INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

RESTORATION OF THE INTESTINAL mucosal barrier following a variety of injurious or inflammatory events is a critical component of innate mucosal defense (25). In previous studies, we have begun to elucidate the pathways by which the prostanoids PGE₂ and PGI₂ stimulate recovery of barrier function in porcine ischemia-injured ileal mucosa. In particular, we have noted that this recovery process appears to be related to events localized to the paracellular space (4-6) rather than reparative events such as epithelial restitution and villous contraction. However, studies by other groups have shown that PGE₂ is permissive for growth factor-stimulated restitution (29) and PGE₂ stimulates contraction of uninjured villi and crypts (9), suggesting that the reparative actions of PGs are multiple and complex. In porcine ileal mucosa subjected to 45 min of ischemia, villous contraction and epithelial restitution are nearly complete within 60 min of injury, and yet PGs are able to stimulate continued elevations in transepithelial resistance (TER) after 60 min. These elevations in TER are correlated with decreased transmucosal flux of the paracellular probes mannitol and inulin and electron microscopic evidence of closure of paracellular spaces in restituted epithelium (4, 6). Furthermore, PG-induced elevations in TER are inhibited by cytochalasin D (5), an agent that initiates cytoskeletal contraction and opening of tight junctions at the appropriate dosages (14).

The mechanisms by which PGs stimulate closure of paracellular spaces are not fully characterized, although we know that sharp elevations in Cl secretion precede recovery and that inhibition of Cl secretion with the loop diuretic bumetanide attenuates mucosal recovery (4). The role of Cl secretion in recovery of paracellular resistance is unclear, although it is conceivable that this event results in a transmucosal osmotic gradient. Indeed, mucosal osmotic loads have been shown to stimulate elevations in TER in normal guinea pig ileum (12) and recovery of TER in ischemia-injured porcine ileal mucosa (4). We have speculated that initial repair of tight junctions would have to precede their subsequent closure and recovery of TER (6).

PG signaling mechanisms that might result in tight junction repair include their second messengers cAMP and Ca²⁺ (5), both of which have been shown to alter tight junction structure in Necturus gallbladder (7, 24). Additional signaling intermediates that we have investigated are tyrosine kinases and protein kinase C (6). Although genistein augmented PG-induced mucosal recovery, this did not appear to relate to its ability to inhibit tyrosine kinases, and inhibition of protein kinase C had no effect on PG-stimulated mucosal recovery (6). However, recent evidence suggests that phosphatidylinositol 3-kinase (PI3K) is intimately involved in regulation of tight junction assembly (27) and preferentially binds to specific regions of the transmembrane protein occludin via its p85 regulatory subunit (20). Therefore, in the present study, we sought to provide further evidence for a selective action of PGs on recovery of paracellular resistance and to determine if PI3K plays a role in this reparative process. Our data show that inhibition of PI3K completely inhibits the action of PGs, which is correlated with inhibition of the ability of PGs to restore localization of the tight junction integral membrane protein occludin and the cytoplasmic plaque protein zonula occludens-1 (ZO-1) to interepithelial junctions.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Experimental animal surgeries. All studies were approved by the North Carolina State University Institutional Animal Care and Use Committee. Six- to eight-week-old Yorkshire crossbred pigs of either sex were housed singularly and maintained on a commercial pelleted feed. Pigs were held off feed for 24 h before experimental surgery. General anesthesia was induced with xylazine (1.5 mg/kg im), ketamine (11 mg/kg im), and pentobarbital (15 mg/kg iv) and was maintained with intermittent infusion of pentobarbital (6-8 mg · kg¹ · h¹). Pigs were placed on heating pads and ventilated with 100% O₂ via a tracheotomy by using a time-cycled ventilator. The jugular vein and carotid artery were cannulated, and blood gas analysis was performed to confirm normal pH and partial pressures of CO₂ and O₂. Lactated Ringer solution was administered intravenously at a maintenance rate of 15 ml · kg¹ · h¹. Blood pressure was continuously monitored via a transducer connected to the carotid artery. The ileum was approached via a ventral midline incision. Ileal segments were delineated by ligating the intestinal lumen at 10-cm intervals. Loops were randomly designated as control or ischemic loops. The latter were subjected to ischemia by clamping the local mesenteric blood supply for 45 min.

Ussing chamber studies. After the ischemic period, the mucosa was stripped from the seromuscular layer in oxygenated (95% O₂-5% CO₂) Ringer solution and mounted in 3.14-cm²-aperture Ussing chambers, as described in a previous study (2). Tissues were bathed on the serosal and mucosal sides with 10 ml Ringer solution. The serosal bathing solution contained 10 mM glucose and was osmotically balanced on the mucosal side with 10 mM mannitol. Bathing solutions were oxygenated (95% O₂-5% CO₂), circulated in water-jacketed reservoirs, and maintained at 37°C. The spontaneous potential difference (PD) was measured by using Ringer-agar bridges connected to calomel electrodes, and the PD was short-circuited through Ag-AgCl electrodes by using a voltage clamp that corrected for fluid resistance. Resistance ( · cm²) was calculated from the spontaneous PD and short-circuit current (I_sc). If the spontaneous PD was between 1.0 and 1.0 mV, tissues were current clamped at ±100 µA for 5 s and the PD was recorded. I_sc and PD were recorded every 15 min for 180 min.

Experimental treatments. Tissues were bathed in Ringer containing 5 µM indomethacin to prevent PG production while stripping mucosa from the seromuscular tissues, and indomethacin was added to the serosal and mucosal bathing solutions in the same concentration before mounting tissues on Ussing chambers. Other treatments that were added to the serosal and mucosal bathing solutions before baseline electrical measurements were the PI3K inhibitors wortmannin (10 nM) and LY-294002 (10 µM). Baseline electrical readings were taken for 30 min, after which 1 µM 16,16-dimethyl-PGE₂ (Sigma Chemical, St. Louis, MO) and 1 µM carbacyclin (the stable analog of PGI₂) were added to the serosal bathing solution. In studies assessing the role of osmotic gradients, 100-300 mosmol/kgH₂O urea was added to either the mucosal or serosal side of tissues. In studies in which hydrostatic pressures were applied, the volume of fluid was incrementally increased on the serosal side of tissues by 2-6 ml.

Isotopic mannitol and Na+ flux studies. All fluxes were conducted under short-circuit conditions (tissues clamped to 0 mV). Dual transmucosal mannitol and Na⁺ fluxes were performed on tissues paired according to their initial conductance readings (within 25% of each other). [³H]mannitol (0.2 µCi/ml diluted in 10 mM mannitol) or [¹⁴C]inulin was placed on the mucosal side of tissues and 0.3 µCi/ml ²²Na was placed on the serosal side of tissues following an initial 30-min equilibration period. One 60-min flux was subsequently conducted from 60 to 120 min of the experimental recovery period by taking samples from the side opposite to that of isotope addition and counted for ³H or ²²Na in a scintillation counter. Mucosal-to-serosal fluxes (J_ms) of mannitol or inulin and serosal-to-mucosal fluxes (J_sm) of Na⁺ were calculated by using standard equations (1, 2).

Electron and light microscopy. Tissues were taken at 0, 30, 60, 120, and 180 min for routine histological evaluation. Tissues were sectioned (5 µm) and stained with hematoxylin and eosin. For each tissue, three sections were evaluated. Four well-oriented villi were identified in each section. The height of the villus and the width at the midpoint of the villus were obtained by using a light microscope with an ocular micrometer. For height measurements, the base of the villus was defined as the intersection between adjacent villi at the opening of the crypt. For villi in which the height of one side of the villus was disparate from the other side, an average height was recorded. In addition, the height of the epithelial-covered portion of each villus was measured. The surface area of the villus was calculated by using the formula for the surface area of a cylinder. The formula was modified by subtracting the area of the base of the villus and multiplying by a factor accounting for the variable position at which each villus was cross-sectioned. In addition, the formula was modified by a factor that accounted for the hemispherical shape of the upper portion of the villus (1). The percentage of the villous surface area that remained denuded was calculated from the total surface area of the villus and the surface area of the villus covered by epithelium. The percentage of denuded villous surface area was used as an index of epithelial restitution.

Epithelial isolation. Tissues were rinsed with 30 ml of cold CO₂-saturated PBS and subsequently dropped into a tube containing CO₂-saturated citrate phosphate buffer (in mM: 96 NaCl, 1.5 KCl, 27.0 Na citrate, 5.6 KH₂PO₄, and 8.0 Na₂HPO₄). The tube was capped immediately and incubated at 37°C in a water bath for 20 min. The tissue was then transferred to a tube containing CO₂-EDTA buffer (in mM: 137 NaCl, 2.7 KCl, 1.5 KH₂PO₄, 8.0 Na₂HPO₄, 1.5 tetrasodium EDTA, and 2.5 glucose) and incubated at 37°C in a water bath for 30 min. Tissues were vortexed, after which a histological sample was submitted to check for the degree of epithelial sloughing. Tissues were subsequently centrifuged at 2,000 rpm for 10 min, and the pellet was solubilized in EDTA buffer in preparation for Western blotting.

Gel electrophoresis and Western blotting. Isolated epithelium from control and ischemia-injured mucosa treated with indomethacin (5 µM), indomethacin (5 µM) and PGs (1 µM), or indomethacin (5 µM), PGs (1 µM), and wortmannin (10 nM) and recovered for 120 min in oxygenated Ringer was snap frozen and stored at 70°C before SDS-PAGE. Tissue aliquots were thawed at 4°C and added to 3 ml chilled lysis buffer, including protease inhibitors (0.5 mM Pefabloc, 0.1 mM 4-nitrophenyl phosphate, 0.04 mM -glycerophosphate, 0.1 mM Na₃VO₄, 40 µg/ml bestatin, 2 µg/ml aprotinin, 0.54 µg/ml leupeptin, and 0.7 µg/ml pepstatin A) at 4°C. This mixture was homogenized on ice and then centrifuged at 4°C, and the supernatant was saved. Protein analysis of extract aliquots was performed (DC protein assay; Bio-Rad, Hercules, CA). Tissue extracts (amounts equalized by protein concentration) were mixed with an equal volume of 2× SDS-PAGE sample buffer and boiled for 4 min. Lysates were loaded on a 10% SDS-polyacrylamide gel, and electrophoresis was carried out according to standard protocols. Proteins were transferred to a nitrocellulose membrane (Hybond ECL; Amersham Life Science, Birmingham, UK) by using an electroblotting minitransfer apparatus. Membranes were blocked at room temperature for 60 min in Tris-buffered saline plus 0.05% Tween 20 and 5% dry powdered milk. Membranes were washed and then incubated for 60 min in primary antibody. After being washed, the membranes were incubated for 45 min with horseradish peroxidase-conjugated secondary antibody. After additional washes, the membranes were developed for visualization of protein by the addition of enhanced chemiluminescence reagent (Amersham, Piscataway, NJ). Densitometry was performed by using appropriate software (IP gel; Scanalytics, Fairfax, VA).

Immunofluorescence microscopy. Tissues were fixed in 10% neutral-buffered formalin for 24 h, transferred to 70% ethanol, routinely processed for paraffin embedding, and cut into 5-µM sections. Slides were subsequently deparaffinized and rehydrated. Epitope retrieval was done by boiling the specimens in citrate buffer (pH 6.0) for 10 min, then allowing specimens to cool for 25 min at room temperature. Sections were blocked with 2% BSA and washed with BLOTTO and PBS, after which they were incubated in primary rabbit polyclonal anti-occludin, primary rabbit polyclonal anti-ZO-1, or an isotype control for rabbit primary antibody (negative control) for 1 h on ice. Sections were then incubated with goat anti-rabbit IgG Cy3 conjugate for 30 min in the dark. Sections were mounted, and well-oriented villi were examined with an immunofluorescence microscope.

Data analysis. Data were reported as means ± SE. All data were analyzed by using an ANOVA for repeated measures, except where the peak response was analyzed by using a standard one-way ANOVA or paired t-test (Sigmastat; Jandel Scientific, San Rafael, CA). A Tukey's test was used to determine differences between treatments following ANOVA. Flux data was subjected to linear regression analysis, and the correlation coefficient (R) was assessed for significance. P < 0.05 was considered significant for all analyses.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Application of 1 µM 16,16-dimethyl-PGE₂ and 1 µM carbacyclin (a stable analog of PGI₂) to mucosal sheets of porcine ileum injured by 45 min of ischemia and bathed in 5 µM indomethacin resulted in recovery of control levels of TER within 30 min, whereas ischemia-injured tissues exposed to indomethacin alone showed minimal elevations of TER over a 180-min recovery period (Fig. 1A). As we have shown in previous reports, this PG-induced recovery was preceded by a sharp elevation in I_sc (Fig. 1B) attributable to secretion of Cl (4). As in previous studies (6), there was no difference in the histological appearance of repairing tissues treated with indomethacin compared with those additionally treated with PGs (Fig. 2), which was confirmed by showing no significant difference in the degree of epithelial restitution (Table 1). In fact, restitution was nearly complete within 60 min, suggesting that the peak effects of PGs between 90 and 120 min were related to events localized to the paracellular space.

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Electrical responses of ischemia-injured tissues treated with indomethacin (indo, 5 µM) and 16,16-dimethyl-PGE2 and carbacyclin (PGs, 1 µM). A: serosal addition of PGs to ischemia-injured tissues after an initial 30-min equilibration period resulted in rapid recovery of control levels of transepithelial resistance (TER), whereas tissues treated with indomethacin alone had little evidence of recovery. B: elevations in TER were preceded by significant elevations in short-circuit current (Isc), which is associated with Cl secretion in this tissue. Plotted values represent means ± SE; n = 8. The significance of the elevations in TER and Isc in the presence of PGs was determined by using 2-way ANOVA on repeated measures (P < 0.05).

View larger version (46K): [in this window] [in a new window]   Fig. 2.   Histological appearance of ischemia-injured porcine ileal mucosa. A: ischemia for 45 min resulted in lifting and sloughing of epithelium from the tips of villi. B: after a 60-min recovery period in an Ussing chamber in the presence of indomethacin (5 µM), villi have contracted and epithelial restitution is nearly complete. C: tissues treated with indomethacin and PGs (added after a 30-min equilibration period and recovered for an additional 30 min) have a similar histological appearance to tissues treated with indomethacin alone. Bar = 100 µm.

To further explore the possibility that PG-induced changes in TER were paracellular in nature, we measured J_ms of the paracellular probes [³H]mannitol and [¹⁴C]inulin as well as J_sm of ²²Na⁺ between 60 and 120 min of the recovery period (when PG-treated tissues reached maximum TER values). Flux of these probes was significantly greater in ischemia-injured tissues treated with indomethacin alone compared with tissues treated additionally with PGs (Fig. 3). We then assessed the correlation between the flux of mannitol or inulin and that of Na⁺ as a method of assessing the contribution of changes in paracellular permeability (accounted for by mannitol or inulin flux) to changes in TER (accounted for by J_sm of Na⁺), as previously described (15). We first confirmed that J_sm of Na⁺ closely correlated with changes in TER in tissues treated with indomethacin or indomethacin/PGs (R = 0.76, P < 0.001, data not shown). We subsequently documented a significant and linear correlation between fluxes of the paracellular probes and J_sm of Na⁺ (Fig. 4), indicating that changes in resistance were indeed reflective of changes in paracellular permeability.

View larger version (12K): [in this window] [in a new window]   Fig. 3.   Evaluation of the effects of indomethacin and PGs on serosal-to-mucosal fluxes (Jsm) of 22Na and mucosal-to-serosal fluxes (Jms) of [3H]mannitol and [14C]inulin. Fluxes were commenced 30 min after the addition of PGs and were conducted over a 1-h period. Tissues in the presence of indomethacin (5 µM) and PGs (1 µM) had significantly reduced Jsm 22Na over a 1-h time period compared with tissues treated with indomethacin alone (A). Similar results were obtained for Jms of the paracellular probes [3H]mannitol (B) and [14C]inulin (C). Plotted values represent means ± SE; n = 8. *P < 0.05 vs. indomethacin-treated tissues.

View larger version (13K): [in this window] [in a new window]   Fig. 4.   Correlation between Jsm 22Na and Jms [3H]mannitol or Jms [14C]inulin in ischemia-injured tissues treated with indomethacin (5 µM) and PGs (1 µM) or indomethacin alone. There was a significant correlation between Jsm 22Na and Jms mannitol (A) or Jms inulin (B), suggesting that changes in TER (which correlate closely with Jsm 22Na) were related to changes in paracellular permeability (as indicated by changes in Jms of the paracellular probes mannitol and inulin). The correlation coefficient (R) and its significance (P values) are indicated adjacent to linear regression plots.

Although the experiments thus far indicated an action of PGs on the paracellular space, we wanted more direct evidence of the involvement of the paracellular structures in the recovery response. Therefore, we performed a series of experiments in which we added increasing levels of serosal hydrostatic pressure by raising the fluid level of the serosal reservoir. We postulated that this would dilate paracellular spaces and apical tight junctions, thereby nullifying the effects of PGs. Accordingly, there was a pressure-dependent decrease in the PG-induced recovery of TER, with 6 cm serosal pressure nullifying the effects of PGs on injured tissues (Fig. 5). This action was not attributable to disruption of Cl secretion, because there was no significant reduction of I_sc by 6 cm serosal pressure. Tissues taken during peak TER levels in response to PGs showed ultrastructural evidence of closely apposed tight junctions compared with tissues treated with indomethacin alone. Furthermore, tissues subjected to 6 cm of serosal pressure in the presence of PGs also showed dilatation of paracellular structures (Fig. 6). These observations were confirmed morphometrically by showing pressure-dependent increases in the area of the paracellular space (Fig. 7). There was no effect of hydrostatic pressure on normal tissues (data not shown), suggesting that hydrostatic pressure selectively affected tissues in the process of recovering paracellular resistance.

View larger version (24K): [in this window] [in a new window]   Fig. 5.   Effects of serosal hydrostatic pressure on recovery of TER. A: recovery response of ischemia-injured tissues treated with indomethacin (5 µM) and PGs (1 µM) was marginally reduced by 2-4 cmH2O of serosal pressure, whereas 6 cmH2O of hydrostatic pressure fully and significantly inhibited recovery of TER. B: changes in Isc in response to serosal pressure did not appear to be correlated with changes in TER. In particular, 6 cmH2O caused a small increase in Isc but fully inhibited TER, suggesting that the effects of serosal pressure are related to mechanical effects on the paracellular space rather than alterations in Cl secretion. Plotted values represent means ± SE; n = 8. The significant reduction in TER in the presence of PGs/6 cmH2O compared with tissues in the presence of PGs alone was determined by using 2-way ANOVA on repeated measures (P < 0.05).

View larger version (179K): [in this window] [in a new window]   Fig. 6.   Electron micrographs of tissues exposed to indomethacin, PGs, and serosal pressure. A: ischemia-injured tissues after a 120-min in vitro recovery period in the presence of indomethacin (5 µM) have dilated tight junctions (arrows) and paracellular spaces. B: tissues additionally treated with PGs (1 µM) have closely apposed tight junctions (arrows). C: application of 6 cmH2O serosal hydrostatic pressure to tissues treated with both indomethacin and PGs results in dilatation of tight junctions (arrows) and paracellular spaces. Bar = 2 µm. D: increased magnification of the interepithelial junctional region of ischemia-injured tissues treated with indomethacin for 120 min. Note dilatation of the tight junction (arrow). E: same high power magnification of the interepithelial junctional region of ischemia-injured tissues treated with indomethacin and PGs revealed closely apposed tight junctions (arrow), whereas tissues exposed to 6 cmH2O serosal hydrostatic pressure had dilated tight junctions (arrow). Bar = 1 µm.

View larger version (12K): [in this window] [in a new window]   Fig. 7.   Area of the paracellular space in the region of the tight junction based on electron microscopic images. The area of the paracellular space was dramatically reduced in tissues treated with indomethacin (5 µM) and PGs (1 µM) compared with tissues treated with indomethacin alone. Application of 2-4 cmH2O serosal pressure had no significant effect on area of the paracellular space compared with tissues treated with indomethacin and PGs, but 6 cmH2O hydrostatic pressure caused a significant elevation in the area of the paracellular space. Plotted values represent means ± SE; n = 8. *P < 0.05 vs. ischemia/ indomethacin tissues. P < 0.05 vs. all other treatment groups, as determined by 1-way ANOVA and a post-hoc Tukey's test.

In further experiments, we attempted to elucidate some of the mechanisms involved in PG-induced recovery of paracellular resistance. In previous studies, we have suggested that increases in Cl secretion that precede recovery of TER may result in development of an osmotic gradient across the mucosa (4, 6). To test this hypothesis, we applied increasing doses of urea on the mucosal surface of ischemia-injured tissues treated with indomethacin and compared the effects of these treatments with that of the PGs. Accordingly, we noted dose-dependent increases in recovery of TER with mucosal application of urea that peaked with application of 200 mosmol/kgH₂O (Fig. 8). Application of other osmotic agents to the mucosal surface of tissues, including mannitol (300 mosmol/kgH₂O) and lactulose (300 mosmol/kgH₂O), resulted in similar increases in TER in ischemia-injured mucosa (peak TER in response to mannitol, 66 ± 4  · cm², n = 6; peak TER in response to lactulose, 65 ± 2  · cm², n = 3). To demonstrate the importance of the direction of the osmotic gradient, we applied 300 mosmol/kgH₂O urea to the serosal surface of ischemia-injured tissues and saw a reduction rather than an increase in recovery of TER. We next reasoned that, if PGs were setting up a mucosal-to-serosal osmotic gradient, the effect of the PGs should be reversed with serosal application of urea. In support of this premise, application of 300 mosmol/kgH₂O urea to the serosal surface of recovering tissues fully inhibited the action of PGs on injured tissues (Fig. 8).

View larger version (41K): [in this window] [in a new window]   Fig. 8.   Electrical responses of tissues subjected to osmotic loads of urea. Dose-dependent increases in TER were noted in response to 100-200 mosmol/kgH2O urea on the mucosal surface of tissues, whereas 300 mosmol/kgH2O urea appeared to have no further effect. Conversely, serosal application of 300 mosmol/kgH2O reduced TER below levels of tissues treated with indomethacin alone and fully inhibited recovery of tissues in response to PGs, suggesting that the orientation of osmotic gradients in recovering tissue is critical. Plotted values represent means ± SE; n = 8. The significant increases in TER in the presence of 200-300 mosmol/kgH2O mucosal urea and the significant inhibitory effect of 300 mosmol/kgH2O serosal urea were determined by using 2-way ANOVA on repeated measures (P < 0.05).

In previous studies (6), we have postulated that tight junction reassembly would be required to initiate recovery of TER. Because of studies (27) implicating PI3K in tight junction assembly, we were particularly interested in this signaling pathway. Application of the PI3K inhibitor wortmannin (10 nM) completely inhibited PG-induced recovery but had no effect on I_sc in PG-treated tissues (Fig. 9). To rule out an effect of wortmannin on restitution, we calculated the percentage of denuded mucosa during in vitro recovery, as in our initial experiments. Following a 60-min recovery period, there was no significant effect of wortmannin on percentage of denuded mucosa (1.9 ± 1.4%) compared with other treatment groups (Table 1), suggesting that wortmannin inhibited paracellular effects of PG addition. However, wortmannin appeared to reduce the small recovery response of ischemia-injured tissues treated with indomethacin alone, suggesting the possibility of nonspecific toxic effects of wortmannin. Therefore, we also assessed the effects of the alternative PI3K inhibitor LY-294002 (10 µM). This agent fully inhibited recovery of TER in PG-treated tissues. However, LY-294002 did not fully inhibit the PG-stimulated elevations in I_sc, which remained significantly elevated compared with tissues treated with indomethacin alone (Fig. 10). LY-294002 appeared to have no effect on TER or I_sc measurements when applied to ischemia-injured tissues treated with indomethacin in the absence of PGs.

View larger version (22K): [in this window] [in a new window]   Fig. 9.   Electrical responses of tissues to inhibition of phosphatidylinositol 3-kinase (PI3K). A: pretreatment with the PI3K inhibitor wortmannin (10 nM) fully inhibited recovery of TER in tissues treated with indomethacin (5 µM) and PGs (1 µM). B: Isc in tissues treated with indomethacin and PGs was no different from Isc levels in tissues additionally treated with wortmannin, suggesting that inhibition of TER did not relate to blockade of Isc. Significant inhibitory effect of wortmannin on tissues treated with indomethacin and PGs was determined by using 2-way ANOVA on repeated measures (P < 0.05).

View larger version (23K): [in this window] [in a new window]   Fig. 10.   Electrical responses of tissues to inhibition of PI3K with LY-290042. A: tissues pretreated with the alternative PI3K inhibitor LY-290042 (10 µM) had no evidence of recovery of TER in response to treatment with PGs, whereas LY-290042 had no effect on tissues treated solely with indomethacin. B: Isc in tissues pretreated with indomethacin and PGs was partially inhibited by LY-290042. However, in previous studies (4), we have shown that complete inhibition of Isc is required to inhibit recovery of TER, suggesting that inhibition of recovery of TER by LY-290042 is related to other mechanisms involving PI3K. Significant inhibitory effect of LY-290042 on tissues treated with indomethacin and PGs was determined by using 2-way ANOVA on repeated measures (P < 0.05).

Since we have postulated that PG-induced Cl secretion sets up an osmotic gradient that is in turn responsible for at least part of the recovery of paracellular resistance, we wanted to determine whether the PI3K inhibitor LY-294002 would also inhibit urea-stimulated recovery. Therefore, tissues were treated with 200 mosmol/kgH₂O urea on their mucosal surface in the presence or absence of LY-294002 (10 µM). The PI3K inhibitor LY-294002 inhibited the effect of urea (Fig. 11), suggesting that PI3K signaling is required for osmotic load-induced recovery of TER, similar to that of PG-induced recovery of TER. Similar results were obtained in tissues pretreated with wortmannin (data not shown).

View larger version (30K): [in this window] [in a new window]   Fig. 11.   Effect of inhibition of PI3K on urea-induced recovery. Treatment of ischemia-injured tissues with 200 mosmol/kgH2O urea on the mucosal side stimulated recovery of TER, as in previous experiments. However, tissues pretreated with the PI3K inhibitor LY-294002 (10 µM) failed to recover when subjected to mucosal urea, suggesting an important role for PI3K in osmotic load-stimulated mucosal recovery.

In additional experiments assessing PI3K-mediated events, we sought to further define the role of PGs and PI3K inhibitors on select components of the tight junction, since it is this structure that is largely responsible for regulating paracellular permeability (13). Therefore, we assessed the tissue expression of the tight junction transmembrane proteins occludin and claudin-5 after 120 min of recovery in the presence of PGs and wortmannin. We used a technique to isolate epithelial cells from remaining mucosal elements to be sure that we were not detecting tight junction proteins from other tissues such as endothelium. Microscopic studies confirmed complete epithelial separation from mucosal villi after the isolation procedure (data not shown). Indomethacin appeared to reduce the expression of claudin-5 in ischemia-injured mucosal epithelium, but there was no apparent effect of any of our treatments on occludin expression (Fig. 12). However, in further studies using immunofluorescence microscopy, we noted differences in the distribution of occludin in the various treatment groups (Fig. 13). In particular, we noted interepithelial localization of occludin labeling in control epithelium that was disrupted in ischemia-injured tissue bathed in indomethacin (5 µM) for 120 min. Treatment with PGs (1 µM) appeared to restore the normal interepithelial junctional distribution of occludin, whereas pretreatment of tissues with wortmannin (10 nM) inhibited the ability of PGs to restore occludin distribution. To seek further evidence of tight junction structural restoration in the presence of PGs, we also performed immunofluorescence experiments to assess the localization of the tight junction cytoplasmic plaque protein ZO-1. These experiments revealed highly selective localization of ZO-1 to the tight junction in control tissues and ischemia-injured tissues exposed to PGs (1 µM). In contrast, ischemia-injured tissues recovered in the presence of indomethacin (5 µM) alone or with wortmannin (10 nM) had evidence of diffuse staining in the apical region of recovering epithelial cells (Fig. 14).

View larger version (29K): [in this window] [in a new window]   Fig. 12.   Immunoblots for tight junction proteins. Tissues were subjected to a 120-min in vitro recovery period, including control tissue (cont) and ischemia-injured tissues treated with indomethacin (indo, 5 µM), indomethacin and PGs (1 µM), or indomethacin, PGs, and wortmannin (wort, 10 nM). Epithelium was isolated from mucosa before Western blotting. A: representative Western blot for claudin-5 (a 22-kDa protein located between markers for 31.6 and 17.8 kDa) showed evidence of reduced expression of claudin-5 in ischemia-injured epithelium treated with indomethacin compared with other treatments. B: occludin is a protein of variable molecular mass depending on the degree of phosphorylation, typically in the range of 65-75 kDa (26). There are no apparent differences in expression of occludin (multiple bands located between markers for 79 and 41 kDa) between the different treatment groups.

View larger version (133K): [in this window] [in a new window]   Fig. 13.   Immunofluorescence microscopic evaluation of ischemia-injured tissues for occludin. A: normal mucosa has evidence of accumulation of occludin at the lateral membrane of cells, particularly toward the apical surface of the epithelium where interepithelial junctions reside (arrows). B: ischemia-injured mucosa following a 120-min in vitro recovery period in the presence of indomethacin (5 µM). Note the disorganized appearance of occludin fluorescence, with a lack of accumulation of occludin at the region of the interepithelial junctions. C: tissues treated with PGs (1 µM) have a pattern of occludin fluorescence that resembles that of normal tissues (arrows). D: tissues pretreated with indomethacin and wortmannin (10 nM) and subsequently treated with PGs have poorly organized occludin fluorescence similar to that of tissues treated with indomethacin alone. Bar = 5 µm.

View larger version (46K): [in this window] [in a new window]   Fig. 14.   Immunofluorescence microscopic evaluation of ischemia-injured tissues for zonula occludens-1 (ZO-1). Normal mucosa has ZO-1 exclusively localized to the region of the tight junction (A), whereas ischemia-injured mucosa exposed to indomethacin (5 µM) for 120 min has evidence of diffuse ZO-1 fluorescence at the apical region of recovering epithelial cells (B). Treatment of ischemia-injured tissues with indomethacin and PGs (1 µM) for 120 min restores localization of ZO-1 to the tight junctions (C), an effect that is inhibited by pretreatment of tissues with wortmannin (D). Bar = 5 µm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Mechanisms believed to be critical for recovery of injured epithelium include restitution (18, 21) and, in the case of small intestinal mucosa, villous contraction (19). Restitution is a broad term that denotes recovery of an intact monolayer of epithelium across a previously denuded region of the mucosa (25). Thus restitution may be broken down into epithelial migratory events and tight junction resealing events. PGs have not been extensively linked to villous contraction or epithelial migration, although there is evidence that PGE₂ stimulates contraction of villi in normal mucosa (9) and that the cyclooxygenase inhibitor piroxicam suppresses epithelial migration stimulated by growth factors in cultured intestinal epithelial cells (29). However, we have not found any evidence for an effect of PGs on either villous contraction or epithelial migration in ischemia-injured porcine ileal mucosa. For example, tissues exposed to 45 min of ischemia have histological evidence of a complete epithelial monolayer after 60 min of in vitro recovery in tissues regardless of whether they are treated with indomethacin alone or indomethacin and exogenous PGs (Fig. 2). Nonetheless, PGs stimulate significant elevations in TER and reductions in permeability to mannitol and inulin, leading us to focus on potential effects of PGs on paracellular structures. The present studies provide further evidence for an effect of PGs on the paracellular space. For example, J_sm of Na⁺ (which reflects changes in TER) significantly correlated with J_ms of the paracellular probes inulin and mannitol. Furthermore, quantitation of the dimensions of the junctional region of the paracellular space revealed significant reductions in the area of this space in response to PGs, whereas serosal hydrostatic pressure significantly increased the area of the paracellular space and inhibited the actions of PGs on recovery of TER.

Since tight junctions largely regulate paracellular permeability, it is likely that at least a component of the action of PGs is directed at these structures. Immunofluorescence of occludin and ZO-1 would tend to support this conclusion, since localization of these tight junction proteins to the region of the interepithelial junctions was associated with peak TER in response to PGs. However, it is also possible that PGs have an effect on the subjunctional paracellular space, the collapse of which might be responsible for a component of the recovery of TER. The importance of the proximity of epithelial lateral membranes has previously been shown to influence measurements of TER (10, 11), and the experiments with serosal pressure support the idea that dilating the paracellular space reduces the ability of PGs to stimulate recovery of TER. However, electron micrographs also showed evidence of dilation of the tight junction in response to serosal pressure, making it difficult to separate the effects of this maneuver on the paracellular space and the tight junction. Similarly, ischemia-injured tissues treated with indomethacin alone had dilated tight junctions and paracellular spaces, whereas those treated with PGs had closely apposed tight junctions and paracellular spaces. However, it is likely that tight junction resealing precedes collapse of the subjacent paracellular space because the continued presence of a dilated tight junction would allow extracellular fluid to enter the paracellular space.

Mechanisms of tight junction resealing following ischemia have not been fully characterized. First, it is likely that tight junctions have to reassemble following ischemic injury, since ischemia or associated ATP depletion disrupts tight junction integrity (16, 17). Reassembly of tight junctions following events such as ATP depletion involves localization of integral membrane proteins such as occludin to the apical-most aspect of the lateral epithelial membrane, along with colocalization of cytoplasmic proteins such as ZO-1 (28). Similarly, studies utilizing a calcium switch (chelation and subsequent repletion of calcium) to disrupt and allow recovery of tight junctions documented the critical role of integral membrane proteins in orchestrating reassembly of tight junctions (22). Although we do not know if these same mechanisms are responsible for PG-stimulated recovery of TER, we do have evidence on immunofluorescence that PGs restore the distribution of the tight junction integral membrane protein occludin and the cytoplasmic plaque protein ZO-1 to the region of the interepithelial junction. However, there was no difference in the expression of occludin in response to PG treatment, suggesting that PGs stimulate movement of preexisting occludin and ZO-1 dispersed throughout the cell during ischemic injury to the interepithelial junction during recovery.

In the present study, we were able to use PI3K inhibitors to functionally separate changes in I_sc and TER, both of which are stimulated by PGs. We know from previous studies that inhibition of I_sc and the associated secretion of Cl largely blocks the action of PGs on recovery of TER (4). However, it now appears that PI3K-mediated events are critical for recovery of TER despite the continued presence of elevations in I_sc. Inhibitors of PI3K also blocked recovery of TER in response to mucosal osmotic loads of urea. Together, this data suggests that PI3K-mediated events are downstream of mechanisms resulting in a mucosal-to-serosal osmotic gradients, including mucosal urea and PG-induced Cl secretion (proposed model shown in Fig. 15). However, as an alternate possibility, it is also conceivable that PGs require both elevations in I_sc and intact PI3K signaling to stimulate recovery of TER. As far as the specific mechanisms involved in PI3K-sensitive recovery of TER, this will have to await further study. However, previous studies demonstrating preferential binding of the p38 regulatory domain of PI3K to occludin (20) and a role for PI3K in junctional actin rearrangement (23) suggest potential important functions of this enzyme in tight junction resealing.

View larger version (59K): [in this window] [in a new window]   Fig. 15.   Proposed model of PG signaling pathways. We have previously shown that PGE2 interacts with EP2 and EP3 receptors, which would be expected to activate protein kinase A (PKA) or stimulate increases in intracellular Ca2+, respectively (3). These second messengers would then phosphorylate apical Cl and Na+ channels, in the case of PKA, or result in opening of basolateral K+ channels, in the case of increased intracellular Ca2+ levels. The combined effect of these second messenger-ion channel interactions is secretion of Cl and blockade of Na+ absorption, which we believe results in a mucosal-to-serosal osmotic gradient (4). The effects of PGE2 can be enhanced with the additional application of PGI2, which appears to activate cholinergic nerves (5), and genistein, which directly enhances Cl secretion via the CFTR (6). As shown in the present studies, the addition of PGE2 and PGI2 to the serosal surface of ischemia-injured mucosa or the application of artificial osmotic gradients to the mucosal surface of tissues stimulates recovery of TER. These effects are prevented with the application of PI3K inhibitors, suggesting that this enzyme is downstream of secretory and osmotic processes. Furthermore, we have shown that PGs restore interepithelial junctional localization of occludin and ZO-1, which is inhibited by the PI3K blocker wortmannin, suggesting possible mechanisms whereby PGs may stimulate tight junction recovery following ischemic injury.