Differences in uterine and serum metabolome associated with metritis in dairy cows

Objectives were to evaluate differences in the uterine and serum metabolomes associated with metritis in dairy cows. Vaginal discharge was evaluated using a Metricheck device (Simcro) at 5, 7, and 11 d in milk (DIM; herd 1) or 4, 6, 8, 10, and 12 DIM (herd 2). Cows with reddish or brownish, watery, and fetid discharge were diagnosed with metritis (n = 24). Cows with metritis were paired with herdmates without metritis (i.e., clear mucous vaginal discharge or clear lochia with ≤50% of pus) based on DIM and parity (n = 24). Day of metritis diagnosis was considered study d 0. All cows diagnosed with metritis received antimicrobial therapy. The metabolome of uterine lavage collected on d 0 and 5, and serum samples collected on d 0 were evaluated using untargeted gas chromatography time-of-flight mass spectrometry. Normalized data were subjected to multivariate canonical analysis of population using the MultBiplotR and MixOmics packages in R Studio. Uni-variate analyses including t -test, principal component analyses, partial least squares discriminant analyses, and pathway analyses were conducted using Metabo-analyst. The uterine metabolome differed between cows with and without metritis on d 0. Differences in the uterine metabolome associated with metritis on d 0 were related to the metabolism of butanoate, amino acids (i.e., glycine, serine, threonine, alanine, aspartate, and glutamate), glycolysis and gluconeogenesis, and the tricarboxylic acid cycle. No differences in the serum metabolome were observed between cows diagnosed with metritis and counterparts without metritis on d 0. Similarly, no differences in uterine metabolome were observed between cows with metritis and counterparts not diagnosed with metritis on d 5. These results indicate that the establishment of metritis in dairy cows is


INTRODUCTION
Metritis is a uterine disease of great importance to the dairy industry because it is highly prevalent (Pinedo et al., 2020;Merenda et al., 2021) and has detrimental effects on milk production, reproduction, survival, and welfare of dairy cows (Stojkov et al., 2015;Barragan et al., 2018;Figueiredo et al., 2021).Considering all the losses, a case of metritis has been estimated to cost between US$156 and $949 (Lima et al., 2019;Pérez-Báez et al., 2021;Silva et al., 2021).Metritis is caused by polymicrobial infection and is associated with major disturbances of uterine bacterial communities, with an increase of gram-negative bacteria from the genera Fusobacterium, Bacteroides, and Porphyromonas (Jeon et al., 2015;Galvão et al., 2019).Furthermore, whole metagenomic sequencing analyses have depicted changes in the function of uterine bacterial communities of cows with metritis, including metabolism of proteins and carbohydrates (Bicalho et al., 2017).Additionally, the uterine microbiome of cows with metritis has been shown to converge with that of cows without metritis between 6 and 12 d after diagnosis, and antimicrobial treatment has been shown to hasten that process (Galvão et al., 2020).
Considering the fluctuations in structure and function of uterine microbial populations, cellular debris resulting from pathogen activity and local immune response, and exudate secondary to extensive tissue damage (Bicalho et al., 2017;Sicsic et al., 2018), the composition of uterine contents is expected to differ between cows with metritis and counterparts without metritis.Metabolome analysis is a useful approach to generate insights on potential disease mechanisms that involve key metabolites secreted as a result of the interaction between pathogens and the host.Analyses of specific metabolites in serum and urine collected from postpartum dairy cows have found differences between those with and without metritis around the week of diagnosis (Dervishi et al., 2018(Dervishi et al., , 2021)).For instance, cows with metritis had greater serum concentrations of inflammatory and mitotoxicity biomarkers, such as haptoglobin and glutamate dehydrogenase, compared with cows without metritis (Dervishi et al., 2021).Moreover, evaluation of the urine metabolome has revealed that cows with metritis presented greater concentrations of several amino acids compared with cows without metritis (Dervishi et al., 2018).Despite knowledge regarding systemic differences in the metabolome associated with metritis, local differences in the uterine metabolome of dairy cows at metritis diagnosis and subsequent progression remain unexplored.Investigation of the serum metabolome in cows with metritis, using an untargeted approach, has also not been reported.In addition to advancing the current understanding of biological processes underlying uterine diseases in cattle, concurrent evaluation of uterine and serum metabolome is expected to further elucidate the interaction between local infection and inflammation and systemic changes associated with metritis.Furthermore, information regarding individual metabolites or metabolic pathways implicated in the establishment and progression of disease may play a pivotal role in the development of novel targets for disease prevention and therapy (Wang et al., 2011;Chen et al., 2016;Wishart, 2016;Jing et al., 2022).
The hypotheses of the present study were that the uterine and serum metabolome profiles of dairy cows with metritis differ from those of cows without metritis at the time of disease diagnosis.Moreover, we hypothesized that the uterine metabolome in cows treated with antimicrobials following diagnosis of metritis remains different from that of cows without metritis 5 d after the onset of treatment.The main objective of this study was to characterize the uterine and serum metabolomes associated with metritis at the time of diagnosis.A secondary objective was to evaluate the shift in the uterine metabolome from diagnosis until 5 d after treatment, compared with cows without metritis.

MATERIALS AND METHODS
All procedures involving animals were approved by the Animal Care and Use Committee of the University of Florida (protocol no. 201810204).

Study Population, Housing, and Management
This study was conducted from February to November 2018 in 2 dairy farms located in Florida, in the United States.The number of lactating cows and herd milk yield were 5,270 and 11,000 kg (herd 1) and 2,500 and 12,049 kg (herd 2).Lactating Holstein cows were milked thrice daily and fed a TMR twice daily to meet or exceed the nutritional requirements of a 650-kg cow producing 40 kg/d of 3.5% fat-corrected milk (NRC, 2001).Nutrient contents of the diet for each farm were as follows.

Experimental Design and Sample Size
This was a prospective cohort study.Cows calving from February to November 2018 in 2 dairy farms were followed until 5 d after metritis diagnosis.Sample size calculation was not performed because a formal approach has not been defined for metabolome studies, considering the inability to estimate mean differences and variability for all metabolites that comprise the acquired database.Conversely, sample size was defined based on previous studies in which the urine metabolomes and uterine microbiomes of dairy cows with and without metritis were compared.Dervishi et al. (2018) characterized differences in the urine metabolome between cows with and without metritis using an untargeted approach as described in the present study, with 6 cows per group.Jeon et al. (2015) depicted differences in the uterine microbiome between cows with and without metritis based on sequencing of the 16S rRNA gene with 12 cows per group.Inclusion of a larger number of cows per group (n = 24) was expected to ensure sufficient power for the characterization of differences in both uterine and serum metabolome associated with metritis in the current study.

Definition of Metritis and Sampling
Metritis was diagnosed based on visual evaluation of vaginal discharge at 5, 7, and 11 DIM (herd 1) or 4, 6, 8, 10, and 12 DIM (herd 2) using a Metricheck device (Simcro).Vaginal discharge was scored using a 5-point scale (1 = clear mucus or lochia; 2 = clear mucus with flecks of pus; 3 = mucopurulent discharge with <50% of pus; 4 = mucopurulent discharge with ≥50% of pus or without fetid reddish mucous discharge; 5 = watery, reddish or brownish, and fetid discharge) described for dairy cows (adapted from Chenault et al., 2004).Metritis was defined as the presence of vaginal discharge score 5 (MET; n = 24: 17 cows in herd 1 and 7 cows in herd 2; 6 primiparous and 18 multiparous cows).With the exception of 2 cows diagnosed with hyperketonemia and 6 cows with pyrexia (>39.5°C) on the day of metritis diagnosis, metritis was the sole clinical disease diagnosed in cows enrolled in this study.Cows without metritis (vaginal discharge score ≤3) and without any clinical disease including mastitis, hyperketonemia, lameness, displaced abomasum, indigestion, clinical hypocalcemia, and fever at enrollment were selected based on parity and calving date to serve as negative control (NOMET; n = 24: 17 cows in herd 1 and 7 cows in herd 2; 4 primiparous and 20 multiparous cows).The day of metritis diagnosis and pairing was considered study d 0. All cows diagnosed with metritis received systemic antimicrobial therapy starting on d 0, either 6.6 mg/kg of BW of ceftiofur crystalline free acid (Excede sterile suspension, Zoetis) subcutaneously injected in the posterior aspect of the ear where it attaches to the head (i.e., base of the ear) as recommended by the manufacturer twice 72 h apart (n = 21), or 11 mg/kg of BW of ampicillin trihydrate (Polyflex, Boehringer Ingelheim Vetmedica) intramuscularly injected once daily during 5 consecutive days (n = 3).
On d 0 before initiation of antimicrobial treatments, all cows were subjected to a low-volume uterine lavage performed by a single technician.Briefly, cows were palpated rectally for stabilization of the cervix, the vulva was cleaned with paper towels and alcohol (70% vol/vol), and a single-use plastic round-tip pipette (UterFlush pipettes, Van Beek) was introduced into the vagina at a 45° angle and manipulated through the cervix.A total of 30 mL of sterile saline solution (0.9% sodium chloride irrigation, Baxter) was infused into the uterine lumen using a 60-mL syringe (Covidien).Uterine contents were homogenized, retrieved into the same 60-mL syringe, and transferred to a sterile 15-mL conical tube (VWR).Blood was sampled on d 0 by puncture of the coccygeal vein or artery using evacu-ated tubes without anticoagulant (Vacutainer, Becton Dickinson).A second uterine lavage was performed on d 5 in all cows.All samples were collected after the first milking between 0630 and 0830 h.Samples were placed on ice immediately after collection and transported to the laboratory within 6 h of collection.Uterine lavage samples were aliquoted into 2-mL microcentrifuge tubes (Eppendorf) and stored at −80°C until assayed.Blood tubes were centrifuged at 4,000 × g for 15 min at 4°C, and serum was separated, aliquoted into 2-mL microcentrifuge tubes, and stored at −80°C until assayed.

Metabolome Analyses
Uterine lavage and serum samples were submitted for metabolome analyses to the University of California's West Coast Metabolomics Center (Davis, CA).Samples were analyzed in a single batch by untargeted gas chromatography with time-of-flight mass spectrometry (GC-TOF-MS).The methodology used for GC-TOF-MS has been described previously (Fiehn et al., 2008;Fiehn, 2016).A column comprising 95% dimethyl/5diphenyl polysiloxanesne was used, and helium gas was selected as carrier.Column temperatures were set between 50 and 330°C at a flow-rate of 1 mL/ min.Initial oven temperature was set at 50°C, followed by a 20°C increase per minute up to a final temperature of 330°C, which was held constant for a period of 5 min.Finally, injection temperature was set to begin at 50°C and to increase to 250°C by increments of 12°C per second.Retention of primary metabolites was evaluated using default settings from ChromaTOF v. 2.32, and quantification was reported as peak height.

Statistical Analyses
Separate data sets containing peak heights for all detectable metabolites from uterine lavage on d 0, uterine lavage on d 5, and serum on d 0 were obtained via GC-TOF-MS.Each metabolite was identified based on its mass and charge relationship.Metabolites were annotated using the PubChem (https: / / pubchem .ncbi.nlm.nih.gov/), Kyoto Encyclopedia of Genes and Genomes (https: / / www .genome.jp/kegg), and Human Metabolome (https: / / hmdb .ca/ ) databases.Differences in the uterine metabolome associated with group and time were analyzed using multivariate canonical analysis of populations (CAP) based on the Mahalanobis distance as described previously (Krzanowski, 1989;Amaro et al., 2004Amaro et al., , 2008)), and multivariate ANOVA (MANOVA) using the Mult-BiplotR and MixOmics packages in R Studio (R Core Team, 2021).Calculation of the Mahalanobis distance maximizes the differences between groups (metritis vs. no metritis) by normalizing it according to the variability within group.Statistical models used for CAP included the fixed effects of group (MET and NO-MET) and day (d 0 and 5), whereas statistical models used for MANOVA included 4 groups generated by the combination of the effects of group and day (MET d 0, MET d 5, NOMET d 0, and NOMET d 5).Data were standardized using the Z-score normalization method before CAP and MANOVA.Metabolites with P ≤ 0.05 and quality of representation index ≥ 0.99 (Varas et al., 2005;Demey et al., 2008;Sierra et al., 2017) were deemed different between cows with and without metritis.
Differences in metabolomes associated with metritis within each time point (i.e., MET vs. NOMET on d 0 for uterine lavage and serum) were further explored by univariate analyses including partial least squaresdiscriminant analysis (PLS-DA) coupled with permutational MANOVA using 2,000 permutations via Metaboanalyst 5.0 (Pang et al., 2021).A t-test was also performed when permutational MANOVA was statistically significant (P ≤ 0.05).Variable importance projection scores were calculated from PLS-DA.Residues distribution was evaluated visually.Data were logtransformed and auto-scaled for all univariate analyses.All P-values obtained from t-tests were adjusted for false discovery rate (FDR; Benjamini and Hochberg, 1995) to account for multiple comparisons.Differences with FDR-adjusted P ≤ 0.05 were considered statistically significant.
Metabolites with FDR-adjusted P ≤ 0.05 on t-test were used for enriched pathway analyses based on the Kyoto Encyclopedia of Genes and Genomes database for Bos taurus using Metaboanalyst 5.0 (Pang et al., 2021), compound network analyses using Metscape 2 (Karnovsky et al., 2012) within the CytoScape 3.8 platform, and identification of affected cellular processes using the Ingenuity Pathway Analysis Metabolomics platform (Krämer et al., 2014).Metabolic pathways with FDR-adjusted P ≤ 0.10 were considered to be affected (Wieder et al., 2021).Compound network analyses and evaluation of affected cellular processes were only performed for comparisons in which more than one metabolic pathway was deemed to be affected in enriched pathway analyses (i.e., MET vs. NOMET on d 0 only).Cellular functions with P ≤ 0.05 and calculated Z-score were considered to be affected, with Z-score >0 indicating functions that were upregulated in cows with metritis and Z-score <0 indicating functions that were downregulated in cows with metritis.

Differences in Uterine Metabolome Associated with Metritis
A total of 884 unique metabolites were identified by GC-TOF-MS in uterine lavage samples collected on d 0 or 5 (Supplemental Table S1; https: / / doi .org/ 10 .6084/m9 .figshare.21350556;Figueiredo, 2022b).From MANOVA, 661 metabolites had P ≤ 0.05, including 142 that were annotated based on publicly available metabolome databases.A total of 35 metabolites with P ≤ 0.05 and quality of representation index ≥0.99 were associated with metritis and time, all of which were annotated based on publicly available metabolome databases (Supplemental Table S2; https: / / doi .org/ 10 .6084/m9 .figshare.21350538;Figueiredo, 2022c).Results from CAP indicate that the uterine metabolome of cows with metritis presented large differences compared with the metabolome of cows without metritis on d 0 (Figure 1).
A total of 81 annotated metabolites were deemed to be associated with metritis on d 0 based on variable importance score >1 (Supplemental Table S4; https: / / doi .org/ 10 .6084/m9 .figshare.21350547;Figueiredo, 2022e).The top 20 metabolites detected by PLS-DA between MET d 0 and NOMET d 0 are summarized in Figure 2B.Enriched pathway analyses indicated that differences in metabolome between MET d 0 and NOMET d 0 were linked primarily to lipid, amino acid, and carbohydrate metabolism (Table 2; Supplemental Table S5; https: / / doi .org/ 10 .6084/m9 .figshare.21350541;Figueiredo, 2022f).Because the metabolites selected for enriched pathway analyses (i.e., annotated metabolites with adjusted P ≤ 0.05 on t-test) were involved in various pathways, results from compound network analyses revealed connections between different metabolic pathways associated with metritis on the day of diagnosis (Figure 3).For instance, butanoate metabolism was associated with metritis on d 0 because of the increased abundance of 4-aminobutanoate, 2-oxoglutarate, and succinate, and decrease abundance  S6; https: / / doi .org/ 10 .6084/m9 .figshare.21350553;Figueiredo, 2022g).At d 5, the uterine metabolome of cows with and without metritis converged, as indicated by the overlap of the 95% CI for both groups (Figure 1).Permutational MANOVA was not statistically significant (P = 0.82) for the comparison between MET and NOMET at d 5; therefore, t-test was not performed.

DISCUSSION
Cows with metritis have marked differences in uterine metabolome at the time of diagnosis compared with cows without metritis matched by parity and days postpartum at the time of sampling, including biomarkers associated with the metabolism of amino acids, lipids, and carbohydrates.Furthermore, processes such as cellular development and survival, signaling, and transport were affected in the uteri of cows with metritis at the time of diagnosis.These differences in the uterine metabolome associated with metritis are potentially a result of bacterial infection and the immune response to infection, which is not surprising considering that metritis is a polymicrobial infection of the uterus associated with intense local immune response and extensive tissue damage (Bicalho et al., 2017;Sicsic et al., 2018).Although differences in metabolomes consistent with bacterial metabolism and immune activation were anticipated, results from the present study highlight specific pathways at play in cows with metritis among several other pathways also linked to bacterial metabolism and immune activation.Identification of pivotal pathways and metabolites involved in the interaction between host and important pathogens provide targets for future intervention.
At the time of diagnosis, the uterine metabolomes of cows with metritis had altered concentrations of metabolites associated with the metabolism of amino acids such as the glycine, serine, and threonine pathway, and the alanine, aspartate, and glutamate pathway.The metabolism of such key amino acids is related to various functions including the initiation and contribution to pathways related to the metabolism of butanoate (e.g., butyrate) and the TCA cycle.Glycine was present in a smaller concentration in the uterine metabolome of cows with metritis relative to the uterine metabolome of cows without metritis at the time of diagnosis.Glycine is a primary nutrient for bacteria (Wright and Hungate, 1967).Therefore, the greater bacterial load present in the uterine microbiome of cows with metritis (Cunha et al., 2018;Galvão et al., 2020) could explain the reduced concentration of glycine, as this amino acid is metabolized for energy production.The metabolization of glycine is essential for the production of serine and threonine, which are important amino acids that can be used in the metabolism of pyruvate, cysteine, purines, and pyrimidines (Lowry et al., 1987;Narkewicz et al., 1996).The decrease in glycine concentration, coupled with the increase of serine and threonine concentration in the uterine metabolome of cows with metritis, as observed herein, indicates bacterial growth and proliferation, given that such compounds are essential for nucleotide production and are precursors for other nutrients.Decreased concentrations of citrate and pyruvate, and increased concentrations of 4-aminobutanoate, 2-oxoglutarate, and succinate, were also observed in the uteri of cows with metritis and were associated with the metabolism of alanine, aspartate, and glutamate.Such differences in important metabolites associated with butanoate metabolism and the TCA cycle indicate the important intermediate role of amino acid metabolism on lipid and carbohydrate metabolism.
At the time of diagnosis, cows with metritis had different concentrations of metabolites associated with the metabolism of lipids, such as butanoate metabolism.Various precursors within butanoate metabolism, such as 4-aminobutanoate (e.g., gamma-aminobutyric acid, GABA) and 3-hydroxybutanoate (e.g., β-hydroxybutyrate) are produced by several fermentation processes performed by obligate anaerobic bacteria.Therefore, the increase in butanoate derivates could be a direct consequence of the increased bacterial load in cows with metritis compared with nonaffected counterparts (Cunha et al., 2018;Galvão et al., 2020).The metabolism of butanoate is mainly associated with the degradation of fatty acids for the production of components for ATP production through the TCA cycle, and it is also an important precursor amino acid metabolism.Gamma-aminobutyric acid was observed in increased concentration in the uterine metabolome of cows with metritis at the time of diagnosis.Although the role of GABA in the uterus is not clear, it was reported that approximately 90% of Bacteroides genomes harbor the glutamic acid decarboxylase system for GABA production in the human gut (Strandwitz et al., 2019).Moreover, Bacteroides use GABA as a source of nutrients in culture-based experiments (Strandwitz et al., 2019) and to maintain bacterial cell integrity and viability   Cows diagnosed with metritis from 4 to 12 d postpartum (MET; n = 24; 17 cows in herd 1, and 7 cows in herd 2) were compared with cows not diagnosed with metritis paired with MET cows based on parity and calving date (NOMET; n = 24; 17 cows in herd 1, and 7 cows in herd 2).
2 Number of metabolites associated with metritis in the present data set [metabolites with P ≤ 0.05 after adjustment for false discovery rate (FDR) on t-test] out of the total number of metabolites described for each pathway according to the Kyoto Encyclopedia of Genes and Genomes database (https: / / www .genome.jp/kegg/ ).
3 P-values adjusted for multiple comparisons using FDR on enriched pathway analysis.
under acid stress situations (Otaru et al., 2021).The increase in GABA concentration in the uterus indicates greater bacterial abundance associated with metritis, and suggests that the uterine environment at the time of metritis diagnosis can be hostile to certain bacteria, as GABA production is thought to be increased under stress scenarios by metabolizing glutamate and putrescine (Sequerra et al., 2007).Furthermore, reduced concentrations of butyrate were observed in the uteri of cows with metritis at the time of diagnosis.Jeon et al. (2021) reported that butyrate is a preferred carbon source used by Fusobacterium necrophorum, a highly prevalent bacterium in the uterine microbiome of cows with metritis (Jeon et al., 2015;Galvão et al., 2019).Therefore, the reduction of butyrate observed in the uterus of cows with metritis at the time of diagnosis, coupled with increased concentrations of GABA and its precursors, indicates increased bacterial load and activity in the uteri of cows with metritis.
Pathways related to carbohydrate metabolism (i.e., glycolysis/gluconeogenesis and TCA cycle pathways) were enriched, as indicated primarily because glucose and pyruvate were observed in smaller concentrations in the uterus of cows with metritis at the time of diagnosis.Conversely, the concentration of succinate was increased in the uterus of cows with metritis compared with nonaffected counterparts.Glucose is a key carbon source for eukaryote and prokaryote cells and plays an important role in the central metabolic pathways of immune cells and bacteria (Newsholme et al., 1986;Richardson et al., 2015).Therefore, a decrease in glucose concentration is likely a result of immune cell activation (O'Rourke and Rider, 1989;Kvidera et al., 2017) and glucose utilization for bacterial growth (Frantz and McCallum, 1979;Kotarski and Salyers, 1981).Succinate is generated in mitochondria via the TCA cycle, and an increase in succinate was expected following glucose metabolism by host cells or bacteria.As an ex- Structures in blue correspond to metabolites identified using gas chromatography with time-of-flight mass spectrometry; orange structures correspond to remaining metabolites that compose the metabolic pathway not recognized in the database.Large hexagons represent metabolites with greater concentration in the uterine metabolome of cows with metritis compared with cows without metritis, and small hexagons represent metabolites with lower concentration in the uterine metabolome of cows with metritis compared with cows without metritis.Metabolic pathways observed in the uterine metabolome of cows with metritis are connected by key metabolites that are components of multiple pathways.
ample, metabolization of glucose by Bacteroides results in increased concentration of succinate (Kotarski and Salyers, 1981).In contrast, pyruvate, which is the end product of glycolysis, is reduced in cows with metritis.This could be a result of utilization of pyruvate by major uterine pathogens.A recent study has reported that pyruvate is a preferred carbon source of Fusobacterium necrophorum (Jeon et al., 2021).Altogether, the pattern observed herein could be a result of immune cell activation and bacterial metabolite utilization for proliferation in the uterus of cows with metritis, leading to uterine metabolome differences such as decreased concentrations of glucose and pyruvate, and increased concentrations of succinate.
Altogether, differences in the uterine metabolome on the day of diagnosis illustrate possible interactions between the host immune response and growing pathogens.The biological processes described here could be translated as mechanisms to subvert immune cell actions and thrive under restrictive environments by bacteria, while immune cells trigger mechanisms of recruitment and activation.Although affected metabolic pathways and their associated molecular functions were described separately, multiple metabolites previously mentioned are shared among various processes, indicating a link between pathways in a global manner, as shown in the compound network analysis.Such connection among metabolic pathways has been previously reported (Rytter et al., 2021), and it has been proposed that the bacterial central metabolism is composed of 3 main metabolic processes (i.e., glycolysis and gluconeogenesis, the TCA cycle, and the pentose phosphate pathway) that share a large proportion of molecules among them (Richardson et al., 2015).It is possible that biosynthetic reactions result in precursors that prompt bacterial proliferation and activation of virulence factors.
The uterine metabolome of cows with metritis converged with that observed in cows without metritis at d 5. Consistent with our findings, the uterine microbiome of cows with metritis has been shown to converge with that of cows without metritis between 6 and 12 d after diagnosis, and it has been reported that antimicrobial treatment hastens that process (Galvão et al., 2020).A limitation of the present study is that cows with metritis that were not treated with antimicrobials were not included, and, therefore, the experimental approach does not allow for distinction of the effects of time and spontaneous disease progression from those induced by antimicrobial treatment.Further research is warranted for evaluation of the uterine metabolome progression independent of antimicrobial treatment and for char-acterization of differences in the uterine metabolome caused by systemic antibiotics.
Contrary to the initial hypothesis, no major differences in the serum metabolome were observed at the time of disease diagnosis between cows with metritis and counterparts without metritis.To our knowledge, data regarding the untargeted comparison of serum metabolome between cows with and without metritis are not available in the literature.Using a targeted metabolome approach, Dervishi et al. (2021) observed a greater concentration of haptoglobin in the serum of cows with metritis, whereas cows without metritis had greater concentrations of calcium, urea, and albumin.Several other studies have evaluated individual metabolites and minerals in cows with and without metritis.For instance, Martinez et al. (2012) reported reduced serum calcium concentration in cows with metritis compared with cows without metritis.Hailemariam et al. (2018) reported differences in serum amino acid profiles, including glycine, serine, threonine, and ornithine associated with metritis at the week of diagnosis.Galvão et al. (2010) also reported greater concentrations of nonesterified fatty acids and BHB in serum of cows with uterine diseases compared with nonaffected cows.The objective of the present study was to characterize the uterine and serum metabolome in a broader manner, rather than via limited searches within specific clusters of metabolites or differences in concentrations of individual metabolites.To fulfill this purpose, an untargeted metabolomic technique was employed.Our t-test analyses identified 19 metabolites with different abundances in serum between cows with and without metritis.Despite differences in the concentrations of specific metabolites in serum at the time of diagnosis, however, current results indicate that metritis is not associated with massive differences in the metabolome profile.The severity of uterine diseases in dairy cows varies widely, ranging from no signs of systemic illness to presence of fever, decreased dry matter intake, delayed cure followed by persistence of uterine inflammation, and death associated with systemic illness and concurrent diseases (Benzaquen et al., 2007;Huzzey et al., 2007;Figueiredo et al., 2021).In the present study, only 2 cows with metritis had concurrent hyperketonemia, and none died within 200 DIM, suggesting that these cases of metritis were not severe.It is possible that comparison of cows with metritis including only those with signs of systemic illness and healthy counterparts would reveal major differences in serum metabolomes.However, such comparison requires a slightly different approach and goes beyond the scope of the study presented here.

CONCLUSIONS
Metritis was associated with differences in the uterine metabolome at the time of disease diagnosis that were no longer evident 5 d after diagnosis and onset of treatment.Shifts in the uterine metabolome in cows with metritis were associated with disturbances in amino acid, lipid, and carbohydrate metabolism.These differences in metabolome are speculated to be associated with bacterial proliferation and activation of the immune response.Results from the present study indicate that amino acid degradation, particularly of ornithine, may be an important source of energy for maintenance of pathogens in the uterus.By-products from ornithine degradation, such as putrescine and spermidine, were observed in greater concentrations in uterine contents from cows with metritis, which has been associated with activation of virulence factors.Current results also suggest that GABA may be an important source of energy for proliferation of Bacteroides in cows with metritis.Moreover, it is possible that increased availability of succinate in the uteri of cows with metritis plays a role as a source of energy for bacterial proliferation and as a mechanism for pathogens to limit the host immune response, considering its anti-inflammatory properties.No major differences in the serum metabolome were observed between cows with and without metritis at the time of diagnosis.Finally, results from this study indicate that metabolites with increased availability in the uteri of cows without metritis, such as lactic acid, pyruvic acid, and ribonic acid, should be investigated as potential strategies to prevent or treat cows with metritis.

Figueiredo
Figueiredo et al.: METABOLOME OF COWS WITH METRITIS

Figueiredo
Figueiredo et al.: METABOLOME OF COWS WITH METRITIS

Figueiredo
Figueiredo et al.: METABOLOME OF COWS WITH METRITIS

Figure 2 .
Figure 2. Comparison of uterine metabolome on d 0 between cows diagnosed with metritis from 4 to 12 d postpartum (MET; n = 24; 17 cows in herd 1, and 7 cows in herd 2) and cows not diagnosed with metritis paired with MET cows based on parity and calving date (NOMET; n = 24; 17 cows in herd 1, and 7 cows in herd 2).Day of metritis diagnosis and pairing was considered study d 0. Panel A: results from partial least squares-discriminant analysis (PLS-DA); P < 0.001 for permutational multivariate ANOVA with 2,000 permutations.PC1 and PC2 = principal component 1 and 2, respectively.Panel B: top 20 uterine metabolites detected by PLS-DA ranked by variable importance projection (VIP) scores; red bars represent metabolites with increased abundance in MET; green bars represent metabolites with increased abundance in NOMET.Uterine metabolome of cows with and without metritis differs at the time of diagnosis.

Figueiredo
Figueiredo et al.: METABOLOME OF COWS WITH METRITIS

Table 1 .
Figueiredo et al.: METABOLOME OF COWS WITH METRITIS Uterine metabolites associated with metritis on the day of diagnosis (d 0) 2P-values from t-test were adjusted for false discovery rate.Adj.=adjusted.Figueiredo et al.: METABOLOME OF COWS WITH METRITIS