INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

UNDER NORMAL CONDITIONS, the gastrointestinal (GI) mucosa is a highly selective barrier that prevents the passage of toxic proinflammatory molecules from the gut lumen into the mucosa and the circulation (12, 20, 27). Under abnormal conditions, loss of GI barrier integrity can permit the penetration of normally excluded luminal substances (e.g., endotoxin and microbes) into or across the mucosa, which can lead to the initiation and/or perpetuation of inflammatory processes and mucosal injury. This injury and the ensuing loss of mucosal barrier integrity have been implicated in the pathophysiology of a wide range of inflammatory disorders including inflammatory bowel disease (IBD) (12, 16, 20, 27). The pathogenesis of mucosal barrier dysfunction in IBD remains poorly understood, but several studies, including ours (2, 6, 8), have shown that chronic gut inflammation is associated with high levels of reactive oxygen metabolites (ROMs) and that these oxidants appear to be involved in causing mucosal barrier dysfunction (2, 8, 26, 30). However, the precise biochemical mechanisms have not been established.

While investigating these mechanisms with monolayers of intestinal cells as a model of barrier function, we (2, 5-8) demonstrated that oxidant-induced barrier disruption is dependent on the disruption of the cytoskeleton. For example, we observed that loss of barrier integrity required disruption of the polymerized (F)-actin cytoskeleton. The idea that cytoskeletal instability in general and actin filament disruption in particular could be major contributing factors to loss of barrier integrity is consistent with other studies in which we have shown the importance of cytoskeletal stability in GI healing in vivo (3, 4) as well as in vitro (2, 5-10a, 25). It is also consistent with the known roles of actin in cell function. Actin is one of the most abundant proteins in the eukaryotic cells, with the ability to polymerize into filaments of highly dynamic -double helices (11, 38). In epithelial cells, such filaments constitute a dense cross-linked actin cortex located on the inner side of the plasma membrane. The actin cytoskeleton also contains filamentous stress fibers that traverse the cytosol and short fibers that extend into the lamellipodia in motile cells. This structural component is essential in maintaining normal cellular physiology, structure, locomotion, and support functions (2, 11, 29, 38, 39).

Our current investigation also derives from recent reports that elevated levels of peroxynitrite (ONOO) may be an essential factor in tissue injury during IBD (23, 34, 35). Indeed, studies from our laboratory on ethanol-induced intestinal injury, as well as those from other laboratories, have shown that inducible nitric oxide (NO) synthase (iNOS) activation can lead to NO overproduction (8, 14, 36, 40) and that the injurious effects of NO overproduction appear to be mediated by ONOO, the product of the reaction of NO with superoxide anions (O·) (7, 18, 22, 34). Accordingly, the objectives of the current study were to determine 1) whether disruption of the actin cytoskeleton and loss of barrier integrity after exposure to oxidants such as H₂O₂ are associated with elevated iNOS activity and with elevated levels of reactive nitrogen metabolites (RNMs) such as NO and ONOO, 2) whether exposure of cells to these RNMs mimics the effects of oxidants, and 3) whether agents that scavenge these RNMs (antioxidants) or inhibit their formation (iNOS inhibitors) are protective.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Cell culture. Intestinal Caco-2 cells (American Type Culture Collection, Manassas, VA) grown for barrier integrity work were split at a ratio of 1:2 and seeded at a density of 200,000 cells/cm² in 0.4-µm collagen I cell culture inserts (0.3-cm² growth surface; Biocoat, Becton Dickinson Labware, Bedford, MA), and experiments were performed at least 7 days after confluence. The utility, maintenance, and characterization of this cell line and the preparation of the monolayers of Caco-2 cells have been previously described (9, 17, 39).

Determination of epithelial barrier function by fluorometry. The barrier integrity of Caco-2 monolayers was determined by measuring the apical-to-basolateral flux of fluorescein sulfonic acid [(FSA); 200 µg/ml; 478 Da; Molecular Probes, Eugene, OR] as previously described (2, 6, 39). After the treatments, fluorescent signals from the samples were quantitated with a fluorescence multiplate reader (FL 600, Bio-Tek Instruments) with the excitation and emission spectra for FSA set as excitation = 485 nm and emission = 530 nm. Clearance (Cl) was calculated with the formula Cl (nl · h¹ · cm²) = F_ab/([FSA]_a × S), where F_ab is the apical-to-basolateral flux of FSA in light units per hour, [FSA]_a is the concentration at baseline in light units per nanoliter, and S is the surface area (0.3 cm²) (2, 8). Simultaneous controls were performed with each experiment.

Assay of NOS activity. Cells grown to confluence were removed by scraping, centrifuged, and homogenized on ice in a buffer containing 50 mM Tris · HCl, 0.1 mM EDTA, 0.1 mM EGTA, 12 mM 2-mercaptoethanol, and 1 mM phenylmethylsulfonyl fluoride (pH 7.4). The conversion of L-[³H]arginine (L-Arg; Amersham, Arlington Heights, IL) to L-[³H]citrulline was measured in the homogenates by scintillation counting as described previously (7, 8, 36). As we previously reported (7, 8), experiments in the absence of NADPH and the presence of the NOS inhibitor N^G-nitro-L-arginine (1 mM) were used to assess the extent of L-[³H]citrulline formation that was independent of any NOS activity. Experiments in the presence of NADPH, without Ca²⁺ and with 5 mM EGTA, determined Ca²⁺-independent NOS (iNOS) activity. Experiments in the presence of NADPH and Ca²⁺ determined Ca²⁺-dependent NOS [constitutive NOS (cNOS)] activity. In selected experiments, we added the isoform-selective iNOS inhibitor L-N⁶-(1-iminoethyl)lysine (L-NIL, 1 mM). Protein concentrations were determined by the Bradford method (13).

Western blot analysis of the level of iNOS protein. After treatment, the cells were washed once with cold PBS, scraped in 1 ml of cold PBS, and harvested in a standard anti-protease cocktail. For immunoblotting, samples (25 µg protein/lane) were added to SDS buffer (250 mM Tris · HCl, pH 6.8, 2% glycerol, and 5% mercaptoethanol), boiled for 5 min, and then separated on 7.5% SDS-PAGE gels. Subsequently, the proteins were transferred to nitrocellulose membranes and blocked in 3% BSA for 1 h, followed by several washes in Tris-buffered saline. The immunoblotted proteins were incubated for 2 h in Tween 20, Tris-buffered saline, and 1% BSA with the primary antibody (mouse monoclonal anti-human iNOS at 1:3,000 dilution; Santa Cruz Biotechnology, Santa Cruz, CA). A horseradish peroxidase (HRP)-conjugated goat anti-mouse antibody (Molecular Probes) was used as a secondary antibody at 1:3,000 dilution. The membranes were visualized by enhanced chemiluminescence (Amersham) and autoradiography (7, 37).

Chemiluminescence analysis of NO concentration in cultures. NO production was assessed by a novel chemiluminescence procedure (1, 7). Briefly, cells were homogenized by sonication, and the endogenous nitrate (NO) and nitrite (NO) and the metabolic degradation products of NO were then reduced to NO with the use of vanadium(III) (Sigma, St. Louis, MO) and HCl at 90°C before the measurement of NO concentration by chemiluminescence analysis. Chemiluminescence was measured with a Sievers NO 280 analyzer (NOA, Boulder, CO). NO (expressed in µM) was calculated by comparison with the chemiluminescence of a standard solution of NaNO₂. The absolute NO values were reported as micromoles per 10⁶ cells.

Determination of cell oxidative stress. Oxidative stress was assessed by measuring the conversion of a nonfluorescent compound, 2',7'-dichlorofluorescein diacetate (DCFD; Molecular Probes), into the fluorescent dye dichlorofluorescein (DCF) as previously described (8). The dependence of the assay on O· generation was shown by adding an active superoxide radical scavenger, superoxide dismutase (SOD, 300 U/ml), or an inactive superoxide radical scavenger, heat-inactivated SOD (iSOD). Briefly, monolayers grown in 96-well plates were preincubated with the membrane-permeant DCFD (10 µg/ml for 30 min) before the subsequent treatments. Fluorescent signals (i.e., DCF fluorescence) from the samples were quantitated with a fluorescence multiplate reader set at an excitation wavelength of 485 nm and an emission wavelength of 530 nm.

Immunofluorescent staining and high-resolution laser scanning confocal microscopy of the actin cytoskeleton. Cells from monolayers were fixed in cytoskeleton stabilization buffer and then postfixed in 95% ethanol as previously described (2, 9). Monolayers of cells were subsequently processed for incubation with FITC-phalloidin (specific for F-actin staining; Sigma) at 1:40 dilution for 1 h at 37°C and then were mounted in Aquamount. The samples were examined by both standard fluorescence microscopy and ultra high-resolution laser scanning confocal microscopy (LSCM; Carl Zeiss). Cell monolayers on slides were analyzed in a blinded fashion using LSCM with a ×63 oil immersion plan-apochromat objective, NA 1.4 (Zeiss). An argon laser (wavelength = 488 nm) was used to examine FITC-labeled cells, and the cytoskeletal elements were examined for their overall morphology, orientation, and disruption as previously described (2, 9).

Actin fractionation and quantitative Western immunoblotting of F-actin and monomeric actin. F-actin and monomeric (G)-actin were isolated as we previously described (2, 5). Briefly, cells were pelleted by centrifugation at low speed (700 rpm for 7 min at 4°C) and resuspended in actin stabilization-extraction buffer (0.1 M PIPES, pH 6.9, 30% glycerol, 5% DMSO, 1 mM MgSO₄, 10 µg/ml of a standard anti-protease cocktail, 1 mM EGTA, and 1% Triton X-100) at room temperature for 20 min. F- and G-actin were separated after a series of ultracentrifugation and extraction steps. Fractionated actin samples were flash frozen in liquid N₂ and stored at 70°C until immunoblotting was performed. For immunoblotting, samples (5 µg of protein) were placed in a standard SDS buffer, boiled for 5 min, and then subjected to PAGE (7.5% gel) (9). To quantify the relative levels of actin, the optical density (OD) of the bands corresponding to the immunoradiolabeled actin was measured with a laser densitometer (2).

Immunoblotting determination of actin oxidation and actin nitration. Oxidation and nitration of the actin cytoskeleton were assessed, respectively, by measuring protein carbonyl and nitrotyrosine formation. Carbonylation and nitrotyrosination of actin were determined in a manner similar to the quantitative blotting of actin (2, 8). To avoid the unwanted oxidation of the actin samples, all buffers contained 0.5 mM dithiothreitol and 20 mM 4,5-dihydroxy-1,3-benzenedisulfonic acid (Sigma). To determine the carbonyl content, samples were blotted to a polyvinylidene difluoride membrane and then subjected to successive 5-min incubations in 2 N HCl and 2,4-dinitrophenylhydrazine (DNPH; 100 µg/ml in 2 N HCl; Sigma). Membranes were then washed three times in 2 N HCl and subsequently washed seven times in 100% methanol (5 min each), followed by blocking for 1 h in 5% BSA in 10× PBS-Tween 20 (PBS-T). Immunologic evaluation of carbonyl formation was performed for 1 h in 1% BSA-PBS-T buffer containing anti-DNPH (1:25,000 dilution; Molecular Probes). The membranes were then incubated with a HRP-conjugated secondary antibody (1:4,000 dilution; Molecular Probes) for 1 h. To determine nitrotyrosine content, after the blocking step listed above (i.e., BSA-PBS-T buffer), the membranes were probed for nitrotyrosine by incubation with 2 µg/ml of monoclonal anti-nitrotyrosine antibody for 1 h (Upstate Biotechnology, Lake Placid, NY) followed by the HRP-conjugated secondary antibody as described for carbonylation. The wash steps and film exposure were as in a standard Western protocol (8, 9). The relative levels of oxidized or nitrated actin were then quantified by measuring the OD of the bands corresponding to anti-DNPH or anti-nitrotyrosine immunoreactivity with a laser densitometer. Immunoreactivity was expressed as the percentage of carbonyl or nitrotyrosine formation (OD) in the treatment group compared with the maximally oxidized or nitrated tubulin standard. Oxidized or nitrated tubulin standards were run concurrently with the corresponding treatment groups.

Statistical analysis. Data are presented as means ± SE. All experiments were carried out with a sample size of at least 4-6 observations/group. Statistical analysis between or among groups was carried out with analysis of variance followed by Dunnett's multiple-range test (19). Correlational analyses were done with the Pearson test for parametric analysis or, when applicable, the Spearman test for nonparametric analysis. A P value < 0.05 was deemed to represent statistical significance.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Evidence that oxidant-induced leakiness of the intestinal barrier involves activation of iNOS. We first confirmed our earlier finding (2) that exposure of Caco-2 cell monolayers for 30 min to increasing concentrations of H₂O₂ causes hyperpermeability of the intestinal barrier in the monolayers in a dose-dependent manner. This is indicated by the increased clearance of FSA (Fig. 1). We now show that preincubation (1 h) with a selective iNOS inhibitor (L-NIL, 1 mM) significantly attenuates (72%) this effect. This inhibition was significantly lower (50%) at the higher oxidant doses: FSA clearance (in nl · h¹ · cm²) = 1,287 ± 72 for L-NIL + 5 mM H₂O₂ vs. 2,589 ± 89 for 5 mM H₂O₂ alone. A substrate for NOS, L-Arg (3 mM, 48-h exposure), by itself did not significantly affect permeability, but it did synergize with a nondamaging concentration of H₂O₂ (0.05 mM) to disrupt monolayer barrier integrity. Moreover, L-Arg potentiated the loss of monolayer barrier integrity in the presence of damaging H₂O₂ concentrations (0.5 mM H₂O₂ is shown). In both cases, potentiation was prevented by L-NIL.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   Injurious effects of H2O2 and protective effects of pretreatment with L-N6-(iminoethyl)lysine (L-NIL). L-NIL, a selective inhibitor of inducible nitric oxide synthase (iNOS), protected against loss of Caco-2 monolayer barrier integrity as assessed by fluorescein sulfonic acid (FSA) clearance. Monolayers (n = 6 observations/group for Figs. 1-11) were pretreated with 1 mM L-NIL 1 h before subsequent exposure to noninjurious (0.05 mM) or injurious (0.5 mM) concentrations of H2O2 for 30 min. In selected experiments, monolayers were preincubated with L-arginine (L-Arg) 48 h before the aforementioned treatments. FSA clearance was calculated as apical to basolateral flux of FSA divided by the concentration of probe in the apical chamber. *P < 0.05 compared with vehicle or 0.05 mM H2O2. +P < 0.05 compared with 0.5 mM H2O2 alone. ++P < 0.05 compared with L-Arg + H2O2. &P < 0.05 compared with L-NIL + H2O2. #P < 0.05 compared with the corresponding 0.05 mM or 0.5 mM H2O2 alone.

Evidence that NO, oxidative stress, and ONOO are also involved in oxidant-induced monolayer barrier dysfunction. Pretreatment of monolayers with the NO and ONOO scavengers urate (66%) and L-cysteine (74%) or the O· scavenger SOD (72%), which is similar to L-NIL, significantly attenuated H₂O₂-induced monolayer hyperpermeability (Fig. 2). Pretreatment with iSOD was not protective. iSOD by itself did not injure cells (clearance = 28 ± 9 vs. 22 ± 8 nl · h¹ · cm² for vehicle). The failure to elicit 100% protection by the aforementioned antioxidants was not due to technical problems because the addition of catalase, an H₂O₂ scavenger (1,000 U/ml), elicited 99% protection.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Protective effect of antioxidants against H2O2-induced disruption of barrier integrity in Caco-2 monolayers. Permeability of monolayers was again assessed by FSA clearance. Monolayers were pretreated with 0.5 mM urate, 1 mM L-cysteine, or 300 U/ml superoxide dismutase (SOD; a superoxide anion scavenger) or heat-treated (inactive) SOD (iSOD), before exposure to H2O2. Clearance was determined as in Fig. 1. *P < 0.05 compared with vehicle. +P < 0.05 compared with H2O2 alone. ++P < 0.05 compared with SOD-pretreated monolayers.

1

1

View larger version (14K): [in this window] [in a new window]   Fig. 3.   A: upregulation of iNOS activity induced by H2O2 in Caco-2 monolayers. Cell monolayers were pretreated for 1 h with L-NIL (1 mM) or vehicle and subsequently exposed to H2O2 for 30 min. *P < 0.05 compared with vehicle. +P < 0.05 compared with H2O2. B: effects of H2O2 on iNOS protein levels in Caco-2 cell monolayers as shown by Western immunoblotting. Monolayers were exposed to nondamaging (0.05 mM) or damaging (0.5 mM) H2O2 or control (vehicle) for 30 min. Damaging concentrations of H2O2 resulted in a large increase in levels of iNOS protein in cell fractions. In contrast, control and nondamaging concentrations of H2O2 were associated with a normal and basal level of iNOS (~130 kDa). Western blots of cell monolayer fractions were processed sequentially with monoclonal anti-human iNOS and horseradish peroxidase-conjugated secondary antibodies. The region of gel shown was between the 126,000 and 218,000 prestained molecular weights, which were run in adjacent lanes; ~ denotes approximate molecular weight for iNOS. C: concentrations of nitric oxide (NO) in the supernatant of homogenates of intestinal cell monolayers exposed to H2O2. NO levels were assessed by a sensitive chemiluminescence assay (see MATERIALS AND METHODS). Conditions were identical to those in A. *P < 0.05 compared with vehicle. +P < 0.05 compared with H2O2.

1

1

View larger version (13K): [in this window] [in a new window]   Fig. 4.   Time course for induction of iNOS and increases in NO. Data are means ± SE. Units: pmol/mg protein for iNOS activity; 103 × optical density for iNOS protein levels; µmoles/106 cells for NO levels. Cells were exposed to 0.5 mM H2O2 at time 0.

View larger version (19K): [in this window] [in a new window]   Fig. 5.   Oxidative stress in cell monolayers that is induced by H2O2 is attenuated by the iNOS inhibitor L-NIL or by antioxidants (L-cysteine or SOD). Oxidative stress was assessed by the dichlorofluorescein (DCF) fluorescence assay. Conditions were as in Figs. 2 and 3A. *P < 0.05 compared with vehicle. +P < 0.05 compared with H2O2. ++P < 0.05 compared with SOD-preexposed monolayers.

NO- and ONOO-dependent mechanisms in the loss of F-actin cytoskeletal integrity. Pretreatment of Caco-2 monolayers with L-NIL or with the above noted antioxidants protected the F-actin cytoskeleton against most of the H₂O₂-induced damage, an outcome that was quantitated with the use of LSCM and by calculating changes in the percentage of cells displaying normal actin (Fig. 6). A 0.5 mM dose of H₂O₂ decreased the proportion of cells showing normal actin by 49%. The extent of this damage was inhibited 90% by L-NIL, 84% by urate, 81% by L-cysteine, and 88% by SOD. These effects on the actin cytoskeleton by L-NIL and antioxidants paralleled their protective effects on barrier function.

View larger version (20K): [in this window] [in a new window]   Fig. 6.   Protective effects of L-NIL or antioxidants against oxidant-induced actin cytoskeletal disruption. L-NIL- or antioxidant-pretreated cells maintained a high percentage of Caco-2 cells displaying normal F-actin cytoskeleton in monolayers exposed to oxidant (H2O2) as assessed by high-resolution laser scanning confocal microscopy (LSCM). Conditions were as in Figs. 2 and 3A. *P < 0.05 compared with vehicle. +P < 0.05 compared with H2O2. ++P < 0.05 compared with SOD-pretreated monolayers.

View larger version (73K): [in this window] [in a new window]   Fig. 7.   Fluorescent staining of the F-actin cytoskeleton assessed by high-resolution LSCM in intestinal cell monolayers incubated with isotonic saline (control; a), 0.5 mM H2O2 (b), or 1 mM L-NIL (c) followed by H2O2. Controls revealed a normal, continuous, and smooth architecture of the actin ring or cortex at areas of cell-cell contact. In contrast, H2O2-exposed monolayers exhibited areas of actin disruption, beading, and fragmentation. In cells pretreated with L-NIL, actin appeared normal and resembled the morphology detected in the control monolayers. Bar, 25 µm.

View larger version (21K): [in this window] [in a new window]   Fig. 8.   A: quantitative immunoblotting analysis of the levels of polymerized F-actin (an index of stability) and monomeric G-actin (an index of disassembly) in Caco-2 cell monolayers. Monolayers were preincubated with L-NIL or antioxidants before incubation with 0.5 mM H2O2. Conditions were similar to those in Figs. 2 and 3A. Actin fractions were analyzed by SDS-PAGE and processed for autoradiography and then for densitometry to obtain the optical density of the bands corresponding to immunoradiolabeled actin. Percent polymerization of actin = [F/total actin pool (F + G)]. *P < 0.05 compared with vehicle. +P < 0.05 compared with H2O2. ++P < 0.05 compared with SOD-preincubated monolayers. B: Western immunoblot photomicrograph of the F-actin (Triton insoluble) extracts after identical treatments as in A. F-actin fractions were analyzed by SDS-PAGE and Western immunoblots with monoclonal anti-actin as primary antibody followed by horseradish peroxidase-conjugated secondary antibody and subsequently processed for X-ray film exposure. The corresponding optical densities for F-acting fractions are, from right to left: Control (isotonic saline), 11,135; 0.5 mM H2O2, 7,843; L-NIL + 0.5 mM H2O2, 10,369; actin standard (43 kDa); second actin standard; SOD + 0.5 mM H2O2, 10,094; iSOD + 0.5 mM H2O2, 7,903; L-cysteine + 0.5 mM H2O2, 10,232.

View larger version (15K): [in this window] [in a new window]   Fig. 9.   A: immunoblot analysis of both anti-dinitrophenylhydrazone (DNP) and anti-nitrotyrosine immunoreactivity of the cytoskeletal F-actin pool in Caco-2 monolayers exposed to H2O2. These represent, respectively, carbonylation (oxidation or anti-DNP) or nitration of actin. Oxidation or nitration was expressed as the following fraction: carbonyl or nitrotyrosine formation (i.e., optical density) in the treatment group divided by the oxidized (or nitrated) actin standard. *P < 0.05 compared with vehicle. B: quantitative immunoblotting showing the protective effects of L-NIL and antioxidants against H2O2-induced nitration of polymerized F-actin in Caco-2 cells. Conditions as in Figs. 2 and 3A. Nitration was expressed as described in A. *P < 0.05 compared with vehicle. +P < 0.05 compared with H2O2. ++P < 0.05 compared with SOD-pretreated monolayers.

ONOO compounds mimic the ability of H₂O₂ to cause F-actin cytoskeletal instability and barrier disruption. If ONOO mediates oxidant-induced damage, then chemically authentic ONOO or compounds capable of generating ONOO should mimic H₂O₂ and cause similar disruption of the F-actin cytoskeleton and monolayer barrier, and these effects should be inhibited by the same antioxidants that inhibit the disruption that is caused by H₂O₂. First, we found that ONOO and ONOO generators [e.g., 3-morpholinosydnonimine (SIN-1), a NO and O· donor, and S-nitroso-N-acetylpenicillamine, a NO donor] in combination with xanthine plus xanthine oxidase (an O· donor) significantly and dose dependently disrupted the monolayer barrier (data not shown), which confirmed our previous findings (7) that these ONOO compounds increase FSA clearance. ONOO generator systems were then added to the cell culture media at a pH of 7.4. To promote the stability of authentic added ONOO in solution, ONOO (180 mM stock in 0.3 M NaOH) was added to the cell culture media to a final pH of 7.6. Pilot studies confirmed that there were no adverse effects of a pH of 7.6 on the cytoskeleton or on monolayer barrier function. We also found that antioxidants such as L-cysteine, urate, and SOD significantly prevented barrier disruption resulting from ONOO mimetics (data not shown), confirming our previous findings.

View larger version (17K): [in this window] [in a new window]   Fig. 10.   Disassembly and disruption of the actin induced by chemically synthesized peroxynitrite (ONOO) and ONOO donors (A) and its prevention by antioxidants (B) in Caco-2 monolayers as determined by immunoblotting analysis of F- and G-actin. Monolayers were preincubated with antioxidants or in control conditions and then exposed to authentic ONOO (30 min) or ONOO generators, 3-morpholinosydnonimine (SIN-1, 1 h) or a combination of S-nitroso-N-acetyl penicillamine (SNAP, 6 h) plus xanthine (X, 1 mM) and xanthine oxidase (XO, 100 mU/mL). Percent polymerization of actin = [F / total actin pool (F + G)]. *P < 0.05 compared with vehicle. +P < 0.05 compared with ONOO or SIN-1 alone. ++P < 0.05 compared with SOD pretreated.

View this table: [in this window] [in a new window]   Table 1.   ONOO compound-induced cytoskeletal F-actin oxidation and nitration and its prevention by antioxidants

View larger version (75K): [in this window] [in a new window]   Fig. 11.   Fluorescent staining of F-actin cytoskeleton by fluorescein-conjugated phalloidin assessed by high-resolution laser confocal microscopy from intestinal cell monolayers incubated with (a) isotonic saline (control), (b) 0.1 mM ONOO, or (c) L-cysteine and then ONOO. ONOO exposure produced the disruption, condensation, and beading of the actin "ring" of the cytoskeleton. In contrast, the actin structure in anti-oxidant (L-cysteine) pretreated cells appears intact and resembles the normal architecture seen in controls. Bar = 25 µm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Together, our present findings support our main conclusion that the H₂O₂-induced disruption of the F-actin cytoskeleton of Caco-2 cells and the consequent disruption of the permeability barrier of Caco-2 monolayers require activation of an iNOS-driven pathway and increased levels of RNMs such as NO and ONOO, which appear to mediate this damage through nitration and oxidation of a 43-kDa actin molecule. A second conclusion, and a novel finding, is that exposure of these intestinal cells to oxidants, a process that models the oxidative stress that occurs during the acute IBD attack, can, surprisingly, further increase cellular synthesis of RNMs and ROMs. A third conclusion, also novel, is that oxidants such as H₂O₂, HOCl, and ONOO can rapidly upregulate iNOS enzyme activity and NO levels.

Our primary conclusion is relevant to IBD. It extends our previous investigation into the role of oxidants in the pathophysiological mechanisms of this disease. Although the primary etiology of IBD is multifactorial, gut leakiness and diffusion of intraluminal proinflammatory antigens (e.g., bacterial products) are considered to be reasonable initial steps for subsequent intestinal inflammation. Similarly, in our monolayer model, oxidants induced barrier hyperpermeability. We (2) previously traced the in vitro cause of this monolayer hyperpermeability back to disruption of the functioning and architectural integrity of actin filaments. We now show that iNOS upregulation and oxidation and nitration of the subunit components of the actin cytoskeleton are a major part of the mechanism for the disruption of the F-actin filaments. In this view, the connection between iNOS and actin nitration is through the reaction

(1)

an idea that is supported by both the existing literature (21, 31, 34) and our finding here that DCF fluorescence is quenched by both O· scavengers (SOD) and by NO and ONOO scavengers.

Two factors further enhance the validity of this primary conclusion. First, the conclusion is supported in the present study by three independent lines of investigation: 1) H₂O₂ concentrations that cause actin damage and intestinal hyperpermeability simultaneously activate iNOS and increase RNMs and oxidative stress; 2) three different exogenously added RNMs mimicked all the effects of H₂O₂; and 3) both iNOS inhibitors and antioxidants that scavenge RNMs prevented or substantially attenuated the injurious changes that resulted from exposure either to H₂O₂ or to RNM compounds. Second, we found robust positive correlations between increases in RNMs, increases in oxidative stress, RNM-associated nitration and disruption of the F-actin cytoskeleton, and increases in monolayer permeability. Correlation analysis showed a significant (P < 0.05) and robust correlation between NO and nitrotyrosine levels (r = 0.97) and between nitrotyrosine levels and either actin disruption (r = 0.98) or actin disassembly (r = 0.94). Similar to nitration, oxidation, as measured by either carbonyl levels or DCF fluorescence, predicted actin disruption (r values of 0.92 and 0.96, respectively). There was also a significant positive correlation (r = 0.95) between H₂O₂-induced barrier disruption (FSA) and actin disassembly. Finally, the two markers for actin integrity (i.e., percentage of polymerization and percentage of normal actin) strongly correlated with each other (r = 0.96). Although it is possible that the increase in ONOO levels observed after exposure of the cells to H₂O₂ derived from sources other than iNOS and its reaction product NO, our data suggest that this would likely be a relatively minor source because the iNOS inhibitor L-NIL nearly completely abolished oxidant-induced nitration injury to the actin cytoskeleton and maintained barrier integrity. Overall, the biochemical cause of injury to the actin network appears to be the nitration and oxidation of its 43-kDa protein subunits.

It might be argued that the antioxidants L-cysteine and urate are not specific to RNMs and might inhibit the synthesis of ONOO (see eq. 1) by scavenging O· as SOD does. Although this could conceivably occur, there would still be overwhelming evidence that NO and ONOO are involved in oxidant-induced damage in our model. First, ONOO mimetics (authentic ONOO or ONOO-generating systems) mimic H₂O₂. Second, nitration as measured by nitrotyrosine formation on actin protein occurs in Caco-2 cells after the addition of oxidants (H₂O₂, ONOO). Third, a selective iNOS inhibitor (L-NIL) prevents nitration. Fourth, our own previous experiments (7) involving the addition of L-cysteine to a test tube containing authentic ONOO (in which L-cysteine completely scavenged the ONOO) and other literature (15, 22, 32-34) clearly indicate that L-cysteine and urate are capable of scavenging RNMs. Moreover, they must be doing so here because urate and L-cysteine prevented both the F-actin nitration and the F-actin depolymerization that was caused when we directly added authentic ONOO to our monolayers. Fifth, the free sulfhydryl groups of L-cysteine react with ONOO with a rate constant (5,900 M¹s¹) that is over 1,000-fold greater than its rate constants for reaction with other oxygen species such as H₂O₂ or O· (~4 M¹s¹)(32), making it more likely that L-cysteine is effectively scavenging ONOO at a greater rate than O·.

Our primary conclusion is consistent with recent studies showing that O· reacts rapidly with NO to generate ONOO in vivo (21, 31). These in vivo studies have led to the proposal that the formation of ONOO radical is key in the pathogenesis of IBD and a variety of other inflammatory GI and systemic disorders (23, 28, 37), but the target proteins were unclear. For example, tissue nitration, which was detected by immunofluorescent staining of nitrotyrosine, has been associated with the inflamed human mucosa in IBD (28, 37) and was linked with the upregulation of iNOS (37). Some nitrated tissue proteins (e.g., SOD and glutathione) have been detected in vivo in non-GI models such as the inflamed lung (23, 32). Moreover, it appears that ONOO-induced tissue nitration (nitrotyrosination) involves the addition of nitro groups to the ortho position of tyrosine residues (22). We also previously showed (7) that ethanol induces tubulin nitration and oxidation in vitro in monolayers of human intestinal cells. The current study suggests that actin molecules and the F-actin cytoskeleton are key target proteins of oxidant-induced nitration. On the basis of our findings with ethanol (7, 8), it seems likely that tubulin and the microtubule cytoskeleton are also key target proteins. These conjectures are further supported by previous studies (2, 6) in which we showed that phalloidin and taxol, agents that prevent oxidative damage to the F-actin and the microtubule cytoskeletons, respectively, also prevent barrier disruption.

Interestingly, cytoskeletal injury does not need to affect all cells in the monolayer to elicit intestinal leakiness. The data presented in Fig. 6 indicate that significant damage occurs when only ~50% of cells in the monolayer no longer show a normal actin cytoskeleton. This is accomplished under oxidative conditions in which there is a 65% increase in actin nitration (Fig. 9A).

Finally, our primary conclusion is consistent with our most recent in vivo studies (48) in which immunoblotting analysis of the mucosal pinch biopsy specimens of inflamed intestinal tissues from IBD patients showed increased tissue levels of NO and nitrotyrosination of actin.

Our second conclusion is also relevant to IBD and suggests a novel positive feedback mechanism that could, if it occurred in vivo, very likely overwhelm endogenous antioxidant defenses and either initiate or sustain the acute IBD attack. This positive feedback is seen in the ability of three separate oxidants (H₂O₂, HOCl, and ONOO) to upregulate iNOS, which then synthesizes NO and ONOO. Although the reasons behind the presence of such a positive feedback mechanism in our model are unclear (oxidants may be activating cellular stress responses), the existence of this mechanism is consistent with the current characterization of the pathophysiology of IBD (26, 28). This is especially true for the transition from the inactive to active (flare-up) phases of inflammation in IBD in which intestinal oxidants and proinflammatory molecules periodically create a vicious cycle that leads to sustained inflammation and tissue damage. In particular, the natural course of IBD involves recurrent episodes of the inactive phase (in which there are no neutrophils) followed by acute exacerbation (flare-up) that is characterized by mucosal infiltration of large numbers of leukocytes including neutrophils. These plasma cells are capable of producing large quantities of ROMs (e.g., H₂O₂ and HOCl) and RNMs (e.g., ONOO), reactive species that create a vicious cycle and sustain an inflammatory cascade. A positive feedback loop, such as the one we observed in Caco-2 cells, could play a key role in establishing such a vicious cycle.

Our third conclusion is that increases in the level or activity of iNOS can occur rapidly. This conclusion is supported by parallel increases in three separate variables: iNOS protein levels, iNOS enzyme activity, and NO levels. The findings of recent studies in endothelial cells, as well as one in vivo study in rat gastric mucosal cells, are also consistent with our finding of rapid iNOS upregulation (41-43). In the study that used isolated rat gastric mucosal cells, low basal levels of iNOS were noted in control (untreated) mucosa (41), whereas after challenge with endotoxin, significant increases in iNOS activity were detected in the mucosal cells within 1 h, followed by peak levels at 2-4 h. Similarly, other studies on endothelial cells showed detectable levels of iNOS activity as early as 60-90 min after H₂O₂ (0.1-1 mM) challenge (42, 43). In yet another study in endothelial cells, a slight basal expression of iNOS protein (and iNOS mRNA as detected by RT-PCR) was shown in unstimulated cells (44). Furthermore, it seems unlikely that oxidant-induced increases in cNOS activity are occurring and confounding our finding that Ca²⁺-independent iNOS is upregulated, because we found that neither oxidants nor L-NIL affect Ca²⁺-dependent cNOS activity.

A question that remains to be answered is how iNOS might be activated so rapidly. We now suggest three mechanisms by which the rapid iNOS upregulation might occur: protein synthesis starting from a preexisting mRNA pool; upregulation of inactive iNOS enzyme molecules by any of several well-known cellular mechanisms such as phosphorylation-dephosphorylation; and iNOS dimerization. The first mechanism requires a basal constitutive level of iNOS mRNA in unstimulated cells, as was suggested by a recent study in endothelial cells (44). In general, protein expression (i.e., transcription) from a "standing pool of mRNA" does not require more than 30-40 min.

A second proposed mechanism is the phosphorylation of the iNOS enzyme as a means of rapidly regulating its enzymatic activity. NOS contains consensus sequences for sites of protein phosphorylation (45, 46); tyrosine phosphorylation of Ca²⁺-independent NOS has been shown in vitro and in endothelial cells after a variety of stimuli, and it was proposed that this mechanism could rapidly regulate the activity of NOS (45, 46).

A third alternative mechanism is the rapid assembly of the two known monomeric domains of iNOS into an active dimer, which is known to be required for NOS catalytic activity. Specifically, pools of inactive monomeric iNOS would be available from a standing intracellular protein pool in unstimulated cells. These monomers can be rapidly assembled by an appropriate stimulus into an active dimer (47). Future studies will be needed to investigate which of these mechanisms is operative in our model and in intestinal cells in general.

Although ONOO is relatively short-lived at physiological pH (7.4), we believe that our methods, results, and conclusions drawn from the ONOO systems that we have employed are valid for several reasons. 1) Our findings are consistent with several reports from the literature that indicate that ONOO or its footprints have been detected in vivo in inflamed mucosal tissue from patients with IBD as well as from patients with other systemic inflammatory disorders such as in the lung (23, 26, 28, 35, 37). For example, ONOO has been detected in the inflamed colonic mucosa by immunofluorescence analysis (28). Thus, although ONOO is short-lived, its footprints have been detectable in vivo under a variety of pathophysiological conditions and, apparently, enough ONOO is around for a long enough time for this to occur. 2) ONOO generator systems (SIN-1 and SNAP-xanthine-xanthine oxidase were used in our studies) caused nitration damage to actin and to the permeability barrier of our monolayers at pH 7.4. Both of these models are known to spontaneously generate ONOO in vitro (25, 34). Moreover, the extent of the damage was equivalent to damage induced by oxidants (H₂O₂ or authentic ONOO at pH 7.6). 3) Added authentic ONOO caused a degree of damage at pH 7.6 similar to that caused by ONOO generators at pH 7.4. At pH 7.6, ONOO decreased over 30 min from 100% of added ONOO to ~38% as our laboratory previously reported (7). Thus a substantial fraction of the added ONOO remains in contact with our monolayers and could very well lead to injury.

In summary, our studies to date indicate that the disruption of intestinal barrier function induced by oxidants is caused by iNOS upregulation, RNMs, and oxidative injury to the actin cytoskeleton and the microtubule cytoskeleton. If this mechanism can be demonstrated in vivo, then our in vitro findings would suggest intracellular mechanisms (e.g., iNOS inhibitors and ONOO scavengers) that might serve as targets for the development of novel therapies for IBD.