Reassembly of the tight junction after oxidative stress depends on tyrosine kinase activity.

Oxidative stress compromises the tight junction, but the mechanisms underlying its recovery remain unclear. We developed a model in which oxidative stress reversibly disrupts the tight junction. Exposure of Madin-Darby canine kidney cells to hydrogen peroxide markedly reduced transepithelial resistance and disrupted the staining patterns of the tight junction proteins ZO-1 and occludin. These changes were reversed by catalase. The short-term reassembly of tight junctions was not dependent on new protein synthesis, suggesting that recovery occurs through re-utilization of existing proteins. Although ATP levels were reduced, the reduction was insufficient to explain the observed changes, since a comparable reduction of ATP levels (with 2-deoxy-D-glucose) did not induce these changes. The intracellular hydrogen peroxide scavenger pyruvate protected Madin-Darby canine kidney cells from loss of transepithelial resistance as did the heavy metal scavenger N,N,N',N'-tetrakis(2-pyridylmethyl)ethylenediamine. Of a wide variety of agents examined, only tyrosine kinase inhibitors and protein kinase C inhibitors markedly inhibited tight junction reassembly. During reassembly, tyrosine phosphorylation in or near the lateral membrane, was detected by immunofluorescence. The tyrosine kinase inhibitors genistein and PP-2 inhibited the recovery of transepithelial resistance and perturbed the relocalization of ZO-1 and occludin to the tight junction, indicating that tyrosine kinases, possibly members of the Src family, are critical for reassembly after oxidative stress.

Many disease states of the kidney such as ischemia/reperfusion, inflammation, or toxic injury to the kidney and gut lead to loss of the epithelial barrier. Reactive oxygen species are directly involved in the pathophysiology of some of these diseases. Targets for hydrogen peroxide (H 2 O 2 ) 1 include DNA, proteins, and lipids. Reported effects of hydrogen peroxide on renal epithelial cell lines include DNA damage with induction of apoptosis or necrosis (1)(2)(3), decreased activity of membrane transporters (4), and membrane lipid peroxidation (5). Hydrogen peroxide decreases transepithelial resistance (TER) and increases transcellular permeability of MDCK cell monolayers (6,7), which is also well documented in non-kidney epithelial cell lines (8 -11) and endothelial cell lines (12)(13)(14)(15). However, the biochemical and subcellular effects of hydrogen peroxide on tight junction (TJ) proteins have not been studied, nor have they been distinguished from effects due to partial ATP depletion. Short-term depletion and repletion of intracellular ATP in cultured cells has been used to study the disassembly and/or reassembly of junctions as a model of organ ischemia and reperfusion (16 -20). However, little is known about the reassembly of the tight junction after exposure to reactive oxygen species due to the lack of a reproducible model system.
We have now developed and analyzed a model of reversible hydrogen peroxide-induced disassembly and reassembly of the TJ in vitro and show that the reassembly pathway, as monitored by physiological, biochemical, and immunocytochemical parameters, depends upon tyrosine kinase activity. Furthermore, the cellular and biochemical consequences of H 2 O 2 exposure are distinct from those caused by ATP depletion-repletion.

EXPERIMENTAL PROCEDURES
Cell Culture and Materials-Madin-Darby canine kidney cells (MDCK, type II) were obtained from American Type Tissue Culture Collection (Rockville, MD) and maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% fetal calf serum, 50 IU/ml penicillin, and 50 g/ml streptomycin. Cells were incubated at 37°C in 5% CO 2 and were passaged weekly. DMEM was from Cellgro (Herndon, VA), streptomycin, penicillin, and fetal calf serum from Sigma. Plasticware was from Falcon (Lincoln Park, NJ). Transwells were obtained from Costar (Cambridge, MA), the ohm meter (Millicell ERS) from Millipore (Bedford, MA). Anti-ZO-1 rat monoclonal antibody (R40.76) was kindly supplied by Dr. D. Goodenough (Harvard Medical School, Boston). Anti-E-cadherin antibodies were isolated from hybridoma cells (rr-1) kindly provided by Barry Gumbiner (University of California, San Francisco). Anti-ZO-2 and anti-occludin polyclonal rabbit antibodies were from Zymed Laboratories Inc. (San Francisco, CA), antiphosphotyrosine antibody was from Upstate (4G10), Upstate Biotechnology, Inc., Lake Placid, NY. All other reagents used in these experiments were of analytical grade.
Hydrogen Peroxide Experiments-Confluent monolayers were cultured in DMEM without fetal calf serum 24 h prior to the exposure with specific concentrations of hydrogen peroxide and throughout the experiment. To terminate the exposure to hydrogen peroxide, catalase was added into the media at a final concentration of 5000 units/ml. Control experiments were performed with monolayers in serum free DMEM with and without catalase or Me 2 SO.
Trypan Blue Exclusion Assay-After treatment of MDCK monolayers with 5 mM H 2 O 2 or 2-deoxy-D-glucose (dGlc) for various times, cells were washed with PBS twice, and exposed to 0.4% trypan blue in PBS supplemented with 1.5 mM CaCl 2 and 2 mM MgCl 2 for 5 min. The cells were examined using a Nikon Diaphot inverted microscope, and the number of viable and nonviable cells was determined.
Apoptosis Assay-MDCK cells were grown to confluency on glass coverslips and exposed to H 2 O 2 with or without treatment with catalase, genistein, or PP-2. After fixation of the monolayers with 4% paraformaldehyde in PBS for 30 min, apoptosis was detected with the ApopTag TM Peroxidase kit (Intergen, Purchase, NY), labeling free 3Ј-OH termini of DNA strand breaks, according to the manufacturers instructions. Apoptotic cells were counted under an inverted microscope (Nikon).
ATP Depletion and Repletion-Depletion of intracellular ATP was achieved by using the glycolytic inhibitor 2-deoxy-D-glucose. In brief, confluent MDCK monolayers were serum starved in serum-free DMEM for 24 h, washed with PBS three times, and exposed to PBS containing 1.5 mM CaCl 2 , 2 mM MgCl 2 , and 12 mM 2-deoxy-D-glucose for different time periods. Repletion of intracellular ATP levels was achieved by changing the media to serum-free DMEM. Control experiments were performed in serum-free DMEM alone.
TER Measurements-MDCK cells were plated on polycarbonate filters (Transwell, Costar) at confluent density (roughly 2 ϫ 10 5 cells/cm 2 ) and maintained in serum containing media for 48 h to establish tight monolayers. 24 h before TER measurement cells were exposed to serum-free DMEM. In some experiments, known scavengers of hydrogen peroxide or modulators of intracellular signaling pathways were preincubated in serum-free DMEM before exposure to H 2 O 2 on both sides of the transwell. TER was measured at different time points after treatment with these reagents and/or hydrogen peroxide with a Millipore electrical resistance system, as described previously (21). Measurements were expressed in ohm ϫ cm 2 , as a mean of the original readings after subtraction of background values. Basal TER measurement of confluent monolayers were recorded between 150 and 300 ohm ϫ cm 2 .
ATP Measurements-Measurements of total cellular ATP content were performed with a luciferase-based ATP determination kit (Sigma). After rinsing the monolayer with PBS three times, 300 l of somatic cell ATP-releasing agent solubilized the cells. Cells were immediately scraped, aspirated, and cleared of insoluble material by spinning for 5 min at 10,000 rpm. The ATP content of 100 l of supernatant was measured with 100 l of Sigmas diluted luciferase-containing ATP assay mixture in an automated chemiluminescence counter (Berthold, Bad Wildholz, Germany). ATP levels are expressed as percent of the initial value after subtraction of background readings.
Immunocytochemistry-Confluent monolayers of MDCK cells were grown on clear coverslips. At specific time points during the experiment, the monolayers were rinsed twice with PBS, fixed with 100% methanol (Ϫ80°C) for 10 min for ZO-1, occludin, and E-cadherin staining, or 4% paraformaldehyde for 20 min at room temperature (for anti-phosphotyrosine staining), and stored in PBS at 4°C. After blocking (PBS, 5% (v/v) goat serum, 3% (w/v) bovine serum albumin) for 1 h, cells were incubated for 2 h with primary antibodies. After washing (4 ϫ 5 min in PBS ϩ 0,05 Triton X-100) the monolayers were incubated with Texas red or fluorescein isothiocyanate-conjugated secondary antibodies for 1 h. The cells were mounted in Fluoromount (Southern Biotechnology Associates Inc., Birmingham, AL). The filters were viewed through a ϫ100 oil immersion objective with a laser scanning confocal system (model MRC-1024/2p, Bio-Rad, Cambridge, MA) coupled to a Zeiss Axiovert microscope. Images were processed using Photoshop software (Adobe, San Jose, CA).
Statistics-The data are given as mean values Ϯ S.E. (n), where n refers to the number of measurements performed. The paired Student's t test was used to compare mean values within one experimental series. Data from two groups were compared by unpaired t test. A p value of Ͻ0.05 was accepted to indicate statistical significance. a 45-min incubation and 20 h for a 60-min incubation (data not shown).

H 2 O 2 Reversibly Decreases Transepithelial Resistance in MDCK Cell Monolayers in a Concentration and Time
Recovery from Hydrogen Peroxide Is Not Dependent on New Protein Synthesis-To test whether the recovery of TER after treatment with hydrogen peroxide and catalase is dependent on synthesis of new TJ or other proteins, MDCK monolayers were coincubated with the protein synthesis inhibitor cycloheximide (50 g/ml) and 5 mM H 2 O 2 before scavenging with catalase. Recovery of TER was nearly identical in cycloheximidetreated cells over the initial 6 h (Fig. 1D). At later time points, TER began to drop in the cycloheximide-treated monolayers, most likely reflecting the general effects of inhibiting protein synthesis. This result indicates that, for the early recovery of barrier function, new protein synthesis is not necessary.
Cell Viability of MDCK Cells after Exposure to H 2 O 2 Depended Upon Exposure Time-Trypan blue uptake was observed in 0.1 Ϯ 0.1% of the cells under baseline conditions. The number of trypan blue positive cells increased to 0.7 Ϯ 0.1% after 30 min exposure to 5 mM H 2 O 2 and 2.5 Ϯ 0.6% after 6 h ( Fig. 2A). After scavenging of H 2 O 2 with catalase for 5.5 h, the number of trypan blue positive cells remained at the same value for 30 min of H 2 O 2 exposure (0.7 Ϯ 0.3%, n ϭ 10). PP-2 and genistein did not increase the number of trypan blue positive cells (Fig. 2B).
The number of apoptotic cells was 0.18 Ϯ 0.1% under control conditions and did not increase significantly after 30 min of exposure to 5 mM H 2 O 2 or after exposure to PP-2 or genistein for 6 h with or without catalase (0.2% Ϯ 0.1%, Fig. 2C). After 6 h exposure to hydrogen peroxide 1.1 Ϯ 0.06% of the cells stained positive for apoptosis.
The Initial Fall in TER after Exposure to Hydrogen Peroxide for 30 min Was Reduced by Intra-and Extracellular Scavengers of Hydrogen Peroxide-After establishing the conditions under which a constant fall and recovery of TER occurred reproducibly (5 mM H 2 O 2 for 30 min), different scavengers or inhibitors of intracellular signaling pathways were used to determine whether the TER reduction was due to oxidative stress, and to investigate specific targets for hydrogen peroxide.
Pyruvate scavenges hydrogen peroxide by direct nonenzymatic reduction of H 2 O 2 to water while undergoing decarboxylation at the 1-carbon position (22). Pyruvate is also readily taken up by cells and can thus serve as an intracellular scavenger of H 2 O 2 . Co-incubation of pyruvate (5 mM) with hydrogen peroxide completely abolished the loss of TER, indicating that pyruvate is effectively scavenging intracellular H 2 O 2 (Fig. 3A, n ϭ 4). In another set of experiments MDCK cells were preincubated with 5 mM pyruvate in serum-free DMEM for 16 h. Before exposure to H 2 O 2 , the cells were washed with PBS to eliminate extracellular pyruvate. This preincubation with pyruvate significantly reduced the drop in TER to 62 Ϯ 5% of control. Recovery to baseline after addition of catalase was accelerated to ϳ2 versus 5 h under control conditions (Fig. 3A, n ϭ 4). Pyruvate alone did not influence TER. Other scavengers were preincubated for 1 h and coincubated with hydrogen peroxide for 30 min and the fall in TER was recorded (Fig. 3B). The reducing agent dithiothreitol reduced the initial fall of TER to 57 Ϯ 7% (dithiothreitol, 0.5 mM, n ϭ 7). Superoxide dismutase (30 units/ml, n ϭ 7) had a similar effect in reducing the fall to 50 Ϯ 4%. In marked contrast, the membrane permeable hydrogen peroxide and hydroxyl radical scavenger dimethylthiourea (10 mM, n ϭ 7) and the hydroxyl radical scavenger 5,5-dimethylpyrroline N-oxide (10 mM, n ϭ 7) had no effect on the initial fall in TER.
Two Heavy Metal Chelators Reduce the Fall in TER after Exposure to H 2 O 2 -Preincubation for 1 h before exposure to H 2 O 2 , with the cell permeable heavy metal chelator TPEN (N,N,NЈ,NЈ-tetrakis(2-pyridylmethyl)ethylenediamine, 1 M), which acts as an intracellular scavenger, reduced the fall in TER at 30 min to 58 Ϯ 3% (Fig. 3C). Furthermore, the subsequent decline of TER was significantly reduced and recovery from the exposure to H 2 O 2 was faster and complete after ϳ2 h (Fig. 3B, n ϭ 6). The cell permeable iron chelator deferoxamine (0.1 mM) was preincubated for 16 h and subsequently reduced the initial fall in TER to 56 Ϯ 7% (Fig. 3C, n ϭ 6). Exposure to TPEN or deferoxamine alone had no effect on TER.
Tyrosine Kinase or PKC Inhibition Prevents Recovery of TER-TJ assembly is known to be modulated by a variety of signaling mechanisms (9,16,17,(23)(24)(25)(26)(27)(28). In a broad screen of many inhibitors of signaling events known to modulate the TJ, the most consistent results were obtained with the two tyrosine kinase inhibitors PP-2 (4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo [3,4-d]pyrimidine, 1 or 10 M) and genistein (4Ј,5,7trihydroxyisoflavone, 50 or 100 M), both of which inhibit Src family tyrosine kinases, although genistein is also a broad spectrum tyrosine kinase inhibitor. MDCK cells were preincubated with these agents for 1 h before being exposed to 5 mM H 2 O 2 . The drop of TER was not different from a control experiment, but the recovery after 5000 units/ml catalase was virtually abolished (Fig. 4, A and B, n ϭ 6), indicating a role for tyrosine kinase in the reassembly of the TJ but not the initial disassembly. Control experiments with PP-2, genistein, or Me 2 SO alone showed no changes in baseline TER (n ϭ 4, data not shown). Thus, tyrosine kinase activity, most likely due to a Src family member, appears to be important for the recovery of the TJ after exposure to hydrogen peroxide. PKC has also been shown to regulate TJ assembly in other models (24,29,30). Three PKC inhibitors were tested for their ability to inhibit the recovery of TER after exposure to hydrogen peroxide. Calphostin C (500 nM, n ϭ 7), bisindolylmaleimide I (500 nM, n ϭ 7), and chelerythrine (1 M, n ϭ 7) reduced TER at 6 h by 73 Ϯ 5, 55 Ϯ 2, and 42 Ϯ 5%, respectively, when compared with control (Fig.  4C) (31,32). In MDCK cells, ATP levels have been shown to decrease with 1 mM H 2 O 2 (7). In our experiments, exposure of MDCK monolayers to 5 mM H 2 O 2 or 12 mM 2-deoxy-D-glucose for 30 min resulted in a partial ATP depletion to 21.4 Ϯ 4 and 20.7 Ϯ 2% of control, respectively (Fig. 5,  A and B, n ϭ 4 -8). Furthermore, the time course for the recovery of cellular ATP content after treatment with either 5000 units/ml catalase (H 2 O 2 group) or repletion with control media (dGlc group) after 30 min was similar. Media changes or catalase treatment had no effect on the intracellular ATP concentration (98.7 Ϯ 3 and 96.3 Ϯ 4% after after 6 h, respectively). Therefore, 12 mM dGlc was used in the following experiments to examine the effect of partial ATP depletion on TER, extractability and localization of TJ proteins.
In marked contrast to the effects of 5 mM H 2 O 2 on TER (see Fig. 1), treatment with 12 mM dGlc was not associated with a detectable drop in transepithelial resistance. On the contrary, at later time points, a rise in TER of partially ATP depleted monolayers to 25.5 Ϯ 5% above the TER of repleted monolayers was observed (Fig. 5C, n ϭ 4). Consistent with this finding, in dGlc-treated cells, no increase in trypan blue uptake was recorded (n ϭ 10). The finding that TER is unchanged in cells ATP depleted to 20% of control levels is consistent with previous work that indicates that the threshold for ATP depletioninduced declines in TER occurs at around 5-10% of normal ATP levels (4,5,7,17,33).
After induces a reversible, partial breakdown of ZO-1 in the TJ after 30 min (Fig. 6A, b, f, and j). Prolonged exposure of MDCK cells to 5 mM H 2 O 2 resulted in a discontinuous ZO-1 staining pattern throughout the monolayer (Fig. 6A, c, g, and k, n ϭ 4 for each observation). Occludin and E-cadherin showed a similar pattern of redistribution and broadening of the staining. Treatment with 5000 units/ml catalase prevented the disruption of the TJ and after 6 h a normal distribution of all three junctional proteins was demonstrated (Fig. 6A, d, h, and l). The exposure of MDCK cells to 12 mM dGlc had no effect on the distribution of ZO-1, occludin, or E-cadherin, indicating that these changes, like the drop in TER, could not be explained by lower ATP levels (Fig. 6B, a and b, data shown for ZO-1 only, n ϭ 3 for each observation). The ZO-1 staining remained continuous but appeared slightly more jagged during later time points.
Genistein and PP-2 Prevent the Reformation of the Tight Junction after Oxidative Stress-Immunofluorescence of MDCK monolayers after double staining with anti-phosphotyrosine and anti-ZO-1 antibodies showed intact staining of the tight junction junction 6 h after exposure to H 2 O 2 and catalase or catalase alone (Fig. 7, a and b). In contrast, MDCK cell monolayers that were pretreated with 10 M PP-2 (Fig. 7c) or 100 M genistein (Fig. 7e) for 1 h and exposed to the same oxidative stress and catalase in the presence of the inhibitors showed a marked disruption in ZO-1 staining after 6 h that resembled the disruption without catalase scavenging (see Fig.  6c), consistent with the data indicating that these agents inhibit TER recovery (Fig. 4). Both inhibitors had no effect on the ZO-1 staining in the presence of catalase without oxidative stress (Fig. 7, d and f ).
Hydrogen Peroxide Induces Tyrosine Phosphorylation along the Lateral Membrane during Disassembly and Reassembly of the Tight Junction-Tyrosine phosphorylation occurred in the lateral membrane after short-term exposure to hydrogen peroxide (Fig. 8A, b) when compared with untreated monolayers (Fig. 8A, a), as described before (34,35). Furthermore, strong tyrosine phosphorylation occurred in the lateral cell border 6 h after exposure to H 2 O 2 and scavenging with catalase (Fig. 8A, f, same sample as Fig. 7a). Analysis of z-sections revealed that the staining for tyrosine phosphorylation and ZO-1 overlapped in the tight junction although additional phosphotyrosine staining was observed in the lateral membrane and the cytosol of the cells (Fig. 8B). Both PP-2 and genistein prevented this tyrosine phosphorylation (Fig. 8A, d and e), giving rise to a very weak and diffuse staining pattern which resembled the tyrosine phosphorylation of monolayers treated with H 2 O 2 alone (Fig. 8A, c). Catalase treatment alone did not induce tyrosine phosphorylation in the cell junction (Fig. 8A, g). These experiments imply that tyrosine phosphorylation, possibly in the lateral membrane near or in the TJ, occurs during TJ recovery, and that the ability of catalase to induce the recovery of the tight junction possibly requires these phosphorylation events. DISCUSSION Oxidative stress is known to disturb the permeability barrier in renal epithelial cells in culture. Transepithelial resistance and intracellular ATP levels fall after exposure to hydrogen peroxide, whereas the permeability for small solutes across the monolayer increases (6,7). Initially, a reduced intracellular ATP level was thought to cause the disruption of the junction (36), but later studies identified other potential regulators such as intracellular calcium activity (37), pH (38), PKC (39), and calcium-independent phospholipase A 2 (40) during oxidative stress. Until now, the mechanisms of tight junction reassembly after oxidative stress that lead to a reconstituted and functional epithelial cell barrier have not been investigated. Here, we present a new model in which the disruptive effect of hydrogen peroxide on physiological, biochemical, and immunocytochemical parameters of TJ integrity is reversible. H 2 O 2 (5 mM) consistently dropped transepithelial resistance to 23% of control. Thereafter, the epithelial cell monolayers completely recovered to baseline within 6 h. After recovery, the MDCK monolayers showed a stable TER for several days. The exposure time of 30 min used in our experiments was associated with only a small decrease in cell viability. This finding is in good agreement with previous studies in epithelial cells that show a high resistance against oxidative stress in MDCK cells (5,32). The small decrease in viability did not correlate with the observed changes in transepithelial resistance. After full recovery of TER about the same number of non-viable or apoptotic cells were observed as after 30 min of hydrogen peroxide exposure. Importantly, the initial recovery did not depend on new protein synthesis, which implies that the recovery does not depend on development of de novo tight junctions. Increasing the intracellular concentration of the H 2 O 2 scavenger pyruvate reduced the drop in TER by 46%, indicating an intracellular effect of H 2 O 2 . Several extracellular hydrogen peroxide scavengers were also effective in reducing the initial fall in TER.
Interestingly, superoxide dismutase had a significant effect on the initial fall in TER. Superoxide dismutase catalyzes the dismutation of superoxide anion into hydrogen peroxide. Superoxide anion is generated proportionally to the oxygen ten- Note, that TER of dGlc exposed monolayers rose above TER of short-term exposed monolayers. Data are mean Ϯ S.E. sion in mitochondria or after activation of NADPH oxidase. Both pathways could be activated under oxidative stress. Mitochondrial function is severely impaired by hydrogen peroxide which results in additional generation of reactive oxidative species. Also, activation of NADPH oxidase by arachidonic acid has been described in kidney epithelial cells (41) and arachidonic acid is known to be increased under oxidative stress (40,42). The induction of superoxide anion production by hydrogen peroxide could therefore increase the intracellular hydrogen peroxide even further.
H 2 O 2 decreases the intracellular ATP level in proportion to concentration and duration of exposure (7,32). A number of groups have shown a decrease of intracellular ATP levels to 16 -34% of control under similar oxidative stress in renal epithelial cell lines (2,5,7). Our observations confirm this reported decrease in ATP, showing an ATP level of 21% of control after 30 min incubation with 5 mM H 2 O 2 . Interestingly, this ATP level is higher than ATP levels obtained with the established ATP depletion model of TJ disassembly in which both glycolysis and mitochondrial oxidative phosphorylation are inhibited. The two most widely used reagents for ATP depletion, 2-deoxy-D-glucose and antimycin A, consistently lower intracellular ATP levels in MDCK cells to 2-5% of the initial value within minutes (16,17,43). This "maximal" ATP depletion induces the disassembly of the tight junction, and increases the association of TJ proteins with the cytoskeleton (17). The reversibility of the effects of ATP depletion on the TJ of MDCK cells has been established in recent years (16,17,33,44). During the ATP repletion phase, the reassembly of the tight junction depends on tyrosine phosphorylation of TJ proteins (16). To confirm that, in our model, the decrease of the intracellular ATP level is not responsible for the disassembly of the TJ, ATP levels were reduced to the level observed after exposure to hydrogen peroxide. Treatment with 2-deoxy-D-glucose for 30 min decreased ATP to the same level as hydrogen peroxide. This partial ATP depletion to 20% of the initial value failed to cause a decrease in transepithelial resistance. Although these findings indicate that other factors are more likely to contribute to the disassembly of the tight junction, they do not rule out an influence of the decreased ATP level on the regulation of the tight junction under oxidative stress. Reassembly of the tight junction after partial ATP depletion could not be monitored since treatment with dGlc did not disturb the tight junction. Hence, it is conceivable that the reassembly of the TJ after oxidative stress and total ATP depletion could be regulated in a similar fashion.
It is well recognized that hydrogen peroxide can act as a protein-tyrosine phosphatase inhibitor (10,45,46). In epithelial cells, exposure to hydrogen peroxide or phosphatase inhibitors has been shown to decrease TER and increase permeability, which could be prevented by tyrosine kinase inhibition (34,35). It was suggested that the general increase in tyrosine phosphorylation could induce the disassembly of the tight junction, although specific phosphorylation events in the tight junction were not described. Interestingly, both genistein and PP-2 did not affect the disassembly of the tight junction in our study but strikingly prevented the recovery. This suggests an important role for tyrosine phosphorylation also in the regulation of TJ reassembly after oxidative stress. Since disassembly and reassembly of the tight junction are regulated by many, yet most likely distinct, signaling pathways, our findings led us to investigate the behavior of tight junction proteins during the reassembly of the junction. Furthermore, the specific inhibition of PP-2 implies a role for Src family tyrosine kinases in this process.
Tyrosine phosphorylation that co-localized with ZO-1 as well as the lateral membrane was observed during disassembly and during the recovery period but not after 6 h of exposure to H 2 O 2 . This tyrosine phosphorylation and recovery of continuous staining of ZO-1 after recovery of the TJ were prevented by PP-2 and genistein. This suggests that tyrosine phosphorylation of tight junction proteins or their regulatory proteins is critical during the recovery of the tight junction. However, a specific target for tyrosine phosphorylation within the tight junction was not identified.
Tyrosine phosphorylation of occludin, ZO-2, and ZO-3 is known to occur during the reassembly of the TJ after total ATP depletion followed by repletion. This tyrosine phosphorylation was inhibited by genistein (16). In a different model, transfection of Ras into MDCK cells induced a decrease of TER together with a redistribution of occludin, ZO-1, and claudin-1 into the cytosol (47). After treatment of these cells with the mitogenactivated protein kinase kinase (MEK1) inhibitor PD98059, occludin and ZO-1 but not claudin-1 were tyrosine phosphorylated. Subsequently, all three proteins were recruited to the cell membrane. ZO-1 was also tyrosine phosphorylated during redistribution into the TJ together with actin after treatment of human epidermal carcinoma cell lines with epidermal growth factor (48). In another in vivo model strong tyrosine phosphorylation of ZO-1 was noted after perfusion of newborn rat kidneys with protamine sulfate. The tyrosine phosphorylation correlated with the rapid replacement of the slit diaphragms between glomerular epithelial foot processes with typical tight junctions (49). Although these model systems are very different in nature, these studies, as does ours, suggest that tyrosine phosphorylation of the tight junction proteins is involved in the targeting of these proteins into the TJ.
Two groups have reported evidence for tyrosine phosphorylation of junctional proteins during the disassembly of the junction after treatment with the tyrosine phosphatase inhibitors vanadate and phenylarsine (35,50). Our study indicates a role for tyrosine phosphorylation during the reassembly process. It thus seems likely that phosphorylation events are involved in the regulation of both disassembly and reassembly.
Interestingly, our data also suggests another regulatory mechanism for TJ reassembly. Although we chose to focus on the role of tyrosine kinases in TJ assembly, in a broad screen of inhibitors, the only others that were found to significantly and consistently inhibit TER recovery after oxidative stress were those active against PKC. This suggests that serine/threonine phosphorylation of TJ proteins may be important for their incorporation into the TJ during recovery from oxidative stress.
During disassembly of the junctional complex, occludin was found to be serine/threonine phosphorylated in endothelial cells after exposure to oxidative stress (15). Other studies in FIG. 8. Immunostaining for phosphotyrosine. A, short-term exposure to H 2 O 2 caused increased tyrosine phosphorylation in the region of the lateral membrane (b) when compared with control (a). Strong tyrosine phosphorylation in the same region also occurred during the recovery period (f) which was diminished in the PP-2 and genistein-treated monolayers (d and e, same samples as shown in Fig. 7). Monolayers after exposure to H 2 O 2 for 6 h (c) or catalase alone (g) showed a similar background staining. Bar, 30 m. B, the tyrosine phosphorylation was further analyzed in the z axis by confocal microscopy. Top panel, ZO-1 staining of the TJ after reassembly; middle panel, tyrosine phosphorylation of the same z axis; bottom panel, co-localization of both staining patterns after independent sampling of both emission wavelengths. Although co-localization occurred, tyrosine phosphorylation can be observed throughout the lateral membrane and in the cytosol. The slide surface is indicated with horizontal bars, the lateral cell borders with vertical arrows. Bar, 10 m. epithelial cells have shown that activation of PKC and serine/ threonine phosphorylation of tight junction proteins occurs during (re)-assembly of the tight junction. Activation of PKC stimulates TER development while the inhibition of PKC induces disassembly of the TJ and inhibits reassembly (24,29,30). Tight junction proteins may be a direct target of PKC. PKC is a member of the atypical PKC subgroup and has been localized to the TJ (29). PKC activation was also detected during TJ reassembly after the calcium switch. Under these conditions, the TJ protein occludin was serine/threonine phosphorylated but not tyrosine phosphorylated (51).
In conclusion, we describe the first reversible model for the study of oxidative stress on the TJ in epithelial cells. Our results identify the important role of tyrosine kinases during the reassembly of the tight junction and establishes differences from other models of tight junction assembly. Future experiments with the presented model system will allow the study of the phosphorylation status of specific TJ proteins after shortterm oxidative stress. Also, the specific role of PKC during the reassembly of the tight junction remains to be investigated in detail.