In Crohn's disease, one of the two major forms of inflammatory bowel diseases in human beings, persistent and chronic inflammation promotes fibrotic processes thereby facilitating formation of strictures, the most common indication for surgical intervention in this disorder. The pathogenesis of Crohn's disease-associated fibrosis is not fully understood, but variants of genes involved in the recognition of microbial components/products [e.g. CARD15 (caspase-activating recruitment domain 15) and ATG16L1 (autophagy-related 16-like 1)] are associated with this phenotype, and experimental evidence suggests that intestinal fibrosis results from an altered balance between deposition of ECM (extracellular matrix) and degradation of ECM by proteases. Studies have also contributed to identify the main phenotypic and functional alterations of cells involved in the fibrogenic process, as well as molecules that stimulate such cells to produce elevated amounts of collagen and other ECM-related proteins. In the present review, we assess the current knowledge about cellular and molecular mediators of intestinal fibrosis and describe results of recent studies aimed at testing the preventive/therapeutic effect of compounds in experimental models of intestinal fibrosis.
Crohn's disease is a chronic inflammatory disorder that can affect any part of the gastrointestinal tract, even though lesions are more common in the distal ileum and right colon. The behaviour and natural history of Crohn's disease are highly heterogeneous and are characterized by an initial transmural inflammation, which can lead to excessive fibrosis and formation of strictures . Strictures can be single or multiple and cause partial or complete obstruction of the lumen , thus representing the main indication for surgery in Crohn's disease patients . Overall, more than one-third of Crohn's disease patients undergo at least one surgical procedure, and 70% of Crohn's disease patients with a fibrostricturing phenotype need a resection over 10 years. It is also known that within 1 year of surgery, recurrence occurs in 70–90% of Crohn's disease patients and more than half of these patients develop new strictures, which could be responsible for further resections .
The pathogenesis of fibrostrictures in Crohn's disease remains unknown. This is because mechanisms responsible for intestinal fibrosis are difficult to analyse because of the relatively late presentation of clinical signs and symptoms of this complication. By the time Crohn's disease patients become symptomatic, fibrosis is well established and this limits the possibility of evaluating the sequence of pathogenic events that drive fibrogenesis. Researchers have largely used animal models of intestinal fibrosis to investigate the initiation and progression of disease, but they do not recapitulate the major features of fibrostrictures seen in Crohn's disease and therefore could have little relevance to the human condition . However, the available data delineate a complex scenario in which various cell types and multiple molecules contribute to sustain the fibrogenic process in the gut .
In the present review, we examine the available data on the mechanisms and factors that are involved in the development of intestinal fibrosis and discuss novel drugs which could interfere with major fibrogenic pathways.
GENETIC BASIS OF INTESTINAL FIBROSIS
The fact that fibrostrictures can develop in a very short time after the initial diagnosis of Crohn's disease and, in some patients, tend to reoccur following intestinal resection for fibrostricturing disease, whereas other patients never develop strictures, suggests the existence of a genetic background predisposing to the development of such a complication [7,8]. Several studies have found a strong genotype–phenotype association between three major variants of the CARD15 (caspase-activating recruitment domain 15) gene (encoding R702W, G908R and L1007fsinsC) and fibrostrictures [7–11]. CARD15 encodes NOD2 (nucleotide-binding oligomerization domain-containing protein 2), a member of the Apaf-1 (apoptotic protease-activating factor 1)/Nod1 family of caspase-recruitment domain-containing proteins, which is expressed by epithelial cells, mesenchymal cells, endothelial cells and monocytes [12,13]. NOD2 acts as a cytosolic pattern-recognition receptor, which is activated by the peptidoglycan fragment muramyl dipeptide . CARD15/NOD2 variants are independent predictive factors for ileal location of the Crohn's disease lesions, fibrostricturing phenotype and need for surgery , as well as for post-operative recurrence . In patients with ileal localization of the lesions, CARD15/NOD2 variants increase the risk of fibrostenosing [10,16], even though this does not, however, exclude the possibility that the association between CARD15 variants and development of fibrostrictures may be due to the effect of a neighbouring gene in linkage disequilibrium with CARD15.
Since NOD2 frameshift mutations in patients with Crohn's disease are located in the leucine-rich region of the gene and result in a 33-amino-acid truncation of the protein , these variants may inhibit interaction of NOD2 with other intracellular proteins. In this context, NOD2 can interact with proteins involved in the activation of autophagy [e.g. ATG16L1 (autophagy-related 16-like 1)], a conserved catabolic process whereby, in response to nutrient deprivation or stress, a cell degrades proteins and/or organelles via the lysosomal pathway [18,19]. In cells infected with Shigella flexneri, ATG16L1 is recruited to bacterial entry sites by NOD2 and this membrane co-localization is abrogated when NOD2 is replaced by the inactive frameshift SNP (single nucleotide polymorphism) encoding L1007fsinsC, which retains ATG16L1 in the cytosol . The link between NOD2 and ATG16L1 in the activation of autophagy could be relevant for intestinal fibrogenesis, as the most common and well-studied genetic variant of ATG16L1 (rs2241880; leading to a T300A conversion) exhibits a strong association with the fibrostricturing phenotype in both paediatric and adult Crohn's disease patients [20,21]. How the NOD2 and ATG16L1 variants facilitate the development of fibrostrictures remains unknown. A possibility is that CARD15 and/or ATG16L1 mutations can alter the responsiveness of immune cells to bacterial components/products, thereby amplifying inflammatory signals that could eventually stimulate mesenchymal cells to make enormous amounts of collagen and other fibrogenic molecules .
Association with Crohn's disease-related intestinal strictures has been also reported for two SNPs (encoding V249I and T280M) in the gene coding for the receptor of CX3CL1 (CX3C chemokine ligand 1)/fractalkine [23,24] and variants of genes for TGFβ (transforming growth factor β), IL (interleukin)-10, IL-23 receptor, TLRs (Toll-like receptors), MMPs (matrix metalloproteinases) and TIMPs (tissue inhibitors of MMPs) [25,26].
Although the available data support the hypothesis that development of intestinal strictures in Crohn's disease is genetically influenced, further studies are needed to clarify whether and how such genetic changes facilitate switching of the inflammatory pattern towards a fibrostricturing phenotype during the course of the disease.
CELLULAR MEDIATORS OF INTESTINAL FIBROSIS
In the fibrotic tissue of Crohn's disease patients, there are uniformly a greater number of collagen-producing mesenchymal cells, and this alteration is secondary to increased proliferation and decreased apoptosis of such cells . Inflammation is the major stimulus for the production of collagen by mesenchymal cells, even though it remains unclear whether these cells need continuous exposure to the inflammatory microenvironment to promote fibrosis or whether the ‘fibrogenic’ phenotype results from the interaction between genetic and immune-inflammatory stimuli. The latter hypothesis appears to be more likely as drugs that are effective in controlling mucosal inflammation do not prevent the development of fibrosis [28,29] and the major molecular pathways driving inflammatory responses are different from those that induce collagen synthesis [30,31]. Similarly, we do not yet have a clear definition of the ‘fibrogenic phenotype’ of intestinal mesenchymal cells, as the normal intestine contains a heterogeneous population of mesenchymal cells, some of which synthesize significant amounts of collagen. In particular, both vimentin- and α-SMA (α-smooth muscle actin)-expressing subepithelial myofibroblasts and vimentin-positive fibroblasts, which are present in the submucosa and serosa, are the primary sites of expression of collagen mRNA and protein in the normal intestine [2,32]. In Crohn's disease, these cell types are the major source of the collagen which accumulates in the muscularis layers [33,34]. It is, however, unclear whether these cells derive from the activation and expansion of fibroblasts or myofibroblasts residing in connective tissue between smooth muscle bundles, from the migration of activated fibroblasts and myofibroblasts from the mucosa or submucosa to the muscularis or from activation of smooth muscle cells or interstitial cells of Cajal. Development of strictures in Crohn's disease is also favoured by overgrowth of the muscularis mucosa and muscularis propria . Thickening of the muscularis layers is marked by increased numbers of vimentin-positive cells in these areas. In fibrotic samples, entire layers of histologically normal muscularis are populated primarily by vimentin-positive/α-SMA-negative and vimentin/α-SMA-double positive fibroblasts and myofibroblasts rather than normal vimentin-negative smooth muscle cells , thus suggesting that muscularis overgrowth in Crohn's disease involves a change in enteric smooth muscle cells towards a fibroblast or myofibroblast phenotype.
Increased levels of collagen types I, III, IV and V RNA and protein are seen in strictured intestine of Crohn's disease patients . Increased levels of mRNAs encoding collagens I, III and V are also detectable in inflamed mucosa of patients with ulcerative colitis, but in this disorder, development of strictures is less common than in Crohn's disease . Changes in post-transcriptional regulation of collagen are crucial in the development of fibrostrictures. Indeed, intestinal fibrosis seems to stem from the imbalance between deposition of ECM (extracellular matrix) proteins (e.g. collagen, vitronectin, fibronectin, osteopontin and thrombospondin) and degradation of ECM by various proteases including MMPs [38–40]. Most MMPs are secreted as inactive proenzymes, and their activation requires proteolytic processing of the N-terminal domain of the protein . The activities of most MMPs are very low or negligible in the normal intestine and inflammatory cytokines, growth factors, hormones, cell–cell and cell–matrix interactions can increase the expression and activity of such enzymes . Because elevated levels of MMPs are seen in Crohn's disease , we can exclude the possibility that ECM accumulation seen in Crohn's disease fibrotic tissue relies on defects in MMP production. MMP activity is also regulated by their natural inhibitors, TIMPs [43,44]. Therefore, during the course of chronic inflammation, marked induction of TIMPs could block the activity of MMPs, with the downstream effect of facilitating fibrogenetic processes. Consistent with this is the demonstration that TIMP-1 is overexpressed in intestinal myofibroblasts isolated from strictured tissue of Crohn's disease patients  and in the colon of mice with chronic colitis-driven intestinal fibrosis .
MOLECULAR MEDIATORS OF INTESTINAL FIBROSIS
Experimental evidence indicates that mesenchymal cells can produce copious amounts of collagen and other ECM-related proteins in response to stimulation with growth factors, cytokines and PAMPs (pathogen-associated molecular patterns) [47,48]. Once activated, mesenchymal cells produce additional molecules that could promote autocrine regulation of pathways controlling major steps in fibrogenesis [49,50]. Intestinal fibroblasts isolated from patients with Crohn's disease express TLRs 2, 3, 4, 6 and 7, and stimulation of fibroblasts with TLR ligands promotes the differentiation of such cells into collagen-producing myofibroblasts [51,52]. Moreover, TLR ligands enhance production of fibronectin and α-SMA in human intestinal myofibroblasts . Intestinal myofibroblasts respond to TLR ligands with increased production of CXCL8 (CXC chemokine ligand 8), which is involved in the recruitment of neutrophils to inflamed tissues and exhibits angiogenic and angiostatic activities that could contribute to the progression of intestinal fibrosis [51,53,54].
TGFβ1, an immunosuppressive cytokine produced by multiple immune and non-immune cells, is one of the major fibrogenic factors in the gut  given its ability to promote ECM-related protein production . TGFβ1 significantly increases collagen type III synthesis by lamina propria fibroblasts isolated from Crohn's disease-associated strictures , stimulates TIMP-1 production and suppresses MMP expression . Moreover, TGFβ1 promotes synthesis of fibronectin and type I collagen and enhances fibroblast contractile activity . The biological function of TGFβ1 is mediated by interaction of the cytokine with TGFβ receptor type II, an event that promotes phosphorylation of Smad2/3, two intracellular proteins. Following phosphorylation, Smad2/3 interact with Smad4, and the resulting complex translocates to the nucleus, thus controlling expression of target genes [59,60]. The TGFβ1-associated Smad pathway is negatively regulated by inhibitory Smad proteins (Smad6 and Smad7), which bind to the TGFβ receptor I and prevent phosphorylation of Smad2/3 [59,60]. TGFβ1 can also activate ERK (extracellular-signal-regulated kinase) 1/2, JNK (c-Jun N-terminal kinase), and p38 MAPK (mitogen-activated kinase) in various cell types. There is also evidence for cross-talk between MAPK and Smad pathways, given that ERK1/2, JNK and p38 MAPK are capable of either activating or inhibiting Smad signalling depending on the cell context analysed [61–65]. The mucosa overlying Crohn's disease strictures contains elevated levels of TGFβ1 RNA transcripts, phosphorylated Smad2/3 and TIMP-1 and low levels of Smad7, MMP-12 and MMP-3 compared with those seen in mucosa overlying non-stricture areas of the same patients . These findings would suggest that the profibrogenic effect of TGFβ1 is mediated by the Smad pathway and compounds that enhance TGFβ1-associated Smad signalling can attenuate inflammatory responses but promote fibrogenesis. This could be the case with Mongersen, a specific Smad7 antisense oligonucleotide-containing oral drug, which has been recently used with success in patients with active, non-stricturing, non-perforating Crohn's disease . No stricture was seen in Mongersen-treated patients [67,68]. Although the safety profile of Mongersen needs to be confirmed in larger studies in which patients will be treated for longer periods, the above data suggest that the in vivo pathogenesis of Crohn's disease strictures is more complex than that suggested by the in vitro evidence, and that TGFβ1 can use the MAPK-dependent pathway rather than Smad signalling for controlling ECM deposition and fibrogenesis. This hypothesis is supported by the demonstration that inhibition of ERK1/2 activation in Crohn's disease fibroblasts reduces basal collagen expression and contractile activity and attenuates the inducing effect of TGFβ1 on fibronectin and collagen I production .
Proliferation of myofibroblasts and smooth muscle cells and production of collagen by these cell types are also stimulated by IGF-1 (insulin-like growth factor 1) [69–75], CTGF (connective tissue growth factor), PDGF (platelet-derived growth factor) and multiple inflammatory cytokines produced in excess in Crohn's disease [27,49,76].
EXPERIMENTAL THERAPEUTIC TARGETS IN INTESTINAL FIBROSIS
There is currently no effective medical therapy for prevention and cure of Crohn's disease-associated fibrostrictures, and surgery is still the only treatment for symptomatic/complicated strictures. The limited knowledge of the factors/mechanisms leading to intestinal fibrosis has largely hampered the development of anti-fibrotic drugs, but evidence has been accumulated to show that some compounds could help to attenuate the fibrogenic process. For example, it has been demonstrated that trans-resveratrol, also known as 3,5,4′-trihydroxy-trans-stilbene, a natural stilbenoid found at high concentration in skins of red grapes and berries, has anti-fibrotic effects in the gut. In particular, trans-resveratrol decreases proliferation and collagen I synthesis by rat intestinal smooth muscle cells and fibroblasts in vitro , and this effect has been linked to the ability of resveratrol to inhibit IGF-1R (IGF-1 receptor) activation and reduce the responsiveness of fibroblasts to IGF-1 . In a model of chronic intestinal inflammation-driven fibrosis induced by injection the PG–PS (peptidoglycan–polysaccharide) complex in the intestinal wall of rats, resveratrol treatment reduces expansion of the submucosa and disruption of the muscularis externa by bands of collagen without altering the inflammatory infiltrate . At the highest dose (100 mg/kg per day), resveratrol reduces significantly expression of TGFβ1 RNA, and other profibrotic factors (i.e. procollagen type I, procollagen type III and IGF-1). The basic mechanism by which resveratrol interferes with intestinal fibrogenesis is unknown, but studies in other systems have shown that the anti-fibrotic effect of this compound is partly mediated by activation of the NAD-dependent protein deacetylase SIRT1 (sirtuin 1) and relies on the block of Smad3 function . In this context, we have recently shown that SIRT1 expression is reduced in Crohn's disease mucosa  and provided preliminary evidence linking the defective Smad7-dependent TGFβ1 activity with diminished SIRT1 expression in Crohn's disease (S. Sedda and G. Monteleone, unpublished work).
Another compound that modulates the TGFβ1 pathway and could have anti-fibrotic effects is cilengitide. Cilengitide is an inhibitor of αVβ3, an integrin expressed by muscle cells. αVβ3 regulates IGF-1-dependent proliferation of muscle cells and contributes to excess hyperplasia in intestinal strictures in Crohn's disease . Occupancy of αVβ3 by integrin ligands (e.g. vitronectin and fibronectin) stimulates smooth muscle proliferation by maximizing the intensity and duration of IGF-1-stimulated IGF-1R activation and effects . Cilengitide reduces levels of active TGFβ1 and, in an animal model of TNBS (2,4,6-trinitrobenzenesulfonic acid)-driven colitis-induced fibrogenesis, decreases collagen I production and inhibits the development of fibrosis over a 6-week period .
Another pathway of interest for therapeutic intervention is the RAS (renin–angiotensin system) that regulates multiple biological functions, including cell growth, inflammation and fibrosis . ACE (angiotensin-converting enzyme) is expressed by intestinal epithelial cells and appears to be involved in the control of intestinal epithelial cell apoptosis [84,85]. The effects of ACE and AngII (angiotensin II) are mediated through a series of cell-surface AngII receptors, which are expressed in the human intestinal mucosa . Transanal administration of enalaprilat, a pharmacological ACE inhibitor, to mice with DSS (dextran sulfate sodium)-induced chronic colitis reduces synthesis of inflammatory cytokines and colonic fibrosis , reinforcing the notion that ACE is a central mediator in the pathogenesis of fibrosis. In line with this are the studies showing significantly reduced fibrosis formation with the ACE inhibitor captopril in the TNBS-induced model of colitis  and reduced tissue collagen content with captoril and lisinopril in a TNBS-induced colitis rat model . These studies also showed that ACE inhibitors reduced, but did not abrogate, intestinal fibrosis [87–89], raising the possibility that fibrosis is mediated by additional mechanisms other than those controlled by ACE inhibitor treatment.
Many in vitro and in vivo studies have tested the anti-inflammatory and anti-fibrotic effects of the widely used cholesterol level-lowering HMG-CoA (3-hydroxy-3-methylglutaryl-CoA) reductase inhibitors (statins) in various organs. However, atorvastatin is the only statin that has been investigated in patients with active Crohn's disease. Although the studies involved a small number of patients, a clinical benefit was seen following treatment with the drug [90,91]. Larger randomized controlled trials are needed to confirm these preliminary results, as well as to investigate effects on fibrogenesis as a primary outcome.
Fibrosis remains one of the less understood complications of Crohn's disease, in terms of pathogenesis, early diagnosis and treatment. More work is therefore needed in the near future to dissect the major mechanisms leading to this complication and identify target molecules for effective preventive or curative therapies. Unfortunately, no experimental model of intestinal fibrosis recapitulates the major features of Crohn's disease-associated fibrostrictures, and the most accredited models of acute and chronic colitis, which have largely advanced our understanding of the tissue-damaging immune-inflammatory response in Crohn's disease, have been developed in mice, a species that is quite resistant to fibrosis. Further work is needed not only to establish which model should be used to track major changes in phenotype and number of mesenchymal cells during acute and chronic injury from onset to progression to fibrosis, but also to ascertain the real impact that genetic changes and environmental stimuli exert in the development of this process.
As revealed by previous studies, TGFβ1 apparently plays a major role in the intestinal fibrogenesis and could be the main therapeutic target. In this context, however, it is noteworthy that TGFβ1 is also the most powerful immunosuppressor in the gut and anti-TGFβ1 therapy could improve the course of fibrosis and, at the same time, promote the development of, or exacerbate, other complications of Crohn's disease.
G.M. is funded by grants from Giuliani SpA, Broad Medical Research Foundation, Novo Nordisk, Teva, Sirtris, Lycera and Sofar, and personal fees from AbbVie and Zambon outside the submitted work.
autophagy-related 16-like 1
caspase-activating recruitment domain 15
insulin-like growth factor 1
c-Jun N-terminal kinase
nucleotide-binding oligomerization domain-containing protein 2
α-smooth muscle actin
single nucleotide polymorphism
transforming growth factor β
tissue inhibitors of MMP
Giovanni Monteleone is the owner of a patent related to the use of Smad7 antisense oligonucleotides in Crohn's disease (PCT Pub. No. WO2004/087920).