|Home | About | Journals | Submit | Contact Us | Français|
Escherichia coli infection of the endometrium causes uterine disease after parturition and is associated with prolonged luteal phases of the ovarian cycle in cattle. Termination of the luteal phase is initiated by prostaglandin F2α (PGF) from oxytocin-stimulated endometrial epithelial cells. Compared with normal animals, the peripheral plasma of animals with E. coli infection of the endometrium had higher concentrations of lipopolysaccharide (LPS) and prostaglandin E2 (PGE), but not PGF. Endometrial explants accumulated predominantly PGE in the culture medium in response to LPS and this effect was not reversed by oxytocin. Endometrial cells expressed the TLR4/CD14/MD-2 receptor complex necessary to detect LPS. Epithelial and stromal cells treated with LPS had higher steady-state media concentrations of PGE rather than PGF. Arachadonic acid is liberated from cell membranes by phospholipase 2 (PLA2) enzymes and converted to prostaglandins by synthase enzymes. Treatment of epithelial and stromal cells with LPS did not change the levels of PGE or PGF synthase enzymes. However, LPS stimulated increased levels of PLA2 group VI but not PLA2 group IV C immunoreactive protein in epithelial cells. Endometrial cells expressed the EP2 and EP4 receptors necessary to respond to PGE, which regulates inflammation as well as being luteotropic. In conclusion, LPS detection by endometrial cells stimulated the accumulation of PGE rather than PGF, providing a mechanism to explain prolonged luteal phases in animals with uterine disease, and this PGE may also be important for regulating inflammatory responses in the endometrium.
The mammalian uterus is usually a sterile environment but it is readily contaminated with bacteria during coitus or parturition. The consequences of uterine infections range from pelvic inflammatory disease (PID), to chronic endometritis and infertility (1-3). In cattle, bacterial contamination of the uterine lumen is ubiquitous after parturition, and up to 40% of animals develop PID and 20% have endometritis (4, 5). Infection of the endometrium with Escherichia coli precedes infection by other pathogenic bacteria and viruses, and is associated with the severity of PID and the impact on fertility (5-7).
Uterine infections cause infertility not only by disrupting endometrial health but also affecting ovarian cycles. Bacterial toxins act at the hypothalamus or pituitary to suppress gonadotrophin release (8); and, in the ovary perturb follicle growth and function (4, 9, 10). A common observation is that animals with uterine disease have prolonged luteal phases, leading to delayed conception (5, 11). The length of the luteal phase in ruminants is dependent on oxytocin (OT) binding endometrial epithelial cell OT receptors to initiate prostaglandin F2α synthesis and luteolysis (12, 13). Prostaglandins are produced by prostaglandin endoperoxide H synthases (PTGS1 and PTGS2) and cyclooxygenation of arachadonic acid (AA) into prostaglandin H2 (12, 13). The AA is liberated from cell membranes by phospholipase A2 (PLA2) enzymes, which include PLA2 group IV (PLA2G4A and PLA2G4C) and group VI (PLA2G6) in the bovine endometrium (14). There is a sub-cellular localisation of OT-stimulated pathways in epithelial cells, with PLA2G4C associated with PGE synthesis and PLA2G6 linked to PGF, although PLA2G4A does not appear to be important (14). Once AA is converted to prostaglandin H2, prostaglandin F synthase (PGFS) and E synthase (PGES), produce prostaglandin F2α (PGF) and prostaglandin E2 (PGE), respectively (15). Whilst PGF is luteolytic in ruminants, PGE is luteotrophic (16). Under physiological conditions, the concentrations of PGF increase in the medium of epithelial cells and only PGE is found in appreciable amounts with stromal cells (17, 18).
Innate immune responses to pathogens are the driving force for the clearance of bacteria, regulation of inflammation, and maintenance of endometrial health (17, 19). The Toll-like receptors are central to innate immunity, recognising and responding to pathogens. The lipopolysaccharide (LPS) of E. coli is detected by the TLR4, MD-2, CD14 receptor complex on endometrial cells, which induces transcription of genes associated with inflammation such as nitric oxide synthase (NOS2) and PTGS2 (17, 19). In vitro, LPS stimulated increased concentrations of PGE in the medium of cultured stromal cells, but we noted increased concentrations of PGE with epithelial cells (17). Increased concentrations of PGE in the uterine lumen have also been reported in cattle with endometrial infections (20, 21). PGE plays an important role in the immune response as well as in endocrine function, acting through EP2 and EP4 receptors to control inflammation (22-24).
The aim of the present study was to test the hypothesis that E. coli LPS induces preferential accumulation of PGE rather than PGF in the endometrium. The effect of E. coli LPS on prostaglandin concentrations was explored in postpartum animals, and using endometrial explants and cells. The mechanisms regulating the balance between PGF and PGE were then examined in detail using purified populations of endometrial cells
To test the role of E. coli we exploited the knowledge that E. coli infection is most common in the first two weeks after parturition (5, 7). Forty five postpartum Holstein-Friesian cows (median parity 3; range: 1 to 9) were examined using previously described procedures (4, 5). Animals with non-uterine bacterial infections were excluded from the study. All procedures were carried out in compliance with the Animals (Scientific Procedures) Act 1986, and experimental protocols were approved by the Local Ethical Review Committee. To evaluate bacterial infection a swab was collected from the uterine body of each animal 7 and 14 days after parturition using a validated method (4, 5). Bacteria were cultured aerobically and anaerobically, and identified using standard tests. The growth density of each bacterial species was scored semi-quantitatively by estimating the number of colony forming units (CFU) on the culture plate from each swab as follows: 0, none; 1, < 10 CFU; 2, 10 to 100 CFU; 3, 101 to 500 CFU; and 4, > 500 CFU (4).
Blood samples were collected 7 and 14 days post partum, into evacuated plain and heparinised glass tubes (BD Vacutainer Systems, Plymouth, UK) and transported on ice to the laboratory. Plasma was separated from blood in the heparinised tubes by centrifugation at 3500 × g for 10 min and the supernatant stored at −20°C. Serum was obtained by allowing the blood in the plain tubes to coagulate at room temperature for at least 1 h before centrifugation at 3500 × g for 10 min, and the supernatant stored at −20°C.
Concentrations of LPS were measured in serum using the Kinetic-QCL Limulus Amoebocyte Lysate (LAL) kit (Lonza, Wokingham, UK) following the manufacturers guidelines with modifications, as described (5, 20). Briefly, samples were thawed, diluted to 1:10 in endotoxin-free water and heated in a water bath at 75°C for 30 min. Samples were then mixed with the LAL substrate reagent in duplicate in 96-well endotoxin-free microplates (Becton Dickinson, Franklin Lakes, NJ) and LPS measured using a kinetic microplate reader with appropriate software (WinKQCL, Lonza). Internal recovery as determined using positively spiked samples was > 80%.
Concentrations of the acute phase proteins haptoglobin and α1-acid glycoprotein (AGP) were measured in plasma using previously described methods adapted for 96-well plates (Invitrogen, Paiseley, UK) (4, 5, 25, 26). The limits of detection for haptoaglobin and AGP were 36 and 200 μg/tube, respectively, and the intra and inter-assay CVs were 10.8% and 11.8%, and 9.6% and 14.8%, respectively. Concentrations of prostaglandins were measured by radioimmunoassay, as described below.
Bovine uteri were collected from post-pubertal non-pregnant animals with no evidence of genital disease or microbial infection at a local abattoir and kept on ice until further processing in the laboratory. The stage of the reproductive cycle was determined by observation of ovarian morphology and genital tracts with an ovarian Stage I corpus luteum were selected for endometrial culture (27). The endometrium from the horn ipsilateral to the corpus luteum was cut into strips and placed into serum-free RPMI-1640 medium (Sigma, Poole, UK) supplemented with 50 IU/ml of penicillin, 50 μg/ml of streptomycin, 2.5 μg/ml of Amphotericin B, and 240 U/ml of Nystatin (Sigma).
For tissue explants, the strips were chopped into 1-mm3 pieces and placed into Hanks balanced salt solution (HBSS; Sigma) as previously described (28). For each treatment (see below), 50 mg of tissue was weighed in triplicate and transferred onto sterile tissue-lined metal grids in 6-well plates (Nunc, Nottingham, UK) with 4.25 ml/well serum-free RPMI 1640. Tissue explants were incubated at 37°C, 5% CO2 in air, in a humidified incubator overnight, and then supernatants removed and replaced with fresh media.
For cell isolation, the endometrial strips were cut into smaller pieces and placed into HBSS as previously described (29). Briefly, tissue was digested in 25 ml sterile digestive solution, made by dissolving 50 mg trypsin III (Roche, Welwyn, UK), 50 mg collagenase II (Sigma), 100 mg BSA (Sigma) and 10 mg DNase I (Sigma) in 100ml HBSS. Following 1.5 h incubation in a shaking water bath at 37°C, the cell suspension was filtered through a 40 μm mesh (Fisher Scientific, Loughborough, UK) to remove undigested material and the filtrate was resuspended in washing medium, comprising of HBSS with 10% fetal bovine serum (FBS, Sigma). The suspension was centrifuged at 100 × g for 10 min and following two further washes in washing medium the cells were resuspended in RPMI-1640 medium containing 10% FBS, 50 IU/ml of penicillin, 50 μg/ml of streptomycin and 2.5 μg/ml of Amphotericin B. The cells were plated at a density of 1 × 105 cells/ml in 24-well plates (Helena Bioscience, Gateshead, UK). To obtain separate stromal and epithelial cell populations, the cell suspension was removed 18 h after plating, which allowed selective attachment of stromal cells (29). The removed cell suspension was then re-plated and incubated allowing epithelial cells to adhere (30). Stromal and epithelial cell populations were distinguished by cell morphology as previously described (29). The culture media was changed every 48 h until the cells reached confluence. Cell cultures were maintained at 37°C, 5% CO2 in air, in a humidified incubator.
Endometrial explant, and confluent stromal and epithelial cells were challenged with arachadonic acid (AA, 100 □M; Sigma), oxytocin (OT, 100 nM; Baychem, Redwood City, CA) or LPS (1 or 3 μg/ml LPS from E. coli O55:B5; Sigma), individually or in combinations and for the period of time indicated in Results. The AA, OT and LPS concentrations were validated previously (17). The LPS concentrations also reflected those found in the uterine lumen of infected animals (5, 7). The 6 and 24 h time points were selected on the basis of preliminary experiments, and previous experience that the earliest detection of increased PGE and maximal concentrations were detected at 6 and 24 h, respectively (17). Treatments were replicated at least twice and experiments were performed on at least three separate occasions. Cell numbers were assessed colorimetrically by the mitochondria-dependent reduction of MTT (Sigma) to formazan, as previously described (31). Culture supernatants were harvested and frozen until PG determination.
Total RNA was isolated from cell cultures using the RNeasy Mini Kit (Qiagen, Crawley UK) and, following DNase treatment (Promega, Southampton UK) to ensure no DNA contamination, quantitated using a NanoDrop® ND-1000 spectrophotometer (NanoDrop Technologies Inc, DE). Following quantification, RNA was reverse transcribed into First Strand cDNA using SuperScript II RNase H− Reverse Transcriptase (Invitrogen) according to the manufacturers' protocols. Using 50 ng of cDNA, amplification used the following conditions: denaturation for 5 min at 94°C, followed by 39 cycles of 94°C for 30 sec, 54-56°C (depending on primer Tm) for 30 sec and 72°C for 30 sec, followed by a final extension of 5 min at 72°C. Water and RNA were used as negative controls, and DNA from whole endometrium as a positive control. Intron-spanning gene-specific primers (Table 1) were designed using sequences published in the National Center for Biotechnology Information database (NCBI, Bethesda, MD) using Primer3 software (32), and purchased from MWG (Ebersberg, Germany). The amplicons were separated on 2% agarose gels to confirm the amplification of distinct bands and to assess the expression of each gene. The amplicons were sequenced using an ABI 3100 genetic analyzer and Bigdye Terminator 3.1 (ABI; Foster City, CA), and were 95-99% homologous to NCBI database sequences.
Supernatants and sera were analysed for steady-state PGE, PGF and the PGF metabolite (PGFM, 15-keto-13,14-dihydro-prostaglandin F2α) concentrations by radio-immunoassay (RIA) as described (33). Briefly, samples were diluted in 0.05 M Tris buffer containing 0.1% gelatin and 0.01% sodium azide. Standards and tritiated tracers were purchased from Sigma and Amersham International PLC (Amersham, UK), respectively. The antisera were generous gifts from Prof H. Kindahl (PGFM; Swedish University of Agricultural Sciences, Uppsala, Sweden) and Professor N.L. Poyser (PGE and PGF; University of Edinburgh, UK) and their cross-reactivity have been reported (34, 35). The limits of detection for PGE, PGF and PGFM were 2 pg/tube, 1 pg/tube and 1 pg/tube, respectively. The intra- and inter-assay coefficients of variation were 4.4% and 7.8% for PGE, 5.1% and 9.7% for PGF, and 7.6% and 14.3% for PGFM, respectively.
Proteins from lysates of cultured cells were normalised to 1μg/μl using a NanoDrop ND-1000 spectrophotometer and separated (10 μg/lane) using 10% (vol/vol) SDS-PAGE. Following electrophoresis, proteins were transferred to Immobilon-polyvinylidine difluoride (PVDF) membrane (Millipore Corp, Bedford, USA) using the omniPAGE electroblotting system (Cleaver Scientific, Rugby, UK). Prestained molecular weight markers were run in parallel lanes (Bio-Rad Laboratories, Inc.). After transfer of proteins to PVDF, non-specific sites were blocked using a solution of 5% (wt/vol) skim milk in Tris buffered saline (TBS) overnight at 4°C with gentle agitation. Membranes were probed with antibodies for PGFS (kindly gifted from Dr. Michel Forter, Université Laval, Quebec, Canada); PTGS2 (Santa Cruz Biotechnology, Santa Cruz, CA); membrane-bound and cytosolic PGES (mPGES-1, mPGES-2, cPGES; Caymen Chemical, Ann Arbor, MI); PLA2G6 (Caymen); or PLA2G4C (kindly gifted from Prof. C.C. Leslie, National Jewish Medical and Research Centre, Denver, CO, USA). Antibodies were selected based on recognition of a single immunoreactive protein of appropriate molecular weight, testing with available blocking peptides, and previous detection of the proteins in the bovine endometrium (14, 36, 37). Each antibody was optimised using a positive control of whole endometrial tissue lysate, and an isotype matched antibody negative control. Primary antibodies were used at 1:1000 dilutions in TBS for 2 h with gentle agitation. Following incubation, membranes were washed 5 × 10 min in TBST (TBS, 0.1% Tween 20, pH 7.6). Membranes were then incubated in secondary horseradish peroxidase-conjugated antibody (DakoCytomation, Glostrup, Denmark) for 1.5 h and washed for 3 × 15 min in TBST. Steady-state levels of immunoreactive proteins were visualised using enhanced chemiluminescence (ECL Western Blotting Substrate; Pierce, Rockford, IL). Protein loading was evaluated and normalised by examining β-actin protein levels using a β-actin antibody (Santa Cruz) and horseradish peroxidase goat anti-mouse IgG (DakoCytomation). Densitometric quantification of immunoreactive bands was carried out using Quantity One® analysis software (BioRad, version 4.6.1).
Data were analysed using ANOVA, except for non-parametric score data which were analysed by the Mann-Whitney test. Results are quoted as mean ± SEM, and significance attributed when P < 0.05.
Animals with disparate uterine E. coli growth densities a week after parturition were identified retrospectively from the database (Fig. 1a; n = 8 “normal” animals with no E. coli; n = 7 “infected” animals with > 500 E. coli CFU/swab). As expected with dynamic infections, by the second week E. coli could be cultured from the uterus of some normal animals and the infected animals had fewer bacteria. In the first week, the infected animals had higher peripheral circulation concentrations of LPS (Fig. 1b), haptoglobin (Fig. 1c) and α1-acid glycoprotein (Fig. 1d) than normal animals. The PGFM concentrations did not differ between the two groups (Fig. 1e) but there were higher concentrations of PGE in the infected animals (Fig. 1f).
To investigate whether the higher concentration of PGE in animals with E. coli infection was a response to LPS and whether this could be reversed by OT, endometrial explants were pre-treated with media or LPS (1 μg/ml) for 24 h, the media removed and the explants treated with media or OT (100 nM) for a further 24 h. Explants media accumulated predominantly PGF in response to OT and predominantly PGE in response to LPS (P < 0.05, Fig. 2a). The ratio of PGE:PGF concentration was 0.45 in response to OT and 2.75 following LPS treatment. However, explants media accumulated predominantly PGE in response to OT if they were pre-treated with LPS (P < 0.05, Fig. 2a) and the ratio of PGE:PGF was 1.95. The concentrations of LPS employed reflect those measured in the uterine lumen of diseased animals (5, 7), but pre-treatment of explants with 0.3 μg/ml LPS followed by treatment with OT had similar effects to those seen with 1 μg/ml (65.9 ± 17.4 ng/ml PGE and 43.8 ± 10.8 ng/ml PGF). At lower concentrations of LPS (0.03 μg/ml pre-treatment), the difference between PGE and PGF concentrations was less (39.9 ± 10.1 ng/ml PGE and 37.9 ± 9.9 ng/ml PGF).
To investigate which endometrial cells were associated with the LPS-induced changes in prostaglandin concentrations, endometrial stromal and epithelial cells were isolated and cell purity confirmed by morphology and the absence of expression of the CD45 panleukocyte marker (data not shown), as previously described (17). The lack of cross-contamination between stromal and epithelial cells, and physiological function was confirmed by stimulating the cells with OT and AA. As expected, stromal cell media accumulated PGE but little PGF (Fig. 2b), and epithelial cells acumulated PGF and little PGE (Fig. 2c). The concentration of PGF in epithelial cell medium in the presence of OT and AA exceeded the concentration of PGE associated with stromal cells (P < 0.001). For LPS treatment, preliminary experiments indicated that 24 h was the optimum culture period for endometrial cells, although there was accumulation of prostaglandin after 6 h of culture with LPS (data not shown). Cell numbers did not differ significantly amongst treatments, and LPS did not affect cell survival compared with control for epithelial cells (1.19 ± 0.15 vs 1.21 ± 0.06 × 105 cells/ml) or stromal cells (1.20 ± 0.13 vs 1.17 ± 0.11 × 105 cells/ml). Furthermore, increased PGF and PGE concentrations were detected with stromal and epithelial cells treated with LPS in the absence of AA (Fig. 2b-c). The stromal cells had increased concentrations of PGE and little PGF when treated with LPS (P < 0.05, Fig. 2b). The epithelial cell medium also had higher concentrations of PGE than PGF in response to LPS (P < 0.05, Fig. 2c). Indeed, the ratio of PGE:PGF concentration in epithelial cells increased from 0.01 for OT, to 4.95 with 1 μg/ml LPS.
Epithelial and stromal cells expressed mRNA for the synthase enzymes PGES and PGFS (Fig. 3). Immunoblotting was used to explore if the preferential accumulation of PGE over PGF was due to differential regulation of the levels of prostaglandin synthase proteins. Immunoreactive PGFS protein was detected in untreated and LPS-challenged stromal and epithelial cells, after 6 and 24 h (Fig. 4a-d). However, LPS did not significantly change the level of PGFS protein in either cell type (Fig. 4e-f). PTGS2 protein was detected in stromal and epithelial cells (Fig 4c-e), and the levels of PTGS2 protein increased in stromal cells following LPS treatment. Of the three PGE synthases examined by immunoblotting, mPGES1 was not detectable in stromal or epithelial cells, with or without LPS treatment for 6 or 24 h. Immunoreactive cPGES and mPGES2 protein was detected in stromal and epithelial cells at 6 h (Fig. 4a-b) but only cPGES was detectable in epithelial cells at 24 h (data not shown). The protein levels did not differ significantly between treatments at 6 or 24 h (densitometry data not shown).
As the LPS-induced accumulation of PGE was not associated with changes in the synthase enzymes, the role of PLA2 enzymes was analysed in stromal and epithelial cells. Immunoreactive PLA2G6 protein was detected in stromal and epithelial cells at both time points (Fig. 4c-f), and LPS treatment increased epithelial cell PLA2G6 protein levels about three-fold. Both cell types had detectable PLA2G4C protein at 6 h (Fig. 4a-b) but PLA2G4C was only identified in epithelial cells at 24 h (data not shown). Following LPS treatment, there was no PLA2G4C protein identified in stromal cells at 6 h (Fig. 4a) and no significant change in PLA2G4C protein levels in epithelial cells (data not shown).
The actions of PGE are mediated through PGE receptors (38, 39). Stromal and epithelial cells expressed EP2 and EP4 mRNA (Fig. 3). Furthermore, both cell types expressed the immune mediator IL-6 and the LPS receptor complex comprising of TLR4, MD-2 and CD14.
Bacterial infections of the endometrium cause diseases in humans and animals that range from PID to chronic endometritis and infertility (1-3). In cattle, PID is particularly associated with E. coli, which paves the way for other pathogens to cause endometrial damage and disrupt ovarian cycles, including extended luteal phases (5, 6, 11). In ruminants, PGF is luteolytic, whilst PGE is luteotropic (13, 16). In the present study, animals with E. coli infection after parturition had more LPS, acute phase proteins and PGE but not PGFM in peripheral circulation. The association between LPS and PGE was supported in vitro using endometrial explants and purified populations of stromal and epithelial cells. In particular, epithelial cell culture media preferentially accumulated PGE rather than PGF. The underlying mechanism for the switch in prostaglandin concentrations was increased levels of PLA2G6 protein in epithelial cells treated with LPS, rather than changes in the level of PGES or PGFS proteins. Endometrial cells play a key role in detecting E. coli in the reproductive tract as they expressed the TLR4/CD14/MD-2 complex required to bind LPS, and express the EP2 and EP4 receptors to respond to PGE. Indeed, PGE may also have an important role regulating endometritis in cattle and other species with PID. The preferential accumulation of luteotropic PGE in response to LPS, rather than luteolytic PGF, provides a mechanism to explain the extended luteal phases and infertility associated with PID.
Increased concentrations of PGE and LPS have been noted previously in the uterine lumen of postpartum cattle with bacterial infections (5, 20, 21). The present study aimed to determine if the accumulation of PGE in the endometrium was associated specifically with E. coli and LPS. Studying a group of animals the first week after parturition provided an opportunity to identify several animals that had no bacteria or just E. coli cultured from the uterine lumen. This disparate bacterial grouping was not maintained the following week because animals eliminate bacteria and acquire new infections throughout the postpartum period (4, 5). Animals with E. coli had more LPS and acute phase proteins haptoglobin and α1-acid glycoprotein in the peripheral circulation, confirming the disparity from normal animals. The infected animals also had more PGE but not PGFM in the peripheral circulation. Thus, E. coli infection of the endometrium was associated with increased concentrations of LPS and PGE.
To test the hypothesis that E. coli LPS induces preferential accumulation of PGE rather than PGF in the endometrium, we examined endometrial explants and purified populations of cells. Tissue explant medium had higher concentrations of PGF than PGE in response to oxytocin, as expected from our previous work (17). Conversely, the explants produced more PGE than PGF in response to LPS, similar to the observations in the live animals. Furthermore, OT did not reverse the effect of LPS on prostaglandin accumulation over a range of LPS concentrations. The next step was to determine which endometrial cells were responsible for the changes in prostaglandin accumulation in response to LPS.
Under physiological conditions, epithelial and stromal cells performed as expected by predominantly accumulating PGF and PGE in the medium, respectively, when treated with OT and AA. We previously exploited this polarised accumulation of prostaglandins to study pathophysiological mechanisms and demonstrated that LPS stimulated stromal and epithelial cell accumulation of PGE and PGF, respectively, in a concentration dependent manner (17). The present study confirmed that LPS increased stromal cell PGE steady-state concentrations, but the epithelial cells also had higher concentrations of PGE than PGF, switching the PGE:PGF ratio from 0.01 following OT/AA treatment to 4.95 following 1 □g/ml LPS. These changes in PGE:PGF ratio between LPS and OT may be important mechanism for disruption of luteolysis, which is initiated by PGF from epithelial cells under the control of OT (13, 12). In the present and previous study, increased steady-state concentrations of prostaglandins required supplementation of the cell culture media with AA when the endometrial cells were treated with OT but not when treated with LPS or E. coli (17). It appears that LPS and OT exploit different prostaglandin synthesis pathways under pathological and physiological situations, respectively. The increased concentrations of PGE may have a wider relevance than just in cattle, because PGE is widely involved in the physiological function and pathology of the human endometrium (44). Bacterial infections in women are a common cause of pre-term labour as well as PID (41, 42). Sexually transmitted infections are also widespread in the human population with an estimated 350 million new cases each year across the World (43).
An obvious mechanism for the favoured accumulation of PGE over PGF would be changes in the corresponding synthase enzymes (45). Endometrial cells expressed PGES and PGFS mRNA, so the accumulation of PGE could be associated with increased levels of PGES protein and/or down-regulation of PGFS. The synthase enzymes were examined 6 and 24 h after LPS treatment to reflect the time points at which increased PGE concentrations were first detected and maximal, respectively. However, PGFS protein levels did not differ significantly between cell types, treatments, or time points. Similarly, there were no significant changes in mPGES2 or cPGES protein levels in the endometrial cells.
The synthase enzymes use PGH as their substrate and there are PTGS1 and PTGS2 cyclooxygenase enzymes convert AA to PGH. PTGS1 is unlikely to be important in the bovine endometrium, because most prostaglandin synthesis is processed through PTGS2 (46, 47). Furthermore, the level of PTGS2 is not a limiting factor in the endometrium as protein levels are already high, or increased by OT or LPS (17, 48). PTGS2 may play a role in the ratio of PGE:PGF synthesis as it preferentially binds to mPGES1 (45) and PTGS2-derived PGE plays a role in inflammation following bacterial challenge (49, 50). However, the absence of detectable levels of mPGES-1 makes it unlikely that this is the mechanism in the present study. The consistent levels of mPGES2 and cPGES protein, despite treatment with LPS, are probably because these synthases are usually constitutively expressed (45). In the absence of changes in cyclooxygenase or synthase enzymes to explain the PGF to PGE switch, we next considered the source of AA.
The synthesis of prostanoids starts with the liberation of AA from membrane stores by PLA2 enzymes, and there are at least 14 groups with considerable differences in expression between species and tissues (51). The PLA2 enzymes involved in bovine endometrial function include PLA2G4C and PLA2G6 (14). In the present study, stromal and epithelial cells had low levels of PLA2G4C protein in agreement with previous observations (14); and, protein levels were not affected by LPS. However, the level of PLA2G6 protein was increased 6 and 24 h after treatment of epithelial cells with LPS. The increased PLA2G6 protein associated with LPS and accumulation of PGE is in contrast to oxytocin-induction of PLA2G6 leading to accumulation of PGF (14). The differences between studies may be a consequence of cell culture methods, the stage of ovarian cycle when the endometrial cells were collected, or differences between OT and LPS challenge. However, even OT increased the PGE:PGF ratio in epithelial cells over-expressing PLA2G6 (14). Further, immune challenges of murine macrophages or human monocytes stimulated accumulation of PGE via the PLA2G6 enzyme (52, 53). There are also temporal effects as PGE accumulation in an acute inflammation model in rats was associated with a rapid increase in PLA2G6 protein levels between 6 and 24 h, before the response subsided (54).
Murine macrophages treated with LPS had increased PGE concentrations associated with LPS binding the TLR4/CD14/MD-2 complex and acting via a MyD88/NF-IL6 pathway (50). The TLR4/CD14/MD-2 complex was present in the bovine endometrial cells and LPS stimulated endometrial cell expression of inflammatory mediators such as tumour necrosis factor α, nitric oxide and interleukin 6, similar to Uematsu et al (17, 50). Furthermore, tumour necrosis factor α and nitric oxide preferentially stimulate PGE accumulation in the media of treated bovine endometrial cells (55, 56). So, these inflammatory mediators may augment the LPS-induced expression of prostanoid pathway enzymes.
The endometrial cells expressed the EP2 and EP4 receptors for PGE that are usually associated with the down regulation of inflammation (38, 39, 57). Thus, the PGE produced by the stromal or epithelial cells may have a paracrine role to regulate endometritis in cattle and other species, including humans (44). In addition, in most biological systems feedback loops are important and the endometrial cell expression of EP2 and EP4 receptors may be important for regulation of the immune response to E. coli, which is the case in mice treated with LPS (57).
In conclusion, animals with E. coli infection after parturition had more LPS, acute phase proteins and PGE in peripheral circulation. The association between infection and PGE was supported in vitro because LPS treatment of endometrial explants and cells preferentially increased PGE rather than PGF concentrations. Surprisingly, the epithelial cell medium accumulated more PGE than PGF, whereas with OT the reverse was observed. This switch in prostaglandin accumulation was associated with an increased level of PLA2G6 protein in epithelial cells, rather than changes in the levels of PGES or PGFS. The endometrial cells play a key role in the response to bacterial challenge in the reproductive tract as they express the TLR4/CD14/MD-2 complex required to detect LPS, and EP2 and EP4 receptors to bind PGE. The PGE may also be important for regulating endometrial inflammation associated with PID in cattle, but this mechanism needs further investigation in other species. Finally, the switch in steady-state prostaglandin concentrations from the luteolytic F series to the luteotropic E series provides a mechanism to explain the extended luteal phases associated with uterine disease and infertility in cattle.
Grant support: This work was supported by the Biotechnology and Biological Sciences Research Council (BBSRC; Grant S19795). Sheldon is a BBSRC Research Development Fellow (Grant BB/D02028X/1).
Disclosure summary: The authors have nothing to disclose.