NOD2 regulates CXCR3-dependent CD8⁺ T cell accumulation in intestinal tissues with acute injury

Abstract

Polymorphisms in NOD2 confer risk for Crohn’s disease (CD), characterized by intestinal inflammation. How NOD2 regulates both inflammatory and regulatory intestinal T cells, which are critical to intestinal immune homeostasis, is not well-understood. Anti-CD3 monoclonal antibody (mAb) administration is used as therapy in human autoimmune diseases, as well as a model of transient intestinal injury. The stages of T cell activation, intestinal injury, and subsequent T tolerance are dependent on migration of T cells into the small intestinal (SI) lamina propria. Upon anti-CD3 mAb treatment of mice, we found that NOD2 was required for optimal small intestinal IL-10 production, in particular from CD8⁺ T cells. This requirement was associated with a critical role for NOD2 in SI CD8⁺ T cell accumulation and induction of the CXCR3 ligands CXCL9 and CXCL10, which regulate T cell migration. NOD2 was required in both the hematopoietic and non-hematopoietic compartments for optimal expression of CXCR3 ligands in intestinal tissues. NOD2 synergized with IFN-γ to induce CXCL9 and CXCL10 secretion in dendritic cells, macrophages and intestinal stromal cells in vitro. Consistent with the in vitro studies, during anti-CD3 mAb treatment in vivo, CXCR3 blockade, CD8⁺ T cell depletion or IFN-γ neutralization each inhibited SI CD8⁺ T cell recruitment, and reduced chemokine expression and IL-10 expression. Thus NOD2 synergizes with IFN-γ to promote CXCL9 and CXCL10 expression, thereby amplifying CXCR3-dependent SI CD8⁺ T cell migration during T cell activation, which in turn contributes to induction of both inflammatory and regulatory T cell outcomes in the intestinal environment.

Introduction

Inflammatory bowel diseases (IBD), including Crohn’s disease (CD) and ulcerative colitis, are characterized by leukocyte accumulation in intestinal tissues (1). Of the common genetic variants identified to date, loss-of-function polymorphisms in nucleotide oligomerization domain 2 (NOD2), an intracellular sensor of the bacterial cell wall component peptidoglycan, confer the greatest susceptibility for developing CD (1). The mechanisms through which NOD2 regulates intestinal immune homeostasis are incompletely understood, although multiple mechanisms are likely involved, including through NOD2 regulation of intestinal epithelial cells, myeloid-derived cells, and T cell differentiation (1–4). Interestingly, although loss-of-function in NOD2 confers risk for CD (1) and intestinal inflammation in specific experimental mouse models (5), there is increasing evidence that loss of NOD2 function may be beneficial in certain situations, such as with infectious challenges (2, 6). For example, mice expressing the CD-associated L1007fsinsC NOD2 polymorphism demonstrate decreased inflammation and lethality after infection with Enterococcus faecalis (6), and T-cell intrinsic NOD2 deficiency protects mice from Toxoplasma gondii associated colitis (2). Further supporting this beneficial effect is that human carriers of NOD2 polymorphisms that result in decreased NOD2 expression (7) are less likely to have chronic disease from Mycobacterium leprae (8). This beneficial effect may help explain the relatively frequent presence of loss-of-function NOD2 polymorphisms in the population. Therefore, the inflammation associated with certain infectious exposures or acute injury appears to be attenuated with decreased NOD2 expression or function.

Anti-CD3 monoclonal antibody (mAb) treatment is being studied in ongoing trials for various human immune-mediated diseases, including IBD, type I diabetes mellitus (T1DM), psoriatic arthritis and graft-versus-host disease (GVHD) (9). This treatment results in T cell activation (10), transient intestinal injury (11) and induction of regulatory T cell populations (e.g. IL-10-producing T cells, FoxP3⁺ Tregs) in the small intestine (SI) (12–15), thereby highlighting the regulation of critical stages of intestinal T cell differentiation. Both the intestinal inflammation and induction of intestinal regulatory T cells are dependent upon T cell recruitment into the intestinal lamina propria (13, 14, 16). Importantly, the regulatory T cells generated upon anti-CD3 mAb treatment can mediate protection of systemic immune-mediated diseases, including GVHD (17), skin graft rejection (18), T1DM (19) and autoimmune encephalomyelitis (20). Furthermore, the systemic protection under these conditions is dependent upon the generation of regulatory T cells within the intestinal lamina propria (13). Loss-of-function Leu1007insC NOD2 CD patients were found to have decreased FoxP3⁺ Tregs in colonic tissue compared to WT NOD2 CD patients (21), pointing to the possibility of dysregulation in the generation of intestinal-derived regulatory T cell populations in the absence of NOD2 function or expression.

To dissect the role of NOD2 in mediating intestinal T cell responses in vivo, we selected the clinically relevant anti-CD3 mAb treatment model. We found that NOD2 was critical for the induction of IL-10-producing CD8⁺ T cells in the small intestinal lamina propria; this was due to a NOD2 requirement for intestinal CD8⁺ T cell accumulation during anti-CD3 mAb treatment. The T cell trafficking CXCR3 ligands CXCL9 and CXCL10 were dramatically decreased in NOD2^−/− mice after anti-CD3 mAb treatment. Consistently, CXCR3 blockade inhibited CD8⁺ T cell recruitment to the SI with anti-CD3 mAb injection, which led to attenuation of small intestinal chemokines and cytokines (e.g. IL-10). NOD2 expression in the hematopoietic and non-hematopoietic cell compartments was necessary for optimal CXCL9 and CXCL10 production in intestinal tissues upon anti-CD3 mAb injection. Interestingly, NOD2 synergized with IFN-γ to significantly enhance CXCL9 and CXCL10 expression in bone marrow-derived dendritic cells (BMDC), bone marrow-derived macrophages (BMM) and intestinal stromal cells in vitro. T cells are a significant source of IFN-γ upon anti-CD3 activation; consistently depletion of CD8⁺ T cells or neutralization of IFN-γ reduced intestinal expression of chemokines and ultimately IL-10 during anti-CD3 mAb injection. NOD2 deficiency similarly attenuated chemokine induction and T cell infiltration in a separate CXCR3-dependent acute intestinal injury model, the piroxicam-induced colitis model in IL-10^−/− mice. Taken together, NOD2 is critical for the increased injury-induced chemokine expression in intestinal tissues, in particular CXCL9 and CXCL10, which in turn mediates amplification of CXCR3-dependent T cell recruitment to the intestinal lamina propria. This recruitment, in turn, regulates both the inflammatory and regulatory T cell outcomes within the intestinal lamina propria.

Materials and Methods

Mice

NOD2^−/− mice (Jackson Laboratory, Bar Harbor, ME) were crossed with IL-10-GFP reporter mice (12) or C57BL/6 Thy1.1^+/+ mice (Jackson Laboratory). Mice were maintained in a specific pathogen-free facility and used between 2–5 months of age. Experiments were performed in agreement with the Yale University Institutional Animal Care and Use Committee and according to National Institutes of Health guidelines for animal use.

Abs and staining reagents

The following Abs were used on a LSR II (BD Biosciences, San Jose, CA): allophycocyanin (APC)-Cy7- and APC-labeled anti-CD4, eFluor 650NC- and FITC-labeled anti-CD8, eFluor450-labeled anti-CD3, eFluor 650NC-labeled anti-MHCII, PE and PE-Cy7-labeled Thy1.2, PerCP-labeled Thy1.1, and PerCP-Cy5.5 and APC-labeled anti-CXCR3 (eBioscience, San Diego, CA). CXCR3-173 (anti-CXCR3 neutralizing antibody) (Biolegend, San Diego, CA), 145-2C11 (anti-CD3), 2.43 (anti-CD8), and XMG1.2 (anti-IFN-γ) (BIO X Cell, West Lebanon, NH) were used in vivo. The following were assessed by ELISA: IL-10, IL-17A, IFN-γ (BD Biosciences), CXCL9 (R&D Systems, Minneapolis, MN) and CXCL10 (eBioscience).

Anti-CD3 mAb treatment model

Mice were injected i.p. with 15 μg of anti-CD3 mAb or isotype control hamster IgG at 0 and 48h. Mice were analyzed 4 hours after the final injection. To block CXCR3 signaling, mice were given 100 μg anti-CXCR3 mAb (or Armenian hamster IgG isotype control) i.p. 2h before each anti-CD3 mAb injection. To deplete CD8⁺ T cells, mice were given 250 μg of anti-CD8 mAb (or rat IgG2b isotype control) i.p. 48h before the first anti-CD3 mAb injection and CD8⁺ T cell depletion was confirmed by flow cytometry. To neutralize IFN-γ, mice were given 500 μg anti-IFN-γ (or rat IgG1 isotype control) i.p. 24h before each anti-CD3 mAb injection. In some cases mice were first provided with a 4 week antibiotic regimen in drinking water consisting of vancomycin hydrochloride (500 mg/L; Hospira, Inc, Lake Forest, IL), ampicillin (1g/L; DAVA Pharmaceuticals, Inc, Fort Lee, NJ), metronidazole (1g/L; Teva Pharmaceuticals, Sellersville, PA, 18960) and neomycin sulfate (1g/L; MP Biomedicals, Solon, Ohio) as per (22).

FITC-dextran permeability assay

Mice were orally gavaged with FITC-dextran (40 mg/100 g body weight) 4h before sacrifice. Serum concentration of FITC-dextran was measured by fluorometer at 488 nm.

Intestinal lamina propria cell isolation

Proximal small intestinal (first 8 cm of SI) lamina propria cells were isolated as previously described (23). In brief, the proximal SI was cut longitudinally and then into ~1 mm pieces. SI pieces were washed thoroughly (ice-cold PBS, 5% FCS) and then digested (PBS, 5 mM EDTA, 5% FCS) at 37°C in a rotating incubator to remove epithelial cells. The supernatants containing epithelial cells and intraepithelial lymphocytes were discarded, and intestines were washed twice in ice-cold PBS to remove residual EDTA. The remaining tissue was then incubated for 1h at 37°C in a rotating incubator in collagenase buffer consisting of RPMI 1640, 10% FCS, 200 U/ml collagenase VIII (Sigma-Aldrich), 2 mM glutamine, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 10 mM HEPES, 50 μM 2-ME, 40 μ;g/ml gentamicin, and 50 U/ml penicillin-50 μg/ml streptomycin. The cells were then filtered through a 40 μM filter (BD Biosciences) and washed twice in RPMI media. For intestinal stromal cells, intestinal lamina propria cells were plated overnight. The next day, nonadherent cells were removed and the adherent fibroblast-like cells were cultured for 7 days. The intestinal stromal cells expressed α-SMA, vimentin, P4HA1, desmin and Myh10 consistent with a myofibroblast phenotype. In contrast, they were negative for hematopoietic markers (e.g. CD45, CD11b, CD11c). They did not contain T cell or B cell contamination as assessed by CD3 and CD19, respectively.

Generation and culture of BMDC and BMM

BM single cell suspensions were cultured in complete RPMI 1640 media containing 20 ng/mL GM-CSF (PeproTech, Rocky Hill, NJ) (for BMDC) and 10% L929-conditioned medium (for BMM). Cultures were fed fresh medium every 3 days, and used at 6 to 8 days. Purity was >98% as assessed by flow cytometry. BMDC were CD11c⁺ and F4/80⁻, while BMM were CD11b⁺, F4/80⁺ and CD11c⁻. IFN-γ (PeproTech), muramyl dipeptide (MDP) (Bachem, King of Prussia, PA), and lipid A (Peptides International, Louisville, KY) were used in vitro.

Tissue mRNA expression and protein analysis

Total RNA (Trizol, Life Technologies, Carlsbad, CA) from cells or homogenized organ tissue was isolated, reverse transcribed, and quantitative PCR was performed as previously described (23). Each sample was run in duplicate and normalized to GAPDH. Primers sequences are shown in Supplementary Table I. For protein analysis, tissue was suspended in a Triton lysis buffer and homogenized (VWR International, Radnor, PA), and ELISA was performed.

Bone marrow chimeras

Donor BM cells (1×10⁷) were adoptively transferred into lethally irradiated recipients (1100 cGy total body irradiation) between 8–12 weeks of age. Mice were analyzed 6 weeks later. We confirmed engraftment in the blood (90.1±1.2%) and intestinal lamina propria (71.2±2.3%). The presence of radioresistant Thy1+ cells in intestinal tissues has been observed by others in such sites as isolated lymphoid follicles (24).

Statistical analyses

Statistical comparisons were assessed using a two-tailed Student’s t test. Bonferroni correction was applied for multiple comparisons. Values of p < 0.05 were considered significant.

Results

NOD2 is required for accumulation of IL-10-producing cells in the SI of mice during anti-CD3 mAb treatment

Anti-CD3 mAb treatment induces an immunoregulatory environment marked by induction of IL-10-producing cells mainly in the SI (12). To determine the role of NOD2 in generating intestinal regulatory T cell populations, we crossed NOD2^−/− mice with IL-10-GFP reporter mice, and compared NOD2^−/− IL-10-GFP and NOD2^+/− IL-10-GFP littermate controls. As expected, the percentage of IL-10-producing cells significantly increased in the SI LP of NOD2^+/− mice after anti-CD3 mAb treatment (Fig 1A&B). In contrast, the accumulation of IL-10-producing cells in the SI of NOD2^−/− mice was significantly decreased relative to littermate control mice (Fig 1A&B). This was associated with a decreased number of IL-10-producing cells in the SI of NOD2^−/− mice (Fig 1C). Consistent with the pattern of IL-10-producing cells in the SI, induction in SI IL-10 mRNA (Fig 1D) and serum IL-10 (Fig 1E) after anti-CD3 mAb treatment was decreased in NOD2^−/− mice relative to littermate controls. Of note is that FoxP3⁺ Tregs can also be induced in the SI with anti-CD3 mAb injection (12, 13). Furthermore, a study in human loss-of-function Leu1007insC NOD2 CD patients identified decreased colonic FoxP3⁺ CD4⁺ Tregs in patients with active CD (21). However, we did not observe differences in the percentage of FoxP3⁺ Tregs in CD4⁺ T cells at baseline or after anti-CD3 mAb treatment in NOD2^+/− compared to NOD2^−/− mice (data not shown). In summary, NOD2 is required for the induction of IL-10-producing cells in the SI of mice upon anti-CD3 mAb treatment.

Figure 1. NOD2 is required for optimal accumulation of IL-10-producing cells in the SI of mice during anti-CD3 mAb treatment.

NOD2^+/− IL-10-GFP and NOD2^−/− IL-10-GFP mice were treated with 15 μg of anti-CD3 mAb or IgG isotype control at 0 and 48 h. Four hours after the last injection animals were sacrificed. (A) Representative flow cytometry plots of IL-10–producing cells in the SI LP (gated on live cells). (B) Percentage of IL-10-producing cells in the SI. (C) Number of IL-10-producing cells in the SI (mean+SEM; n=3 per group; representative of 5 independent experiments). (D) IL-10 mRNA expression in the SI. (E) Serum IL-10. D–E are shown as mean+SEM; n=6 per group from two independent experiments. Comparisons are between anti-CD3 mAb and IgG treatment in the same genotype or as indicated. *, p<0.05; **, p<0.01; ***, p<0.001.

CD8⁺ T cells are a major source of IL-10-producing cells in the SI of mice during anti- CD3 mAb treatment

We next sought to dissect which small intestinal lamina propria cells were producing IL-10 upon anti-CD3 mAb injection, and which of these IL-10-producing cell subsets was decreased in NOD2^−/− mice. We found that upon anti-CD3 mAb injection, ~90% of IL-10⁺ cells were T cells (Thy1.2⁺ cells); the majority of these IL-10-producing T cells were CD8⁺ T cells (Fig 2A&B). Relative to CD4⁺ T cells, regulation of intestinal CD8⁺ T cells during anti-CD3 mAb treatment has not been well-studied. However, treatment of T1DM patients with anti-CD3 mAb leads to increased circulating regulatory CD8⁺ T cell populations (25), highlighting the importance of CD8⁺ T cells in therapy for human disease. We found the percentage (Fig 2C) and number (Fig 2D) of IL-10-producing CD8⁺ T cells in the SI of NOD2^+/− mice was increased with anti-CD3 mAb treatment; this increase was significantly reduced in NOD2^−/− mice.

Figure 2. CD8⁺ T cells are the major source of IL-10-producing cells in the SI of mice during anti-CD3 mAb treatment.

Mice were treated with anti-CD3 mAb as in Fig 1. (A) Representative flow cytometry gated on IL-10–producing cells in the SI LP. Surface CD3 is internalized with anti-CD3 mAb treatment, whereas Thy1.2 expression remains intact. Cells were stained for Thy1.2, CD4 and CD8 expression. (B) Pie chart of the proportion of IL-10-producing cell types in the SI after anti-CD3 mAb injection. (mean+SEM; n=3). (C) Percentage and (D) number of IL-10-producing CD8⁺ T cells (CD4⁻Thy1.2⁺) in the SI of NOD2^+/− and NOD2^−/− mice (mean+SEM; n=3 per group). Data are representative of at least 5 independent experiments. **, p<0.01.

NOD2 is required for CD8⁺ T cell accumulation in the SI during anti-CD3 mAb treatment

Prior studies showed that T cells accumulate in the proximal SI a few hours after anti-CD3 mAb treatment (13, 14). The decreased number of IL-10-producing CD8⁺ T cells in NOD2^−/− mice after anti-CD3 mAb treatment NOD2 may be due to a defect in IL-10 induction in the CD8⁺ T cells within the SI, or due to an overall decrease in accumulation of CD8⁺ T cells in the SI. To dissect these two possibilities, we first gated on CD8⁺ T cells isolated from the SI lamina propria after anti-CD3 mAb treatment and found that the percentage of IL-10-producing cells within the CD8⁺ T cells that were present in the SI of NOD2^+/− and NOD2^−/− mice was equivalent (Fig 3A&B). On the other hand, the dramatic increase in the percentage (Fig 3C) and number (Fig 3D) of CD8⁺ T cells observed within the SI of NOD2^+/− mice after anti-CD3 mAb treatment was significantly diminished in NOD2^−/− mice, highlighting a role for NOD2 in optimal CD8⁺ T cell accumulation in the SI. Of note is that as expected, the %CD8⁺ T cells in the lamina propria (LP) at baseline is low; the increase in CD8⁺ T cells is observed specifically upon anti-CD3 mAb treatment. We ensured that the increase in LP CD8⁺ T cells upon anti-CD3 mAb treatment was not due to contamination by intraepithelial lymphocytes (IEL) through staining for characteristic IEL populations (TCRγδ and TCRαβ CD8αα T cells, data not shown). The CD8⁺ T cells accumulating in the lamina propria express CD45 and Thy1.2, and are MHCII⁻ (Supplementary Fig 1). For experiments that ensue, we gate on live CD8⁺CD4⁻Thy1.2⁺ (MHCII⁻) cells to examine CD8⁺ T cells (Supplementary Fig 1). Given the decreased CD8⁺ T cells in the SI, we examined other T cell-derived cytokines and observed that IFN-γ and IL-17A mRNA expression in the SI was also decreased in NOD2^−/− mice after anti-CD3 mAb treatment (Fig 3E). Therefore, upon anti-CD3 mAb treatment, NOD2 is required for the significant increase in CD8⁺ T cells in the SI lamina propria, and for the subsequent increase in cytokines and IL-10-producing CD8⁺ T cells that occurs.

Figure 3. NOD2 is required for CD8⁺ T cell migration to and cytokine production in the SI upon anti-CD3 mAb treatment.

Mice were treated with anti-CD3 mAb as in Fig 1. (A) Representative flow cytometry plots of cells from the SI gated on CD8⁺ T cells (CD4⁻Thy1.2⁺). (B) Percentage of IL-10-GFP⁺ cells in CD8⁺ T cells in the SI of NOD2^+/− and NOD2^−/− mice after anti-CD3 mAb treatment. (C) Percentage and (D) number of CD8⁺ T cells in the SI lamina propria. (mean+SEM; n=3 per group; representative of 5 independent experiments). (E) IFN-γ and IL-17A mRNA expression in the SI (mean+SEM; n=6–9 per group from 3 independent experiments). Comparisons are between anti-CD3 mAb and IgG treatment in the same genotype or as indicated. (F) Oral FITC-dextran was administered 4h before sacrifice (coincident with the 48h anti-CD3 mAb treatment time point) and measured in the serum at the time of sacrifice. (G–I) NOD2^+/− IL-10-GFP mice were administered an antibiotic (ABX) regimen consisting of vancomycin, metronidazole, ampicillin, and neomycin for 4 weeks in the drinking water. Antibiotic-treated or –untreated (SPF, specific pathogen free) mice were then treated with anti-CD3 mAb as in Fig 1 and examined for: (G) percentage of IL-10-GFP⁺ cells (gated on live cells), (H) percentage of IL-10⁺CD8⁺ T cells, and (I) percentage of CD8⁺ T cells in the SI lamina propria. *, p<0.05; **, p<0.01; ***, p<0.001; NS, not significant.

Optimal anti-CD3-mediated increases in IL-10-producing cells and CD8⁺ T cell accumulation in the SI lamina propria depends on intestinal microbiota

The dependency on NOD2 for optimal accumulation of IL-10-producing CD8⁺ T cells in the SI lamina propria upon anti-CD3 mAb treatment highlighted a likely role for intestinal microbiota in mediating these outcomes. Consistent with prior reports (26), we observe increased intestinal permeability after anti-CD3 treatment (Fig 3F), thereby increasing exposure of lamina propria cells to intestinal microbiota. To directly assess the role of intestinal bacteria in the anti-CD3 treatment-mediated outcomes observed, mice were administered a commonly utilized oral antibiotic regimen consisting of vancomycin, metronidazole, ampicillin, and neomycin for 4 weeks in the drinking water (22). The antibiotic treatment resulted in a 96.6±1.3% reduction in bacterial 16s rRNA levels in the feces. Upon anti-CD3 treatment, mice treated with antibiotics demonstrated significantly decreased accumulation of SI lamina propria IL-10-producing cells (Fig 3G), IL-10-producing CD8⁺ T cells (Fig 3H) and total CD8⁺ T cells (Fig 3I) compared to non-antibiotic treated mice. Taken together, intestinal bacteria contribute to the increased IL10-producing CD8⁺ T cells in the small intestine upon anti-CD3 treatment.

NOD2 is required for optimal CXCL9 and CXCL10 expression in the SI after anti-CD3 mAb treatment

To define the mechanism through which NOD2 mediates the increased CD8⁺ T cell accumulation in the SI after anti-CD3 mAb treatment, we evaluated pertinent trafficking molecules, including integrins and chemokines. CD18, CD11a, α4, α4β7, and β1 play an important role in T cell trafficking to intestinal tissues (27). In some cases the expression of these integrins increased with anti-CD3 mAb (Supplementary Fig 2). However, the expression of the assessed integrins on CD8⁺ T cells did not differ between NOD2^−/− and littermate controls at baseline or with anti-CD3 mAb treatment in the SI, or in lymphoid organs from which they might be recruited such as mesenteric lymph nodes (MLN) or spleen (Supplementary Fig 2).

We next considered the regulation of chemokines in NOD2^−/− mice during anti-CD3 mAb treatment. CXCR3 regulates T cell migration to inflammatory sites (28). The CXCR3 receptor has three ligands, CXCL9, CXCL10 and CXCL11. Both CXCL9 and CXCL10 were significantly induced in the SI and MLN of NOD2^+/− mice after anti-CD3 mAb treatment; this dramatic increase was not observed in NOD2^−/− mice (Fig 4A&B). CXCL11 was induced to an equivalent degree in the SI of NOD2^+/− and NOD2^−/− mice after anti-CD3 mAb treatment (data not shown). In contrast to the decreased CXCL9 and CXCL10 in NOD2^−/− mice upon anti-CD3 mAb treatment, CXCR3 expression on CD8⁺ T cells did not differ between NOD2^−/− mice and littermate controls (Fig 4C&D). Therefore, NOD2 is critical for the significant induction of CXCR3 ligands in intestinal lymphoid tissues observed with anti-CD3 mAb treatment.

Figure 4. NOD2 is required for optimal CXCR3 ligand induction in the SI upon anti-CD3 mAb treatment.

(A–D) NOD2^+/− and NOD2^−/− mice were treated with anti-CD3 mAb as in Fig 1. CXCL9 and CXCL10 mRNA levels in (A) SI and (B) MLN (mean+SEM; n=6 per group; representative of two independent experiments). (C–D) Percentage of CD8⁺ T cells expressing CXCR3 in SI, MLN and spleen was assessed. (C) Representative flow cytometry plots for CXCR3 expression on CD8⁺ T cells (solid black line). Low CXCR3-expressing MHCII⁺ cells are shown for comparison (shaded grey histogram). (D) Summary graph for CXCR3-expressing cells within CD8⁺ T cells (mean+SEM; n=3 per group; representative of two independent experiments). *, p<0.05; **, p<0.01; ***, p<0.001.

CD8⁺ T cell accumulation in the SI and subsequent outcomes upon anti-CD3 mAb treatment are CXCR3-dependent

To establish that CXCR3 is playing a definitive role in CD8⁺ T cell accumulation in the SI during anti-CD3 mAb treatment, and in the induction of cytokines, including IL-10, associated with this accumulation, CXCR3 blocking antibody was injected prior to anti-CD3 mAb treatment. CXCR3 blockade significantly attenuated the accumulation of CD8⁺ T cells in the SI (Fig 5A&B). Anti-CXCR3 mAb administration in the absence of anti-CD3 mAb treatment (with Armenian hamster IgG isotype control) did not alter the percentage or number of CD8⁺ T cells in the SI (data not shown). Importantly, CXCR3 blockade also resulted in decreased intestinal CXCL9 and CXCL10 mRNA expression (Fig 5C), and decreased IFN-γ, IL-17A and IL-10 expression in the SI (Fig 5D) and serum (Fig 5E) during anti-CD3 mAb treatment. In contrast, we did not observe differences in LP CD8⁺ T cell apoptosis with anti-CD3 mAb treatment between NOD2^+/− and NOD2^−/− mice (data not shown). Moreover, while LP CD8⁺ T cell proliferation (assessed by Ki67 expression) increased significantly upon anti-CD3 mAb treatment, there was no difference between NOD2^+/− and NOD2^−/− mice in this induced proliferation (data not shown). Thus, CXCR3 is critical for the T cell migration into the SI lamina propria, amplification of its own chemokine ligands, and the ultimate induction of IL-10 that occurs upon anti-CD3 mAb treatment.

Figure 5. CXCR3 is essential for the amplification of SI CD8⁺ T cell accumulation, chemokines and subsequent IL-10 induction upon anti-CD3 mAb treatment.

To block CXCR3, mice were treated 100 μg anti-CXCR3 mAb (or Armenian hamster IgG isotype control) i.p. 2h before each anti-CD3 mAb injection (at 0 and 48h). Mice were harvested 4h after the 2^nd anti-CD3 mAb injection. (A) Percentage and (B) number of CD8⁺ T cells in the SI LP (mean+SEM; n=3 per group; representative of 3 independent experiments). (C) CXCL9 and CXCL10 mRNA expression in the SI. (D) IL-10, IFN-γ, and IL-17A mRNA expression in SI. (E) Serum levels of IL-10, IFN-γ, and IL-17A (mean + SEM; n = 6/group from 2 independent experiments for C–E). *, p<0.05; **, p<0.01; ***, p<0.001.

NOD2 in hematopoietic and non-hematopoietic compartments is required for optimal CD8⁺ T cell accumulation in the SI upon anti-CD3 mAb treatment

As NOD2 is expressed in both hematopoietic and non-hematopoietic cells, we sought to define the contributions of these cells types to the accumulation of CD8⁺ T cells and IL-10 production in the SI after anti-CD3mAb treatment. To address this, we generated bone marrow chimeras in which either the donors or recipients were NOD2 deficient. On transfer of NOD2^−/− Thy1.1⁺ into WT Thy1.2⁺ mice, accumulation of NOD2^−/− donor CD8⁺ T cells in the SI was significantly attenuated upon anti-CD3 mAb injection relative to donor NOD2^+/− Thy1.1⁺ (Fig 6A). Induction of CXCL9 and CXCL10 expression was reduced in NOD2^−/− donors compared to NOD2^+/− donors upon anti-CD3 mAb injection (Fig 6B), consistent with the decreased CD8⁺ T cell accumulation in these mice. Moreover, IL-10, IFN-γ and IL-17A intestinal mRNA expression (Fig 6C) and serum levels (Fig 6D) were decreased in NOD2^−/− compared to NOD2^+/− donors. In the reciprocal transfer of WT Thy1.2⁺ into NOD2^−/− Thy1.1⁺ mice, similar reductions were observed relative to transfer of WT Thy1.2⁺ into NOD2^+/− Thy1.1⁺ mice (Fig 6E–H). Therefore, NOD2 expression is required in the hematopoietic and non-hematopoietic cell compartments for the CD8⁺ T cell accumulation, and induced chemokine and cytokine expression observed with anti-CD3 mAb injection.

Figure 6. NOD2 in hematopoietic and non-hematopoietic cells is required for optimal CD8⁺ T cell migration into the SI lamina propria upon anti-CD3 mAb treatment.

Chimeric mice were generated and treated with anti-CD3 mAb as in Fig 1. (A–D) Transfer of NOD2^+/− or NOD2^−/− Thy1.1⁺ BM cells into irradiated Thy1.2⁺ WT mice. (E–H) Transfer of WT Thy1.2⁺ BM cells into NOD2^+/− or NOD2^−/− Thy1.1⁺ mice. (A&E) Percentage of CD8⁺ T cells donor cells in the SI LP. (B&F) Chemokine mRNA expression in SI. (C&G) Cytokine mRNA expression in SI. (D&H) Serum cytokine levels. Data are mean+SEM; n=3 per group; representative of 2 independent experiments. *, p<0.05; **, p<0.01; ***, p<0.001.

NOD2 stimulation synergistically enhances IFN-γ-induced CXCL9 and CXCL10 expression in BMM, BMDC and intestinal stromal cells in vitro

Given the dependence on NOD2 in both hematopoietic and non-hematopoietic cells for chemokine production and CD8⁺ T cell accumulation in the SI during anti-CD3 mAb treatment, we evaluated the ability of NOD2 stimulation to induce CXCL9 and CXCL10 in both hematopoietic (e.g. BMM, BMDC) and non-hematopoietic (e.g. intestinal stromal) cells. NOD2 stimulation alone with MDP induced a low level of expression of both these chemokines in BMM and BMDC (Fig 7A&B), although not in intestinal stromal cells (Fig 7C). NOD2 can synergize with other molecules for certain downstream outcomes (29). CD8⁺ T cells significantly increase in the SI upon anti-CD3 activation (Fig 3C&D), and CXCR3-dependent blockade of CD8⁺ T cells led to decreased intestinal CXCL9 and CXCL10 (Fig 5C). IFN-γ can induce CXCL9 and CXCL10 (28), and activated CD8⁺ T cells are an important source of IFN-γ. We therefore evaluated interactions between MDP and IFN-γ and observed a clear synergy between MDP and IFN-γ in inducing CXCL9 and CXCL10 secretion from BMM (Fig 7A), BMDC (Fig 7B) and intestinal stromal cells (Fig 7C). This synergy was absent in BMM, BMDC and intestinal stromal cells from NOD2^−/− mice (Fig 7). In contrast, synergy between NOD2 and TLR4 (Fig 7A–C) and other TLRs (data not shown) was not observed for CXCL9 or CXCL10 secretion. Therefore, NOD2 synergizes with IFN-γ, a cytokine secreted particularly by activated T cells, to produce chemokines from myeloid-derived cells and intestinal stromal cells that can then further enhance T cell recruitment into intestinal tissues.

Figure 7. NOD2 stimulation synergistically enhances IFN-γ-induced CXCL9 and CXCL10 expression in BMM, BMDC and intestinal stromal cells in vitro.

(A) BMM, (B) BMDC or (C) intestinal stromal cells from NOD2^+/− or NOD2^−/− mice were stimulated for 24h with MDP (10 μg/ml for BMM & BMDC cells; 1μg/ml intestinal stromal cells), 1 ng/ml IFN-γ, or 0.1 μg/ml lipid A, alone or combination as indicated. Supernatants were assayed for CXCL9 and CXCL10. Data shown as mean+SEM, n=3 per group, and representative of 3 independent experiments. Significance is compared for conditions examining synergy as indicated (Bonferroni correction applied) and between NOD2^+/− and NOD2^−/− cells. **, p<0.01; ***, p<0.001.

Anti-CD3 mAb treatment-dependent intestinal outcomes requires CD8⁺ T cells and IFN-γ

We next sought to clearly establish in vivo that CD8⁺ T cells and IFN-γ are each required for the intestinal outcomes observed with anti-CD3 mAb treatment. We therefore injected mice with anti-CD8 mAb to deplete CD8⁺ T cells prior to anti-CD3 mAb treatment; CD8⁺ T cells were effectively depleted in the SI (Fig 8A), MLN and spleen (data not shown). CD8⁺ T cell depletion significantly decreased the induction of CXCL9 and CXCL10 mRNA (Fig 8B) and protein (Fig 8C) in the SI, and of IL-10, IFN-γ, and IL-17A expression in the SI (Fig 8D) upon anti-CD3 mAb treatment. IFN-γ blockade also significantly attenuated the accumulation of CD8⁺ T cells in the SI (Fig 8E), induction of CXCL9 and CXCL10 (Fig 8F&G), and IL-10, IFN-γ, and IL-17A in the SI (Fig 8H) during anti-CD3 mAb treatment. Thus, CD8⁺ T cells and IFN-γ are critical for the optimal induction of intestinal chemokines and cytokines, including of IL-10, upon anti-CD3 mAb treatment in vivo, highlighting the ability of activated CD8⁺ T cells to amplify their own recruitment to the SI through modulation of chemokine induction (Fig 9).

Figure 8. Depletion of CD8⁺ T cells or neutralization of IFN-γ attenuates intestinal outcomes upon anti-CD3 mAb treatment.

(A–D) CD8⁺ T cells were depleted by anti-CD8 mAb injection (or treated with rat IgG2b isotype control) as per Materials & Methods. (E–H) IFN-γ was neutralized by anti-IFN-γ mAb injection (or treated with rat IgG1 isotype control) as per Materials & Methods. (A&E) Percentage of CD8⁺ T cells in the SI LP (mean+SEM; n=3 per group; representative of 2 independent experiments). (B&F) CXCL9 and CXCL10 mRNA expression in the SI. (C&G) CXCL9 and CXCL10 protein expression per mg SI tissue. (D&H) IL-10, IFN-γ, and IL-17A mRNA expression in the SI (mean + SEM; n = 6/group from 2 independent experiments for B-D & F-H). *, p<0.05; **, p<0.01; ***, p<0.001.

Figure 9. Proposed model for mechanisms of NOD2-mediated CXCR3-dependent T cell migration into intestinal tissues.

1. Acute activation of circulating and intestinal CD8⁺ T cells by anti-CD3 mAb; 2. IFN-γ secretion by activated T cells; 3. IFN-γ (T cell-derived) and MDP (microbiota-derived; stimulates intracellular NOD2) synergize to increase CXCL9 and CXCL10 secretion from myeloid-derived and intestinal stromal cells; 4. CXCR3-mediated chemotaxis of CD8⁺ T cells into the intestinal lamina propria secondary to increased CXCL9 and CXCL10 expression; 5. Newly recruited CD8⁺ T cells amplify recruitment of additional CD8⁺ T cell into intestinal tissues through the above sequence of events.

NOD2 promotes T cell accumulation in the colon in CXCR3-dependent acute piroxicam-induced colitis

To determine if NOD2 regulation of T cell accumulation, and T-cell mediated injury extends to additional intestinal injury models, we selected a colitis model known to be dependent on CXCR3-CXCL10 interactions. In IL-10^−/− mice, CXCL10 is highly expressed at sites of colitis and CXCL10 neutralization can attenuate colitis severity (30). Short-term exposure of IL-10^−/− mice to the NSAID piroxicam simulates an environmental trigger that can play a role in human IBD, and results in a synchronous, rapid induction of colitis (31) that is similarly CXCL10-dependent (32). Piroxicam-fed IL-10^−/− mice demonstrated rapid weight loss (Supplementary Fig 3A), colon shortening (Supplementary Fig 3B) and moderate to severe colitis (Supplementary Fig 3C&D); such changes were markedly attenuated in NOD2^−/− IL-10^−/− mice. Consistent with these findings, T cell infiltration into the colon (Supplementary Fig 3E) and MLN (Supplementary Fig 3F) of NOD2^−/− IL-10^−/− mice was significantly decreased compared to IL-10^−/− mice. Consistent with the decreased T cell recruitment, colonic CXCL9 and CXCL10 expression was significantly decreased in NOD2^−/− IL-10^−/− mice compared to IL-10^−/− mice (Supplementary Fig 3G), as was colonic TNF-α, IFN-γ and IL-17A mRNA expression (Supplementary Fig 3H). These results indicate that similar to the NOD2 requirement for T cell recruitment and chemokine and cytokine production in the SI during anti-CD3 mAb injection, NOD2 is required for T cell accumulation and chemokine and cytokine production in the colon in acute piroxicam-induced colitis in IL- 10^−/− mice.

Discussion

In this study, we demonstrate a novel role for NOD2 in CD8⁺ T cell accumulation in the intestinal lamina propria during acute intestinal injury, which contributes to consequences with respect to inflammatory and regulatory T cell outcomes. NOD2 is required for optimal CD8⁺ T cell migration into the small intestinal lamina propria during anti-CD3 mAb treatment, which then enhances CXCL9 and CXCL10 induction, thereby amplifying CXCR3-dependent CD8⁺ T cell recruitment. The NOD2 ligand MDP synergizes with IFN-γ (produced by activated T cells), but not lipid A, to induce CXCL9 and CXCL10 in BMM, BMDC and intestinal stromal cells. We show that these interactions between CD8⁺ T cells, IFN-γ, NOD2, intestinal bacteria and CXCR3 in vivo are critical in driving the CD8⁺ T cell recruitment amplification loop (Fig 9). Therefore, we find that through its contributions in multiple cell subsets to the regulation of T cell trafficking, NOD2 is required for both the inflammatory and regulatory T cell outcomes observed in the intestinal environment.

NOD2 can play a complex role in intestinal immune outcomes in vivo, with varying contributions in different cell subsets. On the one hand NOD2 confers protection in certain situations, such as GVHD (33) and Helicobacter hepaticus-driven inflammation (34). On the other hand, NOD2 contributes to the adverse outcomes in other situations, such as in Enterococcus faecalis infection (6), in Toxoplasma gondii-associated colitis (2), and in the colitis observed in older IL-10^−/− mice (associated with contributions to cytokine secretion in macrophages) (35). We now define contributions of NOD2 to both inflammatory and regulatory T cell outcomes through its regulation of T cell-recruiting chemokines, which in turn initiates a self-amplifying loop of T cell migration into the intestinal lamina propria, thereby providing insight into at least one mechanism for these dual effects.

The ability of NOD2 to regulate chemokines and thereby cell recruitment into the intestinal lamina propria has been previously described in the context of C. rodentium infection; stromal cell-derived CCL2 induction and monocyte recruitment is NOD2-dependent (36). The CXCR3-CXCL9/10 axis is important in T cell trafficking and effector T cell generation (28). Accordingly, we now find a clear role for NOD2 in the hematopoietic and non-hematopoietic compartment in vivo and in myeloid-derived cells and intestinal stromal cells in vitro for induction of chemokines critical for intestinal recruitment of T cells. Interestingly, NOD2 in stromal cells can also directly contribute to T cell differentiation (37). Of note is that our studies demonstrate that CXCR3 is required for optimal induction of its own ligands within intestinal tissues during anti-CD3 mAb treatment, consistent with the amplification loop that occurs during this acute form of injury (Fig 9). Blockade of CXCR3-CXCL10 can protect mice from colitis (30, 32, 38); we now identify a clear role for these interactions in clinically relevant anti-CD3 mAb therapy. Moreover, a fully humanized anti-CXCL10 antibody is in development for treating IBD (39). Our studies demonstrate that while blockade of the CXCR3-CXCL9/10 axis can downregulate proinflammatory cytokines, it can also impact on the generation of IL-10-producing T cells that occurs in the intestinal environment, which may ultimately have significant consequences on protective mechanisms given the important role of the intestine for broad regulatory T cell outcomes.

In contrast to IEL which consist of predominantly CD8⁺ T cells, lamina propria lymphocytes are predominantly CD4⁺ T cells. However, under anti-CD3 treatment conditions, we observed a significant increase in the percentage of lamina propria CD8⁺ T cells. This increase has been previously described with anti-CD3 treatment (14). Moreover, a dramatic increase in lamina propria CD8⁺ T cells has been observed during viral infections, such as with vesicular stomatitis virus (40), highlighting that CD8⁺ T cells can accumulate in the intestinal lamina propria under select conditions of T cell activation. While we find that optimal intestinal lamina propria CD8⁺ T cell accumulation is dependent upon CXCR3-mediated migration (Fig 5), there are likely additional factors contributing to both CD8⁺ T cell accumulation and induction of cytokines and chemokines during anti-CD3 mAb treatment. Other chemokine receptors (e.g. CCR6) can contribute to T cell migration into the small intestine during injury (14), as can various adhesion molecules, a subset of which we found to be upregulated upon anti-CD3 mAb treatment (Supplementary Fig. 2). We did not observe differences between cell death (annexin V+) or proliferation (Ki67+) in SI LP CD8⁺ T cells between NOD2^+/− and NOD2^−/− mice during anti-CD3 mAb treatment (data not shown). Therefore, it appears that at least at the time point examined, the majority of CD8⁺ T cell accumulation is attributable to trafficking into the SI LP. However, it remains possible that differences in cell death, proliferation or T cell extrusion into the intestinal lumen (14) may be detected through alternative approaches or may exist at later time points, which could then contribute to differences between T cell accumulation, and cytokine and chemokine induction between NOD2^+/− and NOD2^−/− mice.

Anti-CD3 mAb treatment has been undergoing therapeutic trials for a number of human immune-mediated diseases, with T1DM being the most well-investigated (9, 19). Patients receiving anti-CD3 therapy demonstrate an increase in circulating regulatory CD8⁺ T cells (25). Interestingly, studies in humanized mouse models have shown that the anti-CD3 mAb-mediated protection is dependent upon T cell migration into the SI, where it acquires a regulatory phenotype; this intestinal T cell migration is required for the anti-CD3 mAb-mediated regulatory T cell protection in systemic diseases (13). The anti-CD3 mAb injection model in mice has also been used to simulate viral infection (14); that anti-CD3 mAb results in an acute transient inflammation and subsequent intestinal-dependent induction of T cell tolerance highlights the unique and important role of the local intestinal environment for the proper differentiation of protective T cells. Therefore, identifying those factors that are required for T cell recruitment into the small intestinal lamina propria during acute T cell activation is essential to understanding the ability of anti-CD3 mAb therapy and other acute intestinal T cell activation conditions to mediate their beneficial effects. Here we have found that CD8⁺ T cells are the main source of IL-10-producing cells within the intestinal lamina propria upon anti-CD3 mAb treatment, and that NOD2 is required for the CXCR3-dependent T cell migration into the SI that ultimately leads to these IL-10-producing CD8⁺ T cells; the function and regulation of CD8⁺ T cells in intestinal injury have been investigated much less than have CD4⁺ T cells. Taken together, NOD2 promotes CXCL9 and CXCL10 expression in the SI, thereby amplifying CXCR3-dependent T cell migration to the intestine, which is important for both pathological and protective intestinal immune responses.

Abbreviations

BMDC

bone marrow dendritic cells

BMM

macrophages

CD

Crohn’s disease

GVHD

graft-versus-host disease

IBD

Inflammatory bowel diseases

mAb

monoclonal antibody

LP

lamina propria

MDP

muramyl dipeptide

NOD2

nucleotide oligomerization domain 2

PRR

pattern recognition receptors

SI

small intestine

T1DM

type 1 diabetes mellitus