Noncanonical STAT3 Activation Regulates Excess TGF-β1 and Collagen I Expression in Muscle of Stricturing Crohn’s Disease

The clinical course of 30–50% of patients with Crohn’s disease (CD) is complicated by the development of fibrosis (Montreal Class B2) and bowel obstruction (1). Genome-wide association studies (GWAS) have identified loci that confer susceptibility for the development of CD and ulcerative colitis or modify the course of disease (2–4). GWAS provides information on risk loci but has not yet provided robust insight into risk variants and how these result in CD or a particular CD phenotype (5–7). Pathway analysis, however, indicates that a number of gene variants are involved Th17/IL-23 signaling including IL-23R, JAK2, and STAT3 (8, 9). STAT3 is a critical signaling intermediate in the pathways activated by a number of cytokines that are inhibited by increased expression of suppressor of cytokine signaling (SOCS) proteins (8, 9). Activated mesenchymal cells produce IL-6, IL-10, IL-12, IL-13, and IL-17a and TNF-α.

Activation of mesenchymal cells following injury and inflammation results in the disordered wound healing that leads to fibrosis in susceptible patients. Activated mesenchymal cells, including smooth muscle and subepithelial myofibroblasts, in patients with fibrostenotic Montreal B2 CD produce high levels of the fibrogenic cytokine, TGF-β1 that induces increased extracellular matrix protein production, including collagen I, and cellular proliferation that results in architectural distortion and scarring (10–13). These changes are specific to strictured intestine compared with that in the normal resection margin in the same patient with fibrostenotic disease (12, 14). In contrast, patients with Montreal B1 and B3 phenotype do not have elevated levels of TGF-β1 or collagen I in the affected intestine compared with unaffected intestine (12). Smooth muscle of muscularis propria represents the largest compartment of activated mesenchymal cells in the setting of CD. However, the molecular mechanisms regulating excess TGF-β1 and collagen I and the development of fibrosis in Montreal B2 patients are not known.

Human mesenchymal cells are activated by TGF-β1. Once activated, autocrine TGF-β1 production increases and elicits excess extracellular matrix production including collagen I. Activation of intestinal mesenchymal cells also increases their autocrine production of a number cytokines including IL-6, IL-10, IL-12, IL-13, and IL-17a and TNF-α (15). Synthesis of IL-6 increases also in response to IL-1β and TNF-α (16, 17). Cytokines act through their specific receptors and/or through the common gp130 (gp130) receptor coupled to Jak phosphorylation. Jak phosphorylation elicits activation and dimerization of STATs. STAT3 has two potential phosphorylation sites, STAT3(Y705) and STAT3(S727) (18–21). Phosphorylation of these two different amino acid residues can yield similar, enhancing or even opposite cellular effects depending on cell types and organs. Our understanding of STAT3 activity is undergoing significant paradigm shifts (20). The traditional notion that “inactive” STAT3 existed as cytosolic monomers and is activated by dimerization following STAT3(Y705) phosphorylation has been superseded by the understanding that 1) nonphosphorylated STAT3 exists as dimers and can be active, 2) phosphorylation of either STAT3(Y705), STAT3(S727), or both may confer STAT3 transcriptional activity, 3) STAT3 activity can be exerted by nuclear transcriptional effects or sequestration in signaling endosomes, and 4) STAT3 function is epigenetically regulated by acetylation on STAT3(K140) and/or STAT3(K685) (22). The ability of activated STAT3 to regulate gene transcription depends on the presence of consensus STAT3 binding elements in the 5′-untranslated region promoter regions of target genes. Consensus STAT3 binding elements, TT(N4)AA and TT(N5)AA are present in the 5′ region of the TGFB1 gene (23).

In this paper we delineate the mechanism whereby increased autocrine cytokine production, IL-6, in strictured intestinal muscle of fibrostenotic Crohn’s disease results in abnormal STAT3 phosphorylation. Abnormal STAT3 phosphorylation regulates increased TGFB1, COL1A1, and connective tissue growth factor (CTGF) gene expression and increased cellular proliferation. This mechanism is unique to strictured intestinal muscle in patients with Montreal B2 CD. In normal muscle cells from susceptible patients, IL-6 phenocopied the STAT3-TGFB1 and COL1A1 pathobiology present in strictured intestine and was used to probe the regulation via STAT3. The clinical significance of these findings is that abnormal STAT3 activation is unique to the strictured intestinal muscle in patients with Montreal B2 CD and can be restored to normal levels by selective inhibition of STAT3(S727) activation. This approach could therefore diminish downstream TGF-β1 and TGF-β1–induced fibrosis and stricture formation in susceptible patients.

Intestine was obtained from patients undergoing ileal/ileal-colonic resection for CD (Table I). All patient specimens including in this analysis were from patients with Montreal Classification L1, B2, or L3, B2 determined by computerized tomography enterography, or magnetic resonance imaging enterography. Comparison is made to patients with strictly B1 and B3 disease and without overlap with coexistent fibrostenosis and penetrating disease. Phenotype was confirmed by pathology. Intestinal specimens from non-CD patients undergoing surgery were used as comparison. Smooth muscle cells (SMC) from muscularis propria were enzymatically isolated from the circular muscle layer of affected regions of ileum, the histologically normal proximal ileal resection margin in the same patient or from non-CD subject’s intestine (24).

Isolated SMC were used to prepare RNA and whole-cell lysates or placed into primary cell culture as reported and validated previously (24, 25). Nonmuscle cells including epithelial cells, endothelial cells, neurons, and leukocytes are not detected in cells isolated in this fashion. These cells possess a smooth muscle phenotype: immunostaining for SMC but not fibroblast markers, expression of γ-enteric actin, and contractile ability (24, 26). Each characteristic is retained by muscle cells in primary culture (24, 26).

Primary cultures of SMC were treated with IL-6 for various periods of time in the presence and absence of the selective STAT3 inhibitor Stattic (10 μM) or following transfection with wild-type or mutant STAT3 plasmids.

Human studies were approved by the Virginia Commonwealth University Institutional Review Board. All patients provided informed consent.

Cytokine levels were measured in cell lysates prepared from SMC isolated from patients with distinct CD phenotypes and compared with that from non-CD subjects. IL-6 levels were measured by ELISA (Qiagen, Valencia, CA). Results were expressed as picograms per microgram total cell protein.

Quantitative real-time PCR was used to measure RNA transcripts of TGF-β1, Collagen1α1, and CTGF. Primers were used for human TGFB1: Hs00998133_m1, COL1A1: Hs00164004_m1, and CTGF: Hs00170014_m1 (Applied Bio Systems, Foster City, CA). Results were calculated using the 2^−ΔΔCt method based on GAPDH (Hs03929097_g1) amplification, which remains stable across the regions and phenotypes examined (14, 27).

Cell lysates were prepared as described previously (28–30). The level of phospho-STAT3(Y705), phospho-STAT3(S727), total STAT3, SOCS2, SOCS3 (Cell Signaling Technologies, Danvers, MA), and collagen I (Santa Cruz Biotechnology, Santa Cruz, CA) in lysates was measured and normalized to β-actin (28, 31).

Chromatin immunoprecipitation (ChIP) assay were performed using nuclear extracts from freshly isolated SMC of non-CD subjects and from patients with stricturing Crohn’s disease. In other experiments, primary cultured muscle cells were was used after transfection with various STAT3 mutants and treatment of cells with IL-6 (10ng/ml) for 6 h prior to extraction of genomic DNA using TRIzol (Invitrogen, Carlsbad, CA) for quantitative analysis. ChIP-grade Abs against total STAT3, phospho-STAT3(S727), or phospho-STAT3(Y705) (Cell Signaling Technology, Boston, MA) were used in addition to control rabbit IgG. PCR was performed with primers specific for the promoter region of TGFB1 gene (Switchgear, Menlo Park, CA). Results were calculated from input, which is referred to PCR without immunoprecipitation: %Input(normalized) = %Input(STAT3) − %Input(control IgG).

Primary cultured cells were transfected with dominant-negative (dn)STAT3(S727A), constitutively active STAT3(S727E), dnSTAT3(Y705F), or wild-type STAT3 as control [a gift from Dr. Too, National University of Singapore, Singapore (32)]. Transfection was achieved using X-treme transfection reagent (Roche, Franklin, MA) in serum-free medium for 24 h.

Primary-cultured cells were transfected with dual-reporter constructs for promoter reporter clone for human TGFB1 (HPRM13178-PG04) using EndoFectin Lenti transfection reagent (EFP1003; GeneCopoeia, Rockville, MD). One microgram of DNA was used to transfect cultured cells. After 16 h, cells were treated with 10 ng/ml IL-6 for 6 h. Medium was collected for dual Gaussia luciferase and secreted alkaline phosphatase luminescence assays. Luminescence assays were performed in triplicate using a Wallac Victor2 1420 Multilabel counter (Perkin-Elmer Life Sciences, Waltham, MA).

Twelve-micrometer cryosections of human intestine were used as described previously (12). Human IL-6 polyclonal goat IgG (R&D Systems, Minneapolis, MN), Collagen I, α-smooth muscle actin (α-SMA; Sigma-Aldrich, St. Louis, MO), phosphorylated STAT3(S727), and phosphorylated STAT3(Y705) were examined by immunofluorescence using specific Abs (Cell Signal Technology, Beverly, MA) and Alexa Fluor 594– and 488–conjugated secondary Abs (Molecular Probes, Eugene, OR). Nuclei were counterstained with DAPI. Digital images were obtained using a Leica TCS-SP2 AOBS confocal laser scanning microscope.

SMC proliferation was measured by MTT assay in quiescent cultured SMC as described previously (24).

Colitis was induced in C57BL/6J mice as reported by the instillation of 2,4,6 trinitrobenzene sulfonic acid (TNBS) according to a protocol approved by the Institutional Animal Care and Use Committee (20, 33). Mice were given 6 mg TNBS in 100 μl 50%/EtOH or vehicle control intrarectally. The participation of STAT3 phosphorylation on TGF-β1 and collagen I expression, and the development of early fibrosis in colitis was examined using 3.75 mg/kg/d i.p. Stattic or vehicle control i.p. Mice were sacrificed after 7 d, and the colons were harvested for analysis of gross and histologic damage scores as reported previously (27). Briefly, the severity of inflammation was assessed by scoring of macroscopic damage and histological evaluation. Macroscopic damage scores measured the presence and severity of adhesions (score 0–2), the maximum thickness of the bowel wall (in micrometers), and the absence or presence of diarrhea (0–1). Histological scoring evaluated the extent of destruction of normal mucosal architecture (0–3), the presence and degree of cellular infiltration (0–3), the extent of muscle thickening (0–3), the presence or absence of crypt abscesses (0–1), and the presence or absence of goblet cell mucus (0–1). The average thickness of the muscularis propria was measured using image scanning micrometry at five villous bases per section by blinded reviewers and reported in micrometers.

Total collagen content in the muscularis propria of mouse colon was detected with Sirius red collagen detection kit (Chondrex, Redmond, WA). Muscle cells of mouse colon were homogenized in tissue protein extraction reagent buffer (Thermal Science, Amarillo, TX), incubated on ice for 15 min, and centrifuged for 5 min at 10,600 × g at 4°C. Each protein sample was diluted in 0.5 M acetic acid to a final concentration (100 μg/ml). OD was read at 530 nm. Results were calculated based on collagen per 100 μg/ml protein.

Myeloperoxidase (MPO) activity was quantified in the same regions of human intestine and mouse colon from which muscle cells were isolated (Invitrogen, Grand Island, NY) as reported previously (12). Briefly, equal protein amounts (100 μg/ml) of cell lysate of human and mouse colon were homogenized in 30 μl tissue protein extraction reagent buffer/mg lysate, incubated on ice for 15 min, and centrifuged for 5 min at 10,000 rpm at 4°C. Fluorescence was measured with a Wallac Victor2 1420 Multilabel counter (Perkin-Elmer Life Sciences) with excitation and emission at 485 and 530 nm, respectively, for the 3′-(P-aminophenyl)fluoroscein-based assay and excitation and emission at 530 and 590 nm, respectively, for the Amplex UltraRed assay. Results were expressed as fold change when normalized to the control sample.

Values represent means ± SEM of n experiments, where n represents the number of experiments on samples or cells derived from separate subjects. Statistical significance was tested by Student t test for either paired or unpaired data as appropriate.

The levels of IL-6, IL-10, IL-12, IL-13, IL-17a, and TNF-α were measured in muscle cells isolated from the affected regions and normal intestine in patients with B1, B2, and B3 phenotype CD, in active ulcerative colitis, and in normal intestine of non-Crohn’s subjects (Fig. 1, Tables I, II). IL-6 production by SMC normal subjects in vivo was low, 23.5 ± 1.18 pg/μg protein. IL-6 production in patients with Montreal B1 phenotype CD was increased but was similar between normal proximal resection margin and affected ileal segments, 20.4 ± 1.02 and 26.4 ± 1.32 pg/μg protein, respectively. IL-6 production in patients with Montreal B3 phenotype CD was similar and unchanged between normal resection margin and affected ileum and unchanged from non-CD subjects: 29.1 ± 1.46 and 36.1 ± 1.80 pg/μg protein, respectively. In contrast, IL-6 production in the normal resection margin of patients with Montreal B2 stricturing disease were similar to normal subjects 32.8 ± 1.64 pg/μg protein but was increased in the affected strictured ileum, 62.3 ± 3.11 pg/μg protein (Fig. 1A).

These cytokines can activate STAT3 signaling either through their specific cytokine receptors and/or the common gp130 receptor. STAT3 activation was explored further in these studies on patients with Montreal B2 fibrostenotic CD using IL-6 as a STAT3 activator because the levels of IL-6 in strictured intestinal muscle were higher compared with normal intestine in the same patient. This difference between the normal intestine and affected region in Crohn’s was unique to IL-6 compared with the other cytokines measured.

In strictured intestine, immunoreactive IL-6 was increased in α-SMA–positive muscle cells compared with the normal resection margin in Montreal B2 CD patient (Fig. 1B). A similar pattern was seen for immunoreactive collagen I in α-SMA–positive muscle cells.

The notion that fibrosis progresses even in the absence of continued inflammation was examined by direct measurement of mucosal inflammation. The relative levels of mucosal inflammation in patients with each phenotype of CD were assessed by MPO activity and compared that in the mucosa of non-CD subjects with normal intestine. A significant increase in MPO activity was seen only in patients with inflammatory B1 phenotype CD and not with fibrostenotic B2 or penetrating B3 CD (Fig. 1C).

STAT3 possesses two phosphorylation sites that are known relevant to function: STAT3(Y705) and STAT3(S727) (19). In muscle cells isolated from the normal resection margin in patients with Montreal B2, STAT3(Y705) phosphorylation was high and STAT3(S727) phosphorylation was low (Fig. 2A, 2B). The opposite was seen in strictured intestine where STAT3(Y705) phosphorylation decreased 65 ± 3% and STAT3(S727) phosphorylation increased 540 ± 110% (Fig. 2A, 2B). A similar pattern was seen when STAT3(Y705) and STAT3(S727) phosphorylation was examined using immunofluorescence (Fig. 2C). A differential phosphorylation pattern of phospho-STAT3(S727) and phospho-STAT3(Y705) emerged when muscle cells of normal proximal resection margin and affected ileum were compared between patients with Montreal B1 and B2 but not B3 phenotype disease (Fig. 2D).

In patients with Montreal B2 phenotype, the expected increase in SOCS3 in response to increased autocrine cytokine production and active cytokine signaling was lost. SOCS3 decreased 40 ± 5% in strictured ileum compared with normal resection margin (Fig. 2E). No change in the level of SOCS2 was seen (Fig. 2E).

Both phosphorylated and unphosphorylated STAT3 isoforms can translocate to the nucleus and act as transcription factors (20). We examined the TGFB1 gene DNA–binding activity of total STAT3, STAT3(S727), and STAT3(Y705) using ChIP assays in muscle cells from non-Crohn’s subjects, muscle cells of normal resection margin, and strictured intestine in the same patient.

DNA-binding activity of total STAT3 for the STAT3 binding elements in the TGFB1 promoter was low in normal subjects, 0.009% input (Fig. 3A). Compared with DNA binding in normal subjects, DNA-binding activity increased 2.9 ± 0.2-fold in normal resection margin in patient with B2 CD, and 16.2 ± 0.6-fold in strictured intestine (Fig. 3A). DNA-binding activity of phospho-STAT3(Y705) was low in the muscle of normal subjects, 0.025% input and was unchanged in the muscle of normal resection margin or strictured intestine (Fig. 3B). In contrast, DNA-binding activity of phospho-STAT3(S727) was low in non-CD subjects, 0.008% input, and increased 11 ± 0.1-fold in muscle cells of normal resection margin and 17 ± 1.2-fold in muscle cells of strictured intestine (Fig. 3C). These results indicate that TGFB1 gene DNA–binding activity of STAT3 resides predominately within phospho-STAT3(S727) and not phospho-STAT3(Y705).

Treatment of primary cultures of normal intestinal muscle cells of patients with Montreal B2 CD with IL-6 elicited dose-dependent phosphorylation of STAT3(S727) but not STAT3(Y705) (Fig. 4A). In the presence of 50 μM Stattic, a nonpeptide small molecule inhibitor of STAT3 Src homology 2 domain function, STAT3(S727) phosphorylation induced by 20 ng/ml IL-6 was inhibited 3.5 ± 0.18-fold, STAT3(Y705) phosphorylation increased 2.4 ± 0.12-fold, and SOCS3 levels increased 3.2 ± 0.16-fold (Fig. 4A) (34).

We next examined the significance of STAT3(S727) and STAT3(Y705) phosphorylation using immunofluorescence in cells transfected with wild-type STAT3, dnSTAT3(S727A) or STAT3(Y705F), and constitutively active STAT3(S727E) mutants. IL-6 (10 ng/ml) induced STAT3(S727) phosphorylation and nuclear translocation (Fig. 4B). Transfection of cells with dnSTAT3(S727A) mutant did not affect basal STAT3(S727) phosphorylation, but the STAT3(S727) phosphorylation and translocation in response to IL-6 was lost (Fig. 4B). Transfection of cells with constitutively active STAT3(S727E) resulted in STAT3(S727) phosphorylation and nuclear translocation without addition of IL-6 (Fig. 4B).

The mechanisms by which phosphorylated STAT3 regulates TGFB1 gene expression was examined using ChIP assay and with dual luciferase reporter assay. STAT3-DNA–binding activity was low in control wild-type transfected muscle cells, 0.02 ± 0.001% input, and was increased to 0.12 ± 0.006% input with 10 ng/ml IL-6 (Fig. 4C). In cells transfected with dnSTAT3(S727A), basal DNA–binding activity was low, 0.01 ± 0.001%input. DNA binding of STAT3(S727) to the TGFB1 promoter in response to IL-6 was lost in STAT3(S727A)-transfected cells (Fig. 4C). In cells transfected with constitutively active STAT3(S727E), basal DNA–binding activity was high even in the absence of exogenous IL-6. Interestingly, in cells transfected with dnSTAT3(Y705F), the DNA-binding activity of STAT3(S727) was low and decreased further in response to IL-6 (Fig. 4C).

Transcriptional activity of the TGFB1 gene was increased 2.21 ± 0.11-fold by IL-6 (10 μM) over untreated cells or control pEZX-PG04–transfected cells and was abolished in the presence of the selective STAT3 inhibitor Stattic (50 μM) (Fig. 4D). Transcriptional activity of TGFB1 gene was also examined in cells transfected with the STAT3 mutants. In STAT3 wild-type transfected cells, IL-6 increased TGFB1 transcriptional activity by 0.7 ± 0.04-fold over untreated cells (Fig. 4E). In cells transfected with dnSTAT3(S727A), IL-6–induced TGFB1 transcriptional activity was lost, whereas in cells transfected with constitutively active STAT3(S727E), basal TGFB1 gene transcriptional activity was increased 1 ± 0.05-fold over wild-type transfected cells without the need for added IL-6 and was similar to IL-6–stimulated transcriptional activity in wild-type transfected cells. In the cells transfected with the STAT3(Y705F) mutant, transcriptional activity was similar to STAT3 wild-type transfected untreated cells; the ability of IL-6 to increase TGFB1 transcriptional activity was also lost in STAT3(Y705F)-transfected cells.

Treatment of muscle cells with 10 ng/ml IL-6 also elicited a time-dependent increase in TGF-β1, COL1A1, and CTGF transcripts (Fig. 5A). The maximal IL-6–induced increase in TGF-β1, COL1A1, and CTGF transcripts was inhibited in the presence of the STAT3(S727) phosphorylation inhibitor, 50 μM Stattic (Fig. 5B). In cells transfected with WT STAT3, IL-6 increased COL1A1 (Fig. 5C) and CTGF transcripts (Fig. 5D) expression compared with untreated muscle cells. In cells transfected with dnSTAT3(S727A), basal COL1A1 and CTGF transcripts were similar to control cells, but the ability of IL-6 to increase transcript levels was lost (Fig. 5C, 5D). In contrast, expression of a dnSTAT3(Y705F) had no effect on IL-6–stimulated levels of COL1A1 but did inhibit IL-6–induced TGF-β1and CTGF expression (Figs. 4F, 5D). When cells were transfected with constitutively active STAT3(S727E) basal TGF-β1, COL1A1 and CTGF expression was increased even in the absence of IL-6 treatment (Figs. 4F, 5C, 5D).

Another feature of activated mesenchymal cells in the strictured intestine in patients with Montreal B2 CD is increased rates of cellular proliferation. The role of STAT-3 in this increased proliferation that contributes to the development of fibrosis was investigated in primary cultures muscle cells using the STAT3 inhibitor, Stattic. Proliferation in untreated quiescent cells from strictured intestine was increased 115 ± 11% over that in cells isolated from normal intestine in the same patient. Increased proliferation of muscle cell from strictured intestine was abolished in the presence of Stattic (Fig. 5E).

To determine the role of STAT3(S727) phosphorylation on development of fibrosis, the murine model of TNBS-induced colitis in C57BL/6 mice was used. Mice were treated with Stattic 3.75 mg/kg/d i.p. for 7 d after intrarectal instillation of TNBS. Macroscopic damage scores increased along with microscopic scores in mice following TNBS treatment (Fig. 6A). Treatment of mice with Stattic improved both macroscopic damage and microscopic damage particularly in the development of fibrosis and thickening of the bowel wall (Fig. 6B). Stattic treatment did not have a negative impact on mucosa and related microscopic scores (Fig. 6A).

Increased phosphorylation of STAT3(S727) seen in the TNBS-induced colitis murine model can be inhibited by the treatment of STAT3 inhibitor Stattic within 7 d, whereas the phospho-STAT3(Y705) protein level is restored by the treatment of Stattic (Fig. 6C). This was associated with resultant change of collagen production (Fig. 6D). Immunofluorescent staining of the colon tissue slices from these animals also confirmed differential expression patterns of pSTAT3(S727) and pSTAT3(Y705) seen on Western blot results in the muscle layer (Fig. 6E). We also examined the effect of Stattic on profibrotic genes expression in these animals. Transcript levels of TGF-β1, COL1A1, and CTGF were significantly decreased in the Stattic-treated group after TNBS-induced colitis compared with the control group (Fig. 6F). Interestingly, compared with the increased MPO activity in TNBS-colitis group, Stattic suppressed upregulation of intestinal inflammation induced by TNBS instillation (Fig. 6G).

The analysis of complex polygenic disorders like CD, in contrast to single-gene diseases, requires consideration of the functional interrelationship of multiple disease susceptibility genes and pathobiology. In this study, we have characterized a cellular phenotype that links molecular mechanism with identified risk variants that lie within a common candidate network, for example, JAK-STAT3-TGFB1 that uniquely characterizes patients with Montreal B2 CD and is not seen in patients with other CD phenotypes (12, 14). One strength of our study design is to compare affected intestine, including strictures, with normal resection margin in the same patient and to normal non-CD subjects. It is worth noting that the characteristic phenotype and molecular mechanisms we have identified are independent of the presence or absence of inflammation in the affected intestine and are independent of the medications used by any patient.

Patients with Montreal B2 CD have increased IL-6 production and a distinctive, “noncanonical” activation pattern of STAT3 in strictured intestinal muscle. The specific pattern of STAT3(S727) activation results in increased TGFB1-DNA binding affinity and transcription and increased production of TGF-β1 and collagen I. Both contribute to the pathobiology of fibrosis. It is worth noting that in each paired specimen of normal resection margin and strictured segment, the increase in STAT3(S727) phosphorylation and autocrine TGF-β1 production along with IL-6 production was consistent and occurred irrespective of medication exposure. These observations confirm the central role and delineate one key mechanism of TGF-β1–mediated fibrogenesis in Montreal B2 CD (12, 14, 33).

The consensus STAT3 binding motif present in the 5′-untranslated region of the TGFB1 gene is also present for the SOCS3 and COL1A1 genes. Although this was not investigated directly in our study, the concomitant inappropriate decrease in SOCS3 and loss of negative feedback on active STAT3 could further contribute to dysregulated signaling and excess TGF-β1–dependent collagen I production leading to fibrosis. Typically, SOCS3 is more often associated with IL-6 signaling than SOCS2 (35, 36). The increased production of collagen I in strictures would then reflect the joint effect of indirect TGF-β1–induced collagen I production and direct STAT3 transcriptional regulation of the COL1A1 promoter. This notion is supported by our data obtained from STAT3 mutant–transfected cells where both direct and indirect effects on TGFB1 and COL1A1 gene expression were revealed.

Once immune competent mesenchymal cells are activated, including muscle cells, autocrine IL-6 production increases. IL-6–dependent production of excess TGF-β1, collagen I, CTGF, and increased proliferation, like fibrosis itself, proceed independently of ongoing inflammation via the autocrine pathway that we have identified. This has important implications on efficacy of treatment of Montreal B2 CD patients given the limitations of our current pharmacologic armamentarium to immune suppressing and anti-inflammatory agents and how they alter the transcriptome in CD patients (37). Altered STAT3 signaling in mesenchymal cells is a common theme in diseases associated with organ fibrosis including idiopathic pulmonary fibrosis (38). In the liver Ogata et al. (39) also implicated a loss of SOCS3 that promotes fibrosis by enhancing STAT-3–mediated TGF-β1 production. It is also clear the role of STAT3 in the pathobiology of CD is different in epithelial or immune cells than in mesenchymal cells where loss of intestinal epithelial cell STAT3 leads to more severe chronic inflammation by promoting T cell STAT3 activation (40).

In CD the Th17-related candidate network involving JAK and STAT3 mutations that confers risk of developing CD can also be activated by the Th17 cytokine, IL-23, via IL-23R. The present study implicates autocrine IL-6 and activation of STAT3(S727) in the pathogenesis of intestinal fibrosis. The functional genomic mechanisms of these candidate network polymorphisms remain unknown at present. In the current study, it should be noted that STAT3 polymorphisms located in the disequilibrium block linked to susceptibility of CD, rs744166 in intron 2, rs8074524 in intron 3, rs2293152 in intron 11, and rs957970 in intron 23, do not encode STAT3(Y705) or STAT3(S727) amino acids. These identified polymorphisms may exert cis-eQTL effects instead. Interestingly, STAT3(S727), Rs115474585, STAT3(Y705), and rs139701269 polymorphisms have been identified in patients with hyper-IgE syndrome and are linked to defects in Th17 cell differentiation (41). By itself STAT3 gene represents a shared risk locus. The causative variants within STAT3 gene have not been clearly identified likely becaus of high linkage disequilibrium with another nearby causal polymorphism.

Studies that seek to understand how the complex genetics of CD confer pathobiological risk face several challenges. One is to understand the pathobiology of a complex genetic disorder in which multiple combinations of risk variants present in a patient can on aggregate lead to disease susceptibility, disease modification, or disease phenotype. This is evident from studies that demonstrated carriers of combined STAT3 and Jak variants, or STAT3 and Tyk2 variants, have a higher risk of CD compared with individuals carrying only major alleles of each variant (42, 43). A second challenge is that any specific genetic variant may have different pathobiologic consequences depending on cell type. This is evident for STAT3. In lamina propria mononuclear cells, both total STAT3 and phosphorylated STAT3(Y705) levels are increased in colonic CD; however, the patient phenotype is not considered (44). Similarly, phosphorylated STAT3(Y705) is increased in colonic mucosal T cells in patients with CD compared with peripheral T cells in healthy volunteers (45). Although general STAT3 deletion is a perinatal lethal mutation (46), conditional knockout of STAT3 in bone marrow cells causes a granulomatous transmural inflammation of intestine and colon similar to CD with altered innate immune responses and increased NF-κB activation (47). These differences may be accounted for by different STAT3 activation pattern, STAT3(S727), seen in intestinal muscle coupled to fibrosis rather than STAT3(Y705) seen in epithelial and T cells coupled to inflammation.

Our understanding of STAT3 activity has recently undergone significant paradigm shifts (19, 20). The traditional notion of canonical and noncanonical STAT3 signaling whereby “inactive” STAT3 existed as cytosolic monomers and is activated by dimerization following Y705 phosphorylation has been superseded by the understanding that 1) nonphosphorylated STAT3 exists as dimers and can be active, 2) phosphorylation of either Y705, S727 or both may confer STAT3 transcriptional activity, 3) STAT3 activity can be exerted by nuclear transcriptional effects or sequestration in signaling endosomes, and 4) STAT3 function is epigenetically regulated by acetylation on K140 or K685 (21, 22, 48, 49). The canonical activation of STAT3 relies on phosphorylation of Y705 resulting in dimerization, nuclear translocation, and activation of target genes. It is now clear, including from data obtained in our study, that noncanonical activation of STAT3 via phosphorylation of STAT3(S727) can also result in nuclear translocation and regulation of target genes. This has been found for activation of NF-κB in the development of colitis-associated cancers (50). Physiologic function has also been attributed to phospho-STAT3(Y705) localization to sequestering endosomes (20). This notion may be supported by the findings in this paper where the basal cytosolic localization of phospho-STAT3(Y705) to discrete loci is lost when a dnSTAT3(Y705F) mutant is expressed (Fig. 4B, right panel). Similarly, the differing effects of expression of a dnSTAT3(Y705F) mutant on TGF-β1, COL1A1, and CTGF expression in response to IL-6 suggest cooperativity between phospho-STAT3(S727) and phospho-STAT3(Y705) in regulating transcription. Some functional transcriptional cooperativity must exist between the two isoforms as evidenced by the differential effects of dn-STAT3(Y705F) mutant on TGFB1 and CTGF but not COL1A1 genes. STAT3 function is also epigenetically regulated via acetylation on K140 and K685 by CBP and via deacetylation by HDAC1 (51). Although these modifications do not affect phosphorylation status, they can alter stability of functional STAT3 dimers and their DNA binding.

In summary, in the strictured segment of ileum of patients with Montreal B2 phenotype CD, excess IL-6 production, low SOCS3 levels and preferential STAT3(S727) transcription factor activation lead to increased and unrestrained TGF-β1, collagen I, and CTGF expression. This abnormal STAT3 activation in these patients may reflect the potential functional outcome from polymorphisms in one candidate network in CD.

The authors have no financial conflicts of interest.