Dysfunctional Extracellular Matrix Remodeling Supports Perianal Fistulizing Crohn′s Disease by a Mechanoregulated Activation of the Epithelial-to-Mesenchymal Transition

Abstract

Background and Aims

Perianal fistula represents one of the most disabling manifestations of Crohn′s disease (CD) due to complete destruction of the affected mucosa, which is replaced by granulation tissue and associated with changes in tissue organization. To date, the molecular mechanisms underlying perianal fistula formation are not well defined. Here, we dissected the tissue changes in the fistula area and addressed whether a dysregulation of extracellular matrix (ECM) homeostasis can support fistula formation.

Methods

Surgical specimens from perianal fistula tissue and the surrounding region of fistulizing CD were analyzed histologically and by RNA sequencing. Genes significantly modulated were validated by real-time polymerase chain reaction, Western blot, and immunofluorescence assays. The effect of the protein product of TNF-stimulated gene-6 (TSG-6) on cell morphology, phenotype, and ECM organization was investigated with endogenous lentivirus-induced overexpression of TSG-6 in Caco-2 cells and with exogenous addition of recombinant human TSG-6 protein to primary fibroblasts from region surrounding fistula. Proliferative and migratory assays were performed.

Results

A markedly different organization of ECM was found across fistula and surrounding fistula regions with an increased expression of integrins and matrix metalloproteinases and hyaluronan (HA) staining in the fistula, associated with increased newly synthesized collagen fibers and mechanosensitive proteins. Among dysregulated genes associated with ECM, TNFAI6 (gene encoding for TSG-6) was as significantly upregulated in the fistula compared with area surrounding fistula, where it promoted the pathological formation of complexes between heavy chains from inter-alpha-inhibitor and HA responsible for the formation of a crosslinked ECM. There was a positive correlation between TNFAI6 expression and expression of mechanosensitive genes in fistula tissue. The overexpression of TSG-6 in Caco-2 cells promoted migration, epithelial–mesenchymal transition, transcription factor SNAI1, and HA synthase (HAs) levels, while in fibroblasts, isolated from the area surrounding the fistula, it promoted an activated phenotype. Moreover, the enrichment of an HA scaffold with recombinant human TSG-6 protein promoted collagen release and increase of SNAI1, ITGA4, ITGA42B, and PTK2B genes, the latter being involved in the transduction of responses to mechanical stimuli.

Conclusions

By mediating changes in the ECM organization, TSG-6 triggers the epithelial–mesenchymal transition transcription factor SNAI1 through the activation of mechanosensitive proteins. These data point to regulators of ECM as new potential targets for the treatment of CD perianal fistula.

Graphical abstract

Summary.

Perianal fistula is disabling manifestation of Crohn′s disease of unknown etiopathogenesis. This study shows that in the fistula area the pathological extracellular matrix organization mediated by TNF-stimulated gene-6 triggers epithelial–mesenchymal transition processes and paves the way for new targets for the treatment of perianal fistula.

Crohn’s disease (CD) is a chronic bowel-relapsing inflammatory disorder¹ caused by transmural inflammation, that can affect any part of the gastrointestinal tract. CD phenotype can be variable according to risk of progression and depending on disease location behavior.² Perianal fistulae represent one of the most disabling manifestations of CD and is often associated with rectal disease involvement.³ Moreover, fistula development represents a risk factor for more aggressive disease progression.⁴ CD patients with perianal fistulae experience symptoms (eg, pain and fecal incontinence) that strongly impact on quality of life, with the risk of proctectomy and permanent ostomy in the most severe cases.⁵ Current therapeutic options for CD perianal fistulae are limited and include antibiotics, immunomodulators, biologic agents, and mesenchymal stem cells,⁶ in combination with surgical management. Nevertheless, the surgical removal remains the mainstay of fistula treatment, although it does not provide long-lasting results.⁷ The etiopathogenesis of perianal fistula in CD is still unknown. The affected mucosa is destroyed and ulcerated and replaced by degranulation tissue. It appears infiltrated by inflammatory cells and myofibroblasts that, in the attempt to repair the damaged tissue, deposit collagen and other components of the extracellular matrix (ECM), with consequent tissue remodeling characterized by a high stiffness. Although some fistulae are not clearly associated with inflammatory processes, fistula development appears to be the result of the progression of inflammation through the mucosal wall, leading to the formation of unusual tunnel-like structures.⁸ Fistula formation in CD is likely due to the epithelial–mesenchymal transition (EMT) of intestinal fibroblasts impairing the ability of these cells to repair mucosal membrane damage.^9,10 Specifically, by undergoing EMT, intestinal epithelial cells gain invasive and migratory properties and lose their cell polarity and cell-cell contacts, thus acquiring features of myofibroblasts.11, 12, 13, 14, 15, 16 However, the molecular mechanisms triggering EMT still remain to be investigated. An impaired expression of matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases was observed in CD fistulae,¹⁷ supporting a dysregulated attempt of myofibroblasts to repair the damaged tissue that culminates in ECM remodeling. Of note, ECM, which is highly hydrated and composed of proteins and polysaccharides, in particular the glycosaminoglycan hyaluronan (HA), not only provides essential physical scaffolding for the cellular constituents, but also initiates crucial biochemical cues that are necessary for tissue morphogenesis, differentiation, and homeostasis.¹⁸ In fact, HA associates with proteoglycan molecules that bind growth factors and also interacts directly with cell surface receptors to elicit downstream intracellular signal transduction and participates in the regulation of gene transcription.¹⁹ The control of HA synthesis is, therefore, critical in ECM and cell biology. Alterations in the biochemical structure of the ECM also drive changes in tissue stiffness, which in turn could support the EMT process by creating a positive feedback loop.²⁰ Nevertheless, whether and how ECM disorganization contributes to EMT processes in fistula formation has not been investigated.

Here, we show, via a comparative transcriptome analysis between fistula tissue and its surrounding healthy region (named hereafter perifistula) in perianal fistulizing CD, that there is a different composition and architecture of the ECM between the 2 regions, with an increased expression of matrix-associated regulators and mechanosensitive proteins in the fistula tract compared with the perifistula. To address the functional role of ECM changes in the fistula, we focused on the increase of the protein product of TNF-stimulated gene-6 (TSG-6), an ∼35-kDa protein that, in addition to exerting strong anti-inflammatory and tissue-protective properties,²¹ serves also as a regulator of ECM organization/function and is implicated in the transduction of extracellular signaling into cells.²² By forming covalent complexes between HA and the heavy chains (HCs) of inter-alpha-inhibitor (IαI) (termed HC•HA), TSG-6 participates in altering the structural organization of HA in the ECM of fistula. Our in vitro functional assays indicated that the epithelial overexpression of TSG-6 promotes HC•TSG-6 complex production, an intermediate in the formation of HC•HA and promotes the induction of the EMT transcription factor SNAI1 through the activation of mechanosensitive proteins. Therefore, we propose that the mechanochemical signaling of the ECM in the fistula can promote the perpetuation of the disease by supporting the EMT process, thus unveiling new pathophysiological processes of fistula formation and paving the way for new potential therapeutic approaches.

Results

ECM Organization Discriminates Fistula From Perifistula Tissue

In order to investigate the overall tissue changes in perianal fistula, we performed transcriptome analysis by RNA sequencing (RNA-seq) of resected tissue collected from both fistula and nonlesioned tissue (perifistula) of CD patients (n = 8 for both groups). To ensure correct tissue identification and harvesting, fistula and perifistula tissues were first examined histologically. As shown in Figure 1A, the fistula area (outlined by white dotted lines) was surrounded by granulation tissue and characterized by the presence of inflammatory cell infiltration as also indicated by the upregulation of inflammatory cytokines (interleukin [IL]-6, tumor necrosis factor α, and transforming growth factor β1 [TGFβ1]) compared with perifistula area (Figure 1B). Both analyses revealed clear differences in the morphology and inflammatory status between the 2 analyzed areas. Accordingly, t-distributed stochastic neighbor embedding multidimensional scaling of transcriptomic data revealed extensive clustering of fistula and perifistula areas within the lower left or upper right corners of sample dispersion, respectively (Figure 1C). Out of the 8172 genes that were significantly (P < .5) differentially expressed, 3785 genes were upregulated and 4387 genes were downregulated in fistula compared with the perifistula as illustrated in the MA plot (Figure 1D) and by a hierarchically clustered heatmap (Figure 1E). Gene set enrichment analysis on the significantly differentially expressed genes showed involvement in signaling pathways associated with multiple biological processes (ie, based on the Gene Ontology categories being modulated with statistical significance) (Figure 1F). Not surprisingly, gene sets related to the signatures of EMT, immune response, and ECM were significantly enriched in the fistula compared with the perifistula tissue, consistent with the histological features. In addition, in the fistula we noticed an enhancement of genes associated with HA biosynthetic process, ECM assembly, and negative regulators of signal transduction, suggesting differences in ECM organization and cell-matrix interaction between the 2 areas. To further explore this, we looked at the (significant) differential genes within ECM/adhesion-related Gene Ontology categories. The hierarchically clustered heatmap (Figure 2A) in the fistula area revealed a significant decrease of integrins; adhesion molecules; a significant modulation of the HA pathway characterized by an increase of HA synthase enzymes, HAS1 and HAS2; and a decrease of its receptors and binding proteins (ie, CD44 and HA and proteoglycan link protein 1 [HAPLIN1]). The increase of HAS1 and HAS2 and the decrease of CD44 were further validated on a larger number of patients (n = 20 by real-time polymerase chain reaction [RT-PCR] analysis) (Figure 2B).

Figure 1.

CD-associated fistula displays a distinct transcriptome profile compared with perifistula. (A) Histological representative images underline the differences from the same patient between perifistula and fistula (outlined by white dotted lines) regions by hematoxylin and eosin. Scale bar = 100 μm. (B) Relative transcript levels of IL-6,tumor necrosis factor α, and TGFβ1. The results were normalized to GAPDH and data presented as mean ± SEM.∗P < .05, ∗∗P < .01 by Mann-Whitney test. n = 22 patients. (C) t-Distributed stochastic neighbor embedding plot visualizing cluster assignments of samples, in which samples are projected in t-distributed stochastic neighbor embedding space. (D) Volcano plot representing upregulated and downregulated genes in fistula vs perifistula samples. (E) Heatmap of differentially expressed genes in fistula and perifistula samples. (F) Gene Ontology (GO) plot–enriched pathways in fistula vs perifistula biopsies with respective P values and enrichment score (n = 8, analysis between perifistula and fistula regions from the same patient).

Figure 2.

ECM reorganization in CD-associated fistulae. (A) Heatmap highlighting the differences in genes related to ECM in fistula vs perifistula biopsies (n = 8, analysis between perifistula and fistula regions from the same patient). (B) HAS1, HAS2, and CD44 gene expression by qRT-PCR in fistula vs perifistula tract from the same patient by qRT-PCR (n = 14). (C) Sirius red staining showing collagen deposition in fistula tract from pooled samples of CD patients (n = 15) and in healthy tissue (HT) of anus from non-IBD patients (n = 7). Scale bar = 100 μm. In the right panel, the percentage of total collagen deposition is shown. (D) Representatives images of the polarized light microscopy showing the different type of collagen fibers (green-yellow and red) between fistula (n = 15) and HT (n = 7) areas and their enrichment in percentage in the right panel. (E) Heatmap representing the expression of collagen genes in fistula and perifistula areas from the same patient (n = 8). (F) Relative transcript levels by RNA-seq analysis (n = 8 patients) and (G) mRNA levels by qRT-PCR of MMP9, MMP13, MMP19, MMP25, and MMP28 in fistula and perifistula areas (n = 10 patients). Biological triplicates for each group. Data are presented as mean ± SEM. ∗P < .05, ∗∗P < .01, ∗∗∗P < .001, and ∗∗∗∗ P < .0001 by Mann-Whitney test (B–D, G).

Consistent with the modulation of collagen gene expression, Sirius red staining showed a visible accumulation of collagen deposition in the fistula region, as confirmed by the semiquantitative quantification in Figure 2C. Notably, the analysis of collagen using polarizing microscopy, which informs on the maturation state of the collagen fibers,²³ revealed an accumulation in the fistula of newly synthesized collagen fibers with green/yellow birefringence (P < .032) compared with perifistula region, which was enriched with thick collagen fibers with red birefringence (P < .0001) typical of long-term collagen fibers (Figure 2D). Hierarchical clustering of collagen genes, differentially expressed between fistula and perifistula tracts (P < .05), showed an upregulation of multiple types of collagens in the fistula compared with the perifistula; this included fibrillar collagens such as collagen III (COL3A1), collagen V (COL5A1 and COL5A2), and collagen XXIV (COL24A1); the fibrillar-associated (with interrupted triple helices, known as FACIT) collagens IX (COL9A1), XII (COL12A1), XIX (COL19A1), and XXII (COL22A1); the basement membrane (BM) collagen IV (COL4A1 and COL4A2); and the BM-associated collagen VI (COL6A1, COL6A2, and COL6A3) (Figure 2E). Conversely, collagen XVII (COL17A1), a transmembrane collagen mediating the attachment of keratinocytes to the underlying BM, was markedly downregulated (Figure 2E) along with COL4A5 and COL4A6, which form a type of collagen IV specifically associated with the intestinal epithelial BM. ECM remodeling is finely regulated by a dynamic equilibrium of its production and degradation by the activity of MMPs. In this regard, RNA-seq showed that MMP9, MMP13, MMP19, MMP25, and MMP26 were significantly upregulated in the fistula region, indicative of increased breakdown of extracellular collagens during tissue degeneration (Figure 2F and G). Overall, changes to the collagen network and increased synthesis of HA are likely to impact the tensile strength and swelling properties of the tissue, perhaps explaining the stiffening of the fistula region observed clinically.

TSG-6 Expression Increases in the Fistula Tract

To gain better insight into ECM dysfunction in CD fistula, we further analyzed the core matrisome genes as described by Naba et al.²⁴ Interestingly, within the list of significantly affected genes, as well as collagen subunits, proteoglycans, and glycoproteins involved in the architecture of ECM, we detected that TNFAIP6, encoding for the TSG-6 protein, a regulator of ECM organization/function and cell signaling, was significantly upregulated in the fistula area (Figure 3A). Its upregulation was confirmed on a larger cohort of perianal tissue sample (n = 20) at the RNA level by quantitative RT-PCR (qRT-PCR) (Figure 3B) and at the protein level by Western blot when compared with healthy region (P < .05) (Figure 3C). Although TSG-6 exerts anti-inflammatory and tissue protective functions,21 it also catalyzes a covalent modification of HA by facilitating the transfer of HC via a transesterification reaction. TSG-6 induced HC•HA complex formation were observed in several pathological conditions including inflammatory bowel disease (IBD).^21,25, 26, 27, 28 To explore if TSG-6 participates in the HA modification, first we assessed colocalization of both TSG-6 and HA (using biotin-HABP) using immunofluorescence on tissue. The immunostaining revealed that TSG-6 and HA were coexpressed by both epithelium and stromal compartments in the fistula (Figure 3D, right panel). In the perifistula, conversely, the immunoreactivity for TSG-6 was not detected in the epithelium, and scarcely in the stroma, whereas HA was everywhere abundant (Figure 3D, left panel). Then, we evaluated the formation of HC•HA, by performing a co-immunostaining of HA with an anti-IαI antibody that recognizes HCs (HC1, HC2, and HC3) and bikunin (Figure 4). In the perifistula area, IαI staining was completely negative (Figure 4, left panel), while in the fistula region it was strongly positive and colocalized with HABP (Figure 4, middle panel). No difference was found in the staining of HA, TSG-6, and IαI between the perifistula region and healthy tissue collected from non-IBD patients (Figure 4, right panel), indicating no precommitted alterations of ECM organization in the area surrounding the fistula.

Figure 3.

TSG-6 is upregulated in fistula compared with perifistula regions and mediates pathological HC•HA complex formation. (A) Heatmap representing the expression of matrisome core genes (n = 8 patients). (B) Relative transcript levels of TNFAIP6 in fistula and perifistula areas from the same patients (n = 20). (C) Representative Western blot analysis of TSG-6 and vinculin as control, in whole cell lysate from fistula and perifistula areas. Below, TSG-6 protein signal intensity quantification relative to vinculin (n = 5 patients for each group). (D) Immunofluorescence staining of TSG-6 (red) and HA (green) in fistula area (left panel) and in fistula tract (right panel) from the same patient. DAPI = nuclei in blue. Scale bar = 100 μm. Data are presented as mean ± SEM. ∗P < .05, ∗∗P < .01 by Mann-Whitney test.

Figure 4.

Immunostaining for HA and Iαl expression in fistula and healthy anal tissue. Representative images of immunofluorescence staining of Iαl or TSG-6 (red) and HA (green) in perifistula (left panel) and fistula (middle panel) regions from the same CD patient and in healthy anal tissue of non-IBD patients (right panel). DAPI = nuclei in blue. Scale bar = 50 μm.

Overall, these results indicate that the upregulation of TSG-6 in the fistula tract correlated with the accumulation of HA and likely participates in the formation of pathological HC•HA complexes.

TSG-6 Drives EMT in Caco-2 Cells

A previous study demonstrated that a transient upregulation of TSG-6 in renal tubular epithelial cells promotes the EMT process by modulating HA assembly.²⁹ In order to investigate whether the marked expression of TSG-6 in epithelial cells plays a role in the EMT process associated with fistula formation, we overexpressed TSG-6 by lentiviral transduction in epithelial Caco-2 cells, which do not express TSG-6 under normal conditions (Figure 5A). After 72 hours of transduction, cells were collected and tested for EMT-related genes, proliferation, and migration.

Figure 5.

TSG-6 induces Snail and enhances the migratory capacity of Caco-2 cells. (A) Relative transcript levels of TNFAIP6 (which encodes the TSG-6 protein) in control (EV) and TSG-6 OE Caco-2 cells. Data are representative of 3 independent experiments, with biological triplicates for each group. (B) Western blot analysis of Caco-2 cells with EV and TSG-6 OE cells. (C) Western blot analysis of Caco-2 cell supernatants pretreated with 0.1 M NaOH or with 0.05 U chondroitinase ABC lyase. Shown are that 40 kDa and 150 kDa are the molecular weights expected for TSG-6 free molecules and the complexed TSG-6 forms, respectively. Data are representative of 3 independent experiments, with biological triplicates for each group. (D) Expression levels of CDH1 (which encodes E-cadherin) and SNAI1 (which encodes Snail) in fistula and perifistula regions from the same patients by RNA-seq analysis (left panel) (n = 8) and qRT-PCR (right panel) (n = 22). (E) Relative transcript levels of CDH1 and SNAI1 in control (EV) and TSG-6 OE Caco-2 cells. (F) [³H]-thymidine–based proliferation assay of control (EV) or TSG-6 OE Caco-2 cells. (G) Bright-field images of wound healing assay performed in control or TSG-6 OE Caco-2 cells for 24 and 48 hours, and quantification of the wound closure respective to time zero. Representative images of 3 independent experiments, with biological triplicates for each group. Data are presented as mean ± SEM. ∗P < .05, ∗∗P < .01, ∗∗∗P < .001, and ∗∗∗∗ P < .0001 by Mann-Whitney test (A, C, E) and 1-way analysis of variance, with Tukey’s multiple comparisons (D, G).

TSG-6 expression was validated by quantitative RT-PCR, which showed a 60-fold-increase of expression in the TSG-6-overexpressing (OE) cells compared with cells that had integrated the expression cassette not containing the TSG-6 complementary DNA (cDNA) sequence (empty vector [EV], P < .001) (Figure 5A). The increased levels of TSG-6 were detected at an apparent molecular weight (MW) of ∼40 kDa (close to that expected for the free protein) by Western blot analysis on cell lysates (Figure 5B). To test whether TSG-6 overexpression promoted HA•HC formation in the extracellular compartment, we analyzed the TSG-6 species present in the cell culture supernatants. While the levels of free TSG-6 were low (but detectable) in TSG-6 OE cells, there was an intense band at an apparent MW of ∼150 kDa that was greatly reduced in intensity by NaOH treatment but not by chondroitinase ABC lyase (Figure 5C). This is consistent with this ∼150-kDa species corresponding to a covalent HC•TSG-6 complex, in which the HC and TSG-6 proteins are linked via an ester bond, as has been described previously.21,37 Importantly, HC•TSG-6 is a well-established intermediate in HC•HA formation.

We next assessed the expression of 2 EMT markers, E-cadherin (CDH1), an epithelial marker, significantly decreased, and Snail (SNAI1), a mesenchymal marker significantly enriched in the fistula tract as shown by RNA-seq (Figure 5D, left panel) and then validated on a larger cohort of perianal tissue samples by qRT-PCR (Figure 5D, right panel). Accordingly, SNAI1 expression increased in TSG-6 OE cells compared with EV cells (P < .05), while no difference in CDH1 expression was found between 2 cell populations (Figure 5E). Moreover, the overexpression of TSG-6 significantly attenuated the proliferative capacity of Caco-2 cells 48 hours after plating (Figure 5F) and enhanced the migratory ability of Caco-2 cells at 24 and 48 hours (P = .0003 and P = .0081, respectively) (Figure 5H and G). To determine if the biological response of Caco-2 cells to TSG-6 was dose-responsive, we conducted the migration assay in the presence of a low (5 ng) and high (10 ng) concentration of recombinant human TSG-6 protein (rhTSG-6). Both concentrations significantly promoted cell migration 24 and 48 hours postscratch (Figure 6). Collectively, these data suggested that the overexpression of TSG-6 promotes the formation of HC•TSG-6 complexes in the extracellular compartment and participates in cellular processes typically observed in cells undergoing EMT by promoting epithelial cell transdifferentiation and migration but reducing proliferation.

Figure 6.

TSG-6 induces migration in Caco-2 cells in a dose-dependent manner. Bright-field images of migration assay performed in Caco-2 cells for 24 and 48 hours in presence of low (5 ng) and high (10 ng) concentrations of rhTSG-6 and quantification of the wound closure respective to time zero. Representative images of 3 independent experiments, with biological triplicates for each group. Data are presented as mean ± SEM. ∗P < .05, ∗∗P < .01, ∗∗∗P < .001, and ∗∗∗∗ P < .0001 by 1-way analysis of variance, with Tukey’s multiple comparisons.

TSG-6 Is Required in the EMT

To further corroborate TSG-6 as a mesenchymal gene associated to the mesenchymal transition process, we analyzed its expression during the reprogramming process of mesenchymal–epithelial transition (MET), when mesenchymal genes are repressed through the generation of human induced pluripotent stem cell (iPSC)–derived intestinal organoids (iHOs) from adult fibroblasts. To this end, commercial dermal fibroblasts isolated from a skin of an female healthy donor (HD) were reprogrammed by expression of the transcription factors of OCT4, SOX2, KLF4, and c-MYC.^30,31 iPSC-like colonies were individually picked 4 weeks later, expanded, and characterized. The clones showed an iPSC-like colony morphology, a normal 46,XX karyotype, and stemness markers and were able to differentiate in vitro, as shown by positivity to all 3 germ-layer markers (Figure 7A–D). HD-iPSCs were forced to differentiate into iHOs,³² and iPSC-derived iHOs were analyzed after 30 days in Matrigel (BD Biosciences, San Jose, CA) (Figure 8A). They showed a crypt-villus architecture of the intestine (Figure 8B) and the presence of typical cell types of the intestinal tissue, including cytokeratin 19, epithelial cellular adhesion molecule, E-cadherin, and structural protein of tight junctions zona occludens-1 indicating the presence of apical tight junctions facing the lumen of the organoid (Figure 8C–E). They also showed ubiquitous expression of enterocyte marker villin, with specific subset of cells expressing mucin 2 and chromogranin A, markers associated with the goblet and enteroendocrine cells, respectively, confirming the correct differentiation of iPSCs into iHOs (Figure 8F–H). According with Caco-2 results, TSG-6 expression was scarce in iHOs (Figure 8I–J). TSG-6 messenger RNA (mRNA) expression level was highly expressed in correlation with the levels of the transcriptional factor SNAI1 on fibroblasts, and both genes were drastically reduced as well as vimentin, an mesenchymal marker, when iHOs were well established (Figure 8L), as demonstrated by a concomitant upregulation of key epithelial genes such as CDH1 and LGR5, recognized as markers of differentiated epithelial cells and adult stem cells, respectively. Overall, these data pointed out TSG-6 as a gene required/involved in the conversion of epithelial cells into mesenchymal cells independently of tissue of origin.

Figure 7.

Characterization of HD-iPSCs. (A) Representative images of iPSC-like colony morphology of iPSC cells cultured under feeder-free conditions. Scale bar = 50 μm. (B) HD-iPSC exhibited a normal diploid karyotype (46,XX). (C) Immunostaining for stemness markers (green): OCT4, NANOG, and SOX-2. Scale bar = 15 μm. (D) Immunostaining for markers of the 3 germ layers (green): endoderm (GATA4), mesoderm (smooth muscle actin SMA), and ectoderm (NESTIN). Nuclei are counterstained with DAPI (blue). Scale bar = 15 μm.

Figure 8.

TSG-6 expression in iHOs derived from human dermal iPSCs. (A) Schematic reprogramming process of MET, when mesenchymal genes are repressed through the generation of human iPSC-derived iHOs from adult fibroblasts. (B) Representative image of cryosection iPSC-derived iHO (C, D) immunostained for the presence of cytokeratin 19 and villin indicating the presence of a brush border and columnar epithelial cells; (E–H) zona occludens-1 and epithelial cellular adhesion molecule indicating the presence of the tight junction facing the lumen of the organoid; villin for enterocytes; anti-mucin-2 indicating the presence of goblet cells; and chromogranin A indicating the presence of enteroendocrine cells; and (I) TSG-6. DAPI = nuclei in blue. Scale bars = 100 and 20 μm. (J) Relative transcript levels of TNFAIP6 encoding for TSG-6, SNAI1 encoding for Snail, Vimentin, CDH1 encoding E-cadherin, and LGR5 in fibroblasts and iHOs by qRT-PCR. Data are presented as mean ± SEM. ∗P < .05, ∗∗∗P < .001 by Mann-Whitney test (n = 3).

TSG-6 Overexpression Impacts Fibroblast Phenotype

Upon fistula formation, the transdifferentiation of epithelial cells into mesenchymal cells compensates for the pathological inhibition of the migration of the tissue-resident fibroblasts.^9,10 During this process, transdifferentiated cells gain migratory properties and a high capacity to synthesize ECM components. Interestingly, resident fistula-derived fibroblasts (FFs) expressed higher amounts of TSG-6 compared with those isolated from perifistula (perifistula-derived fibroblasts [PFFs]) (P < .01) together with an increase in SNAI1 (a transcriptional factor that controls fibroblast migration) expression (Figure 9A). To determine if the overexpression of TSG-6 induced SNAI1 upregulation, we overexpressed TSG-6 in PFF and quantified its mRNA expression levels (Figure 9B). SNAI1 mRNA levels increased in PFF TSG-6 OE cells compared with EV control cells (Figure 9B). In addition, TSG-6 OE PFF cells displayed elevated expression of HAS2, and although not significant, expression of ACTA2 (commonly referred to as alpha-smooth muscle actin), a marker of fibroblast activation (Figure 9C). These molecular changes were also accompanied by the acquisition of different cellular morphology. While PFF EV control cells appeared spindle shaped with cytoplasmic bipolar projections, TSG-6 OE PFF cells showed enlarged cytoplasm with plump nuclei characterized by finely dispersed chromatin and conspicuous nucleoli (Figure 9D [left panel] and E) indicative of a transcriptionally active status. Interestingly, the overexpression of TSG-6 in the PFFs drove morphological changes that resembled those of the FFs, as also evidenced by F-actin immunolabeling (Figure 9D, right panel).

Figure 9.

TSG-6 overexpression in PFFs triggers the acquisition of fistula-like features. (A) Relative transcript levels of TNFAIP6 (which encodes for TSG-6 protein) and SNAI1 (which encodes for Snail) in PFFs or FFs isolated from the same patient (n = 5). (B) Relative transcript levels of TNFAIP6 and SNAI1 in PFF control (EV) and overexpressing TSG-6. Data are representative of 2 independent experiments, with biological triplicates for each group (n = 3). (C) Relative transcript levels of HAS2 (encoding for HA synthase 2) and ACTA2 (encoding for α-smooth muscle actin) in control (EV) or TSG-6 OE perifistula-derived human fibroblasts. (D) Bright-field images (left panel) and F-actin immunolabeling (right panel) of human fibroblasts derived from fistula and perifistula areas of the same patient and of perifistula fibroblasts overexpressing TSG-6. Data are representative of 2 independent experiments, with biological triplicates for each group (n = 3). DAPI = nuclei in blue. Scale bar = 30 μm. (E) Nuclei surface (cm²) of human fibroblasts derived was quantified by ImageJ. Data are presented as mean ± SEM. ∗P < .05, ∗∗P < .01, and ∗∗∗P < .001 by Mann-Whitney test (A, D, E) and 1-way analysis of variance (F, H).

Changes in ECM Organization Drive the Activation of Mechanosensors

Changes of ECM structure can translate into the activation of intracellular responses via collagen fibers and focal adhesions through a process defined as mechanotransduction, which in turn triggers the activation of several nuclear transcription factors.³³ Our RNA-seq highlighted a significant upregulation of several genes that mediate mechanotransduction-mediated signaling pathways including integrins and protein tyrosine kinases in the fistula compared with the perifistula tissue (Figure 10A). The paired comparison analysis of RNA-seq data between perifistula and fistula areas from the same patient revealed ITGA4, ITGA2B, ITGB7, FUT8, NEDD9, and PTK2B highly upregulated in the fistula (Figure 10B). The correlation heatmap depicting the measure of dependence between multiple variables, showed a positive correlation of TSG-6 (TNFAIP6) with ITGA4 and ITGA2B, genes encoding for integrins, and protein tyrosine kinase 2B gene (PTK2B) (P < .05) (Figure 10C).

Figure 10.

TSG-6 expression in the fistula positively correlates with mediators of mechanotransduction. (A) Heatmap representing the expression of genes related to mechanotransduction in fistula and perifistula biopsies. (B) Paired analysis of mechanosensitive proteins in fistula and perifistula areas. Gene expression of ITGA2, ITGA2B, ITGA4, PTK2B, FUT8, ITGAL, ITGB7, ITGA3, and NEDD9. For all qRT-PCR results, data reflect mean ± SEM from 3 biological replicates; the results were normalized to GAPDH and P values were measured by paired t test (n = 10 patients for each group). (C) Correlation matrix displaying gene-gene Pearson correlation coefficients. Ellipses show gene expression value distributions with red and blue ellipses representing positive and negative correlations, respectively. Cells marked with X represent not significant correlations. n = 8 patients for each group.

To explore if HA accumulation and overexpression of TSG-6 can impact ECM organization promoting the activation of mechanosensors in fibroblasts, we plated PFFs on top of a HA-based HyStem hydrogel scaffold, in the absence (nonsupplemented) and presence (supplemented) of rhTSG-6 (Figure 11A) at a concentration that is known to crosslink and stiffen HA.^22,34 After 24 hours at 37°C, the cell culture supernatants were analyzed for collagen content and the cells for ITGA4, ITGA2B, PTK2B, and SNAI1 expression. Interestingly, when cultured on the scaffold, PFFs significantly increased the collagen release (P < .05), and this was elevated further in the presence of rhTSG-6 (P < .001) (Figure 11B). Because IL-6 is a potent inducer of collagen production in fibroblasts,³⁵ we checked its levels in the cell culture supernatants. Accordingly, IL-6 levels increased in correlation with collagen amounts in plated PFF cells on both scaffolds (Figure 11C). Interestingly, seeding cells on the nonsupplemented scaffold promoted the transcription of SNAI1, ITGA2B, ITGA4, and PTK2B compared with cells plated without scaffold (no scaffold) (Figure 11D–G). However, the enrichment of the scaffold with rhTSG-6 significantly augmented the expression of all these genes compared with the nonsupplemented scaffold or on the plate without scaffold (P < .005), to a level similar to the FFs (Figure 11D–G). Similar results were obtained with non-IBD healthy fibroblasts (data not shown), indicating that the remodeling of ECM contributes to the activation of mechanosensors that in turn activate nuclear transcription factors.

Figure 11.

The ECM reorganization mediated by HA•TSG-6 complex promotes the activation of mechanosensors in PFFs. (A) Schematic representation of experimental procedure, in which PFFs were plated on HA scaffold ± rhTSG-6. (B) Quantification of collagen deposition in PFFs plated on scaffold ± TSG-6. (C) IL-6 (pg/mL) levels in PFFs plated on scaffold ± TSG-6. (D–G) Relative transcript level of SNAI1 (encoding for Snail), ITGA2B (encoding for integrin alpha 2B), ITGA4 (encoding for integrin alpha-4), and PTK2 (encoding for protein tyrosine kinase 2) in FFs vs PFFs plated or not on scaffold ± rhTSG-6. Data are presented as mean ± SEM from 3 independent experiments, with biological triplicates for each group (n = 3). ∗P < .05, ∗∗P < .01, ∗∗∗P < .001, and ∗∗∗∗ P < .0001 by 1-way analysis of variance with Tukey’s multiple comparison.

Discussion

In this study, we provided evidence that an altered organization of the ECM in the perianal fistula of CD, exhibiting high collagen deposition, increased HA synthesis, and the presence of HA-binding proteoglycans/proteins, may participate in sustaining the pathological epithelial-to-mesenchymal transition events underlying perianal fistula formation. A dysregulated EMT is associated with a variety of pathologies including fibrotic diseases,³⁶ cancer,³⁷ and fistula,¹⁵ in which it intervenes in replacing the impaired activity of resident fibroblasts to repair intestinal damage.^9,10 Several studies reported the increased levels of IL-13 and TGFβ as the critical cues in triggering EMT in the perianal fistula of CD patients, thus supporting the chronic inflammatory processes as important in promoting fistula formation. The perianal fistula area, in fact, is infiltrated by inflammatory cells including T cells, B cells, and macrophages that sustain these EMT events by releasing a plethora of proinflammatory mediators.⁸ Accordingly, our data showed evidence of an increased immune response in the fistula area and upregulation of genes associated with the EMT processes, including IL-13 (Figure 12). To date, standard medical management has been traditionally based on the use of antibiotics and immunomodulators, which includes the use of biologic agents⁶ targeting the massive immune infiltration. However, incomplete (or a lack of) response to medication is frequent in these patients, for whom surgical removal of the fistula becomes necessary with disabling consequences.⁶ This suggests, therefore, that dampening inflammation is not enough to restore mucosal homeostasis and to arrest the EMT processes in perianal fistulizing CD patients.

Figure 12.

Relative transcript levels of IL-13 by qRT-PCR. Data are presented as mean ± SEM from 3 biological replicates; the results were normalized to GAPDH, and P values were measured by paired t test (n = 10 patients for each group).

However, what other factors contribute to these pathological processes remain unknown. Here, we showed for the first time that changes in ECM organization likely promote mechanical signals supporting EMT. ECM remodeling is essential in mucosal repair consequent to an inflammatory response and granulation tissue formation, and in maintaining a correct re-epithelialization of the tissue; for example, the polarization along the apical-basal axis of the epithelium is mediated by the attachment to the basement membranes. Several collagens such as COL17A1, COL4A5, and COL4A6 that stabilize the attachment of intestinal epithelial cells to BM were markedly downregulated in the fistula area. The loss of cell-BM contacts is a first hallmark of EMT.³⁷ Therefore, EMT is highly influenced and controlled by the surrounding ECM. We found deposits of newly synthetized fibril-forming interstitial collagens (mainly type III) and BM collagens (type IV) to be increased in the fistula; these collagens consist of multiple isoforms that support cell adhesion and maintain normal tissue architecture and function.³⁸ The expression of MMP9, MMP13, MMP19, MMP25, and MMP26 critical to regulate ECM degradation are increased. Moreover, the fistula area is enriched in the polysaccharide HA that is likely associated with the formation of a pathogenic matrix. HA is an essential structural component of the ECM that, along with bound chondroitin sulfate proteoglycans and link proteins (such as HAPLN1), regulates tissue turgor and thus contributes to the biomechanical properties of tissues. HA biosynthesis is dysregulated in inflammation and in response to tissue damage, and invasion by tumor cells or pathogens^19,39; it is upregulated in many pathological conditions, including IBD,^40,41 and accumulates in the lungs of patients with influenza A and SARS-CoV-2 infection,^42,43 which in the latter is driven by IL-13. Here, 2 enzymes that mediate HA synthesis, HAS1 and HAS2, were found to be increased in perianal fistula of CD patients, whereas HAPLN1 was decreased, indicative of ECM remodeling. TSG-6, which is often induced in inflammation,^21,44,45 was also upregulated in the fistula area. TSG-6 is a multifunctional, ∼35-kDa, secreted protein that interacts with a wide range of ligands and has been found to have anti-inflammatory properties in many different contexts,²¹ including in models of colitis,^46,47 and also has tissue protective effects, for example, promoting wound healing.^21,48, 49, 50, 51 In a previous study,⁵² we demonstrated that TSG-6 plays a crucial role in the mucosal healing processes mediated by HA. One of TSG-6’s functions is to act as an enzyme in the covalent transfer of HCs from IαI onto HA, where HC•TSG-6 complexes act as intermediates in this process.^21,53 The TSG-6–mediated production of HC•HA leads to the formation of a crosslinked, expanded, HA matrix that is essential for ovulation and fertilization,^21,54 and for gut asymmetry during development⁵⁵; the crosslinking, and thus stabilization of the matrix, likely occurs through HC-HC interactions⁵⁶ and via binding of HCs to pentraxin-3.⁵⁷ Moreover, HC•HA complexes (containing pentraxin-3) are thought to be responsible for the potent anti-inflammatory, antifibrotic, and antiangiogenic properties of amniotic membrane.²⁶ The formation of HC•HA also occurs in tissues during inflammation (ie, at sites where TSG-6 is expressed and where IαI has ingressed from the blood circulation).²¹ Whether the HC•HA complexes are protective or drive pathology is likely dependent on their compositions (ie, which HCs are covalently attached to HA and whether accessory molecules, such as pentraxin-3, are bound) and the context of their formation. For example, in lung conditions, such as cystic fibrosis,⁵⁸ airway hyperresponsiveness,⁵⁹ and influenza A infection,⁴³ HC•HA contributes to a pathogenic matrix, with increased numbers of associated immune cells. The finding here that there is increased staining for HA, IαI, and TSG-6 in fistula strongly suggest that HC•HA complexes are being formed, such that crosslinking interactions have the potential to alter the mechanical properties of the tissue (eg, in conjunction with collagen deposition). Furthermore, TSG-6 in absence of IαI⁶⁰ has been found to have a dramatic effect on HA structure, collapsing and stiffening the HA network.^21,34 Thus, the relative amounts of TSG-6 and IαI could tune the mechanical properties of the HA matrix.

TSG-6–mediated crosslinking of HA has been found to enhance its interaction with CD44 on leukocytes^60,61 and may be responsible for some of TSG-6’s anti-inflammatory effects^52,62,63 HC•HA has also been implicated in supporting immune cell adhesion, which in this case it is likely associated with pathology.^25,43,58,59 Thus, given the different activities for HA/TSG-6 and HC•HA complexes, it is difficult to interpret the decrease in expression of CD44 in the fistula area seen in the present study.

Interestingly, TSG-6 overexpression in Caco-2 cells led to HC•TSG-6 complex formation in the culture supernatants. Although the experiments were performed in serum-free conditions without adding exogenous IαI, there was clearly sufficient IαI present to lead to the formation of HC•TSG-6, as this species was much more abundant than the free TSG-6 protein; in this regard, it is very difficult to remove IαI because it tightly adheres to the cell surface. Given that it is well established that HA is synthesized by epithelial cells, it seems likely that HC•HA complexes will be formed, in which these, for example, have been implicated in the TGFβ–mediated differentiation of fibroblasts to myofibroblasts.⁶⁴ In this regard, TSG-6 overexpression in Caco-2 cell significantly induced the expression of the EMT transcription factor Snail. Accordingly, TSG-6 OE cells exhibited high migratory capacities but reduced the proliferative capacity of Caco-2 cells, an effect also observed in human endothelial cells.⁶⁵ The overexpression of TSG-6 in the PFFs promoted their activation. Indeed, the cells displayed increased levels of ACTA-2, gene encoding for the α-smooth muscle actin marker of myofibroblast activation and of SNAI1, and morphological changes resembling those of FFs. The involvement of TSG-6 in inducing these alterations in the cellular morphology and cytoskeletal architecture is also supported by our previous work demonstrating that the enrichment of ECM with TSG-6 in mesenchymal stem cells promoted the reorganization and polymerization of the actin cytoskeleton.²² The marked reduction of TSG-6 in the reverse process from mesenchymal to epithelial reprogramming observed on iPSC-derived iHOs further corroborated it as a mesenchymal gene involved in the transition process of the fibroblasts independently from tissue of origin. As discussed previously, it is reasonable to assume that the modification of HA by TSG-6 induces changes in the biomechanical properties of the ECM, which in turn propagates mechanotransduction via actin cytoskeleton modifications and activation of intracellular signaling pathways.^33,40 In support of this hypothesis, the enrichment of the HA scaffold with TSG-6 activated some integrins that serve as mechanosensitive proteins in PFFs.

Cell adhesion to the ECM is, indeed, regulated by integrins: transmembrane receptors that interact with matrix components (eg, fibronectin and vitronectin) via their extracellular domains and also interact with actin via their cytoplasmic. Several of these integrins, in particular, integrin alpha 4 (ITGA4) and integrin alpha 4 subunit 2b (ITGA42B) and the associated protein-tyrosine kinase 2-beta (PTK2B) that are all involved in activating pathways leading to actin cytoskeletal rearrangements and transcriptional alterations,⁶⁶ were upregulated in the fistula area. Moreover, their expression was induced after the enrichment of ECM with HA and TSG-6 in a quantity demonstrated to make the HC•HA complex formation.³⁴ Nevertheless, HA alone was enough to stimulate collagen deposition, and the supplementation with rhTSG-6 further enhanced this finding. Although further studies are necessary to gain insight into the molecular mechanisms involved, it is reasonable to hypothesize that upon initiation of inflammation the intestinal mucosa responds to damage by synthesizing HA, TSG-6, and other ECM components in an attempt to activate tissue repair. However, pathological HC•HA complex formation impacts the ECM composition and function activating mechanoregulated signaling pathways that lead to the promotion of EMT.²⁰

HCs of IαI interact with integrin ligands and thus could modulate the structure of matrix in a variety of ways.⁵⁶ Moreover, HC•HA complexes have been found to directly participate in fibroblast to myofibroblast differentiation⁶⁴ and may regulate the bioavailability of growth factors involved in these events such as TGFβ.⁵⁶ Together, this suggests that in the fistula once the dysfunctional ECM is established, mechanical properties and composition of ECM drive EMT. In turn, EMT sustains changes in the mechanics and composition of the ECM fueling EMT independently of inflammatory processes, thus creating a feedback loop between ECM remodeling and EMT processes. This scenario paves the way for a new management of perianal fistula in CD patients. Indeed, the therapeutic approach may evolve from anti-inflammatory therapies and use of antibiotics to a combination treatment strategy of conventional drugs with regulators of ECM. Although the biological drugs (eg, infliximab and adalimumab) have as primary targets the blocking of the inflammatory cascade, there is evidence they also alter the composition and organization of the ECM⁶⁷ by modulating MMPs expression.^68,69 However, these effects on the ECM are rather weak and unspecific and do not take into account the overall changes of ECM mechanical properties. Indeed, a significant proportion of patients only partially respond to these therapies meaning that no single treatment can successfully manage fistula healing. A recent consensus guideline has been designed to simplify the classification of perianal fistulae and to include proactive medical and surgical approaches.⁷⁰ Although there is no histological stratification of the perianal fistula, ECM analysis could inform the choice of combined therapy (ie, to stop ECM-dependent EMT events), and may also allow the design of new strategies for generating antibodies against ECM molecules and using them for novel combined therapies.

Materials and Methods

Surgical Human Tissue Collection

Perianal fistulae involving the anal tract, classified using the Parks criteria based on their relationship to the external and internal anal sphincters, were analyzed in 20 patients diagnosed with CD (Table 1). CD was diagnosed based on clinical, biological, endoscopic, and histological criteria (Crohn’s Disease Activity Index score). Surgical specimens from fistula and the nonlesional area named as perifistula, approximately 3 cm from the fistula, were collected from all patients after giving written informed consent. Surgical specimens from healthy mucosa of patients affected from cancer in the upper part of the anus near the rectum were used as a control group [non-IBD] (n = 7). This study was approved by the ethical committee of the Humanitas Clinical and Research Center (no. 2406/2019) and conducted following national and international guidelines.

Table 1.

Clinical Profile of Patients

	CD ( n = 22)	Non-IBD (n = 7)
Sex
Male	16	4
Female	6	3
Age, y	33 (22–62)	45 (20-68)
CD location
Ileum	13	—
Colon	6	—
Ileocolon	2	—
Rectum	—	7
Upper	1	—
Previous therapies
Azathioprine	4	—
Infliximab	7	—
Adalimumab	7	—
Vedolizumab	4	—
Systemic corticosteroids	11	—
Smoking status
Current smoker	8	—
Never smoked	9	—
Past smoker	2	—
Previous surgeries
Abdominal IBD-related surgeries	9	—
Proctological surgeries	14	—
Familiarity for IBD	4	—

Values are n or median (range).

CD, Crohn’s disease; IBD, inflammatory bowel disease.

Cell Culture

The Caco-2 cell line (passages between 18 and 24) (HTB-37; ATCC, Manassas, VA) was cultured in Dulbecco’s modified Eagle medium (DMEM) (Gibco, Grand Island, NY) supplemented with 10% (v/v) fetal calf serum (FCS) (EuroClone, Pero, Italy), 1 mmol/L L-glutamine (Lonza, Basel, Switzerland), 1 mmol/L sodium pyruvate (EuroClone), 0.1 mmol/L nonessential amino acids (Lonza), and 100 U/mL antibiotics (penicillin/streptomycin with amphotericin; Lonza) at 37°C in 5% CO₂. Before performing the experiments, the cells were tested for mycoplasma.

Human dermal fibroblasts from a normal donor (Thermo Fisher Scientific, Waltham, MA) were maintained in high-glucose DMEM, supplemented with 10% fetal bovine serum (FBS) (Lonza) at 37°C in humidified air with 5% CO₂.

Human iPSCs were cultured in feeder-free conditions, using StemFlex medium (Thermo Fisher Scientific) on vitronectin-coated 6-well plates (BD Biosciences) at 37°C in humidified air with 5% CO₂.

Isolation of Primary Intestinal Fibroblasts

Surgical specimens were washed with Hank’s Balanced Salt Solution (HBSS) (Gibco) containing 10% FCS, 1% antibiotics, and 20 mM HEPES (pH 7.3) (EuroClone) and cut into small pieces. The epithelial layer was removed by incubating the intestinal fragments with Hepes-buffered HBSS containing 2 mM EDTA solution, at 37°C, for 45 minutes, under gentle agitation. Digestion with HBSS + 5% (v/v) FCS, 5 mM CaCl₂, 1 mg/mL collagenase/dispase (Roche, Basel, Switzerland), and 40 μg/mL DNase (Roche Diagnostics) at 37°C for 30 minutes was then performed. The digested fragments were further disrupted through 100-μm filters (BD falcon) with a 1-mL syringe plunger and washed with 10 mL RPMI medium supplemented with 10% (v/v) FCS and 20 mM Hepes pH 7.3. The single-cell suspension was then filtered through 70-μm filters. Cells were cultured in Gibco M106 supplemented with 10% (v/v) Low Serum Growth Supplement (Thermo Fisher Scientific) and 1% (w/v) antibiotics (penicillin/streptomycin with amphotericin) at 37°C, 5% CO₂, and used between passages 2 and 5. All cell lines were tested for mycoplasma.

Cell Reprogramming

Fibroblasts were reprogrammed by an overnight infection at an multiplicity of infection of 1 in the presence of 4 ng/mL Polybrene (Sigma-Aldrich, St Louis, MO) with a third-generation lentiviral vector described by Sommer et al⁷¹ 19096035 carrying the OCT4, SOX-2, and KLF4 reprogramming factors. The following day, the medium was replaced, and after 48 hours cells were seeded onto mouse embryonic fibroblast feeder layer with ES medium, which was changed daily. iPSC-like colonies were individually picked 4 weeks later and cultured in feeder-free conditions.

Chromosome Preparation and Analysis

Chromosome analysis was done on slide preparations of cells grown on coverslips. Briefly, cell cultures were treated with KaryoMAX Colcemid solution (Thermo Fisher Scientific) at a final concentration of 0.1 μg/mL for 2 hours at 37°C. After hypotonic treatment with 0.075 M KCl and fixation in methanol/acetic acid (3:1 v/v), the slides were air-dried and mounted in Eukitt (Sigma-Aldrich). Chromosome counts and karyotype analyses were done on metaphases stained with VECTASHIELD mounting medium with DAPI (Vector Laboratories, Burlingame, CA) for Q banding.

Images were captured using an Olympus BX61 research microscope (Olympus, Tokyo, Japan) equipped with a cooled charge-coupled device (CCD) camera and analyzed with Applied Imaging software CytoVision (V 4.5.4; master system with mouse karyotyping; CytoVision, Praetoria, South Africa). At least 10 karyotypes were analyzed.

Embryoid Body Formation

Semi-confluent human iPSCs in 1 well of a 6-well plate were dissociated by incubation with 2 mg/mL dispase (STEMCELL Technologies, Vancouver, BC, Canada) at 37°C for 2–3 minutes, or until the colonies began to lift off of the plate. The plates were gently shaken to help dislodge the colonies. The dispase solution was diluted in phosphate-buffered saline (PBS) and the suspension was transferred to a 15-mL centrifuge tube. The cell aggregates were allowed to settle by gravity, and the dispase solution was gently aspirated. The aggregates were washed 3 more times in PBS before being resuspended in 4 mL of Essential 6 medium supplemented with 5 μM Y-27632 ROCK inhibitor. The suspension was gently transferred to each well of a 6-well ultra-low attachment plate (Corning Life Sciences, Acton, MA). After 1 week in suspension, embryoid body formations were transferred to Matrigel-coated dishes and cultured in Essential 6 medium (Thermo Fisher Scientific) for an additional 2–3 weeks. Then, the cells were stained with antibodies against markers of all 3 embryonic germline layers and analyzed by immunofluorescence.

iHO Generation

The differentiation experiments will be performed essentially as described by Lees et al. 2019.³² Definitive Endoderm differentiation starts by replacing StemFlex medium (Thermo Fisher Scientific) with StemFlex medium supplemented with 10 ng/mL activin A (ActA) and 12 ng/mL basic fibroblast growth factor (bFGF). After 2 days, the medium is replaced with StemFlex medium supplemented with 100 ng/mL ActA, 100 ng/mL bFGF, 10 ng/mL BMP-4, 10 μM LY294002, and 3 μM CHIR99021. The following day, the medium is replaced with RPMI-1640 medium (Lonza) supplemented with 1% serum-free supplement B27 (Thermo Fisher Scientific), 100 ng/mL ActA, and 100 ng/mL bFGF. The differentiation is continued with hindgut induction for 4 days in RPMI/serum-free supplement B27 containing 6 μM CHIR990212 and 3 μM retinoic acid. For 3-dimensional organoid formation, cells will be collected at day 10 and embedded in a small volume of a basement membrane matrix of Matrigel Growth Factor Reduced (BD Biosciences) and seeded into a 60-μL drop/well of 24-well plate. After drop solidification, Advanced DMEM/F12 containing N2 supplement (Thermo Fisher Scientific), B27 serum-free supplement and supplemented with 500 ng/mL R-spondin1, 100 ng/mL noggin, 100 ng/mL epidermal growth factor, 3 μM CHIR990212, 2.5 mM prostaglandin E2, and 10 mM Y-27632 was added. The cytokines and supplements used are shown in Table 1. Medium was changed every 4 days, and organoids was passaged mechanically every 7–10 days. For media change alone Y-27632 was not added; it is only required when seeding/splitting. Organoid cultures was analyzed after 20–30 days in Matrigel Growth Factor Reduced (d30/40 from human HD-iPSCs).

Culture of Primary Fibroblasts on Scaffolds

HyStem-C Cell Culture HA scaffold, prepared following the manufacturer‘s instructions (HYSC010; Sigma-Aldrich), was supplemented with normal human serum (5% v/v) and rhTSG-6 (R&D Systems) at a final concentration of 0.9 μM.²² Thereafter, PFFs at passage 3 were plated on top of the HA scaffold at a density of 8 × 10⁴ cells/cm² and cultured for 24 hours.

Collagen Measurement

Primary fibroblasts were plated in 6-well plates with 2 mL of total medium. Once the cells reached 90% of confluence, cell supernatants were collected by centrifuging at 400 g for 5 minutes and stored at –80°C. Total soluble collagen was quantified in cell supernatants determined fluorometrically using the Sircol soluble collagen assay (Biocolor, Newtownabbey, United Kingdom). Collagen was measured as μg/mL based on a set of collagen standards.

Lentiviral Overexpression of TSG-6

The TSG-6 cDNA was cleaved from pCDH_TSG6 (XbaI – BamHI blunted sites),⁷² then cloned immediately upstream to the pLenti-C-mGFP (blunted BamHI site) lentiviral vector from OriGene (Rockville, MD). In this vector, the expression of the exogenous cDNA was driven by the cytomegalovirus promoter with the expression of green fluorescent protein. The 2 packaging plasmids psPAX2 and pMD-VSVG were used to produce lentiviral pseudo-particles in HEK 293T cells. Exponentially growing HEK293T cells were transfected with calcium phosphate with the 3-vector lentiviral system. A total of 72 hours after transfection, the supernatant containing the recombinant viruses were collected, filtered (0.45 μm), and centrifuged at 80,000 g for 3 hours at 4ºC. Then, lentiviral particles were used to infect Caco-2 cells or primary fibroblasts that finally were proceeded to fluorescence-activated cell sorting for sorting green fluorescent protein–positive cells using a BD FACS Melody. Postsorting, the cells were collected in tissue culture medium and plated in T25 cell culture flasks.

Cell Proliferation Assay

Caco-2 cells were infected with lentiviral particles generated with a transfer vector for constitutive overexpression of TSG-6 (Caco-2 TSG-6 OE) or a control empty vector (Caco-2 EV).

Cells were seeded in 96-well plates (5 × 10⁴ cells per well) for 24–48 hours, then were incubated with [³H]-thymidine (PerkinElmer, Waltham, MA) (0.01 mCi/mL) for 16 hours. Cells were harvested at 24 and 48 hours and a scintillation β-counter was used to measure the incorporation of [³H]-thymidine (PerkinElmer). Results were expressed as a percentage of cell proliferation normalized on time zero point.

In Vitro Wound Healing Assay

Wound healing capacity was assessed by a scratch assay on a Caco-2 cell monolayer as previously reported.⁷³ Briefly, 5×10⁴ Caco-2 TSG6-OE or Caco-2 EV were seeded into 24-well culture inserts and incubated at 37°C in a humidified atmosphere with 5% CO₂. After 24 hours, the culture inserts were gently removed using sterile tweezers, and a scratch was made in the monolayers. To evaluate the cells’ migratory capacity in the absence of cell proliferation, the cells were preincubated with or without mitomycin C (30 μg/mL; Sigma-Aldrich) for 2h at 37°C. The effects of exogenous TSG-6 on migratory capacity of Caco-2 cells were tested by stimulating the cells with recombinant human TSG-6 (R&D Systems, Minneapolis, MN) (5 ng/well or 10 ng/well). Bright-field images of the wounded area were acquired immediately before stimulation (0 hour time point), and after 24 and 48 hours to monitor the closure of the wounded area, using an inverted microscope (Olympus IX51). The percentage of wound closure was calculated by ImageJ software (ImageJ2 2.3.0; National Institutes of Health, Bethesda, MD) as (Area initial − Area final)/Area initial × 100.

RNA Sequencing and Analysis

Total RNA was extracted from samples using the RNeasy Lipid Tissue Mini kit (Qiagen, Hilden, Germany), according to the manufacturer’s instructions. and checked for quality using Tapestation 4200 (Agilent, Santa Clara, CA). RNA-seq libraries were created using the TrueSeq Stranded mRNA Sample Prep Kit (Illumina, San Diego, CA) starting from 400 ng of RNA. Paired-end sequence reads (150 bp in length) were generated on a NextSeq 500 (Illumina). The sequencing reads were processed to remove Illumina barcodes and aligned to the human genome assembly (hg38) using STAR v.2.5.1b with default parameters. FastQ files were mapped to the hg38 human reference genome with the STAR aligner v2.7.3a.⁷⁴ Differential gene expression was performed with DESeq2.⁷⁵ Functional enrichment was performed with the gene set enrichment analysis and Ingenuity Pathway Analysis (Qiagen). Heatmaps were generated with ComplexHeatmap and multidimensional scaling of transcriptomic data was performed with the rtsne R package by the t-distributed stochastic neighbor embedding machine learning algorithm. Differentially expressed genes were selected based on the following parameters: P value ≤.05 (upregulation or downregulation). Statistical and downstream bioinformatics analyses were performed within the R environment (R 3.6.3; R Foundation for Statistical Computing, Vienna, Austria). RNA-seq data have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus data repository with the accession number GSE157020. cDNA synthesis was carried out using 1 μg of total RNA with the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Bedford, MA). RT-PCR for validation was performed using SYBR Green Master Mix (Applied Biosystems) and detected by using Viia7 Detection system (Applied Biosystems). GAPDH was used as a housekeeping gene and the relative abundance was expressed as 2^−ΔCt. The specific oligonucleotide primers used are listed in Table 2.

Table 2.

List of Real-Time Polymerase Chain Reaction Primers

Gene	Forward	Reverse
TNFAIP6	5′ TTTCTCTTGCTATGGGAAGACAC 3′	5′ GAGCTTGTATTTGCCAGACCG 3′
CDH1	5′ CGAGAGCTACACGTTCACGG 3′	5′ GGGTGTCGAGGGAAAAATAGG 3′
CD44	5′ CTGCCGCTTTGCAGGTGTA 3′	5′ CATTGTGGGCAAGGTGCTATT 3′
IL6	5′ACTCACCTCTTCAGAACGAATTG 3′	5′ CCATCTTTGGAAGGTTCAGGTTG 3′
HAS1	5′ GAGCCTCTTCGCGTACCTG 3′	5′ CCTCCTGGTAGGCGGAGAT 3′
HAS2	5′ CTCTTTTGGACTGTATGGTGCC 3′	5′ AGGGTAGGTTAGCCTTTTCACA 3′
ITGA2	5′ ACAAGCGTTACTGTGAAGCG 3′	5′ GGGCCAGGAGACCTAAGAAATAA 3′
ITGA2B	5′ CACCTCAGCATCCACCTTCCG 3′	5′ GTCTGCGATCCCGCTTGTGATG 3′
ITGA4	5′ CACAACACGCTGTTCGGCTA 3′	5′ CGATCCTGCATCTGTAAATCGC 3′
PTK2B	5′ CCCCTGAGTCGAGTAAAGTTGG 3′	5′ GATACGCACGTCCTCCTTTTC 3′
SNAI1	5′ TCGGAAGCCTAACTACAGCGA 3′	5′ AGATGAGCATTGGCAGCGAG 3′
TNFalpha	5′ CCTCTCTCTAATCAGCCCTCTG 3′	5′ GAGGACCTGGGAGTAGATGAG 3′
GAPDH	5′ GGAGCGAGATCCCTCCAAAAT 3′	5′ GGCTGTTGTCATACTTCTCATGG 3′
ACTA2	5′ ACTGCCTTGGTGTGTGACAA 3′	5′ TCCCAGTTGGTGATGATGCC 3′
FUT8	5′ GGTGGATGGGAGACTGTATTT 3′	5′ GTCTTCTGGTACAGCCAAGGGT 3′
ITGA3	5′ TGTGGCTTGGAGTGACTGTG 3′	5′ TCATTGCCTCGCACGTAGC 3′
ITGAL	5′ CGTGGTGTATGAGAAGCAGATG 3′	5′ GCTCAGAGTCTTCTGCAGGGAT 3′
ITGB7	5′ CAGCTTTCACCATGTGCTGTC 3′	5′ CCAGCAGCCGGGACACATTT 3′
NEDD9	5′ GGAGGAGTTTGAGAGGCAACA 3′	5′ CTCACGCCACTGTTTGTGGT 3′
MMP9	5′ TGTACCGCTATGGTTACACTCG 3′	5′ GGCAGGGACAGTTGCTTCT 3′
MMP13	5′ GTTTGGTCCGATGTAACTCCTCT 3′	5′ CATCTCCTCCATAATTTGGC 3′
MMP19	5′ CTGCTGTCTACTCGCCTCGA 3′	5′ GAGAGGCCAATAGAGAGCTG 3′
MMP25	5′ CCTCGGCGCAGGACGTGA 3′	5′ GCGAACCTCTGCATGACTTTG 3′
LGR5	5′ GAGTTACGTCTTGCGGGAAAC 3′	5′TGGGTACGTGTCTTAGCTGATTA 3′
VIMENTIN	5′GGACCAGCTAACCAACGACA 3′	5′ AAGGTCAAGACGTGCCAGAG 3′

Histological Staining for Collagen Deposition

For histological assessment human samples were fixed in 4% (v/v) formalin, processed, and embedded in paraffin, and 2-μm-thick sections were stained with hematoxylin and eosin.

Collagen deposition was evaluated by Sirius red staining performed on 5-μm-thick formalin-fixed paraffin embedded (FFPE) tissue sections. Second harmonic generation analysis of collagen fibers was conducted by imaging Sirius red–stained slides with the Olympus BX-51 fluorescent microscope using the U-ANT and U-POT polarizers (Olympus), and images were quantified with ImageJ.⁷⁶

Immunofluorescence Staining

On Primary Fibroblasts

Primary fibroblasts seeded on polylysine-coated glass coverslips were fixed in 4% (v/v) paraformaldehyde (PFA) for 10 minutes at room temperature and permeabilized with 0.1% (v/v) Triton X-100 in PBS for 20 minutes at room temperature (all from Sigma-Aldrich). The fixed cells, preincubated for 15 minutes with 1% (w/v) bovine serum albumin, were incubated with Alexa Fluor488-conjugated Phalloidin (F-actin) (1:40; Thermo Fisher Scientific; A12379) for 30 minutes at room temperature. DAPI (1:25,000; Invitrogen) was used for nuclear staining. Coverslips were finally mounted with Fluorescence Mounting Medium (S3023; Dako, Santa Clara, CA). Control slides were obtained by omitting the use of primary antibodies. The images were acquired by Leica SP8 Confocal Microscopy (Leica, Wetzlar, Germany) using 60× objective and analyzed by using ImageJ.

On Human iPSCs, Embryoid Body Formations, and Organoids

Human iPSCs and embryoid body formations were fixed in 4% (v/v) PFA for 10 minutes at room temperature, washed with PBS, and permeabilized in 0.3% Triton X-100 in PBS for 10 minutes at room temperature, whereas intestinal organoids were fixed with 4% PFA, frozen, embedded in Optimal cutting temperature (OCT) compound, and then cut into 10-mm sections and attached to glass slides. Sections were permeabilized in 0.3% Triton X-100 in PBS for 10 minutes at room temperature and blocked in PBS containing 5% FBS for 1 hour before starting the incubation with primary antibodies. Anti-OCT4 (1:200; ab18976; Abcam, Cambridge, United Kingdom), anti-NANOG (1:500; ab62734; Abcam), anti-SOX-2 (1:200; ab97959; Abcam), anti-Nestin (1:200; ab22035; Abcam), anti-α-smooth muscle actin (1:100; ab5694; Abcam), and anti-GATA4 (1:150; ab134057; Abcam). After primary antibody incubation, samples were washed with PBS and incubated with secondary Alexa Fluor 488-conjugated antibodies (Life Technologies, Carlsbad, CA), diluted 1:2000. Samples were also counterstained with DAPI (200 μg/mL). Slides were observed using an Olympus BX61 research microscope equipped with a cooled CCD camera. Images were captured and analyzed with Applied Imaging software CytoVision.

After differentiation, intestinal organoids were fixed with 4% (v/v) PFA, frozen, embedded in OCT compound, and then cut into 10-mm sections and attached to glass slides. Sections were permeabilized in 0.3% Triton X-100 in PBS for 10 minutes at room temperature, blocked in PBS containing 5% FBS for 1 hour, and reacted with the following primary antibodies: anti-chromogranin A (1:250, ab85554; Abcam), recombinant anti-MUC2 antibody (EPR23479-47) (1:400, ab272692; Abcam), recombinant anti-EpCAM antibody (E144) (1:200, ab32392; Abcam), anti-cytokeratin 19 (1:500 ab7754; Abcam), anti-villin (1.400, ab97512; Abcam), and anti-Zo1 (1:50, 61-7300; Life Technologies) at room temperature for 1 hour. After primary antibody incubation, samples were washed with PBS and incubated at room temperature for 45 minutes with secondary Alexa Fluor 488– or Alexa Fluor 555–conjugated antibodies (1:2000; Life Technologies). Slides were mounted in VECTASHIELD mounting medium with DAPI (Vector Laboratories) and then were observed using an Olympus BX61 research microscope equipped with a cooled CCD camera. Images were captured and analyzed with Applied Imaging software CytoVision.

On Frozen Tissue Sections

Frozen fistula and perifistula sections (4 μm) collected on SuperFrost glass slides (Thermo Fisher Scientific) were fixed in 4% (v/v) PFA for 10 minutes at room temperature and permeabilized with 0.1% (v/v) Triton X-100 in PBS (all from Sigma-Aldrich) for 20 minutes at room temperature. The sections were preincubated for 15 minutes with 1% (w/v) bovine serum albumin, and incubated with anti-IαI antibody (1:50; Dako) overnight at 4°C, followed by incubation for 30 minutes at room temperature with Alexa Fluor 647–conjugated goat anti-rabbit (1:1000; Invitrogen; A21245). Subsequently, fixed tissues were incubated with biotinylated HA binding protein (biotin-HABP; 2.5 μg/mL; HoKuto Co, Nagano, Japan)⁷⁷ for 1 hour at room temperature, followed by incubation with Alexa Fluor 488 streptavidin (1:1000; Invitrogen; S32354). DAPI (1:25,000; Invitrogen) was used for nuclear staining. Coverslips were finally mounted with Fluorescence Mounting Medium (Dako; S3023). Control slides were obtained by omitting the use of primary antibodies. The images were acquired by Leica SP8 Confocal Microscopy using 60× objective and analyzed by using ImageJ.

On FFPE Tissue Sections

The 2 μm tissue sections were prepared using standard FFPE fixation and embedding techniques, then antigen retrieval was conducted for 20 minutes at 98°C using Tris EDTA buffer pH = 9. The tissues, preincubated for 1 hour with 2% (w/v) FCS, were incubated with biotin-HABP (2.5 μg/mL; HoKuto Co) and RAH-1 (rabbit anti-human antibody against TSG-6, 1:500 dilution)⁷⁸ overnight at 4°C, followed by 30 minutes of incubation at room temperature with Alexa Fluor 488 streptavidin (1:1000; Invitrogen; S32354) and Alexa Fluor 647–conjugated goat anti-rabbit (1:1000; Invitrogen; A21245), respectively. DAPI (1:5000; Invitrogen) was used for nuclear staining. Coverslips were finally mounted with Fluorescence Mounting Medium (Dako; S3023). Control slides were obtained by omitting the use of primary antibodies. The images were acquired by Leica SP8 Confocal Microscopy using a 20× and analyzed by using ImageJ.

Western Blot Analysis

Protein extracts from biopsies or cell pellets were solubilized using RIPA buffer—10 mM Tris-HCl (pH 8.0), 1 mM EDTA, 140 mM NaCl, 1% (v/v) Triton X-100, 0.1% (v/v) sodium deoxycholate, 0.1% (w/v) sodium dodecyl sulfate—supplemented with protein inhibitor cocktail (1:50; Thermo Fisher Scientific; A32963). The 30 μg protein extracts were resolved by 8% sodium dodecyl sulfate polyacrylamide gel electrophoresis according to the manufacturer’s protocols (Bio-Rad, Hercules, CA). After incubation with 5% (w/v) nonfat milk in Tris-buffered saline with Tween 20 (TBS-T; 10 mM Tris, pH 8.0, 150 mM NaCl, 0.5% (v/v) Tween-20) for 60 minutes, the membranes were incubated with antibodies against TSG-6 (1:1000; RAH-1), or vinculin (1:1000; Santa Cruz Biotechnology, Dallas, TX; sc-25336), or Actin (1:1000; Santa Cruz Biotechnology; sc-1615) overnight at 4°C. Then, after several washes with TBS-T, the membranes were incubated with horseradish peroxidase–conjugated anti-rabbit or anti-mouse antibodies respectively (1:2500; Santa Cruz Biotechnology) for 1 hour at room temperature, followed by development using enhanced chemiluminescence substrate (EuroClone) according to the manufacturer’s protocols. Finally, the immunoreactivity was detected by Chemidoc (Bio-Rad) using Quantity One software.

Proteins in the cell culture supernatant were pretreated with 0.1M NaOH for 10 minutes at room temperature (neutralization was performed with 0.1M HCl) or with 0.05 U chondroitinase ABC (Sigma-Aldrich; C2905) for 4 hours at 37°C and concentrated using StrataClean Resin (1:30; Agilent; 400714). Proteins were separated on 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred onto nitrocellulose membrane according to the manufacturer’s protocols (Bio-Rad). After overnight incubation at 4°C with 5% (w/v) nonfat milk in TBS-T, the membranes were incubated with the antibody against TSG-6 (ap_RAH-2 rabbit anti-human antibody 1:800) overnight at 4°C. The rabbit anti-human antibody RAH-2 was affinity purified (ap) by incubation with the peptide CTSTGNKNFLAGRFSHL (used for immunization), immobilized via the N-terminal cysteine to SulfoLink Coupling Resin (Thermo Fisher Scientific), where bound IgG was eluted at pH 2 and concentrated with buffer exchange into PBS. Then, membranes were incubated with horseradish peroxidase–conjugated anti-rabbit (1:1000; Santa Cruz Biotechnology) for 1 hour at room temperature. The development and the detection of the immunoreactivity were performed as reported previously.

Statistical Analysis

Statistical analyses were performed with GraphPad Prism version 8 (GraphPad Software, San Diego, CA) using Student t test or Mann-Whitney U test for the comparison between 2 groups, whereas 1-way or 2-way analysis of variance with Bonferroni posttest analysis was used for the comparison of multiple groups and paired Student t tests for comparisons between 2 means. The data were expressed as mean ± SEM. Statistical significance was defined as P < .05.