Abstract
The cyclooxygenase-2 (COX-2) selective inhibitor celecoxib is widely used in the treatment of pain and inflammation. Celecoxib has been explored as a possible treatment of liver fibrosis with contradictory results, depending on the model. The present study reports the effect of celecoxib in a 5-week carbon tetrachloride (CCl4)-induced liver fibrosis mouse model. Celecoxib alone and in combination with inhibitors of the enzyme-soluble epoxide hydrolase (sEH), as well as a dual inhibitor that targets both COX-2 and sEH, were administered via osmotic minipump to mice receiving intraperitoneal injections of CCl4. Collagen deposition was elevated in the mice treated with both celecoxib and CCl4 compared with the control or CCl4-only groups, as assessed by trichrome staining. Histopathology revealed more extensive fibrosis and cell death in the animals treated with both celecoxib and CCl4 compared with all other experimental groups. Although some markers of fibrosis, such as matrix metalloprotease, were unchanged or lowered in the animals treated with both celecoxib and CCl4, overall, hepatic fibrosis was more severe in this group. Cotreatment with celecoxib and an inhibitor of sEH or treatment with a dual inhibitor of COX-2 and sEH decreased the elevated levels of fibrotic markers observed in the group that received both celecoxib and CCl4. Oxylipid analysis revealed that celecoxib reduced the level of prostaglandin E2 relative to the CCl4 only group. Overall, celecoxib treatment did not decrease liver fibrosis in CCl4-treated mice.
Introduction
Celecoxib is a nonsteroidal anti-inflammatory drug (NSAID) used by millions of patients to alleviate the pain and inflammation associated with diseases such as rheumatoid arthritis (Bessone et al., 2016). Reviews of controlled trials have found no significant difference in the incidence of liver damage between patients administered celecoxib and those receiving placebo (Bessone et al., 2016).
Liver fibrosis is the result of a normally beneficial wound-healing process that can be initiated by toxicants such as ethanol or carbon tetrachloride (Liedtke et al., 2013). This inflammatory process involves activation of resident macrophages called hepatic stellate cells (HSCs) and the recruitment of macrophages, both of which express proinflammatory signaling molecules, along with enzymes and structural proteins that remodel the extracellular matrix (Pellicoro et al., 2014). This remodeling includes the increased deposition of matrix proteins such as collagen, as well as changes in the populations of metalloproteases (Pellicoro et al., 2014). If damage from exposure to the toxicant continues, liver fibrosis will alter the architecture of the organ and lead to liver failure (Liedtke et al., 2013).
Because of its anti-inflammatory effect, celecoxib has been explored as a possible therapy in several models of liver fibrosis, such as the thioacetamide (TAA)- and carbon tetrachloride (CCl4)-induced rodent models and the surgical bile duct ligation (BDL) rodent model. CCl4 acts primarily through an increase in hepatic lipid peroxidation and oxidative stress, whereas TAA acts primarily through an increase in oxidative stress, processes that damage hepatocytes and trigger fibrosis (Martínez et al., 2014). BDL is a surgical model in which the bile duct is partially ligated, leading to cholestasis, liver damage, and fibrosis (Martínez et al., 2014). In some models, treatment with celecoxib has resulted in a reduction in fibrosis and inflammation, whereas in others, including some rat CCl4-induced models, celecoxib has worsened liver damage and fibrosis (Hui et al., 2006; Paik et al., 2009; Chávez et al., 2010).
Celecoxib targets cyclooxygenase 2 (COX-2), an enzyme that metabolizes arachidonic acid to a class of oxidized fatty acids called prostaglandins (Shi and Klotz, 2008). These oxylipids have diverse effects in the liver, but many COX-2 metabolites increase inflammation and portal hypertension (Sacerdoti et al., 2015). Arachidonic acid is also metabolized by cytochrome P450s (P450s) to form the epoxyeicosatrienoic acids (EETs) (Morisseau and Hammock, 2013). The EETs have been investigated in several disease models and have been found to be anti-inflammatory, organ protective, and antifibrotic in heart and kidney models of fibrosis (Morisseau and Hammock, 2013). The EETs are further metabolized by soluble epoxide hydrolase (sEH) to the dihydroxyeicosatrienoic acids that are less lipophilic and more readily conjugated and excreted by the organism (Morisseau and Hammock, 2013).
We previously modulated the oxylipids in a CCl4-induced model of hepatic fibrosis through dietary manipulation of lipid intake as well as inhibition of sEH, which blocks the major route of metabolism of the EETs and other epoxy fatty acids (Harris et al., 2015, 2016). In general, perturbation of the oxylipids with sEH inhibitors reduced collagen deposition in addition to the expression and activity of profibrotic matrix metalloproteases (MMPs) (Harris et al., 2015, 2016). These results raised the question of how celecoxib, another modulator of oxylipids, would impact fibrosis in the CCl4 model and whether the tools we developed for inhibiting both sEH and COX-2 would alter the observed effects of COX-2 inhibition.
In this study, we treated mice with CCl4 over a 5-week period to induce liver fibrosis. Interestingly, markers of fibrosis were elevated in the mice that received both celecoxib and CCl4 compared with those animals receiving CCl4 alone. Modulation of the EETs by either an sEH inhibitor or a dual inhibitor of COX-2 and sEH blunted some aspects of the profibrotic effect of COX-2 inhibition in the CCl4 background.
Materials and Methods
Animals.
Male C57BL/6NCrl mice (∼25 g) were obtained from Charles River, Inc. (Wilmington, MA) 1 week before the experiment and kept on a 12-hour light/dark cycle. The animal protocol was approved by the University of California Davis Institutional Animal Care and Use Committee, and the study was performed in accordance with the National Institutes of Health guide for the care and use of laboratory animals.
Chemicals.
All commercial chemicals were purchased from Sigma (St. Louis, MO) unless otherwise noted. TPPU, 1-(4-trifluoromethoxyphenyl)-3-(1-propionylpiperidin-4-yl)-urea and PTUPB, 4-(5-phenyl-3-{3-[3-(4-trifluoromethylphenyl)-ureido]-propyl}-pyrazol-1-yl)-benzenesulfonamide, were synthesized according to the previous procedures (Rose et al., 2010; Hwang et al., 2011). Celecoxib was a gift from Pfizer (New York, NY). Structures of TPPU, PTUPB, and celecoxib are given in Supplemental Fig. S1.
Experimental Design.
The mice were randomly divided into five experimental groups (six animals/group): 1) control, 2) CCl4 only, 3) celecoxib + CCl4, 4) celecoxib + TPPU + CCl4, 5) PTUPB + CCl4. CCl4 was administered after a protocol described in detail in Constandinou et al. (2005). Briefly, CCl4 was mixed 1:1 (v/v) with Neobee M5 (Sigma) and injected i.p. every 5 days for a total of seven injections of 80 mg/kg. Mice were euthanized 3 days after the final injection. Celecoxib, PTUPB, and a combination of celecoxib and TPPU were administered subcutaneously via Alzet model 2006 osmotic minipumps (Cupertino, CA). These pumps produce a 3.6 µl/day continuous flow rate for up to 45 days. The control group received Neobee M5 only. The control group and CCl4-only group were implanted with minipumps filled with the 1:1 (v/v) PEG400:DMSO solution.
Plasma Blood-Level Analysis.
Mouse plasma (10 µl) was mixed with 50 µl of water containing 0.1% formic acid. The liquid-liquid extraction and LC-MS/MS analysis were performed as previously described (Harris et al., 2016).
Determination of Liver Tissue Hydroxyproline Levels.
Hydroxyproline content was determined according to a method described in Reddy and Enwemeka (1996). Briefly, 10 µl of 10 N NaOH was added to 40 µl of 0.33 mg/µl liver tissue homogenate, and the solution was autoclaved at 120°C for 25 minutes. After the reaction cooled to room temperature, chloramine T reagent (0.84% chloramine-T, 42 mM sodium acetate, 2.6 mM citric acid, and 39.5% isopropanol) was added (450 µl), bringing the volume to 500 µl. After incubation at room temperature for 25 minutes, DMAB reagent, 15% 4-(dimethylamino)benzaldehyde in 2:1 (v/v) isopropanol/perchloric acid mixture was added (500 µl), bringing the volume to 1 ml. The solution was then incubated at 65°C for 20 minutes, and the absorbances of the samples measured at 550 nm.
mRNA Transcript Analysis.
Liver tissue was immediately stored in RNA later solution (Thermo Fisher Scientific, Waltham, MA) for 24 hours at 4°C before freezing at −80°C. Tissue was homogenized using a roto stator grinder (IKA Works Inc., Wilmington, NC) and passed through a QIAshredder column (Qiagen, Valencia, CA). Total RNA was purified using RNeasy kit (Qiagen). cDNA synthesis from total RNA was performed by the QuantiTect Reverse Transcription Kit (Qiagen). RT-PCR was performed using an Applied Biosciences Fast 7500 Real-Time PCR System (Foster City, CA). The following Taqman probes were purchased from Life Technologies Corporation: GAPDH, Mm99999915_g1; TGFB1, Mm01178820_m1; MMP-2, Mm00439498_m1; MMP-9, Mm00442991_m1; collagen 1A1, Mm00801666_g1; tissue inhibitors of metalloproteases (TIMP-1), Mm00441818_m1 (Life Technologies Corp., Grand Island, NY). GAPDH was used as the internal control.
Histopathology Scoring.
Liver samples were immersion-fixed in 10% neutral-buffered formalin for 48 hours. Samples were stored in 70% (v/v) ethanol in distilled deionized water before routine processing. Tissues were processed, embedded in paraffin, sectioned, and stained with H&E by the histopathology service in the UC Davis Veterinary Medical Teaching Hospital (Davis, CA). H&E-stained sections of liver were scored for inflammation, fibrosis, and cellular damage by a board-certified laboratory animal pathologist (D.M.I.). The scoring system is reported in Supplemental Table S1.
Trichrome Staining and Immunohistochemistry.
αSmooth muscle actin (SMA) and F4/80 immunohistochemistry and trichrome staining were performed by the Genomic Pathology Laboratory at UC Davis and quantitated using imaging analysis software from Aperio (Sausalito, CA) using the IHC and positive pixel count algorithms.
Oxylipid Analysis.
Oxylipids in the COX and lipoxygenase (LOX) branches of the arachidonic acid cascade were extracted and detected as described (Harris et al., 2016). Oxylipids in the P450 branch of the arachidonic acid cascade were detected as follows. Samples were analyzed on a Waters Acquity ultraperformance liquid chromatograph coupled to a Waters Xevo TQ-S Mass Spectrometer in negative electrospray ionization mode. Samples were injected (5 µl) and separated using a Phenomenex Kinetex column (150 × 2.1 mm; 1.7 µm) at 40°C using the following mobile-phase gradient, consisting of water (A) and acetonitrile (B), each containing 0.1% acetic acid: initial conditions of 65:35 A:B for 2.9 minutes (t = 2.9 minutes), changing to 45:55 at 3 minutes (t = 3 minutes) and decreasing to 35:65 over 5.5 minutes (t = 8.5), decreasing to 5:95 over 4 minutes (t = 12.5), holding at 5:95 for 1 minute (t = 13.5), reverting to initial conditions of 65:35 in 0.1 minute (t = 13.6 minutes) and re-equilibrating for 1.9 minutes (t = 15.5 minutes). Mass spectral analysis was accomplished using a capillary voltage of 3 kV, a desolvation temperature of 200°C, a desolvation gas flow of 800 liters/hour, a cone gas flow of 150 liters/hour, nebulizer pressure of 6 bars, and collision gas flow of 0.15 ml/min. Mass transitions for each EET were as follows: 14,15-EET (319.0 > 219.0), 11,12-EET (319.0 > 167.0), and 8,9-EET (319.0 > 155.0). Analyte concentration was quantified against an internal standard calibration curve, normalizing analyte response to 11,12-EET-d11 (330.2 > 167.2), and 12-(3-cyclohexyl-ureido)dodecanoic acid (341.3 > 216.4) to correct the extraction efficiency and instrument response, respectively. Dwell time for each analyte was 25 milliseconds. The detailed information for all analytes are reported in Supplemental Table S2.
Statistics.
Normality was assessed by the Shapiro-Wilk test. Significance was determined by one-way analysis of variance, followed by the Dunnett multiple comparison test. All statistical calculations were performed using SigmaPlot (Systat Software, Inc., San Jose, CA).
Results
Subcutaneous Osmotic Minipumps Produced High Plasma Concentrations of Celecoxib.
Male C57BL mice were implanted with osmotic minipumps 2 days before the start of CCl4 exposure via intraperitoneal injections as described in Materials and Methods. The five experimental groups were: 1) control, 2) CCl4 only, 3) celecoxib + CCl4, 4) celecoxib + TPPU + CCl4, and 5) PTUPB + CCl4. The drugs were loaded into minipumps so that the calculated dose of each compound would be 10 mg/kg per day.
When we examined plasma levels of the drugs, we found that the TPPU and celecoxib levels in the celecoxib + TPPU + CCl4 group were 1.11 ± 0.018 and 1.32 ± 0.039 µM, respectively. The plasma concentration of celecoxib in the celecoxib + CCl4 group was 1.58 ± 0.47 µM, and the PTUPB concentration in the PTUPB + CCl4 group was 0.65 ± 0.027 µM. These plasma concentrations are far above the IC50 of for celecoxib and TPPU with their respective target mouse enzymes and approximately half the IC50 for PTUPB with human COX-2 and far above the IC50 for PTUPB with human sEH. Despite developing robust fibrosis after the 5-week CCl4 treatment, the weight of the mice increased, with no statistical difference between the groups (P < 0.05), the CCl4 toxicity being less severe than other models (Supplemental Fig. S2).
Celecoxib Increased Collagen Deposition and Liver Damage in CCl4-Treated Mice.
Trichrome staining revealed that the celecoxib + CCl4 had a higher amount of collagen deposition than the CCl4-only group (Fig. 1, A–F). The PTUPB + CCl4 group had a reduced level of collagen deposition compared with the celecoxib + CCl4 group, although still higher than the control group (Fig. 1F). Hydroxyproline levels in hepatic tissue from the celecoxib + CCl4 group was the same as the CCl4-only group (P > 0.05), and the levels from both the celecoxib + TPPU + CCl4 and PTUPB + CCl4 groups were lower than the CCl4 only group (Fig. 1G).
We next performed a histopathological assessment of the extent of liver fibrosis and damage (Table 1). The categories were defined as follows: vacuolation—lacy or vacuolated cytoplasm that distends the hepatocyte; necrosis—single cell death of hepatocytes in centrilobular to random areas; lipofuscinosis—accumulation of lipid pigments, associated with nondegradable cellular breakdown products in macrophages; fibrosis—fibroblast proliferation with increased deposition of collagen around centrilobular regions occasionally connecting between centrilobular regions (bridges) or extending into the surrounding parenchyma (dissecting); inflammation—mixed neutrophilic and mononuclear inflammation around predominantly centrilobular regions. The slides were read in a blind fashion. The scoring system is reported in Supplemental Table S1. The celecoxib + CCl4 group had the highest composite lesion scores owing to greater indices of hepatocellular damage (vacuolar degeneration, necrosis, fibrosis, regeneration). Examination of the celecoxib + TPPU + CCl4 group showed that the addition of TPPU dampened the effect of celecoxib (P = 0.0117), predominantly by decreasing hepatic necrosis (P = 0.0256) and regeneration (P = 0.0134).
Celecoxib Did Not Significantly Raise the Level of Profibrotic Markers in CCl4- Treated Mice.
To gauge further the extent of fibrosis, we measured the mRNA level of common markers by RTPCR (Fig. 2). The expression levels of collagen 1A1 (COL 1A1) was increased in the CCl4-only group compared with the control group (P < 0.05) and further increased in the celecoxib + CCl4 group, but this difference did not reach statistical significance (P > 0.05) (Fig. 2A). The mRNA expression of MMP-2 and MMP-9 were also elevated in the CCl4-only group (Fig. 2, B and C). For MMP-2, the three treatment groups were not different from the CCl4-only group (Fig. 2B). In the case of MMP-9, levels in the celecoxib + CCl4, celecoxib + TPPU + CCl4 and PTUPB + CCl4 groups did not differ from the CCl4-only group, but were also no different than the control group (P > 0.05) (Fig. 2C). TIMP-1 was elevated in the CCl4-only group, but the further increase observed in the celecoxib + CCl4, celecoxib + TPPU + CCl4, and PTUPB + CCl4 groups did not reach significance (Fig. 2D). For TGF-β, the levels of the celecoxib + CCl4 and celecoxib + TPPU + CCl4 groups were lower than the CCl4-only group, but this difference did not reach statistical significance (Fig. 2E).
Celecoxib Increased αSMA But Not F4/80 Staining in CCl4-Treated Mice.
F4/80 is a mouse cell surface protein expressed in many populations of macrophages. CCl4 increased the F4/80 immunohistochemical staining relative to the control group (Fig. 3). Whereas the amount of positive staining in the celecoxib + CCl4, PTUPB + CCl4, and celecoxib + TPPU + CCl4 groups were not significantly reduced (P > 0.05) compared with the CCl4-only group (Fig. 3F), these groups were also not different from the control group.
To assess hepatic stellate cell (HSC) activation, we performed IHC on liver tissue with an anti-αSMA (smooth muscle actin) antibody (Fig. 4). HSC activation was greatest in the celecoxib + CCl4 group, where it was significantly increased relative to both the control and CCl4-only groups (P < 0.05). Cotreatment with celecoxib and TPPU or treatment with PTUPB in the CCl4 background decreased the level of HSC activation compared with animals receiving celecoxib and CCl4. In the case of the celecoxib + TPPU + CCl4 group, HSC activation was significantly less than the celecoxib + CCl4 group (P < 0.05). In the PTUPB + CCl4 group, the level of HSC activation was not statistically different from the CCl4-only group (P < 0.05), unlike the celecoxib + CCl4 group.
Celecoxib Lowered the Tissue Level of COX Metabolites of Arachidonic Acid in CCl4-Treated Mice.
We next examined the hepatic oxylipids by LC-MS/MS after solid-phase extraction of lipids from tissue lysates. A simplified diagram of these pathways is given in Supplemental Fig. S1. Although the differences did not reach statistical significance for every analyte, CCl4 treatment displayed the same trend across several COX metabolites, raising their concentration relative to the control group (Fig. 5; Supplemental Table S4). Three metabolites in the COX branch of the arachidonic acid cascade were significantly altered in the celecoxib + CCl4 group relative to either the control or the CCl4-only group (Fig. 5, A–D). The prostaglandin E2 and PGD2 levels were significantly higher in the CCl4-only group compared with the celecoxib + CCl4 group. Although they did not reach the level of statistical significance, prostaglandin J2 (PGJ2) and 15-deoxy-Δ12,14-PGJ2 showed similar trends, the celecoxib + CCl4 group being lower than the CCl4-only group. Interestingly, the LOX metabolite of arachidonic acid, lipoxin A4 (LXA4), was significantly elevated by celecoxib (Fig. 5E). The complete oxylipid profile is summarized in Supplemental Table S4.
Finally, we determined the effect of the treatments on hepatic EET levels (Supplemental Fig. S3; Supplemental Table S4). Although the differences in the levels of these analytes did not achieve statistical significance between any of the groups, the means for 8,9-, 11,12-, and 14,15-EET were elevated in CCl4-only group relative to the control group (Supplemental Fig. S3, A–C). The celecoxib + CCl4 group had lowered 11,12- and 14,15- levels relative to the control group (Supplemental Fig. S3, B and C). In the celecoxib + TPPU + CCl4 and PTUPB + CCl4 groups, the EET levels were increased relative to the celecoxib + CCl4 group (Supplemental Fig. S3, A–C).
Discussion
There is a current disagreement in the literature regarding the use of celecoxib as a treatment of liver damage and fibrosis (Hui et al., 2004; Paik et al., 2009; Yu et al., 2009; Wen et al., 2014). Relevant to this study, opposing results have been reported in CCl4 models of liver fibrosis. In a short-term model, celecoxib reduced liver inflammation and damage during the early phase of CCl4 toxicity (Washino et al., 2010). A similar short-term study found that celecoxib reduced markers of lipid peroxidation but had no effect on hepatic toxicity or necrosis (Ekor et al., 2013). Chávez et al. (2010) reported that celecoxib reduced fibrosis as judged by collagen deposition and other biochemical markers; however, Hui et al. (2006) reported that treatment with celecoxib increased hepatic fibrosis, gauged by collagen deposition and MMP production. In their discussion of these CCl4 studies, Chávez et al. (2010) suggested that differences in CCl4 model parameters might explain the conflicting results, in particular, the dose and length of time of toxicant exposure.
In the current study, we obtained results similar to those reported by Hui et al. (2006) after treatment with celecoxib in their CCl4 rat model of liver fibrosis, as well as those obtained by Reilly et al. (2001) using COX-2 knockout mice in an acetaminophen (APAP) model of acute liver damage. Like Hui et al. (2006), we found that celecoxib treatment increased the amount of collagen deposition by approximately 25% compared with the CCl4-only group. We also observed an increase in αSMA staining, a marker for activated HSCs, in the celecoxib + CCl4 group compared with the CCl4-only group, although our model produced an almost 2-fold increase compared with their reported 40% increase (Hui et al., 2006); however, we did not see an increase in MMP-2 or MMP-9 mRNA expression in the celecoxib + CCl4 group compared with the CCl4-only group, unlike Hui and colleagues. It should be kept in mind that Hui et al. treated with a much higher dose of CCl4 (i.e., 3200 mg/kg delivered twice weekly) compared with our 80 mg/kg delivered twice weekly. Our study also parallels results using COX-2 transgenic mice in a liver damage model. When COX-2–deficient mice were given a single bolus of APAP in an acute liver damage model (Reilly et al., 2001), the COX−/− and COX−/+ animals displayed greater bridging perivenous hepatocyte necrosis than did wild-type (Reilly et al., 2001). Similarly, we observed an increase in hepatic necrosis in the celecoxib + CCl4 group compared with the CCl4-only group. Although the mechanism of toxicity of APAP and CCl4 are different, this may indicate that celecoxib accelerates hepatic necrosis in models that involve severe oxidative stress.
We used two pharmacologic tools in addition to celecoxib to perturb the oxylipid homeostasis, TPPU and PTUPB. TPPU is a potent inhibitor of soluble epoxide hydrolase, an enzyme that metabolizes epoxy fatty acids (Rose et al., 2010). In previous studies, we have reported that TPPU treatment reduces hepatic fibrosis (Harris et al., 2015, 2016). PTUPB contains the pharmacophores from both celecoxib and TPPU and inhibits both enzymes (Hwang et al., 2011). Treatment with PTUPB or cotreatment with TPPU and celecoxib decreased the elevated collagen deposition observed in the celecoxib + CCl4 group. The most striking result obtained with these inhibitors concerns the elevated αSMA in the celecoxib + CCl4 group. Although not returning αSMA levels to that of the control group, cotreatment with the sEH inhibitor attenuated the elevated αSMA expression observed in the celecoxib + CCl4 group. Changes in associated oxylipid mediators are discussed subsequently herein. We saw a slight reduction in F4/80 expression, a marker of macrophages, in the celecoxib + CCl4 group compared with the CCl4-only group, indicating that the increase in fibrosis in the celecoxib + CCl4 group is not due to increased macrophage migration.
The downstream COX-2 metabolites prostaglandin E2 (PGE2) and 15-deoxy-Δ12,14-PGJ2 have been proposed to mediate both the profibrotic and antifibrotic effects of COX-2 inhibition or gene deletion, depending on the model. In general, prostaglandins have organ-protective effects in the liver (Reilly et al., 2001), and PGE2 reduced collagen production in a human cell culture model of HSCs (Hui et al., 2004). Results that support a profibrotic role for COX-2 downstream metabolites include the observation that plasma PGE2 levels were elevated after treatment with TAA (Wen et al., 2014). In a separate study using the BDL and TAA models, PGE2 levels were increased in the fibrotic animals and reduced after treatment with celecoxib, with an accompanying attenuation of fibrosis (Paik et al., 2009). 15-deoxy-Δ12,14-PGJ2 is another potential profibrotic downstream COX-2 metabolite elevated in the serum of fibrotic rats treated with CCl4 (Planagumà et al., 2005); however, 15-deoxy-Δ12,14-PGJ2 has also been shown to induce apoptosis of human myofibroblasts, a process that contributes to the resolution of fibrosis (Li et al., 2001).
We performed an analysis of oxylipids of the arachidonic acid cascade to determine the effect of COX-2 and sEH inhibition. Overall, COX metabolites were elevated in the CCl4-only group, although the levels did not reach statistical significance in most cases. PGE2 levels were elevated in the CCl4-only group and substantially lowered in the celecoxib + CCl4 group, as expected for a COX inhibitor. As outlined before, extensive fibrosis was seen in both these groups; the greatest degree of fibrosis was found in the celecoxib + CCl4 group, where the lowest PGE2 level was detected. Consistent with an anti-inflammatory effect of celecoxib in the liver, LXA4 was greatly elevated by celecoxib. Treatment with TPPU and celecoxib or the dual COX-2/sEH inhibitor in the CCl4 background raised PGE2 levels and lowed LXA4 levels compared with the celecoxib + CCl4 group, in both cases bringing the levels of these lipid mediators closer to those observed in the control group. It is possible that a drastic reduction in inflammatory processes in the CCl4 background results in even greater damage to the liver and that, depending on the inflammatory state of the liver, an elevation in PGE2 levels will either increase or decrease liver damage and fibrosis, which might explain some of the contradictory conclusions regarding the role of PGE2 in different liver fibrosis models.
Wide variation was seen in the PGJ2 and 15-deoxy-Δ12,14-PGJ2 levels, but the trend of the metabolites was similar. Whereas CCl4 elevated their mean levels compared with the control group, levels were slightly decreased in other groups. Based on these data, the 15-deoxy-Δ12,14-PGJ2 levels do not explain the increased fibrosis in the celecoxib + CCl4 group or the attenuation in fibrosis after TPPU cotreatment.
sEH inhibitors, and presumably the underlying elevation in epoxy fatty acid chemical mediators, have reduced pathologic fibrosis in several models (Kim et al., 2014; Hye Khan et al., 2016; Liao et al., 2016; Zhou et al., 2016), including ischemia-driven heart failure (Sirish et al., 2013). Earlier we demonstrated that soluble epoxide hydrolase inhibitor (sEHI) reduced hepatic fibrosis in the same CCl4 model used in this study (Harris et al., 2015, 2016); however, unlike some other pathologic changes reversed by sEHI, the effects were not enhanced by a diet depleted in ω-6 relative to ω-3 fatty acids.
sEH inhibitors synergize with NSAIDs in many biologic systems (Schmelzer et al., 2006; Hwang et al., 2011; Zhang et al., 2014; Hye Khan et al., 2016). Surprisingly, sEH inhibitors prevent and reverse gastrointestinal erosion caused by the NSAIDs diclofenac (Goswami et al., 2016, 2017), and indomethacin (Goswami, unpublished), and in mice they block some of the cardiovascular effects caused by COX-2 inhibitors, possibly by returning the enhanced blood levels of thromboxane A2 to prostaglandin I2 to more normal levels (Schmelzer et al., 2006). Since celecoxib is a selective COX-2 inhibitor and PTUPB is a highly selective COX-2 inhibitor, we expected a strong positive synergistic effect between the sEHI TPPU and celecoxib and a dramatic reduction in fibrosis with PTUPB. Surprisingly, the profibrotic effect of celecoxib in our CCl4 model overpowered possible positive synergistic interaction with sEHI, giving only moderate efficacy in reducing fibrosis and fibrosis markers. This result suggests that celecoxib in not an attractive drug to treat hepatic fibrosis in a system driven by pathologic peroxidation; however, both sEHI alone and dual COX/sEH inhibitors may be worth examining in hepatic fibrosis models driven by agents other than CCl4. In patients with hepatic fibrosis who are on celecoxib or other NSAIDs for a different disease, sEHI may ameliorate exacerbation of hepatic fibrosis by celecoxib.
Finally, although differences observed in the EETs did not achieve statistical significance, the trends suggest that the sEH inhibitors were able to elevate hepatic EET levels, although CCl4 treatment also raised the level of these oxylipids. Given the variation in the data, one must use caution in interpretation, but it is possible that the slight increase in EET levels after CCl4 treatment is insufficient to counteract a profibrotic environment and that the further increase in EETs caused by sEH inhibition tips the balance, partially countering the profibrotic effects of celecoxib in the CCl4 background. In the oxylipin, as well as the mRNA expression and histologic analyses, small sample sizes might be responsible for trends not reaching significance in this study. Given the variability of fibrotic and inflammatory responses in chronic studies, small group sizes can limit the ability to detect differences between groups.
Because of the differences in animal models and the dose of celecoxib used, the effectiveness of COX-2 inhibition as a treatment of liver fibrosis is questionable. In the current study, we have an indication that oxidized fatty acids other than COX-2 metabolites of arachidonic acid may be involved in the profibrotic properties of celecoxib in a CCl4 mouse model of liver fibrosis. Given their effectiveness in modulating the EETs, as well as their antifibrotic effects in the liver and other tissues, sEHIs are promising tools for the study of the role of these oxylipids in hepatic fibrosis and damage.
Authorship Contributions
Participated in research design: Harris, Kodani, Hammock.
Conducted experiments: Harris, Rand.
Contributed new reagents or analytic tools: Rand, Yang, Hwang.
Performed data analysis: Harris, Imai.
Wrote or contributed to the writing of the manuscript: Harris, Imai, Hammock.
Footnotes
- Received January 18, 2018.
- Accepted May 18, 2018.
Primary laboratory of origin: Hammock.
This work was funded by the National Institutes of Health (NIH) National Institute of Environment Health Sciences [Grant R37 ES02710]; the National Institute of Environment Health Sciences Superfund Research Program [Grant P42 ES04699]; the National Institute of Diabetes and Digestive and Kidney Diseases [Grant R01 DK103616]; the NIH West Coast Comprehensive Metabolomics Resources Core [Grant U24 DK097154]; the National Institutes of Health Training Program [Grants T32 HL86350, T32 CA108459]; the Institute of General Medical Sciences Pharmacology Training Program [T32 GM099608]; and the American Association for Cancer research [Grant 16-40-18-RAND].
↵This article has supplemental material available at molpharm.aspetjournals.org.
Abbreviations
- APAP
- acetaminophen
- BDL
- bile duct ligation
- CCl4
- carbon tetrachloride
- COX
- cyclooxygenase
- EET
- epoxyeicosatrienoic acid
- HSC
- hepatic stellate cell
- IHC
- immunohistochemistry
- LC-MS/MS
- liquid chromatography-tandem mass spectrometry
- LOX
- lipoxygenase
- LXA4
- lipoxin A4
- MMP
- matrix metalloprotease
- NSAID
- nonsteroidal anti-inflammatory drug
- P450
- cytochrome P450
- PGD2
- prostaglandin D2
- PGE2
- prostaglandin E2
- PGJ2
- prostaglandin J2
- PTUPB
- 4-(5-phenyl-3-{3-[3-(4-trifluoromethylphenyl)-ureido]-propyl}-pyrazol-1-yl)-benzenesulfonamide
- sEH
- soluble epoxide hydrolase
- sEHI
- soluble epoxide hydrolase inhibitor
- SMA
- smooth muscle actin
- TAA
- thioacetamide
- TIMP
- tissue inhibitor of metalloproteases
- TGF
- transforming growth factor
- TPPU
- 1-trifluoromethoxyphenyl-3-(1-propionylpiperidin-4-yl) urea
- Copyright © 2018 by The American Society for Pharmacology and Experimental Therapeutics