Nod2-Dependent Regulation of Innate and Adaptive Immunity in the Intestinal Tract

See allHide authors and affiliations

Science  04 Feb 2005:
Vol. 307, Issue 5710, pp. 731-734
DOI: 10.1126/science.1104911


The gene encoding the Nod2 protein is frequently mutated in Crohn's disease (CD) patients, although the physiological function of Nod2 in the intestine remains elusive. Here we show that protective immunity mediated by Nod2 recognition of bacterial muramyl dipeptide is abolished in Nod2-deficient mice. These animals are susceptible to bacterial infection via the oral route but not through intravenous or peritoneal delivery. Nod2 is required for the expression of a subgroup of intestinal anti-microbial peptides, known as cryptdins. The Nod2 protein is thus a critical regulator of bacterial immunity within the intestine, providing a possible mechanism for Nod2 mutations in CD.

Homozygous mutations in Nod2, a member of the nucleotide-binding oligomerization domain–leucine-rich repeat (NOD-LRR) family of proteins, are highly correlated with the incidence of Crohn's disease (CD), which suggests that Nod2 plays an important role in intestinal immunity (1, 2). These CD-associated genetic variants are deficient in their ability to sense muramyl dipeptide (MDP), a conserved structure in bacterial peptidoglycan that is normally recognized by Nod2 (35). Consistent with these functional studies, homozygosity and compound heterozygosity increase the relative risk of developing CD by as much as 40-fold as compared to simple heterozygosity (∼2- to 4-fold) (1, 2, 6). Nevertheless, the physiological role of Nod2 in intestinal immunity remains unclear.

To assess the function of Nod2, we generated Nod2–/– mice using a targeting construct to replace the NOD, which is essential for activation of the protein (fig. S1, A and B) (7). Nod2–/– mice were outwardly healthy and displayed normal lymphoid and myeloid cellular composition in the thymus and spleen (fig. S2). The animals also displayed no overt symptoms of intestinal inflammation when observed for up to 6 months, which was confirmed by histological examination (fig. S3), and there was no significantly enhanced susceptibility to colitis in the dextran sulfate–induced model (fig. S4).

Detection of MDP by Nod2 was confirmed by stimulation of wild-type and Nod2–/– bone marrow–derived macrophages (BMDMs) with the following Toll-like receptor (TLR) agonists: lipopolysaccharide (LPS, TLR4 ligand); Pam3CS(K)4 (synthetic lipopeptide, TLR2 ligand); and double-stranded RNA [polyinosinic:polycytidylic acid (poly I:C), TLR3 ligand] in the presence or absence of MDP (812). The synergistic effect of MDP and TLR ligands for the production of interleuken-6 (IL-6) and IL-12 p40 in wild-type macrophages was dependent on the dose of MDP added (Fig. 1, A and B, and fig. S5F) and was absent in Nod2–/– macrophages, revealing that Nod2 is required for the detection of MDP in macrophages. The stimulation of wild-type macrophages with MDP, which results in the activation of nuclear factor κB (NF-κB), p38, and extracellular signal–regulated kinase (ERK), was strongly reduced in Nod2–/– macrophages, again indicating that Nod2 is required for MDP signaling (Fig. 1C). In light of the expression of Nod2 in CD40+/CD86+ dendritic cells (DCs) (13), we tested the potential involvement of the protein in the detection of MDP by DCs. Although immature bone marrow–derived dendritic cells (BMDCs) from wild-type mice were capable of secreting the proinflammatory cytokines IL-6 and tumor necrosis factor–α (TNF-α), these cytokines were not detectable from Nod2–/– immature BMDCs (Fig. 1D). MDP alone did not induce cytokine production from BMDMs.

Fig. 1.

Nod2 cooperates with TLR signaling and is required for MDP recognition in vivo. (A) BMDMs from wild-type (WT) and Nod2–/– mice were stimulated with LPS (10 ng/ml) (from Alexis) or poly(I:C) (100 μg/ml) in the presence of MDP (0, 1, or 10 μg/ml) for 24 hours. Production of IL-6 was assessed by ELISA. (B) Production of IL-12 p40 was assessed as in (A). (C) BMDMs from wild-type and Nod2–/– mice were stimulated with MDP (10 μg/ml) for indicated periods. Total cell lysates were prepared and blotted with antibodies against phosphorylated (p) and unphosphorylated forms of IκBα, p38, and ERK1/2. (D) Immature BMDCs were stimulated with MDP (25 μg/ml) for 20 hours, and production of IL-6 and TNF-α was assessed by ELISA. (E) Wild-type (n = 8) and Nod2–/– (n = 6) mice (8 to 12 weeks old) were primed with MDP (250 μg) and challenged with LPS (250 μg, Ultrapure LPS from InvivoGen). The survival of each mouse genotype was plotted. As a control, wild-type (n = 11) and Nod2–/– (n = 8) mice were challenged with LPS without MDP priming.

Previous studies have shown that pretreatment with MDP sensitizes mice to endotoxic shock induced by LPS injection (14). Although pretreatment with 250 μg of MDP and subsequent highly purified LPS injection were fatal in wild-type mice by day 6, Nod2–/– mice were resistant to LPS challenge (Fig. 1E). LPS injection without MDP priming resulted in the similar survival of both wild-type and Nod2–/– animals. These indicate an essential role for Nod2 in detecting MDP in vivo.

Nod2 has recently been proposed to serve as a negative regulator of TLR2 signaling in generating the Th1 phenotype (15). In testing the response of Nod2–/– BMDMs for the production of inflammatory cytokines, including IL-6, IL-12 p40, TNF-α, and IL-1β, after stimulation of various TLR ligands, including TLR2 ligands (fig. S5, A to E), we failed to observe any significant diminution in the response of Nod2–/– macrophages to either LPS and TLR2 ligands (fig. S5, A to C). We also observed a synergistic effect of lipopeptide Pam3CS(K)4 (TLR2 ligand) and MDP in wild-type macrophages but not in Nod2–/– macrophages (fig. S5F). These results suggest that any potential negative role of Nod2 in the TLR2 response is not a universal phenomenon. Determination of whether there is a contribution to TLR2 response from specific experimental details, genetic background, or cell types examined will require further study.

In light of previous studies showing the role of TLRs in the adaptive immune system (16), we immunized wild-type and Nod2–/– mice with MDP (100 μg per mouse) and with human serum albumin (HSA) (50 μg per mouse) by intraperitoneal injection. Three weeks later, we boosted with MDP (20 μg per mouse) and HSA (10 μg per mouse). Serum samples were obtained 2 weeks after the first immunization and 1 week after the boost, and antigen-specific serum immunoglobulin (Ig) was assessed by enzyme-linked immunosorbent assay (ELISA). Although HSA-specific antibodies were detected 2 weeks after immunization in wild-type mice, Nod2–/– animals displayed a severe deficiency in the production of antigen-specific Ig, specifically in IgG1, which is the predominant Nod2-dependent isotype (Fig. 2, A and B). The defect in antigen-specific Ig production was observed even after boosting mice with MDP and HSA. Nod2–/– mice failed to produce HSA-specific Ig and IgG1 (Fig. 2, C and D). In contrast, Nod2–/– mice were capable of producing HSA-specific Ig and IgG1 at a similar level to wild-type mice when they were immunized with resiquimod (R-848, a synthetic ligand for mouse TLR7) (Fig. 2, E and F) (17). These results indicate that Nod2 is able to activate adaptive immunity and to mediate adjuvant activity in the production of antibody to T cell–dependent antigens.

Fig. 2.

Impaired antigen-specific immunoglobulin production in Nod2–/– mice immunized with MDP. (A) Wild-type (n = 5) and Nod2–/– (n = 4) mice (8 to 12 weeks old) were immunized with HSA (100 μg) and MDP (50 μg) and bled 2 weeks after immunization. Antigen-specific Ig was measured by ELISA. The P value was determined by Student's t-test. (B) Antigen-specific IgG1 was assessed by ELISA as in (A). (C) The immunized mice were boosted 3 weeks later with HSA (20 μg) and MDP (10 μg) and bled 1 week later. Antigen-specific Ig was measured by ELISA. (D) Antigen-specific IgG1 was assessed by ELISA as in (C). (E) Wild-type (n = 5) and Nod2–/– (n = 4) mice (8 to 12 weeks old) were immunized with HSA (100 μg) and R-848 (100 nmol) and bled 2 weeks after immunization. Antigen-specific Ig was measured as in (A). (F) Antigen-specific IgG1 was assessed by ELISA as in (E).

To investigate whether Nod2 plays a specific role in the innate immune response against bacterial infection, wild-type and Nod2–/– mice were challenged with the Gram-positive intracellular bacterium Listeria monocytogenes by intravenous injection. No significant difference was observed in the number of bacteria recovered from both the liver and the spleen 24 or 48 hours later (Fig. 3A). Similarly, serum IL-6 production, which normally accompanies bacterial infection, was not elevated. (Fig. 3B). Because L. monocytogenes naturally infects humans by the intestinal route, mice were also challenged with L. monocytogenes by intraperitoneal injection, and again no significant difference in survival between wild-type and Nod2–/– mice was seen (Fig. 3C). Consistent with these data, in vitro infection of macrophages from Nod2–/– mice with L. monocytogenes induced equivalent level of TNF-α and resulted in similar intracellular bacterial growth and killing (fig. S6, A to C). In contrast, Nod2–/– mice challenged with L. monocytogenes via intragastric dosing were susceptible to infection and showed significantly greater numbers of bacteria recovered from both the liver and the spleen than did wild-type mice (Fig. 3D). These results suggest that Nod2 plays a pivotal and specific role in protecting against bacterial infection in the intestine. The number of L. monocytogenes in intestinal Peyer's patches was not significantly different between wild-type and Nod2–/– mice, which suggests that Nod2 might be involved in a Peyer's patch–independent route of bacterial invasion (fig. S7). Expression of Nod2 in the terminal ileum was examined by using purified villi and crypts (Fig. 3E) (18). Reverse transcription polymerase chain reaction (RT-PCR) revealed that Nod2 is highly expressed in crypts but not in villi, which is consistent with a recent report showing expression in human intestinal Paneth cells (Fig. 3F) (19, 20).

Fig. 3.

Nod2 is required for the expression of cryptdins and for protection against intracellular bacterial infection in the intestine. (A) Wild-type and Nod2–/– mice (8 to 12 weeks old) were infected with 3 × 103 L. monocytogenes (strain 10403S) intravenously, and the number of bacteria in the liver and the spleen was counted 24 and 48 hours after infection. (B) Wild-type (n = 5) and Nod2–/– (n = 5) mice were infected with 104 L. monocytogenes intravenously and serum IL-6 was measured by ELISA. (C) Wild-type (n = 14) and Nod2–/– (n = 13) mice were infected with 106 L. monocytogenes intraperitoneally and the survival of the mice was plotted. (D) Wild-type (n = 10) and Nod2–/– (n = 10) mice were infected with 109 L. monocytogenes intragastrically by gastric gavage, and the number of bacteria in the liver and the spleen was counted 72 hours after infection. The P value was determined by Mann-Whitney test. Circles indicate each animal and bars indicate the mean value. (E) Preparation of mouse small intestinal crypts and villi. Individual mouse crypts were stained for 2 min with 0.25% amido black (marked by an arrow). (F) Villi and crypts were isolated from the terminal ileum of wild-type and Nod2–/– mice, and the expression of Nod2 was examined by RT-PCR. Glyceraldehyde phosphate dehydrogenase (GAPDH) and Defcr3 were used for loading control and quality of the crypt preparation, respectively. (G to I) Wild-type and Nod2–/– mice (8 to 12 weeks old) were infected with 109 L. monocytogenes intragastrically by gastric gavage. mRNA was extracted from the terminal ileum from Nod2–/– mice before (n = 5) and 24 hours after (n = 4) infection and from littermate wild-type mice before (n = 9) and 24 hours after (n = 4) infection. The expression of Defcr4, Defcr-rs10, and Defcr5 was examined by quantitative real-time PCR.

To investigate potential genes that might be induced by Nod2 during intestinal infection, we isolated RNA samples from both the wild-type and Nod2–/– terminal ileum before and after Listeria infection by gastric gavage and screened them by microarray analysis (table S2). The up-regulation of candidate genes, which might explain the susceptibility of Nod2–/– mice to bacterial infection, was confirmed by quantitative real-time PCR (Fig. 3, G to I). The most significant difference was in the expression of a subgroup of cryptdins (called α-defensins in humans). Thus, expression of defensin-related cryptdin 4 (Defcr4) and Defcr-related sequence 10 (Defcr-rs10) was very low in Nod2–/– mice and was further reduced after infection in Nod2–/– animals relative to wild-type mice (Fig. 3, G and H). Cryptdins are antimicrobial peptides that are preferentially produced in intestinal Paneth cells, and their antimicrobial activity is important in suppressing infection with pathogenic bacteria, including L. monocytogenes (21) and Mycobacterium paratuberculosis, an organism implicated in CD (22, 23). Of the cryptdins, Defcr4 has the most potent bactericidal activity (24), with its expression being highest in the lower ileum, in contrast to other cryptdins (25, 26). By comparison, Defcr5 was expressed normally in Nod2–/– mice both before and after infection (Fig. 3I) (25).

Our results indicate that Nod2 is essential in the detection of bacterial MDP and is capable of activating the adaptive immune system by acting as an adjuvant receptor for antibody production, either directly or by enhancing the production of α-defensins (27, 28) or other immunostimulatory molecules. Therefore, Nod2 is critical in protecting the host from intestinal bacterial infection. More specifically, we reveal an important role for Nod2 in the regulation of a subgroup of cryptdins, offering a plausible mechanism to explain the association between Nod2 and susceptibility to CD. Murine cryptdins represent a more diverse family than those of human α-defensins and are already known to be critical in the innate immune responses to bacterial infection (29). CD-associated Nod2 mutations predispose primarily to ileal lesions (3033), corresponding to the location of Paneth cells. Recent reports suggest that the expression of α-defensins is diminished in human CD patients, particularly those who have Nod2 gene mutations (34, 35). However, it remains to be established whether a defect in Paneth cell function is the only possible mechanism by which Nod2 mutations might associate with the development of CD in humans. Nevertheless, it seems reasonable to suggest that mutations in Nod2 might promote CD through defective regulation of responses to commensal and/or pathogenic bacteria, rather than acting as an initiating factor for disease. Further studies may resolve this issue and may lead to the development of more rational therapeutic approaches for treating CD.

Supporting Online Material

Materials and Methods

Figs. S1 to S7

Tables S1 and S2


References and Notes

View Abstract

Navigate This Article