Pathogenic and protective roles of MyD88 in leukocytes and epithelial cells in mouse models of inflammatory bowel disease

Mice

C57BL/6, C57BL/6MyD88^−/−, C57BL/6RAG1^−/− (RAG1^−/−), 129SvEv, 129SvEvMyD88^−/−, 129SvEvRAG2^−/− (RAG2^−/−) and 129SvEvRAG2^−/−MyD88^−/− mice were bred and maintained under specific-pathogen free conditions. C57BL/6MyD88^−/− mice were backcrossed onto a 129SvEvRAG2^−/− background using a microsatellite marker-assisted speed congenic approach ²⁴. 129SvEvRAG2^−/−MyD88^+/− mice from the F5 and F6 generations were intercrossed to produce 129SvEvRAG2^−/−MyD88^−/− and littermate controls. Experiments were performed in accordance with the UK Animal Procedures Act (1986).

Bone marrow chimeras

Bone marrow was isolated from either 129SvEvRAG2^−/− or 129SvEvRAG2^−/− MyD88^−/− mice, and 5 × 10⁶ cells injected IV into gamma-irradiated (5.5Gy, 550 rad) 129SvEvRAG2^−/− mice, which were left to reconstitute for 12wk.

Induction and regulation of innate immune typhlocolitis

Induction of T cell dependent colitis in C57BL/6 mice

C57BL/6 mice were infected with Helicobacter hepaticus as above and injected with 1 mg anti-IL-10R monoclonal antibody IP on days 0, 7, 14 and 21 post-infection ¹⁷. Mice were sacrificed and assessed for disease on day 28 post-infection.

Induction of IBD by adoptive transfer of naive CD4⁺ T cells

CD4⁺ T cells were purified from spleens and labeled with a cocktail of fluorescent antibodies as described ^{18, 25}. Naïve (CD4⁺CD45RB^HI) and regulatory (CD4⁺CD45RB^LOCD25⁺) T cell populations were isolated using a Mo-Flo cell sorter (DakoCytomation) to >99% purity. For colitis induction, C57BL/6RAG1^−/− mice were adoptively transferred with 4 × 10⁵ CD4⁺CD45RB^HI T cells IP and sacrificed upon the onset of symptoms of severe colitis (>6 wks post-transfer).

Assessment of intestinal inflammation

Samples of cecum and colon were fixed in buffered 10% formalin solution. 4 μm paraffin-embedded sections were stained with H&E and inflammation evaluated in a blinded fashion using a scoring system described previously ²⁵; where 0-3 = normal; 4-6 = mild inflammation; 6-9 = moderate inflammation; 9-12 = severe inflammation. Colon scores represent the mean score of the proximal, mid and distal sections.

Quantitation of Helicobacter hepaticus colonization

Upon sacrifice, cecal contents were immediately collected in PBS and stored at −20°C. DNA was purified from cecal contents with the DNA Stool Kit (QIAGEN) as per manufacturer’s instructions. Hh DNA was quantitated using a Q-PCR method based on the cdtB gene ¹⁴.

Immunofluorescent staining of colon sections

Colon sections were snap frozen in OCT compound (Tissue-Tek) and cryosectioned to 6μm. Slides were fixed in 2% paraformaldehyde solution, washed with PBS and blocked with PBS containing 3% goat serum ,1%BSA and 0.3% Triton X-100 (Sigma Aldrich)). Following washing, sections were stained with a polyclonal anti-Hh serum (rabbit) followed by Alexa 633 conjugated goat anti-rabbit IgG (Invitrogen). Sections were co-stained with FITC-conjugated mouse anti-E-Cadherin (BD Biosciences) and mounted in ProLong Gold Antifade Reagent with DAPI (Invitrogen). Sections were imaged using a 63× oil immersion lens on a Zeiss LSM510 confocal microscope. Data were analyzed using LSM 5 Image Software (Zeiss).

Quantitation of cytokines and antimicrobial peptides (AMP) in colonic tissue

Colonic protein and RNA was isolated from frozen colonic tissue as described ¹⁸. Protein concentrations were measured using cytometric bead assay (BD Biosciences) (TNFα and IFNγ) or with Luminex 100 assay (Bio-Rad Laboratories) (IL-1β) as described ¹⁸. Real-Time PCR using a Chromo4 detection system (MJ Research) was used to determine colonic cytokine mRNA levels as described ¹⁸. Taqman Gene Expression Assays (Applied Biosystems) were used for the quantification of antimicrobial peptides RegIIIβ, RegIIIγ and S1008a.

Flow cytometry

Preparations of spleen, mesenteric lymph node and lamina propria leukocytes were isolated as described previously ²⁶ and enumerated by hemocytometer. For analysis of innate leukocytes, cells were stained with fluorescent antibodies against CD11b, Gr1 (Ly6G) and CD11c ^{14, 18}. For phenotypic characterization of lymphocytes, cells were surface stained with fluorescent antibodies against CD4, TcRβ and CD25, followed by intracellular staining with fluorochrome-labeled anti-mouse FoxP3 or isotype control (eBioscience) as described ^{18, 25}. Data were analysed with FlowJo software (Tree Star, Inc.).

Statistics

Statistical analysis was performed with Prism 5.0 (GraphPad Software). The non-parametric Mann Whitney test was used for all statistical comparisons, with the exception of log-rank (Mantel-Cox) test used for Kaplan-Meier plots. Differences were considered statistically significant at P < 0.05.

Helicobacter hepaticus (Hh)-induced innate immune intestinal inflammation is MyD88-dependent

Infection of lymphocyte-deficient 129SvEvRAG2^−/− (RAG2^−/−) mice with Hh triggers potent intestinal and systemic innate inflammation ¹⁴. To assess the role of MyD88 signaling in chronic innate inflammatory responses in the gut, we generated RAG2^−/−MyD88^−/− double-knock out mice. Even when maintained under strict SPF conditions, the RAG2^−/−MyD88^−/− mice exhibited poor fitness with a progressive mortality in excess of 70%, as opposed to the 100% survival of RAG2^−/−MyD88^+/− littermate controls (Figure 1A). RAG2^−/−MyD88^−/− mice died suddenly without signs of wasting disease and did not have overt signs of immune pathology. Despite this issue, we were able to infect a sufficient number of RAG2^−/−MyD88^−/− mice to assess the role of MyD88 signaling in the development of innate immune IBD.

In contrast to MyD88-sufficient littermate controls, Hh-infected RAG2^−/−MyD88^−/− mice displayed no cecal or colonic inflammation (Figure 1B,C), demonstrating that Hh-driven innate inflammation is strictly MyD88-dependent. We also found Hh-triggered activation of systemic innate inflammation was completely MyD88-dependent, as shown by the absence of splenomegaly and granulocyte accumulation in RAG2^−/−MyD88^−/− mice (Figure 2A-D).

We hypothesized that in the absence of MyD88-driven pro-inflammatory signals, infected mice might exhibit outgrowth of Hh, as has been reported following infection of MyD88^−/− mice with intestinal pathogen Citrobacter rodentium ⁸. However, we found that RAG2^−/−MyD88^−/− mice exhibited decreased Hh colonization (Figure 3). The increased Hh colonization observed in MyD88-competent mice suggests intestinal inflammation may confer a competitive advantage to Hh, as has been described for other enteric pathogens ^{27, 28}.

Hematopoietic MyD88 signaling is required for innate intestinal inflammation

Intestinal epithelial cells and leukocytes of the lamina propria express TLRs and MyD88 and much current debate has focused on their relative contribution to the development of intestinal inflammation ^{29, 30}. We therefore created bone marrow chimeras to selectively deplete MyD88 from the hematopoietic cellular compartment. Thus, we irradiated RAG2^−/− mice and reconstituted them with bone marrow from RAG2^−/−MyD88^−/− or RAG2^−/− control donor mice. Bone marrow chimeras were then infected with Hh and assessed for disease development.

We found a complete absence of intestinal inflammation in the chimeric mice with selective ablation of MyD88 in hematopoietic cells, as opposed to the severe typhlocolitis observed in MyD88-competent bone marrow chimeras (Figure 4A,B). Following Hh infection, hematopoietic MyD88^−/− chimeras were resistant to crypt hyperplasia, goblet cell depletion and leukocytic infiltrate of the lamina propria indicating that epithelial cell MyD88-signaling alone was insufficient to evoke the innate inflammatory response to Hh. Moreover, levels of pro-inflammatory cytokines including TNFα, IFNγ, IL-1β, IL-23p19, IL-17A, IL-22, and IL-6 were significantly reduced in the colons of Hh infected chimeric mice lacking hematopoietic MyD88 signaling (Figure 4C). Concordant with the absence of intestinal inflammation, mice with selective depletion of MyD88 from hematopoietic cells were also refractory to Hh-driven systemic inflammation (Supplementary Figure 1). As with the complete RAG2^−/−MyD88^−/− mice, the absence of Hh-driven disease was not due to a lack of Hh colonization and lack of MyD88-signaling from the hematopoietic compartment did not lead to Hh outgrowth or to increased epithelial colonization (Figure 5A, Supplementary Figure 2). Together, these findings reveal an indispensable role for MyD88-signaling receptors upon hematopoietic cells for the induction of innate inflammatory responses to Hh.

Due to the poor fitness of RAG2^−/−MyD88^−/− mice (Figure 1A), we were unable to perform the inverse selective ablation of MyD88 in non-hematopoietic, radioresistant cells. Notably, in stark contrast to complete RAG2^−/−MyD88^−/− mice (Figure 1A) selective depletion of MyD88 from the hematopoietic compartment did not predispose to mortality, either prior to, or following, Hh infection (Figure 5B). Together, these findings suggest a critical role for epithelial MyD88 signals for host survival under specific pathogen-free conditions. We therefore examined expression of the antimicrobial peptides (AMP) RegIIIβ, RegIIIγ and S100A8 in the colons of mice with or without intact epithelial MyD88 signaling following infection with Hh. Whereas Hh-infected RAG2^−/−MyD88^−/− mice exhibited a marked reduction in AMP expression as compared to RAG2^−/− controls (Figure 5C), chimeric mice selectively lacking MyD88 in the hematopoietic compartment alone maintained normal AMP expression following Hh infection (Figure 5D). Thus, epithelial MyD88 signaling efficiently drives AMP expression in the colon and this correlates with survival under SPF conditions.

T cell-dependent intestinal inflammation is MyD88-dependent

Following infection with Helicobacter hepaticus, C57BL/6 mice mount a range of Hh-specific CD4⁺ T cell responses, comprising both regulatory and effector T cell populations ³¹. Blockade of IL-10 signaling by treatment with anti-IL-10R blocking mAb disrupts the dominant effects of regulation, revealing colitogenic CD4⁺ T responses that mediate chronic intestinal inflammation ¹⁷. We used this model to address whether MyD88 was required for the development of such pathogenic T cell responses in the gut. As expected, infection with Hh alone did not provoke any intestinal pathology in WT or MyD88^−/− mice (Figure 6A,B). However, wild type mice infected with Hh and concomitantly treated with anti-IL-10R mAb developed severe typhlocolitis that was characterized by epithelial hyperplasia, edema, crypt abscesses and dense leukocytic infiltrates (Figure 6A-C). By contrast, MyD88^−/− mice infected with Hh and treated with anti-IL-10R mAb exhibited no significant signs of cecal or colonic inflammation (Figure 6A-C).

Similarly, both splenomegaly and granulocyte accumulation induced by Hh + anti-IL-10R mAb treatment of immune competent mice were completely MyD88-dependent (Figure 6D,E). As uninfected wild type and MyD88^−/− mice harbour comparable numbers and proportions of splenic granulocytes (³²and data not shown) this resistance to systemic inflammation is due to failure to activate and/or recruit these inflammatory leukocytes rather than any intrinsic neutropenia. These findings demonstrated that both mucosal and systemic inflammatory responses to Hh were MyD88 dependent.

It was recently reported that MyD88 signaling within T cells was required for their pathogenic function in the intestine ³³. However, a subsequent study found no such requirement for T cell intrinsic MyD88 signals for intestinal pathology ³⁴. Therefore, using the classical ‘T cell transfer’ model of colitis, we compared the ability of naïve CD45RB^HI CD4⁺ T cells isolated from wild type or MyD88^−/− mice to induce colitis upon adoptive transfer to lymphopenic C57/BL6RAG1^−/− recipients. Consistent with the findings of Tomita et. al. ³⁴, we found MyD88 signaling by T cells was dispensable for the induction of T cell transfer colitis (Supplementary Fig 3), with equivalent levels of disease incidence and severity induced following adoptive transfer of wild type or MyD88^−/− CD45RB^HI CD4⁺ T cells. This suggests that the resistance of MyD88^−/− mice to Hh-driven T cell-dependent inflammation is due to a defect in innate leukocyte activation, rather than intrinsically hypo-responsive T cells.

Intrinsic MyD88 signals are dispensable for the suppression of immune pathology by regulatory T cells

A number of studies have reported that regulatory T (Treg) cells selectively express TLR and that TLR agonists may directly inhibit CD4⁺CD25⁺ Treg suppressive function ^21-23. However, given that it was recently reported that MyD88^−/− Treg lacked regulatory function in vivo ³³, whether TLR/MyD88 signaling positively or negatively modulates Treg remains contentious.

We have shown previously that CD4⁺CD25⁺ Treg cells suppress Hh-driven intestinal inflammation ¹⁴. Therefore, we examined whether MyD88 signaling by Treg was required for this regulation. Immediately following infection with Hh, cohorts of RAG2^−/− mice received either wild type or MyD88^−/− CD4⁺CD25⁺ Treg cells and were sacrificed ten weeks later. We found that mice transferred with either wild type or MyD88^−/− CD4⁺CD25⁺ Treg cells exhibited comparable attenuation of Hh driven typhlitis and colitis (Figure 7A-C) and also exhibited similar decreases in splenomegaly (Figure 7D). These findings demonstrated that MyD88^−/− Treg exhibited no defect in their ability to suppress either intestinal or systemic inflammation. Finally, we found no difference in the accumulation of either wild type or MyD88^−/− FoxP3 cells in the mesenteric lymph node (Figure 7E). These findings are consistent with the comparable frequency of FoxP3⁺ CD4⁺ T cells in WT and MyD88^−/− mice (Supplementary Figure 4). Together, these results indicate that T cell intrinsic MyD88 signals are not required for development of CD4⁺CD25⁺ Treg cells or for the expression of their suppressor function in vivo.