If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Chronic postpartum uterine infection detrimentally affects subsequent fertility. Nonsteroidal anti-inflammatory drugs (NSAID) are used to alleviate pain and treat inflammatory conditions in transition dairy cows with varying success. To screen the efficacy of NSAID in the absence of animal experiments, we have established an in vitro model to study uterine inflammation. Inflammation was induced in cultured bovine endometrial epithelial cells by challenging cells with an inflammation cocktail: lipopolysaccharide and proinflammatory cytokines, interleukin-1β (IL1β) and tumor necrosis factor α (TNFα). Release of the inflammation markers, serum amyloid A (SAA) and α-1-acid glycoprotein (αAGP), was measured by ELISA. Concentration of these markers was used to indicate the effectiveness in dampening inflammation of 5 NSAID: meloxicam, flunixin meglumine, aspirin, ketoprofen, and tolfenamic acid. Three NSAID, meloxicam, flunixin meglumine, and tolfenamic acid, were successful at dampening the release of SAA and αAGP into cell-culture supernatant, and the corresponding treated cells were selected for down-stream mRNA expression analysis. Expression of 192 genes involved in regulation of inflammatory pathways were investigated using Nanostring. Of the genes investigated, 81 were above the mRNA expression-analysis threshold criteria and were included in expression analysis. All SAA genes investigated (SAA2, SAA3, M-SAA3.2) were upregulated in response to the inflammation cocktail, relative to mRNA expression in control cells; however, AGP mRNA expression was below the expression analysis threshold and was, therefore, excluded from analysis. Treatment with NSAID downregulated genes involved in regulating chemokine signaling (e.g., CXCL2, CXCR4, CXCL5, and CXCL16) and genes that regulate the eicosanoid pathway (e.g., LTA4H, PTGS2, PLA2G4A, and PTGDS). Of the 5 NSAID investigated, meloxicam, flunixin meglumine, and tolfenamic acid are recommended for further investigation into treatment of postpartum uterine inflammation. The results from this study confirm the immunomodulatory properties of the endometrial epithelium in response to inflammatory stimuli and suggest that NSAID may be beneficial in alleviating uterine inflammation.
). Clinical and subclinical endometritis both contribute to reproductive impairment, with cows exhibiting a decreased chance of pregnancy to first insemination and a reduced likelihood of pregnancy at 150 DIM compared with those without endometritis (
). Furthermore, the diagnostic technique and timing of sampling to measure endometritis have different associations with reproductive outcomes due to differences in sensitivity and specificity, which are reviewed by
A growing body of research indicates the possibility of using markers of inflammation, such as acute-phase proteins, to diagnose subclinical uterine inflammation, as they increase during infectious and inflammatory conditions (
). These studies suggest use of SAA as a diagnostic marker of subclinical disease. Another acute-phase protein, α-1 acid glycoprotein (αAGP), also known as orosomucoid, is produced in response to various stimuli including wound healing, physical trauma, and infection (
), all of which occur in the postpartum uterus. Cows with a fetid vaginal mucous odor due to high growth densities of pathogenic bacteria in the postpartum uterus had higher circulating αAGP compared with cows with normal vaginal mucus odor (
Treatment of postpartum inflammation with nonsteroidal anti-inflammatory drugs (NSAID) has had varying success due to differences in efficacy with drug choice, mode of action (i.e., cyclooxygenase or prostaglandin-endoperoxide synthase 1 or 2 inhibition), as well as timing and duration of treatment (
). A large-scale experiment investigating treatment with oral meloxicam at calving resulted in improved milk production in the first 3 d of lactation and reduced the risk of subclinical mastitis and culling (
demonstrated that endometrial inflammation was not affected by meloxicam, despite improvements in circulating indicators of inflammation (i.e., improved PMN function and decreased serum haptoglobin concentrations).
Flunixin meglumine is a nonselective (COX-1) and COX-2 inhibitor with a bioavailability of 76% (range, 44 to 119%) and a half-life between 3.1 and 26 h depending on the dose and injection route (
). However, in a subsequent study, the opposite effect on retained placenta was reported; cows treated with flunixin meglumine immediately (within 12 h) after calving demonstrated a reduced risk of culling and exhibited a lower incidence of retained placenta compared with control animals (2.3 vs. 9.3%, respectively;
), highlighting differences in physiological effects with the timing of NSAID administration and cow population. Ketoprofen is another nonselective COX-1 and COX-2 inhibitor that has been used to treat peripartal inflammation (
), and it binds irreversibly to COX; at low doses it may be more specific for COX-1 than COX-2. Aspirin is slowly absorbed (absorption half-life 3 h) but rapidly eliminated (terminal half-life of 32 min;
From a practical perspective, there are different milk and meat withholding times depending on the NSAID, which affect how suitable they are as on-farm solutions to treating inflammation in dairy cows. Furthermore, large-scale investigation of NSAID in vivo is costly and has ethical restrictions; thus, in vitro experimentation can be used to better understand biological responses of drugs before animal studies (
). We hypothesize that endometrial epithelial cells treatment with NSAID may diminish inflammatory markers upregulated during uterine inflammation. Therefore, the aim of this study is to establish an in vitro model of uterine inflammation (pilot experiment) and use it to investigate the efficacy of NSAID before experimentation in vivo (in vitro NSAID model).
MATERIALS AND METHODS
No human or animal subjects were used, so this analysis did not require approval by an Institutional Animal Care and Use Committee or Institutional Review Board.
Bovine Endometrial Epithelial Cell Culture
The bovine endometrial epithelial cell line (bEEL) was kindly gifted from Michel A. Fortier (Université Laval, Québec, Canada). Cells were grown in RPMI media (Sigma-Aldrich, R8758) containing 10% fetal bovine serum (FBS; Gibco, Thermo Fisher Scientific), penicillin-streptomycin (pen-strep) solution (1,000 IU/mL penicillin/mL 10,000 µg streptomycin/mL; Gibco, Thermo Fisher Scientific, 15140122), and incubated at 37°C and 5% CO2.
A pilot experiment was used to screen cell-culture parameters for their use in the in vitro NSAID model. The parameters screened included surface area (1.9 and 1.1 cm2; 24-well and 48-well cell culture plates) and volume (500 and 250 μL of cell-culture media), cell density (3.5 × 104, 7.5 × 104, 1.5 × 105, and 1 × 106 cells/well), a concentration of LPS (10 and 1,000 ng/mL), and the addition of an inflammation cocktail [LPS from Escherichia coli serotype O111:B4 (1 μg/mL; Sigma, L2630), bovine IL-1β (10 ng/mL; MediRay, RP106B), and bovine TNFα (50 ng/mL; MediRay, RP0055B)]. The different parameters were screened for expression of SAA and αAGP upregulation in comparison to untreated control cells. This resulted in the selection of the following parameters: 1.1-cm2 surface area [48-well culture plate (Nunc, Thermo Fisher Scientific, 150687)] with 250 μL of RPMI cell-culture media, 1.5 × 105 cells/well, and the inflammation cocktail. A time-course experiment using the selected parameters and SAA as the output, measured by ELISA, was used to determine the optimal time the in vitro NSAID experiment should incubate.
Enzyme-Linked Immunosorbent Assay
Chosen markers of inflammation, SAA and αAGP, were measured by commercially available ELISA kits (MyBiosource; MBS2702647 and MBS2702511, respectively). The ELISA kits were used according to manufacturers' instructions with both SAA and αAGP kits having a similar protocol.
Briefly, the kit and samples were brought to room temperature (18–25°C). Cell-culture supernatant was diluted 1:2,000 and 1:500 for SAA and αAGP, respectively, standard curves were made by 2-fold serial dilution, and 100 μL was pipetted into each well of the pre-coated ELISA plate. The plate was covered and incubated for 1 h at 37°C after which time, the detection antibody was added and incubated again for 1 h at 37°C. The plate was then washed, and horseradish peroxidase conjugate was added, and the plate incubated for 30 min at 37°C. Following this, the plate was washed 5 times and tetramethylbenzidine substrate added and incubated for 15 min at 37°C to allow color to develop, then stop solution was added and the plate was read at 450 nm (VersaMax microplate reader, Molecular Devices).
Protein concentrations were calculated using the SoftMax Pro v5.4 Data Analysis Software. External standard curves were used with a dynamic range of 0.156–10 ng/mL with a minimum sensitivity of 0.069 ng/mL for SAA and 15.6–1,000 ng/mL and minimum sensitivity of 7 ng/mL for αAGP. The intra-assay coefficient of variation (CV)% is stated as <10%, which was achieved within strips of the precoated plates and, therefore, this data was used. However, the intra-assay CV% was >40% between strips of the SAA kits, which was determined too high to use for the in vitro NSAID model, and an alternative kit was sourced. The interassay CV% is stated as <12%; however, all technical replicates were analyzed using one kit to avoid interassay variability.
In Vitro NSAID Model
Using the parameters established in the pilot experiment, cells were seeded at a density of 1.5 × 105 cells/well into a 48-well cell-culture plate and grown in RPMI medium containing 10% FBS and 1% penicillin-streptomycin for 24 h to allow cells to attach. Following attachment, the cell-culture medium was removed and replaced with one of 2 media types. The untreated (control) media was RPMI media without FBS or penicillin-streptomycin. The treated (inflamed) media also contained LPS (1 μg/mL), bovine IL-1β (10 ng/mL), and bovine TNFα (50 ng/mL). Cells were left to incubate for 3 h before the addition of an NSAID treatment. The final concentrations of NSAID were selected based on published pharmacokinetics data (see references below). Therefore, NSAID were purchased as powdered chemicals (Sigma; flunixin meglumine F0429, meloxicam PHR1799, ketoprofen K1751, aspirin A2093, and tolfenamic acid T0535), dissolved in dimethyl sulfoxide (DMSO; ensuring ≤0.1% DMSO used in assay), and diluted in RPMI media without FBS at the time of plating: 50 μM flunixin meglumine (flunixin;
). The NSAID were applied in triplicate to control wells (NSAID only) and inflamed wells (inflamed + NSAID). The in vitro model was repeated 3 times using bovine endometrial epithelial cell line cells at passage number 16–21, and each experiment included control (0.1% DMSO) and inflamed wells without the addition of NSAID in triplicate as experimental controls.
After the addition of the NSAID, the cells were incubated at 37°C and 5% CO2 for a further 24 h. This incubation time was determined using an SAA time course experiment and, once completed, cell culture supernatant (250 μL) was collected from the plates into 1.5-mL centrifuge tubes. Supernatant was centrifuged at room temperature at 3,000 × g for 10 min to pellet any cell debris, then transferred into a new 1.5-mL tube and stored at −80°C until protein analysis. Cells from the in vitro experiments were lysed in 20 μL of RLT buffer (Qiagen, Bio-Strategy) + 10% β-mercaptoethanol and stored at −80°C awaiting mRNA expression analysis.
Due to quality issues with newer batches of the originally used brand of SAA ELISA kits, detection of SAA protein in supernatant collected from the in vitro NSAID model was achieved using a different commercially available ELISA kit to that during validation in the pilot experiment (Tridelta Development, distributed by Invitrogen; KAA0021). Samples were analyzed using the manufacturer's protocol, similar to the protocol above with the exception that samples were not diluted and 50 μL of sample was used. After addition of the stop solution, the plate was read at 450 nm (VersaMax microplate reader), and concentrations of SAA were calculated using an in-run standard curve with a dynamic range of 18.8–300 ng/mL. The intra- and interassay CV% was 7.5 and 12.1%, respectively, and the analytical sensitivity was 1.5 μg/mL.
Detection of αAGP in supernatant collected from the in vitro model experiments was achieved using the same method as described previously (MyBiosource), however, due to the high variability between ELISA kits, cells treated with 3 selected NSAID (flunixin, meloxicam, and tolfenamic acid) were analyzed for αAGP again using a single kit to reduce interassay variability and a relative fold change calculated as with data for SAA using the formula: log2 Relative Fold Change = log2(inflamed + NSAID − inflamed) − log2(control + NSAID − control).
Cell Viability Assay
Bovine endometrial epithelial cells (passage 16 and 21) were grown in 96-well plates (2.5 × 104 cells/well) in full RPMI culture media. After 24 h, the media was removed and replaced with inflamed or control media (0.1% DMSO) and cultured for 3 h at 37°C and 5% CO2. Anti-inflammatory drugs were added at the concentrations described earlier in quadruplicate and incubated for 24 h. To obtain a standard curve, cells were seeded at 2-fold serial dilutions in duplicate from 1 × 105 to 6.25 × 103 cells/well. CellTiter 96 AQueous One Solution Assay (20 μL) was added to each well and incubated for 3.5 h at 37°C and 5% CO2, and absorbance was measured at 490 nm using a VersaMAX microplate reader (Bio-Strategy).
Bovine endometrial epithelial cells (2 × 105 cells/well) were seeded onto round, sterile glass coverslips (22 mm diameter) in 12-well plates, left for 48 h to adhere, then washed twice with PBS to remove nonadherent cells and debris. Cells were fixed with ice-cold (−20°C) methanol for 5 min on ice, then washed once with PBS. Cells were incubated at room temperature (approximately 18°C) for 30 min with blocking solution [1% (wt/vol) trim milk powder dissolved in PBS + 0.05% Tween 20], and incubated at room temperature for 1 h with a primary antibody: anticytokeratin-18 (KRT18, 1 μg/mL; SAB1404012, Sigma-Aldrich), antiepithelial cellular adhesion molecule (EpCAM, 1 μg/mL; ab71916, Abcam), anti-SAA (5 μg/mL; MA5–41676, ThermoFisher), or corresponding IgG isotype controls (data not shown), diluted in blocking solution. After incubation, cells were washed 5 times with PBS + 0.05% Tween 20 and incubated with the corresponding secondary antibody diluted in blocking solution [2 μg/mL; goat anti-rabbit IgG, Alexa Fluor 488 (A11008; ThermoFisher) or goat anti-mouse IgG, Alexa Fluor 488 (A11029, ThermoFisher)] plus 4′,6-diamidino-2-phenylindole (DAPI; 1 μg/mL; D8417, Sigma-Aldrich) for nuclear counter-staining. Secondary antibody-only controls were also included to confirm specificity (data not shown). Cells were incubated for 1 h in a dark humidity chamber at room temperature and then washed 5 times with PBS + 0.05% Tween 20. Coverslips were mounted overnight on microscope slides using ProLong Gold Antifade Mountant (ThermoFisher) at room temperature. Slides were stored in the dark at 4°C and imaged within a week of mounting, using an Eclipse Ti-U inverted epifluorescence microscope (Nikon).
Quantification of SAA in control and inflamed cells was undertaken using an LSM 900 Airyscan 2 (Zeiss) by imaging 3 random fields and using a z-series at 0.78 μM intervals. A summed projection of fluorescence in 16 sections was used to calculate the fluorescence intensities and divided by the total area measured to calculate an average fluorescence intensity per μM2.
RNA from experimental cell lysate was extracted using a Qiagen RNeasy micro kit (Bio-Strategy) following manufacturers' protocols with some modifications as follows. Technical triplicates within a cell-culture experiment were pipetted vigorously to lyse cells and then pooled, resulting in 60 μL of lysed cells (final n = 3/treatment). To this, 290 μL of RLT buffer + 10% β-mercaptoethanol was added resulting in a final volume of 350 μL. One volume (350 μL) of 70% molecular-grade ethanol was added to each sample and pipetted onto a RNeasy micro column, then centrifuged for 20 s at 10,000 × g at room temperature (∼20°C). The RNA was washed with RW1 (Qiagen), RPE (Qiagen), and 80% molecular-grade ethanol and dried according to the manufacturers' protocol using a centrifugation speed of 10,000 × g at room temperature (∼20°C). The RNA was eluted with 20 μL of RNase-free water. RNA integrity was measured using a Bioanalyser 2100 (Agilent Technologies), the average RNA integrity number was 10.0 (rounded mean with SD = 0.12). RNA was stored at −80°C awaiting mRNA expression analysis.
Expression of mRNA
Gene expression analysis from RNA of 3 NSAID treatments: flunixin, meloxicam, and tolfenamic acid (n = 3/treatment) was achieved using the nCounter Analysis System (Nanostring Technologies). Gene expression probes were multiplexed using nCounter Elements TagSets and customized probes for 192 genes, which included 11 endogenous control genes (Supplemental Table S1; https://doi.org/10.6084/m9.figshare.17343248.v2;
). Briefly, a MasterMix was prepared by mixing 70 μL of hybridization buffer and 7 μL each of Probe A and B Working Pools. Reactions were prepared in a 12-tube strip by pipetting 8 μL of MasterMix and 7 μL (not standardized; total of 2.3 ± 0.4 μg, mean ± SD) of RNA per reaction. The tubes were mixed and capped and incubated in a preheated thermocycler at 67°C for 21 h. After hybridization, excess probe was removed, and the probe-target complexes aligned and immobilized in the nCounter cartridge. These steps were performed automatically by the nCounter Prep Station. Sample cartridges were placed in the nCounter Digital Analyzer for data collection, as where fluorescent barcodes on the surface of the cartridge are counted and tabulated for each target molecule.
For analysis, the raw data (RCC files) were imported into nSolver Analysis Software v4.0 (http://www.nanostring.com/products/nSolver) and underwent the software's sample quality control routine. ‘Background thresholding' was used to subtract a background value calculated using the geometric mean of the counts of 6 negative internal controls from each sample. Positive control normalization was performed by using the geometric mean of 6 positive internal controls to compute the normalization factor. The normalization factor of all samples was inside the 0.2 to 5.5 range.
Reference gene normalization was performed using the geometric mean of counts for the reference genes ACTB, B2M, GAPDH, RPL10, RPL13A, RPL15, RPL19, UXT, and YWHAZ. Two reference genes, GUSB and SMUG1, were removed from the normalization due to low counts. The average of these geometric means across all lanes was used as the reference against which each lane is normalized. A normalization factor was then calculated for each of the lanes based on the geometric mean of counts for the reference gene in each lane relative to the average geometric mean of counts for the reference genes across all lanes. This normalization factor was then used to adjust the counts for each gene target and controls in the associated lane. The normalization factor of all samples was inside the 0.15 to 3 range.
Normalized data were screened for genes above the threshold of background expression, and the resulting 81 genes were further analyzed. Data from selected genes were log2 transformed, and ratios between treatment groups were calculated. These were used to calculate relative fold changes in mRNA expression using the formula:
Statistical significance of protein data was determined using GraphPad Prism 8.2.1 (GraphPad Software Inc.). An unpaired t-test was used to determine significance of the concentration of SAA and αAGP in response to the inflamed media (containing of LPS, TNFα, and IL1β) during pilot experimentation and for SAA quantification using immunofluorescence staining in the NSAID experiment. A one-way ANOVA using Šídák's multiple comparisons test was used to determine the significance between cell-culture supernatant collected from cells treated with the NSAID only (i.e., control) and the NSAID combined with the inflamed. Results are presented as means ± the standard error of the mean (SEM).
Gene expression data were log2 transformed and ratios for treatment comparisons were loaded into R version 4.0.2 for analysis. Hypothesis testing of the data was carried out using permutation ANOVA as implemented in the lmPerm R package version 2.1 (
In vitro culture of endometrial epithelial cells can be used to gain a greater physiological understanding of the inflammatory response that may occur during a uterine infection in vivo. The aim of this study was to design an in vitro model of inflammation to test the efficacy of NSAID. Two markers, SAA and αAGP, were selected as indicators of inflammation, release of which was induced by an ‘inflammation cocktail' of LPS, TNFα, and IL1β. The model was used to investigate how effective the 5 NSAID (ketoprofen, aspirin, flunixin, meloxicam, or tolfenamic acid) were at reducing the release of the selected inflammation markers, as well as expression of a panel of genes known to be involved in inflammatory pathways in response to 3 selected NSAID.
We have successfully established a model to investigate several NSAID in vitro, however, the parameters are very specific for our chosen inflammation markers. We demonstrate increased release of SAA and αAGP in cell culture supernatant in response to the ‘inflammation cocktail' (Figure 1) and an optimal incubation time of 24 h for the NSAID experiments due to concentrations of SAA in the supernatant (Figure 2). The parameters were used to demonstrate the effectiveness of NSAID in dampening these markers using the in vitro NSAID model as a proxy for uterine infection in vivo (Figure 3). Lipopolysaccharide treatment alone was not sufficient to induce the same effect (data not shown); however, this lack of effect on SAA has previously been demonstrated in vitro using primary bovine endometrial cells (
). The bovine endometrial epithelial cell line cells used in the current study were validated using immunofluorescence (Figure 4, Figure 5) and to investigate prostaglandin synthesis in response to individual treatment with LPS, IL1β, or TNFα (
). The use of the inflammation cocktail and SAA and αAGP in response to NSAID treatment is unique to this study, further, the method used to measure analytes should be considered. Our NSAID model adds to the growing body of evidence that they may be useful for treating postpartum uterine infection.
We determined that of the 5 NSAID investigated, meloxicam, flunixin, and tolfenamic acid were best at reducing inflammation using our in vitro model: SAA concentration was lower (P < 0.01) in the supernatant of inflamed cells treated with meloxicam (5 μM), flunixin meglumine (50 μM), and tolfenamic acid (50 μM) relative to control cells (Figure 6A). Tolfenamic acid also prevented an increase (P < 0.05) in the concentration of αAGP in the supernatant of inflamed cells (Figure 6B). Herein, we discuss the use of these NSAID to treat uterine inflammation in dairy cows; it has been reported with varying success, likely owing to differences in chemical structure and preferences of binding to COX subtypes (i.e., COX-1 or COX-2), whereby ketoprofen, aspirin, and flunixin meglumine are dual COX-1/COX-2 inhibitors, and meloxicam and tolfenamic acid are preferential COX2 inhibitors. Further, the pharmacokinetics, particularly the ability to reach uterine tissue, differs between NSAID, as demonstrated by varying success in vivo (
). Another study using repeated flunixin administration on 6 occasions during early postpartum, twice daily on the first 2 d and once daily for the subsequent 2, along with antimicrobial treatment, reduced pyrexia, increased chance of uterine involution by 90 d postpartum, and returned to estrus compared with nontreated animals (
). Meloxicam treatment (4 administrations: once daily on d 10 to 13 postpartum) reduced circulating indicators of inflammation and energy metabolism (reduced β-hydroxybutyrate and haptoglobin relative to control cows) but did not affect uterine inflammatory status (
). Treatment of uterine inflammation with tolfenamic acid in postpartum cows has not been investigated; however, the drug is used to treat dysmenorrhea in humans, owing to its action in COX inhibition in the uterus (
). Although in vitro experimentation cannot replace in vivo drug trials, our results indicate that this model is useful for unraveling the intricacies of drug efficacy and suggest that tolfenamic acid, meloxicam, and flunixin meglumine should be used for further investigation in vivo for treating uterine inflammation.
The results for αAGP suggest it might not be a suitable marker of inflammation for the in vitro model. The results of αAGP in the in vitro NSAID model were not consistent with the pilot experiment results; αAGP was greater in the supernatant of cells treated with the inflammation cocktail during the establishment of the in vitro model, however, αAGP was not different between control and inflamed cells (P > 0.05) in experiments investigating the effect of our chosen NSAID (Figure 3B). Although we detected a significant (P = 0.03) overall treatment effect on αAGP concentration, we observed no significant differences in αAGP concentration in cells treated with NSAID and cells treated with the corresponding NSAID plus inflammation cocktail. Due to the large variation in αAGP concentration between experiments, it was not useful in determining the effectiveness of NSAID treatment. Interestingly, this was consistent with αAGP expression, whereby, expression was too low and did not meet the criteria for inclusion in statistical analysis. Analysis of the normalized data indicated large variation between groups, and we detected no significant treatment effects (data not shown). Alpha-AGP is expressed by the liver and peripheral tissues (e.g., adipose tissue) to regulate inflammation (
). Importantly, circulating αAGP in plasma or serum occurs as several different glycoforms that have differing functions during acute inflammation, indicating the complexity of measuring it as a biomarker (
). Our results indicate it is unlikely that endometrial epithelial cells transcribe αAGP in high amounts at 24 h but may outside of this window, therefore, it is not an efficient biomarker of inflammation in the NSAID model. However, we conclude that αAGP may be a useful marker in vivo due to the importance of neutrophils for recovery of the postpartum uterus and in the etiology of uterine infection, but glycoforms should be considered due to their different functional effects.
The cell viability results indicate minimal cell death in the in vitro model, with the exception of treatment with tolfenamic acid, which significantly (P > 0.05) reduced cell viability relative to both the untreated and inflammation cocktail treated controls (Figure 7). This is consistent with a study demonstrating that meloxicam reduces cell viability of mammary epithelial cells in vitro (
), however, we did not see the same effects with our meloxicam treatment. Further, treatment with the inflammation cocktail numerically increased cell viability, which only reached significance when cells were treated with aspirin and tolfenamic acid. This is inconsistent with previous studies that demonstrated LPS treatment increases viability, but viability is reduced when cells are treated with LPS in combination with an NSAID relative to the NSAID-only control (
). We demonstrate that the treatments had little effect on cell viability, suggesting the concentrations we have used are not cytotoxic, thereby indicating the lower SAA and αAGP concentrations in response to NSAID treatment are not a result of cell death.
The mRNA expression results also support that our model may be useful in investigating uterine inflammation in vitro. Of the 181 target genes investigated in this study, 81 were expressed above the background and analyzed for the effect of treatment (Supplemental Table S2; https://doi.org/10.6084/m9.figshare.17343248.v2; Crookenden et al., 2021).
The expression of several genes involved in inflammation was downregulated in cells treated with NSAID relative to control cells, which is consistent with dampening of the endometrial epithelial cell inflammatory response. Relative fold-change analysis indicated genes involved in immune regulation and the eicosanoid pathway were downregulated (P < 0.05) by NSAID treatment (Figure 8); ABCA1, CASP8, CPNE3, CXCL16, CXCL2, CXCL5, CXCR4, GNAI2, IL6, LTA4H, LTF, LYN, M-SAA3.2, NFKB1, PTGS2, SAA2, SAA3, SLC2A1, and TAP demonstrated lower expression and PLA2G4A (P = 0.11) and BOLA (P = 0.09) tended to be lower in NSAID-treated cells relative to control cells. Of those mentioned above, members of the chemokine signaling pathway, CXCL16, CXCL2, CXCL5, and CXCR4, had increased expression in endometrium of cows with postpartum uterine inflammation compared with healthy cows using microarray analysis (
). The CXCL16 gene is primarily expressed by macrophages and dendritic cells where it mediates bacterial recognition and binding, but it is also expressed by epidermal epithelial cells as a mediator of innate immune activity (
). Further, NFκB is an important regulator of inflammatory processes and downregulation of NFKB1 by NSAID, as observed in the current study, is in support of an in vitro investigation of flunixin meglumine action by
. The importance of the regulation of genes involved in chemokine signaling for the endometrial immune response suggests that altered expression due to NSAID application may ameliorate uterine pathologies with underlying inflammatory etiology.
Gene expression of SAA in cells treated with flunixin meglumine, meloxicam, and tolfenamic acid corresponded with protein, suggesting its potential use as a marker of uterine inflammation. Expression of 3 SAA genes (SAA2, SAA3, M-SAA3.2) was greatly upregulated by inflammation (Figure 9) and dampened by NSAID treatment in cell culture supernatant (Figure 8). However, an increase of intracellular SAA in response to the ‘inflammation cocktail' could not be detected using immunofluorescence (Figure 5), likely due to the quick release of SAA from the cells. Gene expression of SAA4, which is constitutively expressed by the liver (
), did not meet the expression threshold criteria for analysis (data not shown). The isoform SAA2 is a major acute-phase protein produced by the liver, whereas SAA3 and M-SAA3.2 are produced extrahepatically, in response to proinflammatory cytokines including TNFα and IL1β (
), however, the current study is the first to demonstrate all 3 of these SAA genes are upregulated by endometrial epithelial cells during inflammation. In concordance with our results, SAA3 expression has been shown to increase >2,000-fold in endometrial epithelium after 24 h exposure to E. coli, along with mRNA expression of tracheal antimicrobial peptide (TAP), which increased >200-fold relative to pre-stimulation with bacteria (
). Hence, we suggest that SAA should be further investigated as a marker of postpartum inflammation in the context of endometritis.
The mRNA expression results confirmed regulation of the eicosanoid pathway, which suggests the model is valid for use in investigating the action of NSAID. Cells treated with the inflammation cocktail demonstrated increased mRNA expression of proinflammatory genes and genes that regulate the innate immune response as well as enzymes involved in prostaglandin synthesis (Figure 9). Further, mRNA expression is consistent with previous work that indicated PTGS2, which is also known as COX2, and IL6 expression, is upregulated when endometrial epithelial cells are treated with LPS, IL-1β, or TNFα (
). It is possible that the combination treatment of the proinflammatory mediators used in the current study was sufficient to upregulate TBXAS1, the enzyme product of which (thromboxane A synthase 1) catalyzes the conversion of prostaglandin H to thromboxane A. Thromboxane A synthase 1 is involved in drug metabolism, and polymorphisms in TBXAS1 are linked with NSAID-induced pathologies (
). Flunixin meglumine, meloxicam, and tolfenamic acid lowered expression of PTGS2 and only tolfenamic acid lowered mRNA expression of TBXAS1 relative to control cells (Figure 8), demonstrating their regulation of the COX-2 pathway. We propose that the in vitro model employed in the current study can be used as a proxy to investigate the inflammatory response of the uterus and further understand uterine physiology in response to drug treatment.
We observed differences in mRNA expression depending on the NSAID used, highlighting potential differences in efficacy for treating uterine inflammation. We detected 3 genes that demonstrated (P < 0.05) lower (LDHA, NCOR1) or higher (PAFAH2) relative fold change in mRNA expression in cells treated with flunixin meglumine and tolfenamic acid compared with control cells. The protein product of LDHA is lactate dehydrogenase-A, a subunit of the enzyme lactate dehydrogenase (LDH-5), which catalyzes pyruvate to lactate, and nuclear corepressor 1 (encoded by NCOR1) is a transcription factor; both LDHA and NCOR1 mRNA expression increase in endometrium during hypoxia and increased protein expression is associated with human endometrial pathologies including cancer and endometriosis, respectively, relative to healthy tissue (
). The gene PAFAH2 encodes an acetylhydrolase (AH) enzyme that degrades platelet activating factor (PAF), a potent inflammatory phospholipid and an important trigger of the uterine inflammatory response (
). Although we did not investigate expression of PAF, PAF-AH is important for terminating PAF-induced inflammation, which is likely an important regulatory mechanism of chronic uterine inflammation. These additional mediators of inflammation, that are either up- or downregulated in response to flunixin meglumine or tolfenamic acid treatment, could indicate differences from meloxicam when using to treat uterine inflammation.
We have established an in vitro model of uterine inflammation and investigated the efficacy of 5 NSAID. We determined that flunixin meglumine, meloxicam, and tolfenamic acid are recommended for further investigation in vivo due to their ability to prevent an increase in uterine inflammation in vitro. Serum amyloid A was a useful marker of uterine inflammation and was lowered in response to NSAID treatment, which was in agreement with mRNA expression of 3 SAA isoforms. Expression of other genes decreased by NSAID treatment included genes involved in the eicosanoid pathway and inflammatory mediators, providing evidence that these drugs affect endometrial inflammatory processes and may act this way during uterine infection.
This work was supported by funding from a partnership between New Zealand dairy farmers through DairyNZ Inc. (Hamilton, New Zealand) and the Ministry of Business, Innovation and Employment (contract #DRCX1302, Wellington, New Zealand), with matched co-funding from the AgResearch Strategic Science Investment Fund (Lincoln, New Zealand). We thank Paul Maclean, AgResearch, for his statistical support and Francis Phimister, AgResearch, for the secondary antibodies used in the immunofluorescence staining. We also acknowledge Matthew Savoian, Manawatū Microscopy and Imaging Centre (Palmerston North, New Zealand), for his expertise with the immunofluorescence imaging. The authors have not stated any conflicts of interest.
Eicosanoid pathway expression in bovine endometrial epithelial and stromal cells in response to lipopolysaccharide, interleukin 1 beta, and tumor necrosis factor alpha.