Transforming Growth Factor-β Promotes Pro-fibrotic Behavior by Serosal Fibroblasts via PKC and ERK1/2 Mitogen Activated Protein Kinase Cell Signaling

Abstract

Objective:

To assess the role of fibroblasts, transforming growth factor (TGF)-β, and cell signal pathways in promoting fibrosis in Crohn's disease (CD).

Summary Background Data:

Intestinal strictures are a major source of morbidity in CD. Fibroblasts found at sites of stricture promote fibrogenesis. The mechanisms underlying this pro-fibrotic behavior remain elusive.

Methods:

Fibroblasts were isolated from strictured and macroscopically normal serosa in patients with CD and from normal serosa in patients with colorectal cancer. Whole cell connective tissue growth factor (CTGF) and fibronectin expression were determined by Western blot analysis. Fibroblast type I collagen expression was evaluated by real-time PCR, while fibroblast contractile activity was measured using fibroblast populated collagen lattices. Cells were stimulated with TGF-β₁ and inhibitors of the protein kinase C (PKC) and ERK 1/2 mitogen activated protein (MAP) kinase cell signaling pathways.

Results:

Stricture fibroblasts displayed enhanced constitutive expression of fibronectin. TGF-β promoted fibroblast CTGF, fibronectin, and type I collagen expression and enhanced fibroblast contractile activity. Inhibition of PKC reduced basal collagen expression and contractile activity in Crohn's fibroblasts and attenuated the effect of TGF-β on fibroblast CTGF, fibronectin, and collagen I expression as well as fibroblast contractility. ERK 1/2 inhibition had a similar effect on TGF-β-induced CTGF and fibronectin expression.

Conclusions:

TGF-β is a critical pro-fibrotic growth factor in CD, and its effects are mediated via PKC and ERK 1/2 MAP kinase cell signaling. These pathways may represent novel therapeutic targets for patients with CD characterized by recurrent intestinal stricture formation.

TGF-β is a critical pro-fibrotic growth factor in Crohn's disease. Inhibition of the cell signalling pathways by which its effects are mediated may reduce stricture formation.

Intestinal strictures in Crohn's disease (CD) are the end product of chronic transmural inflammation characterized by dysregulated wound healing that results in excessive and abnormal deposition of extracellular matrix (ECM).¹ Abnormal contraction of this ECM leads to scar formation, tissue distortion, and ultimately intestinal obstruction. It is increasingly understood that more than one third of patients with CD have a distinct fibrostenosing phenotype that results in recurrent intestinal stricture formation.² Most patients will require surgical intervention; however, this is rarely curative as up to 34% will have a symptomatic recurrence within 3 years.^3,4 Repeated intervention risks perioperative morbidity and places patients at risk for significant bowel loss.

In CD, intestinal fibroblasts display phenotypic variations that promote regional fibrosis. Fibroblasts isolated from sites of stricture have an increased capacity to produce and reorganize collagen.^5,6 These essential steps in stricture formation may, at least in part, be promoted by fibroblast overexpression of adhesion molecules (eg, intracellular adhesion molecule-1) and pro-fibrotic growth factors such as vascular endothelial growth factor and connective tissue growth factor (CTGF).^7–9 Overall, however, regulation of fibroblast contribution to fibrogenesis remains poorly understood. It is probable that differential cytokine expression drives development of abnormal pro-fibrotic fibroblast populations in CD. The transforming growth factor (TGF)-β family have a universal role in the development of fibrosis with the TGF-β₁ subtype, in particular being implicated in pathogenic fibrosis.¹⁰ Overexpression of TGF-β and its receptors in both the intestinal wall, and in fibroblast cultures taken from sites of intestinal stricture, in patients with CD suggests a potential regulatory role for this cytokine in intestinal fibrogenesis.^11,12 It is known that TGF-β promotes collagen expression by intestinal fibroblasts and smooth muscle cells,^5,13 and we have recently shown that TGF-β stimulates fibroblast expression of pro-fibrotic growth factors (vascular endothelial growth factor, CTGF).^8,9 Whether TGF-β activity promotes other critical events in stricture formation such as fibroblast contractile activity and expression of ECM components such as fibronectin remains to be established.

TGF-β is known to bind membrane receptors and signals via Smad proteins to the nucleus where gene transcription is altered.¹⁴ While Smad proteins may be critical for TGF-β cell signaling, it is increasingly apparent that interaction with other signal cascades is necessary for regulation of target gene expression. Such cascades have been shown to modulate TGF-β target gene expression without directly affecting Smad activity.¹⁵ The ERK 1/2 mitogen activated protein (MAP) kinase pathway is one such cascade that has been implicated in pathogenic fibrosis in other disorders. In scleroderma, a condition that shares many immunologic features with CD, protein kinase C (PKC) and ERK1/2 MAP kinase are required for TGF-β cell signaling in cutaneous fibroblasts.¹⁶

The primary hypothesis in the current study was that overexpression of TGF-β and its associated downstream pro-fibrotic growth factors underlies enhanced expression and contraction of ECM by intestinal fibroblasts in CD. The second hypothesis was that PKC and ERK 1/2 are required for TGF-β cell signaling (Fig. 1). Targeting these pathways might not only inhibit induction of CTGF but also key endpoints in fibrosis such as ECM synthesis and contraction.

FIGURE 1. TGF-β cell signaling via Smad proteins and ERK 1/2 MAP kinase pathways. The ERK 1/2 pathway is activated by phosphorylation in a PKC-dependent or -independent manner and signals to the nucleus where gene transcription is regulated. Go6850 is a broad spectrum PKC inhibitor, PD98059 inhibits MEK 1, while Uo126 inhibits both MEK 1 and 2 subtypes.

MATERIALS AND METHODS

Patients

The study protocol was approved by the institutional ethics committee and written informed consent obtained from all patients. Eleven patients with CD (average age, 42 years) who were undergoing resection of ileal or ileocolonic strictures were recruited. Five patients who were undergoing intestinal resection for colorectal cancer (average age, 64 years) were also recruited. All surgical procedures were performed by the same consultant surgeon.

Materials

RPMI medium, penicillin, streptomycin, l-glutamine, and fetal calf serum (FCS) were purchased from Gibco Life Technologies Ltd. (Paisley, UK). Human TGF-β₁ was purchased from R&D Systems (Oxford, UK). The inhibitors Uo126, PD98059, and Bisindolylmaleimide (Go6850) were all purchased from Calbiochem (Darmstadt, Germany). Goat antihuman CTGF was purchased from Santa Cruz Biotechnology Inc. (Heidelberg, Germany). Mouse antihuman fibronectin and rat tail collagen type I were purchased from Becton Dickinson Biosciences (Cambridge, UK). p44/42 (ERK 1/2) and phospho-p44/42 (ERK 1/2) antibodies were purchased from Cell Signaling Technology (Beverly, MA). Reagents for cDNA synthesis were purchased from Invitrogen Life Technologies (Paisley, UK). Primers for real-time PCR were purchased from Applied Biosystems (Warrington, UK). All other reagents were purchased from the Sigma-Aldrich Company Ltd. (Poole, UK) unless otherwise stated.

Fibroblast Isolation and Culture

Intestinal fibroblast cultures were established using a primary explant technique. Patients with CD had 1 cm² serosal biopsies taken from sites of stricture (“stricture”), and also adjacent macroscopically normal bowel (“nonstricture”) within the resection specimen. Single serosal biopsies were taken from normal colon in patients undergoing resection for colorectal cancer (“control”). All biopsies were taken prior to bowel devascularization. Primary fibroblast cultures were then established using a primary explant technique previously described.^6–8,17 Confluent cells were characterized by their morphologic features and their immunohistochemical staining properties for vimentin, α-smooth muscle actin, and desmin as previously described.^17,18

For all experiments, fibroblasts were used between the second and fifth passages and maintained in low serum medium (1% FCS). Where inhibitors were used, the vehicle was controlled for at equal concentrations.

Protein Extraction and Western Blot Analysis

Total protein was isolated from 2 × 10⁶ fibroblasts using NP-40 isolation solution and Western blot analysis performed as previously described.^7,17 For CTGF studies, protein (40 μg) was resolved on a 12% SDS gel (75 minutes at 140 V) prior to transfer at 100 V for 80 minutes. Blots were incubated with primary antibody (goat antihuman CTGF) at a concentration of 1:100. For fibronectin, protein (40 μg) was resolved on an 8% SDS gel (210 minutes at 140 V) and transferred at 140 V for 90 minutes. Blots were incubated with primary antibody (murine antihuman fibronectin) at a concentration of 1:1000. Equal protein loading was confirmed by staining for Coomassie Blue and by β-actin expression.

For ERK 1/2 and phospho-ERK 1/2 studies, protein was isolated from whole cell lysates using SDS sample buffer containing 62.5 mmol/L Tris-HCL pH 6.8, 2% SDS, 10% glycerol, 0.1% bromophenol blue, and 50 mmol/L DTT. Samples were resolved on a 12% SDS gel (75 minutes at 140 V) and transferred at 100 V for 1 hour. Membranes were incubated overnight at 4°C with rabbit antihuman p44/42 (ERK 1/2) MAP kinase antibody (1:1000) and protein bands visualized following exposure to the secondary antibody. Stripped blots were reprobed with rabbit antihuman phospho-p44/42 (ERK 1/2) MAP kinase (Thr202/Tyr204) antibody (1:1000) at 4°C overnight. Blots were again incubated with goat anti-rabbit IgG (1:5000) and protein visualized using ECL.

Protein density on scanned Western blots was determined using the Un-Scan-It gel automated digitizing system, version 5.1 (Silk Scientific Inc., Orem, UT).

RNA Extraction and Real-Time Polymerase Chain Reaction

Total RNA was extracted from fibroblasts using an acid-guanidinium-phenol-chloroform method (TRIzol Reagent, Invitrogen) according to the manufacturer's protocol.

RNA concentrations were assessed by spectrophotometry (Beckman DU530 spectrophotometer, Beckman Coulter Inc., Fullerton, CA) at the 260:280 ratio. RNA quality was assessed by electrophoresis in a 1% agarose gel run at 80 V for 75 minutes.

cDNA for real-time PCR was prepared from RNA following treatment with DNAase I. Total RNA (1μg) was used to synthesize cDNA using random primers and the cDNA template amplified by PCR in a thermal cycler (Perkin Elmer 7700, Norwalk, CT).

Real-time PCR was performed on an ABI PRISM 7900HT Sequence Detection system (Applied Biosystems) using the following primers for collagen I: (f) 5′-TGT TGG CCC AAG AGG TCC T-3′ and (r) 5′-CAC CGG GCT CTC CCT TAT C-3′. The TaqMan probe sequence was 5′-TGG CCC ACA AGG CAT TCG TGG-3′. The endogenous control 18S RNA was purchased from Applied Biosystems as a predeveloped assay reagent (PDAR). Cycling conditions were as follows: step 1, 2 minutes at 50°C; step 2, 10 minutes at 95°C; step 3, 15 seconds at 95°C; step 4, 1 minute at 60°C; repeat step 3 × 40 cycles. Probes were labeled with 5′ FAM and 3′ TAMRA as a quencher with the exception of the ribosomal probe, which was labeled with 5′ VIC to facilitate dual reporter assay.

Fibroblast Populated Collagen Lattice

Free-floating fibroblast populated collagen lattices (FPCL) were prepared as previously described in 24-well sterile culture plates.⁶ Nonstricture fibroblasts were mixed with rat tail collagen type I and RPMI containing 1% FCS to give final concentrations of 6.25 × 10⁴/mL fibroblasts and 1.25 mg/mL collagen in a total volume of 0.5 mL. Incubation at 37°C, 5% CO₂ for 25 minutes, caused the mixture to set and the resultant gel was freed from the well by gentle manipulation with a pipette tip and refloated in 0.5 mL RPMI culture medium containing 1% FCS and test substances (TGF-β₁, Go6850, PD98059). Medium was replaced on a daily basis. Contracting lattices were photographed at 24-hour intervals over a light box from a fixed height using a mounted digital camera (Powershot A70, Canon Inc.). Analysis of calibrated images allowed area determination in cm² (Image J Software, Version 1.26t, NIH).

Statistical Analysis

Unless otherwise stated, data are expressed as mean ± SEM, and n represents the number of patients. For Western blot experiments, densitometry values for protein bands were adjusted for β-actin or total ERK 1/2 protein expression as appropriate, and statistical analysis performed on the resultant values. For collagen expression and contraction studies, statistical analysis was performed on raw data. Data were analyzed on an IBM-compatible personal computer using Minitab 13 (Minitab Inc., Coventry, UK). An analysis of variance (ANOVA) or Student t test was performed as appropriate and considered significant at the P < 0.05 level. The Mann-Whitney U test was used for nonparametric data.

RESULTS

Fibroblast Immunohistochemistry

Immunohistochemistry performed on confluent cells demonstrated typical fibroblast staining properties for vimentin (+), desmin (−), and α-smooth muscle actin (−).¹⁷

Fibroblast Expression of CTGF and Fibronectin

Fibroblasts isolated from sites of stricture in patients with CD showed a 2.7-fold increased constitutive expression of fibronectin when compared with fibroblasts cultured from nonstrictured intestine in the same patients (Fig. 2A, B). We have previously reported an increased expression of CTGF by stricture fibroblasts.⁹ This trend was again seen and correlated with an overexpression of fibronectin by stricture fibroblasts (Fig. 2A). To investigate the regulation of fibronectin expression, fibroblasts were stimulated with TGF-β₁. We have previously shown TGF-β₁ to maximally induce CTGF in intestinal fibroblasts at a dose of 1 ng/mL.⁹ Following 24-hour stimulation with TGF-β₁ at this dose, nonstricture fibroblasts behaved like stricture fibroblasts through a significantly enhanced fibronectin expression (Fig. 2A, B). TGF-β₁ had only a minimal effect in inducing fibronectin in stricture fibroblasts (Fig. 2A, B). When stimulated with TGF-β₁, nonstricture and control fibroblasts behaved similarly, showing a 2- to 3-fold increased expression of fibronectin (Fig. 2C).

FIGURE 2. CTGF and fibronectin expression by intestinal fibroblasts. Basal and TGF-β₁ (1 ng/mL for 24 hours) induced CTGF and fibronectin expression were assessed in control and Crohn's fibroblasts by Western blot analysis of whole cell protein extracts. Equal protein loading was confirmed by staining for Coommaisse blue and β-actin. Western blots were scanned and assessed for protein densitometry (B, C). A, Representative Western blots from 2 patients with Crohn's disease comparing CTGF and fibronectin expression by fibroblasts cultured from strictured and nonstrictured intestine. Fibronectin protein band densitometry is shown (B) and is expressed relative to unstimulated fibronectin levels in nonstricture fibroblasts (n = 5 patients with Crohn's disease). Data are mean ± SEM. *P < 0.05 versus unstimulated nonstricture fibroblasts (analysis of variance [ANOVA]). C, Relative induction of fibronectin by TGF-β₁ in control (n = 3) and nonstricture (n = 5) fibroblasts. Data are mean ± SEM. *P < 0.05 versus unstimulated cells (Student t test).

Effect of TGF-β₁ on Fibroblast Collagen Type I Expression and Contraction

Collagen expression following TGF-β₁ (1 ng/mL for 24 hours) stimulation was greater in nonstricture than stricture Crohn's fibroblasts (Fig. 3A). Organization and contraction of deposited collagen is a key event in fibrosis and stricture formation. Using the fibroblast populated collagen lattice, we assessed cytokine regulation of fibroblast contractile activity in vitro. TGF-β₁ induced an immediate and sustained increase in nonstricture fibroblast contractility at a dose of 1 ng/mL (Fig. 3B). This equated to a 15% increase in contractility that persisted to day 5 when collagen contraction had ceased. Increasing the dose of TGF-β₁ did not promote further contraction. Comparative contractility assays were not performed in stricture or control fibroblasts.

FIGURE 3. Collagen I expression and contraction by intestinal fibroblasts following stimulation with TGF-β₁. (A) Relative induction of collagen type I assessed by real-time PCR following stimulation with TGF-β₁ (1ng/mL) for 24 hours in nonstricture (n = 7) and stricture (n = 4) fibroblasts. Data are mean + SEM. *P < 0.05 relative to unstimulated fibroblasts (Student t test). (B) TGF-β₁ stimulated collagen lattice contraction by nonstricture fibroblasts (n = 3). The area of contracting lattices is shown after 5 days of culture in the presence of TGF-β₁ and is expressed as a percentage of the starting FPCL size. Medium containing 1% FCS acted as a control. Data are mean ± SEM. *P < 0.05 relative to 1% FCS control (analysis of variance [ANOVA]).

Phospho-ERK 1/2 Expression

TGF-β₁ (1 ng/mL) rapidly activated ERK 1/2 MAP kinase in Crohn's fibroblasts (Fig. 4). ERK 1/2 phosphorylation was seen within 30 minutes of treatment, thus demonstrating a role for this pathway in TGF-β cell signaling.

FIGURE 4. Phosphorylation of ERK 1/2 MAP kinase in nonstricture and stricture fibroblasts following stimulation with TGF-β₁ for 30 minutes. Protein band densitometry for phosphorylated ERK 1/2 is shown graphically below representative Western blots and is adjusted for total ERK 1/2 expression.

Effect of PKC and ERK 1/2 MAPK Inhibition on CTGF and Fibronectin Expression

Having shown TGF-β₁ to maximally promote CTGF and fibronectin expression in nonstricture and control fibroblasts, we proceeded to investigate TGF-β₁ cell signaling events using these cell types. We used the broad-spectrum PKC inhibitor Go6850 and, to target the ERK 1/2 MAPK cascade, the MEK 1/2 inhibitor Uo126 and the MEK 1 inhibitor PD98059. Incubation of fibroblasts with Go6850 resulted in a greater than 50% inhibition of the ability of TGF-β₁ to induce CTGF (Fig. 5A). Inhibition of MEK 1 and 2 subtypes with Uo126 at a dose of 10 μmol/L (not shown), and maximally at a dose of 50 μmol/L, produced a similar response (Fig. 5A). Inhibition of the MEK 1 subtype alone with PD98059 was equally effective, suggesting a specific requirement for this subtype in the induction of CTGF by TGF-β₁. Targeting PKC and ERK 1/2 cell signaling proved equally effective in inhibiting the induction of fibronectin by TGF-β₁ (Fig. 5B).

FIGURE 5. Effect of PKC and ERK 1/2 inhibition on the TGF-β₁ induction of CTGF (A, n = 10), and fibronectin (B, n = 7) in control and nonstricture fibroblasts. Cells were pretreated with Go6850 (Go, 10 μmol/L), Uo126 (Uo, 50 μmol/L), and PD98059 (PD, 50 μmol/L) for 45 minutes prior to the addition of TGF-β₁ (1 ng/mL) for 24 hours. Fibroblast CTGF and fibronectin expression were determined by Western blot analysis. Protein band densitometry is shown graphically beneath representative Western blots. Data are mean ± SEM. *P < 0.05 versus TGF-β₁ alone (analysis of variance [ANOVA]).

Effect of PKC and ERK 1/2 MAPK Inhibition on Collagen Expression and Contraction

Inhibition of PKC produced a 60% inhibition of basal collagen I expression and significantly inhibited TGF-β₁ induced collagen expression by nonstricture fibroblasts (Fig. 6A, B). PD98059 had no effect suggesting that the p42/44 MAPK pathway is not required for type I collagen expression by intestinal fibroblasts. The addition of Go6850 to contracting collagen lattices produced a significant inhibitory effect, resulting in a 15% to 18% inhibition of basal and TGF-β₁-induced matrix contraction (Fig. 7). ERK 1/2 blockade with PD98059 had an inconsistent inhibitory effect that failed to reach statistical significance (Fig. 7).

FIGURE 6. Effect of PKC and ERK 1/2 inhibition on collagen expression by nonstricture fibroblasts cultured from patients with Crohn's disease. The effect of Go6850 (Go, 10 μmol/L) and PD98059 (PD, 50 μmol/L) on basal (A, n = 4) and TGF-β₁ (1 ng/mL) (B, n = 5) induced collagen type I expression after 24 hours was assessed using real-time PCR. Median, interquartile range, and range are indicated by horizontal bar, rectangular box, and error bars, respectively. *P < 0.05 versus TGF-β alone (Mann-Whitney U test).

FIGURE 7. The effect of Go6850 and PD98059 on basal (A, n = 6), and TGF-β₁ (B, n = 3) induced contractile activity by nonstricture fibroblasts. Fibroblast populated collagen lattices were treated with TGF-β₁ (1 ng/mL), Go6850 (10 μmol/L), and PD98059 (50 μmol/L). Data are shown following 3 days contraction and are plotted as the percentage inhibition of contraction, relative to unstimulated (A) or TGF-β₁ (B) induced contraction. Data are mean ± SEM. *P < 0.05 versus unstimulated (A) or TGF-β₁ (B) alone (analysis of variance [ANOVA]).

DISCUSSION

Intestinal strictures in CD are characterized by excessive accumulation of abnormal collagen subtypes that are subsequently organized and contracted by fibroblasts. We and others have previously shown that such strictures are characterized by the appearance of discrete fibroblast subpopulations that display pro-fibrotic phenotypic variations including overexpression of pro-fibrotic growth factors and collagen, as well as increased collagen contractility.^5–9,19

Dermal fibroblasts and those isolated from neighboring macroscopically normal bowel in the same patients are not characterized by these abnormal pro-fibrotic characteristics.⁷ This suggests that fibroblast activity at stricture sites does not represent a constitutive abnormality but rather a response to local factors such as cytokines and growth factors secreted by acute inflammatory cells and neighboring fibroblasts. One such cytokine may be TGF-β, a pro-fibrotic growth factor that is overexpressed at sites of stricture in CD.¹¹ By comparing fibroblast behavior from sites of stricture to that of fibroblasts from neighboring healthy intestine in patients with CD, we assessed the role of TGF-β in stricture pathogenesis.

We first assessed fibronectin expression by serosal fibroblasts in CD. Fibronectin is a structural glycoprotein that is found in abundance during normal wound healing. It regulates collagen orientation as well as facilitating cell-cell and cell-matrix interaction, which promotes cell adhesion and migration. During wound healing, fibroblasts acquire features of smooth muscle cells with increased expression of α-smooth muscle actin that allows generation of greater contractile forces. De novo expression of cellular ED-A variant fibronectin promotes this differentiation into the myofibroblast phenotype, while organization of extracellular fibronectin into fibrils further facilitates contractile force generation.^20,21 In CD, increased fibronectin has been demonstrated at sites of intestinal stricture, while serum levels of fibronectin correlate with fibrostenosing disease behavior.^22–24 The present study has now shown that when compared with fibroblasts from adjacent macroscopically normal intestine, fibroblasts from strictures express significantly more fibronectin. This increased constitutive fibronectin expression may facilitate tissue contraction at sites of stricture. To investigate whether this enhanced expression represents a constitutive variation or a response to intestinal cytokine stimulation, cells were stimulated with the pro-fibrotic growth factor TGF-β₁. Stricture fibroblasts responded only minimally to TGF-β while, when stimulated, nonstricture and control fibroblasts behaved like stricture fibroblasts through enhanced expression of fibronectin. These varying responses may represent a phenotypic alteration by stricture fibroblasts in response to localized chronic TGF-β stimulation in vivo, while the response of nonstricture and control fibroblasts reflects relative naïveté to TGF-β. Increased basal fibronectin expression mirrored increased basal CTGF expression seen in stricture fibroblasts. CTGF is a downstream mediator of TGF-β and its overexpression may in part drive the increased synthesis of fibronectin.²⁵

Type I collagen is the predominant collagen subtype found at sites of stricture;¹ and in the current study, we have shown that its expression by intestinal fibroblasts is driven by TGF-β. The response to TGF-β was maximal in nonstricture fibroblasts and again suggests a relative naivety of these cells to TGF-β stimulation when compared with stricture fibroblasts, which showed a lesser and nonsignificant response. Furthermore, using the fibroblast populated collagen lattice, we have demonstrated that TGF-β promotes contraction of deposited collagen by intestinal fibroblasts. The FPCL is an established in vitro model of wound healing whereby cultured fibroblasts bind to and reorganize collagen into a more compact arrangement with resultant lattice contraction.^20,21 Stricture fibroblasts are characterized by an inherent enhanced ability to organize and contract deposited collagen.⁶ The current findings suggest that chronic in vivo fibroblast stimulation by TGF-β promotes the development of this pro-fibrotic characteristic. TGF-β may induce enhanced contractility by driving fibroblast differentiation to the more contractile myofibroblast phenotype.²⁰ Immunohistochemical staining performed on paraffin-embedded FPCLs showed that TGF-β stimulation was not associated with increased cellular expression of α-smooth muscle actin (data not shown). This can be explained by the lack of a tethering or counter force within a free-floating collagen lattice. Absence of this natural mechanical feedback results in fibroblasts failing to acquire the myofibroblast phenotype.²⁰ This does, however, suggest that TGF-β confers an enhanced contractile ability through mechanisms other than generation of α-smooth-muscle-actin fibers. Whether TGF-β acts directly or through other mediators to increase fibroblast contractility is unclear. CTGF may be one such mediator; in ocular fibroblasts, CTGF promotes enhanced collagen contractility in a similar manner to TGF-β.²⁶

Having demonstrated critical mechanisms by which TGF-β₁ promotes pro-fibrotic behavior by intestinal fibroblasts, the current study went on to assess the cell signaling pathways involved. Targeting cell signaling cascades might allow inhibition of specific TGF-β actions without compromising many of its broad physiological effects. PKC and MAPK cell signal cascades are increasingly recognized as a potential therapeutic targets, and inhibitors of these pathways have been assessed in in vivo and clinical trials.^27–30 In CD, MAP kinase pathways have been implicated in acute inflammation, and their inhibition in clinical trials has shown promising results.^31,32 However, the role of these pathways in intestinal fibrogenesis is unclear.

The current study has now shown that PKC and ERK 1/2 pathways are required for TGF-β induction of CTGF in intestinal fibroblasts. Furthermore, inhibition of CTGF was mirrored by an equivalent inhibition of fibronectin and suggests a role for CTGF in TGF-β induction of fibronectin. PKC inhibition proved effective in inhibiting TGF-β induced collagen synthesis and both basal and TGF-β induced collagen contraction by Crohn's fibroblasts. These novel findings suggest a critical role for this signaling molecule in multiple aspects of intestinal fibrogenesis.

Development of abnormal fibroblast subpopulations that display an enhanced capacity to express pro-fibrotic growth factors and ECM components, and to contract deposited collagen appears to be a critical step in stricture formation in CD.

TGF-β₁ powerfully promotes these characteristics in intestinal fibroblasts. Overexpression of TGF-β in CD may drive development of abnormal fibroblast populations characterized by pro-fibrotic phenotypic variations. Targeting TGF-β as an anti-fibrotic strategy may not be clinically advantageous due to the multiple physiologic effects of this cytokine. Thus, targeting downstream mediators may be more appropriate. PKC and ERK 1/2 MAP kinase cell signaling pathways have been shown in this study to be critically involved in fibroblast expression of growth factors and ECM components, as well as fibroblast contractile activity. These novel findings identify PKC and ERK 1/2 MAPK signaling as potential therapeutic targets to inhibit the pro-fibrotic effects of TGF-β in patients with recurrent CD strictures.

Discussions

Dr. Jeppsson: Thank you very much for a nice presentation and an elegant study. Your and your group are to be complemented for your continuous effort to try to solve the problems with stricture formation in Crohn's disease. Your results may also be applicable to other situations with formation of strictures, for example, secondary to other types of chronic inflammation and infection.

I have a few questions:

Biopsies were taken from patients at operation with bowel resection. What was their state of inflammation, because you have previously shown that the state of the inflammation is important for the development of strictures. The same holds true for steroid therapy. I would like to have more information on this.

Does it make any difference whether biopsies were taken from small or large intestines? Most strictures appear in the small bowel. Do you have controls also for colon, since colonic tissue is different?

You found that there was an increased contractile activity in your fibroblasts. You have previously shown that this may be associated with an increased activity of myofibroblasts. How was this in your present study? It has also been previously seen that there is an altered apoptosis in myofibroblasts in patients with a stricture formation. Have you any data on this?

What is the primary event in this story? Is there a role for bacteria? You know that there is a derangement in intestinal barrier function in Crohn′s disease patients. Have you considered to expose your fibroblasts to bacteria and see how they react to different types of bacteria and if that reaction is different in fibroblasts taken from bowel without chronic inflammation?

Finally, your blocking experiments are very intriguing, and I wonder if you could tell us a bit more about the perspective of using them in a clinical situation.

Once again, thank you very much for your presentation, and I thank the association for giving me the opportunity to discuss the paper.

Dr. Mulsow: Thank you, Dr. Jeppsson for your comments and questions.

All of the 11 patients enrolled with Crohn's disease underwent elective surgery for obstructive symptoms relating to intestinal strictures. Disease duration and preoperative therapies, including the use of steroids, varied between patients. We did not see any correlation between these variables and fibroblast behavior in vitro. However, it is not possible to draw conclusions from a patient group of this size in whom a number of different treatment combinations have been employed and for different durations. With larger patient numbers, trends might emerge relating preoperative therapy with varying fibroblast behavior. Nonetheless, despite different preoperative treatment regimens, stricture fibroblasts consistently demonstrated similar pro-fibrotic features.

With respect to biopsy site, all patients with Crohn's disease had either pure ileal or ileocolonic strictures. Control fibroblasts were isolated from macroscopically normal colon in patients undergoing elective resection of colorectal cancer. In this paper, we did not directly compare behavior between control and Crohn's fibroblasts. Patients with Crohn's disease acted as their own internal control in that we compared fibroblast behavior from sites of stricture to that of those from neighboring macroscopically normal but anatomically similar intestine. Furthermore, the response to TGF-β was affected only by the presence or absence of fibrosis and not by the cell site of origin; for example, ileal nonstricture fibroblasts behaved similarly to colonic control fibroblasts. This is consistent with our previous reports and suggests that fibroblast behavior varies according to the presence of fibrosis but not according to the anatomic site of origin (ileal versus colonic). “Control” fibroblasts from patients with colorectal cancer were used to replicate studies of cell signaling once they had been shown to behave similarly to nonstricture CD fibroblasts following stimulation with TGF-β.

Our primary cell cultures showed typical immunohistochemical staining properties for fibroblasts with strong positivity for vimentin. There was some weak staining for α-smooth muscle actin that may reflect the presence of myofibroblasts within the cell population. This is not surprising and probably reflects ongoing differentiation between these cell types. Importantly, cells stained negatively for desmin, which is a marker for smooth muscle cells. Within the contraction model, fibroblasts do not differentiate to the myofibroblast cell type as is reflected by the lack of staining for α-smooth muscle actin. Fibroblasts must be tethered to undergo differentiation. Free-floating collagen gels lack this mechanical tension and in that respect do not mimic events in vivo. This does, however, suggest that enhanced contractility seen following stimulation with TGF-β occurs through mechanisms other than differentiation to the myofibroblast cell-type.

Our current understanding of Crohn's disease suggests that an abnormal immunologic response to bacterial products in susceptible individuals could play a role in stricture pathogenesis. Changes in the cytokine milieu following stimulation by bacterial products may ultimately promote fibrotic behavior by intestinal fibroblasts. We have not studied fibroblast responses to bacterial product exposure but may do so in the future, with particular focus on the endpoints studied in this paper.

Lastly, to the clinical applicability of our findings, MAP kinase pathways are increasingly recognized as potential therapeutic targets in a variety of disease states. The inhibitors used in this study were primarily developed for use in the oncology setting and have been used in vivo in animal and in human trials. An orally given MEK 1 and 2 inhibitor is now available and has undergone assessment in oncology trials. The challenge for us is to take our in vitro findings and assess their relevance in the in vivo setting. The main obstacle is a lack of a satisfactory animal model of chronic intestinal fibrosis. Developing such a model will be a focus of our future work.

Dr. Mortensen: Thank you for an elegant study, nicely presented. I want to ask, were any of the Crohn's disease patients phenotyped; and if they were, do you think that would have made any difference? In other words do you think this is an acquired change in the fibroblasts, or could there be some ultimately phenotypic influence? And then just to carry on with the question about clinical relevance, would the use of the agents have any effect on impaired healing of anastomoses, for example?

Dr. Mulsow: We suspect that the changes seen in stricture fibroblasts represent local phenotypic rather than constitutive genetic variations. This is based on our finding that stricture fibroblasts behave in isolation; fibroblasts from the same patients with Crohn's disease isolated from neighboring healthy intestine or from skin do not display their pro-fibrotic characteristics. This suggests that local factors, for example, cytokine activity, underlie the variation in fibroblast behavior.

Abnormal cytokine expression in response to luminal antigens may, of course, reflect genotypic variations. A hypothesis that has been raised for patients with NOD2 mutations who develop fibrostenosing disease is that the immune response distal to the mutation promotes a shift away from pro-inflammatory cytokine expression toward a pro-fibrotic state with enhanced TGF-β expression. This is an area that warrants further investigation.

With respect to clinical applicability, I am not aware of any direct evidence linking PKC, ERK1/2 inhibition with impaired wound healing. Clearly, however, any future use of such agents would have to be carefully planned to allow maximal antifibrotic activity without compromising natural wound healing processes, particularly in the perioperative setting.

Dr. Senninger: I have a comment and a question. The comment would be that, given all the efforts that you are spending on cultures, you should definitely combine your investigation with permeability measurements, just to see whether these changes you are describing in the fibroblast, result in any type of different messages through the epithelial lining.

My question would be, considering Crohn's disease as being a member of the chronic inflammatory bowel diseases, could you have a chance to look at ulcerative colitis specimens as well, because the strictures would be uncommon there. However, lots of fibroblast behaviors appear to be similar between the two diseases.

Dr. Mulsow: Thank you, Dr. Senninger, for your comments and suggestion. In this particular study, our primary aim was to assess mechanisms regulating previously established fibrotic behavior by fibroblasts in Crohn's disease. We did not look directly at fibroblast behavior in ulcerative colitis; however, much of the background to this work arose from our previous findings demonstrating varying behavior by fibroblasts from these two patient groups. For example, in our previously reported collagen contraction study, stricture fibroblasts from patients with Crohn's disease displayed an enhanced basal contractility when compared with either those from patients with ulcerative colitis or noninflammatory controls. Similarly, fibroblasts from patients with Crohn's disease express more cellular intracellular adhesion molecule-1 than those from patients with ulcerative colitis. These important differences may in part explain the tendency to fibrosis seen in Crohn's disease but not in ulcerative colitis.