The orphan nuclear receptor ROR alpha and group 3 innate lymphoid cells drive fibrosis in a mouse model of Crohn’s disease

Abstract

Fibrosis is the result of dysregulated tissue regeneration and is characterized by excessive accumulation of matrix proteins that become detrimental to tissue function. In Crohn’s disease, this manifests itself as recurrent gastrointestinal strictures for which there is no effective therapy beyond surgical intervention. Using a model of infection-induced chronic gut inflammation, we show that Rora-deficient mice are protected from fibrosis; infected intestinal tissues display diminished pathology, attenuated collagen deposition and reduced fibroblast accumulation. Although Rora is best known for its role in ILC2 development, we find that Salmonella-induced fibrosis is independent of eosinophils, STAT6 signaling and Th2 cytokine production arguing that this process is largely ILC2-independent. Instead, we observe reduced levels of ILC3- and T cell-derived IL-17A and IL-22 in infected gut tissues. Furthermore, using Rora^sg/sg/Rag1^−/− bone marrow chimeric mice, we show that restoring ILC function is sufficient to re-establish IL-17A and IL-22 production and a profibrotic phenotype. Our results show that RORα-dependent ILC3 functions are pivotal in mediating gut fibrosis and they offer an avenue for therapeutic intervention in Crohn’s-like diseases.

Introduction

Fibrosis is the pathological deposition of extra-cellular matrix (ECM) associated with persistent inflammation. The contribution of the hematopoietic system to this process has been widely investigated and numerous leukocyte subsets are implicated in driving fibrotic remodeling including M2 macrophages, neutrophils, eosinophils, mast cells and lymphocytes (1, 2). In particular, type 2 immune responses have long been associated with tissue fibrosis and IL-4 and IL-13 signaling has been shown to drive tissue remodeling and ECM deposition in the liver in response to Schistosoma mansoni infection (3). In mucosal tissues both CD4⁺ T helper 2 (Th2) and Th17 cells have been linked to the development of tissue fibrosis. For example, in a mouse model of chronic gut inflammation and fibrosis, an overlapping production of IL-13 and IL-17A coincides with the appearance of fibrotic tissue, which suggests the pathology may be driven by a mixed Th2 and Th17 response (4).

Innate lymphoid cells (ILCs) are a recently identified subset of leukocytes and a prominent source of cytokines with profiles similar to those of T helper cells, yet these cells lack antigen-specific receptors and their temporal production of cytokines precedes adaptive immune responses. ILCs can be classified based on their transcript and cytokine profiles; for instance, group 2 ILCs (ILC2s) are GATA3- and RORα-dependent and produce IL-5 and IL-13 while group 3 ILCs (ILC3s) express RORγt and are a source of IL-17A and IL-22. ILCs have broad functions in host defense responses against pathogens, maintenance of tissue homeostasis, and pathologies such as allergic asthma and colitis (5). Recently, ILC2s were shown to have the capacity to drive excessive lung collagen deposition; likewise, analyses of idiopathic pulmonary fibrosis patient bronchoalveolar lavage fluid (BAL) revealed an expansion of ILC2s, providing correlative evidence that ILCs may be pathogenic in fibrotic remodeling (6). Despite their pathogenic role in these models, ILC2s have also been shown to play a critical role in tissue repair. For example, in mouse models of lung influenza infection (7) and DSS-induced colitis (8) ILC2s aid tissue repair through their production of amphiregulin.

Similarly, ILC3s in the gut demonstrate protective or pathogenic roles depending on context. ILC3s are necessary for pathogen clearance (9, 10) and gut tissue repair following chemotherapy-induced injury. This is likely through IL-22-dependent activation of epithelial progenitors (11), but ILC3s can also promote intestinal tumor formation in a chronic inflammatory setting (12). In summary, there are data to support both beneficial and pathogenic roles for ILC2s and ILC3s in tissue remodeling.

Crohn’s disease (CD) is primarily characterized as a Th17-driven disease (13), and distinct IL-23-dependent ILCs that produce IL-17A and IL-22 are also elevated in CD patient samples (14, 15) arguing for a role for Th17 cytokines. Conversely, other studies have suggested that ILC2s and eosinophils are contributing factors to CD progression through an IL-13 dependent accumulation of matrix-producing fibroblasts (16, 17). Thus, there is evidence to support both Th2 and Th17 responses in the pathology of CD. However, the direct involvement of ILCs in driving excessive tissue remodeling and ultimately fibrosis in the gut remains unclear.

RORα is a transcription factor known to regulate ILC2 development. It is closely related to RORγt, which is a critical regulator of Th17 and ILC3 subsets including lymphoid tissue-inducer cells (18–20). Moreover, in the Th17 lineage, RORγt and RORα are co-expressed and function synergistically for lineage maintenance and cytokine production (21). Previously we showed that Rora^sg/sg bone marrow transplant (BMT) chimeric mice provide a suitable model of ILC2 deficiency as they selectively lack ILC2s while other leukocyte lineages, including other ILCs, are present at steady state (22). Furthermore, Rora^sg/sg BMT animals are protected from type 2 immune lung pathology (22–24). In this report, we used a similar strategy to evaluate their role in a Salmonella-induced model of gut fibrosis. Strikingly, we show that RORα is required for a fibrotic response. Surprisingly, this is not through regulation of ILC2-dependent production of type 2 cytokines, but rather through a previously unrecognized role for RORα in the production of cytokines from ILC3s. We also find that restoration of ILCs in Rora^sg/sg BMT mice is sufficient to reestablish susceptibility to fibrotic disease. These data provide evidence that ILC3s contribute to fibrosis, and offer RORα as a new therapeutic target for fibrotic tissue remodelling.

Results

Salmonella-induced gut pathology is dependent on hematopoietic expression of RORα

We generated Rora^sg/sg BMT mice and evaluated how they respond to infection with the ΔaroA strain of Salmonella Typhimurium, which induces robust ECM deposition in the cecum (25). We find no differences in pathogen burdens in the ceca or spleens of WT BMT and Rora^sg/sg BMT mice at days 7, 21, and 35 postinfection (pi), indicating that Rora is dispensable for Salmonella clearance (Fig. 1A). In addition, we find similar degrees of mucosal thickening, edema, and inflammatory infiltrates in these animals at day 7 pi (Fig. 1B). In contrast, at 21 days pi, during peak fibrosis, we observe pronounced epithelial cell hyperplasia and pathological remodeling of normal tissue architecture, including disruption and thickening of the basal lamina in WT BMT ceca, whereas we find a striking attenuation in collagen deposition and dampened immunopathology in cecal sections of Rora^sg/sg BMT mice. Similarly, by day 35 pi, much of the pathology and inflammation is resolved in Rora^sg/sg BMT mice, with mucosa thickness returning to levels similar to uninfected samples (albeit with some residual collagen in the submucosal regions) (Fig. 1B). Conversely, WT BMT cecal sections reveal persistent tissue inflammation and lymphoid cell aggregates at this time point (Fig. 1B, C).

Figure 1. Hematopoietic expression of Rora mediates S. Typhimurium ΔaroA driven cecal immunopathology and fibrosis.

WT and Rora^sg/sg BMT mice were infected with Salmonella and sacrificed 7, 21, and 35 days pi. (A) Salmonella colonization of ceca and spleens of infected animals. CFU, colony-forming unit. (B) Masson’s trichrome (MT)-stained cecal tissue sections. L, lumen; SE, surface epithelium; M, mucosa; SM, submucosa. Arrowheads indicate submucosal collagen accumulation visualized by blue staining. Scale bar, 200 μm. (C) Pathology scores of the luminal, epithelial, mucosal, and submucosal subsections of cecal tissue. *1, p = 0.012 (n=5 per group); *2, p= 0.0016 (n=6, 7 per group). (D) Transcript levels of pro-inflammatory cytokines Tnfa and Il6 in ceca normalized to Gapdh. TNFa: *3, p=0.022 (n=5 per group); *4, p=0.027 (n=5, 6 per group); *5, p=0.002 (n=5 per group); *6, p = 0.032 (n=11, 13 per group; data from 2 independent experiments) (E) Protein levels of TNF, IL-6, MCP-1, IFN-γ, IL-12p70, and IL-10 in naïve and 21 days pi cecal homogenates as determined by cytometric bead array (CBA) normalized to total tissue protein. *7, p=0.0017; *8, p=0.012; *9, p=0.0010; *10, p=0.040 (n=10, 11; data from 2 independent experiments). Significance determined by unpaired student’s t test.

In evaluating cytokine production, we find Tnfa and Il6 transcripts are present at similar levels at day 7 pi, but are significantly reduced in Rora^sg/sg BMT cecal samples at the day 21 and 35 time points (Fig. 1D). Consistent with these mRNA results, quantification of cytokines in cecal homogenates reveals significantly less TNF, IL-6, MCP-1, and IL-10 (but comparable levels of IL-12p70 and IFN-γ) in Rora^sg/sg BMT samples when compared with their WT BMT counterparts at day 21 pi (Fig. 1E). In summary we find that deletion of Rora leads to attenuated inflammatory responses associated with fibrosis and these correlate with a more rapid resolution of immunopathology and tissue architecture during the late phases of chronic Salmonella infection.

Fibrosis is attenuated in Rora^sg/sg BMT mice during chronic infection

Consistent with a role for RORα in driving fibrosis we find significantly reduced levels of Col1a2 and Tgfb1 transcripts and TGF-β1 protein in Rora^sg/sg BMT ceca at days 21 and 35 pi (Fig. 2A, B). Immunofluorescent staining for type 1 collagen (Col1) in WT BMT tissues indicates that the most pronounced collagen deposition occurs in the submucosal (SM) region of the cecum, with more modest accumulation in the mucosa, 21 and 35 days pi. Strikingly, there is less SM Col1 staining in the Rora^sg/sg BMT ceca (Fig. 2C) as quantified by evaluation of the fold change increase in surface area of SM Col1 staining normalized to total tissue surface area (Fig. 2D). Consistent with these findings, we also observe reduced type 3 collagen (Col3) staining in the submucosal regions of the cecum of Rora^sg/sg BMT mice compared with WT BMT tissue (Fig. 2E). Finally, we visualized fibroblast accumulation at day 21 pi by staining for the common fibroblast markers vimentin and desmin (25). We find substantial accumulation of vimentin and desmin positive cells across the mucosal and SM regions of WT BMT ceca; desmin staining is particularly enriched within the basal lamina which is thicker in WT than in Rora^sg/sg BMT samples (Fig. 2F). Quantification of vimentin⁺ and desmin⁺ cells in these immunofluorescent images reveals a reduction in fibroblast numbers in Rora^sg/sg BMT cecum compared with WT BMT tissues (Fig. 2G). Taken together our results indicate that, by all criteria, Rora^sg/sg BMT animals are protected from fibrosis with reduced collagen and fibroblast accumulation in the cecum.

Figure 2. Attenuated collagen deposition and reduced fibroblast cell numbers in cecal tissue of chronically infected Rora^sg/sg BMT animals.

(A) Cecal transcript levels of Col1a2 and Tgfb1 normalized to Gapdh. *1, p=0.026 (n=5 per group); *2, p=0.016 (n=6,7 per group); *3, p=0.029 (n=5 per group); *4, p=0.015 (n=6, 7 per group) (B) TGF-β1 protein levels of cecal homogenates normalized to total protein. *5, p=0.0044 (n=10 per group, data from 2 independent experiments); *6, p=0.023 (n=5, 6 per group) (C) Immunofluorescence images of cecal sections stained for type 1 collagen (Col1, magenta) and nuclei (DAPI, cyan). Scale bar, 100 μm. (D) Quantification of the submucosal collagen surface area normalized to total tissue surface area expressed as fold change of uninfected. *7, p=2.96×10⁻⁴ (n=5 per group); *8, p= 5.57×10⁻⁴ (n=6, 7 per group) (E) Immunofluorescence images of cecal sections stained for type 3 collagen (Col3, red) and nuclei (DAPI, cyan). Scale bar, 100 μm. (F) Desmin (magenta), vimentin (blue), and nuclear (DAPI, cyan) staining for fibroblasts in cecal tissues 21 days pi. Arrowheads indicate desmin⁺ staining in the basal lamina. Scale bar, 100 μm. (G) Quantification of vimentin⁺ or desmin⁺ fibroblasts per representative field of view (FOV). Significance determined by unpaired Student’s t test.

Eosinophils and STAT6 signaling are dispensable in Salmonella induced fibrosis

Chronic Salmonella infection is known to induce a Th1/Th17 response (25), but the contribution of type 2 immune cell cytokines in this model has not been thoroughly addressed. In our cytokine analysis, however, we find that IL-5 and IL-13 transcript levels in fibrotic WT BMT cecal tissues were similar to infected tissues of Rora^sg/sg BMT animals, and importantly, were reduced or unaltered when compared with uninfected samples (Fig. S1A). While we find that absolute numbers of ILC2s and indeed eosinophils in WT BMT cecal samples expand in response to Salmonella infection, and are significantly reduced in Rora^sg/sg BMT samples, it is noteworthy that the relative proportions of ILC2s and eosinophils (as Th2 effector cells) normalized to all CD45⁺ cells are reduced or unchanged during peak fibrosis when compared to naïve controls (Fig S1B,C). Thus, these data argue against a major role of ILC2s in this fibrotic response.

To further rule out a role of Th2 immunity in Salmonella induced fibrosis, infections were performed in eosinophil-deficient ΔdblGATA mice. These studies reveal similar pathogen burdens in the cecum and spleen (Fig. 3A) as well as comparable evidence of pathology in MT-stained cecal sections of infected animals (Fig. 3B, C). Transcript quantification also reveals similar levels of Col1a2, Tnfa, and Il6 in the eosinophil-deficient tissues, although Tgfb1 and Epx transcripts are significantly reduced (Fig. 3D). We conclude that eosinophils are dispensable in this model.

Figure 3. Eosinophils and STAT6 signaling are dispensable in S. Typhimurium ΔaroA-dependent gut inflammation and fibrosis.

(A–D) Eosinophil deficient ΔdblGATA, (E–H) Stat6^−/− and their respective WT controls were infected with Salmonella and sacrificed 21 days pi. (A) Cecal and splenic Salmonella burdens of infected animals. (n=10 per group; data from two independent experiments) (B) MT-stained cecal sections and (C) their corresponding pathology scores. (D) Transcript levels of Tnfa, Il6, Col1a2, Tgfb1, and Epx of infected cecal tissues normalized to Gapdh. *1, p=0.042; *2, p=0.0014 (n=5 per group). Significance determined by unpaired student’s t test. (E) Pathogen burdens of Salmonella infected ceca and spleens. (n=7, 8; data from single experiment. (F) MT-stained infected cecal tissues and (G) their corresponding pathological scores. (H) Transcript levels of Tnfa, Il6, Col1a2, and Tgfb1 of Salmonella infected ceca normalized to Gapdh. Scale bar, 200 μm.

In many examples of fibrosis, STAT6 plays an essential role as a transcriptional mediator of downstream IL-4/IL-13 receptor signaling involved in the propagation of other type 2 immune cells (1). Following infection with Salmonella, we find that Stat6^−/− and WT animals exhibit similar pathogen burdens in the cecum and the spleen (Fig. 3E) and display a similar degree of pathology in the cecum (Fig. 3F,G). Furthermore, we observe comparable levels of Col1a2, Tgfb1, Tnfa, and Il6 expression in both WT and Stat6^−/−cecal tissues (Fig. 3H). In aggregate, our data indicate that eosinophils and STAT6/Th2 effector function signaling are dispensable for Salmonella-induced gut fibrosis.

RORα mediates IL-17A and IL-22 production in chronic gut inflammation and fibrosis

Although Rorα^sg/sg BMT mice are known to have a profound defect in ILC2 development(22, 26), one previous study indicated that RORα is also required by Th17 cells for normal cytokine production (21). In addition, although Rora^sg/sg BMT animals have normal ILC3 numbers in the small intestine and mesenteric lymph nodes (mLNs) at steady state (22), several studies have detected high Rora transcript levels in gut ILC3s (27, 28) or ILC3-like subsets (29, 30). Despite these observations, the importance of RORα expression in regulating ILC3s in the context of inflammatory diseases is unknown. We therefore evaluated Th17 and ILC3 associated cytokines in infected Rora^sg/sg BMT ceca. These mice exhibit a decrease in cecal Il17a and Il22 transcripts at days 21 and 35, and a corresponding reduction in IL-17A and IL-22 protein in cecal homogenates (Fig. 4A, B). Consistent with this observation, we previously reported significantly reduced levels of IL-17A in the lungs of Rora^sg/sg BMT mice in both Ova-and HDM-induced models of allergic disease (23). We conclude that Th17-biased immune responses are highly attenuated in RORα-deficient mice.

Figure 4. ILC3 production of IL-17A and IL-22 is Rora dependent in response to chronic Salmonella infection.

(A) Transcripts of Il17a and Il22 in WT and Rora^sg/sg BMT ceca normalized to Gapdh. *1, p=0.029 (n=5 per group); *2, p=0.018 (n= 11, 13 per group; data from two independent experiments) (B) Protein levels of IL-17A and IL-22 in cecal homogenates normalized to total protein. *3, p=0.035 (n=10 per group; data from two independent experiments); *4, p=0.034 (n=6, 7 per group); *5, p = 0.020 (n=10, 11 per group; data from two independent experiment); *6, p=0.036 (n=6 per group). (C) Total ILC3s (CD45⁺ Lin^neg CD90^high KLRG1^neg) per mLN 21 days pi. (D–F) Intracellular staining of IL-17A and IL-22 of mLN leukocyte subsets following restimulation with PMA, ionomycin, and BFA. Donor versus recipient hematopoietic cells determined by expression of congenic CD45. Percentages of IL-17A⁺ and IL-22⁺ mLN ILC3s (D) and CD3e⁺ lymphocytes (E). (F) Comparison of peripheral blood chimerism with percent contribution of total IL-17A⁺ or IL22⁺ events from BM-donor derived versus radio-resistant recipient hematopoietic cells. *7, p=1.14×10⁻⁶; *8, p= 0.0041; *9, p=6.98×10⁻⁴; *10, p=0.0022; *11, p=0.0056 (n=5, 7 per group). One of three independent experiments. Percentages of IL-17A⁺ and IL-22⁺ gut ILC3s (G) and CD3e⁺ lymphocytes (H). *12, p=1.29×10⁻⁴; *13, p= 0.035; *14, p=1.17×10⁻⁴ (n=5 per group). One of two independent experiments. Significance determined by unpaired Student’s t test.

Next, we characterized IL-17A- and IL-22-producing mLN populations by flow cytometry (Fig. S2). Total mLN ILC3 numbers are comparable in WT and Rora^sg/sg BMT mLNs at 21 days pi (Fig 4C). Intriguingly, despite their normal frequency, we find highly attenuated IL-17A and IL-22 production in ILC3s of Rorα^sg/sg BMT mice (Fig. 4D). This is further supported by evaluation of residual, radio-resistant WT ILC3 present in the mLN of Rora^sg/sg BMT chimeric mice. These cells represent roughly 10% of the mLN ILC3 as determined by allotype marker staining and produce normal levels of IL-17A and IL-22 (Fig. 4D). This suggests that defective cytokine production by Rora^sg/sg ILC3s is cell-intrinsic and independent of potential alterations in the composition of the inflammatory environment. Consistent with previous characterization of RORα in Th17 cells (21), Rora^sg/sg CD3⁺ T cells isolated from mLNs of infected animals have reduced IL-17A but unaltered IL-22 production compared to WT T cells (Fig. 4E).

Peripheral blood analyses of BM chimeras reveals approximately 90% donor-derived CD45.2⁺ cells, and 10% CD45.1⁺ radio-resistant host-derived cells (Fig. 4F). More selective analyses of IL-17A⁺ mLN cells in WT BMT mice reveals 73% IL-17A⁺ cell chimerism suggesting there is a subset of tissue-specific, radio-resistant IL-17A-producing cells (Fig. 4F). Interestingly, these cells are more frequent in Rora^sg/sg BMT chimeras, with only ~55% of total IL-17A⁺ cells originating from donor BM cells. Host and donor contributions to IL-22⁺ cells in both WT and Rora^sg/sg BMT mice overlap with their respective peripheral blood chimerism indicating a lack of IL-22⁺ cell enrichment in the tissue-resident, radio-resistant population (Fig. 4F). Consistent with mLN cell analyses, ILC3s isolated from the ceca of Rora^sg/sg BMT mice have reduced IL-17A and IL-22 production compared with WT BMT ILC3s (Fig. 4G); RORα-deficient cecal CD3⁺ T-cells also display attenuated IL-17A but normal IL-22 expression (Fig. 4H). In summary, we conclude that there is a clear defect in cytokine production by Th17 and ILC3 cells in Rora^sg/sg BMT mice.

Salmonella-induced gut fibrosis is dependent on Rora⁺ ILC3

To further evaluate the ILC- versus Th cell-dependent contributions to fibrosis in our experimental model, we generated mixed BM chimeras by reconstituting lethally irradiated mice with equal numbers of BM cells from Rag1^−/− mice combined with either WT or Rora^sg/sg BM cells. The rationale was that co-transplanting Rag1^−/− and Rora^sg/sg BM cells would restore normal ILC3s (Rag1^−/−-derived) without altering the Rora^sg/sg–derived T cell compartment. At 21 days pi we find cecal and splenic Salmonella burdens in these mice are similar (Fig. 5A). MT-stained cecal tissues from these Rag1^−/−/Rora^sg/sg BMT animals reveal immunopathology and fibrosis comparable to Rag1^−/−/WT BMT animals (Fig. 5B), including a similar degree epithelial cell hyperplasia, inflammatory infiltrates, and collagen deposition in the submucosal regions, which is represented in the pathology scores (Fig. 5C). Further, we find similar transcript levels of Col1a2, Tgfb1, Tnfa and Il6 in infected ceca of Rag1^−/−/WT and Rag1^−/−/Rora^sg/sg BMT animals (Fig. 5D). Quantification of IL-17A and IL-22 in cecal homogenates indicates that these cytokines are restored to comparable levels in Rag1^−/−/WT BMT and of Rag1^−/−/Rora^sg/sg BMT samples (Fig. 5E). In summary, these data argue that Rora-expressing ILC3s are sufficient to cause fibrosis.

Figure 5. Restoring the innate cell compartment in Rora^sg/sg BMT mice is sufficient to cause fibrosis.

S. Typhimurium ΔaroA infection of Rag1^−/−/WT and Rag1^−/−/Rora^sg/sg mixed BMT chimeras. (A) Bacterial counts of ceca and spleens 21 days pi. (B) MT-stained cecal tissue with (C) corresponding pathology scores. (D) Relative mRNA expression of Tnfa, Il6, Col1a2, and Tgfb1 in infected ceca normalized to Gapdh. (E) IL-17A and IL-22 protein quantification in cecal homogenates. Representative data from two independent experiments (n = 5 or 7 per experiment). Scale bar, 200 μm.

Finally, to determine if IL-17A is a critical pro-fibrotic factor in this model, we treated Salmonella-infected mice with neutralizing antibodies against IL-17A. Although we find that the Salmonella burdens in the spleens and the ceca of isotype control and anti-IL-17A treated mice were unaltered (Fig. 6A), we observe reduced pathology in the ceca of animals treated with antibodies against Il-17A (Fig. 6B) as MT-stained cecal sections exhibit an attenuation in epithelial remodeling and fibrotic scarring (Fig. 6C). Moreover, IL-17A neutralization resulted in a decrease in SM collagen accumulation (Fig. 6D). Our results indicate fibrosis during the late phases of chronic Salmonella infection is IL-17A dependent.

Figure 6. Neutralizing antibodies against IL-17A attenuates fibrosis following Salmonella infection.

Isotype control (iso) or neutralizing antibodies against IL-17A (α-IL-17A) were administered intraperitoneally to infected mice and sacrificed at day 21. (A) Salmonella counts in ceca and spleens. (B–C) Cecal pathology scores and MT-stained tissue sections. L, lumen; SE, surface epithelium; M, mucosa; SM, submucosa. *1, p=1.10×10⁻⁴ (n=12, 13 per group, results from two independent experiments). Scale bar, 200 μm. (D) Quantification of submucosal collagen surface area normalized to total tissue surface area. *2, p=7.80×10⁻⁴ (n=12, 13 per group, results from two independent experiments). Significance determined by unpaired Student’s t test.

Discussion

This report provides important insights into Rora-dependent immune function and Crohn’s-like fibrotic disease (Fig 4S). Using a model of Salmonella induced intestinal fibrosis, we find that hematopoietic deletion of Rora protects animals from fibrogenesis in the gut. Assessment of tissue pathology and cytokine analysis during the pre-fibrotic phase at day 7 pi suggests early inflammatory responses were comparable in WT and Rora^sg/sg BMT animals. Furthermore, Salmonella burdens in the gut and spleen during initial colonization were unaltered by hematopoietic Rora expression. These observations suggest that RORα does not influence initial inflammatory responses, or the host’s ability to clear the infection, but rather plays a specific role in generating a pro-fibrotic response in the gut during late stages of chronic Salmonella infection. The apparent disconnect between bacterial colonization the severity of fibrosis is consistent with previous reports which demonstrate that intestinal fibrosis does not require the presence of Salmonella as a persistent inflammatory stimulus (31, 32). Instead, the pathology becomes self-propagating and is largely irrespective of interventions that attempt to reduce pathogen burdens or repress early inflammatory responses.

Although gut fibrosis that develops following chronic Salmonella infection strongly correlates with a Th17 response (25), we initially sought to examine the contributions of type 2 immune cell-associated cytokines in fibrosis. This is relevant for several reasons. First, in a mouse model of TNBS-induced colitis, peak IL-4 and IL-13 cytokine levels were reported during the late phases of inflammation and were associated with tissue remodeling and fibrosis (4). Second, ILC2s are well known to recruit eosinophils by their secretion of IL-5 (33); eosinophilia has been linked previously to stricture complications in CD patients and IL-13 has been reported to promote gut fibroblast activation (16). Therefore, while it is likely ILC2s are irrelevant during the early phase of Salmonella infection, their role in downstream fibrotic remodeling has not been thoroughly investigated. While we observed a profound attenuation in collagen deposition in infected Rora^sg/sg BMT ILC2 deficient animals, IL-13 and IL-5 levels were not markedly altered in fibrotic gut tissue when compared to uninfected controls. In support of these observations, we find that fibrosis is not diminished in cecal tissues of infected ΔdblGATA or Stat6^−/− mice, which suggests that eosinophils and STAT6 signaling as mediators of Th2 effector cell functions are dispensable for Salmonella-induced fibrosis. Theses findings argue strongly for a Th2/ILC2-independent role for RORα in fibrosis. In examining other eosinophil derived factors previously implicated in IBD, the reduction in Epx1 was fully expected since infected ΔdblGATA mice lack eosinophils and therefore, this selective marker. Eosinophils are also a well-known source of TGF-β1 and therefore it is perhaps not surprising that its production is lower in these animals. The fact that loss of eosinophils as a source of TGF-β1 has no effect on the fibrotic outcome, however, is intriguing. One could argue that the residual levels of TGF-β1 we observe in the eosinophil-deficient mice is necessary but not sufficient to induce fibrosis and that a second signal is required. Based on our antibody suppression experiments we would argue that IL-17A is one such factor and a critical driver of fibrotic disease.

Consistent with a previous report identifying RORα as an important transcriptional mediator of Th17 cell cytokine production (21), we find that IL-17A and IL-22 levels were diminished in infected cecal tissue of Rora^sg/sg BMT mice compared with WT samples. Interestingly, normal IL-17A and IL-22 production by ILC3s is Rora dependent while IL-17A expression by T cells is also impaired in Salmonella-infected Rora^sg/sg BMT animals. This defect in cytokine production by RORα-deficient ILC3s and Th17 cells is further illustrated by a ~50% increase in the relative proportion of radio-resistant, host-derived IL-17⁺ cell numbers in mLNs of Rora^sg/sg BMT animals when compared to WT controls. The persistence of WT radio-resistant subsets represents a caveat in our BMT model. However, despite the presence of a larger relative population of IL-17A producing radio-resistant host-derived cells, the cumulative effect of deleting Rora in donor cells is sufficient to attenuate the fibrotic phenotype. To delineate the contributions of ILC3s and Th17 cells to the phenotype in our disease model, we restored WT innate cells by co-transplanting mixed Rag1^−/− and Rora^sg/sg BM cells into recipient animals. Following Salmonella infection, these mice develop fibrotic disease in the gut and have IL-17A and IL-22 levels similar to their Rag1^−/−/WT BMT counterparts. These data demonstrates that ILC3s are sufficient to cause fibrosis, but does not formally rule out an additional role of Th17 cells. New technological approaches to selectively deplete ILC3s or ablate Rora in ILC3s would be necessary to definitively address this question.

While we have shown that antibody mediated neutralization of IL-17A protects against fibrosis in a Salmonella-induced model, the clinical importance of these findings may be questioned since anti-IL-17A antibody treatment failed as a therapy in a clinical trial of Crohn’s disease patients due to lack of efficacy and adverse side effects (34). However, Crohn’s disease manifests itself in many ways and fibrosis is only detected in a subset of these patients. Importantly, patients with “stricture causing obstructive symptoms” were excluded from this clinical trial. With this in mind, we would argue that the effects of anti-IL-17A on stricture formation and maintenance in humans remain an open question and that further study is justified.

Although RORα and RORγt are related members of a subfamily of nuclear orphan receptors, each possess distinct and some overlapping functions in hematopoietic development. Unlike Rorc deletion, which ablates ILC3s and Th17 cells, the effect of Rora deletion on these subsets appears to be far more subtle and limited to an attenuation of cytokine production. There has been strong interest in developing RORγt inhibitors as a therapy for Th17- or ILC3- mediated disorders (35–37). However, there are some concerns with this strategy as RORγt⁺ regulatory T cells were found to be critical in limiting disease severity in experimental colitis (38, 39). Moreover, RORγt regulates thymocyte differentiation and aged Rorc^−/− mice exhibit a very high frequency of T cell lymphoma (40). Our findings argue that RORα may serve as a more attractive therapeutic target for Crohn’s disease associated fibrosis and that development of selective inhibitors of RORα (that spare RORγt) is warranted.

Materials and Methods

Study design

Our aim was to investigate the role of RORα dependent immunity in gut fibrotic responses. We employed a bone marrow transplant model for hematopoietic Rora deficiency in mice as well as other conventional mutant knockout strains in a chronic infection model of intestinal fibrosis. Age and sex matched mice were randomly divided into recipient groups for BM transplantation. Infection experiments included four to eight mice per group and were performed at least two times with the exception of experiments pertaining to Stat6^−/− mice (Fig. 3E–3H), which were from a single experiment.

Mice

C57BL/6J, B6.SJL-Ptprc^aPepc^b/BoyJ (CD45.1), B6.C3(Cg)-Rora^sg/J (Rora^sg/sg), C.129S1(B6)-Gata1^tm6Sho/J (ΔdblGATA), C.129S2-Stat6^tm1Gru/J (Stat6^−/−), and B6.129S7-Rag1^tm1Mom/J (Rag1^−/−) mice were maintained in a specific pathogen-free environment at the Biomedical Research Centre. BM chimeras were generated by lethally irradiating and reconstituting CD45.1 animals with two million BM cells isolated from WT or Rora^sg/sg littermates. For Rag1^−/− co-transplant experiments, WT or Rora^sg/sg BM cells were mixed 1:1 with Rag1^−/− BM cells for reconstitution. Animals were used at least 8 weeks after transplantation. All experiments performed were approved by the UBC Animal Care Committee.

Infection model

Mice were treated with 20 mg of streptomycin by oral gavage 24 hours prior to oral infection with 3 × 10⁶ Salmonella Typhimurium ΔaroA CFU in 100 μL PBS. At various time points, tissues were collected and homogenized in sterile PBS; serial dilutions of the homogenates were plated on LB agar plates (100 μg/mL streptomycin) for bacterial enumeration. For IL-17A neutralizing experiments, mice were injected intraperitoneally with 200 μg of antibodies against IL-17A (Janssen, CNTO 8096), or isotype control (Janssen, CNTO 2407) three times per week beginning at 7 days after infection, and sacrificed at day 21 pi for tissue collection.

Histology and immunohistochemistry

Formalin fixed and paraffin embedded tissues were cut into 5-μm sections for Masson’s trichrome (MT) staining. Cecal pathology was scored as previously described (25). For immunostaining, tissue sections were deparaffinized, rehydrated, and underwent antigen retrieval; antibodies against type 1 collagen (ABcam), type 3 collagen (Fitzgerald), vimentin (ABcam), and desmin (Santa Cruz) were used, followed by incubation with Alexa Fluor-conjugated secondaries (Life Technologies). Sections were then mounted using ProLong Gold Antifade with DAPI (Life Technologies). Optical z-stack images were captured on a Leica SP5X confocal microscope and analyzed using ImageJ.

Quantitative RT-PCR

Total RNA extracted from terminal ends of cecal tissue using Trizol (Invitrogen) was reverse transcribed with a high-capacity cDNA RT kit (Thermo). Quantitative real-time PCR was performed using SYBR green chemistry (KAPA) and gene-specific primer pairs (supplemental table 1) on an AB7900 RT-PCR system.

Cytokine detection

Concentrations of TNF, IL-6, MCP-1, IFN-γ, IL-12p70, IL-10, and IL-17A in cecal homogenates were determined using a cytometric bead array (BD Biosciences); TGF-β1 and IL-22 levels were determined by plate ELISA kits (eBiosciences) according to manufacturer’s instructions.

Immune analysis and flow cytometry

Mesenteric lymph nodes were excised and passed through a 70-μm filter to generate single cell suspensions. Leukocytes were isolated from ceca by flushing luminal contents, mincing and incubating tissues in 1.5 U/mL collagenase D and 2.4 U/mL dispase II (Roche) for 30 minutes at 37°C with gentle rotation; samples were then passed through a 70 μm filter and hematopoietic cells were enriched by Percoll (GE) separation. Cells were restimulated using 50 ng/mL PMA and 750 ng/mL ionomycin (Sigma) in the presence of brefeldin A (eBiosciences) for 4 hours. Samples were then incubated with 5 μg/mL anti-CD16/32 (2.4G2, in house) to block non-specific antibody binding. Fluorescence-conjugated antibodies to CD45.2 (104), CD8 (53.67), CD11b (M1/70), CD11c (N418), CD19 (1D3), NK1.1 (PK136), Gr1 (RB6-8C5), Ter119, CD3e (145-2C11), CD90.2 (30-H12), KLRG1 (2F1), Sca-1 (D7), IL-17A (eBio17B7), and IL-22 (1H8PWSR) were used. A fixation and permeabilization buffer set (eBiosciences) was used for intracellular cytokine staining and a fixable viability dye (eBiosciences) was used for dead cell exclusion. Data was acquired on a BD LSRII, and analyzed using FlowJo.

Statistics

Results are presented as mean +/− SEM; statistical significance was determined by unpaired Student’s t test.

Supplementary Material

Supplementary Data and Figures

Fig. S1. Analysis of type 2 immune responses in Rora^sg/sg BMT mice following Salmonella infection.

Fig. S2. Gating strategy for flow cytometric analysis.

Fig. S3. Isotype control for immunohistochemical tissue staining.

Fig. S4. Summary of effects of RORα deficiency on ILC2, ILC3 and Th17 cell development and function.

Table S1. Primer sequences (5′-3′) used for transcript detection by quantitative RT-PCR.