Ursodeoxycholic acid protects against intestinal barrier breakdown by promoting enterocyte migration via EGFR- and COX-2-dependent mechanisms

Abstract

The intestinal barrier is often disrupted in disease states, and intestinal barrier failure leads to sepsis. Ursodeoxycholic acid (UDCA) is a bile acid that may protect the intestinal barrier. We hypothesized that UDCA would protect the intestinal epithelium in injury models. To test this hypothesis, we utilized an in vitro wound-healing assay and a mouse model of intestinal barrier injury. We found that UDCA stimulates intestinal epithelial cell migration in vitro, and this migration was blocked by inhibition of cyclooxygenase 2 (COX-2), epidermal growth factor receptor (EGFR), or ERK. Furthermore, UDCA stimulated both COX-2 induction and EGFR phosphorylation. In vivo UDCA protected the intestinal barrier from LPS-induced injury as measured by FITC dextran leakage into the serum. Using 5-bromo-2′-deoxyuridine and 5-ethynyl-2′-deoxyuridine injections, we found that UDCA stimulated intestinal epithelial cell migration in these animals. These effects were blocked with either administration of Rofecoxib, a COX-2 inhibitor, or in EGFR-dominant negative Velvet mice, wherein UDCA had no effect on LPS-induced injury. Finally, we found increased COX-2 and phosphorylated ERK levels in LPS animals also treated with UDCA. Taken together, these data suggest that UDCA can stimulate intestinal epithelial cell migration and protect against acute intestinal injury via an EGFR- and COX-2-dependent mechanism. UDCA may be an effective treatment to prevent the early onset of gut-origin sepsis.

NEW & NOTEWORTHY In this study, we show that the secondary bile acid ursodeoxycholic acid stimulates intestinal epithelial cell migration after cellular injury and also protects the intestinal barrier in an acute rodent injury model, neither of which has been previously reported. These effects are dependent on epidermal growth factor receptor activation and downstream cyclooxygenase 2 upregulation in the small intestine. This provides a potential treatment for acute, gut-origin sepsis as seen in diseases such as necrotizing enterocolitis.

INTRODUCTION

Intestinal barrier breakdown is a critical turning point in the pathogenesis of sepsis, resulting in disruption of nutrient absorption, translocation of intraluminal bacteria, and propagation of a pathological host response (29). Under pathological conditions, cell death overcomes mechanisms of repair, such as migration and proliferation. This allows for translocation of bacteria and toxins through an impaired intestinal barrier, exacerbating systemic inflammation and leading to multisystem organ failure and sepsis. A potential mechanism protecting against this injury is the activity of bile acid. Bile acids are cholesterol-based molecules synthesized in the liver and secreted into the gut to aid in lipid digestion. They also play a large role in mediating intestinal epithelial cell signaling (13, 33) and may play a role in maintaining the intestinal barrier.

Bile acids are found in numerous forms. Primary bile acids, cholic and chenodeoxycholic acid, are secreted from the liver, where they can be transformed into secondary bile acids, deoxycholic and lithocholic acid, by intraluminal bacteria. These secondary bile acids are more hydrophobic and cytotoxic than primary bile acids (51), which may influence their effects on the intestine. For example, high levels of deoxycholic acid increase ileal damage in experimental necrotizing enterocolitis (21). On the other hand, decreased concentrations of secondary bile acids have been found in feces of patients with inflammatory bowel disease (IBD), whereas the severity of experimental colitis has been associated with hydrophobicity of fecal bile acids (15, 33, 51). More hydrophilic bile acids, including ursodeoxycholic acid (UDCA), exhibit anti-inflammatory effects on epithelial cells in vitro (3, 33, 43, 58). UDCA is a unique, hydrophilic secondary bile acid with known antiapoptotic effects (2, 33) that may aid the colonic intestinal barrier in Crohn’s disease models (5). Because the circulating bile acid pool changes in healthy vs. disease states, this suggests that bile acids may play a key role in intestinal barrier integrity.

UDCA, while making up only 3% of the human bile acid pool (38), has been used in the treatment of various cholestatic diseases in humans, including cholestasis (46), cholelithiasis (36), cholangiopathy (9), primary sclerosis cholangitis (34), primary biliary cirrhosis (8), and chronic liver disease (20). It has also been investigated for the treatment of intestinal inflammatory disorders, such as IBD, ileitis, and colon cancer (5, 16, 25, 35, 40, 57). In addition to its antiapoptotic effects, UDCA can stabilize plasma membranes, reduce bile acid cytotoxicity, change the expression of bile acid transporters, and modulate immunological properties associated with autoimmune disorders (2, 38, 45, 46).

The wide variety of beneficial effects of UDCA may be related to its multiple mechanisms of action. UDCA has been shown to interact with cyclooxygenase 2 (COX-2), epidermal growth factor receptor (EGFR), and their downstream mediators in some hepatocyte and enterocyte signaling pathways (10, 25, 28, 42, 46, 55, 56). In addition, COX-2 and EGFR seem to play a role in enterocyte migration (19, 26, 39, 41, 52). In this study, we sought to evaluate the mechanism of UDCA-induced enterocyte restitution after injury and determine the roles of COX-2 and EGFR in this effect. We hypothesized that UDCA promotes enterocyte migration via a COX-2- and EGFR-dependent mechanism and will attenuate intestinal barrier breakdown during experimental peritonitis in vivo.

MATERIALS AND METHODS

Reagents, antibodies, and assays.

Reagents were purchased from the following suppliers: UDCA, COX-2 inhibitor (Rofecoxib), EGFR inhibitor (AG1478), β-actin antibody, LPS from Escherichia coli O127:B8, FITC-dextran 4000, 5-ethynyl-2′-deoxyuridine (EdU), and 5-bromo-2′-deoxyuridine (BrdU) were all from Sigma-Aldrich (St. Louis, MO); MEK/ERK inhibitor (PD98059), phosphatidylinositol 3-kinase (PI3K) inhibitor (LY294002), and AKT inhibitor IV were from EMD Millipore (Billerica, MA); Src inhibitor (PP2), EP2 inhibitor (PF-04418948), and EP1 inhibitor (ONO-8711) were from Cayman Chemical (Ann Arbor, MI); recombinant human epidermal growth factor (EGF) was from PeproTech (Rocky Hill, NJ); transarterial chemoembolization (TACE) inhibitor (TAPI-1) and COX-2 antibody M-19 were from Santa Cruz Biotechnology (Dallas, TX); EP4 antagonist (GW627368) was from Tocris Bioscience (Bristol, UK); MEK/ERK inhibitor U0126 and COX-1 antibody were from Cell Signaling Technologies (Burlingame, CA). Secondary antibodies for Western blot (donkey anti-goat and donkey anti-mouse) were from Li-Cor Biosciences (Lincoln, NE). Secondary antibodies for immunostaining and ProLong Diamond Antifade Mountant with DAPI were from ThermoFisher Scientific (Canoga Park, CA). Dulbecco’s modified eagle medium (DMEM) and RPMI 1640 medium were purchased from Corning (Manassas, VA).

Cell culture.

Rat intestinal epithelial cells (IEC-6) were purchased from American Type Culture Collection (Manassas, VA) and were maintained in Dulbecco’s modified Eagle medium (DMEM) supplemented with heat-inactivated 10% FBS, 1% penicillin-streptomycin, and 0.1% insulin-transferrin-selenium (ITS, Corning). Cells were serum starved with DMEM supplemented with 0.5% FBS and 1% penicillin-streptomycin for 24 h in preparation for experiments. IEC-6 cells were maintained in incubators at 37°C and 5% CO₂-95% air atmosphere. Young adult mouse colon (YAMC) cells and mouse small intestine epithelial cells (MSIE) were maintained at 33°C in RPMI 1640 growth medium supplemented with 5% FBS, 1% penicillin-streptomycin, and IFN-γ (5 IU/ml), as previously described (13). Cells were shifted to 37°C without IFN in 0.5% FBS for 24 h before experiments. Experiments were done within 20 passages of cells.

Cell proliferation assays.

Crystal violet assay was performed as previously described (13). Briefly, IEC-6 cells were plated at 10,000–20,000 cells per well in 12-well plates, allowed to grow in complete media for 24 h, and then serum starved for 24 h. Cells were treated, stained with 0.1% crystal violet, and extracted with 10% aqueous acetic acid solution. A microplate reader (Model 680; Bio-Rad, Hercules, CA) was used to measure absorbance at 570 nm. To confirm crystal violet results, we further assessed proliferation with EdU incorporation using the Click-iT assay kit (Life Technologies, Eugene, OR), as previously described (13).

Cell migration assay.

Circular wounds were made in confluent cell monolayers that had been serum starved for 24 h. Wounds were made using a rotating silicone probe and were ~1 mm² in size, as previously described (11). Cells were treated with or without UDCA and various inhibitors vs. vehicle controls. Images of wounds were obtained at 0 and 6 or 16 h. Wound area was measured using FIJI ImageJ processing and analysis software (49), and wound closure was compared with control.

Cell survival assay.

Cells were plated in 96-well plates at 20,000 cells/well, serum starved for 48 h, and treated with increasing doses of UDCA. CellTiter-Blue Cell Viability Assay (Promega, Madison, WI) was performed per the manufacturer’s protocol.

Western blot analysis.

Cells or intestinal tissue were extracted with ice-cold cell lysis buffer (150 mM NaCl, 50 mM Tris, 1 mM EDTA, 1 mM EGTD, 1% Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate) with 1:100 protease inhibitor (ThermoFisher Scientific, Waltham, MA) and phosphatase inhibitor 2 and 3 (Sigma-Aldrich). SDS-PAGE was used to separate solubilized protein. Gels were transferred to nitrocellulose membranes, blocked in 5% BSA in PBS, incubated in primary antibody, washed with PBS, and incubated with secondary antibody. After being washed, Li-Cor quantitative infrared Odyssey Imaging System was used for membrane development and band density analysis (Image Studio Lite 4.0, Li-Cor).

siRNA transfection.

IEC-6 cells were plated into six-well dishes at 250,000 cells/well for 24 h. Cells were transfected with 10 nM nontargeting siRNA (NT) or siRNA specific to COX-2 or EGFR, purchased from Dharmacon (ON-TARGETplus SMART pool; Lafayette, CO) using Lipofectamine (Life Technologies) in Opti-Mem for 24 h. Cells were then serum starved for an additional 24 h, treated, and subjected to our migration protocol. Parallel cells were transfected and probed for total protein knockdown.

In vivo peritonitis model.

All procedures were designed in adherence to the National Institutes of Health Guidelines on Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee at Children’s Hospital Los Angeles. We used 6- to 8-wk-old male and female C57BL/6 wild-type mice or Velvet mice (EGFR-dominant negative mutant) (14) in all experiments. Each experiment was done with littermate controls. Twenty-four hours before death, mice were injected intraperitoneally with 50 mg/kg BrdU in normal saline (0.9% NaCl, NS). Mice were orally gavaged with 200 µl NS containing 22 mg/ml (175 mg/kg) FITC-dextran with or without 100 mg/kg UDCA and injected with 200 µl NS with or without 30 mg/kg LPS intraperitoneally 16 h before death. An additional group of mice was given Rofecoxib 20 mg/kg in 1% Triton X-100 solution intraperitoneally at the time of NS or LPS injection. EdU (120 µl of 10 µM) was administered intraperitoneally 2 h before death. Blood samples (400 µl) were obtained by cardiac puncture and combined with 25 µl of 0.5M EDTA. Samples were prepared by centrifugation at 10,000 g for 10 min twice. FITC dextran levels were determined by fluorimetry of 50 µl of serum and standardized to animals that did not receive LPS or UDCA. Two centimeters of distal terminal ileum were obtained for histological evaluation and immunofluorescent staining.

Immunofluorescence.

Paraffin-embedded sections (7 µm) of terminal ileum were deparaffinized in xylene and rehydrated in ethanol. Sections were blocked with 5% goat serum (Vector Laboratories, Burlingame, CA) and stained with Click-iT EdU Assay Kits (ThermoFisher). EdU and BrdU stains were performed according to the manufacturer’s protocols. Alternatively, sections were incubated with primary rabbit antibody to either COX-2 or phosphorylated ERK (Cell Signaling Technologies) at 4°C overnight. These sections were washed and reincubated with secondary goat anti-rabbit (Cell Signaling Technologies) for 1 h. All slides were mounted with ProLong Diamond Antifade Mountant with DAPI (Life Technologies) and imaged on an upright Leica DM4000B LED microscope using Leica Application Suite Advanced Fluorescence software (Jena, Germany). Positive cells for stains were quantified by a blinded observer evaluating a minimum of 18 villi from at least 6 fields of view.

In vivo migration and proliferation assay.

Mice were injected intraperitoneally with BrdU 24 h before death and EdU 2 h before death. Migration was measured as the distance between the highest BrdU cell and highest EdU cell in each hemivillus divided by time between injections, as previously described (4). Enterocyte proliferation was measured by quantification of EdU-labeled cells per DAPI-labeled cell per villus (47, 48). At least 18 villi were analyzed per animal.

Statistics.

Data were analyzed using either ANOVA or Student’s t-test with Bonferroni correction, where appropriate, using GraphPad Prism 6.0. Data are presented as means ± SE unless otherwise noted. A P value of ≤0.05 was considered significant.

RESULTS

UDCA induces intestinal cell migration while decreasing proliferation in intestinal epithelial cells.

To test whether UDCA exerts a protective effect on the intestinal epithelium, we evaluated two key mechanisms of intestinal homeostasis and wound repair, cell migration and proliferation. To measure migration, circular wounds were created in confluent monolayers of serum-starved rat IEC-6 cells (18). Cells were treated with 200 µM UDCA, and wound area was assessed at 0, 6, and 16 h postinjury. Migration was calculated as the percentage of wound closure relative to control. UDCA significantly increased wound closure relative to control at both 6 and 16 h (20.2 ± 2.6% and 15.7 ± 2.1%, respectively; P < 0.001, n ≥ 6 for each, Fig. 1A). A dose-response experiment of UDCA-induced migration suggested that 200 µM yielded maximal results over 6 h (P < 0.05, n = 3, Fig. 1B). Although a dose of 200 µM UDCA has been used before (57), this is greater than the concentration seen in the intestine under normal physiological conditions (22). Thus we tested the toxicity of UDCA on IEC-6 cells using a modified MTS-based assay. IEC-6 cells were treated with increasing doses of UDCA for 24 h and subjected to the CellTiter-Blue Cell Viability Assay. Cell death was not noted at 200 µM but was seen at concentrations ≥500 µM (P < 0.01, n ≥ 5 for each, Fig. 1C). As a result of these data, we used 200 µM UDCA for the remainder of the experiments.

Fig. 1.

Ursodeoxycholic acid (UDCA) induces enterocyte migration independent of proliferation and cell death. A: after 24 h of serum starvation, confluent intestinal epithelial cells (IEC-6) cells were wounded and treated with 200 µM UDCA. Wound closure was assessed at 6 and 16 h to measure cell migration. B: similar IEC-6 cells were wounded and treated with increasing concentrations of UDCA for 6 h. C: confluent IEC-6 cells were treated for 24 h with increasing doses of UDCA. Cell viability was measured using CellTiter-Blue Cell Viability Assay. D: subconfluent IEC-6 cells were then serum starved for 24 h and treated with varying doses of UDCA. Subconfluent cells were required to allow proliferation measurements. E: cell number at 24 h was measured by crystal violet staining and confirmed using 5-ethynyl-2′-deoxyuridine (EdU) incorporation. F: to confirm that these effects were not IEC-6 cell line specific, migration experiments were repeated using mouse small intestine epithelial cells (MSIE) or young adult mouse colon (YAMC) cells treated with 200 µM UDCA. G: proliferation experiments using the crystal violet method were repeated in MSIE cells. n ≥ 3 for each, *P < 0.05.

To determine the role of enterocyte proliferation, subconfluent, serum-starved IEC-6 cells were treated with 200 µM UDCA for 24 h, fixed, and stained with 0.1% crystal violet. Absorbance was measured to determine cell number. UDCA decreased cell number at 24 h to 79.8 ± 2.9% (P < 0.001, n = 12, Fig. 1D). We confirmed these results using EdU incorporation in IEC-6 cells treated with varying doses of UDCA using the Click-iT assay kit (P < 0.05, n = 3, Fig. 1E). These results, consistent with previous reports (30), suggest that UDCA induces intestinal repair after injury via enterocyte migration, independent of cell proliferation or cell death.

To confirm that these effects were not specific to the IEC-6 cell line, migration experiments were performed with mouse-derived MSIE and YAMC cells, whereas proliferation experiments using the crystal violet method were performed using MSIE cells. Results from these experiments were in agreement with the data from the IEC-6 cells (P < 0.05, n ≥ 3, Fig. 1, F and G).

UDCA-induced enterocyte migration requires EGFR, COX-2, EP receptors, AKT, ERK, and PI3K.

It has been suggested that UDCA can interact with COX-2 and EGFR (10, 25, 28, 55, 56), two molecules implicated in enterocyte migration. To determine the roles of COX-2, EGFR, and their downstream mediators in UDCA-induced enterocyte migration in vitro, confluent IEC-6 cells were wounded and treated with UDCA with and without pharmacological inhibitors. Inhibition of either COX-2 (Rofecoxib, 10 µM) or EGFR (AG1478, 150 nM) blocked UDCA-induced enterocyte migration (n ≥ 3, Fig. 2, A and B). EGF (10 ng/ml) stimulated enterocyte migration, as expected (41). To confirm that these results were not the result of off-target inhibitor effects, we next knocked down either EGFR (Fig. 2C) or COX-2 (Fig. 2D) using specific siRNA and found a similar blockade of UDCA-induced migration (P < 0.05, n ≥ 3 for all). We achieved ~75% and 65% reduction in protein levels of EGFR and COX-2, respectively (Fig. 2E).

Fig. 2.

Ursodeoxycholic acid (UDCA)-induced migration requires cyclooxygenase 2 (COX-2), epidermal growth factor receptor (EGFR), EP receptors, ERK, AKT, and phosphatidylinositol 3-kinase (PI3K). Cells were treated at the time of wounding with pharmacological inhibitors to COX-2 (Rofecoxib) (A) or EGFR (AG1478) (B) or with specific siRNA to EGFR (C) or COX-2 (D), with or without UDCA vs. nonspecific siRNA (NT). E: Western blots of parallel cells transfected with siRNA to either EGFR or COX-2 were probed to assess protein knockdown. EP receptor inhibitors (EP1, 2, 4 using ONO-8711, PF-04418948, or GW627368, respectively) (F) and inhibitors of MEK/ERK (PD98059), AKT (AKT inhibitor IV), or PI3K (LY294002) (G) were also used with or without UDCA (labeled ERKi, AKTi, PI3Ki). H: second MEK/ERK inhibitor, U0126, was used to confirm results from G. Migration was measured at 6 h for all experiments. n ≥ 3 for each, *P < 0.01.

COX-2 is the rate-limiting enzyme in the production of prostaglandin E2 (PGE2). PGE2, in tune, acts on four EP receptors (EP1–4) in the intestine, of which EP1, 2, and 4 appear to be the most important (12, 50). To determine the role of these EP receptors on UDCA-induced migration, IEC-6 cells were treated with specific pharmacological EP receptor inhibitors. UDCA-induced cell migration was attenuated with inhibition of EP1 (ONO-8711, 10 µM), EP2 (PF-04418948, 10 µM), or EP4 (GW627368, 10 µM) (n ≥ 4 for each, Fig. 2F). This suggests that UDCA-induced enterocyte migration is dependent on COX-2-mediated PGE2 release and that signaling through multiple EP receptors is required for this effect.

COX-2 and EGFR share many of the same intracellular mediators, including ERK, AKT, and PI3K (37). Thus, pharmacological inhibitors to MEK/ERK (PD98059, 20µM), AKT (AKT inhibitor IV, 1µM), or PI3K (LY294002, 10µM) were used to treat cells subjected to our wound-healing assay. Each of these inhibitors blocked UDCA-induced migration (n ≥ 3 for each, Fig. 2G). Additionally, a second MEK/ERK inhibitor (U0126, 10µM) was added to ensure no off-target effects (n = 3, Fig. 2H). Taken together, these data suggest a role for ERFR, COX-2, and their downstream mediators in UDCA-induced epithelial cell migration.

UDCA induces COX-2 expression via EGFR independent of ERK.

To further test the importance of COX-2 and EGFR in UDCA-induced enterocyte migration, IEC-6 cells were grown to confluence, serum starved for 24 h, and treated with 200 µM UDCA for increasing times. COX-2 expression was measured by Western blot. UDCA significantly increased COX-2 protein expression by 3 h (1.94 ± 0.16 vs. control, P < 0.01), which was sustained at 12 h (Fig. 3A). We tested multiple doses of UDCA at 3 h and found a significant increase in COX-2 expression at 200 µM and higher (n = 3, P < 0.05, Fig. 3B); however, we noted some increased cellular toxicity in IEC-6 cells exposed to 400 µM for 24 h, consistent with our data from Fig. 1C.

Fig. 3.

Ursodeoxycholic acid (UDCA) induces cyclooxygenase 2 (COX-2) expression via an epidermal growth factor receptor (EGFR)-dependent mechanism. Confluent, serum-starved (IEC-6) cells were treated with 200 µM UDCA for increasing times (A) or increasing doses of UDCA for 3 h (B). COX-2 levels were assessed by Western blot of cell lysates. Cells were pretreated for 1 h with the EGFR inhibitor AG1478 (150 nM) (EGFRi) (C), transfected with siRNA against EGFR (D), or pretreated with the MEK/ERK inhibitor PD98059 (20 µM) (E), the phosphatidylinositol 3-kinase (PI3K) inhibitor LY294002 (10 µM) (F), or the ADAM 17 inhibitor transarterial chemoembolization (TACE) inhibitor (TAPI1) (1 µM) (G) before 200 µM UDCA treatment. NT, nontargeting siRNA control. COX-2 expression was measured at 12 h. H: COX-2 expression was also measured 3 h after UDCA treatment when pretreated with AG1478 or PD98059. For all panels, representative blots are shown above pooled, densitometry data. n ≥ 3 for each, *P < 0.05.

To determine whether UDCA-induced COX-2 upregulation occurs via an EGFR-dependent mechanism, cells were pretreated with the EGFR inhibitor AG1478 before UDCA administration. EGFR inhibition attenuated COX-2 induction by UDCA at 12 h (n ≥ 7, Fig. 3C). EGF administration increased COX-2 protein expression, as expected. Similar findings were also seen with cells transfected with siRNA to specific to EGFR (n = 5, Fig. 3D). Cells were then pretreated with inhibitors to the downstream EGFR mediators ERK and PI3K. Pretreatment with the MEK/ERK inhibitor (n = 7) or the PI3K inhibitor (n = 4) did not affect UDCA-induced COX-2 expression (Fig. 3, E and F), suggesting that these molecules are downstream of COX-2 activation. Because EGFR may be activated by release of soluble EGF through matrix metalloproteinases (MMPs), cells were then pretreated with the ADAM17 MMP inhibitor TAPI1. This treatment also did not alter UDCA-induced COX-2 induction (n = 8, Fig. 3G). These data suggest that UDCA induces COX-2 via an EGFR-dependent, ADAM17-independent, pathway. Because UDCA increases migration at 6 and 16 h, we also tested the effects of EGFR and ERK inhibition on COX-2 activation at 3 h using the same methods. EGFR, but not MEK/ERK inhibitor, blocked UDCA-induced COX-2 induction even at 3 h (n ≥ 5, Fig. 3H).

Interestingly, Fig. 2D showed an increase in baseline migration in cells treated with siRNA against COX-2, whereas blockade of UDCA-induced migration remained. One possible explanation for this would be a compensatory change in COX-1 levels in the setting of COX-2 knockdown. IEC-6 cells were transfected with either siRNA to COX-2 or treated with Rofecoxib, with or without UDCA. Lysates were then probed for COX-1. We found no change in COX-1 levels in any treatment group (n ≥ 4, Fig. 4, A and B).

Fig. 4.

Ursodeoxycholic acid (UDCA) induces epidermal growth factor receptor (EGFR) phosphorylation followed by downstream cyclooxygenase 2 (COX-2) and ERK activation independent of COX-1. Confluent, serum-starved cells were treated with siRNA against COX-2 (A) or Rofecoxib (Rof) (B) followed by UDCA treatment for 3 h. Cell lysates were probed for COX-1 levels. Similar cells were treated with 200 µM UDCA for increasing times (C). Cell lysates were probed for p-EGFR (Tyr 1068). Cells were pretreated for 15 min with Rofecoxib (10 µM) (D) or the MEK/ERK inhibitor PD98059 (ERKi) (E) before stimulation with UDCA for 5 min and probed for p-EGFR with total EGFR used as a loading control. Cells pretreated with the EGFR inhibitor AG1478 (F) or Rofecoxib (G) were stimulated with UDCA for 15 min; the resultant lysates were probed for p-ERK. n ≥ 5 for each, *P < 0.05. C, control; NT, nontargeting siRNA control.

To further assess EGFR, COX-2, and ERK signaling in the UDCA-treated cells, cells were treated with 200 µM UDCA for increasing times. Cell lysates were probed for phosphorylated EGFR (p-EGFR). Phosphorylation of EGFR was maximum at 5 min of UDCA treatment (57% above control, n = 9, P < 0.05, Fig. 4C). Pretreatment with either the COX-2 (n = 7) or MEK/ERK inhibitor (n = 4) did not alter UDCA-induced phosphorylation of EGFR (Fig. 4, D and E). Cells were then treated with UDCA+/− EGFR or COX-2 inhibition and probed for phosphorylation of ERK (p-ERK). Interestingly, although EGFR inhibition blocked ERK activation (n = 5, Fig. 4F), COX-2 inhibition had no effect (P < 0.05, n = 5, Fig. 4G). Taken together, these results suggest that UDCA stimulates phosphorylation of EGFR followed by COX-2 induction and separate, but equally required, activation of ERK without COX-1 involvement.

UDCA increases enterocyte migration and protects against intestinal barrier breakdown in experimental peritonitis.

The in vitro effect of UDCA on cell migration is somewhat modest. However, even small migratory changes may be the difference between an intact intestine and barrier breakdown. To test the effect of UDCA in vivo, we utilized a mouse model of experimental peritonitis. Here, 6-wk-old C57BL/6 wild-type mice were treated with intraperitoneal saline or LPS (30 mg/kg) injection to induce intestinal injury. Mice were orally gavaged with FITC-dextran at the induction of injury with or without 100 mg/kg UDCA. After 16 h, mice were euthanized, and intestine and serum were harvested. We measured intestinal barrier breakdown by FITC-dextran fluorescence in serum. Animals injected with LPS had a 9.7 ± 2.6-fold increase in FITC-dextran detection vs. control (P < 0.02, n = 8), as expected. UDCA treatment significantly attenuated LPS-induced intestinal barrier breakdown to 2.7 ± 0.6-fold (P < 0.03 vs. LPS alone, n = 8, Fig. 5A). We measured villus height on histological ileal sections and found that mice injected with LPS had a reduced villus height compared with saline controls (50.3 ± 1.6 vs. 62.4 ± 1.3 cells/villus, P < 0.01). Treatment with UDCA attenuated the loss of villus height seen with LPS injection (59.6 ± 2.6 cells/villus, n = 6, P < 0.05, Fig. 5B).

Fig. 5.

Ursodeoxycholic acid (UDCA) protects against intestinal barrier breakdown in experimental peritonitis and increases intestinal epithelial cell migration. A: animals were given LPS or saline intraperitoneally and gavaged with FITC-dextran with or without UDCA (100 mg/kg). After 16 h, mice were euthanized, and FITC was measured in the serum (harvested by cardiac puncture). B: histological sections of terminal ileum were hematoxylin and eosin stained and used to measure villus height. At least 15 villi per animal were measured. C: animals were given intraperitoneal 5-bromo-2′-deoxyuridine (BrdU) 24 h and 5-ethynyl-2′-deoxyuridine (EdU) 2 h from the end of the experiment. Histological terminal ileal slides were stained for EdU and DAPI. Total EdU cells/total DAPI cells were counted to assess proliferation. D: migration was assessed on histological ileal sections by measuring the difference between the highest BrdU- and EdU-positive cells divided by the time between injections. E: representative EdU/BrdU-stained segments of terminal ileum are shown (line = 100 µm). n ≥ 4 animals for each, *P < 0.05.

To begin to understand how UDCA may protect the intestine from LPS-induced barrier injury, we evaluated the roles of enterocyte migration and proliferation. We injected mice intraperitoneally with BrdU 24 h and EdU 2 h before death to pulse label the intestinal epithelium and evaluated enterocyte proliferation by quantification of EdU labeled cells divided by DAPI-labeled cells per villus. Enterocyte proliferation did not change with UDCA treatment in LPS-injected mice (n = 6, Fig. 5C). Enterocyte migration was measured by calculating the distance between the highest BrdU-labeled cell and the highest EdU-labeled cell within each villus divided by the time between injections, as previously reported (4). LPS injection decreased cell migration, which was rescued by UDCA treatment (2.4 ± 0.2 vs. 3.3 ± 0.3 µm/h, P < 0.01, n = 4, Fig. 5, D and E). These results suggest that enterocyte migration, but not proliferation, may contribute to the protective effect UDCA has against intestinal barrier breakdown in vivo.

UDCA induces enterocyte migration in vivo via COX-2- and EGFR-dependent mechanisms.

To evaluate the role of COX-2 and EGFR in the effects of UDCA in vivo, histological sections of terminal ileum were stained for COX-2 and phosphorylated ERK (p-ERK), a downstream EGFR target, using immunofluorescence. Mice treated with UDCA showed an increase in epithelial COX-2 expression, whereas those treated with LPS alone showed COX-2 expression predominantly in the subepithelial compartment of the intestinal villi but not in the epithelium. When animals were treated with UDCA and LPS injection, epithelial staining was seen, similar to UDCA alone. These data suggest that epithelial COX-2 expression is increased by UDCA treatment, which may play a role in migration, whereas LPS induces subepithelial COX-2 expression, likely an inflammatory response. Expression of p-ERK was minimal in mice injected with LPS but increased in animals treated with UDCA, consistent with our in vitro data (Fig. 6).

Fig. 6.

Ursodeoxycholic acid (UDCA) stimulates epithelial cyclooxygenase 2 (COX-2) and p-ERK expression not seen with LPS injection alone. Histological sections from animals treated with or without UDCA and LPS were stained using immunofluorescence for COX-2 expression and p-ERK expression. LPS increases subepithelial COX-2 expression (arrow) but does not cause epithelial COX-2 expression (arrow head). Representative slides are shown. n ≥ 3 animals for each condition.

To further elucidate the role of COX-2 in UDCA-induced barrier protection, mice injected with LPS were treated with or without 20 mg/kg Rofecoxib (COX-2 inhibitor, administered intraperitoneally) and 100 mg/kg UDCA. As described above, UDCA protected from LPS-induced intestinal barrier breakdown. In mice treated with Rofecoxib, the protective effect of UDCA on LPS-induced intestinal injury was lost (n = 10, Fig. 7A). We also evaluated enterocyte migration in these animals and found that UDCA did not stimulate enterocyte migration in the presence of Rofecoxib in LPS-injected animals (n = 4, Fig. 7, B and C).

Fig. 7.

Cyclooxygenase 2 (COX-2) and epidermal growth factor receptor (EGFR) play a role in ursodeoxycholic acid (UDCA)-induced enterocyte barrier protection and migration in vivo. A: animals were treated similarly to those in Fig. 4 with the addition of the COX-2 inhibitor Rofecoxib (Rof) (20 mg/kg ip). FITC levels were measured in the serum obtained via cardiac puncture. B: animals were injected with 5-bromo-2′-deoxyuridine (BrdU) and 5-ethynyl-2′-deoxyuridine (EdU) to assess migration as in Fig. 4D. C: representative BrdU and EdU stains are shown. D: EGFR-dominant negative Velvet mice were treated with LPS injection with or without UDCA. FITC-dextran was assessed in the serum. E: migration was assessed with BrdU and EdU staining. F: representative EdU/BrdU sections are shown. n ≥ 4 animals for all experiments, *P < 0.05.

To further assess the role of EGFR in the beneficial effects of UDCA, we used EGFR-dominant negative mutant (Velvet) mice (14). Velvet mice were injected with saline or LPS (30 mg/kg) and orally gavaged with FITC-dextran with or without 100 mg/kg UDCA. Intestinal barrier breakdown was not altered by UDCA treatment in Velvet mice injected with LPS (P < 0.02 vs. control, n = 7, Fig. 7D). There was also no difference in enterocyte migration in these animals with or without UDCA following LPS injection (n = 4, Fig. 7, E and F). Immunofluorescence stains of terminal ileal sections for COX-2 and p-ERK on Velvet animals injected with LPS showed no epithelial expression of COX-2 or p-ERK when UDCA was given (Fig. 8). All animal experiments were also analyzed by sex, and we found no sex-related differences in our results. Taken together, these data suggest that UDCA induces enterocyte migration in vivo and protects against intestinal barrier breakdown via COX-2- and EGFR-dependent mechanisms.

Fig. 8.

Ursodeoxycholic acid (UDCA) does not change cyclooxygenase 2 (COX-2) and p-ERK expression in Velvet mice. Histological sections from Velvet animals treated with or without UDCA and LPS were stained using immunofluorescence for COX-2 expression and p-ERK expression. Representative slides are shown. n = 3 animals for each condition.

DISCUSSION

A number of cellular mechanisms are involved in the maintenance of the intestinal barrier, including limiting apoptosis or promoting cell proliferation and migration. Without these, the intestinal barrier may break down, leading to sepsis and even death. The bile acid UDCA is known to have anti-inflammatory, antiapoptotic, and antioxidative properties (2, 33, 45, 46, 58). Interestingly, UDCA is also implicated in decreasing cellular proliferation (30), which might seem paradoxical given its protective function. In acute injury, however, cells must migrate rapidly to close a leaky intestinal barrier, and migration and proliferation are incompatible at the cellular level. A potential role for UDCA in intestinal cell migration has not been previously tested. Our data suggest that UDCA prevents intestinal barrier breakdown and promotes enterocyte migration via a COX-2- and EGFR-dependent mechanism. We show that UDCA induces migration but not proliferation using in vitro and in vivo models. In vitro UDCA-induced migration was attenuated when cells were pretreated with inhibitors to COX-2, EGFR, and the downstream MEK/ERK cascade. UDCA-induced COX-2 protein expression appears to be EGFR dependent but not ERK dependent. ERK phosphorylation downstream is dependent on EGFR but not on COX-2 levels. This suggests that UDCA induces cell migration through EGFR activation, leading to separate COX-2 upregulation and ERK activation. Both COX-2 and ERK are required for UDCA-induced migration but do not appear to lie on a linear pathway. The migratory effect of UDCA, along with the contribution of COX-2, EGFR, and ERK, was also important in vivo in an acute model of experimental peritonitis. Our proposed pathway based on these data is summarized in Fig. 9. UDCA, or molecules within its pathway of activity, may be useful therapeutic targets in the early prevention of gut-origin sepsis.

Fig. 9.

Proposed pathway through which ursodeoxycholic acid (UDCA) promotes intestinal restitution. EGFR, epidermal growth factor receptor; COX, cyclooxygenase.

UDCA is a hydrophilic, secondary bile acid that is used in humans for the treatment of a variety of hepatobiliary disorders, in part because of its antiapoptotic effects. This has led to its investigation in the treatment of intestinal disorders. Data suggest that, in the colon, UDCA attenuates proinflammatory cytokines and decreases inflammation during DSS-induced (57) and TNBS-induced colitis (35). Additionally, taurine-conjugated UDCA inhibits epithelial apoptosis during DSS-induced colitis (33). UDCA has also been shown to decrease DCA-induced apoptosis in colon cancer (HCT 116) cells (25). Although most studies have focused on colon epithelium, there is some, although limited, data on the role of UDCA in the small intestine using indomethacin-induced ileitis (5, 31, 53). Bernardes-Silva et al. (5) found a protective effect associated with decreased intestinal permeability and oxidative stress as measured by intestinal superoxide production. These studies suggest that the beneficial effects of UDCA are due to its antiapoptotic and anti-inflammatory properties. Although UDCA clearly interacts with the inflammatory pathway, its role in the stimulation of epithelial cell migration is a novel role of UDCA that has not yet been described that may contribute to barrier protection in this acute injury model. In our acute LPS-induced model, UDCA decreased intestinal barrier breakdown, attenuated loss of villus height, and induced enterocyte migration, which is dependent on COX-2 upregulation and EGFR pathway activation.

Interestingly, models used to test UDCA to date represent much more long-term models of injury, requiring days to weeks of treatment before study. Data on the role of UDCA treatment in acute small bowel inflammatory damage and in generalized peritonitis are limited. LPS-induced peritonitis differs from the inflammatory response seen with indomethacin ileitis and DSS colitis in its more acute onset with a rapid induction of high levels of proinflammatory cytokines (44). This type of model may be more consistent with diseases such as necrotizing enterocolitis, in which intestinal injury can occur in a matter of hours.

Interestingly, COX-2 and EGFR, along with downstream mediators such as ERK, have been implicated in enterocyte cell signaling and have been shown to interact with UDCA in antiapoptotic pathways previously (2, 6, 7, 12, 19, 23, 24, 26, 37, 39, 41, 52). However, these interactions are clearly complex. In HCT 116 cells, UDCA treatment protected from apoptosis while inhibiting EGFR and ERK phosphorylation (25). In contrast, ERK phosphorylation is increased in IEC-6 cells when treated with UDCA (30). In fact, reports typically suggest that UDCA treatment decreases EGFR activation in colon and liver cells when studied in cancer models (10). This may represent differences in physiology in cancer vs. normal cells or simply differences within tissue types (small intestine vs. colon or liver). Similarly, UDCA blocks COX-2 expression in colon cancer cells in cell culture and in animal cancer models (27, 28), whereas, in this study, we show that UDCA treatment stimulates both EGFR phosphorylation and COX-2 expression in the small intestine. These differential effects within the colon vs. small intestine may also be secondary to bacterial metabolism of UDCA to other bile acid forms, such as lithocholic acid, which normally occurs in the colon (57). Clearly, bile acid metabolites all interact with the intestine differently (32).

The effect of UDCA on COX-2 expression in the small intestine is not well studied, limited to only a few manuscripts investigating the esophagus and duodenum. The COX-2 inhibitor Celecoxib has been shown to reduce duodenal polyp density in inherited gastrointestinal polypoid diseases, whereas UDCA cotreatment seemed to counteract this effect (56), suggesting that UDCA may function through increased COX-2 activity. However, this combination treatment did decrease the growth of colon adenoma cells in vitro (54). The role of cell migration was not assessed in these studies. In esophageal cancer cells, COX-2 induction was decreased by UDCA pretreatment (1), providing further evidence for the differential effects of UDCA in various parts of the gastrointestinal tract, as well as in normal vs. cancer physiology. COX-1 does not seem to be involved in these effects (Fig. 4, A and B), which we tested after identifying increased baseline migration in cells treated with siRNA against COX-2, although UDCA-induced migration was still blocked. This effect was not noted with Rofecoxib, a specific COX-2 inhibitor. This may reflect another compensatory mechanism that exists in cells devoid of COX-2 for days and may explain why this was not seen with the pharmacological inhibitor. Further exploration of this is beyond the scope of this study. Although increased COX-2 protein expression can, in some systems, be a slow process, there is precedent for rapid induction in 3 h or less, as we have shown (17, 37). Clearly, COX-2 plays numerous roles in the intestine, and the sustained activation seen in Fig. 2A may be involved in other cellular processes outside cell migration. However, this early, UDCA-induced, EGFR-dependent activation of COX-2 seems vital in the migratory response of intestinal epithelial cells.

In this study, we find that UDCA stimulates intestinal epithelial cell migration but not proliferation. We also find that UDCA protects from LPS-induced intestinal barrier breakdown. These changes, both in vitro and in vivo, require UDCA-induced activation of EGFR, COX-2, and ERK, as shown pharmacologically and with EGFR-dominant negative mice. The migratory response to UDCA may represent a new role for UDCA in the intestine, in addition to its anti-inflammatory effects. Understanding this key restorative signaling pathway highlights the potential of UDCA as a therapeutic mediator during acute intestinal inflammation. This may help treat disorders of intestinal injury and gut-origin sepsis specifically in diseases targeting the small bowel, such as necrotizing enterocolitis or terminal ileal involvement seen in inflammatory bowel disease.

GRANTS

This work was supported by the Society of University Surgeons Junior Faculty Award (to C. Gayer), Saban Research Career Development Award (to C. Gayer), and National Institute of Diabetes and Digestive and Kidney Diseases Grant R01-DK-095004 (to M. Frey).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.M.G., O.H.E., M.V.N., M.U.M., P.K., and C.P.G. performed experiments; J.M.G. and C.P.G. drafted manuscript; O.H.E., M.V.N., M.R.F., and C.P.G. analyzed data; O.H.E., M.U.M., and C.P.G. prepared figures; M.R.F. and C.P.G. interpreted results of experiments; M.R.F. and C.P.G. edited and revised manuscript; C.P.G. conceived and designed research; C.P.G. approved final version of manuscript.