Tight junctions (TJs) form a barrier to the paracellular diffusion of ions and solutes across epithelia. Although transmembrane proteins of the claudin family have emerged as critical determinants of TJ permeability, little is known about the signaling pathways that control their expression. The aim of this study was to assess the role of three mitogen-activated protein kinases (MAPKs), i.e., extracellular signal-regulated kinase-1/2 (ERK1/2), c-Jun NH2-terminal kinases (JNKs), and p38 kinases, in the regulation of epithelial barrier function and claudin expression in mammary epithelial cells. Addition of either PD169316 (a p38 inhibitor) or SP600125 (a JNK inhibitor) induced formation of domes (a phenomenon dependent on TJ barrier function) and enhanced transepithelial electrical resistance, whereas U0126 (an inhibitor of the ERK1/2 activators MEK1/MEK2) had no significant effect. Similar results were obtained using mechanistically unrelated p38 or JNK inhibitors. PD169316 increased the expression of claudin-4 and -8, whereas SP600125 increased claudin-4 and -9 and downregulated claudin-8. Silencing of p38α by isoform-specific small interfering RNAs increased claudin-4 and -8 mRNAs, whereas silencing of p38β only increased claudin-4 mRNA. Silencing of either JNK1 or JNK2 increased claudin-9 mRNA expression while decreasing claudin-8 mRNA. Moreover, selective silencing of JNK2 increased claudin-4 and -7 mRNAs. Finally, both PD169316 and SP600125 inhibited the paracellular diffusion of Na+ and Cl− across epithelial monolayers. Collectively, these results provide evidence that inhibition of either p38 or JNK enhances epithelial barrier function by selectively modulating claudin expression, implying that the basal activity of these MAPKs exerts a tonic effect on TJ ionic permeability.
- mitogen-activated protein kinases
- transepithelial electrical resistance
- tight junction
- mammary epithelial cells
- c-Jun NH2-terminal kinase
a fundamental function of epithelial cells is to maintain homeostasis of the internal milieu by regulating the exchange of substances between compositionally distinct body compartments. Movement of solutes, ions, and water across the epithelial barrier occurs through both the transcellular pathway, owing to the asymmetric cellular distribution of membrane pumps and channels, and the paracellular pathway, via tight junctions (TJs). Whereas the contribution of the transcellular route has been characterized in considerable detail, the molecular mechanisms that regulate TJ permeability are still incompletely understood.
TJs consist of a belt-like network of anastomosing strands that form the boundary between the apical and basolateral membrane domains of epithelial cells. Each strand is composed of a row of transmembrane proteins and pairs with a similar strand on an adjacent cell to obliterate the intercellular space. TJs act as a regulated barrier that restricts the diffusion of ions and small hydrophilic molecules through the paracellular space. They consist of an array of transmembrane proteins linked to cytosolic adaptor proteins (63). The claudin family of transmembrane proteins comprises at least 24 isoforms that exhibit tissue-specific distribution patterns. Claudins constitute the building blocks of TJ strands and act as key regulators of their selective permeability properties. In particular, it has been shown that different claudins regulate paracellular permeability to specific ions and that the overall ionic permeability of a TJ is determined by the combination and ratios of different claudin isoforms (3, 16, 28, 69, 72).
TJ permeability is dynamically regulated in response to changing physiological needs (52), and disturbances of epithelial barrier function play a critical role in a number of pathological conditions, such as inflammatory bowel diseases (81). Hence, a better understanding of the cellular mechanisms involved in the regulation of TJ permeability could have important clinical implications. The tissue-specific regulation and physiological plasticity of TJ barrier function suggest the existence of biochemical cues that control epithelial permeability through selective modulation of claudins. However, despite the identification of a variety of different stimuli that either promote or repress the expression of specific claudins (reviewed in Refs. 29 and 72), the intracellular signaling pathways that regulate the molecular composition and permeability of TJs have just begun to be elucidated.
Mitogen-activated protein kinase (MAPK) cascades are among the best characterized intracellular signaling pathways. These cascades consist of a three-kinase module comprising a MAPK kinase kinase (MAPKKK) that activates a MAPK kinase (MAPKK) that, in turn, activates a MAPK. Activated MAPKs translocate to the nucleus, where they phosphorylate and activate transcription factors, thereby modulating gene expression. To date, six distinct groups of MAPKs have been identified in mammalian cells, including extracellular signal-regulated kinase-1 and -2 (ERK1/2), p38 (p38α/β/γ/δ), c-Jun NH2-terminal kinases (JNK1/2/3), and the more recently discovered ERK3/4, ERK5, and ERK7/8 (38). The most extensively characterized groups of MAPKs are ERK1/2, p38, and JNK. Recent evidence indicates that besides their well-established role in the control of cell growth and differentiation, MAPKs are also involved in the regulation of claudin expression and TJ permeability. Thus, for instance, it has been reported that ERK1/2 activation by hepatocyte growth factor, or ERK1/2 overexpression, increases TJ barrier function in low-resistance Madin-Darby canine kidney cells (MDCK II cells) through downregulation of claudin-2 expression (42).
The objective of the present investigation was to dissect the respective roles of ERK1/2, JNK, and p38 MAPKs in the regulation of TJ permeability and claudin expression using a mammary gland-derived epithelial cell line. The choice of this cellular model was guided by the notion that TJs of mammary alveoli are exquisitely sensitive to environmental cues. Thus, at the onset of lactation, in response to a drop in progesterone levels along with a surge in the levels of glucocorticoid hormones and prolactin, TJ permeability markedly decreases to prevent paracellular leakage of milk components from the alveolar lumen. In addition, changes in the ionic composition of milk (e.g., Na+ and K+ concentration) are largely dependent on alterations in TJ permeability (51–53). Of further relevance, activation of MAPK pathways has been shown to play a crucial role in the regulation of gene expression during the postnatal differentiation of the mammary gland (21, 41, 45). Accordingly, numerous reports indicate that mammary epithelial cell lines represent a valuable model system to investigate the factors involved in the regulation of TJ permeability (24, 57, 77, 82).
The results of our study show that inhibition of basal p38 or JNK activity enhances TJ barrier function in mammary epithelial cells by selectively modulating the expression of specific claudin isoforms.
MATERIALS AND METHODS
U0126, PD169316, SP600125, and JNK Inhibitor 1 were purchased from Alexis Biochemicals (Lausen, Switzerland), and BIRB 796 from Axon Medchem (Groningen, The Netherlands). JNK Inhibitor 1 was dissolved in sterile PBS, whereas all other reagents were dissolved in DMSO, stored at −20°C, and protected from light. DMSO alone was used as a control in cultures treated with U0126, PD169316, SP600125, or BIRB 796.
31EG4-2A4 cells (49), a clonal derivative of the nontumorigenic murine 31EG4 mammary epithelial cell line (82), were routinely grown in a 1:1 mixture of Dulbecco's modified Eagle's medium and F12 medium (DMEM-F12, Invitrogen, Basel, Switzerland) supplemented with 5% fetal calf serum (FCS, Invitrogen) and used between passages 10 and 19. For simplicity, 31EG4-2A4 cells will be referred to as 2A4 cells. Canine kidney MDCK II cells (clone 3B5; 48) were grown under the same culture conditions as above.
2A4 or MDCK II cells were resuspended in DMEM-F12 medium supplemented with 1% ITS+ Premix (6.25 μg/ml insulin, 6.25 μg/ml transferrin, 6.25 ng/ml selenious acid, 1.25 mg/ml bovine serum albumin, and 5.35 μg/ml linoleic acid; BD Biosciences, Franklin Lakes, NJ) and 1% FCS (this medium is hereinafter referred to as “assay medium”) and seeded at subconfluent density (1.6 × 106 cells/dish) into 60-mm dishes (Falcon, BD Biosciences). Cells were grown for 48 h to attain confluence, then treated with the indicated pharmacological MAPK inhibitors. After incubation for 96 h (2A4 cells) or 48 h (MDCK II cells), cell monolayers were fixed with 2.5% glutaraldehyde in 100 mM cacodylate buffer (pH 7.4) for 1 h and subsequently washed in cacodylate buffer. The cultures were then inspected using a Nikon Diaphot TMD inverted phase contrast photomicroscope, and all domes in each dish were counted.
For measurement of transepithelial electrical resistance (TER), cells were plated at confluent density (9 × 105 cells/cm2 for 2A4 cells; 5 × 105 cells/cm2 for MDCK II cells) on polycarbonate Transwell filters (0.4-μm pore size, Corning Life Sciences, Lowell, MA) in assay medium. TER was measured using a Millicell-ERS volt-ohm meter (Millipore, Billerica, MA), and monolayer TER values (expressed as Ω × cm2 ± SE) were obtained by subtracting blank (cell-free) filter readings and multiplying by the surface area of the filter (46).
For determination of the equivalent short-circuit current (Isc), 2A4 cells were resuspended in assay medium, plated at confluent density (9 × 105 cells/cm2) on polycarbonate Transwell filters (0.4-μm pore size, Corning Life Sciences), and grown for 24 h before addition of the indicated pharmacological MAPK inhibitors. After a 96-h incubation, total Isc was calculated according to Ohm's law from the values of transepithelial potential difference and resistance measured with a Millicell device (Millipore). Ouabain-insensitive Isc was measured 30 min after addition of 1 mM ouabain to the basal compartment, as described previously (47). Results are given as the mean ± SE of 3 independent experiments. All experiments were performed on cells at the same passage. Comparisons between two groups were performed by Mann-Whitney U-test for unpaired data or by paired Student's t-test. Multiple group analyses were done using ANOVA and Tukey's test for post hoc analysis.
Western blot analysis.
Confluent cultures of 2A4 cells in 60-mm dishes were incubated in lysis buffer (50 mM Tris·HCl, pH 8.0, 150 mM NaCl, 5 mM EDTA, pH 8.0, 1% Triton X-100, and 1% NP-40) containing both Complete Protease Inhibitor and PhosphoStop Phosphatase Inhibitor cocktails (Roche, Mannheim, Germany) for 30 min on ice. Lysed cells were scraped, transferred to an Eppendorf tube, and centrifuged. The supernatant was collected and protein content measured by the BCA Protein assay kit (Pierce, Rockford, IL). Equal amounts of total extracts (20 μg) were separated by 10% SDS-PAGE before transfer onto either polyvinylidene difluoride or nitrocellulose membranes using the iBlot Dry Blotting System (Invitrogen). To control for loading differences, proteins were both stained with Coomassie blue and immunoblotted for total actin. Membranes were blocked for 3 h at room temperature in Tris-buffered saline, pH 7.6, containing 0.2% (vol/vol) Tween 20 (TBS-Tween) and 5% (wt/vol) non-fat milk powder (blocking buffer), then incubated overnight at 4°C with the following primary antibodies diluted in TBS-Tween containing 5% BSA: rabbit sera (Cell Signaling, Danvers, MA) against c-Jun (clone 60A8, 1:2,000), phospho-c-Jun (Ser63) (1:2,000), MAPKAPK-2 (1:4,000), phospho-MAPKAPK-2 (Thr334) (clone 27B7, 1:500), p44/p42 MAPK (ERK1/2) (clone 137F5, 1:4,000), and phospho-p44/p42 MAPK (Thr202/Tyr204) (1:1,000); mouse sera (Invitrogen) against claudin-1 (clone 2H10D10, 1 μg/ml) and claudin-2 (clone 12H12, 1 μg/ml); rabbit sera (Invitrogen) against claudin-4 (1 μg/ml), claudin-7 (1 μg/ml) and claudin-8 (1 μg/ml); rabbit serum (Aviva Systems Biology, San Diego, CA) against claudin-9 (2 μg/ml); rabbit serum against total actin (gift of G. Gabbiani, Geneva; 1:2,000). After extensive washing in TBS-Tween, the membranes were incubated for 1 h at room temperature with horseradish peroxidase-conjugated secondary antibodies (GE Healthcare, Otelfingen, Switzerland), diluted 1:6,000 in blocking buffer. Membranes were then washed extensively in TBS-Tween and antigen-antibody complexes were detected by enhanced chemiluminescence (ECL Plus, GE Healthcare) using the ChemiDoc XRS System (Bio-Rad, Reinach, Switzerland).
Quantitative real-time RT-PCR.
Total RNA was extracted from 2A4 cells using NucleoSpin RNA II (Macherey-Nagel, Düren, Germany) and subsequently treated with DNase I. RNA integrity was assessed using RNA 6000 nanochips with an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). First-strand complementary DNA was synthesized from 1 μg of purified RNA using the SuperScript II RNase H(−) reverse transcriptase (Invitrogen) and random hexadeoxynucleotides. Reactions were set up in 384-well plates using a 7900HT Real-Time PCR System (Applied Biosystems, Foster City, CA), and all samples were assayed in triplicate. Optical data obtained were analyzed using the default and variable parameters available in the Sequence Detection Systems software (SDS, version 2.2.2; Applied Biosystems). Expression level of target genes was normalized according to geNorm (70), using as endogenous control genes mouse hypoxanthine-guanine phosphoribosyl transferase, mouse ribosomal protein S9, mouse β-actin, mouse glucuronidase-β, and mouse β-tubulin. Primer sequences are listed in supplemental data Table I (supplemental material can be found online at the American J Physiology-Cell Physiology Web site).
Synthetic small interfering RNAs (siRNAs) against mRNAs encoding for mouse JNK and p38 isoforms were obtained from Qiagen (Hombrechtikon, Switzerland). The following target sequences were used: for p38α, 5′-CTCCTTTACTATCTTTCTCAA-3′; for p38β, 5′-CTGAGCGATGAGCATGTTCAA-3′; for JNK1, 5′-CTCAGAGCATAACAAACTTAA-3′; for JNK2, 5′-ACCGTCATATATCATATCTTA-3′. 2A4 cells (4 × 105) were resuspended in assay medium and seeded into 35-mm culture dishes (Corning Life Sciences). After 24-h incubation at 37°C in a humidified 5% CO2 atmosphere, cells were transfected with 10 nM siRNAs using HiPerfect Transfection Reagent (Qiagen), according to the manufacturer's instructions. Cells were harvested 72 h after transfection, and total RNA was extracted as described above. Knockdown efficacy was evaluated using quantitative real-time RT-PCR. Both nontransfected cells and cells transfected with 10 nM nonsilencing siRNA (AllStars Negative Control, Qiagen) were used as negative controls.
Transepithelial ionic gradients.
2A4 cells were resuspended in assay medium, plated at confluent density (9 × 105 cells/cm2) on polycarbonate Transwell filters (0.4-μm pore size, Corning Life Sciences), and grown for 24 h before addition of the indicated pharmacological MAPK inhibitors. After a 96-h incubation, the ability of 2A4 cells to maintain transepithelial Na+ and Cl− gradients was determined by substituting assay medium with one of the following solutions: 1) NaCl-containing defined medium (120 mM NaCl, 5 mM KCl, 4 mM NaHCO3, 1 mM CaCl2, 1 mM MgSO4, 0.2 mM NaH2PO4, 0.15 mM Na2HPO4, 5 mM glucose, 10 mM lactate, 1 mM pyruvate, and 20 mM HEPES, pH 7.4); 2) defined medium in which NaCl was replaced by N-methyl-d-glucamine-Cl and other Na+ salts were replaced by K+ salts (nominally Na+-free medium); and 3) defined medium in which NaCl was replaced by Na-gluconate, KCl by K-gluconate, and CaCl2 by CaSO4 (nominally Cl−-free medium). Briefly, the cells were extensively washed with either Na+- or Cl−-free defined medium before addition of NaCl-containing defined medium in the apical compartment and either Na+-free medium or Cl−-free medium in the basal compartment, or the reverse. After 6-h incubation at 37°C in a humidified 5% CO2 atmosphere, Na+ and Cl− concentration in apical and basolateral media was measured using ion-specific electrodes in a Synchron LX20 apparatus (Beckman Coulter, Fullerton, CA) (16). Results are expressed mM ± SE.
Pharmacological inhibition of p38 or JNK pathways enhances tight junction barrier function in mammary epithelial cells.
To investigate the role of ERK1/2, JNKs, and p38 MAPKs in the control of TJ permeability, we used the 2A4 murine mammary epithelial cell line (49). 2A4 cells are a clonal subpopulation of the 31EG4 cell line, which has been used extensively to investigate the regulation of TJ barrier function (77, 82). In postconfluent monolayer cultures, 2A4 cells form numerous blister-like structures, or domes, generated by focal accumulation of fluid between the basal cell surface and the underlying substrate, a phenomenon that is largely dependent on the barrier function of TJs (17). To get insight into the role of MAPK pathways in the regulation of TJ permeability, we first examined the impact of small-molecule MAPK inhibitors on dome formation. Initial experiments carried out in routine growth medium (containing 5% FCS) showed that addition of either PD169316, a selective inhibitor of the p38 pathway (66), or SP600125, a specific inhibitor of the JNK pathway (10), increased dome formation by 2A4 cells, whereas U0126, an inhibitor of the upstream ERK1/2 activators MEK1/2 (22), had no obvious effect (data not shown). We next wished to determine the potential effect of PD169316 and SP600125 under conditions that are not permissive for spontaneous dome formation. We found that lowering FCS concentration to 1% resulted in minimal dome formation in untreated monolayers of 2A4 cells, thereby providing a convenient baseline for assessing cellular responses to MAPK inhibitors. 2A4 cells were therefore grown in DMEM-F12 supplemented with 1% FCS (“assay medium”) for the remainder of this study. As illustrated in Fig. 1A, treatment of postconfluent monolayers of 2A4 cells with either the p38 inhibitor PD169316 or the JNK inhibitor SP600125, but not the MEK1/2 inhibitor U0126, promoted the formation of numerous domes. A quantitative analysis demonstrated that PD169316 and SP600125 induced dome formation in a dose-dependent manner, whereas U0126 had no significant effect at the concentrations we used (Fig. 1B).
To determine whether p38 or JNK inhibitors induce dome formation by decreasing the ionic permeability of TJs, we measured TER, a reliable indicator of TJ barrier function, at different time points (24 h to 96 h) after treatment. Incubation of 2A4 cells grown on polycarbonate filters with either PD169316 (1 μM to 5 μM) or SP600125 (2 μM to 10 μM) resulted in a time- and dose-dependent increase of TER, whereas U0126 (2 μM to 10 μM) had no significant effect (Fig. 1C, and data not shown). Notably, the maximum increase of TER was observed after a 96-h incubation with either 2 μM PD169316 or 10 μM SP600125, i.e., the same experimental conditions that yielded optimal induction of dome formation (Fig. 1, B and C, and data not shown). The 96-h time point and the above indicated concentrations of MAPK inhibitors were therefore selected for subsequent experiments.
To confirm that PD169316, SP600125, and U0126 efficiently inhibited the activity of p38, JNK, or MEK1/2 in our system, we assessed their effect on the phosphorylation of MAPK-activated protein kinase-2 (MAPKAPK-2), c-Jun, and ERK1/2, which are immediate downstream targets of p38 (60), JNK (18), and MEK1/2 (61), respectively. To this end, 2A4 cells were pretreated with each MAPK inhibitor for 96h, then briefly exposed to either anisomycin, a well-known activator of p38 and JNK (31), or 20% FCS to activate the MEK-ERK pathway (65). As expected, the p38 inhibitor PD169316 (2 μM) decreased anisomycin-induced MAPKAPK-2 phosphorylation (Fig. 2A), while the JNK inhibitor SP600125 (10 μM) decreased anisomycin-induced c-Jun phosphorylation (Fig. 2B). Pretreatment with the MEK1/2 inhibitor U0126 (10 μM) reduced FCS-induced ERK1/2 phosphorylation (Fig. 2C).
To verify that the effect of PD169316 and SP600125 on TJ barrier function was not unique to 2A4 mammary epithelial cells, we extended our analysis to MDCK II cells, which are a well-established model for studying the regulation of TJ permeability (72). We found that in MDCK II cells as well, PD169316 and SP600125 induced a dose-dependent increase in both dome formation and TER, while U0126 had no significant effect (supplemental data Fig. 1).
PD169316, a pyridinyl imidazole compound, and SP600125, an anthrapyrazole, are known to act by blocking the ATP-binding site of p38 and JNK, respectively (10, 66). Despite their reported selectivity (10, 66), it is possible that these drugs exert some inhibitory activity toward other protein kinases (7, 54). To ascertain that the TER increase induced by treatment with PD169316 or SP600125 was actually mediated by inhibition of p38 or JNK, respectively, we utilized additional pharmacological agents that inhibit these MAPKs through distinct mechanisms. BIRB 796, a diaryl urea compound that stabilizes p38 in a conformation incompatible with ATP binding (58), induced a dose-dependent increase of TER in 2A4 cells at concentrations (20–500 nM) reported to specifically inhibit p38 isoforms (37) (Fig. 3A). Likewise, JNK Inhibitor-1, a cell-permeable peptide that competitively blocks the interaction between JNK and its substrate c-Jun, thereby interrupting the JNK signaling cascade (15), induced a significant dose-dependent increase in TER (Fig. 3B). These findings validated the concept that selective interference with either the p38 or JNK pathway enhances epithelial barrier function in 2A4 mammary epithelial cells.
Inhibition of p38 or JNK pathways differentially modulates claudin expression in mammary epithelial cells.
Since the ionic permeability of TJs is largely determined by the level of expression and ratio of different claudin isoforms (67, 69, 72), we next investigated whether the increase in TER induced by inhibitors of p38 or JNK correlated with changes in claudin expression. Among the 24 claudin isoforms identified to date in mouse epithelial cells (68, 72), claudin-1, -2, -4, -7, -8, and -9 were selected for study because of their well-established role in the regulation of TJ ionic permeability (3, 62, 72) and/or their reported expression in murine mammary gland epithelium (11, 12).
Preliminary analysis by Western blot documented the expression of claudin-1, -4, -7, -8, and -9 in postconfluent monolayers of untreated 2A4 cells (data not shown). In contrast, claudin-2 was not expressed in 2A4 cells, a finding confirmed by real-time RT-PCR (data not shown). We next examined by Western blot the effect of a 96-h incubation with either PD169316 (2 μM) or SP600125 (10 μM) on the expression of claudin-1, -4, -7, -8, and -9. Inhibition of the p38 pathway by PD169316 was associated with increased expression of claudin-4 and pronounced upregulation of claudin-8, while leaving unaffected the levels of claudin-1 , -7, and -9 (Fig. 4A). Inhibition of the JNK pathway by SP600125 resulted in upregulation of claudin-4 and -9 and robust downregulation of claudin-8, without ostensibly altering the expression of claudin-1 and -7 (Fig. 4B).
To determine whether claudin localization was modified on inhibition of either p38 or JNK, we performed immunofluorescence analysis using antibodies to claudin-4, -7, and -8. All claudins examined showed a predominant lateral membrane localization with faint punctuate cytoplasmic staining. This pattern of claudin distribution was not ostensibly altered by treatment with p38 or JNK inhibitors (supplemental data Fig. 2). Because claudin-7 was reported to be expressed at the basolateral membrane in some tissues (11), we more precisely assessed its localization by confocal microscopy. We found that in 2A4 cells, claudin-7 was concentrated at the level of the junctional region, and that this localization was not modified by p38 or JNK inhibitors (supplemental data Fig. 2). Altogether, these findings indicate that treatment with MAPK inhibitors does not cause major alterations in the cellular distribution of claudins.
The activity of p38 and JNK pathways is largely mediated through transcriptional regulation of target genes (18, 59). To determine whether the observed modulation of claudin protein expression was the result of alterations in claudin gene transcription, we measured by real-time RT-PCR the levels of claudin-1, -4, -7, -8, and -9 mRNAs in 2A4 cell cultures after a 96-h incubation with either PD169316 (2 μM) or SP600125 (10 μM). In accordance with the data obtained by Western blot analysis, inhibition of the p38 pathway by PD169316 caused a significant increase of claudin-4 mRNA level and a more pronounced upregulation of claudin-8 mRNA transcription, without significantly modifying the mRNA levels of claudin-1, -7, and -9 (Fig. 5A and data not shown). Inhibition of the JNK pathway by SP600125 induced a slight but significant increase in claudin-7 mRNA, a marked upregulation of claudin-4 and -9 mRNA, and a robust downregulation of claudin-8 mRNA, without modulating mRNA levels of claudin-1 (Fig. 5B and data not shown).
To broaden our analysis of claudin modulation by MAPK inhibitors, we subsequently screened by RT-PCR for the expression of eight additional claudin isoforms, including claudin-5, -6, -11, -12, -14, -15, -16, and -17. Claudin-5, -6, -11, -14, -15, and -16 were not expressed by 2A4 mammary epithelial cells, while claudin-17 was barely detectable. In contrast, claudin-12 was expressed at substantial levels, which were further slightly incremented on inhibition of the p38 pathway by PD169316 (data not shown).
Among the four known isoforms of p38 MAPK, p38α and p38β are ubiquitously expressed, whereas p38γ and p38δ have a tissue-specific expression pattern (20). Among the three known members of the JNK family, JNK1 and JNK2 are widely expressed, whereas expression of JNK3 is largely restricted to neurons (13). To further validate the specificity of the pharmacological inhibitors of p38 and JNK, and to define the kinase isoforms responsible for modulation of claudin expression, we silenced p38α, p38β, JNK1, or JNK2 with siRNAs specifically targeting individual MAPK isoforms. We then measured by real-time RT-PCR the resulting changes in mRNA levels of p38 and JNK isoforms, as well as mRNA levels of claudin-1, -4, -7, -8, -9, and -12.
Isoform-specific siRNAs decreased the expression of p38α and p38β mRNAs by 55% and 59%, respectively (Fig. 6A). Albeit slightly less efficient, siRNAs to JNK isoforms led to a significant reduction of JNK1 (43%) and JNK2 (44%) transcripts (Fig. 6B). Importantly, the mRNA level of each MAPK isoform was not affected by the silencing of a related isoform (Fig. 6, A and B). Silencing of either p38α or p38β was associated with increased expression of claudin-4 mRNA but did not affect the expression of claudin-1, -7, -9, and -12 mRNAs (Fig. 6C and data not shown). Downregulation of p38α, but not of p38β, robustly increased claudin-8 mRNA expression (Fig. 6C). Silencing of either JNK1 or JNK2 strongly increased claudin-9 mRNA expression, decreased claudin-8 mRNA expression, but did not modulate the expression of claudin-1 mRNA (Fig. 6D and data not shown). Downregulation of JNK2, but not of JNK1, was associated with a significant increase of claudin-4, -7, and -12 transcripts (Fig. 6D and data not shown). These results therefore confirmed the validity of the data obtained using pharmacological MAPK inhibitors. They also suggest that specific p38 and JNK isoforms play a differential role in the regulation of claudin expression.
Inhibition of p38 or JNK pathways enhances the ability of mammary epithelial cells to maintain transepithelial Na+ and Cl− gradients.
Individual claudins isoforms, such as claudin-4, -7, -8, and -9, have been shown to restrict paracellular permeability to specific ions (1, 62, 71, 80). To evaluate the functional consequences of the changes in claudin expression reported above, we investigated whether inhibition of p38 or JNK could attenuate the dissipation of an experimentally imposed transepithelial Na+ or Cl− gradient. For this purpose, postconfluent cultures of 2A4 cells on Transwell filters were washed with either Na+- or Cl−-free defined medium (see materials and methods) before addition of NaCl-containing defined medium in the apical compartment and either Na+-free medium or Cl−-free medium in the basal compartment, or the reverse. Preliminary tests carried out in untreated cultures after 2 h, 4 h, and 6 h of incubation in asymmetric solutions showed a progressive dissipation of initial ion gradients, with partial equilibration of Na+ and Cl− concentrations in the apical and basolateral medium being obtained at 6 h (data not shown). Accordingly, the 6-h time point was selected for subsequent experiments aimed at assessing the potential effect of MAPK inhibitors on transepithelial Na+ and Cl− gradients. As illustrated in Fig. 7, cultures of 2A4 cells preincubated with either PD169316 (2 μM) or SP600125 (10 μM) for 96 h to inhibit the p38 and JNK pathway, respectively, maintained significantly more robust transepithelial Na+ and Cl− gradients than control cultures.
The observed bidirectional effect of MAPK inhibitors on experimentally imposed transepithelial ion gradients strongly suggested a modulation of paracellular permeability. Nonetheless, to rule out the potential contribution of components of the transcellular pathway, such as vectorial Na+ transport driven by Na+-K+-ATPase, we assessed the effect of ouabain, an inhibitor of Na+-K+-ATPase. Since ouabain treatment could deleteriously affect cell viability when applied over the relatively long time lapse (6h) of the transepithelial ion gradient assay, we chose to assess its effect in an instantaneous measurement, the equivalent Isc. We found that addition of 1 mM ouabain to the lower compartment of 2A4 cells grown on semipermeable filters almost fully inhibited Isc and thereby transepithelial ion transport (control: 2.6 ± 0.1 to −2.0 ± 1.5; 2 μM PD169316: 1.1 ± 0.4 to −0.7 ± 0.6; 10 μM SP600125: 1.2 ± 0.2 to −0.2 ± 0.5; values expressed as μA/cm2 ± SE). However, ouabain did not abrogate the increase in TER induced by MAPK inhibitors (control: 0.03 ± 0.01 to 0.02 ± 0.00; 2 μM PD169316: 0.29 ± 0.03 to 0.24 ± 0.03; 10 μM SP600125: 0.37 ± 0.05 to 0.27 ± 0.04; values expressed as kΩ × cm2 ± SE). These results indicate that transcellular ion transport is likely to have a negligible impact on the observed modifications of transepithelial ion gradients in response to MAPK inhibitors.
Collectively, the experiments described above provide evidence that the modulation of claudin expression observed following inhibition of p38 or JNK correlates with decreased epithelial permeability toward Na+ and Cl−.
The role of intracellular signaling pathways in the regulation of TJ permeability and claudin expression has only recently begun to be elucidated. Studies published so far have focused largely on the MEK-ERK pathway. Specifically, it has been reported that growth factor-dependent activation of ERK1/2 causes an increase in TER associated with modulation of claudin expression in MDCK II cells (23, 42, 64), as well as in T84 intestinal epithelial cells (33). At variance with these findings, IL17- or HIV-1 Tat-induced activation of ERK1/2 has been shown to result in decreased TJ barrier function and altered claudin expression in intestinal (36) or retinal pigment (6) epithelial cells, respectively. Overall, available knowledge suggests that activation of the MEK-ERK cascade modulates epithelial permeability in a ligand-dependent and tissue-specific manner.
Although the involvement of p38 in the regulation of epithelial barrier function has been less thoroughly investigated, circumstantial evidence indicates that activation of this MAPK may increase TJ permeability. Thus, inhibition of p38 was reported to prevent the disruption of TJ barrier induced by various stimuli in different epithelial cell types (8, 43, 55–57). To our knowledge, a role for JNK in the control of TJ function has been previously documented in a single study (30) showing that inhibition of JNK signaling reduces claudin-2 expression in MDCK II cells.
Building on our initial findings that pharmacological inhibition of either p38 or JNK in 2A4 mammary epithelial cells induces dome formation (a phenomenon largely dependent on the barrier function of TJs), we assessed the effect of small-molecule kinase inhibitors on TER, a reliable indicator of the tightness of an epithelial sheet. Both PD169316 and SP600125, which act by blocking the ATP-binding site of p38 and JNK, respectively, induced a time- and dose-dependent increase of TER in 2A4 mammary epithelial cells. Similar results were obtained using kidney-derived MDCK cells. The efficacy and selectivity of PD169316 and SP600125 was confirmed both by assessing the phosphorylation of specific substrates of p38 or JNK, and by using additional, mechanistically unrelated p38 and JNK inhibitors. The finding that inhibition of either p38 or JNK enhances epithelial barrier function strongly suggests that the basal activity of these MAPKs has a tonic effect on TJ ionic permeability.
JNK and p38 were originally identified by their activation in response to environmental stress stimuli. However, subsequent studies demonstrated that these MAPKs are also activated by a number of cytokines or growth factors and are involved in a broad range of cellular responses such as cell proliferation, migration, differentiation, and morphogenesis (34, 50, 78). Notably, it has been reported that unstressed cells have measurable levels of basal JNK activity, which is critical for cell cycle progression (19, 39) and cell motility (35). In our system, low-level signaling by agonists present in serum or produced by 2A4 cells themselves is likely to be responsible for basal levels of activation of p38 and JNK pathways.
It is now well established that members of the claudin family of transmembrane proteins are critical regulators of TJ permeability; in particular, it has been shown that different claudins regulate paracellular permeability to specific ions (72) and that the overall ionic permeability of a TJ is dictated by the combination and ratios of different claudin isoforms (28). In an attempt to elucidate the molecular mechanisms by which p38 and JNK control paracellular ion permeability, we examined the effect of pharmacological inhibitors of these MAPKs on the expression of selected claudins. We found that the p38 inhibitor PD169316 increases the expression of claudin-4 and -8, and that the JNK inhibitor SP600125 increases the expression of claudin-4 and -9 while decreasing claudin-8. Importantly, a similar modulation of claudin expression by PD169316 and SP600125 was observed at the mRNA level. SP600125 also slightly increased mRNA, but not protein levels, of claudin-7. Finally, neither PD169316 nor SP600125 significantly affected the expression of claudin-1. Collectively, our results are in accordance with the recent report that inhibition of p38 attenuates aspirin-mediated decrease in claudin-7 and disruption of TJ barrier function in gastric epithelial cells (55). However, they contrast with the finding that p38 inhibition suppresses claudin-4 expression in salivary gland epithelial cells (26). This discrepancy may reflect tissue-specific differences in the regulation of claudin expression.
Notably, the finding that blocking either p38 or JNK signaling is sufficient to enhance epithelial barrier function implies that both pathways need to be active to maintain sustained levels of paracellular ionic permeability. This notion is also supported by the observation that simultaneous pharmacological inhibition of both p38 and JNK pathways causes only a slight additional increment of TER with respect to the increase induced by inhibition of a single pathway (F. Carrozzino and R. Montesano, unpublished data).
The data obtained by RNA interference further corroborated the conclusion that the observed modulation of claudin expression and TJ ionic permeability are due to selective inhibition of JNK or p38 activity. These experiments also disclosed a differential regulation of claudin expression by siRNAs specifically targeting individual MAPK isoforms. Thus, silencing of p38α but not p38β resulted in upregulation of claudin-8 mRNA, whereas silencing of either p38α or p38β increased claudin-4 transcription. Likewise, silencing of JNK2, but not JNK1, upregulated claudin-4 and -7 mRNAs, whereas silencing of either JNK1 or JNK2 markedly induced claudin-9 transcription, while decreasing claudin-8 gene expression. These results suggest that different isoforms of the same MAPK have specific roles in the regulation of TJ permeability.
What is the physiological relevance of MAPK-dependent changes in claudin expression? Claudin-4 has been shown to selectively decrease TJ permeability to Na+ ions. Upregulation of claudin-4 is therefore expected to be associated with increased TER and reduced paracellular diffusion of Na+, as we have indeed observed on pharmacological inhibition of p38 or JNK. Likewise, our finding that inhibition of p38 pathway in 2A4 cells upregulates claudin-8, increases TER, and reduces paracellular diffusion of Na+ is consistent with previous studies demonstrating that expression of claudin-8 decreases TJ permeability to both monovalent and divalent cations (80). On the other hand, we have also shown that JNK inhibition markedly decreases claudin-8 expression, a result that appears at first sight difficult to reconcile with the concomitantly observed increase in TJ barrier function. However, it is important to underscore that the present analysis does not encompass the whole spectrum of known claudin isoforms, and that interpretation of our results must take into account the potential role of additional claudins. In this respect, increasing evidence indicates that TJ permeability to cations is decreased by the expression of several claudins, including claudin-4 (71), -5 (76), -8 (80), -11 (73), -14 (9), and -19 (4). In contrast, paracellular cation diffusion is greatly facilitated by pore-forming claudins, such as claudin-2 (2, 27), -7 (1), -10b (74), -12 (27), -15 (73), and -16 (32). Because claudin-4 and -8 are likely to represent only a fraction of all cation-specific claudin isoforms affected by inhibition of JNK pathway, it is conceivable that downregulation of claudin-8 is compensated by increased expression of cation-restrictive and/or decreased expression of cation-permissive isoforms. An additional, mutually nonexclusive possibility is that the impact of claudin-4 on TJ cation-selective barrier largely overrides that of claudin-8, so that the net outcome of JNK pathway inhibition is a decreased paracellular cation permeability.
With respect to paracellular anion permeability, we found that inhibition of either p38 or JNK pathways decreases TJ permeability to Cl−. Among the claudins we found to be expressed in 2A4 cells, claudin-7 and -9 have been reported to decrease the paracellular flow of Cl− (1, 62). While the expression of claudin-7 was not altered by inhibition of p38 pathway and only slightly upregulated by inhibition of JNK pathway, the expression of claudin-9 was robustly increased on inhibition of JNK pathway. These findings suggest that blocking the JNK pathway decreases paracellular Cl− permeability, at least in part, through upregulation of claudin-9. On the other hand, it is conceivable that inhibition of the p38 pathway decreases TJ permeability to Cl− by modulating the expression of anion-specific claudins, such as claudin-10a (74), that have not been examined in our study, or other claudins whose ionic permeability properties have not yet been characterized.
Recent studies have shown that claudin phosphorylation by a number of protein kinases (5, 40, 79), including MAP kinases (25), modulates TJ barrier function (3, 72). Whether phosphorylation of specific claudins contributes to the modulation of TJ barrier function in our system remains to be established. It would also be important in future studies to identify MAPK-dependent transcription factors involved in differential regulation of claudin expression, as well as additional signaling pathways responsible for the fine tuning of TJ ionic permeability.
In conclusion, we have shown that inhibition of p38 or JNK pathways in mammary gland epithelial cells enhances TJ barrier function by selectively modulating the expression of specific claudin isoforms. These results imply that the basal activity of p38 or JNK exerts a tonic effect on TJ ionic permeability, and support the notion that inhibitors of JNK or p38 (14, 44) may afford therapeutic benefit in diseases associated with impaired epithelial barrier function (75, 81).
This study was supported by Grants 3100A0-113832/1 and 310000-121938/1 from the Swiss National Science Foundation (to R. Montesano and E. Féraille, respectively).
We thank P. Couleru, J. Rial-Robert, and M. Eissler for excellent technical assistance, A. Britan for support with claudin primers, and N. Dupont for secretarial work. We are grateful to N. Mensi, O. Golaz, I. Garland, and V. Lê (Laboratory of Clinical Chemistry, Geneva University Hospital) for assistance with sample analysis. We also thank D. Chollet (Genomics Platform, NCCR “Frontiers in Genetics,” University of Geneva) for granting access to real-time RT-PCR equipment and help with data analysis.
- Copyright © 2009 the American Physiological Society