Myeloid differentiation primary response gene (MyD) 88 signalling is not essential for intestinal fibrosis development

Dysregulation of the immune response to microbiota is associated with inflammatory bowel disease (IBD), which can trigger intestinal fibrosis. MyD88 is a key component of microbiota signalling but its influence on intestinal fibrosis has not been clarified. Small bowel resections from donor-mice were transplanted subcutaneously into the neck of recipients C57BL/6 B6-MyD88tm1 Aki (MyD88−/−) and C57BL/6-Tg(UBC-green fluorescence protein (GFP))30Scha/J (GFP-Tg). Grafts were explanted up to 21 days after transplantation. Collagen layer thickness was determined using Sirius Red stained slides. In the mouse model of fibrosis collagen deposition and transforming growth factor-beta 1 (TGF-β1) expression was equal in MyD88+/+ and MyD88−/−, indicating that MyD88 was not essential for fibrogenesis. Matrix metalloproteinase (Mmp)9 expression was significantly decreased in grafts transplanted into MyD88−/− recipients compared to MyD88+/+ recipients (0.2 ± 0.1 vs. 153.0 ± 23.1, respectively, p < 0.05), similarly recruitment of neutrophils was significantly reduced (16.3 ± 4.5 vs. 25.4 ± 3.1, respectively, p < 0.05). Development of intestinal fibrosis appears to be independent of MyD88 signalling indicating a minor role of bacterial wall compounds in the process which is in contrast to published concepts and theories. Development of fibrosis appears to be uncoupled from acute inflammation.

has an effect on the risk of IBD 11 . Equally, for components downstream of MyD88 studies show an association between signalling dysfunction and IBD 12,13 . MyD88 can be viewed as a focal point for microbiota signalling involved in the signal transduction of almost all TLRs and causing inflammation. Studies have identified patients with mutations in MYD88 that abolish protein production and result in a primary immunodeficiency syndrome characterized by greater susceptibility to pyogenic bacteria [14][15][16] . Three different autosomal recessive mutations have been described in MYD88-deficient patients suffering from recurrent infections with pyogenic bacteria. Life-threatening infections in subjects with defective MyD88 first occur during early infancy 14 .
Thus, one would also expect an association between MyD88 polymorphisms or its dysregulated expression and a susceptibility to IBD. A total of 163 IBD loci that meet genome-wide significance thresholds were found by analysis of data from 15 genome-wide association studies of CD and UC combined with Immunochip validation 17 . However, disease-associated MyD88 polymorphisms were not described. Further studies showed that the MyD88 polymorphism rs7744 is closely associated with the development of UC and that this polymorphism may be linked with the response to UC therapeutic approaches 18 . The function of rs7744 has not been clearly identified, although minor allele variants is thought to lead to an overexpression of the MyD88 protein 18 , considering the increased mRNA and protein levels of MyD88 in UC patients 19 .
Fibrosis is increasingly being recognised as an important cause of morbidity and mortality in IBD. While the triggering of fibrosis seems to require inflammation as a prerequisite step, the mechanisms involved in the progression of fibrogenesis may be distinct. There is evidence that intestinal fibrogenesis after being initiated is -at least to some extent -a self-perpetuating process 20 . Administration of potent anti-inflammatory agents effectively treats inflammatory flares, but at present there is no specific treatment option for patients with recurrent intestinal fibrosis 21,22 . Notable progress has been made in elucidating therapeutic targets to tackle the inflammatory component of IBD, but the incidence of strictures and of subsequent surgical interventions remains relatively unchanged 23 .
In this current study we describe the role of the MyD88-dependent innate immunity pathway in the fibrogenic process. We applied the rapid and reliable heterotopic transplantation model of intestinal fibrosis 24 to MyD88-deficient mice. Our data suggests that the development and progression of fibrosis in the intestine is not dependent on MyD88 signalling but alludes to the presence of alternative pathways which contribute to this process. This suggests that anti-inflammatory agents -although essential to treat inflammatory flares -are not a sufficient treatment option against IBD-associated fibrosis.

Materials and Methods
All experiments were performed in accordance with relevant guidelines and regulations.
Animals. C57BL/6 B6-MyD88tm1 Aki (MyD88 −/− ) and C57BL/6-Tg(UBC-green fluorescence protein (GFP))30Scha/J (GFP-Tg) mice weighing 20 g were obtained from Jackson Laboratories and bred locally. The animals received standard laboratory mouse food and water ad libitum. They were housed under specific pathogen-free conditions in individually ventilated cages. The experimental protocol was approved by the local Animal Care Committee of the University of Zurich (registration numbers 114/2011 and ZH183/2014).

Heterotopic intestinal transplant model. The heterotopic mouse intestinal transplant model is an adap-
tion of the transplantation model of intestinal fibrosis in rats. Both have been previously described in detail 24,25 . In short, donor small bowel was resected and transplanted subcutaneously into the neck of recipient animals. Intestinal grafts were explanted up to 21 days after transplantation. At explantation, each graft was divided into three equal segments. One segment was fixed in 4% formalin and prepared for histopathological assessment. The other segments were snap frozen in liquid nitrogen and stored at −80 °C until RNA extraction. RNA isolation, cDNA synthesis and real-time PCR. For isolation of total RNA, lysis buffer from the RNeasy ® kit (Qiagen, Hilden, Germany) was added to snap frozen resections. Samples were shredded in M tubes (# 130-093-236, Miltenyi Biotec) in a gentle MACS TM dissociator (Miltenyi Biotec). Total RNA was prepared according to the manufacturer's protocol and stored at −80 °C. Quality of total RNA was controlled with the Agilent 2100 expert EukaryoteTotal RNA Nano Assay. Only RNA with a RNA integrity number > 6 was used for analysis. The isolated RNA was reverse-transcribed using a Reverse Transcription System (Applied Biosystems).
Sirius Red-stained slides were analyzed by bright-field microscopy with an additional polarizing filter. Under polarized light Sirius Red-stained collagen assumes a palette of colors ranging from green to red based on the fibrotic maturation process. Collagen layer thickness was determined by an investigator blinded to the experiment. Thickness was measured in at least eight places in representative areas at 10-fold magnification. Similarly slides were stained with Elastica van Gieson and analyzed using transmission light microscopy. Collagen layer thickness was determined as previously described for Sirius Red staining.
SCIEnTIfIC RepoRtS | 7: 17678 | DOI:10.1038/s41598-017-17755-7 Statistical analysis. Statistical analysis for real time PCR and collagen layer thickness was performed using Kruskal-Wallis one way analysis of variance on ranks, all pairwise multiple comparison procedures, Dunn's method. Comparison between genotypes was performed using the Mann-Whitney-Test for each time point. Differences were considered significant at a p-value of <0.05 (*) and highly significant at a p-value of < 0.01 (**) and p-value of < 0.001 (***).

Ethical considerations. The experimental protocol for animals was approved by the local Animal Care
Committee of the University of Zurich (registration numbers 114/2011 and ZH183/2014).

Development of intestinal fibrosis is not prevented in grafts extracted from GFP-Tg donor animals and transplanted into MyD88 −/− recipients.
To determine the relevance of MyD88 in the development of fibrosis, MyD88 −/− and GFP-Tg mice were used as both donors and recipients for isogeneic transplantation in our heterotopic animal model of intestinal fibrosis. The presence of endogenous GFP in GFP-Tg donor or recipient mice allowed for a genotype-specific staining. Body weight remained unchanged in both MyD88 −/− and GFP-Tg recipients (not shown). Grafts were explanted up to 21 days after transplantation as indicated. 26 isogeneic transplants (GFP-Tg into GFP-Tg mice), 58 isogeneic transplants (GFP-Tg into MyD88 −/− mice) and 47 isogeneic transplants (MyD88 −/− into GFP-Tg mice) were performed. Of 131 transplants histologically evaluable tissue was recovered from all but 7 grafts. To compare grafts with non-transplanted tissue 28 small bowel resections were extracted from GFP-Tg mice and six small bowel resections were extracted from MyD88 −/− mice.
Exaggerated collagen deposition is a histologic and molecular hallmark of human intestinal fibrosis, therefore Col1a1 mRNA expression was analyzed by real-time PCR (Fig. 1A). The development of intestinal fibrosis after heterotopic transplantation was also observed in histological cross sections. Collagen production and deposition in the intestinal transplants was determined by Sirius red staining under transmission and polarizing light microscopy (Fig. 1B). In Sirius red-stained histological cross-sections freshly isolated small intestine was characterized by an open lumen and distinctive epithelial crypts (Fig. 1B). At day 7 after transplantation the lumen of intestinal grafts was obstructed by granulation tissue and fibrotic material. Freshly isolated small intestine was characterized by mostly long-chained collagen (red stain) adjacent to the submucosa. The presence of short-chained collagen (green and yellow stain) increased continuously in the submucosa and in the luminal occlusion after transplantation. Loss of epithelial structures and luminal occlusion was observed irrespective of the genotype of donor and recipient mice. Time dependent increase in collagen was also confirmed by Elastica van Gieson (EvG) staining (Supplementary Figure 1).
The collagen layer thickness in harvested grafts was significantly increased in comparison to the collagen layer thickness in freshly isolated small intestine (Fig. 1C, p < 0.01 (**), calculated from n = number of grafts as indicated). Collagen layer thickness is comparable between grafts in both MyD88 −/− and GFP-Tg recipients over time.
Myd88 is not required for an increase in Hif1α expression after heterotopic small bowel transplantation. MyD88 signaling regulates Hif-1α expression 26 . Therefore Hif1α mRNA expression was analyzed by real-time PCR to determine hypoxia-mediated signalling (Fig. 2). Surprisingly Hif1α mRNA was unchanged for at least 48 hours in grafts of both genotypes. This was followed by a significantly increased Hif1α mRNA expression after 4 days in isogeneic transplants of both GFP-Tg into MyD88 −/− mice as well as MyD88 −/− into GFP-Tg mice compared to freshly isolated small intestine (Fig. 2, 2.90 ± 0.89, vs. 1.00 ± 0.01 and 3.60 ± 0.52 vs. 0.77 ± 0.23, respectively, * p < 0.05, n = 3 each column). The development of Hif1α mRNA expression over time is similar in grafts of both genotypes.
Previously, we could show that macroscopically visible blood vessels from the surrounding tissue extend towards the graft where they form a dense vascular network. In this study we determined the origin of the lumen-obstructing cells in isogeneic transplants (MyD88 −/− into GFP-Tg mice). As expected, the freshly isolated intestinal resections from MyD88 −/− mice are negative for GFP staining ( Supplementary Figure 2A and B). An increase in infiltration of stained GFP cells ensued in a time-dependent manner. Initially, red-stained cells were visible exclusively in the submucosa and seemed to accumulate along the collagen layer. Subsequent time-points showed progressive infiltration along the crypts and into the lumen. As seen in GFP staining, the infiltrating cells frequently enter the graft not in a diffuse, but rather in a more localized fashion, with specific points of access forming along the graft wall (Supplementary Figure 2B). They then pour across the tissue and into the lumen contributing to the total occlusion of the graft. Seven days after transplantation GFP positive blood vessels appear within the GFP negative donor tissue (Supplementary Figure 2C). This result demonstrates an increased infiltration to the graft via blood vessels irrespective of MyD88-driven signalling.

MyD88 deficiency reduces intestinal inflammation but not the development of intestinal fibrosis.
MyD88-signalling is an inducer of MMP-9 expression 27,28 . As MMPs are involved in mucosal fibrosis we determined the expression of tissue remodeling protease MMP-9 and the tissue inhibitor of metalloproteinases TIMP-1 in intestinal transplants by real-time PCR. Mmp9 mRNA expression was increased in a time-dependent manner and was significantly increased in grafts from both GFP-Tg donors transplanted into GFP-Tg recipients and MyD88 −/− donors transplanted into GFP-Tg recipients at day 14 compared to grafts from GFP-Tg donors transplanted into MyD88 −/− recipients at day 14 (Fig. 4A, 133.0 ± 24.0 and 153.0 ± 23.1 vs. 0.15 ± 0.10, respectively, p < 0.05 (*), n = 3 in each column). In contrast, increase of TIMP-1 mRNA expression is similar between grafts of both genotypes over time (Fig. 4B, n = 3 each column).
As MMP-9 is produced by neutrophils in large quantities we determined if Mmp9 mRNA expression coincided with the accumulation of Ly-6G positive neutrophils and performed the relevant immunohistochemistry. As expected, Ly-6G staining was more pronounced in grafts from MyD88 −/− donors transplanted into GFP-Tg recipients at day 7 and 14 compared to grafts from GFP-Tg donors transplanted into MyD88 −/− recipients at day 7 (Fig. 4C) and 14 (Supplementary Figure 3). A significantly increased number of Ly-6G positive neutrophils was determined in grafts from MyD88 −/− donors transplanted into GFP-Tg recipients at day 7 and 14 compared to freshly isolated small bowel (Fig. 4C, 25.4 ± 3.1 and 24.6 ± 10.8 vs. 0.3 ± 0.7, respectively, * p < 0.05, n = as indicated). Further a significantly reduced number of neutrophils in grafts from GFP-Tg donors transplanted into MyD88 −/− recipients at day 7 and 14 compared to grafts from MyD88 −/− donors transplanted into GFP-Tg recipients confirmed the hypothesis that large quantities of MMP-9 were produced by neutrophils (Fig. 4C, 16.3 ± 4.5 vs. 25.4 ± 3.1 and 9.1 ± 5.7 vs. 24.6 ± 10.8, respectively, * p < 0.05, n = as indicated). Accumulation of neutrophils was therefore significantly increased in heterotopic intestinal grafts transplanted into mice capable of MyD88 signalling compared to recipients deficient in MyD88 signalling. Therefore, MyD88 deficiency reduces intestinal inflammation, at least with respect to Mmp9 expression and neutrophil infiltration whereas development of fibrosis is similar in grafts of both MyD88 deficient and wildtype mice.

Discussion
In this study we investigated the role of MyD88 in the development and progression of fibrosis in an animal model of intestinal fibrosis. All hallmarks of intestinal fibrosis were similar in wildtype and MyD88 −/− animals. Specifically, we found increased collagen expression, which intensified over time, virtually identical in both MyD88 −/− and GFP-Tg grafts. Increased collagen deposition in the intestinal wall over time was detected by histological analysis and confirmed the real time PCR data. EvG and Sirius red staining demonstrated an identical network of progressively shorter-chain collagen fibrils accumulating in an IBD-mimicking fashion in grafts from both genotypes. We also found an increase in Tgfβ1 expression over time in both grafts. Histological staining confirmed the increase in TGF-β1 in the obliterated lumen and the surrounding tissue. In addition, real-time PCR revealed a time-dependent increase of Timp1 mRNA in both grafts.
In contrast, pronounced differences between wildtype and MyD88 −/− animals could also be observed. While Mmp9 expression was increased upon development of fibrosis we did observe a significant difference between the grafts from both genotypes. MyD88-signalling is a potent inducer of MMP-9 expression 27,28 . Likewise, GFP-Tg grafts transplanted into MyD88 −/− recipients yielded significantly lower expression levels than the MyD88 −/− grafts in GFP-Tg recipients. MMP-9 is produced plentifully by neutrophils 29 , suggesting that the recipient cells infiltrating the transplant are mainly responsible for the production of this mediator. Histological staining confirms a significantly increased number of Ly-6G positive neutrophils in the obliterated lumen in grafts from MyD88 −/− donors transplanted into GFP-Tg recipients as compared to grafts from GFP-Tg donors transplanted into MyD88 −/− recipients. MMP-9 is considered as a critical component of intestinal inflammation in IBD, but less is known concerning the role of MMP-9 in penetrating or stricturing CD. Recently, we have generated an anti-MMP-9 antibody, CALY-001, and have evaluated its efficacy in fibrogenesis. We showed that inhibiting MMP-9 significantly reduces collagen deposition and intestinal tissue remodelling 30 . This anti-fibrotic effect, associated with the anti-inflammatory effect of MMP-9 inhibitor described in the DSS-colitis model by Marshall and colleagues using an alternative anti-MMP-9 antibody, AB-0046 31 , supports the development of an anti-MMP-9 antibody in patients with penetrating CD.
Several animal models for the study of intestinal fibrosis have been proposed 32 . All of them have some advantages as well as disadvantages and none of them really resembles intestinal fibrosis of CD patients. Intramural injection of peptide glycan-polysaccharide complex at laparotomy has been shown to cause intestinal fibrosis in rats with immunopathological features resembling CD 33 . In this model, bacterial cell wall polymers were shown to stimulate collagen and TGF-b1 expression in intestinal myofibroblasts 34 . In the trinitrobenzene sulfonic acidinduced model of colitis, rats develop colonic fibrosis and stricture formation, but this has not yet been well characterized 35 . Our animal model of intestinal fibrosis may have its own limitations.The observed increase in neutrophils in the heterotopic transplantation model may be affected by immune response to intestinal grafts, rejection, ischemia or hypoxia which contributes to fibrosis in the animal model but not necessarily to fibrosis in IBD patients. Differences in the microbiome are likely to influence development of intestinal fibrosis. The intestinal microbiota that lead to a response by the innate immune system in IBD cause inflammation and may trigger fibrogenesis. However, there are conflicting reports on the importance of the MyD88-dependent signalling in the development of intestinal fibrosis. S. typhimurium ranges among the most prevalent agents of foodborne diseases and provokes severe inflammatory responses and thus may initiate fibrosis. Infection studies with streptomycin pretreated mice showed evidence for MyD88-dependent and MyD88-independent pathways in S. typhimurium colitis 36 . S. typhimurium was shown to efficiently colonize the cecal lumen and to cause the same level of inflammation in both MyD88 +/+ and MyD88 −/−36,37 . S. typhimurium can efficiently induce inflammation even in the absence of MyD88 signalling. Intestinal dendritic cells are believed to sample and present not only S. typhimurium but all commensal bacteria to the gut-associated immune system. MyD88 signaling in mucosal dendritic cells is not required for induction of colitis by S. typhimurium 38 . In contrast, using a less virulent S. typhimurium strain in infection studies MyD88 −/− mice displayed a decrease in collagen and mucosal inflammation in ceca compared to wildtype mice 37 . Submucosal inflammation and fibroblast accumulation was similar in the infected MyD88 −/− and wildtype mice 37 and the authors conclude that fibrosis is not simply a secondary effect of local inflammation and tissue damage. Studies where S. typhimurium infection in mice was eradicated with antibiotics showed evidence that fibrogenesis is preceded by inflammation but eradication of the inflammatory stimulus represses inflammation without preventing fibrosis 20 . This suggests that intestinal fibrosis does not require persistent microbal stimulation and could be independent from the MyD88 pathways from a certain time point onwards.
There are also conflicting reports on the importance of the MyD88-dependent signalling in the development of fibrosis in other organs. Kidney fibrosis studies using various murine models have shown MyD88 to be a key component of this disease, whereby its absence coincided with reduced inflammation and attenuated collagen deposition 39,40 . MyD88-activation was linked with a shift in the immune response towards T helper type 2 differentiation and M2 polarisation of macrophages 39 . This pattern has previously been implicated in the development and progression of fibrosis 41,42 . Confirmatively, MyD88 signalling was found to be essential for the inflammation and fibrosis in a bleomycin-induced model of pulmonary disease 43 . The TLR4-triggered MyD88 pathway was found to be crucial to hepatic fibrosis by potentiating pro-fibrotic TGF-β1 signalling 44 . In contrast, in other studies using different models of fibrosis MyD88 signalling is less relevant for the development and progression of fibrosis. For instance, Leptospira interrogans induces fibrosis in mouse kidney in a MyD88-independent fashion 45 . Similarly, a recent study showed that MyD88, while relevant for the inflammatory process, plays only a minor role in a silica model of lung fibrosis 46 . Critically, MyD88-deficient mice presented a pronounced accumulation of pro-fibrotic cytokines and did not manifest a tempering of collagen deposition.
TLRs and NOD-like receptors are responsible for sensing local microbiota 2 . TLR are not only expressed in macrophages, dendritic cells and intestinal epithelial cells but also in primary colonic myofibroblasts 47 , one of the main producers of extracellular matrix components that are deposited during fibrosis. In mice, MyD88 is essential for the inflammatory responses mediated by almost all TLR family members 48 as MyD88 −/− show no responses to the TLR2 ligands peptidoglycan and lipoprotein 49,50 , to the TLR4 ligand lipopolysaccharide (LPS) 51 , to the TLR5 ligand flagellin 52 , to the TLR7 ligand imidazoquinoline 53 and to the TLR9 ligand CpG DNA 54,55 . On one hand MyD88 signalling is required to limit the bacterial burden and mice with a Myd88-deficiency have increased susceptibility to bacterial and viral infections. On the other hand MyD88-independent TLR signaling pathways have also been identified [56][57][58] .
Furthermore wound healing is impaired in MyD88-deficient mice. Adequate wound healing is the result of an exquisite physiological balance between multiple pro-and anti-fibrotic stimuli on extracellular matrix-producing cells [59][60][61][62] . Excessive tissue repair promotes fibrosis and impairs gastrointestinal function and is a common clinical problem in patients with IBD.
The data presented here suggest that although inflammation is reduced by MyD88 deficiency, the development of intestinal fibrosis is uncoupled from MyD88 signalling. MyD88 is not the only adaptor protein involved in Toll/IL-1 receptor family signalling. Stimulation of TLR3 and TLR4 in the absence of MyD88 still leads to the activation of NF-kB, albeit with delayed kinetics 51 . Interestingly, TLR4 polymorphisms have been associated with an increased susceptibility for IBD 63,64 . The MyD88-independent TLR4 response occurs via the adaptor TIR domain-containing adaptor inducing IFN-beta (TRIF) which then triggers the interferon regulatory factor 3 (IRF3) 57,65 or NF-kB via RIP 66 . Furthermore, TLRs are not the only pattern recognition receptors involved in the innate immune response. Intracellular NOD/CARD-like receptors are also able to bind PAMPs 67 and have been shown, in the case of NOD2/CARD15, to be strictly linked to CD 68 . However, even considering the presence of these alternative pathways, the acute inflammatory response will be strongly reduced without a functional MyD88, as exemplified by a strongly reduced accumulation of neutrophils in our experiments. Development of fibrosis thus appears to be uncoupled from inflammation, contradicting published concepts and theories. This suggests that anti-inflammatory agents -although essential to treat inflammatory flares -are not a sufficient treatment option against IBD associated fibrosis.