Abstract
Ethyl pyruvate (EP) solution ameliorates ileal mucosal hyperpermeability and decreases the expression of several proinflammatory genes in ileal and/or colonic mucosa when it is used instead of Ringer's lactate solution (RLS) to resuscitate mice from hemorrhagic shock. To test the hypothesis that EP can ameliorate gut barrier dysfunction induced by other forms of inflammation, we incubated Caco-2 monolayers for 24 to 48 h with cytomix (a mixture of interferon-γ, tumor necrosis factor-α, and interleukin-1β) in the presence or absence of graded concentrations of EP or sodium pyruvate. Cytomix increased the permeability of Caco-2 monolayers to fluorescein isothiocyanate-labeled dextran (FD4; average molecular mass 4 kDa), but this effect was inhibited by adding 0.1 to 10 mM EP (but not similar concentrations of sodium pyruvate) to the culture medium. EP inhibited several other cytomix-induced phenomena, including nuclear factor-κB activation, inducible nitric oxide synthase mRNA expression, and nitric oxide production. Cytomix altered the expression and localization of the tight junctional proteins, ZO-1 and occludin, but this effect was prevented by EP. Delayed treatment with EP solution instead of RLS ameliorated ileal mucosal hyperpermeability to FD4 and bacterial translocation to mesenteric lymph nodes in mice challenged with lipopolysaccharide (LPS). These data support the view that EP ameliorates cytokine- and/or LPS-induced derangements in intestinal epithelial barrier function.
The simple α-ketocarboxylate pyruvate functions in cells not only as an important intermediate in the metabolism of glucose but also as an endogenous antioxidant and free radical scavenger (Brand, 1997; Brand and Hermfisse, 1997; Biagini et al., 2001). The capacity of pyruvic acid to function as an antioxidant was first reported by Holleman (1904), who showed that α-keto acids with the general structure, R–CO–COOH, reduce hydrogen peroxide (H2O2)3nonenzymatically in a reaction that yields carbon dioxide and water. In the case of pyruvic acid, this oxidative decarboxylation reaction also yields acetate and is both rapid and stoichiometric (Bunton, 1949;Melzer and Schmidt, 1988). In addition to H2O2, pyruvate is also capable of scavenging another highly reactive oxygen species (ROS), namely hydroxyl radical (Dobsak et al., 1999).
Recognition that pyruvate is an effective ROS scavenger prompted numerous investigators to try using this compound as a therapeutic agent for the treatment of various pathological conditions that are thought to be mediated, at least in part, by redox-dependent phenomena. For example, Salahudeen et al. (1991) showed that the i.v. infusion of a solution of sodium pyruvate protects rats from renal parenchymal injury induced by glycerol, a model of acute kidney failure associated with increased production of H2O2. Other investigators reported that treatment with pyruvate ameliorates organ injury or dysfunction in animal models of redox stress, such as transient myocardial (Bunger et al., 1989), intestinal (Cicalese et al., 1999), or hepatic (Sileri et al., 2001) ischemia followed by reperfusion.
Despite these promising findings, the usefulness of pyruvate as a therapeutic agent may be limited by its poor stability in solution (von Korff, 1964). When dissolved water, pyruvate spontaneously undergoes condensation and cyclization reactions to form a variety of products, some of which may be toxic (Montgomery and Webb, 1956). In an effort to circumvent this issue, our laboratory formulated a derivative of pyruvic acid, namely ethyl pyruvate (EP), in a calcium- and potassium-containing balanced salt solution and showed that treatment with this fluid could ameliorate much of the structural and functional damage to the intestinal mucosa caused by mesenteric ischemia and reperfusion in rats (Sims et al., 2001). Interestingly, in this study, treatment with EP seemed to be substantially more effective than treatment with pyruvate. Similar findings indicating that EP is more effective than pyruvate were reported by Varma et al. (1998), who compared the two compounds in an in vitro study of redox-mediated cellular injury.
Recently, our laboratory showed that resuscitation with EP solution instead of Ringer's lactate solution (RLS) prolongs survival and decreases intestinal mucosal injury in rats subjected to hemorrhagic shock (Tawadrous et al., 2002). In a follow-up study, we showed that resuscitation with EP solution instead of RLS decreases activation of the proinflammatory transcription factor NF-κB in liver and colonic mucosa following hemorrhagic shock in mice and also decreases the expression of several proinflammatory genes, including inducible nitric oxide synthase (iNOS), TNF-α, cyclooxygenase-2, and IL-6 in liver, and ileal mucosa and colonic mucosa (Yang et al., 2002a). These latter findings suggested to us that EP may have activity as an anti-inflammatory agent.
To pursue this line of investigation further, we took advantage of prior studies from our laboratory wherein we showed that injecting rats with lipopolysaccharide (LPS) promotes gut barrier dysfunction through an NO-dependent mechanism (Unno et al., 1997c) and incubation of Caco-2 human enterocytic monolayers with “cytomix”, a mixture containing the proinflammatory cytokines IFN-γ, IL-1β, and TNF-α, which increases epithelial permeability via a mechanism that is dependent on NO availability (Chavez et al., 1999a). In this study, we show that adding EP, but not sodium pyruvate, to cytomix-stimulated Caco-2 monolayers down-regulates the induction of iNOS, the production of NO, and the development of hyperpermeability to a hydrophilic macromolecular tracer. Furthermore, we also show that treatment of endotoxemic mice with EP ameliorates gut barrier dysfunction, even when administration of the compound is delayed for 6 h after the injection of LPS.
Materials and Methods
Animals.
The research protocol complied with the regulations regarding animal care as published by the National Institutes of Health and was approved by the Institutional Animal Use and Care Committee of the University of Pittsburgh Medical School. Male C57BL/6J mice, weighing 20 to 25 g (The Jackson Laboratories, Bar Harbor, ME), were maintained at the University of Pittsburgh Animal Research Center with a 12-h light/dark cycle and free access to standard laboratory food and water. Animals were not fasted before the experiments.
Materials.
All chemicals were purchased from Sigma-Aldrich (St. Louis, MO) unless otherwise noted. Dulbecco's minimal essential medium (DMEM) and phosphate-buffered saline (PBS) were from BioWhittaker (Walkersville, MD). Fetal bovine serum (FBS) (<0.05 endotoxin units/ml) was obtained from Hyclone (Logan, UT). IFN-γ, TNF-α, and IL-1β were obtained from Pierce-Endogen (Rockford, IN). Anti-ZO-1 and anti-occludin monoclonal antibodies (Ab) were purchased from BD Translabs (Franklin Lakes, NJ). All secondary Ab used for immunohistochemistry were from Jackson Immunologicals (Bar Harbor, ME). Peroxidase-conjugated anti-mouse IgG (Fc fragment) Ab used for Western blotting was from Sigma-Aldrich (St. Louis, MO). Caco-2 human intestinal epithelial cells were obtained from the American Type Culture Collection (Manassas, VA).
Cell Culture.
Caco-2 cells were routinely maintained on collagen-1 coated Biocoat tissue culture dishes (BD Biosciences, San Jose, CA) at 37°C in a 5% CO2 humidified atmosphere in DMEM supplemented with 10% FBS, penicillin G (100 U/ml), streptomycin (100 μg/ml), pyruvate (2 mM), l-glutamine (4 mM), and nonessential amino acid supplement (2% v/v).
Monolayer Permeability Assays.
Caco-2 human enterocytes (100,000 cells/well) were plated on permeable filters (0.4-μM pore size) in 12-well Transwell bicameral chambers (COSTAR, Corning, NY) and fed biweekly. Permeability studies were carried out using confluent monolayers during the interval from 21 to 28 days after seeding. The permeability probe was fluorescein isothiocyanate-labeled dextran (FD4; average molecular mass 4400 Da). A sterile stock solution of FD4 (25 mg/ml) was prepared by dissolving the compound in HEPES-buffered DMEM complete medium (pH 6.8) and passing it through a filter (0.45-μm pore size). For permeability studies, the medium was aspirated from the apical and basolateral sides of the Transwell chambers. FD4 solution (200 μl) was pipetted into the apical compartments. The medium on the basolateral side of the Transwell chambers was replaced with 500 μl of control medium, medium with cytomix, or medium with cytomix plus 0.01, 0.1, 1, and 10 mM EP or sodium pyruvate. After 24 and 48 h of incubation, 30 μl of medium was aspirated from the basolateral compartments for spectrofluorometric determination of FD4 concentration, as previously described (Menconi et al., 1997). Measurements were made using a PerkinElmer LS-50 fluorescence spectrophotometer (Palo Alto, CA). Samples were diluted with 3 ml of Tris-buffered saline (pH 7.5). Fluorescence at 515 nm (slit width 10 nm) was determined using an excitation wavelength of 492 nm (slit width 2.5 nm) and an integration time of 30 s. The permeability of monolayers was expressed as a clearance with units of nanoliters per centimeter per hour, which was calculated as previously described (Menconi et al., 1997). Concurrent controls were performed with each experiment.
Nuclear Extract Preparations.
Caco-2 enterocytes were plated at 106 cells/well in six-well dishes for 21 days. The Caco-2 cells were incubated with control medium, medium with cytomix, or medium with 10 mM EP plus cytomix. After various intervals of stimulation, the cells were removed from the incubator and immediately placed on ice. Cells were washed once with PBS then harvested in 1 ml of PBS containing 2% FBS using a rubber policeman. The cells were transferred to 1.5-ml microfuge tubes and centrifuged at 14,000g for 10 s. The cell pellet was resuspended in 600 μl of buffer I [10 mM KCl, 1.5 mM MgCl2, 0.3 M sucrose, 500 μM phenylmethylsulfonyl fluoride (PMSF), 1.0 mM sodium orthovanadate, 1 mM dithiothreitol (DTT), and 10 mM Tris · HCl, pH 7.8] and incubated for 15 min. Subsequently, 38.3 μl of 10% NP-40 was the added and the tubes were vortexed at full speed for 10 s. The nuclei were isolated by centrifugation at 310g for 3 min, and the supernatants were aspirated. The nuclear pellets were gently resuspended in 80 μl of buffer II (500 μM PMSF, 1.0 mM sodium orthovanadate, 1 mM DTT, 420 mM KCl, 1.5 mM MgCl2, 20% glycerol, and 10 mM Tris · HCl, pH 7.8). Following 15 min of incubation, nuclear extracts were cleared by centrifugation at 14,000g for 10 min. The supernatants were transferred to new tubes, and the protein concentration was determined using a commercially available Bradford assay (Bio-Rad protein assay; Bio-Rad; Hercules, CA). Nuclear extracts were frozen at −80°C.
Electrophoretic Mobility Shift Assay (EMSA).
The sequence of the double-stranded NF-κB oligonucleotide was as follows: sense, 5′-AGT TGA GGG GAC TTT CCC AGG C-3′; antisense, 3′-TCA ACT CCC CTG AAA GGG TCC G-5′ (Promega, Madison, WI). The oligonucleotides were end-labeled with [γ-32P]ATP (PerkinElmer Life Sciences, Boston, MA) using T4 polynucleotide kinase (Promega). Nuclear protein (3 μg) was incubated with γ-32P-labeled NF-κB probe (1 μl) in 4 μl of 5× bandshift buffer (325 mM NaCl, 5 mM DTT, 0.7 mM EDTA, 40% v/v glycerol, and 65 mM HEPES, pH 8.0) in the presence of 2 μg of poly [d(I-C)] (Boehringer Mannheim, Indianapolis, IN) for 20 min at room temperature. For competition reactions, a 100-fold molar excess of cold oligonucleotide was added simultaneously with labeled probe. Super-shift assays were performed by incubating nuclear extracts with 2 μl of anti-p65 and anti-p50 Ab (Santa Cruz Biotechnology, Santa Cruz, CA) for 1 h before the addition of the radiolabeled probe. The binding reaction mixture was electrophoresed on 4% nondenaturing polyacrylamide gel electrophoresis gels containing 5% glycerol and 1/4× Tris-borate-EDTA buffer. After polyacrylamide gel electrophoresis, the gels were dried and exposed to Biomax-5 film (Kodak, Rochester, NY) at −80°C overnight using an intensifying screen.
Semiquantitative Reverse Transcription-Polymerase Chain Reaction (RT-PCR).
To estimate iNOS mRNA levels, Caco-2 enterocytic monolayers were grown in six-well dishes for 21 days. The Caco-2 cells were incubated for 24 h with control medium, medium plus cytomix, or medium plus cytomix plus 10 mM of EP. Total RNA was extracted from the cells with chloroform and TRI Reagent (Molecular Research Center, Cincinnati, OH) as directed by the manufacturer. The total RNA was treated with DNAFree (Ambion, Houston, TX) as instructed by the manufacturer using 10 units of DNase I/10 μg of RNA. Two micrograms of total RNA was reverse transcribed in a 40-μl reaction volume containing 0.5 μg of oligo(dT)15 (Promega), 1 mM of each dNTP, 15 U avian myeloblastosis virus reverse transcriptase (Promega), and 1 U/μl of RNasin ribonuclease inhibitor (Promega) in 5 mM MgCl2, 50 mM KCl, 0.1% Triton X-100, and 10 mM Tris-HCl (pH 8.0). The reaction mixture was preincubated at 65°C for 10 min before DNA synthesis. The RT reaction was carried out for 50 min at 42°C and was heated to 95°C for 5 min to terminate the reaction. Reaction mixtures (50 μl) for PCR were assembled using 5 μl of cDNA template, 10 units AdvanTaq Plus DNA polymerase (Invitrogen, Carlsbad, CA), 200 μM each dNTP, 1.5 mM MgCl2, and 1.0 μM of each primer in 1× AdvanTaq plus PCR buffer. PCR reactions were performed in a PerkinElmer model 480 thermocycler (Norwalk, CT). Amplification was initiated with 5 min of denaturation at 94°C. Amplification of cDNA for iNOS was carried out by denaturing at 94°C for 45 s, annealing at 61°C for 45 s, and polymerizing at 68°C for 1 min for 25 cycles. After the last cycle of amplification, the samples were incubated at 72°C for 7 min and then held at 4°C. The 5′ and 3′ primers for iNOS were 5′-GCG CCT GGA GGA CCT GGA TGA GA -3′ and 5′-CCC GGG AGG AGC TGA TGG AGT AGA-3′, respectively (Invitrogen); the expected product length was 341-base pair 18S ribosomal RNA was amplified to verify equal loading. For this reaction, the 5′ and 3′ primers were 5′-CCC GGG GAG GTA GTG ACG AAA AAT-3′ and 5′-CGC CCG CTC CCA AGA TCC AAC TAC-3′, respectively; the expected product length was 200 base pairs. For amplifying 18S cDNA, PCR was carried out by denaturing at 94°C for 30 s, annealing at 55°C for 30 s, and polymerizing at 68°C for 1 min for 20 cycles. After the last cycle of amplification, the samples were incubated at 68°C for 3 min and then held at 4°C. Ten microliters of each PCR reaction was electrophoresed on a 2% agarose gel in 1× Tris-acetate-EDTA buffer, scanned in NucleoVision imaging workstation (NucleoTech, San Mateo, CA), and quantified using GelExpert release 3.5.
Measurement of NO Production.
To determine nitrate (NO3−) plus nitrite (NO2−) concentration in culture supernatants, 106 Caco-2 enterocytes were plated in six-well dishes and incubated for 21 days. Confluent monolayers were incubated for 24 h with control medium, medium plus cytomix, or medium plus cytomix plus 10 mM of EP. To first reduce NO3− to NO2−, cadmium filings (0.4–0.7 g/tube; Fluka, Milwaukee, WI) were loaded into 1.5-ml microfuge tubes. The filings were washed twice with 1.0 ml of deionized water, twice with 1.0 ml of 0.1 M HCl, and twice with 1.0 ml of 0.1 M NH4OH. Ten microliters of 30% ZnSO4 was added to 200 μl of culture supernatant, vortexed, incubated at room temp for 15 min, and centrifuged at 14,000g for 5 min. The resulting supernatant was added to a cadmium-containing microcentrifuge tube and incubated at room temperature overnight with constant mixing. The samples were transferred to fresh microcentrifuge tubes and centrifuged again. The supernatants were subsequently assayed for NO2−, using a modified Griess assay as previously described (Vodovotz, 1996).
Western Blot Analyses.
Postconfluent cultures of Caco-2 cells were used 21 to 24 days after plating. The cultures were incubated with control medium, medium with cytomix, or medium with cytomix plus 10 mM EP for 48 h. After washing with ice-cold PBS, the cells were lysed in 1 ml of radioimmunoprecipitation assay buffer [1× PBS, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 0.1 mg/ml PMSF, 1.0 mM sodium orthovanadate, and 1× mammalian protease inhibitor cocktail (Sigma-Aldrich catalog no. P 8340)]. The cells were removed from the tissue culture plate by gentle scraping with a rubber policeman and transferred to a 1.5-ml microfuge tube. The samples were sonicated 3 times for 30 s on ice using a 0.1-W Fisher Scientific sonic dismembrator (Pittsburgh, PA) fitted with a microtip on power setting 3. The lysate was transferred to a microcentrifuge tube and incubated for 30 min on ice. The lysate was centrifuged at 10,000g for 15 min at 4°C, and then the supernatant was transferred to a new tube. Total protein concentration was determined using the Bio-Rad protein reagent.
Equivalent amounts of protein were mixed with Laemlli buffer [20% glycerol, 10% β-mercaptoethanol, 5% SDS, 0.2 M Tris · HCl (pH 6.8), and 0.4% bromophenol blue]. After boiling for 5 to 10 min, the protein samples were centrifuged for 10 s, and the supernatants were electrophoresed at 100 mA for 40 min on 7.5% precast SDS-polyacrylamide gels (Bio-Rad). The size-fractionated proteins were electroblotted onto a Hybond-P PVDF membrane (Amersham Biosciences, Leicester, Denmark), blocked with Blotto [1× Tris-buffered saline (TBS), 5% milk, 0.05% Tween 20, and 0.2% NaN3] for 60 min. The membrane was then incubated at room temperature for 1 h with murine monoclonal anti-ZO-1 or anti-occludin monoclonal Ab diluted 1:2,000 or 1:1,000, respectively, in PBST (1× PBS containing 0.02% Tween 20). After washing three times in 1× PBST, immunoblots were exposed at room temperature for 1 h to a 1:20,000 dilution of anti-mouse horseradish peroxidase-conjugated anti-Ig secondary Ab. Following three washes in PBST and two washes in PBS, the membrane was impregnated with the Enhanced Chemiluminescence substrate (Amersham Biosciences) and used to expose X-ray film according to the manufacturer's instructions. The autoradiographs were captured using a Hewlett Packard (Palo Alto, CA) ScanJet 6300s. Band intensities were quantified by densitometry and expressed as the mean area density using GelExpert 3.5 software (Nucleotech Corporation, San Mateo, CA).
Immunohistochemistry.
All procedures were performed at 0–4°C, all buffers were precooled, and all washes were incubated for 5 min. Caco-2 cells growing on collagen-1 coated eight-well culture slides (BD Biosciences) were used at 14 to 17 days after the monolayers were confluent. Following treatment with control medium, cytomix-containing medium, or cytomix-containing medium plus EP, the monolayers were fixed with methanol for 10 min at 20°C and then allowed to air dry. The cells were stained using anti-ZO-1 or anti-occludin Ab, which were diluted 1:100 in TBS and centrifuged for 2 min at 13,500g. For staining, the cells were washed twice with PBS then overlaid with a volume of the Ab solution that was just sufficient to cover the surface of the slides. The slides were incubated for 1 h then washed four times with PBS. The secondary Ab for ZO-1 was 5 μg/ml tetramethylrhodamine isothiocyanate-conjugated affinity-purified donkey anti-mouse IgG. The secondary Ab for occludin was 15 μg/ml tetramethylrhodamine isothiocyanate-conjugated affinity purified anti-mouse IgG. The secondary Ab were diluted 1:100 in TBS and centrifuged for 2 min at 13,500g, and then layered over the cells and incubated for 1 h. The slides were washed four times in PBS. Nuclei were stained by incubating the cells in 2× saline-sodium citrate buffer containing 500 nM propidium iodide (Molecular Probes, Eugene, OR) for 5 min. The cells were washed twice with 2× saline-sodium citrate buffer, then the excess buffer was drained away and antifade reagent (Molecular Probes) was used to preserve the specimens. The cells were imaged using a Model DM HC inverted fluorescent microscope (Leica, Germany) equipped with a Diagnostic Instruments SPOT-II model 1.4 Digital Microscope Imaging System (Burroughs, MI). Captured images were minimally manipulated for publication using Adobe Photoshop software (San Jose, CA).
Experimental Design for in Vivo Experiments.
Four groups of mice (n = 5 each) were studied. All agents were injected intraperitoneally. Each animal in the control (PBS) group was injected with 1.0 ml of PBS and then with 0.31 ml of RLS 1, 6, and at 12 h later. Each mouse in the LPS + RLS group was injected with 1.0 ml of a well sonicated suspension of Escherichia coliserotype 0111:B4 LPS (0.1 mg/ml; 4 mg/kg) in PBS. One, six, and twelve h later, these mice were injected with 0.31 ml of RLS, a balanced salt solution containing 109 mM NaCl, 4.0 mM KCl, 2.7 mM CaCl2, and 28 mM sodium lactate. Mice in the early EP (LPS + EARLY EP) group were injected with the same dose of LPS suspension and then were injected with 0.31 ml of a solution of EP (3.23 mg/ml; 40 mg/kg) 1, 6 and 12 h later. The EP was dissolved in a balanced salt solution containing 130 mM NaCl, 4 mM KCl, and 2.7 mM CaCl2. Mice in the late EP (LPS + LATE EP) group were injected with the same dose of LPS and then 6 and 12 h later with 0.31 ml/dose of EP solution. Eighteen hours after the injection of LPS (or the PBS vehicle in the control group), the mice were anesthetized with intramuscular injections of sodium pentobarbital (90 mg/kg), and segments of ileum were excised for determination of mucosal permeability (see below). The mesenteric lymph node complex was harvested to measure bacterial translocation (see below). Blood was aspirated from the heart to measure the plasma concentration of alanine aminotransferase (ALT).
Measurement of Intestinal Mucosal Permeability.
Intestinal mucosal permeability to the fluorescent tracer, FD4, was determine using an everted gut sac method, as previously described byWattanasirichaigoon et al. (1999) and subsequently modified by Yang et al. (2002a). Briefly, everted gut sacs were prepared in ice-cold modified Krebs-Henseleit bicarbonate buffer (KHBB; pH 7.4). One end of the gut segment was ligated with a 4.0 silk. The segment was then everted using a thin plastic rod, and the resulting sac was secured with a 4.0 silk suture to the grooved tip of a 3-ml plastic syringe containing KHBB. The everted sac was gently distended by injecting 1.5 ml of KHBB and suspended in a 50-ml beaker containing 40 ml of a solution of FD4 (40 μg/ml) in KHBB. The solution in the beaker was temperature jacketed at 37°C and continuously bubbled with a gas mixture containing 95% O2/5% CO2. A 1.0-ml sample was taken from the beaker before adding the sac to determine the initial external (i.e., mucosal surface) FD4 concentration. The sac was incubated for 30 min in the FD4 solution. At the end of this period, the length of the gut sac was measured. The fluid on the serosal side was aspirated for the determination of FD4 concentration. The serosal and mucosal samples were centrifuged for 10 min at 1000g. Three-hundred-microliter samples of the supernatants were diluted with 2.7 ml of PBS. FD4 concentrations were determined spectrofluorometrically as described above. Permeability was expressed as the mucosal-to-serosal clearance of FD4 and calculated as previously described (Wattanasirichaigoon et al., 1999).
Quantitation of Bacterial Translocation.
The skin was cleaned with a 10% solution of povidone-iodine. Using sterile technique, the abdominal cavity was opened and the viscera were exposed. The mesenteric lymph node complex was removed, weighed, and placed in a Donnce homogenizer fitted with pestle B (Fisher Scientific) containing 1 ml of PBS. The lymph node complex was homogenized with five strokes of the pestle. Three-hundred-microliter aliquots of the homogenate were coated onto plates containing brain-heart agar (BD Biosciences). The plates were examined 24 h later after being aerobically incubated at 37°C. Visible colonies were counted and the extent of translocation expressed as CFU per gram of tissue.
ALT Assay.
Blood (200 μl) was obtained by cardiac puncture and placed in a 0.5-ml centrifugation tube on ice. After being allowed to clot, the sample was centrifuged at 5000g for 3 min. The serum was aspirated and assayed in by the clinical laboratory for ALT concentration.
Assessment of Cell Viability.
Caco-2 cells (100,000 cells/well) were grown on collagen-I coated slides (BD Biosciences) and fed biweekly. The cells were incubated for 48 h with control medium or medium containing cytomix, medium containing 10 mM ethyl pyruvate, or medium containing 10 mM ethyl pyruvate and cytomix. Cell viability was assessed using the LIVE/DEAD viability/cytotoxicity kit from Molecular Probes. After washing the cells twice with phosphate-buffered saline, 100 μl of a solution containing 2 μM calcein-AM and 4 μM ethidium homodimer-1 was added to each well, and the cells were incubated for 45 min. The cells were imaged using a model DM HC inverted fluorescent microscope as described above.
Statistical Methods.
Unless otherwise noted, results are presented as means ± S.E. Permeability data from in vitro studies were always assayed by making comparison to appropriate concurrent controls. In general, data were analyzed using analysis of variance followed by Fisher's protected least significant difference test. Translocation data were analyzed using the Kruskal-Wallis nonparametric analysis of variance and the Mann-Whitney U test. Summary statistics are presented for densitometry results from studies using RT-PCR to estimate iNOS mRNA expression, but these results were not subjected to statistical analysis since the method used was only semiquantitative, and the samples sizes (n = 3–4) were small (Yang et al., 2002a,b). Similarly, summary statistics are presented for densitometry results from Western blots performed to estimate ZO-1 and occludin expression, but these results were not subjected to statistical analysis for the same reasons. Pvalues <0.05 were considered significant.
Results
EP Decreases Cytomix-Induced Hyperpermeability of Caco-2 Monolayers in a Dose-Dependent Manner.
Consistent with previously reported observations (Chavez et al., 1999a), the permeability of Caco-2 monolayers increased following incubation with cytomix for 24 or 48 h (Fig. 1). Addition of EP to the culture conditions, however, prevented the increase in permeability to FD4 induced by cytomix. The protective effect of EP was concentration-dependent, being statistically significant for concentrations equal to or greater than 1.0 mM after a 24-h incubation and 0.1 mM after a 48-h incubation.
Solutions of sodium pyruvate protect against oxidant-mediated injury in a variety of in vitro systems (Andrae et al., 1985; O'Donnell-Tormey et al., 1987; Nath et al., 1995; Borle and Stanko, 1996; Varma et al., 1998). Thus, we considered the possibility that inhibition of cytomix-induced epithelial hyperpermeability would also be observed if a solution of sodium pyruvate was substituted for EP. We also recognized, however, that the culture medium used for growing Caco-2 cells consisted of DMEM that was already supplemented with 2 mM pyruvate among other additives. Accordingly, we considered it to be more likely that EP has effects that are not manifested by the parent compound. To investigate these possibilities, we incubated Caco-2 monolayers for 24 and 48 h under control conditions, with cytomix, or with cytomix plus graded concentrations of sodium pyruvate added to the 2 mM pyruvate already present in the medium (i.e., final concentrations of sodium pyruvate ranging from 2.01 to 12 mM). As depicted in Fig. 2, adding sodium pyruvate failed to protect the monolayers from cytomix-induced hyperpermeability. This finding supports the view that EP has a pharmacological mechanism of action that is not shared by the parent α-ketocarboxylate anion.
Cytomix-Induced Hyperpermeability Is Not Secondary to Cell Death.
Preliminary studies revealed no evidence of increased uptake of trypan blue when Caco-2 monolayers were incubated for 24 or 48 h with cytomix in the presence or absence of EP (data not shown). To obtain additional information regarding the effect of cytomix and/or EP on the viability of Caco-2 cells, we incubated cells growing on culture slides for 48 h with control medium, cytomix, 10 mM EP, or 10 mM EP plus cytomix. Other cells were incubated with 10 mM KCN for 60 min. The cells were then stained with calcein-AM and ethidium homodimer-1. The ester linkage of the former compound is cleaved by esterases present in the cytosol, and the green fluorescent product (calcein) is retained only by viable cells. The nuclear membrane of dead (but not living) cells is permeable to ethidium homerdimer-1. Only very rare cells incubated with the control medium were stained by the red fluorescent ethidium homodimer-1 (Fig.3). The number of dead cells was similarly low following incubation with cytomix in presence or absence of EP. As expected, incubation with KCN killed all of the cells.
Incubation of Cytomix Stimulated Caco-2 Cells with EP Blocks Activation of NF-κB.
Transcriptional activation of iNOS mRNA expression in human enterocytes is partially dependent upon activation of the transcription factor, NF-κB (Salzman et al., 1996; Chavez et al., 1999b). We previously showed that treating mice with EP inhibits NF-κB activation induced by hemorrhagic shock and resuscitation (Yang et al., 2002a). In addition, we also recently reported that EP inhibits NF-κB activation in LPS-stimulated RAW 264.7 murine macrophage-like cells (Ulloa et al., 2002). Accordingly, we sought to determine whether EP blocks NF-κB DNA binding when Caco-2 cells are stimulated with cytomix. Nuclear extracts were prepared from cells incubated for various periods under control conditions, with 5 mM of EP, with cytomix, or with cytomix plus 5 mM EP. Activation of NF-κB was assessed using EMSA. Incubating Caco-2 cells with cytomix for 4 h increased DNA binding of NF-κB; this effect was largely blocked when the cells were exposed to cytomix in the presence of EP (Fig.4A). To confirm the identity of the activated protein-DNA complex, binding assays were carried out with samples that were preincubated with specific Ab directed against p50 and p65, two proteins belonging to the NF-κB family (Bowie et al., 1997). We observed both a super-shifted band and decreased intensity of the NF-κB band with the p65 Ab (Fig. 4B). Moreover, binding of the protein to labeled NF-κB binding element was completely inhibited by a 100-fold excess of unlabeled NF-κB duplex oligonucleotide but not by a similar excess of unlabeled irrelevant (HIF-1) duplex oligonucleotide. We failed to observe a clear super-shift with the anti-p50 Ab; nevertheless, the density of the NF-κB band was somewhat diminished.
Incubation of Cytomix Stimulated Caco-2 Cells with EP Inhibits iNOS Expression and Decreases NO Production.
We previously showed that exposing Caco-2 monolayers to the proinflammatory cytokine, IFN-γ, either alone (Unno et al., 1995) or in combination with TNF-α and IL-1β (Chavez et al., 1999a,b), induces iNOS expression and increases the release of NO by these cells. In view of these findings, we hypothesized that exposing cytomix stimulated Caco-2 cells to ethyl pyruvate might inhibit induction of iNOS mRNA expression and thereby decrease NO production. As shown in Fig.5, A and B, incubation of cytomix-stimulated Caco-2 cells with EP for 24 h decreased steady-state iNOS mRNA levels. In addition, the concentration of NO2− plus NO3− in culture supernatants was significantly decreased when the cells were incubated with cytomix in the presence of EP (Fig. 5C).
EP Inhibits Cytomix-induced Changes in the Expression and Localization of Tight Junction Proteins in Cytomix-Stimulated Caco-2 Monolayers.
The selective barrier to the diffusion of hydrophilic solutes imposed by epithelia is determined in part by the proper formation of tight junctions (zonula occludens) between adjacent cells (Denker and Nigam, 1998; Stevenson, 1999). It is noteworthy, therefore, that our laboratory recently reported that incubating Caco-2 cells with cytomix promotes marked changes in the expression and/or localization of a number of tight junction proteins, including ZO-1 and occludin (Han et al., 2002). Prompted by these findings, we sought to determine whether EP could inhibit the effects of cytomix on the expression and localization of ZO-1 and occludin in Caco-2 cells.
Following 48 h of incubation with cytomix, the expression of ZO-1 in Caco-2 cells decreased to about 50% of the level in cells incubated under control conditions (Fig. 5). Similarly, occludin expression decreased to about 25% of control levels following exposure of Caco-2 cells to cytomix for 48 h. When cells were coincubated with cytomix and 10 mM EP, however, normal levels of both ZO-1 and occludin were preserved.
When Caco-2 monolayers were incubated under control conditions, both ZO-1 and occludin were predominantly localized to the regions of cell-cell boundaries, and immunofluorescent staining for the two proteins was even and continuous (Fig. 6, A and D). Following incubation with cytomix for 48 h, staining for ZO-1 and occludin at cell boundaries became faint, and localization of both of these proteins was more diffuse and discontinuous (Fig.7, B and E). If the Caco-2 monolayers were incubated for 48 h with cytomix in the presence of 10 mM of EP, however, then normal staining patterns for both proteins were largely preserved (Fig. 7, C and F).
EP Protects against LPS-Induced Ileal Mucosal Hyperpermeability and Inhibits Bacterial Translocation and Hepatocellular Injury in Mice.
It is well established that the systemic inflammatory response caused by injecting mice with LPS is associated with marked derangements in gut mucosal barrier function (Deitch et al., 1987; Unno et al., 1997c; Sappington et al., 2002). Therefore, we sought to follow-up our observations made in vitro by examining the effects of EP on LPS-induced alterations in gut barrier function in vivo. Groups of mice were challenged with LPS suspended in PBS or PBS alone. Mice in one of the groups (LPS + EARLY EP) were treated with 40 mg/kg doses of EP at 1, 6 and 12 h after the injection of LPS. Mice in another group (LPS + LATE EP) were treated with 40 mg/kg doses of EP at 6 and 12 h after the injection of LPS. Mice in the LPS + RLS group were treated with the same volume of fluid as the animals in the LPS + EARLY EP group, but rather than receiving a solution of EP, they received RLS instead. Mice in the PBS group were not injected with LPS but received only the vehicle. All assays were performed 18 h after injecting LPS or PBS. As expected, injecting mice with LPS increased ileal mucosal permeability to FD4 (Fig. 8A) and promoted bacterial translocation to mesenteric lymph nodes (Fig. 8B). Treatment with EP, however, ameliorated both of these LPS-induced derangements in gut barrier function, irrespective of whether the delay before starting therapy was 1 or 6 h. Injecting mice with LPS also increased circulating levels of the hepatocellular enzyme, ALT, indicative of damage to the hepatic parenchyma (Fig.8C). This deleterious effect of LPS administration was significantly ameliorated by EP, irrespective of the dosing regimen used.
Discussion
Inflammatory conditions are associated with marked alterations in the barrier function of the intestinal epithelium. For example, the permeability of cultured monolayers of human intestinal epithelial cell lines, such as T84 or Caco-2, is markedly increased following incubation with the proinflammatory cytokine IFN-γ whether used alone (Madara and Stafford, 1989; Adams et al., 1993; Youakim and Ahdieh, 1999) or in combination with other proinflammatory cytokines (Chavez et al., 1999a; Zolotarevsky et al., 2002). Similarly, the induction of localized (Jijon et al., 2000) or systemic (Deitch et al., 1987l; Unno et al., 1997c; Sappington et al., 2002) inflammation promotes intestinal mucosal barrier dysfunction in experimental animals.
In the present report, we showed that a simple compound, EP, ameliorated derangements in intestinal epithelial barrier function whether induced by cytomix in vitro or the proinflammatory bacterial product, LPS, in vivo. In addition, we showed that exposing Caco-2 monolayers to cytomix caused marked alterations in the expression and localization of two key tight junction proteins, ZO-1 and occludin; however, these cytokine-induced changes were largely prevented by EP. Thus, EP was shown to markedly attenuate many of the structural and functional derangements in gut epithelial barrier function induced by an inflammatory milieu.
Proinflammatory cytokines trigger activation of the transcription factor NF-κB in human intestinal epithelial cell lines (Salzman et al., 1996; Parikh et al., 1997, 2000). In this study, we showed that 10 mM EP almost completely blocked cytomix-induced NF-κB activation. This observation replicated similar findings recently reported by Ulloa et al. (2002), who showed that EP inhibits NF-κB activation in LPS-stimulated RAW 264.7 murine macrophage-like cells. Agents such as pyrrolidine dithiocarbamate, which block NF-κB activation, inhibit iNOS induction in cultured intestinal epithelial cells (Salzman et al., 1996; Cavicchi and Whittle, 1999; Chavez et al., 1999b). In the present study, we showed that adding EP to cytomix prevented the increase in steady-state iNOS mRNA levels and the increase in NO production, which were observed when the cells were incubated with cytomix alone.
Although it is possible that the salutary effects of EP on intestinal epithelial barrier function were caused by factors other than or in addition to inhibition of iNOS induction, we think this mechanism is quite likely to be important. Certainly, several lines of evidence support the view that increased production of NO secondary to induction of the enzyme iNOS is a key factor responsible for cytokine-induced epithelial barrier dysfunction in vitro. For instance, various NO donors, including sodium nitroprusside (Salzman et al., 1995),S-nitroso-N-acetylpenicillamine (Salzman et al., 1995; Menconi et al., 1998), and 3-morpholinosydnonimine (Unno et al., 1997b), are known to be capable of increasing the permeability of Caco-2 monolayers. Furthermore, iNOS expression is induced when human enterocyte-like cells are exposed to IFN-γ, particularly in combination with IL-1β and TNF-α (Unno et al., 1995; Linn et al., 1997; Chavez et al., 1999b). Most important, the development of hyperpermeability following the incubation of Caco-2 cells with IFN-γ (Unno et al., 1995) or IFN-γ plus IL-1β and TNF-α (i.e., cytomix) (Chavez et al., 1999a) is prevented by pharmacological agents that inhibit iNOS or scavenge NO.
The adverse effects of NO on intestinal barrier function probably are not caused by the diatomic molecule per se, but rather by a highly reactive derivative, peroxynitrite, formed when NO reacts with superoxide radical anion (Radi et al., 2001). This view is supported by data showing that peroxynitrite or superoxide scavengers ameliorate epithelial hyperpermeability induced by exogenous NO donors (Unno et al., 1997b; Menconi et al., 1998) or cytomix (Chavez et al., 1999a). This notion is further supported by studies showing that mild acidosis exacerbates IFN-γ (Unno et al., 1999) or NO⨪-induced hyperpermeability (Unno et al., 1997a,b), presumably because these conditions favor formation of peroxynitrous acid, which behaves in many ways like the highly reactive hydroxyl radical.
The deleterious effects of NO (or various closely related reactive nitrogen species) on intestinal epithelial barrier function are not limited to highly reductionist in vitro systems. We (Unno et al., 1997c) and others (Chen et al., 1996; Mishima et al., 1998) have shown that pharmacological inhibition of iNOS ameliorates LPS-induced hyperpermeability and/or bacterial translocation in rodents. Confirmatory findings have been reported from studies using (iNOS−/−) mice with targeted disruption of the iNOS gene (Mishima et al., 1997). Along these same lines, our laboratory recently reported that high mobility group-1 B box, a novel proinflammatory cytokine-like molecule, induces ileal mucosal hyperpermeability and bacterial translocation in wild-type but not iNOS−/− mice (Sappington et al., 2002).
Since normal epithelial permeability is maintained and regulated by the tight junctions between adjacent cells (Denker and Nigam, 1998;Stevenson, 1999), a number of laboratories have investigated the effects of NO and/or proinflammatory cytokines on the structure and function of tight junctions. Tight junctions are closely linked to the actin-based cytoskeleton, and our laboratory showed that incubating Caco-2 cells with SNP leads to marked alterations in the immunofluorescent staining of F-actin (Salzman et al., 1995). Subsequently, Youakim and Ahdieh (1999) reported that incubating T84 cells with IFN-γ results in an almost total disappearance of ZO-1. In another study, Cuzzocrea et al. (2000) exposed Madin-Darby canine kidney epithelial cells to a proinflammatory microbial product, zymosan, to induce iNOS expression and showed that NO production was associated with decreased expression of occludin. In parallel in vivo studies, these investigators showed that intraperitoneal administration of zymosan increased ileal mucosal permeability in wild-type (iNOS+/+) mice but not in iNOS−/− mice. Immunofluorescent studies showed marked disruption of ZO-1 and occludin staining patterns in zymosan-challenged iNOS+/+ but not in iNOS−/− mice. These findings support the view that inflammation-induced intestinal epithelial barrier dysfunction is related to altered expression and organization of key tight junction proteins secondary to increased iNOS-dependent NO production. In the present study, many of the elements outlined above—including cytomix-stimulated changes in iNOS expression, NO production, ZO-1, and occludin expression and localization, and epithelial hyperpermeability—were markedly inhibited by EP.
To explain the salutary effects of EP, it is tempting to postulate that scavenging of ROS is the primary mechanism. Pyruvate is an effective scavenger of H2O2 (Bunton, 1949; Melzer and Schmidt, 1988) and hydroxyl radical (Dobsak et al., 1999), and some data support the view that EP is also an effective antioxidant (Varma et al., 1998; Tawadrous et al., 2002). Moreover, redox-mediated events are widely regarded as being important in the activation of NF-κB. This view is supported by numerous studies showing that H2O2 triggers activation of NF-κB in cultured cells (Schoonbroodt et al., 2000;Livolsi et al., 2001; Rahman et al., 2001) and various antioxidants block NF-κB activation in cytokine- or LPS-stimulated cells (Schreck et al., 1992; Oka et al., 2000; Ma and Kinneer, 2002).
These arguments notwithstanding, it seems more probable to us that the beneficial effects of EP are not directly redox-mediated. Several considerations prompt this view. First, the culture medium used by us for growing and maintaining Caco-2 cells contained 2 mM pyruvate, a known ROS scavenger as already mentioned, and yet this medium failed to prevent cytomix-induced hyperpermeability, NF-κB activation, or iNOS induction. Indeed, even when we added additional pyruvate to increase the final pyruvate concentration to as high as 12 mM, we saw no evidence of protection against the increase in epithelial permeability induced by incubation with cytomix. Second, previously reported data suggest that oxidant stress (1–10 mM H2O2) fails to activate NF-κB in Caco-2 (Parikh et al., 2000) or DLD-1 (Salzman et al., 1996) enterocyte-like cells. Furthermore, two ROS scavengers, pyrrolidine dithiocarbamate and dimethyl sulfoxide, fail to block IL-1β-induced NF-κB activation in Caco-2 cells (Parikh et al., 2000). Thus, it seems unlikely that EP blocked NF-κB activation in cytomix-stimulated cells by scavenging ROS. Third, investigators working in an unrelated field—the regulation of insulin secretion by pancreatic islet cells—have reported divergent pharmacological effects of pyruvate, on the one hand, and a pyruvate ester, methyl pyruvate (MP), on the other hand. Specifically, it was shown that MP stimulates insulin secretion by isolated pancreatic islets (Mertz et al., 1996; Zawalich and Zawalich, 1997), whereas pyruvate is not insulinogenic (Sener et al., 1978). To explain the differential effects of these two closely related compounds, it was speculated that the more lipophilic compound, MP, might penetrate the mitochondrial matrix better than pyruvate and thereby support supranormal rates of ATP production. However, recently reported data refute this hypothesis and suggest that pyruvate and MP have distinct biochemical effects in pancreatic β-cells that are unrelated to ATP biosynthesis (Lembert et al., 2001). Although the effects of EP on insulin secretion by pancreatic islet cells have not been reported, it seems probable that the pharmacological effects of EP and MP are similar. In any case, the data obtained by comparing the effects of MP and pyruvate on insulin secretion by cultured islet cells support the view that these two compounds have very different pharmacological actions. The islet cell data just cited, along with the data presented here, support the view that further studies are warranted to better understand the biochemical bases for the distinct pharmacological actions of pyruvate anion and simple aliphatic esters of pyruvic acid.
Footnotes
-
This work was supported by grants from the National Institutes of Health (GM53789, GM37631, and GM58484) and DARPA (N65236-00-1-5434).
-
DOI: 10.1124/jpet.102.043182
- Abbreviations:
- ROS
- reactive oxygen species
- EP
- ethyl pyruvate
- RLS
- Ringer's lactate solution
- NF-κB
- nuclear factor-κB
- iNOS
- inducible nitric oxide synthase
- TNF-α
- tumor necrosis factor-α
- IL
- interleukin
- LPS
- lipopolysaccharide
- IFN-γ
- interferon-γ
- DMEM
- Dulbecco's minimal essential medium
- PBS
- phosphate-buffered saline
- FBS
- fetal bovine serum
- Ab
- antibody
- FD4
- fluorescein isothiocyanate-labeled dextran
- PMSF
- phenylmethylsulfonyl fluoride
- DTT
- dithiothreitol
- EMSA
- electrophoretic mobility shift assay
- RT-PCR
- reverse transcription-polymerase chain reaction
- TBS
- Tris-buffered saline
- PBST
- PBS containing 0.02% Tween 20
- ALT
- alanine aminotransferase
- KHBB
- Krebs-Henseleit bicarbonate buffer
- MP
- methyl pyruvate
- NO2−
- nitrite
- NO3−
- nitrate
- CYTO
- cytomix
- KCN
- potassium cyanide
- Received August 14, 2002.
- Accepted September 25, 2002.
- The American Society for Pharmacology and Experimental Therapeutics