Polychlorinated Biphenyls Disrupt Intestinal Integrity via NADPH Oxidase-Induced Alterations of Tight Junction Protein Expression

Background Polychlorinated biphenyls (PCBs) are widely distributed environmental toxicants that contribute to numerous disease states. The main route of exposure to PCBs is through the gastrointestinal tract; however, little is known about the effects of PCBs on intestinal epithelial barrier functions. Objective The aim of the present study was to address the hypothesis that highly chlorinated PCBs can disrupt gut integrity at the level of tight junction (TJ) proteins. Methods Caco-2 human colon adenocarcinoma cells were exposed to one of the following PCB congeners: PCB153, PCB118, PCB104, and PCB126. We then assessed NAD(P)H oxidase (NOX) activity and expression and the barrier function of Caco-2 cells. In addition, the integrity of intestinal barrier function and expression of TJ proteins were evaluated in C57BL/6 mice exposed to individual PCBs by oral gavage. Results Exposure of Caco-2 cells to individual PCB congeners resulted in activation of NOX and increased permeability of fluorescein isothiocyanate (FITC)-labeled dextran (4 kDa). Treatment with PCB congeners also disrupted expression of TJ proteins zonula occludens-1 (ZO-1) and occludin in Caco-2 cells. Importantly, inhibition of NOX by apocynin significantly protected against PCB-mediated increase in epithelial permeability and alterations of ZO-1 protein expression. Exposure to PCBs also resulted in alterations of gut permeability via decreased expression of TJ proteins in an intact physiological animal model. Conclusions These results suggest that oral exposure to highly chlorinated PCBs disrupts intestinal epithelial integrity and may directly contribute to the systemic effects of these toxicants.

volume 118 | number 7 | July 2010 • Environmental Health Perspectives Research Polychlorinated biphenyls (PCBs) are among the most persistent and widespread environ mental pollutants today. Currently, the amount of PCBs globally distributed through out the environment is estimated at > 1.5 mil lion metric tons (Tanabe 1988), and their bioaccumulation and bioconcentration in the food chain are well established. Indeed, the primary route of PCB exposure for most of the population is through dietary intake, with highfishconsuming populations being at higher risk (Norström et al. 2009). Studies performed on population cohorts in the Great Lakes region consistently report that individu als with high fish consumption have a higher PCB body burden that may be related to the alterations of pre natal and post natal cognitive development, behavior, and childhood IQ (Li et al. 2009;Stewart et al. 2008).
Very little is known about the effects of PCBs on the gastrointestinal (GI) system. We propose that exposure to PCBs through diet may influence individual health outcome via alterations of tight junctions (TJs) that seal together the adjacent intestinal epithelial cells, leading to increased intestinal perme ability. The small intestine is characterized by the presence of villi that contain epithelium, which functions as a highly dynamic and selective barrier between the outside environ ment and underlying tissue. TJs consist of an intricate combination of trans membrane (e.g., occludin) and cytoplasmic accessory proteins [e.g., zonula occludens1 (ZO1)] linked to the actin cytoskeleton. The func tion of TJs depends on the proper expression and localization of these proteins. For exam ple, regional loss of occludin in the intestinal epithelium was noted in an experimental model of jaundiceinduced gut barrier dys function (Yang et al. 2005), and a decrease in ZO1 expression was observed in hemor rhagic shock and resuscitation (Yang et al. 2003). Loss of GI epithelial integrity has been associated with the pathogenesis of several acute and chronic diseases, including diabetes, allergies, asthma, and autoimmune diseases (Liu et al. 2005).
The mechanisms of cellular and tissue toxicity of PCBs are not fully understood. However, PCBs are potent activators of oxi dative stress; therefore, induction of reactive oxygen species (ROS) may be one of the mechanisms contributing to gut dysfunction. Recent evidence implicates NAD(P)H oxi dase (NOX) as an important source of ROS (Harrison and Gongora 2009). Active NOX generates superoxide via one electron transfer from NADH (reduced nicotinamide adenine dinucleotide) or NADPH (nicotinamide adenine dinucleotide phosphate). Excessive activation of NOX has been implicated in the patho genesis of inflammatory tissue injury and disease (ElBenna et al. 2009).
The aim of the present study was to evalu ate the hypothesis that PCBs can disrupt gut integrity at the level of TJs. Mechanistically, our studies focused on the role of PCBs in activation of prooxidative NOX. Using in vitro and in vivo assays, we investigated whether exposure to individual PCB con geners can disrupt integrity of the intestinal epithelium via NOXmediated disruption of TJ proteins.
We used PCBs at concentrations of 1-10 µM, with 5 µM used in most experi ments. Such concentrations of PCBs do not affect cell viability (Eum et al. 2009) and reflect serum PCB levels in acutely exposed human populations (3.4 µM or 1 ppm) (Jensen 1989;Wassermann et al. 1979).
For immunofluorescence, cultures were fixed with icecold 4% formaldehyde for 15 min. Nonspecific binding was blocked by 10% goat serum (Sigma). Cells were washed with Trisbuffered saline (TBS) contain ing 0.05% Tween 20 and incubated over night at 4°C with primary antibody diluted (1:100) in TBS. Cells were then incubated with FITCconjugated goat antimouse IgG, Texas red-conjugated goat antirabbit IgG, or FITCconjugated rabbit antigoat IgG (1:1,000 dilution) for 1 hr at 25°C. Images were viewed using a confocal microscope under identical instrument settings.
Animals. All animal protocols in this study were approved by the Committee on Animal Care at the University of Kentucky. The ani mals were treated humanely and with regard for alleviation of suffering. Male C57BL/6 mice (12-14 weeks of age; Charles River Laboratories, Wilmington, MA) were housed under 12:12 hr light:dark conditions with access to food and water ad libitum. Animals were fasted overnight before PCB treatment (150 µmol/kg). The 150µmol/kg dose results in plasma PCB levels of 5 µM, reflecting the concentration that was used in vitro. Individual PCB congeners were dissolved in tocopherol stripped safflower oil (Dyets Inc., Bethlehem, PA) and administered in a 0.1mL volume via oral gavage using an 18gauge gavage nee dle, 3 in. long, curved, 2.25 mm ball diam eter (Popper and Sons, New Hyde Park, NY). Control animals received safflower oil vehicle. Mice were either sacrificed for tissue analysis or used in intestinal permeability experiments 24 hr after PCB or vehicle treatment.
Statistical analysis. We used SigmaStat 2.0 software (Jandel Corp., San Rafael, CA) for statistical analysis. Comparisons between treatments were made by oneway or two way analysis of variance followed by Tukey's pairwise multiple comparison procedure. Statistical probability of p < 0.05 was consid ered significant.

Individual PCB congeners induce NOX activity via phosphorylation of the p47 subunit.
We investigated the hypothesis whether NOX mediated reactions are involved in PCB induced cellular toxicity. We exposed Caco2 cells to individual PCB congeners at the con centration of 5 µM and assessed NOX activity using the lucigeninenhanced chemilumines cence assay. All measurements were performed at the same time, immediately after the end of PCB exposure. As illustrated in Figure 1, a 5min exposure to all studied PCBs resulted in a significant increase in NOX activity by approximately 40%. However, NOX activity was similar to control values after treatment with PCBs for 15 or 30 min.
Assembly of the active NOX complex is associated with phosphorylation of the p47 sub unit and its translocation from the cytosol into the membrane fraction. Therefore, we measured p47 phosphorylation after PCB treatment of Caco2 cells. Figure 2A indicates that exposure to all studied PCBs for 5-15 min resulted in an increase in phosphorylated p47 (pp47) levels.
Proteins can be phosphorylated on tyrosine, serine, and/or threonine residues. Therefore, we designed additional coimmuno precipitation experiments to determine the type of p47 phosphorylation in response to PCB exposure. Treatment with PCB conge ners for 5 min increased tyrosine, serine, or threonine phosphorylation ( Figure 2B), with tyrosine phosphorylation being the most prominent type of phosphorylation. Pre treatment with the NOX inhibitor apocynin (0.5 mM) appeared to inhibit all three forms of p47 phosphorylation (data not shown).
We also performed immunoreactivity experiments in Caco2 cells treated with indi vidual PCB congeners for 5 min ( Figure 2C). Consistent with Western blotting results, we observed an increase in pp47 immuno reactivity in all PCBtreated cultures. The staining pattern suggests localization of pp47 in the membrane fractions of Caco2 cells ( Figure 2C).

Treatment with individual PCB congeners stimulates p47 translocation and its association with membrane subunits of NOX.
NOX activity requires not only phosphory lation of p47 but also its transfer from the cytoplasm into the membrane fraction and assembly with other NOX subunits into one complex. Therefore, we assessed these events in PCBtreated Caco2 cells. After treatment with individual PCB congeners for 5 min, we fractionated Caco2 cells to isolated mem brane fraction. As shown in Figure 3A, expo sure to individual PCBs markedly increased pp47 levels in cellular membranes.
In addition, we evaluated the interactions between p47 and NOX subunits gp91 and   Figure 3B indicates that p47 associates with gp91 within 5-15 min after treatment. We observed these effects in all PCBexposed groups; however, the interactions between p47 and gp91 appeared to be the greatest in cells exposed to PCB118 and PCB153 for 15 min. In addition, we observed a strongly pronounced interaction between pp47 and p22 NOX subunits at 5min exposure for all studied PCBs ( Figure 3B), further indicating the formation of the active NOX complex. NOX is involved in PCB-induced alterations of ZO-1 protein expression. The barrier function of gut epithelium is regulated by spe cialized TJ protein networks that limit passive paracellular movement molecules between adjacent epithelial cells. To determine the influence of PCBs on TJ integrity, we treated Caco2 cells with vehicle or individual PCB congeners at concentration of 1, 5, or 10 µM for 24 hr and evaluated protein expression of ZO1 and occludin. Occludin is a trans membrane TJ protein, whereas ZO1 belongs to a group of socalled accessory TJ proteins that link the transmembrane proteins with the cytoskeleton.
Exposure to all PCB congeners employed in the present study resulted in a decreased expression of ZO1 ( Figure 4A). We observed these effects in Caco2 cells exposed to indi vidual PCBs at concentrations as low as 1 µM. To assess the role of NOX in PCBmediated alterations of ZO1 expression, we pretreated Caco2 cells with apocynin, a specific NOX inhibitor, 30 min before exposure to indi vidual PCBs. As illustrated in Figure 4B, apocynin significantly protected against a decrease in ZO1 expression in cultures treated with PCB118 and PCB126.
In addition to alterations in ZO1 expres sion, treatment with PCB congeners also decreased occludin expression in Caco2 cells ( Figure 4C). However, apocynin was not effective in protection against these effects, suggesting that NOX is not involved in PCB induced alterations of occludin expression.

Individual PCBs disrupt the barrier integrity of Caco-2 cells via the NOX-dependent mechanism.
To better understand the func tional ramifications of PCBmediated acti vation of NOX, the next series of our experiments focused on permeability across Caco2 monolayers. Cultures were grown to confluence on Transwell filters until the TEER values in the range of 200-250 Ω × cm 2 were achieved around 14-21 days after seeding. Then, cultures were exposed to individual PCB congeners at 5 µM for 24 hr. Treatment with all PCB congeners used in the present study significantly increased Caco2 permeability to FD4 compared with control. Importantly, the PCBinduced permeability changes were atten uated by 30min pre treatment with the NOX inhibitor apocynin at 0.5 mM ( Figure 5).

Oral administration of PCBs alters intestinal permeability and TJ protein expression in an animal model.
In the last series of experi ments, we employed an in vivo experi mental model of PCB exposure in which we adminis tered individual PCBs by oral gavage Figure 2. Treatment with individual PCB congeners increases the expression level and phosphorylation of the p47 NOX subunit. (A) p-p47 levels in Caco-2 cells exposed to DMSO (controls) or PCBs at 5 µM for up to 30 min; Western blotting (WB) was performed using whole-cell extracts immediately after the termination of PCB exposure. (B) p-p47 levels in Caco-2 cells exposed to PCB congeners (5 µM) for 15 min, determined by immunoprecipitation (IP) with phosphotyrosine (p-Tyr), phosphoserine (p-Ser), or phosphothreonine (p-Thr) antibody, and immunoblotting with p47 antibody. The blots in A and B are representative images from at least three experiments. (C) Immunofluorescence analysis of p-p47 in cells exposed to individual PCB congeners for 5 min. Arrows indicate examples of presumed membrane localization of p-p47. The images are representative data from three independent experiments. to resemble the main route of human expo sure through the food chain. Mice exposed to individual PCB congeners (150 µmol/kg body weight) for 24 hr showed statistically signif icant increased gut permeability compared with vehicletreated controls ( Figure 6A). The assay was performed by assessing FD20 flux from the intestinal lumen into the plasma. Elevated plasma levels of FD20 were appar ent as early as 1 hr after FD20 administra tion and reached statistical significance 5 hr after FD20 treatment.
To explain the mechanisms of PCB induced intestinal permeability, we examined the expression of TJ proteins in villi sections of the small intestine of treated mice. The control animals were characterized by promi nent immuno histochemical staining for ZO1 ( Figure 6B) and occludin ( Figure 6C), indi cating borders between adjacent enterocytes (arrows point to dense bars between cells in Figures 6B and C). Oral gavage of PCB153, PCB118, or PCB104 did not appear to affect the overall morphology of villi. In contrast, the morphology of villi was highly distorted in PCB126treated animals and included areas lacking the villus epithelium. Importantly, ZO1 and occludin staining was markedly decreased, especially at the cell-cell borders, in the small intestine of animals exposed to PCB congeners. These alterations are impor tant because loss of TJ protein expression between adjacent enterocytes may correspond to "leaky" epithelium. Indeed, PCBinduced morphological alterations of TJs and focal loss of villous epithelium correspond to functional permeability changes observed in Figure 6A.

Discussion
Although diet remains the primary route of PCB exposure, little is known about the influ ence of PCBs on gut integrity. In the pres ent study, we demonstrate for the first time that individual PCB congeners induce NOX activity leading to TJ disruption and increased gut epithelial opening. This is an important finding, as a "leaky" intestinal mucosal barrier is associated with numerous pathologies (Liu et al. 2005) and may increase PCB entry into the body. Moreover, PCBmediated proinflam matory and oxidativestress-related damage to the mucosal immune system and/or intestinal epithelium may disrupt natural defenses and increase the passage of enteric pathogens across the epithelium into the circulation.
ROS and enhanced tissue oxidative stress appear to underlie the longterm toxicity of PCBs. For example, the markers of enhanced lipid peroxidation have been increased in Yusho victims 35 years after accidental poi soning with PCBs (Shimizu et al. 2007). However, the cellu lar sources of PCBinduced ROS are not fully understood; therefore, we explored the effects of individual PCB congeners on NOX assembly and activity. This focus was supported by the recent obser vation that NOX is involved in PCBinduced ROS production in neutrophils (Myhre et al. 2009). In addition, evidence from our labora tory demonstrated that PCBinduced NOX activity is required for upregulation of cell

Caco-2 permeability (% of control)
No apocynin Apocynin added volume 118 | number 7 | July 2010 • Environmental Health Perspectives adhesion molecules in human brain endothe lial cells (Eum et al. 2009). Results of the present study show that treatment with both coplanar and noncoplanar PCBs markedly increased activity of NOX. The NOX sys tem can be activated through several signal ing pathways, including mitogenactivated protein kinases (MAPKs), protein kinase C (PKC), and Ca 2+ (Myhre et al. 2009); thus, it is likely that different PCBs may stimulate different aspects of this pathway, resulting in final assembly of an active enzyme. Activation of NOX involves PKCmediated phospho rylation and the recruitment of the cytosolic subunit p47 to the membrane. Functionally, pp47 acts as an adaptor protein helping to increase the binding affinity of other NOX components, such as gp91 and p22 (Babior 2004). Consistent with these data from the literature, our results show that both coplanar PCBs (i.e., PCB126) and noncoplanar PCBs (i.e., PCB104, PCB118, and PCB153) stimu late phosphorylation of p47 and its inter action with other key NOX subunit proteins. We next addressed the hypothesis that PCBs may act through the NOX mechanism to elicit TJ modulation within intestinal epi thelium. These experiments were based on the observations that increased ROS can rap idly stimulate compartmental redistribution of occludin and ZO1 in Caco2 cultures (Musch et al. 2006). Studies performed in our labora tory (Zhong et al. 2008) and other (Persidsky et al. 2006) laboratories demonstrated that oxidative stress can alter the expression of TJ proteins acting through Ras and Rho redox responsive elements and activation of protein tyrosine kinase (Haorah et al. 2007). Importantly, exposure to PCBs can up regulate activities of MAPK, phosphatidyl inositol 3kinase, or cSrc (Eum et al. 2004(Eum et al. , 2008(Eum et al. , 2009, that is, signaling mechanisms that may influence integrity of TJs (Rao 2009). Finally, proteolysis of TJ proteins by matrix metallo proteinases and ubiquitination proteasome systems may also be responsible for PCBinduced changes in TJ protein expression (Eum et al. 2008).
We observed that inhibition of NOX by apocynin protected against PCBinduced alterations of ZO1 but not occludin expres sion. Such a differential effect indicates finely tuned regulation of TJ expression in epithelial cells. For example, occludin phosphorylation in epithelial cells is regulated by the balance between protein kinases (e.g., cSrc, PKC zeta, PKC lambda/iota) and protein phosphatases 1 and 2A, as well as protein tyrosine phosphatase 1B (Rao 2009). Thus, this complex system of phosphorylation and dephosphorylation of occludin-rather than NOXdependent reactions-might be the dominant mechanism regulating occludin expression.
Because in vitro studies cannot fully reflect the complexity of cytotoxicity of PCBs, we fur ther investigated the role of PCBs on intestinal integrity using an intact physiological model. The experiments presented here indicate for the first time that oral exposure to PCBs can significantly increase intestinal permeability to FD20. Moreover, this effect was associ ated with altered immuno reactivity for the TJ proteins ZO1 and occludin. Disruption of gut integrity by PCBs may contribute to multiple longterm and shortterm adverse effects. Initial disruption of gut integrity by PCB exposure may result in increased gut permeability to PCBs, resulting in elevated circulating levels of these toxicants. PCBs are believed to cross the gut epithelium via passive diffusion and enter into the lymphatic system (Dulfer et al. 1996). A disrupted gut barrier is likely to increase concentrations of PCBs in the systemic circulation and lead to enhanced PCB accumulation in tissues.
Another consequence of PCBinduced dis ruption of epithelial integrity may be an increase in gut permeability to other substances and pathogens present in the gut mucosa (Ley et al. 2006). Several gut pathogens have shown to be independently associated with alterations in TJs at the intestinal barrier (Hofman 2003). For example, an experimental model of Escherichia coli injection demonstrated an increase in TJ permeability associated with reorganization of the actin cyto skeleton and redistribution of TJ proteins (Qin et al. 2009). Babbin et al. (2009) recently demonstrated that compro mised intestinal barrier function induced by the bacterial virulence factor lympho statin is mediated by the Rho GTPase signaling cascade. (Additionally, high circulating levels of enteric toxins have been shown to stimulate lowlevel systemic inflammation. Chronic, albeit low level, systemic inflammation has been associ ated with several pathologies, including vascular disorders (Pizzi et al. 2008).

Conclusion
Results of the present study demon strate for the first time that the intestinal epithelium is Figure 6. Oral administration of individual PCB congeners disrupts intestinal permeability in vivo. C57BL/6 mice were administered with individual PCBs (150 µmol/kg body weight) by oral gavage, and the control group received stripped safflower oil (vehicle). (A) Intestinal permeability assessed 24 hr after PCB adminis tration using FD-20 that was allowed to circulate for 1, 2, or 5 hr; results are mean ± SE (n = 6). Immunohistochemical staining of ZO-1 (B) and occludin (C) in the small intestine of mice administered PCB congeners as described for A. Sections show individual villi; ZO-1 and occludin immuno reactivity is indicated by brown staining. In control animals, prominent ZO-1 and occludin immunoreactivity is visible at the borders of adjacent epithelial cells (B,C; white arrows). Administration of PCBs resulted in decreased ZO-1 and occludin expression, as indicated by the overall loss of ZO-1 and occludin staining. Administration of PCB126 markedly disrupted the morphology of villi, as indicated by loss of the villus epithelium (B; black arrow). *p < 0.05 compared with vehicle-treated controls. highly susceptible to oral administration of coplanar and noncoplanar PCBs. Exposure to PCBs can modulate intestinal epithelial cell functions by altering TJ protein expres sion and functional permeability. In addition, activation of NOX is a critical event in the signaling of PCBinduced cellu lar stress and cytotoxicity. Thus, our study indicates that oral exposure to coplanar or noncoplanar PCBs present a significant risk to intestinal epithelial integrity and may directly contribute to the systemic effects of these toxicants.