ) and was composed of 7 metabolites, of which 71.2% were carbohydrates (Supplemental Table S2 ; Supplemental Figure S2 C). Plasma PC4 at parturition explained 4.6% of the variation in the data (Supplemental Table S1 ) and was composed of 24 metabolites, of which 45.3% were carbohydrates (Supplemental Table S2 ; Supplemental Figure S2 D). Plasma PC1 ( P = 0.08) and PC4 ( P = 0.04) at parturition were lesser, and PC2 ( P = 0.02) and PC3 ( P = 0.02) at parturition were greater in cows that developed metritis when compared with cows that did not develop metritis (Fig. 1 ; Supplemental Table S4 ). For plasma metabolome at diagnosis, PC1 and PC5 remained in the Bayesian network. Plasma PC1 at diagnosis explained 11.7% of the variation in the data (Supplemental Table S1 ) and was composed of 27 metabolites, of which 63.6% were amino acids (Supplemental Table S2 ; Supplemental Figure S3 A). Plasma PC5 at diagnosis explained 3.9% of the variation in the data (Supplemental Table S1 ) and was composed of 48 metabolites, of which 53.1% were amino acids (Supplemental Table S2 ; Supplemental Figure S3 E). Plasma PC1 at diagnosis ( P < 0.01) was lesser and PC5 at diagnosis ( P < 0.01) was greater in cows that developed metritis when compared with cows that did not develop metritis (Fig. 1 ; Supplemental Table S4 ). For uterine metabolome at parturition, PC5 remained in the Bayesian network. Uterine PC5 at parturition explained 4.0% of the variation in the data (Supplemental Table S1 ) and was composed of 45 metabolites, of which 67.9% were carbohydrates (Supplemental Table S3 ; Supplemental Figure S4 E). Uterine PC5 at parturition was lesser ( P = 0.04) in cows that developed metritis when compared with cows that did not develop metritis (Supplemental Table S4 ). For uterine metabolome at diagnosis, PC2 and PC3 remained in the Bayesian network. Uterine PC2 at diagnosis explained 15.5% of the variation in the data (Supplemental Table S1 ) and was composed of 44 metabolites, of which 28.6% were uncategorized metabolites (Supplemental Table S3 ; Supplemental Figure S5 B). Uterine PC3 at diagnosis explained 7.5% of the variation in the data (Supplemental Table S1 ) and was composed of 43 metabolites, of which 50.2% were amino acids (Supplemental Table S3 ; Supplemental Figure S5 C). Uterine metabolome PC2 at diagnosis ( P < 0.01) was lesser and uterine PC3 at diagnosis ( P < 0.01) was greater in cows that developed metritis when compared with cows that did not develop metritis (Fig. 1 ; Supplemental Table S4 ). For uterine microbiome at parturition, PC1 was composed of 3 bacteria. The 3 bacteria, which explained 100% of the variation in PC1, were Porphyromonas (34.3%), Bacteroides (33.5%), and Fusobacterium (32.2%). For uterine microbiome at diagnosis, PC1 was composed of 3 bacteria. The 3 bacteria, which explained 100% of the variation in PC1, were Bacteroides (34.9%), Porphyromonas (34.5%), and Fusobacterium (30.6%). Bayesian Network A total of 55 edges connected 37 nodes with a strength greater or equal to 60% (Supplemental Table S5 ). Only 39 edges and 28 nodes remained in the final network (Fig. 1 ) either because they were part of a pathway connected to Microbiome PC1 on the day of metritis diagnosis or because they were of biological interest. In the prepartum, ppBW and IFN-γ were not impacted by other nodes. Body weight change and Live cells were negatively impacted by ppBW. Polymorphonuclear cells were negatively impacted by Live cells. Plasma PC1, composed of 75.3% amino acids, was positively impacted by BWC and Live cells, and negatively impacted by PMN. Plasma PC3, composed of 53.8% carbohydrates, was positively impacted by Live cells and negatively impacted by PMN. Interleukin 8 was negatively impacted by plasma PC3. At parturition, plasma PC1, composed of 95.9% amino acids, was not impacted by other nodes. Microbiome PC1, composed of Bacteroides , Porphyromonas , and Fusobacterium , was negatively impacted by prepartum plasma PC3. IL-8 was positively impacted by prepartum IL-8. Plasma PC4, composed of 45.3% carbohydrates, was positively impacted by prepartum Live cells and prepartum plasma PC3. Monocyte activation was positively impacted by plasma PC4. B-cell activation was negatively impacted by Monocyte activation and plasma PC1 and positively impacted by prepartum PMN. Uterine PC5, composed of 67.9% carbohydrates, was negatively impacted by prepartum IFN-γ and positively impacted by plasma PC1. Interferon γ was positively impacted by prepartum IFN-γ. Interleukin 1β was positively impacted by IFN-γ. Plasma PC2, composed of 100.0% lipids, was negatively impacted by BWC, and plasma PC3, composed of 71.2% carbohydrates, was positively impacted by prepartum PMN. On the day of metritis diagnosis, plasma PC1, composed of 63.6% amino acids, was not impacted by other nodes. B-cell activation was positively impacted by parturition B-cell activation. Monocytes MHCII was positively impacted by plasma PC1 and negatively impacted by parturition IL-1β. Plasma PC5, composed of 53.1% amino acids, was positively impacted by parturition plasma PC3. Uterine PC2, composed of 28.6% uncategorized metabolites, was positively impacted by plasma PC1 and negatively impacted by plasma PC5. Polymorphonuclear cells activation was positively impacted by parturition plasma PC2. Polymorphonuclear cells were negatively impacted by B-cell activation, PMN activation, and parturition IL-8 and positively impacted by uterine PC5. Uterine PC3, composed of 50.2% amino acids, was negatively impacted by PMN. Microbiome PC1, composed of Bacteroides , Porphyromonas , and Fusobacterium , was negatively impacted by uterine PC2, and positively impacted by uterine PC3 and parturition microbiome PC1. Discussion Herein, Bayesian networks were applied to infer potential causal relationships between the uterine microbiome, uterine and plasma metabolome, and systemic immune profiling to advance the understanding of metritis development in dairy cows. Bayesian network analysis is a powerful statistical method that can suggest potential causal relationships among variables, which can aid in the development of strong mechanistic hypotheses for further experimental testing [ 21 ]. According to the directionality network, the prepartum immune and metabolic status of cows were impacted by their ppBW. In dairy cows, body weight is strongly correlated with body condition score (BCS), a measure of subcutaneous adiposity [ 11 ], hence, we speculate that heavier cows were fatter. The greater ppBW in cows that developed metritis negatively impacted the prepartum Live cells and the BWC leading to heavier cows experiencing more cell death and greater BW loss. The greater level of fatty acid accumulation in adipocytes from fatter animals activates NADPH oxidase, increasing the production of reactive oxygen species (ROS) and their release to circulation [ 24 ]. Elevated systemic ROS levels cause cell death [ 24 ]. Furthermore, cows with greater BCS experience a more pronounced negative nutrient balance prior to parturition, leading to greater BW loss [ 11 ]. The greater cell death observed in cows that developed metritis led to an increase in prepartum PMN. Following cell death, intracellular components are released to circulation triggering immune cell recruitment [ 25 ]. It is possible that PMN were released from the bone marrow or spleen as a response to cell death-associated oxidative stress [ 26 ]. Greater cell death and PMN led to a reduction in prepartum plasma PC3, which was composed of fructose, tagatose, methionine sulfoxide, glycerol-α-phosphate, and serotonin. All these metabolites except for serotonin had positive loadings (i.e. positive correlations), whereas serotonin had a negative loading, meaning that the observed reduction in prepartum plasma PC3 in cows that developed metritis indicates a reduction in all of these metabolites but an increase in serotonin. Fructose is a sugar that can enter the glycolytic pathway at various points [ 27 ], while glycerol-α-phosphate is an intermediate in glycolysis. The observed decrease of these metabolites may reflect an overall usage of these glycolytic substrates by PMN, which upon activation shift their metabolism from oxidative phosphorylation to glycolysis [ 28 ]. Under conditions of elevated ROS and cell death, methionine residues in proteins are oxidized to methionine sulfoxide [ 29 ]. In an attempt to reverse the oxidative damage, methionine sulfoxide reductases catalyze the reduction of methionine sulfoxide back to methionine, thereby, reducing methionine sulfoxide concentrations [ 29 ]. The observed decrease in methionine sulfoxide levels in cows experiencing greater cell death may indicate a response to a greater oxidative challenge. Although most of the circulating serotonin is synthetized by the enterochromaffin cells of the gastrointestinal tract, leukocytes are able to synthesize and store serotonin [ 30 ]. It is possible that following cell death, intracellular serotonin is released; thus, increasing circulating serotonin levels. Serotonin enhances the production of IL-8 by dendritic cells [ 31 ], which may help explain why plasma PC3 led to an increase in prepartum IL-8. According to the directionality network, the immune and metabolic status of cows at parturition were impacted by changes observed in the prepartum. The greater BW loss in cows that developed metritis led to an increase in plasma PC2 at parturition. Plasma PC2 was composed of fatty acids, indicating that heavier cows were losing more weight, consequently mobilizing more adipose tissue. Furthermore, the greater prepartum cell death and the reduction in prepartum plasma PC3 led to a decrease in plasma PC4 at parturition. Similar to prepartum plasma PC3, plasma PC4 at parturition was mostly composed of tagatose, fructose, and methionine sulfoxide, which indicates that the response to oxidative stress observed in the prepartum was maintained at parturition. The decrease in parturition plasma PC4 led to a reduction in monocyte activation at parturition. It is possible that the reduction in glycolytic substrates hindered monocyte activation. Contrarily to neutrophils, monocytes lack glycogen storages; thus, they depend entirely on extracellular carbohydrates for activation [ 32 ]; making them more sensible to low carbohydrate levels. The decrease in monocyte activation, the decrease in parturition plasma PC1, composed of amino acids, and the increase in prepartum PMN, which was impacted by prepartum cell death, led to an increase in B-cell activation at parturition. Although not included in the Bayesian network analysis presented herein because of the high collinearity with B-cell activation at parturition, B-cell activation prepartum was directly impacted by prepartum cell death (data not shown), indicating that B-cell activation is part of the immune response to cell death [ 25 , 33 ]. The decrease in prepartum plasma PC3 led to an increase in parturition microbiome PC1, composed of Bacteroides, Porphyromonas , and Fusobacterium . In humans, bacteria present in the gut sense serotonin levels, which impact gene expression related to biofilm formation, adhesion, motility, and virulence [ 34 ]. For example, serotonin impacts quorum sensing pathways increasing virulence of Pseudomonas aeruginosa , leading to Pseudomonas aeruginosa overgrowth [ 35 ]. At parturition, dairy cows have degenerative vascular changes in uterine small blood vessels [ 16 ], which has been proposed as a mechanism by which blood components leak into the uterine lumen [ 15 ]. It is possible that circulating serotonin seeps into the uterus and affects the uterine microbiome favoring opportunistic pathogenic bacteria overgrowth. It is worth noting that, on average, uterine samples were collected approximately 12h after parturition, which allowed for bacterial proliferation [ 36 ]. In fact, we previously observed that uterine serotonin was greater in cows with metritis than in cows without metritis, and that was highly correlated with uterine pathogenic bacteria [ 6 ]. According to the directionality network, the immune and metabolic status of cows at the day of metritis diagnosis were impacted by changes observed in the prepartum and at parturition. The greater parturition plasma PC2, composed of fatty acids, led to a greater PMN activation at the day of metritis diagnosis in cows with metritis. Treating bovine neutrophils with long chain fatty acids increased their adhesion and chemotaxis [ 37 ], indicating a possible direct effect of greater circulating fatty acids on PMN activation. The observed greater B-cell activation at parturition was maintained at metritis diagnosis, and together with greater PMN activation at the day of metritis diagnosis, greater parturition IL-8, and lesser parturition uterine PC5, mostly composed of carbohydrates, led to a reduction in circulating PMN at metritis diagnosis. Greater levels of IL-8, which is the main PMN chemokine [ 38 ], and greater expression of CD62L, which is an adhesion molecule on PMN [ 10 ], likely promoted PMN migration to the uterus, leading to a reduction in circulating PMN. In fact, PMN are more abundant in the endometrium of cows with metritis than in the endometrium of cows without metritis [ 39 ]. Despite the greater PMN migration to the uterus in cows with metritis, circulating PMN from cows with metritis have lesser intracellular killing capacity when compared with cows without metritis [ 40 ]. It is possible that these dysfunctional PMN arriving in the uterus are less effective at eliminating pathogenic bacteria, resulting in a persistent chemo attractive signal for additional PMN recruitment. The lesser circulating PMN in cows that developed metritis led to an increase in uterine PC3 on the day of metritis diagnosis in cows with metritis, which consequently led to an increase in Bacteroides , Porphyromonas , and Fusobacterium at diagnosis. Uterine PC3 at diagnosis was mostly composed of amino and organic acids. Upon arrival to tissues, activated PMN release proteolytic enzymes [ 41 ], leading to endometrial damage [ 39 ]. Tissue damage triggers the breakdown of proteins, releasing dipeptides, such as ile-ile and alanine-alanine, and amino acids such as tyrosine and phenylalanine [ 42 ]. Furthermore, cellular stress and metabolic alterations during tissue injury leads to an increase in several other metabolites, including organic acids such as succinic and fumaric acids [ 43 ]. Greater tissue damage creates an environment that may promote opportunistic pathogenic bacterial proliferation by increasing substrate availability [ 44 ], which might explain how the greater levels of uterine PC3 in cows with metritis positively impacted the observed increase in Bacteroides , Porphyromonas , and Fusobacterium . The greater parturition IL-1β together with the lesser diagnosis plasma PC1, composed of amino and organic acids, led to a reduction in monocytes MHCII expression at diagnosis. The lesser circulating amino acids may impair monocyte antigen presentation capacity. It is also possible that these cells were experiencing immune tolerance. The greater parturition IL-1β was positively impacted by parturition IFN-γ, which was positively impacted by prepartum IFN-γ. Prepartum IFN-γ was not influenced by any node. The origin of the greater prepartum IFN-γ in cows that developed metritis is, therefore, not clear. In obese non-ruminants, T-cells surrounding necrotic adipocytes produce high levels of IFN-γ [ 45 ]; thus, we hypothesize that the greater prepartum IFN-γ in cows that developed metritis may originate from adipocyte-associated T-cells. Nonetheless, the persistent systemic inflammation observed since the prepartum led to a reduction in monocyte antigen presentation capacity. In a previous study, we hypothesized that the reduction in monocyte antigen presentation capacity observed postpartum in cows with metritis could be due to immune tolerance [ 10 ]. In fact, prolonged inflammatory responses are known for leading to immune tolerance [ 46 ], particularly reducing monocytes and macrophages activation [ 47 ] but not PMN activation [ 48 ]. It is noteworthy that we did not detect any influence of the reduction in monocyte antigen presentation capacity on the abundance of opportunistic pathogenic bacteria; however, it is possible that if circulatory monocytes are experiencing immune tolerance, uterine monocytes and macrophages could also be experiencing tolerance. Monocytes and macrophages play a key role in the resolution of inflammation [ 49 ]; therefore, their impaired response could hinder the resolution of inflammation, potentially leading to greater tissue damage by PMN. Conclusions This observational study provides insights into the complex relationship between ppBW, BW loss, systemic immune profiling, and uterine and plasma metabolome around parturition, and how they may promote uterine opportunistic pathogens growth leading to metritis development. Altogether, directionality network analysis indicates that greater prepartum BW led to greater BW loss and prepartum systemic cellular death, which led to an increase in systemic inflammation and immune activation. The aforementioned changes led to an increase in PMN extravasation, which led to an increase in uterine metabolomic changes associated with tissue damage. These changes led to an increase in Bacteroides , Porphyromonas and Fusobacterium in cows that developed metritis, indicating that excessive tissue damage may promote opportunistic pathogenic bacterial growth. Our working hypothesis is illustrated in Fig. 2 . Experimental research needs to be conducted to validate the mechanistic hypotheses generated herein. In conclusion, the directionality network showed that prepartum BW led to immune dysregulation and plasma and uterine metabolomic changes leading to opportunistic pathogens overgrowth in cows that developed metritis. Abbreviations BW: Body weight ppBW: Prepartum body weight BWC: Daily body weight change PCA: Principal component analysis sPCA: Sparse Principal component analysis BIC: Bayesian information criterion PC: Principal component BCS: Body condition score ROS: Reactive oxygen species Declarations Ethics approval and consent to participate All procedures involving cows were approved by the Institutional Animal Care and Use Committee of the University of Florida; protocol number 201910623. Consent for publication Not applicable. Availability of data and material The raw sequence data used in the current study were generated during our previous study [6] and are available in the NCBI repository under BioProject OR883023 (www.ncbi.nlm.nih.gov/nuccore/OR883023) - OR883397 (www.ncbi.nlm.nih.gov/nuccore/OR883397). The metabolomics dataset used in the current study was generated during our two previous studies [6, 13] and are available in the NIH Common Fund’s National Metabolomics Data Repository website, the Metabolomics Workbench repository [50]. Uterine metabolomics dataset [6] can be found under Study ID ST002994, http://dx.doi.org/10.21228/M8S425. Plasma metabolomics dataset [13] can be found under Study ID ST002556; http://dx.doi.org/10.21228/M8PF0K . The Metabolomics Workbench is supported by NIH grant U2C-DK119886 and OT2-OD030544 grants. Competing interests The authors declare that they have no competing interests. Funding This work was supported by U.S. Department of Agriculture Grant # 2019-67015-29836, Accession No: 1019435. Authors' contributions JEPS, CDN, SJJ, RCB, JPD, and KNG initiated the study design and provided scientific input during manuscript writing. RSB and RCC provided scientific input during manuscript writing. SC, JGP, ACM, and TDG collected the on-farm data. SC and FC performed the laboratory work. SC performed the statistical analyses. KNG and FC provided their scientific input during statistical analyses. SC and KNG wrote the manuscript. All authors edited and approved the manuscript. Acknowledgements We would like to thank the general manager, Mr. Eric Williams, and the staff of the University of Florida Dairy Unit for allowing the use of their animals and facilities. References Pinedo P, Santos JEP, Chebel RC, Galvão KN, Schuenemann GM, Bicalho RC, et al. Early-lactation diseases and fertility in 2 seasons of calving across US dairy herds. J Dairy Sci. 2020;103:10560–76. Barragan AA, Piñeiro JM, Schuenemann GM, Rajala-Schultz PJ, Sanders DE, Lakritz J, et al. Assessment of daily activity patterns and biomarkers of pain, inflammation, and stress in lactating dairy cows diagnosed with clinical metritis. J Dairy Sci. 2018;101:8248–58. Figueiredo CC, Merenda VR, de Oliveira EB, Lima FS, Chebel RC, Galvão KN, et al. Failure of clinical cure in dairy cows treated for metritis is associated with reduced productive and reproductive performance. J Dairy Sci. 2021;104:7056–70. Pérez-Báez J, Risco CA, Chebel RC, Gomes GC, Greco LF, Tao S, et al. Association of dry matter intake and energy balance prepartum and postpartum with health disorders postpartum: Part I. Calving disorders and metritis. J Dairy Sci. 2019;102:9138–50. Jeon SJ, Vieira-Neto A, Gobikrushanth M, Daetz R, Mingoti RD, Parize ACB, et al. Uterine microbiota progression from calving until establishment of metritis in dairy cows. Appl Environ Microbiol. 2015;81:6324–32. Casaro S, Prim JG, Gonzalez TD, Cunha F, Bisinotto RS, Chebel RC, et al. Integrating uterine microbiome and metabolome to advance the understanding of the uterine environment in dairy cows with metritis. Animal Microbiome 2024 6:1. 2024;6:1–13. Jeon SJ, Cunha F, Ma X, Martinez N, Vieira-Neto A, Daetz R, et al. Uterine microbiota and immune parameters associated with fever in dairy cows with metritis. PLoS One. 2016;11. Galvão KN, Bicalho RC, Jeon SJ. Symposium review: The uterine microbiome associated with the development of uterine disease in dairy cows. J Dairy Sci. 2019;102:11786–97. Casaro S, Prim J, Gonzalez T, Figueiredo C, Bisinotto R, Chebel R, et al. Blood metabolomics and impacted cellular mechanisms during transition into lactation in dairy cows that develop metritis. J Dairy Sci. 2023. https://doi.org/10.3168/jds.2023-23433. Casaro S, Prim JG, Gonzalez TD, Bisinotto RS, Chebel RC, Marrero MG, et al. Unraveling the immune and metabolic changes associated with metritis in dairy cows. J Dairy Sci. 2023;106:9244–59. Casaro S, Pérez-Báez J, Bisinotto RS, Chebel RC, Prim JG, Gonzalez TD, et al. Association between prepartum body condition score and prepartum and postpartum dry matter intake and energy balance in multiparous Holstein cows. J Dairy Sci. 2024. https://doi.org/10.3168/JDS.2023-24047. Martinez N, Risco CA, Lima FS, Bisinotto RS, Greco LF, Ribeiro ES, et al. Evaluation of peripartal calcium status, energetic profile, and neutrophil function in dairy cows at low or high risk of developing uterine disease. J Dairy Sci. 2012;95:7158–72. Casaro S, Prim JG, Gonzalez TD, Figueiredo CC, Bisinotto RS, Chebel RC, et al. Blood metabolomics and impacted cellular mechanisms during transition into lactation in dairy cows that develop metritis. J Dairy Sci. 2023;106:8098–109. Chu X, Jaeger M, Beumer J, Bakker OB, Aguirre-Gamboa R, Oosting M, et al. Integration of metabolomics, genomics, and immune phenotypes reveals the causal roles of metabolites in disease. Genome Biol. 2021;22. Jeon SJ, Cunha F, Vieira-Neto A, Bicalho RC, Lima S, Bicalho ML, et al. Blood as a route of transmission of uterine pathogens from the gut to the uterus in cows. Microbiome. 2017;5:109. Archbald LF, Schultz RH, Fahning ML, Kurtz HJ, Zemjanis R. A sequential histological study of the post-partum bovine uterus. J Reprod Fertil. 1972;29:133–6. R Core Team. R: A Language and Environment for Statistical Computing. 2023. Rohart F, Gautier B, Singh A, Lê Cao KA. mixOmics: An R package for ’omics feature selection and multiple data integration. PLoS Comput Biol. 2017;13. Shen H, Huang JZ. Sparse principal component analysis via regularized low rank matrix approximation. J Multivar Anal. 2008;99:1015–34. Scutari M. Learning Bayesian Networks with the bnlearn R Package. J Stat Softw. 2010;35:1–22. Yu H, Campbell MT, Zhang Q, Walia H, Morota G. Genomic Bayesian Confirmatory Factor Analysis and Bayesian Network To Characterize a Wide Spectrum of Rice Phenotypes. G3 Genes|Genomes|Genetics. 2019;9:1975–86. Novais FJ de, Yu H, Cesar ASM, Momen M, Poleti MD, Petry B, et al. Multi-omic data integration for the study of production, carcass, and meat quality traits in Nellore cattle. Front Genet. 2022;13:948240. Karnovsky A, Weymouth T, Hull T, Glenn Tarcea V, Scardoni G, Laudanna C, et al. Metscape 2 bioinformatics tool for the analysis and visualization of metabolomics and gene expression data. Bioinformatics. 2012;28:373–80. Furukawa S, Fujita T, Shimabukuro M, Iwaki M, Yamada Y, Nakajima Y, et al. Increased oxidative stress in obesity and its impact on metabolic syndrome. Journal of Clinical Investigation. 2004;114:1752–61. Roh JS, Sohn DH. Damage-Associated Molecular Patterns in Inflammatory Diseases. Immune Netw. 2018;18. Mulder PPG, Vlig M, Boekema BKHL, Stoop MM, Pijpe A, van Zuijlen PPM, et al. Persistent Systemic Inflammation in Patients With Severe Burn Injury Is Accompanied by Influx of Immature Neutrophils and Shifts in T Cell Subsets and Cytokine Profiles. Front Immunol. 2021;11 January:1–13. Dholariya SJ, Orrick JA. Biochemistry, Fructose Metabolism. StatPearls Publishing; 2024. O’Neill LAJ, Kishton RJ, Rathmell J. A guide to immunometabolism for immunologists. Nat Rev Immunol. 2016;16:553–65. Weissbach H, Resnick L, Brot N. Methionine sulfoxide reductases: history and cellular role in protecting against oxidative damage. Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics. 2005;1703:203–12. Baganz NL, Blakely RD. A dialogue between the immune system and brain, spoken in the language of serotonin. ACS Chem Neurosci. 2013;4:48–63. Idzko M, Panther E, Stratz C, Müller T, Bayer H, Zissel G, et al. The serotoninergic receptors of human dendritic cells: identification and coupling to cytokine release. J Immunol. 2004;172:6011–9. O’Neill LAJ. A Broken Krebs Cycle in Macrophages. Immunity. 2015;42:393–4. Gong T, Liu L, Jiang W, Zhou R. DAMP-sensing receptors in sterile inflammation and inflammatory diseases. Nat Rev Immunol. 2020;20:95–112. Everett BA, Tran P, Prindle A. Toward manipulating serotonin signaling via the microbiota–gut–brain axis. Curr Opin Biotechnol. 2022;78:102826. Knecht LD, O’Connor G, Mittal R, Liu XZ, Daftarian P, Deo SK, et al. Serotonin Activates Bacterial Quorum Sensing and Enhances the Virulence of Pseudomonas aeruginosa in the Host. EBioMedicine. 2016;9:161–9. Bacic MK, Smith CJ. Laboratory maintenance and cultivation of bacteroides species. Curr Protoc Microbiol. 2008;Chapter 13 SUPPL. 9. Mena J, Manosalva C, Ramirez R, Chandia L, Carroza D, Loaiza A, et al. Linoleic acid increases adhesion, chemotaxis, granule release, intracellular calcium mobilisation, MAPK phosphorylation and gene expression in bovine neutrophils. Vet Immunol Immunopathol. 2013;151:275–84. Tajima A, Iwase T, Shinji H, Seki K, Mizunoe Y. Inhibition of endothelial interleukin-8 production and neutrophil transmigration by Staphylococcus aureus beta-hemolysin. Infect Immun. 2009;77:327–34. Sicsic R, Goshen T, Dutta R, Kedem-Vaanunu N, Kaplan-Shabtai V, Pasternak Z, et al. Microbial communities and inflammatory response in the endometrium differ between normal and metritic dairy cows at 5-10 days post-partum. Vet Res. 2018;49:77. Hammon DS, Evjen IM, Dhiman TR, Goff JP, Walters JL. Neutrophil function and energy status in Holstein cows with uterine health disorders. Vet Immunol Immunopathol. 2006;113:21–9. Tecchio C, Cassatella MA. Neutrophil-derived chemokines on the road to immunity. Semin Immunol. 2016;28:119–28. Wang J. Neutrophils in tissue injury and repair. Cell Tissue Res. 2018;371:531–9. Zindel J, Kubes P. DAMPs, PAMPs, and LAMPs in Immunity and Sterile Inflammation. Annual Review of Pathology: Mechanisms of Disease. 2020;15:493–518. Anderson CJ, Medina CB, Barron BJ, Karvelyte L, Aaes TL, Lambertz I, et al. Microbes exploit death-induced nutrient release by gut epithelial cells. Nature 2021 596:7871. 2021;596:262–7. Rocha VZ, Folco EJ, Sukhova G, Shimizu K, Gotsman I, Vernon AH, et al. Interferon-γ, a Th1 cytokine, regulates fat inflammation: A role for adaptive immunity in obesity. Circ Res. 2008;103:467–76. Rogovskii V. Immune Tolerance as the Physiologic Counterpart of Chronic Inflammation. Front Immunol. 2020;11 September:1–7. Fernández-Ruiz I, Arnalich F, Cubillos-Zapata C, Hernández-Jiménez E, Moreno-González R, Toledano V, et al. Mitochondrial DAMPs induce endotoxin tolerance in human monocytes: An observation in patients with myocardial infarction. PLoS One. 2014;9. Ariga SK, Abatepaulo FB, Melo ESA, Velasco IT, Da Silva FP, De Lima TM, et al. Endotoxin tolerance drives Neutrophil to infectious site. Shock. 2014;42:168–73. Parihar A, Eubank TD, Doseff AI. Monocytes and macrophages regulate immunity through dynamic networks of survival and cell death. J Innate Immun. 2010;2:204–15. Sud M, Fahy E, Cotter D, Azam K, Vadivelu I, Burant C, et al. Metabolomics Workbench: An international repository for metabolomics data and metadata, metabolite standards, protocols, tutorials and training, and analysis tools. Nucleic Acids Res. 2016;44:D463–70. Additional Declarations No competing interests reported. Supplementary Files SIMaterialsandMethods.pdf SupplementalFigures.pdf SupplementalTableS1.xlsx SupplementalTableS2.xlsx SupplementalTableS3.xlsx SupplementalTableS4.xlsx SupplementalTableS5.xlsx Cite Share Download PDF Status: Published Journal Publication published 09 Jan, 2025 Read the published version in Animal Microbiome → Version 1 posted Editorial decision: Revision requested 11 Sep, 2024 Reviews received at journal 20 Aug, 2024 Reviews received at journal 14 Aug, 2024 Reviews received at journal 14 Aug, 2024 Reviewers agreed at journal 31 Jul, 2024 Reviewers agreed at journal 30 Jul, 2024 Reviewers agreed at journal 30 Jul, 2024 Reviewers invited by journal 30 Jul, 2024 Editor assigned by journal 19 Jun, 2024 Submission checks completed at journal 15 Jun, 2024 First submitted to journal 12 Jun, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4571697","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":320116267,"identity":"81b96653-3f39-478d-8604-1ad87830f28a","order_by":0,"name":"S. Casaro","email":"","orcid":"","institution":"University of Florida","correspondingAuthor":false,"prefix":"","firstName":"S.","middleName":"","lastName":"Casaro","suffix":""},{"id":320116268,"identity":"af369e7e-a1bf-4c45-a556-cc8ae90eef0b","order_by":1,"name":"J. G. Prim","email":"","orcid":"","institution":"Auburn University","correspondingAuthor":false,"prefix":"","firstName":"J.","middleName":"G.","lastName":"Prim","suffix":""},{"id":320116271,"identity":"aecaff27-cb79-4e90-9483-31541b0712e8","order_by":2,"name":"T. D. Gonzalez","email":"","orcid":"","institution":"University of Florida","correspondingAuthor":false,"prefix":"","firstName":"T.","middleName":"D.","lastName":"Gonzalez","suffix":""},{"id":320116277,"identity":"85a785f8-4b2f-45ed-9d7e-7cf8bc82e8b5","order_by":3,"name":"F. Cunha","email":"","orcid":"","institution":"University of Florida","correspondingAuthor":false,"prefix":"","firstName":"F.","middleName":"","lastName":"Cunha","suffix":""},{"id":320116279,"identity":"b12f531d-9803-4501-806c-6df16cd2c771","order_by":4,"name":"A. C. M. Silva","email":"","orcid":"","institution":"University of Florida","correspondingAuthor":false,"prefix":"","firstName":"A.","middleName":"C. M.","lastName":"Silva","suffix":""},{"id":320116281,"identity":"1d7d9301-b7b1-4aaa-af0d-d550e3ebe06f","order_by":5,"name":"H. Yu","email":"","orcid":"","institution":"University of Florida","correspondingAuthor":false,"prefix":"","firstName":"H.","middleName":"","lastName":"Yu","suffix":""},{"id":320116282,"identity":"ddb07a35-26a5-4d8d-9fec-4fae33749e47","order_by":6,"name":"R. S. Bisinotto","email":"","orcid":"","institution":"University of Florida","correspondingAuthor":false,"prefix":"","firstName":"R.","middleName":"S.","lastName":"Bisinotto","suffix":""},{"id":320116283,"identity":"9bc430fa-69e3-446a-b94a-8f3f572d1246","order_by":7,"name":"R. C. Chebel","email":"","orcid":"","institution":"University of Florida","correspondingAuthor":false,"prefix":"","firstName":"R.","middleName":"C.","lastName":"Chebel","suffix":""},{"id":320116284,"identity":"7bdc5626-db5a-413a-8d83-f35bc0b2d74a","order_by":8,"name":"J. E. P. Santos","email":"","orcid":"","institution":"University of Florida","correspondingAuthor":false,"prefix":"","firstName":"J.","middleName":"E. P.","lastName":"Santos","suffix":""},{"id":320116285,"identity":"6795465f-2846-447a-b8fe-5cbc7e6b0a23","order_by":9,"name":"C. D. Nelson","email":"","orcid":"","institution":"University of Florida","correspondingAuthor":false,"prefix":"","firstName":"C.","middleName":"D.","lastName":"Nelson","suffix":""},{"id":320116286,"identity":"e2f99f63-7fbd-44dc-a25f-2698b81df434","order_by":10,"name":"S. J. Jeon","email":"","orcid":"","institution":"Long Island University","correspondingAuthor":false,"prefix":"","firstName":"S.","middleName":"J.","lastName":"Jeon","suffix":""},{"id":320116287,"identity":"2f13c097-4149-4ca1-9663-e1c929eff8bb","order_by":11,"name":"R. C. Bicalho","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"R.","middleName":"C.","lastName":"Bicalho","suffix":""},{"id":320116288,"identity":"5d6e51e2-3fbc-4296-9bac-df64efced06e","order_by":12,"name":"J. P. Driver","email":"","orcid":"","institution":"University of Missouri","correspondingAuthor":false,"prefix":"","firstName":"J.","middleName":"P.","lastName":"Driver","suffix":""},{"id":320116294,"identity":"8228a381-62f7-40d1-8e81-34c294879cf1","order_by":13,"name":"Klibs N. Galvão","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAuklEQVRIiWNgGAWjYFAC5gYgkSDHwHwYyGAjSgsjWIsxA1siiVoSG4jWIu9+sE3iR0Va+oZjQL0fyg4T1mJ4JrFNsudMTi5IC+OMc8RoaUhsNuBtq8jdcL+xgZm3jRgt/Q+bDf+2VaQbAG1h/kuMFnmJxMbHvG05CWAtjMRoMZB42PhY5kya4UygloM959KJsKU/+cDBNxXJ8nzHmA8++FFmTYQtB5A4B3AoQrOlgShlo2AUjIJRMKIBAK9XQIHJEoxIAAAAAElFTkSuQmCC","orcid":"","institution":"University of Florida","correspondingAuthor":true,"prefix":"","firstName":"Klibs","middleName":"N.","lastName":"Galvão","suffix":""}],"badges":[],"createdAt":"2024-06-12 16:45:38","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4571697/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4571697/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s42523-024-00366-9","type":"published","date":"2025-01-09T15:56:54+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":59465574,"identity":"22fe7348-5a38-4a80-9136-90d3526fe151","added_by":"auto","created_at":"2024-07-02 06:27:26","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":8012003,"visible":true,"origin":"","legend":"\u003cp\u003eBayesian network between immune variables, metabolome and microbiome latent variables, and metadata variables bootstrapped 1,000 times to estimate the uncertainty of the edge’s strength and the direction of the network. Hexagons represent nodes and arrows represent edges. Edges showing presence in at least 600 of the 1,000 models (60% strength) are shown. Average model estimates were extracted to determine if the parent node was positively or negatively predicting the child node. Positive estimates are represented by red edges and negative estimates are represented by blue edges. Edge labels represent the strength of each edge. Sparse principal component analyses were performed for plasma and uterine metabolome latent variable identification. Plasma principal components (PCs) indicate plasma latent variables. Uterine PCs indicate uterine latent variables. Principal component analyses were performed on uterine Bacteroides, Fusobacterium, and Porphyromonas for latent variables identification. Microbiome PCs indicate the latent variables composed of Bacteroides, Fusobacterium, and Porphyromonas. Information regarding each latent variable are shown surrounding the Bayesian network. Each pie chart and bar graph represent one latent variable. Colors represent the categories of metabolites, or different microbes, respectively. The center color on each pie chart represents the different timepoints. The “N =” on each pie chart represents the number of metabolites or microbes composing the respective latent variable. Bar graphs represent the top 5 most important metabolites explaining the variation of each latent variable. Percentages between parentheses represent the proportion of the variation in the respective component explained by the metabolite or microbe. Metabolites starting with “(-)” were negatively associated with their respective latent variable. Otherwise, metabolites were positively associated with the latent variable. Hexagons with green borders correspond to variables that differed between cows that developed metritis and cows that did not develop metritis. MET: methionine. PO: phosphate. Hp: hydroxy phenyl. ppBW: prepartum body weight. BW: body weight. PMN: polymorphonuclear cells. IFN-γ: interferon gamma. IL: interleukin. MHC2: major histocompatibility complex class 2. AA: amino acids. CHO: carbohydrates. Metr: metritis. Monocyte activation indicates an increase in cluster of differentiation (CD) 62L. PMN activation indicates an increase in CD62L. B-cell activation indicates a decrease in CD62L.\u003c/p\u003e","description":"","filename":"Figure1MultiOmicsIntegrationNetwork.png","url":"https://assets-eu.researchsquare.com/files/rs-4571697/v1/51d1dfdb92a4bf910d08ed74.png"},{"id":59465572,"identity":"4b647c7f-298b-4d2c-9739-792793a66f00","added_by":"auto","created_at":"2024-07-02 06:27:26","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":9596120,"visible":true,"origin":"","legend":"\u003cp\u003eIllustration of our working hypothesis of the interactions between prepartum body weight, body weight loss, systemic immune profiling, and uterine and plasma metabolome around parturition and how they may affect uterine opportunistic pathogens growth leading to metritis development. Down arrows indicate reduction, and up arrows indicate increase in cows with metritis. ROS: reactive oxygen species. MetO: methionine sulfoxide. CHO: carbohydrates. IL: interleukin. PMN: polymorphonuclear cells. IFN: interferon. MHC: major histocompatibility complex. DMI: dry matter intake. NNB: negative nutrient balance. Mo: monocytes. Mϕ: macrophages. Figure created with Biorender.\u003c/p\u003e","description":"","filename":"Figure2WorkingHypothesisIntegration.png","url":"https://assets-eu.researchsquare.com/files/rs-4571697/v1/dafb838da4bb18d141251563.png"},{"id":73693754,"identity":"1a7d291a-5d17-45e2-8376-feb56cc861ec","added_by":"auto","created_at":"2025-01-13 16:05:06","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":18681403,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4571697/v1/f337a809-dc45-42f4-b7f9-e5424caa9805.pdf"},{"id":59465908,"identity":"512bf0c0-a439-488c-a91c-d73d694c25ca","added_by":"auto","created_at":"2024-07-02 06:35:27","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":185532,"visible":true,"origin":"","legend":"","description":"","filename":"SIMaterialsandMethods.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4571697/v1/8a3e3c98dd8b1d6f1a2f5cd3.pdf"},{"id":59465573,"identity":"01a0f933-18b3-45c0-bd7d-1cd26ee451d5","added_by":"auto","created_at":"2024-07-02 06:27:26","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":1021872,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementalFigures.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4571697/v1/ef17b9b3f40e195d41da407f.pdf"},{"id":59465580,"identity":"6c5b3c46-6d0c-4177-a6a1-256ca3d908ea","added_by":"auto","created_at":"2024-07-02 06:27:27","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":11337,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementalTableS1.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-4571697/v1/57cf2676177ec360b3885df3.xlsx"},{"id":59465912,"identity":"6017699f-c720-4bd7-9196-a2ebb1ea5fcd","added_by":"auto","created_at":"2024-07-02 06:35:28","extension":"xlsx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":25535,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementalTableS2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-4571697/v1/76c623e3ff031eb9a1ae7c63.xlsx"},{"id":59465576,"identity":"743e8fe7-70ba-4917-bfcc-f6833626ef44","added_by":"auto","created_at":"2024-07-02 06:27:26","extension":"xlsx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":27034,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementalTableS3.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-4571697/v1/b28bd4c63708c85d1071289b.xlsx"},{"id":59465575,"identity":"e165ecac-676f-4cc7-b74c-125837cafef9","added_by":"auto","created_at":"2024-07-02 06:27:26","extension":"xlsx","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":13183,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementalTableS4.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-4571697/v1/72d31cb270282d3910d26fbc.xlsx"},{"id":59465577,"identity":"5d261622-2b34-4bbe-8022-5c4202a028cd","added_by":"auto","created_at":"2024-07-02 06:27:26","extension":"xlsx","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":45516,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementalTableS5.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-4571697/v1/ea3b8cb4165ff7284463361c.xlsx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Multi-omics integration and immune profiling identify possible causal networks leading to uterine microbiome dysbiosis in dairy cows that develop metritis","fulltext":[{"header":"Background","content":"\u003cp\u003eMetritis affects 25% of dairy cows [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e], affecting animal wellbeing [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e] and performance, which leads to reduced dairy farm profitability [\u003cspan additionalcitationids=\"CR3\" citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. The uterine microbiome from cows that develop metritis and those that remain healthy do not differ until 2 days postpartum [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e], after which opportunistic pathogens, such as \u003cem\u003eFusobacterium\u003c/em\u003e, \u003cem\u003ePorphyromonas\u003c/em\u003e, and \u003cem\u003eBacteroides\u003c/em\u003e overtake the uterine commensals [\u003cspan additionalcitationids=\"CR6 CR7\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. It is not clear why these opportunistic pathogens thrive and cause metritis; however, the metabolic and immune changes associated with parturition and the onset of lactation seem to play an important role in metritis development [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Around parturition, cows that develop metritis experience a period of negative nutrient balance [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e], which is more pronounced in overconditioned cows [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. This period of negative nutrient balance leads to lesser blood calcium [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e], lesser amino acids [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e], and greater circulating fatty acids concentration [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. These metabolic changes have been associated with persistent systemic inflammation, oxidative stress, cellular damage, and immune dysregulation [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Studies to date have only evaluated how a limited number of metabolites affect the systemic immune response in cows that develop metritis, which do not fully represent the wide variety of metabolic pathways involved in immune processes [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Furthermore, vascular degeneration that occurs shortly after parturition in dairy cows allows for the exchange of metabolites between blood and uterus [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e], thereby raising the potential for blood metabolites to influence the uterine microbiome, or microbial-derived uterine metabolites to influence the systemic immune response. Either of these mechanisms could affect bacterial proliferation in the uterus. We have previously demonstrated that the peripartum plasma [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e] and uterine [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e] metabolome and immune response [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] differ between cows that develop metritis and those that do not; however, the interplay between plasma and uterine metabolome, systemic immune response, and the uterine microbiome has not been investigated. Therefore, the hypothesis of the current study was that plasma and/or uterine metabolites promote opportunistic pathogenic bacterial overgrowth in cows that develop metritis either directly or indirectly by regulating the systemic immune response. Thus, the objective of the current study was to apply directionality networks to infer possible causal relationships between the uterine and plasma metabolome, systemic immune response, and the uterine microbiome to advance the understanding of metritis development.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eThis case-control observational study was conducted at the University of Florida Dairy Unit from September 2019 to March 2020. The cows used for the current study were a subset of cows used in our previous studies [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. All procedures involving cows were approved by the Institutional Animal Care and Use Committee of the University of Florida; protocol number 201910623.\u003c/p\u003e \u003cp\u003eBriefly, cows were enrolled at 240 d of gestation and uterine discharge was evaluated using a Metricheck device (Metricheck, Simcro, New Zealand) at 3\u0026thinsp;\u0026plusmn;\u0026thinsp;1, 7\u0026thinsp;\u0026plusmn;\u0026thinsp;1, 10\u0026thinsp;\u0026plusmn;\u0026thinsp;1 and 13\u0026thinsp;\u0026plusmn;\u0026thinsp;1 days after parturition. Cows with a fetid red-brownish, watery discharge were classified as having metritis (n\u0026thinsp;=\u0026thinsp;50) and paired with cows without metritis (n\u0026thinsp;=\u0026thinsp;50). All cows had blood collected prepartum (14\u0026thinsp;\u0026plusmn;\u0026thinsp;6 d before parturition), at parturition (first 24 hours after parturition), and at the day of metritis diagnosis (7\u0026thinsp;\u0026plusmn;\u0026thinsp;2 d after parturition). All cows had uterine fluid collected at parturition and at diagnosis of metritis. Blood samples were used for flow cytometry, multiplex, and untargeted gas chromatography mass spectrometry. Flow cytometry and multiplex were performed to assess the systemic immune profile and activation status. Gas chromatography mass spectrometry was performed to assess the systemic metabolomic profile. Uterine fluid samples were used for untargeted gas chromatography mass spectrometry and 16S rRNA gene amplification by PCR. Gas chromatography mass spectrometry was performed to assess the uterine metabolomic profile and the 16S rRNA gene was amplified by PCR to assess the uterine microbiome. Prepartum body weight (ppBW) and prepartum daily body weight change (BWC) were evaluated. Full details can be found in SI Materials and Methods.\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eData Pre-processing\u003c/h2\u003e \u003cp\u003eData pre-processing was performed using R [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Datasets included systemic immune data and plasma metabolome prepartum, at parturition, and on the day of metritis diagnosis from two of our previous studies [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e], uterine microbial estimated counts and uterine metabolome at parturition and on the day of metritis diagnosis from our previous study [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Lastly, ppBW and BWC from our previous study [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Because Bayesian networks do not allow for variables with missing values, these were excluded from further analysis. For systemic immune data, only variables with a \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026le;\u0026thinsp;0.10 to the effect of metritis in our previous study [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] and without missing values were selected for further analysis. The variables used herein for prepartum were Live %, Monocytes %, PMN %, B-cell activation (CD62L negative), IL-8, and IFN-y. For parturition the variables were Live %, B-cell activation, Monocyte activation (increase in CD62L MFI), IL-1β, IL-8, and IFN-γ. For the day of diagnosis, the variables were Live %, Singlets %, Monocytes %, PMN %, Monocyte activation, Monocyte MHCII %, PMN activation (increase in CD62L MFI), B-cell activation, and T helper activation (reduction in CD62L MFI).\u003c/p\u003e \u003cp\u003eTo eliminate any possible effect of parity from the analyses, all variables from the datasets were divided by parity group (primiparous and multiparous), log transformed and auto scaled within parity group, and then merged back together.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eLatent variables identification and significance\u003c/h2\u003e \u003cp\u003eLatent variables identification and significance were performed using the mixOmics R package [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Because microbiome and metabolome data are composed of highly correlated variables, principal component analysis (PCA) was applied to microbiome and metabolome data for orthogonal latent variables identification.\u003c/p\u003e \u003cp\u003eFor the five metabolomics datasets, sparse PCA (sPCA) via regularized single value decomposition [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e] was applied using the \u0026lsquo;spca\u0026rsquo; function of the mixOmics package. Briefly, the sPCA from MixOmics is based on singular value decomposition, which is appropriate when dealing with large datasets where not all variables are likely to be equally important. The sparsity is achieved via LASSO penalization, such that latent variables are no longer a linear combination of all original variables but are a linear combination of a subset of variables that maximize the captured variance. Five latent variables were created for each dataset. For each latent variable, optimal number of variables were selected using 3-fold cross-validation with 10 repetitions.\u003c/p\u003e \u003cp\u003eFor the uterine microbiome at parturition and at the day of diagnosis a PCA was applied using the \u0026lsquo;pca\u0026rsquo; function of the mixOmics package selecting the opportunistic pathogenic bacteria with the greatest relative abundance [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e] and estimated counts [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e] in cows with metritis compared with cows without metritis. These were \u003cem\u003eFusobacterium\u003c/em\u003e, \u003cem\u003ePorphyromonas\u003c/em\u003e, and \u003cem\u003eBacteroides\u003c/em\u003e. One latent variable was created for parturition and one for the day of metritis diagnosis. The rationale behind this decision was to observe how different immune and metabolic variables affected the growth of these bacteria since parturition.\u003c/p\u003e \u003cp\u003eLatent variables significance between cows that developed metritis and cows that did not develop metritis were assessed by \u003cem\u003et\u003c/em\u003e-tests using the t.test function of the stats R package [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Statistical significance was considered if \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026le;\u0026thinsp;0.10.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eBayesian Networks\u003c/h2\u003e \u003cp\u003eBayesian networks is a powerful statistical method that models conditional dependencies between variables, which can be interpreted as potential causal relationships among variables. Nonetheless, care must be taken in interpreting these directional relationships as causal because not all the assumptions for causality can be met in an observational study such as that there are no unobserved external variables affecting the variables in the network [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. Bayesian network analysis was performed using the bnlearn R package [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. Highly correlated variables (\u0026gt;\u0026thinsp;0.80 or \u0026lt; -0.80) were excluded to avoid multicollinearity. Prepartum B-cell activation was excluded because they were highly correlated with parturition (Pearson correlation coefficient r\u0026thinsp;=\u0026thinsp;0.83) and diagnosis (r\u0026thinsp;=\u0026thinsp;0.82) B-cell activation. A total of 20 immune variables, 14 metabolomic latent variables, 2 microbiome latent variables, and 2 metadata variables were used to study their probabilistic relationships. Structure learning was performed using hill climbing, tabu, and max\u0026ndash;min hill climbing algorithms. A blacklist based on prior biological knowledge was added to the structure learning algorithms to establish a temporal logic model such that nodes from the day of diagnosis were not allowed to predict nodes at parturition or prepartum, nodes at parturition were not allowed to predict nodes prepartum, and prepartum nodes were not allowed to predict nodes at the day of diagnosis. Each algorithm was bootstrapped 1,000 times and mean and median Bayesian information criterion (BIC) scores were extracted. The algorithm with the mean and median larger BIC scores was chosen since bnlearn re-scales the BIC scores by \u0026minus;\u0026thinsp;2, meaning that the larger the BIC scores, the better the Bayesian network explains the data.\u003c/p\u003e \u003cp\u003eThe best algorithm was then bootstrapped 1,000 times to estimate the uncertainty of the edge\u0026rsquo;s strength and the direction of the network as previously described [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Edges showing presence in at least 60% (strength) among all the 1,000 models were kept in the Bayesian network through model averaging. Model estimates were extracted for each permutation to calculate a mean estimate with a 95% confidence interval for each edge to understand if the parent node was positively or negatively predicting the child node.\u003c/p\u003e \u003cp\u003eOnce the network was created, it was exported and edited using Metscape 2 [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e] within the CytoScape 3.8 platform. Only edges and nodes that were part of the network connected to the latent variable composed of opportunistic pathogenic bacteria at the day of diagnosis were kept in the final network for ease of interpretation.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eLatent variables identification\u003c/h2\u003e \u003cp\u003eWithin each metabolomic dataset, five principal components (PC) were generated as latent variables. These components were sequentially labeled from PC1 to PC5, representing distinct dimensions of variation captured from each dataset. For the microbiome datasets, one PC at parturition and one PC at metritis diagnosis were generated. Only PCs that remained in the Bayesian network will be described. Full details about each PC can be found in SI Results. Each dataset explained variation by each PC can be found in Supplemental Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e. Loadings for each plasma and uterine PCs can be found in Supplemental Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e and Supplemental Table \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003e, respectively. Association between each PC and metritis can be found in Supplemental Table \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003e.\u003c/p\u003e \u003cp\u003eFor plasma metabolome prepartum, PC1 and PC3 remained in the Bayesian network. Plasma prepartum PC1 explained 14.7% of the variation in the data (Supplemental Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and was composed of 49 metabolites, of which 75.3% were amino acids (Supplemental Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e; Supplemental Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eA). Plasma prepartum PC3 explained 4.8% of the variation in the data (Supplemental Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and was composed of 5 metabolites, of which 53.8% were carbohydrates (Supplemental Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e; Supplemental Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eC). Both plasma prepartum PC1 (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.01) and PC3 (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01) were lesser in cows that developed metritis when compared with cows that did not develop metritis (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e; Supplemental Table \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFor plasma metabolome at parturition, PC1, PC2, PC3, and PC4 remained in the Bayesian network. Plasma PC1 at parturition explained 11.0% of the variation in the data (Supplemental Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and was composed of 18 metabolites, of which 95.9% were amino acids (Supplemental Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e; Supplemental Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eA). Plasma PC2 at parturition explained 8.6% of the variation in the data (Supplemental Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and was composed of 6 metabolites, of which 100.0% were lipids (Supplemental Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e; Supplemental Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eB). Plasma PC3 at parturition explained 6.9% of the variation in the data (Supplemental Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and was composed of 7 metabolites, of which 71.2% were carbohydrates (Supplemental Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e; Supplemental Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eC). Plasma PC4 at parturition explained 4.6% of the variation in the data (Supplemental Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and was composed of 24 metabolites, of which 45.3% were carbohydrates (Supplemental Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e; Supplemental Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eD). Plasma PC1 (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.08) and PC4 (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.04) at parturition were lesser, and PC2 (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.02) and PC3 (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.02) at parturition were greater in cows that developed metritis when compared with cows that did not develop metritis (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e; Supplemental Table \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFor plasma metabolome at diagnosis, PC1 and PC5 remained in the Bayesian network. Plasma PC1 at diagnosis explained 11.7% of the variation in the data (Supplemental Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and was composed of 27 metabolites, of which 63.6% were amino acids (Supplemental Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e; Supplemental Figure \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003eA). Plasma PC5 at diagnosis explained 3.9% of the variation in the data (Supplemental Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and was composed of 48 metabolites, of which 53.1% were amino acids (Supplemental Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e; Supplemental Figure \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003eE). Plasma PC1 at diagnosis (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01) was lesser and PC5 at diagnosis (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01) was greater in cows that developed metritis when compared with cows that did not develop metritis (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e; Supplemental Table \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFor uterine metabolome at parturition, PC5 remained in the Bayesian network. Uterine PC5 at parturition explained 4.0% of the variation in the data (Supplemental Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and was composed of 45 metabolites, of which 67.9% were carbohydrates (Supplemental Table \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003e; Supplemental Figure \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003eE). Uterine PC5 at parturition was lesser (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.04) in cows that developed metritis when compared with cows that did not develop metritis (Supplemental Table \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFor uterine metabolome at diagnosis, PC2 and PC3 remained in the Bayesian network. Uterine PC2 at diagnosis explained 15.5% of the variation in the data (Supplemental Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and was composed of 44 metabolites, of which 28.6% were uncategorized metabolites (Supplemental Table \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003e; Supplemental Figure \u003cspan refid=\"MOESM5\" class=\"InternalRef\"\u003eS5\u003c/span\u003eB). Uterine PC3 at diagnosis explained 7.5% of the variation in the data (Supplemental Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and was composed of 43 metabolites, of which 50.2% were amino acids (Supplemental Table \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003e; Supplemental Figure \u003cspan refid=\"MOESM5\" class=\"InternalRef\"\u003eS5\u003c/span\u003eC). Uterine metabolome PC2 at diagnosis (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01) was lesser and uterine PC3 at diagnosis (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01) was greater in cows that developed metritis when compared with cows that did not develop metritis (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e; Supplemental Table \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFor uterine microbiome at parturition, PC1 was composed of 3 bacteria. The 3 bacteria, which explained 100% of the variation in PC1, were \u003cem\u003ePorphyromonas\u003c/em\u003e (34.3%), \u003cem\u003eBacteroides\u003c/em\u003e (33.5%), and \u003cem\u003eFusobacterium\u003c/em\u003e (32.2%). For uterine microbiome at diagnosis, PC1 was composed of 3 bacteria. The 3 bacteria, which explained 100% of the variation in PC1, were \u003cem\u003eBacteroides\u003c/em\u003e (34.9%), \u003cem\u003ePorphyromonas\u003c/em\u003e (34.5%), and \u003cem\u003eFusobacterium\u003c/em\u003e (30.6%).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eBayesian Network\u003c/h2\u003e \u003cp\u003eA total of 55 edges connected 37 nodes with a strength greater or equal to 60% (Supplemental Table \u003cspan refid=\"MOESM5\" class=\"InternalRef\"\u003eS5\u003c/span\u003e). Only 39 edges and 28 nodes remained in the final network (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) either because they were part of a pathway connected to Microbiome PC1 on the day of metritis diagnosis or because they were of biological interest.\u003c/p\u003e \u003cp\u003eIn the prepartum, ppBW and IFN-γ were not impacted by other nodes. Body weight change and Live cells were negatively impacted by ppBW. Polymorphonuclear cells were negatively impacted by Live cells. Plasma PC1, composed of 75.3% amino acids, was positively impacted by BWC and Live cells, and negatively impacted by PMN. Plasma PC3, composed of 53.8% carbohydrates, was positively impacted by Live cells and negatively impacted by PMN. Interleukin 8 was negatively impacted by plasma PC3.\u003c/p\u003e \u003cp\u003eAt parturition, plasma PC1, composed of 95.9% amino acids, was not impacted by other nodes. Microbiome PC1, composed of \u003cem\u003eBacteroides\u003c/em\u003e, \u003cem\u003ePorphyromonas\u003c/em\u003e, and \u003cem\u003eFusobacterium\u003c/em\u003e, was negatively impacted by prepartum plasma PC3. IL-8 was positively impacted by prepartum IL-8. Plasma PC4, composed of 45.3% carbohydrates, was positively impacted by prepartum Live cells and prepartum plasma PC3. Monocyte activation was positively impacted by plasma PC4. B-cell activation was negatively impacted by Monocyte activation and plasma PC1 and positively impacted by prepartum PMN. Uterine PC5, composed of 67.9% carbohydrates, was negatively impacted by prepartum IFN-γ and positively impacted by plasma PC1. Interferon γ was positively impacted by prepartum IFN-γ. Interleukin 1β was positively impacted by IFN-γ. Plasma PC2, composed of 100.0% lipids, was negatively impacted by BWC, and plasma PC3, composed of 71.2% carbohydrates, was positively impacted by prepartum PMN.\u003c/p\u003e \u003cp\u003eOn the day of metritis diagnosis, plasma PC1, composed of 63.6% amino acids, was not impacted by other nodes. B-cell activation was positively impacted by parturition B-cell activation. Monocytes MHCII was positively impacted by plasma PC1 and negatively impacted by parturition IL-1β. Plasma PC5, composed of 53.1% amino acids, was positively impacted by parturition plasma PC3. Uterine PC2, composed of 28.6% uncategorized metabolites, was positively impacted by plasma PC1 and negatively impacted by plasma PC5. Polymorphonuclear cells activation was positively impacted by parturition plasma PC2. Polymorphonuclear cells were negatively impacted by B-cell activation, PMN activation, and parturition IL-8 and positively impacted by uterine PC5. Uterine PC3, composed of 50.2% amino acids, was negatively impacted by PMN. Microbiome PC1, composed of \u003cem\u003eBacteroides\u003c/em\u003e, \u003cem\u003ePorphyromonas\u003c/em\u003e, and \u003cem\u003eFusobacterium\u003c/em\u003e, was negatively impacted by uterine PC2, and positively impacted by uterine PC3 and parturition microbiome PC1.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eHerein, Bayesian networks were applied to infer potential causal relationships between the uterine microbiome, uterine and plasma metabolome, and systemic immune profiling to advance the understanding of metritis development in dairy cows. Bayesian network analysis is a powerful statistical method that can suggest potential causal relationships among variables, which can aid in the development of strong mechanistic hypotheses for further experimental testing [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAccording to the directionality network, the prepartum immune and metabolic status of cows were impacted by their ppBW. In dairy cows, body weight is strongly correlated with body condition score (BCS), a measure of subcutaneous adiposity [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e], hence, we speculate that heavier cows were fatter. The greater ppBW in cows that developed metritis negatively impacted the prepartum Live cells and the BWC leading to heavier cows experiencing more cell death and greater BW loss. The greater level of fatty acid accumulation in adipocytes from fatter animals activates NADPH oxidase, increasing the production of reactive oxygen species (ROS) and their release to circulation [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Elevated systemic ROS levels cause cell death [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Furthermore, cows with greater BCS experience a more pronounced negative nutrient balance prior to parturition, leading to greater BW loss [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. The greater cell death observed in cows that developed metritis led to an increase in prepartum PMN. Following cell death, intracellular components are released to circulation triggering immune cell recruitment [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. It is possible that PMN were released from the bone marrow or spleen as a response to cell death-associated oxidative stress [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. Greater cell death and PMN led to a reduction in prepartum plasma PC3, which was composed of fructose, tagatose, methionine sulfoxide, glycerol-α-phosphate, and serotonin. All these metabolites except for serotonin had positive loadings (i.e. positive correlations), whereas serotonin had a negative loading, meaning that the observed reduction in prepartum plasma PC3 in cows that developed metritis indicates a reduction in all of these metabolites but an increase in serotonin. Fructose is a sugar that can enter the glycolytic pathway at various points [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e], while glycerol-α-phosphate is an intermediate in glycolysis. The observed decrease of these metabolites may reflect an overall usage of these glycolytic substrates by PMN, which upon activation shift their metabolism from oxidative phosphorylation to glycolysis [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. Under conditions of elevated ROS and cell death, methionine residues in proteins are oxidized to methionine sulfoxide [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. In an attempt to reverse the oxidative damage, methionine sulfoxide reductases catalyze the reduction of methionine sulfoxide back to methionine, thereby, reducing methionine sulfoxide concentrations [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. The observed decrease in methionine sulfoxide levels in cows experiencing greater cell death may indicate a response to a greater oxidative challenge. Although most of the circulating serotonin is synthetized by the enterochromaffin cells of the gastrointestinal tract, leukocytes are able to synthesize and store serotonin [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. It is possible that following cell death, intracellular serotonin is released; thus, increasing circulating serotonin levels. Serotonin enhances the production of IL-8 by dendritic cells [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e], which may help explain why plasma PC3 led to an increase in prepartum IL-8.\u003c/p\u003e \u003cp\u003eAccording to the directionality network, the immune and metabolic status of cows at parturition were impacted by changes observed in the prepartum. The greater BW loss in cows that developed metritis led to an increase in plasma PC2 at parturition. Plasma PC2 was composed of fatty acids, indicating that heavier cows were losing more weight, consequently mobilizing more adipose tissue. Furthermore, the greater prepartum cell death and the reduction in prepartum plasma PC3 led to a decrease in plasma PC4 at parturition. Similar to prepartum plasma PC3, plasma PC4 at parturition was mostly composed of tagatose, fructose, and methionine sulfoxide, which indicates that the response to oxidative stress observed in the prepartum was maintained at parturition. The decrease in parturition plasma PC4 led to a reduction in monocyte activation at parturition. It is possible that the reduction in glycolytic substrates hindered monocyte activation. Contrarily to neutrophils, monocytes lack glycogen storages; thus, they depend entirely on extracellular carbohydrates for activation [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]; making them more sensible to low carbohydrate levels. The decrease in monocyte activation, the decrease in parturition plasma PC1, composed of amino acids, and the increase in prepartum PMN, which was impacted by prepartum cell death, led to an increase in B-cell activation at parturition. Although not included in the Bayesian network analysis presented herein because of the high collinearity with B-cell activation at parturition, B-cell activation prepartum was directly impacted by prepartum cell death (data not shown), indicating that B-cell activation is part of the immune response to cell death [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. The decrease in prepartum plasma PC3 led to an increase in parturition microbiome PC1, composed of \u003cem\u003eBacteroides, Porphyromonas\u003c/em\u003e, and \u003cem\u003eFusobacterium\u003c/em\u003e. In humans, bacteria present in the gut sense serotonin levels, which impact gene expression related to biofilm formation, adhesion, motility, and virulence [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. For example, serotonin impacts quorum sensing pathways increasing virulence of \u003cem\u003ePseudomonas aeruginosa\u003c/em\u003e, leading to \u003cem\u003ePseudomonas aeruginosa\u003c/em\u003e overgrowth [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. At parturition, dairy cows have degenerative vascular changes in uterine small blood vessels [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e], which has been proposed as a mechanism by which blood components leak into the uterine lumen [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. It is possible that circulating serotonin seeps into the uterus and affects the uterine microbiome favoring opportunistic pathogenic bacteria overgrowth. It is worth noting that, on average, uterine samples were collected approximately 12h after parturition, which allowed for bacterial proliferation [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. In fact, we previously observed that uterine serotonin was greater in cows with metritis than in cows without metritis, and that was highly correlated with uterine pathogenic bacteria [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAccording to the directionality network, the immune and metabolic status of cows at the day of metritis diagnosis were impacted by changes observed in the prepartum and at parturition. The greater parturition plasma PC2, composed of fatty acids, led to a greater PMN activation at the day of metritis diagnosis in cows with metritis. Treating bovine neutrophils with long chain fatty acids increased their adhesion and chemotaxis [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e], indicating a possible direct effect of greater circulating fatty acids on PMN activation. The observed greater B-cell activation at parturition was maintained at metritis diagnosis, and together with greater PMN activation at the day of metritis diagnosis, greater parturition IL-8, and lesser parturition uterine PC5, mostly composed of carbohydrates, led to a reduction in circulating PMN at metritis diagnosis. Greater levels of IL-8, which is the main PMN chemokine [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e], and greater expression of CD62L, which is an adhesion molecule on PMN [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e], likely promoted PMN migration to the uterus, leading to a reduction in circulating PMN. In fact, PMN are more abundant in the endometrium of cows with metritis than in the endometrium of cows without metritis [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. Despite the greater PMN migration to the uterus in cows with metritis, circulating PMN from cows with metritis have lesser intracellular killing capacity when compared with cows without metritis [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. It is possible that these dysfunctional PMN arriving in the uterus are less effective at eliminating pathogenic bacteria, resulting in a persistent chemo attractive signal for additional PMN recruitment. The lesser circulating PMN in cows that developed metritis led to an increase in uterine PC3 on the day of metritis diagnosis in cows with metritis, which consequently led to an increase in \u003cem\u003eBacteroides\u003c/em\u003e, \u003cem\u003ePorphyromonas\u003c/em\u003e, and \u003cem\u003eFusobacterium\u003c/em\u003e at diagnosis. Uterine PC3 at diagnosis was mostly composed of amino and organic acids. Upon arrival to tissues, activated PMN release proteolytic enzymes [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e], leading to endometrial damage [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. Tissue damage triggers the breakdown of proteins, releasing dipeptides, such as ile-ile and alanine-alanine, and amino acids such as tyrosine and phenylalanine [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. Furthermore, cellular stress and metabolic alterations during tissue injury leads to an increase in several other metabolites, including organic acids such as succinic and fumaric acids [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. Greater tissue damage creates an environment that may promote opportunistic pathogenic bacterial proliferation by increasing substrate availability [\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e], which might explain how the greater levels of uterine PC3 in cows with metritis positively impacted the observed increase in \u003cem\u003eBacteroides\u003c/em\u003e, \u003cem\u003ePorphyromonas\u003c/em\u003e, and \u003cem\u003eFusobacterium\u003c/em\u003e. The greater parturition IL-1β together with the lesser diagnosis plasma PC1, composed of amino and organic acids, led to a reduction in monocytes MHCII expression at diagnosis. The lesser circulating amino acids may impair monocyte antigen presentation capacity. It is also possible that these cells were experiencing immune tolerance. The greater parturition IL-1β was positively impacted by parturition IFN-γ, which was positively impacted by prepartum IFN-γ. Prepartum IFN-γ was not influenced by any node. The origin of the greater prepartum IFN-γ in cows that developed metritis is, therefore, not clear. In obese non-ruminants, T-cells surrounding necrotic adipocytes produce high levels of IFN-γ [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]; thus, we hypothesize that the greater prepartum IFN-γ in cows that developed metritis may originate from adipocyte-associated T-cells. Nonetheless, the persistent systemic inflammation observed since the prepartum led to a reduction in monocyte antigen presentation capacity. In a previous study, we hypothesized that the reduction in monocyte antigen presentation capacity observed postpartum in cows with metritis could be due to immune tolerance [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. In fact, prolonged inflammatory responses are known for leading to immune tolerance [\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e], particularly reducing monocytes and macrophages activation [\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e] but not PMN activation [\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e]. It is noteworthy that we did not detect any influence of the reduction in monocyte antigen presentation capacity on the abundance of opportunistic pathogenic bacteria; however, it is possible that if circulatory monocytes are experiencing immune tolerance, uterine monocytes and macrophages could also be experiencing tolerance. Monocytes and macrophages play a key role in the resolution of inflammation [\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e]; therefore, their impaired response could hinder the resolution of inflammation, potentially leading to greater tissue damage by PMN.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis observational study provides insights into the complex relationship between ppBW, BW loss, systemic immune profiling, and uterine and plasma metabolome around parturition, and how they may promote uterine opportunistic pathogens growth leading to metritis development. Altogether, directionality network analysis indicates that greater prepartum BW led to greater BW loss and prepartum systemic cellular death, which led to an increase in systemic inflammation and immune activation. The aforementioned changes led to an increase in PMN extravasation, which led to an increase in uterine metabolomic changes associated with tissue damage. These changes led to an increase in \u003cem\u003eBacteroides\u003c/em\u003e, \u003cem\u003ePorphyromonas\u003c/em\u003e and \u003cem\u003eFusobacterium\u003c/em\u003e in cows that developed metritis, indicating that excessive tissue damage may promote opportunistic pathogenic bacterial growth. Our working hypothesis is illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Experimental research needs to be conducted to validate the mechanistic hypotheses generated herein. In conclusion, the directionality network showed that prepartum BW led to immune dysregulation and plasma and uterine metabolomic changes leading to opportunistic pathogens overgrowth in cows that developed metritis.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003e\u003cstrong\u003eBW:\u003c/strong\u003e Body weight\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eppBW:\u003c/strong\u003e Prepartum body weight\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBWC:\u003c/strong\u003e Daily body weight change\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePCA:\u003c/strong\u003e Principal component analysis\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003esPCA:\u003c/strong\u003e Sparse Principal component analysis\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBIC:\u0026nbsp;\u003c/strong\u003eBayesian information criterion\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePC:\u003c/strong\u003e Principal component\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBCS:\u003c/strong\u003e Body condition score\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eROS:\u003c/strong\u003e Reactive oxygen species\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cem\u003eEthics approval and consent to participate \u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAll procedures involving cows were approved by the Institutional Animal Care and Use Committee of the University of Florida; protocol number 201910623.\u003c/p\u003e\n\n\u003cp\u003e\u003cem\u003eConsent for publication\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\n\u003cp\u003e\u003cem\u003eAvailability of data and material\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe raw sequence data used in the current study were generated during our previous study [6] and are available in the NCBI repository under BioProject OR883023 (www.ncbi.nlm.nih.gov/nuccore/OR883023) - OR883397 (www.ncbi.nlm.nih.gov/nuccore/OR883397).\u003c/p\u003e\n\u003cp\u003eThe metabolomics dataset used in the current study was generated during our two previous studies [6, 13] and are available in the NIH Common Fund\u0026rsquo;s National Metabolomics Data Repository website, the Metabolomics Workbench repository [50]. Uterine metabolomics dataset [6]\u003cstrong\u003e \u003c/strong\u003ecan be found under Study ID ST002994, http://dx.doi.org/10.21228/M8S425. Plasma metabolomics dataset [13] can be found under Study ID ST002556; http://dx.doi.org/10.21228/M8PF0K . The Metabolomics Workbench is supported by NIH grant U2C-DK119886 and OT2-OD030544 grants.\u003c/p\u003e\n\n\u003cp\u003e\u003cem\u003eCompeting interests\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\n\u003cp\u003e\u003cem\u003eFunding\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by U.S. Department of Agriculture Grant # 2019-67015-29836, Accession No: 1019435.\u003c/p\u003e\n\n\u003cp\u003e\u003cem\u003eAuthors\u0026apos; contributions\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eJEPS, CDN, SJJ, RCB, JPD, and KNG initiated the study design and provided scientific input during manuscript writing. RSB and RCC provided scientific input during manuscript writing. SC, JGP, ACM, and TDG collected the on-farm data. SC and FC performed the laboratory work. SC performed the statistical analyses. KNG and FC provided their scientific input during statistical analyses. SC and KNG wrote the manuscript. All authors edited and approved the manuscript.\u003c/p\u003e\n\n\u003cp\u003e\u003cem\u003eAcknowledgements\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eWe would like to thank the general manager, Mr. Eric Williams, and the staff of the University of Florida Dairy Unit for allowing the use of their animals and facilities.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003ePinedo P, Santos JEP, Chebel RC, Galv\u0026atilde;o KN, Schuenemann GM, Bicalho RC, et al. Early-lactation diseases and fertility in 2 seasons of calving across US dairy herds. J Dairy Sci. 2020;103:10560\u0026ndash;76.\u003c/li\u003e\n\u003cli\u003eBarragan AA, Pi\u0026ntilde;eiro JM, Schuenemann GM, Rajala-Schultz PJ, Sanders DE, Lakritz J, et al. Assessment of daily activity patterns and biomarkers of pain, inflammation, and stress in lactating dairy cows diagnosed with clinical metritis. J Dairy Sci. 2018;101:8248\u0026ndash;58.\u003c/li\u003e\n\u003cli\u003eFigueiredo CC, Merenda VR, de Oliveira EB, Lima FS, Chebel RC, Galv\u0026atilde;o KN, et al. Failure of clinical cure in dairy cows treated for metritis is associated with reduced productive and reproductive performance. J Dairy Sci. 2021;104:7056\u0026ndash;70.\u003c/li\u003e\n\u003cli\u003eP\u0026eacute;rez-B\u0026aacute;ez J, Risco CA, Chebel RC, Gomes GC, Greco LF, Tao S, et al. Association of dry matter intake and energy balance prepartum and postpartum with health disorders postpartum: Part I. Calving disorders and metritis. J Dairy Sci. 2019;102:9138\u0026ndash;50.\u003c/li\u003e\n\u003cli\u003eJeon SJ, Vieira-Neto A, Gobikrushanth M, Daetz R, Mingoti RD, Parize ACB, et al. Uterine microbiota progression from calving until establishment of metritis in dairy cows. Appl Environ Microbiol. 2015;81:6324\u0026ndash;32.\u003c/li\u003e\n\u003cli\u003eCasaro S, Prim JG, Gonzalez TD, Cunha F, Bisinotto RS, Chebel RC, et al. Integrating uterine microbiome and metabolome to advance the understanding of the uterine environment in dairy cows with metritis. Animal Microbiome 2024 6:1. 2024;6:1\u0026ndash;13.\u003c/li\u003e\n\u003cli\u003eJeon SJ, Cunha F, Ma X, Martinez N, Vieira-Neto A, Daetz R, et al. Uterine microbiota and immune parameters associated with fever in dairy cows with metritis. PLoS One. 2016;11.\u003c/li\u003e\n\u003cli\u003eGalv\u0026atilde;o KN, Bicalho RC, Jeon SJ. Symposium review: The uterine microbiome associated with the development of uterine disease in dairy cows. J Dairy Sci. 2019;102:11786\u0026ndash;97.\u003c/li\u003e\n\u003cli\u003eCasaro S, Prim J, Gonzalez T, Figueiredo C, Bisinotto R, Chebel R, et al. Blood metabolomics and impacted cellular mechanisms during transition into lactation in dairy cows that develop metritis. J Dairy Sci. 2023. https://doi.org/10.3168/jds.2023-23433.\u003c/li\u003e\n\u003cli\u003eCasaro S, Prim JG, Gonzalez TD, Bisinotto RS, Chebel RC, Marrero MG, et al. Unraveling the immune and metabolic changes associated with metritis in dairy cows. J Dairy Sci. 2023;106:9244\u0026ndash;59.\u003c/li\u003e\n\u003cli\u003eCasaro S, P\u0026eacute;rez-B\u0026aacute;ez J, Bisinotto RS, Chebel RC, Prim JG, Gonzalez TD, et al. Association between prepartum body condition score and prepartum and postpartum dry matter intake and energy balance in multiparous Holstein cows. J Dairy Sci. 2024. https://doi.org/10.3168/JDS.2023-24047.\u003c/li\u003e\n\u003cli\u003eMartinez N, Risco CA, Lima FS, Bisinotto RS, Greco LF, Ribeiro ES, et al. Evaluation of peripartal calcium status, energetic profile, and neutrophil function in dairy cows at low or high risk of developing uterine disease. J Dairy Sci. 2012;95:7158\u0026ndash;72.\u003c/li\u003e\n\u003cli\u003eCasaro S, Prim JG, Gonzalez TD, Figueiredo CC, Bisinotto RS, Chebel RC, et al. Blood metabolomics and impacted cellular mechanisms during transition into lactation in dairy cows that develop metritis. J Dairy Sci. 2023;106:8098\u0026ndash;109.\u003c/li\u003e\n\u003cli\u003eChu X, Jaeger M, Beumer J, Bakker OB, Aguirre-Gamboa R, Oosting M, et al. Integration of metabolomics, genomics, and immune phenotypes reveals the causal roles of metabolites in disease. Genome Biol. 2021;22.\u003c/li\u003e\n\u003cli\u003eJeon SJ, Cunha F, Vieira-Neto A, Bicalho RC, Lima S, Bicalho ML, et al. Blood as a route of transmission of uterine pathogens from the gut to the uterus in cows. Microbiome. 2017;5:109.\u003c/li\u003e\n\u003cli\u003eArchbald LF, Schultz RH, Fahning ML, Kurtz HJ, Zemjanis R. A sequential histological study of the post-partum bovine uterus. J Reprod Fertil. 1972;29:133\u0026ndash;6.\u003c/li\u003e\n\u003cli\u003eR Core Team. R: A Language and Environment for Statistical Computing. 2023.\u003c/li\u003e\n\u003cli\u003eRohart F, Gautier B, Singh A, L\u0026ecirc; Cao KA. mixOmics: An R package for \u0026rsquo;omics feature selection and multiple data integration. PLoS Comput Biol. 2017;13.\u003c/li\u003e\n\u003cli\u003eShen H, Huang JZ. Sparse principal component analysis via regularized low rank matrix approximation. J Multivar Anal. 2008;99:1015\u0026ndash;34.\u003c/li\u003e\n\u003cli\u003eScutari M. Learning Bayesian Networks with the bnlearn R Package. J Stat Softw. 2010;35:1\u0026ndash;22.\u003c/li\u003e\n\u003cli\u003eYu H, Campbell MT, Zhang Q, Walia H, Morota G. Genomic Bayesian Confirmatory Factor Analysis and Bayesian Network To Characterize a Wide Spectrum of Rice Phenotypes. G3 Genes|Genomes|Genetics. 2019;9:1975\u0026ndash;86.\u003c/li\u003e\n\u003cli\u003eNovais FJ de, Yu H, Cesar ASM, Momen M, Poleti MD, Petry B, et al. Multi-omic data integration for the study of production, carcass, and meat quality traits in Nellore cattle. Front Genet. 2022;13:948240.\u003c/li\u003e\n\u003cli\u003eKarnovsky A, Weymouth T, Hull T, Glenn Tarcea V, Scardoni G, Laudanna C, et al. Metscape 2 bioinformatics tool for the analysis and visualization of metabolomics and gene expression data. Bioinformatics. 2012;28:373\u0026ndash;80.\u003c/li\u003e\n\u003cli\u003eFurukawa S, Fujita T, Shimabukuro M, Iwaki M, Yamada Y, Nakajima Y, et al. Increased oxidative stress in obesity and its impact on metabolic syndrome. Journal of Clinical Investigation. 2004;114:1752\u0026ndash;61.\u003c/li\u003e\n\u003cli\u003eRoh JS, Sohn DH. Damage-Associated Molecular Patterns in Inflammatory Diseases. Immune Netw. 2018;18.\u003c/li\u003e\n\u003cli\u003eMulder PPG, Vlig M, Boekema BKHL, Stoop MM, Pijpe A, van Zuijlen PPM, et al. Persistent Systemic Inflammation in Patients With Severe Burn Injury Is Accompanied by Influx of Immature Neutrophils and Shifts in T Cell Subsets and Cytokine Profiles. Front Immunol. 2021;11 January:1\u0026ndash;13.\u003c/li\u003e\n\u003cli\u003eDholariya SJ, Orrick JA. Biochemistry, Fructose Metabolism. StatPearls Publishing; 2024.\u003c/li\u003e\n\u003cli\u003eO\u0026rsquo;Neill LAJ, Kishton RJ, Rathmell J. A guide to immunometabolism for immunologists. Nat Rev Immunol. 2016;16:553\u0026ndash;65.\u003c/li\u003e\n\u003cli\u003eWeissbach H, Resnick L, Brot N. Methionine sulfoxide reductases: history and cellular role in protecting against oxidative damage. Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics. 2005;1703:203\u0026ndash;12.\u003c/li\u003e\n\u003cli\u003eBaganz NL, Blakely RD. A dialogue between the immune system and brain, spoken in the language of serotonin. ACS Chem Neurosci. 2013;4:48\u0026ndash;63.\u003c/li\u003e\n\u003cli\u003eIdzko M, Panther E, Stratz C, M\u0026uuml;ller T, Bayer H, Zissel G, et al. The serotoninergic receptors of human dendritic cells: identification and coupling to cytokine release. J Immunol. 2004;172:6011\u0026ndash;9.\u003c/li\u003e\n\u003cli\u003eO\u0026rsquo;Neill LAJ. A Broken Krebs Cycle in Macrophages. Immunity. 2015;42:393\u0026ndash;4.\u003c/li\u003e\n\u003cli\u003eGong T, Liu L, Jiang W, Zhou R. DAMP-sensing receptors in sterile inflammation and inflammatory diseases. Nat Rev Immunol. 2020;20:95\u0026ndash;112.\u003c/li\u003e\n\u003cli\u003eEverett BA, Tran P, Prindle A. Toward manipulating serotonin signaling via the microbiota\u0026ndash;gut\u0026ndash;brain axis. Curr Opin Biotechnol. 2022;78:102826.\u003c/li\u003e\n\u003cli\u003eKnecht LD, O\u0026rsquo;Connor G, Mittal R, Liu XZ, Daftarian P, Deo SK, et al. Serotonin Activates Bacterial Quorum Sensing and Enhances the Virulence of Pseudomonas aeruginosa in the Host. EBioMedicine. 2016;9:161\u0026ndash;9.\u003c/li\u003e\n\u003cli\u003eBacic MK, Smith CJ. Laboratory maintenance and cultivation of bacteroides species. Curr Protoc Microbiol. 2008;Chapter 13 SUPPL. 9.\u003c/li\u003e\n\u003cli\u003eMena J, Manosalva C, Ramirez R, Chandia L, Carroza D, Loaiza A, et al. Linoleic acid increases adhesion, chemotaxis, granule release, intracellular calcium mobilisation, MAPK phosphorylation and gene expression in bovine neutrophils. Vet Immunol Immunopathol. 2013;151:275\u0026ndash;84.\u003c/li\u003e\n\u003cli\u003eTajima A, Iwase T, Shinji H, Seki K, Mizunoe Y. Inhibition of endothelial interleukin-8 production and neutrophil transmigration by Staphylococcus aureus beta-hemolysin. Infect Immun. 2009;77:327\u0026ndash;34.\u003c/li\u003e\n\u003cli\u003eSicsic R, Goshen T, Dutta R, Kedem-Vaanunu N, Kaplan-Shabtai V, Pasternak Z, et al. Microbial communities and inflammatory response in the endometrium differ between normal and metritic dairy cows at 5-10 days post-partum. Vet Res. 2018;49:77.\u003c/li\u003e\n\u003cli\u003eHammon DS, Evjen IM, Dhiman TR, Goff JP, Walters JL. Neutrophil function and energy status in Holstein cows with uterine health disorders. Vet Immunol Immunopathol. 2006;113:21\u0026ndash;9.\u003c/li\u003e\n\u003cli\u003eTecchio C, Cassatella MA. Neutrophil-derived chemokines on the road to immunity. Semin Immunol. 2016;28:119\u0026ndash;28.\u003c/li\u003e\n\u003cli\u003eWang J. Neutrophils in tissue injury and repair. Cell Tissue Res. 2018;371:531\u0026ndash;9.\u003c/li\u003e\n\u003cli\u003eZindel J, Kubes P. DAMPs, PAMPs, and LAMPs in Immunity and Sterile Inflammation. Annual Review of Pathology: Mechanisms of Disease. 2020;15:493\u0026ndash;518.\u003c/li\u003e\n\u003cli\u003eAnderson CJ, Medina CB, Barron BJ, Karvelyte L, Aaes TL, Lambertz I, et al. Microbes exploit death-induced nutrient release by gut epithelial cells. Nature 2021 596:7871. 2021;596:262\u0026ndash;7.\u003c/li\u003e\n\u003cli\u003eRocha VZ, Folco EJ, Sukhova G, Shimizu K, Gotsman I, Vernon AH, et al. Interferon-\u0026gamma;, a Th1 cytokine, regulates fat inflammation: A role for adaptive immunity in obesity. Circ Res. 2008;103:467\u0026ndash;76.\u003c/li\u003e\n\u003cli\u003eRogovskii V. Immune Tolerance as the Physiologic Counterpart of Chronic Inflammation. Front Immunol. 2020;11 September:1\u0026ndash;7.\u003c/li\u003e\n\u003cli\u003eFern\u0026aacute;ndez-Ruiz I, Arnalich F, Cubillos-Zapata C, Hern\u0026aacute;ndez-Jim\u0026eacute;nez E, Moreno-Gonz\u0026aacute;lez R, Toledano V, et al. Mitochondrial DAMPs induce endotoxin tolerance in human monocytes: An observation in patients with myocardial infarction. PLoS One. 2014;9.\u003c/li\u003e\n\u003cli\u003eAriga SK, Abatepaulo FB, Melo ESA, Velasco IT, Da Silva FP, De Lima TM, et al. Endotoxin tolerance drives Neutrophil to infectious site. Shock. 2014;42:168\u0026ndash;73.\u003c/li\u003e\n\u003cli\u003eParihar A, Eubank TD, Doseff AI. Monocytes and macrophages regulate immunity through dynamic networks of survival and cell death. J Innate Immun. 2010;2:204\u0026ndash;15.\u003c/li\u003e\n\u003cli\u003eSud M, Fahy E, Cotter D, Azam K, Vadivelu I, Burant C, et al. Metabolomics Workbench: An international repository for metabolomics data and metadata, metabolite standards, protocols, tutorials and training, and analysis tools. Nucleic Acids Res. 2016;44:D463\u0026ndash;70.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"animal-microbiome","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"amic","sideBox":"Learn more about [Animal Microbiome](http://animalmicrobiome.biomedcentral.com)","snPcode":"42523","submissionUrl":"https://submission.nature.com/new-submission/42523/3","title":"Animal Microbiome","twitterHandle":"@bmc","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"microbiome, metabolome, immune dysregulation, multi-omics, causal networks","lastPublishedDoi":"10.21203/rs.3.rs-4571697/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4571697/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eCows that develop metritis experience dysbiosis of their uterine microbiome, where opportunistic pathogens overtake uterine commensals. Whether opportunistic pathogens thrive and cause metritis may be determined by how efficiently the immune system responds to these pathogens. Nonetheless, periparturient cows experience immune dysregulation, which seems to be intensified by prepartum obesity and lipid mobilization Herein, Bayesian networks were applied to investigate the directional correlations between prepartum body weight (BW), BW loss, pre- and postpartum systemic immune profiling and plasma metabolome, and postpartum uterine metabolome and microbiome.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eAccording to the directionality network, greater prepartum BW led to greater BW loss, which led to an increase in plasma fatty acids at parturition, indicating that heavier cows were in lower energy balance than lighter cows. Greater prepartum BW also led to an increase in prepartum systemic cellular death, which led to an increase in systemic inflammation, immune activation, and metabolomic changes associated with oxidative stress prepartum and at parturition, which indicates a positive directional correlation between BW and systemic inflammation. These changes led to an increase in polymorphonuclear cell extravasation postpartum, which led to an increase in uterine metabolomic changes associated with tissue damage, suggesting that excessive polymorphonuclear cell migration to the uterus leads to excessive endometrial damage. These changes led to an increase in pathogenic bacteria in cows that developed metritis, suggesting that excessive tissue damage may disrupt physical barriers or increase substrate availability for bacterial growth.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eThis work provides robust mechanistic hypotheses for how prepartum body weight impacts peripartum immune and metabolic profiles, leading to uterine opportunistic pathogens overgrowth and metritis development.\u003c/p\u003e","manuscriptTitle":"Multi-omics integration and immune profiling identify possible causal networks leading to uterine microbiome dysbiosis in dairy cows that develop metritis","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-07-02 06:27:21","doi":"10.21203/rs.3.rs-4571697/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2024-09-11T07:10:58+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-08-20T09:19:03+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-08-14T16:00:45+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-08-14T14:28:50+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"108004871940875632568548044449592919910","date":"2024-07-31T07:33:25+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"324428634177732238367011256315911483568","date":"2024-07-30T10:18:09+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"255798908660219032357697170403621150774","date":"2024-07-30T09:16:33+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-07-30T07:58:51+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-06-19T21:39:53+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-06-16T02:59:22+00:00","index":"","fulltext":""},{"type":"submitted","content":"Animal Microbiome","date":"2024-06-12T16:44:17+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"animal-microbiome","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"amic","sideBox":"Learn more about [Animal Microbiome](http://animalmicrobiome.biomedcentral.com)","snPcode":"42523","submissionUrl":"https://submission.nature.com/new-submission/42523/3","title":"Animal Microbiome","twitterHandle":"@bmc","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"52f43679-25d4-4d89-9dd6-44a990a1f17d","owner":[],"postedDate":"July 2nd, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-01-13T15:58:29+00:00","versionOfRecord":{"articleIdentity":"rs-4571697","link":"https://doi.org/10.1186/s42523-024-00366-9","journal":{"identity":"animal-microbiome","isVorOnly":false,"title":"Animal Microbiome"},"publishedOn":"2025-01-09 15:56:54","publishedOnDateReadable":"January 9th, 2025"},"versionCreatedAt":"2024-07-02 06:27:21","video":"","vorDoi":"10.1186/s42523-024-00366-9","vorDoiUrl":"https://doi.org/10.1186/s42523-024-00366-9","workflowStages":[]},"version":"v1","identity":"rs-4571697","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4571697","identity":"rs-4571697","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}