Proinflammatory cytokines tumor necrosis factor-α and interferon-γ modulate epithelial barrier function in Madin-Darby canine kidney cells through mitogen activated protein kinase signaling

Tight junction proteins combine to form an important barrier which serves to limit paracellular transport in epithelial cell lines [1]. Important studies have identified occludin [2], junctional adhesion molecule (JAM) [3] and the claudins [4] as tight junction proteins that restrict molecular movement within the paracellular space. These tight junction proteins form a dynamic seal between epithelial cells becoming the principle physical paracellular barrier. The extracellular domains of adjacent occludin or claudin molecules form interactions that restrict diffusion [5]. The claudin family, with now more than twenty members, has garnered much attention due to the heterogeneous expression patterns observed in a variety of epithelia and endothelial cell types. Complex arrays of claudin species with unique distribution patterns are found in each segment of the kidney [6]; for instance the distal tubule contained measurable expression of both claudin-3 and -8. MDCK II cells express claudin-1, -2, -3 and -4 [7], the expression of claudin-2 in the MDCK II cells is in part responsible for its low electrical resistance profile [8].

The response of the kidney epithelium to inflammatory mediators is complex; an early study demonstrated that Tumor Necrosis Factor-α(TNFα) exposure impaired barrier function [9]. The renal epithelium has been shown to produce TNFα as well as other potent proinflammatory cytokines in response to external stressors such as ischemia-reperfusion injury [10, 11]. TNFα mRNA levels increased significantly thirty minutes after ischemia and inhibition of TNFα bioactivity decreases neutrophil infiltration and preserves renal function [12]. Investigation of the proximal tubule LLC-PK cell model shows that in order to produce a compromised epithelium the TNFα dose must elicit apoptosis [13, 14]. The combination of TNFα and Interferon-γ (IFNγ) exposure in model epithelial cell lines such as Caco-2 cells [15, 16] and T84 cells [17, 18] results in loss of TER and increased paracellular permeability. Finally, in a recent MDCK cell study using a model of chronic exposure to TNFα for five days, claudin-1 expression eventually decreased and tight junctions were disrupted [19].

Mitogen-activated protein (MAP) kinases a family of serine-threonine kinases, have a fundamentally important roles as signal transducers. Activation of MAP kinases by various growth factors and cytokines are important molecules involved in modulating cellular responses [20, 21]. In terms of tight junction regulation the role of MAP kinase signaling has been of interest [22, 23]. MAPK kinase (MEK) overexpression led to epithelial dedifferentiation in MDCK-C7 cells [24]. Tight junction biogenesis was inhibited in MDCK cells expressing constitutively active MAP kinase; pharmacological inhibition of MEK1 signaling in these cells permitted tight junction formation [25]. Pharmacological inhibition of MEK, a Ras effector known to phosphorylate extracellular signal-regulated kinase 1 and 2 (ERK1 and ERK2), attenuated dexamethasone-induced tight junction formation in the Con8 mammary tumor cell line [26]. In these studies, the mitogenic effect of MAP kinase activity is logically opposed to tight junction formation. The analysis of the effects of external stimuli on tight junction regulation, specifically the activated signaling pathways, will provide valuable insight into tight junction regulation.

The goal of this present study was to characterize the response of MDCK cells to the combination of TNFα/IFNγ. We hypothesized that TNFα/IFNγ would impair MDCK cell tight junction function. We examined TER, paracellular flux, tight junction protein expression and localization in response to the proinflammatory cytokines. In a variety of disease states inflammation is thought to negatively impact epithelial barrier function, we report that TNFα/IFNγ co-administration to MDCK cell monolayers impaired epithelial barrier function as measured by elevated paracellular flux and produced marked elevation in transepithelial electrical resistance (TER). Occludin, claudin-1 and claudin-3 protein expression was induced by TNFα/IFNγ exposure, whereas claudin-2 levels decreased; tight junction protein localization was modulated contributing to impaired tight junction function. Inhibition of MEK1 and p38 signaling during exposure to TNFα/IFNγ, abrogated these cytokine-induced effects in MDCK cells.

Epithelial cell layers play a critical role by separating physiologically distinct compartments within most major organ systems. The loss of tight junction barrier function by deleterious inflammatory mechanisms is an important problem in renal physiology due to the contribution of these structures to the maintenance of ionic and water balance. Acute renal failure related to ischemic episodes profoundly impairs the renal epithelium. Following ischemia-reperfusion injury, cytoskeletal disruption and loss of epithelial cell polarity contribute to back-leak of glomerular filtrate into the blood [28]. In a recent study following kidney transplant, decreased ZO-1 staining was reported in response to postischemic injury [29]. Importantly, leukocyte infiltration is likely to occur in chronic renal conditions such as diabetes, hypertension, autoimmune disorders leading to production of proinflammatory cytokines [30‐32]. In this study, we employ an important renal cell model, (MDCK cells) to examine the effects of proinflammatory cytokines on barrier function.

Initially, we were interested in determining whether the proinflammatory treatment regiment was altering MDCK barrier function as a cytotoxic effect. TNFα is a known activator of NF-κB, a transcription factor that promotes survival; in a recent report NF-κB activity was inhibited by exogenous expression of Smad7 resulting in elevated apoptosis in MDCK cells [33]. However, it was recently reported that TNFα (30 ng/ml) initiated caspase-8 cleaved PARP that induced apoptosis in serum starved MDCK cells [34]. In the present study, MDCK cells were treated in media containing five percent FBS to minimize serum withdrawal responses, we report that the combination of cytokines used in this study did not significantly induce apoptosis. At the highest doses (30/60 and 100/200 ng/ml) of cytokine treatment there was a moderate elevation in LDH release, however this was less than a ten percent elevation in LDH levels compared to control. Importantly, we report that paracellular flux increased in a graded fashion with increasing dose of TNFα/IFNγ. When renal epithelial cells are exposed to agents that produce necrosis and apoptosis investigators report a decrease in TER along with a subsequent increase in paracellular flux [27, 35], we confirmed this finding in the MDCK system by using a combination of energy starvation and ATP depletion. We find that exposure of MDCK cells to TNFα/IFNγ results in a decrease in ionic permeability which is reported as increased TER values, in fact when MDCK cells are serum and energy starved ionic permeability decreased in response to TNFα/IFNγ. These data suggest that the MDCK cell response to TNFα/IFNγ is distinct from a cytotoxic insult. In support of this concept a recent study using the intestinal epithelial T84 cell line demonstrated that the combination of TNFα/IFNγ increases paracellular permeability in an apoptosis-independent manner [36]. Therefore, although it is feasible to induce cell death in MDCK cells by serum starvation and/or high doses of TNFα for an extended duration, we are confident that the perturbations reported in barrier function were conducted using conditions that would activate NF-κB minimizing induction of apoptotic events. These conditions appear to result in a reorganization of the MDCK cell junctions with minimal loss of junctional proteins.

In the present study we have demonstrated that pharmacological inhibition of MEK1 and p38 signaling in proinflammatory cytokine stimulated MDCK cells functionally protects the barrier function. Several studies indicate that MEK1 signaling increases paracellular permeability, there exists some disparity in observed cellular responses. Recently, a report demonstrated that inhibition of MEK1 signaling did not influence expression of occludin or claudin-1 or affect tight junction function in several breast cancer cell lines [37]. Also, a study using enteropathogenic Escherichia coli, showed that ERK1/2 was activated in T84 cells, but did induce tight junction barrier disruption as measured by TER [38]. However, activation of MEK1 signaling by H₂O₂ exposure in endothelial cells increased permeability and resulted in occludin disorganization [39]. Similar effects were also observed in Caco-1 and MDCK cell lines [40]. In this present study, activation of the ERK1/2 pathway by TNFα/IFNγ treatment produced altered ionic permeability and dynamic changes in junctional protein expression and localization. Additionally, we found that TNFα alone potently decreased MDCK cell ionic permeability while having only minimal impact on paracellular flux. This suggests that the observed junctional responses occur independent of apoptotic or necrotic mechanisms that likely elevate paracellular flux.

Decreased ionic permeability in response to TNFα or TNFα/IFNγ exposure coupled to the increased paracellular flux of non-charged solutes when cytokines were presented in combination is intriguing. We find that inhibition of ERK1/2 signaling increased ionic permeability toward control levels as expected but inhibition of p38 signaling further decreased ionic permeability levels above cytokine treatment alone. This suggests that activation of the p38 pathway is antagonizing ERK1/2-mediated effects on elevated TER in TNFα/IFNγ-treated MDCK cells. While the MAP kinase inhibitors produced divergent effects on cellular ionic permeability measurements both inhibitors protected against increase paracellular flux of non-charged solutes. Several recent reports reveal that ERK1/2 activation in MDCK II cells results in increased TER. For instance, a recent study of cyclosporine A treated MDCK cells produced elevated TER through a MAPK pathway [41]. In another study of MDCK II cells, EGF receptor activation resulted in increased TER with a concomitant decrease in claudin-2 expression [7]. In a recent study of MDCK II cells investigators demonstrate that these cells have endogenously low ERK1/2 activity that corresponds to high expression of claudin-2 [42]. ERK1/2 inhibition in all of these studies prevented elevation of TER in the MDCK II cell line. Recently investigators have determined that claudin-2 forms cation-selective channels in the tight junction complex [43, 44], alteration in claudin-2 expression results in perturbations in ionic permeability. Consistent with these studies we find a dose-dependent decrease in claudin-2 expression in MDCK cells treated with TNFα/IFNγ, this loss of claudin-2 correlates to a substantial reduction in ionic permeability. Elevation in TER was inhibited by treatment with the ERK1/2 inhibitor but not by inhibiting the p38 signaling pathway. These findings are consistent with the current literature demonstrating that claudin-2 levels are regulated following ERK1/2 activation in MDCK cells and its expression level will influence recorded TER from MDCK cultures.

The cellular tight junction response to proinflammatory cytokines is variable based on cell type and numerous physiological variables. Measurable changes in tight junction protein expression or localization that are predicted to play a key role in maintaining barrier function are typically more unpredictable. We report a statistically significant elevation in the protein expression of claudin-1, but not occludin or claudin-3, following exposure to TNFα/IFNγ. However, occludin protein levels are slightly elevated in response to several doses of TNFα/IFNγ tested compared to control. In this study, we report a dose-dependent decrease in claudin-2 expression following exposure to TNFα/IFNγ. The heterogeneous response of tight junctional proteins to cytokine exposure may be due to junctional remodeling which may involve additional protein synthesis and altered turnover rates. In other studies, researchers have reported decreases, increases or no change in tight protein expression following challenge with proinflammatory mediator. For instance: TNFα increased permeability while decreasing ZO-1 expression through increased NFκB signaling, in a study using Caco-2 cells [45]. Investigators report increased paracellular flux with a decrease in TER following TNFα/IFNγ exposure using a mouse cholangiocyte model; interestingly major structural changes to the tight junction proteins (occludin, claudin-1, -3, and ZO-1) were not detected [46]. Finally, using T84 cells investigators find that inhibition of MEK signaling impairs both basal and cytokine-induced tight junction formation demonstrating an increased claudin-1 and claudin-2 protein expression in response to the cytokine IL-17 [47]. Although it might be tractable to predict that exposure to proinflammatory cytokines would be correlated to decreased expression of tight junction proteins, our study is in agreement with other studies, finding moderate effects on expression.

In the present study's examination of tight junction protein localization, treatment with ERK1/2 inhibitor in the presence of TNFα/IFNγ enhanced occludin and claudin-1 expression at the junctional interface but did not significantly affect claudin-2 or claudin-3. In the TNFα/IFNγ treatment group there appears to be increased cytoplasmic staining, possibly related to a lack of tethering related to cytoskeletal rearrangements. Functionally, MDCK cells pretreated with the ERK1/2 inhibitor exhibited no change in flux or TER compared to control cells even in the presence of TNFα/IFNγ. Analysis of tight junction protein levels demonstrated that pretreatment with U0126 in the presence of TNFα/IFNγ induces protein levels similar to control however, when the ERK1/2 inhibitor was added two hours following treatment with TNFα/IFNγ, we observed a similar magnitude of elevation in their expression, similar to TNFα/IFNγ alone. This suggests that early events in the cytokine response are activated and produce lasting effects on MDCK cells.

The actin cytoskeleton maintains an intimate association with tight junctions through scaffolding proteins like ZO-1, a member of the MAGUK family [48, 49]. A recent study using a glomerular epithelial cell model exposed to TNFα for twenty-four hours reports an approximate two-fold elevation in total actin content as determined by a DNase I inhibition assay [50]. This finding is consistent with our results based on perijunctional actin staining in MDCK cells. We report a two-fold elevation of F-actin staining following exposure to TNFα/IFNγ. This increase was prevented by inhibition of ERK1/2 activation. It is also noteworthy that other researchers have identified functional relationship between MEK activation and cytoskeletal organization. MEK-dependent pathway leads to cytoskeleton disruption in Ras-transformed fibroblasts [51]. Additionally studies examining epithelial dedifferentiation show a loss of actin stress fibers [24, 52]. MDCK cells stimulated with HGF/SF produced actin cytoskeleton reorganization through the activation of MAP kinase, resulting in junctional rearrangement [53]. Constitutively active Raf-1 in a salivary epithelial cell model induced actin reorganization and disrupted tight junctions by downregulation of occludin [54]. The association between the cytoskeleton and the tight junction suggests a structural and functional relationship that provides a tractable model for understanding the regulation of barrier function.

We demonstrated that MDCK tight junctions are functionally reorganized in response to TNFα/IFNγ exposure through the activation of ERK1/2. We find the junction tightens by elevating the expression of occludin and claudin-1 in response to TNFα/IFNγ. Additionally, decreased ionic permeability arises primarily through a significant loss of claudin-2 expression due to ERK1/2 activation. Apoptotic and necrotic mechanisms in response to TNFα/IFNγ may contribute in part to the elevated paracellular flux. Based on immunofluorescent findings, occludin and claudin-1 localization appear to be in transition perhaps due to the reorganization of the actin cytoskeleton. Activated ERK1/2 and p38 signaling pathways appear to be responsible in part for both the curative and disruptive patterns observed in this study.