NOD2 regulates hematopoietic cell function during graft-versus-host disease

To address contradictory clinical studies showing that NOD2 SNPs of the allo-HSCT donor can either exacerbate GVHD (Holler et al., 2004, 2006) or ameliorate GVHD (Elmaagacli et al., 2006), we investigated the effect of NOD2 deficiency of allo-BMT donors in our GVHD models. We first assessed whether NOD2 deficiency affects allogeneic donor T cells. We transferred CFSE-labeled B6 WT or B6 NOD2^−/− T cells to lethally irradiated BALB/c recipients and found similar T cell proliferation (Fig. 1 A), T cell activation (Fig. 1 B), and expression of the gut-homing molecule LPAM (Fig. 1 C). Next, we performed mixed leukocyte reactions (MLRs) using CD5⁺-selected B6 WT or B6 NOD2^−/− splenocytes as effectors and irradiated BALB/c splenocytes as stimulators. We found no significant differences in the proliferation rate between WT T cells and NOD2^−/− T cells (Fig. 1 D). In addition, we found no significant differences in the proliferation rate between WT T cells and NOD2^−/− T cells when we added different toll-like receptor (TLR) and NOD2 ligands (MDP, LPS, Pam3Cys, Flagellin, and PGN) alone or in combination with the MLRs (unpublished data). To determine the effect of NOD2 deficiency in allo-BMT donors on the development of GVHD, we first used an MHC-disparate allo-BMT model, B6 (H-2^b)→BALB/c (H-2^d). Lethally irradiated recipients received T cell–depleted BM (TCD-BM), and GVHD was induced by the addition of donor splenic T cells to the allograft. We used NOD2^−/− donors and WT donors as the source for TCD-BM or T cells. We did not detect significant differences in lethal GVHD between the four different groups receiving WT BM + WT T cells, WT BM + NOD2^−/− T cells, NOD2^−/− BM + WT T cells, or NOD2^−/− BM + NOD2^−/− T cells (Fig. 1 E, left). To confirm that our observations were not strain dependent or model dependent, we performed similar experiments in an MHC-matched model with minor histocompatibility disparities, B6 (H-2^b)→LP (H-2^b) (Fig. 1 E, right). We conclude that NOD2 deficiency of the allo-BMT donor (either donor T cells or BM) has no significant impact on the development of GVHD and does not regulate alloactivation of donor T cells. Our data suggest that NOD2 has no cell-intrinsic role in the regulation of T cell activity during GVHD despite the increasing evidence that innate immune receptors, such as TLRs, can modulate T cell function and activation during inflammation (Dabbagh and Lewis, 2003; Caron et al., 2005; MacLeod and Wetzler, 2007).

To investigate the role of NOD2 in regulating GVHD in allo-BMT recipients, we first used the MHC-disparate B10BR (H-2^k)→B6 (H-2^b) model. We found significantly more lethal GVHD in B6 NOD2^−/− allo-BMT recipients as compared with B6 WT allo-BMT recipients (Fig. 2 A, left). We confirmed these findings in the MHC-matched LP (H-2^b)→B6 (H-2^b) model (Fig. 2 A, right). To analyze target organ GVHD, we performed histopathological analyses and found significantly more GVHD in the terminal ileum, colon, and liver, as well as a trend toward more skin GVHD in NOD2^−/− versus WT allo-BMT recipients (Fig. 2 B). The numbers of tissue-infiltrating alloreactive CD3⁺ T cells in GVHD target organs were increased in NOD2^−/− allo-BMT recipients (Fig. 2, C and D). To investigate thymic GVHD, we determined the number of thymocytes and their subsets at day 21 after allo-BMT. We found a significantly reduced number of CD4⁺/CD8⁺ double-positive thymocytes in NOD2^−/− allo-BMT recipients (Fig. 2 E), which is the characteristic feature of thymic GVHD (Hollander et al., 1994). We conclude that NOD2 deficiency in allo-BMT recipients exacerbates systemic and organ GVHD. Our results are in agreement with clinical data connecting NOD2 SNPs in allo-BMT recipients to increased incidence and severity of GVHD (Holler et al., 2004, 2006). A limitation of our study is that in humans the association between NOD2 and GVHD was observed in individuals with heterozygous SNPs, whereas our studies used homozygous NOD2-deficient allo-BMT recipients to investigate the effect of NOD2 deficiency during GVHD. This is similar to studies in mice and men regarding the role of NOD2 in Crohn's disease (Hugot et al., 2001; Watanabe et al., 2004, 2006, 2008; Kobayashi et al., 2005). NOD2 polymorphisms resulting in impaired NOD2 function were associated with increased susceptibility to Crohn's disease. Studies in NOD2-deficient mice demonstrated an increased susceptibility to experimental colitis, whereas mice with a NOD2 transgene resulting in increased NOD2 function were resistant to experimental colitis.

To study the mechanisms underlying increased GVHD in NOD2^−/− allo-BMT recipients, we first transferred CFSE-labeled allogeneic B10BR T cells into B6 WT or B6 NOD2^−/− allo-BMT recipients. We found that the in vivo proliferation of alloreactive donor T cells was significantly increased in NOD2^−/− allo-BMT recipients (Fig. 2 F). Next, we performed allo-BMTs in the B10BR→B6 model and harvested organs at day 14 after allo-BMT. We found that the number of donor T cells was significantly increased in spleen and mesenteric LNs (MLNs) in NOD2^−/− allo-BMT recipients (Fig. 2 G, left). However, the percentages of CD4⁺ T cells, CD8 ⁺ T cells, and CD44⁺/CD62L⁻ effector T cells were not significantly different (Fig. S1). We next determined the alloactivation status of donor T cells by their CD25 expression, which is a reliable marker for alloactivation in our GVHD models (Alpdogan et al., 2003), and found that donor T cells were significantly more activated in NOD2^−/− allo-BMT recipients as compared with WT allo-BMT recipients (Fig. 2 G, middle). When we analyzed FoxP3⁺/CD4⁺ regulatory T cells in the spleen, we found a significantly reduced percentage of FoxP3⁺/CD4⁺ cells (of total CD4⁺ splenocytes) in NOD2^−/− allo-BMT recipients (Fig. 2 G, top right). The percentages of IL-17⁺CD4⁺, IFN-γ⁺CD4⁺, and Tbet⁺CD4⁺ T cells were not different (Fig. S1). There was no difference between the expression of the gut-homing molecule LPAM (α4β7 integrin) on donor T cells in MLNs in NOD2^−/− and WT allo-BMT recipients (Fig. 2 G, bottom right). We detected no differences in donor cell types, such as B cells, NK cells, granulocytes, monocytes, and DCs in NOD2^−/− allo-BMT recipients (Fig. 2 H). We conclude that increased GVHD in NOD2^−/− allo-BMT recipients is the result of enhanced alloactivation and proliferation of donor T cells.

Given that NOD2 is expressed in both hematopoietic and nonhematopoietic cells types, including antigen-presenting cells and intestinal epithelial cells, respectively, it is not clear what relative contributions of these cells types is to the development of GVHD. To address this, we created BM chimeras in which the NOD2 deficiency was only manifest in one of these systems. BM chimeras were created by performing syngeneic BMTs after lethal irradiation, using B6 WT and B6 NOD2^−/− as donors or as recipients (WT→NOD2^−/− or NOD2^−/−→WT). Using B6 Ly5.1⁺ mice as donors, we found the donor chimerism of hematopoietic cell subsets to be >95% in GVHD target organs, as well as in secondary lymphoid organs at day 90 after syngeneic BMT (Fig. S2). We then used the BM chimera, WT mice, and NOD2^−/− mice as allo-BMT recipients in the B10BR→B6 GVHD model. We found that NOD2^−/− and NOD2^−/−→WT chimeric allo-BMT recipients had a greater GVHD-induced mortality rate as compared with WT and WT→NOD2^−/− chimeric allo-BMT recipients (Fig. 3 A, left). We confirmed our findings in the minor mismatched LP→B6 model (Fig. 3 A, right). To analyze target organ GVHD, we performed histopathological analyses and found significantly more GVHD in the terminal ileum, colon, liver, and skin in NOD2^−/−→WT chimeric allo-BMT recipients versus WT→NOD2^−/− chimeric allo-BMT recipients (Fig. 3 B). The numbers of tissue-infiltrating CD3⁺ T cells in GVHD target organs were increased in NOD2^−/−→WT chimeric allo-BMT recipients (Fig. 3 C). We found that the number of donor T cells was significantly increased in the spleen and MLNs in NOD2^−/−→WT chimeric allo-BMT recipients (Fig. 3 D, left). We next determined the alloactivation status of donor T cells by their CD25 expression and found that donor T cells were significantly more activated in NOD2^−/−→WT chimeric allo-BMT recipients as compared with WT→NOD2^−/− chimeric allo-BMT recipients (Fig. 3 D, middle). Additionally, we found a significantly reduced percentage of FoxP3⁺/CD4⁺ regulatory T cells in the spleen (percentage of FoxP3⁺ in total CD4⁺ splenocytes) in NOD2^−/−→WT chimeric allo-BMT recipients (Fig. 3 D, top right). The expression of LPAM on donor T cells in MLNs was similar in both chimeric allo-BMT recipients (Fig. 3 D, bottom right). We conclude that NOD2 deficiency in the hematopoietic system exacerbates GVHD, whereas NOD2 deficiency in the nonhematopoietic system has no significant impact on the development of GVHD in NOD2^−/− allo-BMT recipients.

Based upon our finding that donor T cell proliferation is increased in allo-BMT recipients with NOD2 deficiency in the hematopoietic system, we hypothesized that NOD2^−/− DCs might have a higher activation status and increased ability to induce T cell proliferation during GVHD. We therefore performed allo-BMTs in the B10BR→B6 model and harvested spleens at day 7 after allo-BMT. We found that the absolute numbers of host DC subsets in the spleen were not significantly different before and after allo-BMT in WT versus NOD2^−/− allo-BMT recipients (Fig. 4 A). We then quantified the expression of the activation markers and costimulatory molecules CD40, CD80, and CD86 on host (CD11c⁺H2kb⁺) DCs before allo-BMT and at day 7 after allo-BMT. We found a significantly increased up-regulation of activation markers on host NOD2^−/− DCs in contrast to WT DCs during GVHD (Fig. 4, B and C). The increase in the expression of MHC class II molecules, which was already high at baseline, did not reach statistical significance (unpublished data). To study DC function, we selected splenic DCs from WT and NOD2^−/− allo-BMT recipients with GVHD (day 7) and used them as stimulators in MLRs. We found that B6 NOD2^−/− DCs had a significantly increased ability to induce proliferation of allogeneic B10BR T cells as compared with B6 WT DCs (Fig. 4 D). These data supports the hypothesis that NOD2 can negatively regulate the activity and function of host DCs, resulting in increased alloactivation and proliferation of donor T cells in NOD2^−/− allo-BMT recipients. Our results provide a possible cellular mechanism for recent clinical studies suggesting a regulatory role of NOD2 on the incidence of GVHD and transplant-related mortality in allo-HSCT recipients with SNPs in the gene locus encoding NOD2 (Holler et al., 2004, 2006; Elmaagacli et al., 2006).

In previous studies using NOD2^−/− mice in inflammatory bowel disease models, it was suggested that increased TLR signaling led to a massively increased production of cytokines such as IL-12. (Watanabe et al., 2004, 2006, 2008; Strober et al., 2006; Yang et al., 2007) To determine whether increased cytokine/chemokine production is the underlying mechanism for the enhanced ability of NOD2^−/− DCs to induce T cell proliferation, we performed allo-BMTs in the B10BR→B6 model and measured cytokines/chemokines in serum with a multiplex system at day 7 after allo-BMT. We did not detect significant differences in serum levels of IFN-γ, IL-1α, IL-1β, IL-2, IL-6, IL-7, IL-9, IL-10, IL-12 p40, IL-12 p70, IL-15, IL-17, IP-10, LIX, MCP-1, MIG, MIP-1α, MIP-1β, Rantes, and TNF in NOD2^−/− allo-BMT recipients versus WT allo-BMT recipients (Fig. S3). Although serum levels were not different we cannot exclude differences in cytokine production in specified cell populations. However, our results are in agreement with recent data, which showed that APCs from humans with NOD2 SNPs do not have increased cytokine production (van Heel et al., 2005). Consequently, there might be alternative molecular mechanisms that can enhance APC function and activity in NOD2^−/− allo-BMT recipients during inflammatory diseases such as colitis and GVHD.

Next, we analyzed whether results from our GVHD experiments also apply to other inflammatory diseases. Because the pathophysiology of GVHD and inflammatory bowel disease has many similarities, we performed experiments using the TNBS (2,4,6-trinitrobenzenesulfonic acid) colitis model. TNBS colitis is a transmural inflammation often used as a model for inflammatory bowel disease (Scheiffele and Fuss, 2002). Administration of TNBS leads to modification of intestinal proteins that function as haptens. We rectally challenged B6 WT or B6 NOD2^−/− mice with 5 mg TNBS in 50% ethanol. We found a significantly higher weight loss in NOD2^−/− mice than in WT mice during colitis (Fig. 5 A). We next compared the NOD2^−/−→WT chimeric to the WT→NOD2^−/− chimeric mice (90 d after syngeneic BMT) in the TNBS colitis model. We found significantly more weight loss in NOD2^−/−→WT chimeric mice versus the WT→NOD2^−/− chimeric mice (Fig. 5 B). We harvested colons at day 3 after induction of colitis, performed histopathological analysis, and found significantly higher colitis scores in NOD2^−/−→WT chimeric mice as compared with the WT→NOD2^−/− chimeric mice (Fig. 5, C and D). Our data demonstrate that NOD2 deficiency in the hematopoietic system, but not in the nonhematopoietic system, exacerbates both experimental GVHD and experimental TNBS colitis. These results suggest that loss of intestinal epithelial cell function causing impaired antibacterial resistance, which has been proposed as a mechanism for increased inflammation in NOD2^−/− mice (Kobayashi et al., 2005), is not responsible for the increased GVHD and colitis. Our results are in agreement with previous findings (Watanabe et al., 2004, 2006, 2008; Strober et al., 2006; Yang et al., 2007) that found increased susceptibility to different types of experimental colitis in NOD2^−/− mice, which was the result of the enhanced ability of NOD2^−/− APCs to trigger inflammatory T cell responses.

In response to mixed clinical results on a possible role of NOD2 in antitumor immunity (Mayor et al., 2007; Holler et al., 2008), we performed tumor challenges with hematologic and solid tumors. We studied the role of NOD2 in antitumor responses in two tumor models: acute myeloid leukemia (AML; C1498) and melanoma (B16). When we challenged B6 WT or B6 NOD2^−/− mice intravenously with C1498 or B16, we found no significant differences in survival (unpublished data). Bioluminescence imaging demonstrated that early tumor growth was slightly reduced in B6 NOD2^−/− mice in both models (Fig. S4). We conclude that NOD2^−/− has no major role in antitumor immunity against C1498 AML and B16 melanoma.

Animal protocols were approved by the Memorial Sloan-Kettering Cancer Center Institutional Animal Care and Use Committee. NOD2^−/− mice (Kobayashi et al., 2005) were backcrossed to a B6 background (n = 11). BMT experiments and bioluminescent imaging was performed as described previously (Terwey et al., 2005; Zakrzewski et al., 2006, 2008).

Most antibodies were obtained from BD. DCs were purified by positive selection with CD11c (N418) MicroBeads (Miltenyi Biotec) from digested spleens. Purity was between 90 and 95%. Splenic T cells were enriched by CD5⁺ selection and cocultured. After 48 h, thymidine (1 µCi [0.037 MBq] [³H]thymidine/well) was added and, after 72 h, thymidine uptake was measured (TopCount; Packard).

T cells were selected using CD5 MicroBeads (Miltenyi Biotec), stained with 5 µM CFSE (Invitrogen), and transplanted into lethally irradiated allogeneic recipients.

Serum levels were determined using the Milliplex Mouse Cytokine/Chemokine Immunoassay (Millipore).

Induction of TNBS colitis was performed as published previously (Scheiffele and Fuss, 2002). Severity of colitis was assessed using a semiquantitative scoring system (Scheiffele and Fuss, 2002).

Groupwise comparisons were done with the Mann-Whitney test, one-way ANOVA test, or the Wilcoxon matched pairs test where appropriate. Survival data were analyzed with the Mantel-Cox log-rank test.

Fig. S1 shows that donor T cell subsets are not different in NOD2^−/− versus WT allo-BMT recipients. In Fig. S2, the donor chimerism at day 90 is given. Fig. S3 shows serum levels of cytokines and chemokines during GVHD. Fig. S4 demonstrates that NOD2^−/− mice have intact resistance against experimental tumors.

We would like to thank Renier Brentjens, Richard Flavell, Ivan J. Fuss, Atsushi Kitani, Katia Manova, Yan Nikhamin, Gabriel Nunez, Eric Pamer, Carles Ubeda-Morant, Joan Zhiqiong Yang, and James Young.

This research was supported by National Institutes of Health grants RO1-HL069929 (to M. van den Brink), RO1-CA107096 (to M. van den Brink), RO1-AI080455 (to M. van den Brink), R01 HL084815 (to G.F. Murphy), and PO1-CA33049 (to M. van den Brink). Support was also received from the Ryan Gibson Foundation, the Elsa U. Pardee Foundation, the Byrne Fund, the Emerald Foundation, the Experimental Therapeutics Center of Memorial Sloan-Kettering Cancer Center funded by Mr. William H. Goodwin and Mrs. Alice Goodwin, the Commonwealth Foundation for Cancer Research, the Bobby Zucker Memorial Fund (to M. van den Brink), the Deutsche Forschungsgemeinschaft (to O. Penack), the Deutsche Krebshilfe (to I. Na and A. Ghosh), and the Alexander von Humboldt foundation (to K. Brandl).

The authors have no conflicting financial interests.