A Novel Pharmacological Approach to Enhance the Integrity and Accelerate Restitution of the Intestinal Epithelial Barrier

This study describes a novel pharmacological approach to enhance the intestinal epithelial barrier by a group of isoxazole compounds. These compounds decrease epithelial permeability, prevent cytokine-induced barrier disruption, promote epithelial restitution, and could be promising agents for adjuvant IBD therapy.

Abstract

Background

Disruption of the gut barrier is an essential mechanism of inflammatory bowel diseases (IBDs) contributing to the development of mucosal inflammation. A hallmark of barrier disruption is the disassembly of epithelial adherens junctions (AJs) driven by decreased expression of a major AJ protein, E-cadherin. A group of isoxazole compounds, such as E-cadherin-upregulator (ECU) and ML327, were previously shown to stimulate E-cadherin expression in poorly differentiated human cancer cells. This study was designed to examine whether these isoxazole compounds can enhance and protect model intestinal epithelial barriers in vitro.

Methods

The study was conducted using T84, SK-CO15, and HT-29 human colonic epithelial cell monolayers. Disruption of the epithelial barrier was induced by pro-inflammatory cytokines, tumor necrosis factor-α, and interferon-γ. Barrier integrity and epithelial junction assembly was examined using different permeability assays, immunofluorescence labeling, and confocal microscopy. Epithelial restitution was analyzed using a scratch wound healing assay.

Results

E-cadherin-upregulator and ML327 treatment of intestinal epithelial cell monolayers resulted in several barrier-protective effects, including reduced steady-state epithelial permeability, inhibition of cytokine-induced barrier disruption and junction disassembly, and acceleration of epithelial wound healing. Surprisingly, these effects were not due to upregulation of E-cadherin expression but were mediated by multiple mechanisms including inhibition of junction protein endocytosis, attenuation of cytokine-induced apoptosis, and activation of promigratory Src and AKT signaling.

Conclusions

Our data highlight ECU and ML327 as promising compounds for developing new therapeutic strategies to protect the integrity and accelerate the restitution of the intestinal epithelial barrier in IBD and other inflammatory disorders.

INTRODUCTION

Increased permeability of the intestinal epithelial barrier is a common feature of inflammatory bowel diseases (IBDs), being well documented in both Crohn’s disease (CD) and ulcerative colitis (UC) patients.^1–7 Though the exact role of gut barrier leakiness in the pathogenesis of IBD remains a topic of debate, significant experimental and clinical evidence demonstrates that loss of barrier integrity exaggerates mucosal inflammation via excessive activation of intestinal immune cells by gut luminal microbes.^{4, 7, 8} Importantly, increased permeability of the gut barrier is not limited to IBD and has been described in a broad spectrum of intestinal inflammatory conditions including celiac disease, bacterial enteritis, and graft vs host disease.^8–10 One could therefore suggest that developing pharmacological approaches to protect the integrity of the gut barrier would be a valuable therapeutic strategy to decrease morbidity and mortality of patients with IBD and other types of gastrointestinal inflammatory disorders.

Developing drugs that would preserve normal epithelial permeability in inflamed intestinal mucosa is not an easy task given the multiple mechanisms that drive epithelial barrier disassembly.^{8, 11, 12} Permeability of the intestinal epithelial barrier is determined by specialized cellular structures called junctions; these are classified as tight junctions (TJs), adherens junctions (AJs), and desmosomes.^13–15 The structure and function of all junctional complexes is known to be compromised in the inflamed intestinal mucosa of IBD patients.^{3, 11, 16, 17} A further mechanism of gut barrier disruption during active inflammation involves excessive cell death in the epithelium, leading to the formation of mucosal wounds.^{18, 19} The ideal barrier-protective drugs should possess several beneficial features such as stabilization of different junctional complexes, along with the ability to either prevent the formation, or accelerate the healing of mucosal wounds.

Among the multiple mechanisms mediating the disruption of the gut barrier in inflamed mucosa, downregulation of E-cadherin expression appears to be particularly important. E-cadherin is the major adhesive AJ protein that regulates the assembly of other junctional complexes and is indispensable to the formation and maintenance of the intestinal epithelial barrier.^{14, 20} E-cadherin expression is known to be decreased in different intestinal inflammatory disorders, including Crohn’s disease, ulcerative colitis, celiac disease, and irritable bowel syndrome.^{3, 21–24} Because animal studies demonstrate that selective loss of intestinal epithelial E-cadherin is sufficient to trigger gut leakiness and the development of spontaneous colitis, E-cadherin downregulation is likely to play a causal role in epithelial barrier disruption in the inflamed human intestinal mucosa.²⁵ In addition to regulating the assembly of epithelial junctions, E-cadherin has other important barrier-protective and anti-inflammatory functions in the gut, inhibiting epithelial cell apoptosis and accelerating collective cell migration.^26–29 One could therefore predict that the enhancement of E-cadherin expression would exert beneficial anti-inflammatory effects by attenuating disruption and promoting restitution of the intestinal epithelial barrier.

Downregulation of E-cadherin expression is not unique to inflamed intestinal mucosa; it also is one of the most recognized molecular events during the epithelia-to-mesenchymal transition (EMT), driving the progression of many solid tumors.^{30, 31} Due to the crucial role of EMT in tumorigenesis, several attempts have been made to reverse such phenotypic transition of cancer cells via pharmacological restoration of E-cadherin expression.^{32, 33} The most notable example is a small molecule library screen that identified a group of cell-permeable isoxazole compounds upregulating E-cadherin expression in mesenchymal-type colon and lung cancer cells.³⁴ Among these small molecule compounds, 5-(Furan-2-yl)-N-(pyridine-4-yl)butyl)isoxazole-3-carboxamide (herein referred to as E-cadherin upregulator [ECU]) and 1,2-Dihydro-2-oxo-N-[3-[[(5-phenyl-3-isoxazolyl)carbonyl]amino]propyl]-3-pyridinecarboxamide (referred to as ML327) potently increased E-cadherin mRNA and protein levels and inhibited cancer cell growth and invasion.^34–37 Furthermore, ECU was shown to protect kidney epithelial cells from cisplatin-induced toxicity.³⁸ The described features of these isoxazole compounds allow us to hypothesize that ECU and/or ML327 could have barrier-enhancing and barrier-protective effects in intestinal epithelial cell monolayers. The goal of the present study is to test this hypothesis. Our data revealed several novel activities of ECU and ML327, such as decreasing permeability of the normal intestinal epithelial barrier, attenuating cytokine-induced barrier disruption, and promoting epithelial wound healing.

MATERIALS AND METHODS

Antibodies and Other Reagents

The following monoclonal (mAb) and polyclonal (pAb) antibodies were used to probe apical junctions and detect apoptosis-associated molecules and other signaling proteins: anti-FAK, paxillin, E-cadherin, β-catenin, p120 catenin mAbs (BD Biosciences, San Jose, CA); anti-p-AKT, AKT, p-ERK, ERK, Src, PARP, Mcl-1, Bcl-xl, GAPDH mAbs and anti-p-FAK, p-Src, cleaved PARP, cleaved caspase 3, and cleaved caspase 7 pAbs (Cell Signaling, Beverly, MA); anti-P-cadherin and HNF4α mAb (R&D Systems, Minneapolis, MN); anti-claudin-1, claudin-3, claudin-7, p-Paxillin pAbs, anti-claudin-4, and anti-ZO-1 mAbs (Thermo Fisher Scientific, Waltham, MA); anti-occludin and Bcl-2 pAbs (Proteintech, Rosemont, IL); anti-Noxa mAb and anti-Bax-(NT) pAb (Millipore-Sigma, Burlington, MA); anti-Bim/Bad pAb (Enzo Life Science, Farmingdale, NY). Alexa Fluor 488 conjugated donkey antirabbit and Alexa Fluor 555 conjugated donkey antimouse secondary antibodies were obtained from Thermo Fisher. Horseradish peroxidase (HRP)-conjugated goat antirabbit and antimouse secondary antibodies were obtained from Bio-Rad Laboratories (Hercules, CA). E-cadherin upregulator was purchased from Millipore-Sigma. The ML327 was either purchased from MedChem Express (Monmouth Junction, NJ) or provided by Dr. Alex Waterson (Vanderbilt University, Nashville, TN).

Cell Culture

The T84, Colo-320, SW620, and IEC6 cells were obtained from the American Type Culture Collection (Manassas, VA). The SK-CO15 human colonic epithelial cells³⁹ were a gift from Dr. Enrique Rodriguez-Boulan (Weill Medical College, Cornell University, New York, NY). The HT-29 cF8 cells, a well-differentiated clone of HT-29 cells,^{40, 41} were provided by Dr. Judith M. Ball (College of Veterinary and Biomedical Sciences, Texas A&M University, College Station, TX). The HT-29 cF8 and SK-CO15 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS), N-2-hydroxyethylpiperazine-N-ethanesulfonic acid (HEPES), nonessential amino acids, and penicillin-streptomycin antibiotic. The IEC6 cells were cultured in DMEM supplemented with 10% FBS, HEPES, insulin (10 mg/mL), and penicillin-streptomycin antibiotic. The T84 cells were cultured in DMEM/F12 medium supplemented with 5% FBS and penicillin-streptomycin antibiotic. The Colo-320 and SW620 cells were cultured in Roswell Park Memorial Institute medium supplemented with 10% FBS, HEPES, sodium pyruvate, and penicillin-streptomycin antibiotic.

Measurement of Epithelial Barrier Permeability

Transepithelial electrical resistance (TEER) of cultured T84, HT-29 cF8, and SK-CO15 intestinal epithelial cell monolayers was measured using an EVOMX voltohmmeter (World Precision Instruments, Sarasota, FL). Cells were plated on collagen-coated transwell filters (pore size 3 μm, Thermo Fisher). The resistance of cell-free collagen-coated filters was subtracted from each experimental point. A transmonolayer dextran flux assay was performed as previously described.⁴¹ Briefly, intestinal epithelial cell monolayers growing on transwell filters were apically exposed to 1 mg/mL of Fluorescein Isothiocyanate (FITC)-labeled dextran (4000 Da) in HEPES-buffered Hanks’ balanced salt solution (HBSS). After 120 minutes of incubation, HBSS samples were collected from the lower chamber, and FITC fluorescence intensity was measured using a SpectraMax M2 plate reader (Molecular Devises, San Jose, CA) at excitation and emission wavelengths 485 and 544 nm, respectively. After subtracting the fluorescence of dextran-free HBSS, the amount of FITC dextran translocated across the epithelial cell monolayer was calculated based on a calibration curve using Prism 5 software (GraphPad, La Jolla, CA).

Immunofluorescence Labeling and Confocal Microscopy

To visualize the structure of epithelial junctions, cultured colonic epithelial cell monolayers were fixed and permeabilized with 100% methanol at −20ºC. Fixed samples were blocked for 60 minutes in HBSS containing 1% bovine serum albumin, followed by a 60-minute incubation with primary antibodies. Samples were then washed and incubated with Alexa Fluor 488 conjugated donkey antirabbit and Alexa Fluor 555 conjugated donkey antimouse secondary antibodies for 60 minutes, rinsed with blocking buffer, and mounted on slides with ProLong Antifade mounting reagent (Thermo Fisher). Immunolabeled cell monolayers were imaged using a Leica SP8 confocal microscope (Wentzler, Germany). The Alexa Fluor 488 and 555 signals were acquired sequentially in frame-interlace mode to eliminate cross-talk between channels. Images were processed using Adobe Photoshop. The images shown are representative of at least 3 experiments, with multiple images taken per slide.

Immunoblotting Analysis

Epithelial cell monolayers were scraped and homogenized using a Dounce homogenizer in Radio-Immunoprecipitation Assay buffer (20 mM Tris, 50 mM NaCl, 2 mM EDTA, 2 mM EGTA, 1% sodium deoxycholate, 1% Triton X-100 [TX-100], and 0.1% SDS, pH 7.4) containing a protease inhibitor cocktail and phosphatase inhibitor cocktails 2 and 3 (Millipore-Sigma). The obtained total cell lysates were cleared by centrifugation (20 min at 14,000 × g), diluted with 2x SDS sample loading buffer, and boiled. The SDS-polyacrylamide gel electrophoresis was conducted using a standard protocol, with equal amounts of total protein (10 or 20 μg) loaded per lane. The separated proteins were transferred to nitrocellulose membranes, and the membranes were blocked with 5% nonfat milk. The blocked membranes were incubated overnight with primary antibodies and then exposed to HRP-conjugated secondary antibodies for 1 hour. The labeled proteins were visualized using a standard enhanced chemoluminescence solution and x-ray films.

Scratch Wound Assay

Confluent epithelial cell monolayers were mechanically wounded by making a thin scratch wound with a 200-µL pipette tip. The bottom of the well was marked to define the initial position of the wound, and the monolayers were supplied with fresh cell culture medium. The images of a cell-free area at the marked region were acquired at the indicated times after wounding using an inverted bright field microscope equipped with a camera. The percentage of wound closure was calculated using Image J software (NIH, Bethesda, MD). For biochemical experiments, multiple wounds were created in cell monolayers using a Cell Comb Scratch kit (Millipore-Sigma).

RNA Interference

E-cadherin expression was transiently downregulated using gene-specific Dharmacon SmartPool small interfering (si) RNAs (Horizon Discovery, Waterbeach, UK) as previously described.^41–43 Noncoding siRNA duplex 2 was used as a control. Cells were seeded in 6-well plates at approximately 60% confluence and transfected with siRNA, using DharmaFect 1 transfection reagent and Opti-MEM reduced serum medium. The final siRNA concentration in the medium was 50 nM. Cells were utilized for experiments on days 2 to 4 post-transfection.

Statistical Analysis

All data are expressed as means ± standard error (se) from 3 biological replicates. A 2-tailed, unpaired Student t test was used to compare results obtained with 2 experimental groups (vehicle vs compound treatment). P values <0.05 were considered statistically significant.

RESULTS

Isoxazole Compounds Enhance Assembly of the Paracellular Barrier in Model Intestinal Epithelial Cell Monolayers

Initial experiments sought to examine the effects of isoxazole compounds, ECU and ML327, on the barrier properties of model intestinal epithelial cell monolayers. Three well-differentiated human colonic epithelial cell lines, namely T84, SK-CO15, and HT-29 cF8 cells, were selected for this study. Cells were allowed to form confluent monolayers over the course of 5 to 7 days after plating and were treated thereafter with ECU (10 μM), ML327 (10 μM), or vehicle (DMSO). Transepithelial electrical resistance was measured before and at various times after compound addition, whereas transmonolayer flux of FITC-dextran was examined after 72 hours of the drug exposure. Both ECU and ML327 caused rapid (within 24 h) and sustained (up to 72 h) increase of TEER in all tested epithelial cell monolayers, which reflects a decrease in ionic permeability (Fig. 1A, C, E). Furthermore, either ECU or ML327 exposure significantly inhibited transmonolayer flux of FITC-dextran, indicating attenuated epithelial permeability to large molecules (Fig. 1B, D, F). Immunofluorescence labeling and confocal microscopy demonstrated that ECU and ML327 treatment had no effects on AJ and TJ morphology in T84 cell monolayers (Supplementary Fig. 1). The only noticeable change was the disappearance of intracellular vesicular pools of claudin-4 (Supplementary Fig. 1, arrows) and claudin-7 (data not shown) after compound treatment.

FIGURE 1.

Isoxazole compounds decrease the permeability of colonic epithelial cell monolayers. Confluent T84 (A, B), SK-CO15 (C, D), and HT-29cF8 (E, F) cell monolayers were treated with either vehicle, E-cadherin upregulator (10 μM), or ML327 (10 μM). Transepithelial electrical resistance was measured at the indicated times (A, C, E). Transmonolayer flux of FITC-dextran was determined after 72 hours of ECU and ML327 treatment (B, D, F). Data are presented as a mean ± SE (n = 3); *P < 0.05; **P < 0.01; ***P < 0.001.

Given the previously reported ability of ECU and ML327 to stimulate E-cadherin expression,^34–36 we next investigated whether the observed barrier-enhancing effects of these compounds could be explained by increased levels of different AJ and TJ proteins. Surprisingly, immunoblotting analysis of T84 cells treated for 48 hours with either ECU or ML327 did not show significant upregulation of E-cadherin or other essential junctional proteins (Fig. 2A). In fact, expression of claudin-4 was significantly decreased in ECU-exposed cells, which could explain the aforementioned disappearance of the cytoplasmic pool of this TJ protein (Supplementary Fig. 1). Similar results were obtained after analyzing junctional protein expression in ECU and ML327-treated SK-CO15 cell monolayers (data not shown). In contrast to well-differentiated T84 and SK-CO15 epithelial cells, ECU and ML327 stimulated the expression of E-cadherin and P-cadherin in poorly differentiated mesenchymal-type Colo-320 and SW620 colon cancer cells (Fig. 2B, C).

FIGURE 2.

Isoxazole compounds do not upregulate expression of different AJ/TJ proteins in well-differentiated colonic epithelial cells. Well-differentiated T84 colonic epithelial cells (A), poorly differentiated Colo-320 (B), and SW620 (C) colonic epithelial cell lines were treated for 48 hours and 72 hours with either ECU (10 μM), ML327 (10 μM), or vehicle. The expression of epithelial cadherins and other junctional proteins in total cell lysates was determined using immunoblotting analysis.

To further investigate possible relationships between the E-cadherin level and the barrier-enhancing effects of isoxazole compounds, we downregulated E-cadherin expression in intestinal epithelial cells by using small interfering RNAs. The SK-CO15 cells were used due to the superior efficiency of siRNA-mediated gene knockdown in this cell line.^42–44 Exposure of epithelial cells to E-cadherin-specific siRNAs caused a dramatic (more than 80%) decrease in the level of this protein (Fig. 3A). Such E-cadherin knockdown attenuated TEER development and increased dextran flux, thereby indicating delayed assembly of the epithelial barrier (Fig. 3B, C). Interestingly, ECU exposure promoted barrier formation (increased TEER and decreased dextran flux) in SK-CO15 cell monolayers treated with either control or E-cadherin-specific siRNAs (Fig. 3B, C). The magnitude of these barrier-enhancing effects of ECU seems to be higher in E-cadherin-depleted epithelial cells compared with their controls (Fig. 3B, C). Together, our results demonstrate that ECU and ML327 promote the barrier function of well-differentiated human intestinal epithelial cell monolayers via mechanisms independent of stimulating E-cadherin expression.

FIGURE 3.

E-cadherin upregulator strengthens the paracellular barrier in E-cadherin-depleted colonic epithelial cells. The SK-CO15 cells were transfected with either control or E-cadherin-specific siRNAs, and on day 3, post-transfection was treated with either ECU (10 μM) or vehicle. The efficiency of E-cadherin knockdown was determined by immunoblotting (A). Barrier permeability was examined using TEER measurements at the indicated times (B) and FITC dextran flux assay after 48 hours of the compound treatment (C). Data are presented as a mean ± SE (n = 3); ***P < 0.001.

A previous study suggested that ML327 induces transcriptional reprograming of cancer cells by stimulating the expression and activity of an important transcriptional factor, hepatocyte nuclear factor 4 alpha (HNF4α).³⁵ Therefore, we sought to investigate if HNF4α activation contributes to ECU- and ML327-dependent enhancement of model intestinal epithelial barriers. However, ECU exposure downregulated HNF4α, whereas ML327 did not alter the expression of this transcriptional factor in T84 cell monolayers (Supplementary Fig. 2A) Furthermore, two known pharmacological activators of HNF4α, alverine and benfluorex,⁴⁵ had no effect on TEER development and, in fact, significantly increased transepithelial dextran flux, failing to recapitulate the barrier-enhancing activity of ECU at the same experimental conditions (Supplementary Fig. 2B, C). These data argue against the role of HNF4α in mediating the barrier-enhancing effects of isoxazole compounds in intestinal epithelial monolayers.

Isoxazole Compounds Attenuate Cytokine-induced Disruption of Model Intestinal Epithelial Barrier

Next, we sought to investigate if ECU and ML327 can modulate the disruption of the intestinal epithelial barrier caused by pro-inflammatory cytokines, which are known to accumulate in the inflamed intestinal mucosa of IBD patients. We selected a combination of tumor necrosis factor alpha (TNFα) and interferon-gamma (IFNγ) due to their profound epithelial barrier–damaging effects in different epithelia.^46–50 Co-exposure of T84 cell monolayers to TNFα (10 ng/mL) and IFNγ (50 ng/mL) caused a robust barrier disruption manifested by a decrease in TEER (Fig. 4A, C) and an increase in transepithelial dextran flux (Fig. 4B, D). Pretreatment of cell monolayers with either ECU or ML327 markedly attenuated cytokine-induced disruption of the epithelial barrier (Fig. 4A–D). Despite the fact that ECU and ML327 similarly decreased steady-state epithelial permeability (Fig. 1), ECU displayed superior barrier-protective activity in TNFα/IFNγ-treated T84 cell monolayers (Fig. 4). Because pro-inflammatory cytokines disrupt epithelial barriers by triggering junctional disassembly,^{47, 49} we sought to determine if ECU and ML327 protect AJ and/or TJ integrity in cytokine-treated intestinal epithelial cells. Immunofluorescence labeling and confocal microscopy showed a dramatic disruption of the characteristic “chicken wire” labeling pattern of E-cadherin, ZO-1, and claudin-4 after 48 hours of TNFα and IFNγ exposure, indicative of both AJ and TJ disassembly (Fig. 5, arrows). Specifically, ZO-1 signal disappeared from TJ areas, whereas E-cadherin and claudin-4 were translocated from intercellular contacts into the cytoplasm (Fig. 5, arrows). In strong agreement with our permeability data, ECU and ML327 prevented cytokine-induced AJ and TJ disassembly and cytoplasmic accumulation of junctional proteins (Fig. 5, arrowheads). To understand the mechanisms underlying the observed barrier-protective effects of ECU and ML327, we examined the effects of these compounds on cytokine-induced inhibition of junctional protein expression. In agreement with previously published studies,^{49, 51, 52} we observed significant downregulation of several AJC proteins, including E-cadherin, P-cadherin, p120 catenin, ZO-1, occludin, and claudin-7 in TNFα-treated and IFNγ-treated T84 cell monolayers (Supplementary Fig. 3). The ECU and ML327 treatment partially reversed cytokine-induced downregulation of E-cadherin and claudin-7 expression (Supplementary Fig. 3).

FIGURE 4.

Isoxazole compounds prevent cytokine-induced disruption of the model intestinal epithelial barrier. Confluent T84 cell monolayers were pretreated with either vehicle, ECU (A, B), or ML327 (C, D) for 24 hours. Afterward, cells were exposed to a combination of TNFα (10 ng/mL) and IFNγ (50 ng/mL), both with and without the isoxazole compounds. Transepithelial electrical resistance was measured at the indicated times (A, C). FITC-dextran flux was examined after 48 hours of incubation with cytokines (B, D). Data are presented as a mean ± SE (n = 3); ***P < 0.001.

FIGURE 5.

E-cadherin upregulator and ML237 prevent cytokine-induced disruption of epithelial apical junctions. Confluent T84 cell monolayers were pretreated with either vehicle, ECU (10 μM), or ML327 (10 μM) for 24 hours. Afterward, cells were exposed to a combination of TNFα (10 ng/mL) and IFNγ (50 ng/mL) with or without E-cadherin-enhancing drugs for 48 hours. Cells were fixed, and the structure of AJ and TJ was determined by immunolabeling for E-cadherin, ZO-1, and claudin-4. Arrows highlight disrupted and internalized AJ and TJ in the vehicle-treated cytokine-exposed cells. Arrowheads point at the preserved apical junctions in ECU- and ML327-treated cytokine-exposed cell monolayers. Scale bar, 20 µm.

Because excessive epithelial cell death is known to contribute to cytokine-induced disruption of the gut barrier,^{18, 19} we next assessed if the observed barrier-protective effects of the isoxazole compounds could be mediated by attenuated cell apoptosis. Immunoblotting analysis revealed that 48-hour exposure of T84 cell monolayers to TNFα and IFNγ caused significant cell apoptosis, manifested by the appearance of cleaved PARP, along with cleaved (active) caspases 3 and 7 (Fig. 6A). We also examined the effects of cytokine exposure on the expression of different pro-apoptotic and anti-apoptotic proteins and observed selective upregulation of Noxa and Mcl-1 in TNFα/IFNγ-treated cells (Fig. 6A). Although Noxa is a well-known inducer of mitochondrial apoptosis,⁵³ Mcl-1 is a Noxa-interacting protein with anti-apoptotic activity.⁵⁴ Treatment with ECU or ML327 significantly attenuated apoptosis-related molecular events in cytokine-treated T84 cells by inhibiting PARP cleavage and caspase activation and preventing upregulation of Noxa and Mcl-1 expression (Fig. 6A, B). Interestingly, ECU exhibited more pronounced anti-apoptotic activity compared with ML327 (Fig. 6), which is consistent with the superior barrier protective-effects of this compound in cytokine-treated epithelial cell monolayers (Fig. 4). Overall, this series of experiments demonstrates that isoxazole compounds prevent barrier disruption and epithelial junction disassembly caused by pro-inflammatory cytokines. Their barrier-protective effects are likely mediated by different mechanisms, including the inhibition of apoptosis and the partial restoration of junctional protein expression in cytokine-treated epithelial cells.

FIGURE 6.

Isoxazole compounds attenuate cytokine-induced cell death. Confluent T84 cell monolayers were pretreated with either vehicle, ECU (10 μM), or ML327 (10 μM) for 24 hours. Afterward, cells were exposed to a combination of TNFα (10 ng/mL) and IFNγ (50 ng/mL) with and without isoxazole compounds for an additional 48 hours. The expression of different pro-apoptotic and anti-apoptotic proteins was determined via immunoblotting analysis. Representative immunoblots (A) and densitometric quantification of protein band intensities in the cytokine-treated cells, from 3 independent experiments (B) are shown. Data are presented as a mean ± SE (n = 3); *P < 0.05; **P < 0.01.

ECU and ML327 Stimulate Cell Migration in Wounded Intestinal Epithelial Monolayers

Next, we asked a clinically relevant question about the effects of epithelial barrier–enhancing isoxazole compounds on intestinal mucosal restitution. The restitution process was modeled using a scratch wound healing assay. Confluent T84 cell monolayers pre-incubated with ECU, ML327, or vehicle were wounded and allowed to migrate for 48 hours. Figure 7 shows that both compounds significantly increased the rate of wound healing, with ECU promoting cell migration more efficiently than ML327. Importantly, ECU-dependent and ML327-dependent increases in cell migration were also observed in SK-CO15 cell (Supplementary Fig. 4A, B) and IEC6 rat small intestinal epithelial cell (Supplementary Fig. 4C, D) monolayers, thus indicating this phenomenon is not specific to T84 cells. Next, we sought to elucidate which signaling mechanisms could mediate such accelerated epithelial wound closure. Immunoblotting analysis was used to determine the activation status of major signaling pathways involved in cell-matrix adhesion and epithelial migration. Interestingly, ECU and ML327 demonstrated similar and disparate effects on the promigratory signaling (Fig. 8). Both compounds similarly increased phosphorylation of Src, indicating activation of this kinase that plays key roles in the assembly of cell-matrix adhesions.⁵⁵ However, ECU—but not ML327—promoted activatory phosphorylation of another promigratory kinase, AKT (Fig. 8). Finally, we asked if the prorestitutive activity of the isoxazole compounds is preserved in epithelial cell monolayers treated with pro-inflammatory cytokines. Exposure of T84 cells to a combination of TNFα and IFNγ significantly attenuated scratch wound healing (Supplementary Fig. 5). Remarkably, such cytokine-induced attenuation of epithelial cell migration was completely reversed by either ECU or ML327 treatment (Supplementary Fig. 5). The described experiments reveal novel functions of these isoxazole compounds in promoting the collective intestinal epithelial cell migration and reversing the cytokine-induced block of epithelial restitution in vitro.

FIGURE 7.

E-cadherin upregulator and ML237 promote the collective migration of intestinal epithelial cells. Confluent T84 cell monolayers were pretreated for 24 hours with either vehicle, ECU (10 μM), or ML327 (10 μM), then wounded, and allowed to migrate into the wound area in the presence of drugs. Cell monolayers were photographed, and the percentage of wound closure was calculated at the indicated times. Data are presented as a mean ± SE (n = 3); *P < 0.05; **P < 0.01; ***P < 0.001. Scale bar, 100 µm.

FIGURE 8.

Accelerated wound healing in ECU- and ML327-treated epithelial cells is accompanied by the activation of promigratory signaling events. Confluent T84 cell monolayers were pretreated with either vehicle, ECU (10 μM), or ML327 (10 μM), for 24 hours and were subjected to multiple wounding. Total cell lysates were collected at 12 and 24 hours postwounding, and the expression of different signaling molecules was determined by immunoblotting. Representative immunoblots (A) and densitometric quantification of protein band intensities from 3 independent experiments (B) are shown. Data are presented as a mean ± SE (n = 3); *P < 0.05; **P < 0.01.

DISCUSSION

Disruption of the intestinal epithelial barrier is an important mechanism of IBD, and pharmacological protection of gut barrier integrity could be a valuable supportive treatment for decreasing the morbidity of IBD patients.^{2, 4} In the present study, we describe a novel pharmacological approach for stabilizing the intestinal epithelial barrier using isoxazole small molecules such as ECU and ML327. Our study revealed a triple beneficial effect of these compounds: the tightening of the epithelial barrier under normal homeostatic conditions, the attenuation of cytokine-induced barrier disruption, and the acceleration of epithelial wound healing. Although the goal of the present study was limited to examining the effects of isoxazole compounds on model intestinal epithelial barriers in vitro, we predict that ECU and ML327 could also attenuate barrier disruption and accelerate epithelial restitution during mucosal inflammation in vivo.

E-cadherin-upregulator and ML327 have been previously identified as potent suppressors of EMT in mesenchymal-type colon and lung cancer cell lines by mechanisms involving stimulation of E-cadherin expression.^{34, 35} We confirmed and expanded upon these data by demonstrating that ECU and ML327 stimulate the expression of not only E-cadherin but also P-cadherin proteins in poorly differentiated colon cancer cell lines (Fig. 2B, C). Surprisingly, these compounds upregulated neither E-cadherin nor P-cadherin expression in well-differentiated T84 and SK-CO15 colonic epithelial cell lines that are characterized by high steady-state levels of epithelial cadherins (Fig. 2A, data not shown). Furthermore, ECU potently promoted the formation of the paracellular barrier in E-cadherin-depleted epithelial cell monolayers (Fig. 3B, C), providing additional support for its E-cadherin-independent mechanisms of action. Our findings, together with previously published data, suggest that ECU and ML327 are not bona fide regulators of classical cadherin expression, but they could derepress the expression of these AJ proteins in EMT-like or inflammatory states. Such derepression is likely to occur at the epigenetic level because ML327 was unable to increase E-cadherin expression in RKO and MDA-MB-231 cancer cells due to its promoter hypermethylation.⁵⁶ In colon and lung cancer cells, ECU and its structural derivatives increased the level of histone H4 acetylation without inhibiting histone deacetylating enzymes.³⁴ Consequently, isoxazole compounds could modulate histone acetylation and chromatin organization by yet to be determined mechanisms. It is also predictable that ECU- and ML327-dependent modulation of chromatin organization would have global effects on gene expression in different cell types. Indeed, previous RNA sequencing analysis demonstrated significant numbers of genes either upregulated or downregulated in ML327-treated colon cancer cells.³⁵ Because regulation of epithelial barriers involves complex cross-talk among junctional protein expression, vesicular trafficking, cytoskeletal remodeling, and intracellular signaling,^{11, 20} several of these mechanisms could be affected by ECU- and ML327-induced transcriptional reprogramming of epithelial cells. One such mechanism could involve altered vesicle trafficking of different AJ and TJ proteins. Indeed, we found that isoxazole compounds altered the distribution of claudins between their plasma membrane and cytoplasmic pools, both in control and cytokine-treated epithelial cells (Fig. 5, Supplementary Fig. 1, data not shown). This altered distribution resulted in the retention of claudin proteins at the plasma membrane, thus contributing to the enhancement/stabilization of the epithelial barrier.

Another important finding of this study is the protective effects of ECU and ML327 against cytokine-induced breakdown of model intestinal epithelial barrier. Such barrier-protective actions were observed under conditions of strong pro-inflammatory stimulation caused by the combined actions of TNFα and IFNγ (Figs. 4 & 5), which are known to trigger the disruption and internalization of junctional proteins in different types of epithelial cells^{47, 49} and are notoriously difficult to prevent with either pharmacological or genetic manipulations. Furthermore, elevated levels of TNFα and IFNγ have been implicated in the disruption of gut barrier in patients with IBD, graft vs host disease, and other gastrointestinal inflammatory disorders.^{8, 10, 12, 48, 50} This makes isoxazole compounds attractive candidates for the creation of barrier-protective drugs, which may limit the development of mucosal inflammation. The observed barrier-protective features of ECU and ML327 could be mediated by several mechanisms. One mechanism is a partial preservation of E-cadherin and claudin-7 protein levels in cytokine-treated colonic epithelial cells (Supplementary Fig. 3). This resembles ECU- and ML327-dependent upregulation of E-cadherin expression in poorly differentiated cancer cells (Fig. 2B, C) and may reflect the reversal of cytokine-induced epigenetic inhibition of AJ/TJ protein expression. However, the most robust mechanism underlying the barrier-protective effects of isoxazole compounds involves the attenuation of cytokine-induced epithelial cell apoptosis. Indeed, ECM, and to a lesser extent ML327, demonstrated strong anti-apoptotic actions in TNFα and IFNγ-treated cell monolayers (Fig. 6). These results are consistent with previous findings that ECU inhibits cisplatin-induced death of kidney tubular epithelial cells,³⁸ but is contradictory to published data that ML327 sensitizes colon and breast cancer cells to TNF-related apoptosis-inducing ligand-dependent cell death.^{56, 57} The reason for such conflicting results is unclear; however, it is possible that isoxazole compounds activate distinct pro- and anti-apoptotic signaling pathways in normal epithelial and cancer cells, depending on their differentiation and transformation status.

Accelerated cell death in the actively inflamed mucosa of IBD patients results in the formation of epithelial wounds that, in turn, exaggerate the inflammatory response.⁵⁸ Wound healing is an important event in mucosal restitution and is required for the restoration of gut barrier integrity.^{19, 59} Our results suggest that, in addition to their barrier-protective effects, the isoxazole compounds could have potent prorestitution activity in inflamed intestinal mucosa. Indeed, ECU and ML327 accelerated wound healing in model intestinal epithelial cell monolayers by activating important promigratory signaling pathways (Figs. 7, 8, Supplementary Figs. 4, 5). Such prorestitutive activity of ECU and ML327 could be underscored by similar and unique signaling mechanisms. One similar mechanism involves Src activation by both compounds (Fig. 8). Additionally, ECU—but not ML327—potently activated AKT (Fig. 8). Because AKT is known to play prosurvival roles in different cells, its activation by ECU could have 2 major functional consequences: stimulation of epithelial cell migration and inhibition of cell death. Importantly, the observed prorestitutive effects of ECU and ML327 are not limited to normal colonic epithelial monolayers because these compounds were able to completely prevent the inhibition of epithelial wound healing caused by pro-inflammatory cytokines (Supplementary Fig. 5).

Though our study did not reveal major roles for epithelial cadherins’ upregulation in the barrier-protective and prorestitutive activities of the investigated isoxazole compounds in vitro, we cannot exclude the possibility that these compounds could enhance E-cadherin expression under inflammatory and injurious conditions in vivo. Indeed, ECU was shown to restore E-cadherin levels during acute cisplatin-induced kidney injury in mice.³⁸ Furthermore, our data demonstrated partial restoration of E-cadherin expression in cytokine-exposed epithelial monolayers treated with ECU or ML327 (Supplementary Fig. 3). Generally, stimulation of E-cadherin-based epithelial adhesions should be considered as an attractive approach for the stabilization and restoration of gut barrier during mucosal inflammation. This could be achieved via 2 different mechanisms: upregulation of E-cadherin expression or the activation of E-cadherin-dependent adhesive interactions. Published literature describes several small molecules that stimulate E-cadherin expression. The list includes methotrexate,⁶⁰ sphingosine-1-phosphate,⁶¹ and formononetin,⁶² among others. Additionally, many microRNA species, including miR-200b,^{63, 64} miR-205,⁶⁵ and miR-101,⁶⁶ are known to upregulate E-cadherin levels in different cancer cells. The expression-independent mechanism that stimulates E-cadherin adhesions involves monoclonal E-cadherin-activating antibodies. These antibodies interact with the extracellular domains of E-cadherin, altering protein conformation and promoting E-cadherin-based cell-cell adhesions.^{67, 68} It would be interesting to see if the described E-cadherin function-enhancing approaches could have barrier-protective activities during intestinal inflammation.

In conclusion, our study revealed that the isoxazole small molecules ECU and ML327, previously identified as E-cadherin upregulating drugs in cancer cell lines, can act as potent barrier-enhancing and barrier-protective compounds in intestinal epithelial monolayers. These compounds displayed a combination of beneficial effects in vitro that include enhancement of steady state epithelial barrier, attenuation of cytokine-induced barrier disassembly, and acceleration of epithelial wound healing. These effects were not caused by upregulation of E-cadherin expression but involved other mechanisms including the attenuation of cell death and stimulation of promigratory Src and AKT signaling. The discovered barrier-protective effects of the studied isoxazole compounds reveal them to be attractive candidates for the development of new barrier-protective drugs to manage mucosal inflammation in IBD patients.

Supported by: National Institutes of Health grant R01 DK108278 to AII.

Conflicts of Interest: AII has a patent application relevant to this study.