Characterization of myofibroblasts isolated from the intestine of patients with inflammatory bowel disease

Background: Intestinal fibrosis represents a serious complication of inflammatory bowel diseases (IBD), often necessitating surgical resections. Myofibroblasts are primarily responsible for interstitial matrix accumulation in fibrotic diseases. However intestinal myofibroblasts (IMF) remain inadequately characterized.  The aim was to examine fibroblast markers and fibrosis-associated gene expression in IMF isolated from resected intestine from IBD and control patients. As well as determining the effect of the fibrogenic cytokine TGFβ. Methods: Intestinal resections were obtained (n =35) from consenting patients undergoing elective surgery (2014-16). Primary cultures of IMF were isolated using DTT and EDTA and cultured. Viability and phenotypic characterization of IMF was carried out by flow cytometry and fluorescence microscopy. IMF (passages 3-8) were treated for 24 hours. Cytokines were quantified in IMF by real time PCR and in supernatants using the human pro-inflammatory cytokine panel  Results: All markers and most fibrosis mediators studied were preferentially expressed by IMF compared to mucosal tissue. Metalloproteinases (MMP) 2 and 3, as well as their inhibitor TIMP1, are highly expressed by IMF. They also highly expressed inflammatory mediators, including IL-6, IL-8, CCL2 and PTGS2. Whereas mucosal expression of pro-inflammatory cytokines such as TNFα and IL-17 is increased in IBD, that of fibrosis mediators was not different. Fibrosis-related gene expression in IMF from IBD patients and controls was similar, but IMF from IBD expressed higher levels of several inflammatory genes. IMF from CD and UC had mostly similar expression profiles. TGFβ induced expression of fibrogenic genes αSMA, COL1A1, CTGF, FN1 and LOX. TGFβ-stimulated IMF released increased levels of IL-6, whereas IL-6, IL-8, as well as small amounts of IFN-γ and IL12p70 were produced following stimulation with IL-1β+IL-23. Conclusions: This study extends knowledge about the pathogenesis of fibrosis in IBD. Further research in the identification of mechanisms involved in IMF activation and fibrogenesis are required.


Introduction
The two major forms of inflammatory bowel disease (IBD), Crohn's disease (CD) and ulcerative colitis (UC), are characterized by chronic and relapsing intestinal inflammation that develop as the result of an abnormal regulation of immune responses, likely directed at least in part, against the host commensal gut microbiota [1][2][3] . The prevalence of IBD is estimated to be over 3.6 million persons in North America and Europe, and the incidence is increasing in Asia and Africa 4,5 . Genetic studies have identified over 200 susceptibility loci for IBD, mostly shared between CD and UC 6-8 . These loci are enriched for pathways that interact with environmental factors to modulate intestinal homeostasis, including IL23R, ATG16L1, IRGM and NOD2 9 .
The natural history and clinical course of IBD is extremely heterogeneous. Up to 16% of UC patients require a colectomy within 10 years, whereas about 50% of patients with CD develop complications such as strictures, fistulas, and abscesses that frequently require surgery within the same period 10,11 . Intestinal fibrosis is a common and potentially serious complication of CD, but not UC, that results from the reaction of intestinal tissue to the damage inflicted by chronic inflammation 12 . NOD2 has been long considered the most important genetic predictor of the evolution toward a fibrostenosing phenotype 13,14 . Fibrosis in CD is also associated with polymorphisms of the Jak2, ATG16L1, CX3CR1 and MMP3 genes 14 .
An improved understanding of the cellular and molecular mechanisms that underlie the pathogenesis of intestinal fibrosis is needed. The mechanisms are complex, dynamic, and likely involve multiple cell types and soluble factors 12 . Intestinal myofibroblasts (IMF) play important roles in inflammation and tissue remodeling 15 . The development of fibrosis results from an imbalance in extracellular matrix (ECM) deposition and degradation. Alpha-smooth muscle actin (α-SMA) positive myofibroblasts were identified as the primary cell type responsible for interstitial matrix accumulation in fibrotic diseases 16,17 . A hallmark of mesenchymal cell activation is the acquisition of a myofibroblast phenotype, whereby fibroblasts transform into myofibroblasts acquiring smooth muscle features, most notably the expression of α-SMA and synthesis of mesenchymal cell related matrix proteins. TGF-β1, a prototype of the TGF-β superfamily, is widely considered to be the major profibrogenic cytokine that is responsible for the myofibroblast differentiation and subsequent matrix synthesis 16,18 . Although the TGF-β/ Smad pathway is considered a driving force of fibrosis, myofibroblasts are activated by numerous paracrine mediators in their environment that promote their production of ECM and proliferation including, TGF-β1, PDGF, CTGF, IGFI/II, bFGF and various interleukins: IL-1β, IL-6 and IL-13 17,19 .
Although inflammation is necessary for fibrosis, recent evidence indicates that once initiated, fibrosis in CD can progress independently of inflammation 17 . Consequently, current antiinflammatory treatments may not prevent fibrosis once excessive ECM deposition has commenced 17,20 . Matrix stiffness is capable of further activation of intestinal fibroblasts and can contribute to progression of fibrosis independently of inflammation 21,22 .
There is currently little information about the identity, abundance and characteristics of intestinal mesenchymal cells such as fibroblasts and IMF under normal and pathological conditions. In this study, we examined the expression of fibroblast and IMF molecular markers in the intestine from patients with CD, UC and from non-IBD control patients.

Human intestinal tissue acquisition and consent
The McGill University Health Center's research ethics board approved the study design and consent forms. Intestinal resections were obtained (2014-2016) from patients undergoing elective intestinal surgery who voluntarily gave written informed consent to participate. Written informed consent for publication of the participants/patients' details and/or their images was obtained from the participants/patients/parents/guardian/relative of the participant/patient Resected tissue was obtained from 15 CD and 6 UC patients, as well as from uninvolved surgical specimens (>5 cm from the tumor margin) of 14 control patients undergoing colectomy for carcinoma or polyps. The characteristics of the patient groups are shown in Table 1.

Isolation and culture of human intestinal subepithelial fibroblasts and myofibroblasts
Primary cultures of IMF were isolated and cultured according to the method reported by Mahida et al. 23 . Briefly, tissue specimens were trimmed of fat and thoroughly washed. The mucosa was cut into 0.5 cm pieces and incubated in 1.5 mM DTT (Sigma-Aldrich Canada Ltd, Oakville, ON, Canada) in HBSS for 15 min to remove mucus. Tissue was washed and then incubated in HBSS containing 2 mM EDTA for 2 × 30 min in a shaker at 37°C.
The resulting mucosal samples denuded of epithelial cells were cultured at 37°C, in RPMI-1640 supplemented with 10% fetal calf serum (FCS) to allow myofibroblasts to migrate out, in order to establish primary cultures. Established colonies of IMF were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FCS, 1% non-essential amino acids (Gibco, Carlsbad, CA, USA), and 200 mM glutamine (Sigma). Experiments were performed on passages 3-8.

Viability and Phenotypic Characterization of Primary Myofibroblast Cultures
Cells obtained as above were prepared for flow cytometry as follows. After trypsinization, cells were counted and distributed in 1x10 6 per sample tubes. Cells were washed twice in PBS and blocked with Human BD Fc Block (BD Biosciences, Mississuagua, ON) according to the manufacturer's instructions. Samples were then incubated with eBioscience TM Fixable Viability Dye eFluor 450 (Life Technology Inc., Burlington, ON) for 30 min in ice, washed again in PBS and incubated in PBS containing the surface marker antibodies for 20 min in ice (Table 2). Samples were then permeabilized with BD Cytofix/Cytoperm Buffer (BD Biosciences) for 20 min on ice and then washed with BD Perm/ Wash (BD Biosciences) and incubated with BD Perm/Wash Buffer containing intracellular marker antibodies for 20 min on ice (Table 2). Samples were washed with BD Perm/Wash and   In some experiments, cells were incubated with SMAD (5uM, x 2hr) and/or TGFα (3 ng/ml, overnight).
Real time polymerase chain reaction (qPCR) mRNA from IMF was isolated with RNeasy Mini Plus kit (Qiagen, Toronto, ON, Canada) according to the manufacturer's instructions. cDNA was generated from 1 µg of total RNA using Transcriptor First Strand cDNA Synthesis kit (Sigma-Aldrich Canada Ltd). Real-time RT-PCR was performed by a "StepOnePLus" RT-PCR (Life Technologies, Burlington, ON) using PerfeCTa SYBR green Fast Mix (Quanta Biosciences, Beverly, MA). Primers were as described in Table 3.
Primers were ordered from Integrated DNA Technologies (Coralville, IA). Gene expression was standardized using GAPDH expression. Results were quantified using the ΔCt or 2ΔCt method. Table 3. Sequence of primers used. Statistical analysis Data were analyzed using GraphPad and presented as median. Statistical analysis of the results between groups was determined by the Mann-Whitney test. Significance was set as p<0.05.

Results and discussion
Characterization of primary IMF cultures The viability of IMF cells as determined by flow cytometry was 97.8% (n= 3/group; passages 3-8). No difference was observed between UC, CD and control groups. Immunofluroescence microscopy revealed that the IMF were aSMA+, CD90+, FSP1 +, and negative for hematopoietic and epithelial markers CD45 and EpCAM. Representative images are illustrated in Figure 1.

Expression of fibroblast markers by primary IMF cultures
We examined the gene expression of fibroblast markers in IMF cultures and compared their levels with those in the mucosa. The expression of fibroblast activation protein (FAP) was more than 200 times higher in IMF cultures than in mucosal tissue. Levels of Fibroblast specific protein 1 (FSP1, also known as S100A4) and those of the stromal marker CD90/Thy-1 were 5-8 times greater in IMF than in the mucosa. The myofibroblast marker αSMA in primary IMF cultures was 2-fold the mucosal level ( Figure 2A). We then examined the expression of different mediators implicated in the fibrotic process. As illustrated in Figure 2A, IMF expressed high levels of type 1 collagen (COL1A1) and fibronectin 1 (FN1). Connective tissue growth factor (CTGF/CCN2), a matricellular protein recognized to promote matrix protein deposition and fibrogenesis, was similarly expressed by IMF and mucosa. CCN1, another matricellular protein, was preferentially expressed by IMF.
Fibrosis is a pathological condition characterized by the deposition of excessive or abnormal ECM components, including collagen type I. Metalloproteinases (MMP) are responsible for the degradation of ECM components and are thus play a central role in ECM remodeling. Their activity is controlled by tissue inhibitors of metalloproteinase (TIMP). MMP-2 and MMP-3 are highly expressed in IMF (over 100 times the mucosal levels). However, MMP-9 transcript levels were higher in the mucosa (p=0.05). IMF were found to be an important source of TIMP1, with expression largely exceeded that of the mucosa ( Figure 2B). Accumulating evidence points toward the NAD(P)H oxidase of the Nox family and particularly Nox4 as the predominant enzyme source for ROS generation in fibrotic disease 24 . As shown in Figure 2B, IMF expressed significant levels of NOX4 transcripts.
The inflammatory potential of IMF was then investigated. IMF expressed high levels of pro-inflammatory cytokines IL-6 and IL-8, more than 40 and 30 times the mucosal levels, respectively (Figure 3). The chemokine CCL2 was also more expressed by IMF cultures than mucosal tissue. Transcripts of cytokines such as TNFα, IFNγ, IL-17A and IL-23A were undetectable in most IMF samples analyzed. PTGS2, also known as cyclooxygenase-2 or COX-2, is highly expressed by IMF, approximately 200 times the mucosa levels.

Mucosal gene expression of fibrosis mediators
We then examined mRNA level of fibrosis related genes in the mucosa of IBD patients and of controls. No significant differences were found in fibroblast markers and fibrosis gene expression in mucosa from IBD patients compared to controls. MMP3, MMP9 and TIMP1 expression tended to be higher in IBD tissue, but the differences were not statistically different due to the large variation in IBD mucosa (data not shown). There was no clear expression profile related to the pathology except for TIMP1 which was more expressed in UC than in CD (Figure 4). Pro-inflammatory cytokine expression was greater in mucosa isolated from IBD patients. Expression of IL-17 and TNFα, but not IL-23A, were significantly greater in the intestine from IBD patients (Figure 4). Values for IL-6 and IL-8 were too dispersed to achieve significance.

Figure 2. Comparative expression of fibrosis-related mediators and fibroblast markers by intestinal myofibroblasts (IMF) compared to intestinal mucosa levels in human bowel resections.
Gene expression was determined by real-time qPCR, and data normalized compared to GAPDH expression. Data are presented as medians. * p< 0.05, ** p< 0.005, *** p, 0.00

Fibrosis-related gene expression in IMF from control and IBD patients.
Expression of fibroblast markers and fibrosis-related genes was not different between IMF generated from IBD and control patients ( Figure 5). Increased CTGF expression by IBD IMF was marginally (significant p=0.053). FSP-1 expression was not different between IBD and control groups, but IMF from UC patients expressed lower transcript levels compared to those from controls (p=0.0357).
Inflammatory gene expression in IMF from control and IBD patients IMF from IBD resections expressed higher levels of IL-6 than those from controls (p=0.0357, Figure 6). IL-8 was nearly 100 times more expressed by IMF from IBD but due to the small number of samples the difference did not reach statistical significance. CCL2 expression was also 30 times more in IMFs from IBD patients. There was a trend towards higher PTGS2 expression in IMF from CD patients compared to those from controls and UC patients.

Effect of TGF-β on fibrosis-related gene expression in IMF from control & IBD patients
TGFβ stimulation of IMF resulted in an increased expression of profibrotic genes COL1A1 and CTGF. CCN1 as well as FN1 expression were also upregulated following stimulation with TGFβ ( Figure 7). Lysyl oxidase (LOX), a collagen modifying enzyme required for the cross-linking of collagen, was also induced. There was marked upregulation of αSMA of IMF. On the other hand, TGFβ down-regulated expression of fibroblast markers FSP1 and CD90. TGFβ slightly induced MMP9 and TIMP1 expression, although not significantly. No significant effect of TGFβ was observed on NOX4, IL-6, IL-8 and PTGS2 expression, although IL-6 expression was increased in 4 of 6 IMF cultures.

Modulation of inflammation-related gene expression in IMF by fibrogenic and inflammatory signals
To assess the inflammatory properties of IMF, we determined their cytokine release following fibrogenic and inflammatory stimulation. IMF spontaneously released IL-6 and IL-8 (Figure 8a). TGFβ induced IL-6 release (Figure 8b). Stimulation with IL-1β+IL-23 increased IL-6 and IL-8 production. In addition, small amounts of IFNγ and IL-12p70 were released (Figure 8c).
Fibrosis is a chronic, progressive process characterised by an excessive deposition of collagen and other ECM components.
Intestinal fibrosis is a common complication of CD, an incurable disorder, forcing patients to undergo bowel resections over their lifetime. More than 40% of CD patients with ileal involvement will require one or more resections of strictures 25 . Although less common in UC, longstanding disease is believed to cause fibrosis resulting in altered bowel function 26,27 . In this study, isolation techniques used yielded multiple cell populations present in the intestinal lamina propria including: immune cells, epithelial cells, and MF. This permitted the establishment of primary IMF cultures with a high rate of viability. After the first passage, only MF remained, as identified by their unique phenotype (Figure 1).
A variety of signals promote fibroblast differentiation into IMF and augment ECM expression. TGFβ1 represents the prototype of the profibrogenic mediators, with its unique ability to drive myofibroblast activation through both canonical and non-canonical signaling pathways leading to expression and deposition of ECM 18 .
There is little information about the identity, localization, and abundance of the different intestinal mesenchymal cell types.
Various studies show that fibroblasts isolated from different tissues are morphologically and functionally heterogeneous subpopulations 28 . Several fibroblast markers have been described, but none of them is unique to fibroblasts, and not all fibroblasts express the proposed markers. This lack of specific markers has impeded the characterization of IMF and their putative precursors. Moreover, characterization of these markers under normal and pathological conditions is still lacking. This study aimed to determine the expression of several markers as well as fibrosis related mediators in IMF derived from the CD, UC and control patients.
Among the markers available to identify fibroblasts, Thy-1 (CD90), FAP and FSP1 (S100A4) are the most extensively studied 15 . Our results show that all three markers are highly expressed by IMF compared to the intestinal mucosa. IMF also express higher levels of the myofibroblast marker αSMA.
Heterogeneous expression of surface receptor Thy-1 in fibroblasts from several tissues is well established. Normal lung fibroblasts express Thy-1, whereas myofibroblasts in the fibroblastic foci in idiopathic pulmonary fibrosis lack Thy-1 expression 29 . It has been shown that loss of Thy-1 in human lung fibroblasts induces a fibrogenic phenotype 30 . In contrast to lung fibroblasts, TGFβ up-regulated αSMA expression only in Thy-1+ myometrial and orbital fibroblasts 31 . These results show that the presence rather than the absence of CD90 apparently favors the appearance of a myofibroblast phenotype in response to TGFβ. In this study, we did not observe any difference in CD90 expression between IMF from IBD patients and controls.
Our results reveal that IMF are also enriched in several mediators. COL1A1 and FN1 were highly expressed in IMF. However, expression of the fibrogenic gene CTGF, also known as CCN2, a key mediator of ECM production in pathological fibrotic conditions, was similar to that in the mucosa. This is likely because the epithelium is an important source of CTGF 32 . CCN1/CYR61 expression was augmented in IMF compared to the intestinal mucosa. To our knowledge, this is the first time that CCN1 expression is reported in IMF. CCN1 levels in parenchymal liver cells were relatively low compared to that in hepatic stellate cells and portal myofibroblasts 33 . The same study found that overexpressed CCN1 significantly inhibited production of collagen type I, attenuated TGFβ signaling and induced production of reactive oxygen species (ROS), leading to dose-dependent cellular senescence and apoptosis. On the contrary, CCN1 has been shown to augment TGF-β signaling and contribute to fibrogenic responses to lung injury 34 . Its role in intestinal fibrosis is still unknown.   It was recently demonstrated that CCN1 promotes mucosal healing in murine colitis. Mechanistically, CCN1 induced IL-6 in macrophages and fibroblasts and promoted intestinal epithelial healing 35 . IL-6 is produced by several cell types in the lamina propria. Our data indicates that IMF cultures spontaneously released IL-6 and IL-8. In addition, we found that both fibrogenic and inflammatory stimuli can up-regulate IL-6 production. IL-6 has been shown to induce production of collagen 1 and fibronectin in fibroblasts from normal lungs and in idiopathic pulmonary fibrosis. In vivo neutralization of IL-6 trans-signaling resulted in a reduction in pulmonary inflammation and fibrosis, associated with improvement in respiratory function 36 . Neutralization of autocrine IL-6 reversed STAT3 phosphorylation and normalized expression of TGFβ1 in structured intestinal muscle 37 . • Figure 1 dataset on immunostaining of IMF primary cultures.
• Figure 2 dataset on the expression of fibrosis-related mediators and fibroblast markers by intestinal myofibroblasts (IMF) compared to intestinal mucosa levels in human bowel resections.
• Figure 3 dataset on the inflammation-related gene expression in intestinal myofibroblasts (IMF) and intestinal mucosa resections.
• Figure 4 dataset on the mucosal gene expression from control and inflammatory bowel disease patients.
• Figure 5 dataset on the fibrosis-related gene expression in intestinal myofibroblasts cultures obtained from resected bowel in control and inflammatory bowel disease patients.
• Figure 6 dataset on the inflammatory gene expression in intestinal myofibroblasts obtained from control and inflammatory bowel disease patients.
• Figure 7 dataset on the effect of TGFβ on fibrosis-related gene expression intestinal myofibroblasts obtained from control and inflammatory bowel disease patients.
• Figure 8 dataset on the effect of TGFβ and IL-1β+IL-23 on cytokine production by intestinal myofibroblasts obtained from control and inflammatory bowel disease patients.

Grant information
Support for this research was in the form of funds provided to EG Seidman as a Tier 1 Canada Research Chair in immune mediated gastrointestinal disorders. PS Escobar was provided salary support through the Spanish government with a predoctoral training grant Resolución de la Presidencia de la Agencia Estatal de Investigación, por la que se conceden ayudas a la movilidad predoctoral para la realización de estancias breves en Centros de I+D, convocatoria 2016. S Restillini was provided salary support by the Government of Switzerland through her home institute Geneva University Hospital.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
inflammatory IMFs for UC/CD controls? The authors could also compare their mRNA data to what is already published in the literature.
mRNA levels of some inflammatory cytokines are enriched in IMFs over mucosa, how do the authors explain this, given IMFs are not the primary cytokine producing cells?

5.
Data and text to justify the use of GADPH as the normaliser would be useful. 6.
Figures 7 and 8 are very difficult to follow because of a lack of information in the text and the legend. For example, the title for Figure 7 states, "gene expression intestinal myobroblasts (IMF) obtained from control and inflammatory bowel disease (IBD) patients" yet there is only a single bar for each gene? This should be explained in the legend and then numbers used should be given in the legend. The use of black plus, I think, two shades of grey which do not match the key complicate matter further. Once again it would be more clinically impactful to have compared response between different diseases and phenotypes. Currently numbers are too small to do this.

7.
In Figure 8 have the cultures been pooled? Why are the numbers different between panels A/B and C? What were the criteria for selecting these particular cultures over others?

8.
There is no clear explanation as to how the data presented in Figures 7 and 8 relate to the mRNA data given in earlier Figures. There is a lack of explanation as to why select the various cytokine combinations for stimulation experiments.

9.
Overall there is an absence of protein validation. 10.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Partly

Are the conclusions drawn adequately supported by the results? Partly
Competing Interests: No competing interests were disclosed.