|Home | About | Journals | Submit | Contact Us | Français|
Protein tyrosine phosphatase N2 (PTPN2) has been identified as a Crohn’s disease (CD) candidate gene. However, a role for PTPN2 in the pathogenesis of CD has not been identified. Increased permeability of the intestinal epithelium is believed to contribute prominently to CD. The aim of this study was to determine a possible role for PTPN2 in CD pathogenesis.
Intestinal epithelial cell (IEC) lines, T84 and HT29cl.19a, were used in all studies. Protein analysis was performed by Western blotting and protein knock-down was induced by siRNA. Primary samples were from control and CD patients.
Here, we demonstrate increased PTPN2 expression in CD intestinal biopsies and that the pro-inflammatory cytokine, IFNγ, increases PTPN2 expression and activity in IEC. Moreover, IFNγ-induced STAT1 and STAT3 phosphorylation in IEC is enhanced by PTPN2 knock-down. The cellular energy sensor, AMPK, partially regulates the IFNγ-induced effects on PTPN2. Additionally, PTPN2 knock-down potentiates IFNγ-induced increases in epithelial permeability, accompanied by elevated expression of the pore-forming protein, claudin-2.
PTPN2 is activated by IFNγ and limits IFNγ-induced signalling and consequent barrier defects. These data suggest a functional role for PTPN2 in maintaining the intestinal epithelial barrier and in the pathophysiology of CD.
The family of protein tyrosine phosphatases (PTPs) plays a critical role in regulating fundamental cellular signalling events, including cell proliferation, differentiation and survival1. A single nucleotide polymorphism (SNP) in the gene locus encoding one such phosphatase, protein tyrosine phosphatase N2 (PTPN2), has been associated with Crohn’s Disease (CD), Type I diabetes, and ulcerative colitis (UC)2-6. However, a functional role for PTPN2 in CD pathogenesis remains unidentified. PTPN2, also referred to as T-cell protein tyrosine phosphatase (TC-PTP), exists in two different splice forms, a nuclear 45kD and a cytoplasmic 48kD variant bound to the endoplasmic reticulum. Upon activation, the nuclear variant, which seems more important for PTPN2 catalytic activity7-11, can exit the nucleus12. In fibroblasts, this is dependent upon activation of the cellular energy sensor, adenosine monophosphate-activated protein kinase (AMPK)8. Among the PTPN2 substrates are receptors for epidermal growth factor9, 10, and insulin13 and the Signal transducers and activators of transcription (STAT) 1+37, 11, 14.
Inflammatory bowel disease (IBD) is comprised of CD and UC. Genetic, immunological and bacterial factors are involved in IBD pathogenesis. Evidence suggests that an epithelial barrier defect, coupled with a dysfunctional immune response to commensal flora, drives the development of chronic intestinal inflammation15. Both IBD subgroups are characterized by their cytokine profiles, with elevated levels of interferon gamma (IFNγ) predominating in CD16. The primary mediators of IFNγ-induced signalling are the STATs17, with upregulation of both STAT1+3 observed in IBD18,19. In addition to a possible role in the onset of CD, increased epithelial barrier permeability may also contribute to the major clinical symptom of CD, diarrhea, due to “leak-flux” of electrolytes and water into the lumen. In cell culture studies, IFNγ reduces intestinal epithelial barrier function. Recent publications demonstrated upregulation of the pore forming claudin-2 in colonic crypt epithelial cells in CD20, 21. Moreover, increased claudin-2 levels correlate with increased permeability20, 22-24.
We found that PTPN2 is upregulated both in IFNγ treated IECs and in CD patient biopsies, and this effect in IECs is partially AMPK-dependent. Moreover, PTPN2 knock-down facilitates elevated claudin-2 expression, correlating with a further increase in IFNγ-induced epithelial permeability. These data establish PTPN2 as a protective factor in a cell model of intestinal inflammation and implicate PTPN2 dysfunction in CD pathogenesis.
Human recombinant IFNγ (Roche, Mannheim, Germany), Compound C (CC), AICA-Riboside- 5’-phosphate (AICAR), and monoclonal mouse anti-PTPN2 antibody CF-4, which detects the 45kD and the 48kD isoforms (Calbiochem San Diego, CA), rabbit anti-lamin A/C (Santa Cruz Biotechnology, Santa Cruz, CA), mouse anti-claudin-2, anti-claudin-4, anti-ZO-1 and rabbit anti-occludin antibodies (Zymed Laboratories, Carlsbad, CA) were obtained from the sources noted. Rabbit anti-phospho-AMPKα (Thr172), anti-AMPKα, anti-phospho-STAT1 (Tyr701), anti-STAT1, anti-phospho-STAT3 (Tyr705) and anti-STAT3 antibodies were obtained from Cell Signaling Technologies (Danvers, MA). All other reagents were of analytical grade and acquired commercially.
Human colonic crypt T84 IEC were cultured as described previously25. Human HT29cl.19a IEC were cultured in McCoy’s 5A Medium (JRH, Lenexa, Kansas) supplemented with 10% fetal calf serum. IFNγ was added basolaterally. The AMPK inhibitor, CC (50 μM), and the AMPK activator, AICAR (1 mM), were added bilaterally.
Tissue specimens were prospectively collected from the terminal ileum, colon or rectum of male and female individuals with active CD (n=9), CD in remission (n=7), or from control subjects (n=9). Experimental details are in the Supplementary Methods while patient characteristics are presented in Supplementary Table 1. Written informed consent was obtained before specimen collection and studies were approved by the Cantonal Ethics Committee of the Canton of Zurich.
Total RNA was isolated using RNeasy Plus Mini Kit (Qiagen, Valencia, CA) and RNA expression was analyzed by real-time PCR. Experimental procedures, real-time PCR conditions and primer sequences (Integrated DNA Technologies, Coralville, IA) are described in the Supplementary Methods.
T84 and HT29cl.19a lysates were prepared as described previously25. The protein content was adjusted to ensure the same amount of total protein in each sample.
Separate cytoplasmic and nuclear lysates were obtained using a Nuclear Extraction Kit (Imgenex, San Diego, CA). Briefly, T84 cells were scraped from filters and cytoplasmic proteins were collected using a 1x hypotonic buffer and incubated with 10% detergent solution. The nuclear fraction was collected using a Nuclear Lysis Buffer. Protein content was determined and adjusted as described above.
Western analysis of cell lysates was performed as previously25.
5×105 T84 cells were seeded onto 12-mm Millicell-HA filters for 6 days before stimulation. Experimental conditions and staining protocols are described in detail in the Supplementary Methods.
Phosphatase activity was assessed using the EnzChek Phosphatase Assay Kit (Molecular Probes) according to manufacturer’s instructions and as described in Mattila et al.10. Further details are found in the Supplementary Methods.
T84 (2×106) or HT29cl.19a (1×106) cells were seeded 3 days before transfection. For PTPN2 and AMPKα1 (PRKAA1) genes, 100 pmol of three different annealed Silencer Pre-designed siRNA oligonucleotides (Applied Biosystems; Supplementary Methods) were transfected into T84 or HT29cl.19a cells using the Amaxa nucleofector system (Amaxa, Gaithersburg, MD). After transfection, IEC were cultured on filter membranes for 48 h before treatment. Control siRNA SMARTpool (Upstate Biotechnology/Dharmacon, Chicago, IL) (100 pmol/transfection) was used as negative control.
Transepithelial permeability was measured as 10 kD fluorescein isothiocyanate-dextran (FITC-Dextran) (Sigma St. Louis, MO) flux across IEC monolayers. Following IFNγ treatment, cells were washed (x3) and incubated for 30 min at 37°C with Ringer’s to equilibrate. FITC-Dextran (1 mg/ml) was added to the apical compartment of the monolayer. 1 h later, 100 μl of the basolateral solution was removed and fluorescence was detected using a microplate reader (Molecular Devices).
TER across T84 monolayers was assessed by voltohmeter (WPI, Sarasota, FL) and companion electrodes (Millipore, Bedford, MA). Measurements were calculated in Ω · cm2 and expressed as % control.
Data are presented as means +/- S.E.M. for a series of n experiments. Data are expressed as a percentage of the respective controls, arbitrary units (Figure 1d, Supplementary Figure 3) or fluorescence units (Figure 2A). Statistical analysis was performed by Analysis of Variance (ANOVA) and Student-Newman-Keuls post-test or Student’s t-test, where appropriate, using Graph Pad Instat 3 software (Graph Pad Software, La Jolla, CA). P values < 0.05 were considered significant.
To investigate if IFNγ affects epithelial PTPN2 expression, T84 cells were treated with IFNγ (1000 U/ml) for 24, 48 and 72 h. IFNγ significantly increased PTPN2 mRNA expression (Figure 1a). Western blotting revealed consistently increased levels of cytoplasmic PTPN2 protein (Figure 1b). However, in the nuclear compartment, a significant increase was only observed in 24 h IFNγ-treated cells, with diminishing effects thereafter (Figure 1c). Dose-response studies revealed that 1000 U/ml IFNγ caused a maximal increase in PTPN2 protein by 24 h IFNγ and PTPN2 was not further increased by higher concentrations (Supplementary Figure 1). Previously26, cultured biopsies from patients with active CD have been shown to release IFNγ at levels as high as approximately 1000 U/ml/mg tissue, although the mean level was 75 +/- 215 U/ml/mg tissue. Serum levels of IFNγ in these studies were much lower due to dilution of the high local level of IFNγ originating in the intestinal mucosa. Although much of our data was generated using 1000 U/ml of IFNγ, we do show in Supplementary Figure 1 approximately similar effects of 100 vs. 1000 U/ml of IFNγ on PTPN2 expression in IEC. Therefore, while 1000 U/ml is at the high end of the pathologic range, the experimental use of this concentration is not unreasonable. IFNγ also elevated PTPN2 protein in HT29cl.19a IEC (Supplementary Figure 2a).
To demonstrate the relevance of our studies to human disease, we investigated PTPN2 mRNA levels in human biopsies from macroscopically non-inflamed areas of the terminal ileum, colon or rectum of CD patients in clinical and macroscopic remission, macroscopically inflamed areas of the terminal ileum and colon of CD patients with clinically and macroscopically active disease or healthy control subjects undergoing routine colon cancer screening. We found that PTPN2 mRNA was significantly elevated in active CD patients (p<0.05), but not in CD patients in remission (Figure 1d). Correlating with increased PTPN2 mRNA levels, CD patients with active disease also showed an approximately three-fold increase in IFNγ mRNA levels compared to healthy controls (Supplementary Figure 3). However, the increase did not reach statistical significance due to the marked variability in IFNγ levels among individual samples. In contrast, IFNγ mRNA levels in CD patients in remission were not different from control subjects (Supplementary Figure 3). Microscopic assessment of H&E-stained biopsy specimens did not show signs of inflammation in either control subjects or patients with CD in remission, while signs of acute inflammation were visible in patients with active CD (Supplementary Figure 4). Overall, our findings of apparently altered PTPN2 expression in active CD are of special interest, since they demonstrate the importance of our study to “bona fide” human disease.
We investigated whether IFNγ alters PTPN2 enzymatic activity in T84 cells and found that IFNγ increases PTPN2 activity maximally within 72 h treatment in PTPN2 immunoprecipitates from whole cell lysates (Figure 2a). Next, we studied the activation status of the PTPN2 target proteins, STAT1 and STAT3 by measuring phosphorylation on tyrosine701 and tyrosine705, respectively. Our hypothesis was that increased PTPN2 activity should correlate with decreased phosphorylation of STAT1+3. Western blot analysis revealed that IFNγ treatment for 24 h not only enhanced cytoplasmic and nuclear STAT1 phosphorylation (Figure 2b,c), but also increased STAT1 expression (Figure 2b,c). In keeping with our hypothesis, however, STAT1 phosphorylation subsequently declined in both compartments in parallel with increased PTPN2 activity, albeit with different kinetics. Nuclear STAT1 phosphorylation was no longer significantly increased at 48 h IFNγ treatment and declined to control levels by 72 h. This suggests that PTPN2 activity was capable of significantly dephosphorylating STAT1 in the nuclear compartment as early as 24 to 48 h after IFNγ treatment (p<0.001). In contrast, cytoplasmic STAT1 phosphorylation, although reduced, was still significantly increased even at 48 h treatment (p<0.05). Since prior observations in fibroblasts point to the 48kD variant of PTPN2 being less active than the nuclear variant, we speculate that this difference in the catalytic activity between the PTPN2 isoforms might also be present in T84 IEC. Therefore, a possible explanation for the observation that cytosolic STAT1 phosphorylation already declines after 48 h IFNγ treatment might be that a small, but sufficient, amount of the nuclear PTPN2 variant is already in the cytosolic compartment and/or has already exited the nucleus. Additionally, since STAT1 is a transcription factor, it translocates after its activation from the cytoplasm into the nucleus, where it becomes dephosphorylated by the nuclear PTPN2 isoform, and is, subsequently, not detectable in the cytosolic protein fraction anymore. Similar data were obtained regarding IFNγ-induced cytoplasmic STAT3 phosphorylation (Figure 2d).
The delay in increased activity of PTPN2 is at odds with increased PTPN2 expression at 24 h. A possible explanation for this dichotomy is that whole cell lysates used for the PTPN2 activity assay contained only a small portion of the nuclear protein fraction (Supplementary Figure 5).
It is possible that only low levels of the nuclear 45kD isoform of PTPN2 were present in the whole cell lysates at the earlier timepoints of 24 and 48 h, whereas by 72 h, significant levels of the 45kD variant had exited the nucleus. Therefore, we speculate that in response to IFNγ treatment, a certain amount of the nuclear 45kD PTPN2 variant shifts from the nucleus into the cytoplasm. Moreover, cytoplasmic accumulation of PTPN2 in T84 cells in response to 72 h IFNγ treatment may contribute to STAT1 dephosphorylation in the cytoplasmic cell compartment, as well as to the significantly increased PTPN2 activity (Figures 2a-c).
Though our data suggest that IFNγ increases PTPN2 activity and this correlates with downregulation of IFNγ-induced STAT signalling in IEC, basal PTPN2 is also likely to be involved in STAT dephosphorylation. Further studies will be needed to investigate the effects of IFNγ on the catalytic activity of the different isoforms of PTPN2 in detail.
We next investigated whether PTPN2 knock-down conversely elevated STAT1+3 activity. T84 cells transfected with either PTPN2-specific siRNA or non-specific control siRNA sequences, were treated with IFNγ (24 h). PTPN2 siRNA decreased PTPN2 expression with a maximal reduction of 92 ± 3% (Figure 3a). Non-specific effects on protein levels were not observed, as shown by equivalent levels of the nuclear envelope protein lamin A/C and total STAT1+3 in both subgroups (Figure 3a-c). As expected, IFNγ increased PTPN2 protein (Figure 3a) and STAT1+3 phosphorylation (Figure 3b,c). PTPN2 knock-down amplified STAT1+3 phosphorylation (Figure 3b,c). Similar findings were observed in separately collected cytoplasmic and nuclear protein fractions of 24 h IFNγ-treated T84 monolayers (Figure 3d). siRNA-induced knock-down of PTPN2 in HT29cl.19a cells revealed the same regulatory influence of PTPN2 on STAT1 phosphorylation (Supplementary Figure 2a,b). These data, in addition with those in Figure 2a-d, indicate that PTPN2 is the cellular phosphatase of STAT1+3 in IEC, and that PTPN2 terminates IFNγ-induced STAT signalling.
The cellular energy sensor, AMPK, has been shown in fibroblasts to participate in regulation of PTPN2 nuclear-cytoplasmic shuttling. Therefore, we investigated if AMPK modified the effects of IFNγ on PTPN2 expression, activity and subcellular distribution in IECs. AMPK activity was assessed by phosphorylation of Thr172 on the AMPK catalytic α1-subunit in T84 cells. Western blot analysis revealed that IFNγ induced AMPK activation in a time-dependent manner, reaching a peak at 6 h (Figure 4a). This effect was sensitive to the AMPK inhibitor, compound C (CC; 50 μM), which competes for binding of intracellular ATP to AMPK, and inhibits AMPK phosphorylation by upstream kinases. CC co-treatment significantly inhibited IFNγ-induced AMPK phosphorylation (Figure 4b). To determine if AMPK was involved in the regulation of PTPN2 protein, we stimulated T84 monolayers with IFNγ and/or CC for 72 h. IFNγ increased cytoplasmic and nuclear PTPN2 levels, an effect that was partially diminished by CC. Thus, AMPK mediates, at least in part, IFNγ-induced PTPN2 expression although additional pathways may also be involved (Figure 4c,d).
Since we have shown that STAT1 phosphorylation kinetics can be used as a marker of PTPN2 activity, we surmised that STAT1 phosphorylation in cells treated with IFNγ and CC should be higher than in cells treated with IFNγ alone. Indeed, while IFNγ treatment (72 h) increased cytoplasmic and nuclear STAT1 phosphorylation (Figure 4e,f), CC co-treatment potentiated cytoplasmic (Figure 4e), and nuclear (Figure 4f) STAT1 activity above IFNγ alone. Interestingly, CC alone had no effect on STAT1 phosphorylation. In contrast, AMPK activity alone, as induced by the pharmacological activator AICAR (1 mM), neither induced PTPN2 expression nor altered STAT1 phosphorylation (Supplementary Figure 6). These data indicate that AMPK modification of PTPN2 protein levels and STAT1 dephosphorylation is dependent not only on IFNγ-stimulated AMPK activity, but also on other aspects of the signalling milieu existing in IFNγ- vs. AICAR-treated cells. Therefore, further studies will be required to determine additional pathways that are likely involved in the regulation of IFNγ-induced effects on PTPN2 besides AMPK.
Maximal IFNγ-induced PTPN2 activity was detectable by 72 h. However, the peak of IFNγ-induced AMPK phosphorylation was seen by 6 h. We hypothesized that the 45kD PTPN2 isoform, which likely accounts for the majority of PTPN2 activity, shifts from the nucleus to the cytoplasm in response to IFNγ. Therefore, we examined if IFNγ affects the cellular distribution of PTPN2. In untreated T84 cells, PTPN2 was evident in the cytoplasm and the nucleus (Figure 5a). A similar pattern of PTPN2 distribution was observed in 24 h IFNγ-treated cells (Supplementary Figure 7). This finding supports our Western blot data (Figure 1c) and our hypothesis that nuclear PTPN2 protein is highest from 0-24 h after IFNγ treatment. Therefore, our evidence suggests that 24 h treatment with IFNγ increases PTPN2 protein in both cell compartments independent of alterations in the relative subcellular distribution of PTPN2.
In contrast, after 72 h IFNγ, PTPN2 was more prominent in the cytoplasm than the nucleus compared to untreated control cells (Figure 5a). These findings further support our hypothesis, as well as our previous PTPN2 data (Figures 1b,,2a),2a), suggesting that the majority of nuclear PTPN2 shifts from the nucleus to the cytoplasm in response to 72 h IFNγ and accounts for the significant increase in PTPN2 activity at this timepoint. We further investigated whether AMPK inhibition by CC modulates PTPN2 distribution. In T84 monolayers treated with CC alone, PTPN2 was equally distributed in cytoplasmic and nuclear compartments as seen in untreated control cells. However, CC co-treatment clearly attenuated the IFNγ-induced cytoplasmic accumulation of PTPN2. Though PTPN2 was still prominent in the cytoplasm of co-treated cells, PTPN2 was more visible in the nucleus in comparison to IFNγ treatment alone (Figure 5a). Though these data indicate that the extra-nuclear localization of PTPN2 in response to IFNγ is, at least partially AMPK-dependent, additional pathways are likely involved.
To confirm the regulatory effect of AMPK on PTPN2 we transfected T84 cells with either control or AMPK-α1 targeting siRNAs. In cells transfected with AMPK siRNA, AMPK was significantly lowered by 64 ± 11% (Figure 5b), while lamin A/C and total STAT1 were not affected (Figure 5b-d). IFNγ treatment (24 h) increased PTPN2 expression and STAT1 phosphorylation in control siRNA-transfected cells (Figure 5c,d). However, AMPK-deficient cells showed, in parallel with decreased PTPN2 expression, a potentiation of IFNγ-induced STAT1 phosphorylation (Figure 5c,d). These data confirm that AMPK activity plays a partial but important role in regulating PTPN2 expression and activity in response to IFNγ in IEC.
We next determined whether PTPN2 mediates the IFNγ-induced decrease in barrier function. We assessed epithelial permeability by measuring flux of 10kD FITC-Dextran across T84 and HT29cl.19a monolayers, transfected with either control or PTPN2-specific siRNA and subsequently treated with IFNγ (72 h). PTPN2 siRNA significantly decreased PTPN2 expression (Figure 6a; Supplementary Figure 2a). As expected, IFNγ significantly increased permeability in cells transfected with non-specific siRNA. However, PTPN2 knock-down further exacerbated barrier dysfunction following IFNγ treatment in both IEC lines (Figure 6a, Supplementary Figure 2c). Additionally, loss of PTPN2 further aggravated the IFNγ-induced decrease in monolayer TER (Figure 6b). These data suggest that PTPN2 plays an important role in stabilizing epithelial barrier function during exposure to inflammatory cytokines.
The pore-forming protein, claudin-2, plays an important role in regulating epithelial permeability in IBD20, 21, 23. In T84 cells transfected with control siRNA, IFNγ treatment did not affect claudin-2 expression. However, in parallel with the rise in epithelial permeability, IFNγ significantly increased claudin-2 expression in PTPN2-deficient cells (Figure 6c). In contrast, while IFNγ reduced expression of ZO-1 and occludin , this was not further affected by PTPN2 knock-down. Moreover, claudin-4 levels remained stable under all conditions (Figure 6d). Overall, our data indicate that a lack of PTPN2 enables IFNγ-stimulated expression of claudin-2, which is believed to contribute to increased permeability in IBD. These data thus illustrate a novel role for PTPN2 in protecting epithelial barrier function.
PTPN2 plays a critical role in regulating cytokine signalling in immune cells by inactivating STAT1+37, 11, 14. These findings are supported by observations in PTPN2-/- mice that die within 3-5 weeks from systemic inflammation accompanied by elevated serum levels of IFNγ and TNFα27. We demonstrated that IFNγ increases PTPN2 expression and activity in IEC. Activated PTPN2, in turn, reduces IFNγ-induced STAT1+3 activation. These data were corroborated by PTPN2 knock-down studies. We can speculate that PTPN2 participates in a negative feedback mechanism activated by IFNγ and thus limits cytokine signalling. STAT1+3 affect different target genes and induce different biological effects. STAT1 plays a decisive role in mediating IFNγ-dependent immunomodulatory, antiproliferative and antiviral effects17. In contrast, STAT3 has primarily oncogenic and anti-apoptotic effects28, and plays an important role in colon cancer pathogenesis, which has a higher frequency in IBD patients29. Overall, strict regulation of STAT1+3 signalling is crucial for intestinal homeostasis. Our data establish PTPN2 as a key regulator of STAT1+3 phosphorylation and IFNγ-induced signalling in IEC in particular.
We observed that PTPN2 is regulated in T84 cells, at least in part, via IFNγ-induced AMPK activation. In response to reduced cellular energy levels, AMPK limits energy-consuming and activates energy-generating processes30. Energy consuming processes in the intestine include ion transport and epithelial barrier maintenance, both of which are compromised by IFNγ25, 31, 32, and in IBD33, 34. AMPK is thus a logical mediator of IFNγ signalling. However, AMPK activity alone was unable to affect PTPN2 expression. Therefore, accessory signals activated by IFNγ, in addition to AMPK, may also play an important role in the regulation of PTPN2. Identification of these signals was beyond the scope of this study. Nevertheless, our data indicate that AMPK is necessary, though not sufficient, to mediate important downstream effects of IFNγ relevant to intestinal inflammation.
The ability of IFNγ to increase epithelial permeability and decrease TER was significantly enhanced by PTPN2 knock-down and accompanied by increased claudin-2 expression. The claudins are a family of tight junction proteins that play a crucial role in regulating paracellular permeability22,35. Increased epithelial permeability is believed to contribute substantially to the pathophysiology of CD36. Although the claudins function to seal the tight junction, claudin-2 is unusual as it forms a cation-selective membrane pore for substrates smaller than 4 Angstrom and may therefore directly account for the further decrease in TER across IFNγ-treated PTPN2-deficient T84 monolayers. However, it is not possible that the increased FITC-dextran flux observed in our studies occurs via passage through this specific pore due to the restriction against passage of molecules this size (10 kD). Moreover, increased numbers of claudin-2 molecules do not directly correlate with increased permeability for substrates that exceed its pore size23. However in CD, claudin-2 is upregulated in intestinal crypt epithelium20, while the number of tight junction strand breaks in the colonic epithelium is increased. These breaks, of ≥25 nm, mean that even large macromolecules, such as the 10kD FITC-dextran used here, can pass through20. Moreover, increased claudin-2 correlates with the appearance of discontinuous tight junction strands37. It is plausible that the increased FITC-dextran flux observed reflects such discontinuities. This scenario is compatible with findings showing an essential role for claudin-2 in regulating the structural composition of tight junctions, and consequently in modifying epithelial permeability20-24, 38.
Claudin-2 expression in T84 cells is not increased by IFNγ alone21, 38. However, when PTPN2 expression is repressed, IFNγ increases claudin-2 expression and further elevates epithelial permeability. These findings imply that PTPN2 normally prevents a rise in claudin-2 expression by IFNγ. Moreover, the IFNγ-induced elevation of claudin-2 expression in PTPN2-deficient cells may be mediated by STAT proteins as we have identified a putative STAT1/STAT3 binding sequence in the claudin-2 promoter region39, 40, which is likely needed for STATs to have a regulatory effect on claudin-2 expression. Since, to the best of our knowledge, no STAT binding motif has been described within the claudin-2 promoter, we screened the claudin-2 promoter region39 for such a motif. By blast search, we identified such a palindromic STAT recognition motif, TTCNNNGAA (where N is a variable spacer), 260 to 252bp upstream in the claudin-2 promoter (Supplementary Figure 8). The identified sequence, 5’-TTCCCGGAA-3’ matches the criteria for a possible STAT1 binding motif40 and matches exactly with the STAT3 binding motif in the gene encoding human p21WAF40. Thus, we present a plausible candidate sequence for STAT-dependent regulation of claudin-2 expression. Though the nature of STAT regulation of claudin-2 expression, and the interaction with the putative STAT binding motif are unresolved, these findings support our hypothesis that PTPN2 knock-down leads to increased claudin-2 expression in response to IFNγ, possibly via upregulation of STAT1 and/or STAT3.
Whether PTPN2 restricts the interaction of STAT proteins with various promoters, perhaps by reducing their phosphorylation, remains to be elucidated. Nevertheless, our findings indicate that PTPN2 protects epithelial barrier function during exposure to inflammatory cytokines. Conversely, PTPN2 dysfunction could play an important role in CD, a chronic inflammatory disease featuring elevated levels of IFNγ. A mutation in the PTPN2 gene locus has been associated with CD3, 5, 6 and this mutation could possibly lead to a loss of functional PTPN2 in this disease (although this has not been formally demonstrated and, to the best of our knowledge, functional consequences of the specific PTPN2 SNP have not been reported) and to the downstream consequences described herein. Future studies should identify whether loss of functional PTPN2 permits passage of large molecules that might initiate or exacerbate episodes of intestinal inflammation, across the intestinal epithelium. We also observed increased PTPN2 expression in biopsies from CD patients with active inflammation vs. normal subjects, which is in line with our in vitro data showing that IFNγ increases PTPN2 expression. These data provide proof of principle of altered PTPN2 expression in CD. However, a more detailed study will be required to determine whether there are differing levels of activity of PTPN2 in active and inactive CD tissues. A similar analysis will need to be performed in tissues from patients harbouring the PTPN2 SNP.
In summary, our data show that loss of PTPN2 leads not only to a prolonged and more severe IFNγ signalling cascade, but critically to a defect in intestinal epithelial barrier function. Taken together, these findings suggest that PTPN2 activity is crucial for regulating inflammatory processes and preserving barrier function in the setting of inflammation. These data indicate a mechanistic role for PTPN2 in restricting the pathogenesis of CD, as well as other inflammatory conditions associated with increased intestinal permeability, and provide a functional basis for the identification of PTPN2 as a CD candidate gene.
This work was supported by a Crohn’s and Colitis Foundation of America Senior Research Award to DFM, a Jon I. Isenberg Award to DFM, grants from the German Research Foundation (DFG) to MS and GP, NIH grant DK28305 to KEB, an unrestricted research gift to KEB from the Shape-Up Settlement fund, and by the University of California, San Diego, Digestive Diseases Research Development Center (DK080506).
Author contributions: MS (acquisition of data; analysis and interpretation of data; drafting of the manuscript; statistical analysis; obtained funding); GP (acquisition of data; statistical analysis); AW (acquisition of data; analysis and interpretation of data); MH (technical and material support; acquisition of data); BJ, MJD, GR (material support); KEB (critical revision of the manuscript for important intellectual content; obtained funding); DFM (study concept and design; analysis and interpretation of data; drafting of the manuscript; critical revision of the manuscript for important intellectual content; statistical analysis; obtained funding; study supervision).
The authors declare that they have no competing financial interests.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.