Nod2 Mutation in Crohn's Disease Potentiates NF-κB Activity and IL-1ß Processing

See allHide authors and affiliations

Science  04 Feb 2005:
Vol. 307, Issue 5710, pp. 734-738
DOI: 10.1126/science.1103685

This article has corrections. Please see:


Variants of NOD2, an intracellular sensor of bacteria-derived muramyl dipeptide (MDP), increase susceptibility to Crohn's disease (CD). These variants are thought to be defective in activation of nuclear factor κB (NF-κB) and antibacterial defenses, but CD clinical specimens display elevated NF-κB activity. To illuminate the pathophysiological function of NOD2, we introduced such a variant to the mouse Nod2 locus. Mutant mice exhibited elevated NF-κB activation in response to MDP and more efficient processing and secretion of the cytokine interleukin-1β (IL-1β). These effects are linked to increased susceptibility to bacterial-induced intestinal inflammation and identify NOD2 as a positive regulator of NF-κB activation and IL-1β secretion.

Crohn's disease (CD) is a chronic inflammatory bowel disease (IBD) thought to be caused by genetic and environmental factors that affect host-microbe interactions and production of inflammatory mediators (1, 2). Mutations that increase susceptibility to CD up to 40 times were mapped to the NOD2/CARD15 locus (3, 4). The NOD2 protein contains two N-terminal caspase recruitment domains (CARDs), a nucleotide-binding domain (NBD), and 10 C-terminal leucine-rich repeats (LRRs), and it is expressed mainly by macrophages and dendritic cells (5). NOD2 mediates intracellular recognition of MDP, a building block for bacterial cell walls (6, 7), and can activate NF-κB (5). Macrophages within the intestinal lamina propria of CD patients overproduce NF-κB targets, including the proinflammatory cytokines tumor necrosis factor–α (TNFα) and the interleukins IL-1β and IL-6 (2, 8). Many of the anti-inflammatory drugs used to treat CD inhibit NF-κB activation, which suggests it is a key pathogenic factor (8, 9). However, paradoxically, transient transfection experiments indicate that CD-associated NOD2 variants no longer activate NF-κB in response to MDP (6, 7), which suggests that defective NF-κB activation in macrophages facilitates infection of the lamina propria by enteric bacteria. However, macrophages can activate NF-κB in response to bacteria independently of NOD2 (10), and Nod2 gene ablation did not cause spontaneous intestinal infections or colonic inflammation (11).

To find an explanation for these quandaries and to illuminate the mechanism by which CD-associated NOD2 variants act, we generated mice whose Nod2 locus harbors the homolog of the most common CD susceptibility allele, 3020insC, which encodes a truncated protein lacking the last 33 amino acids (3, 4). This was done through insertion of cytosine at position 2939 (corresponding to 3020 in human NOD2) of the Nod2 open reading frame (Fig. 1, A and B). Homozygous Nod22939iC mice were obtained at the expected mendelian ratio and did not show abnormalities of the gastrointestinal tract (fig. S1) or other organs; they were healthy (12). The mutation had no effect on Nod2 mRNA or protein amounts in bone marrow–derived macrophages (BMDMs) (Fig. 1, C and D).

Fig. 1.

Generation of Nod22939iC mice. (A) Schematic structure of NOD2, sequence of WT, and mutant alleles around the 2939insC mutation, targeting vector, and the targeted locus. Solid boxes represent exons, and lines introns. The Neor cassette was inserted opposite the Nod2 transcription unit. (B) Southern blot analysis of Nco I–digested genomic DNA from F2 mice of the indicated genotypes, where m is mutant allele; +, WT allele. (C) Nod2 mRNA in BMDMs. RNA was converted to cDNA and amplified by using primers for three different regions of Nod2 cDNA. (D) Expression of WT and truncated (m/m) NOD2 proteins. BMDM lysates were immunoblotted with antibodies against NOD2 and actin in order to control loading.

We examined the effect of the Nod22939ic mutation on NF-κB activation in BMDM cultures. The activity of IKK, which is the inhibitor of κB (IκB) kinase, the degradation of IκBα, and NF-κB DNA binding activity were higher in MDP-stimulated Nod22939iC macrophages than in wild-type (WT) cells (Fig. 2A). Only marginal differences in mitogen-activated protein kinases (MAPKs) were observed (fig. S2). No genotype-specific differences in NF-κB activation were observed after macrophage treatment with other microbial components that activate Toll-like receptors (TLRs) (10), including the TLR2 agonists Pam3Cys (tripalmitoyl-S-glyceryl-Cys-Ser-4(Lys)) and peptidoglycan (PGN), the TLR4 agonist lipopolysaccharide (LPS), and the TLR9 agonist nonmethylated CpG-containing DNA (Fig. 2B) (12). Expression of several NF-κB target genes was increased in MDP-treated Nod22939iC macrophages relative to WT counterparts (Fig. 2C). Only minor differences in expression of these genes were observed when macrophages were stimulated with LPS or PGN. Although MDP-induced gene expression of several cytokine genes was increased in Nod22939iC macrophages, only IL-1β secretion was significantly elevated in these cells relative to WT counterparts (Fig. 2, D and E, fig. S3). Secretion of IL-1α was modestly elevated, and neither IL-6 nor TNFα were secreted in response to MDP. The only microbial product that stimulated IL-1β secretion by Nod22939iC macrophages was MDP (Fig. 2E).

Fig. 2.

Nod22939iC macrophages exhibit elevated NF-κB activation and IL-1β secretion in response to MDP. (A) BMDMs from WT and Nod22939iC (m/m) mice were incubated with MDP (1 μg/ml). Where indicated, cytosolic and nuclear extracts were prepared and used to analyze IKK activation (KA), IκBα degradation, and NF-κB DNA binding activity, respectively. Nuclear extract quality was monitored by measuring nuclear factor-Y (NF-Y) DNA binding. (B) BMDMs were stimulated with Pam3Cys (1 μg/ml), LPS (100 ng/ml), or CpG DNA (1 μM) to activate TLR2, 4, and 9, respectively. Where indicated, nuclear extracts were prepared and NF-κB DNA binding activity was analyzed. (C) Expression of NF-κB target genes was examined in Nod22939iC and WT macrophages stimulated with MDP, LPS, or peptidoglycan (PGN from Staphylococcus aureus, 10 μg/ml). After 4 hours, cells were collected, total cytoplasmic RNA was prepared, and gene expression was analyzed by real-time polymerase chain reaction (PCR). Data are presented as the fold increase in mRNA expression in Nod22939iC macrophages relative to WT macrophages, which was given an arbitrary level of 1.0 for each gene. Results are means ± SEM of three independent experiments. (D) Elevated IL-1β secretion in MDP-stimulated Nod22939iC macrophages. WT and Nod22939iC (m/m) BMDMs were stimulated as indicated. After 24 hours, culture supernatants were collected and secreted cytokines were measured. (E) MDP induces IL-1β release by Nod22939iC (m/m) BMDMs. Macrophages were treated with MDP or LPS for 24 hours. Culture supernatants were collected and analyzed by immunoblotting with antibodies against IL-1β and TNFα.

Macrophages involved in CD most likely reside in the lamina propria (2). To expose these cells to enteric bacteria, mice were treated with dextran sodium sulfate (DSS), an agent that kills mucosal epithelial cells and disrupts their barrier function, causing bacterial invasion (13). WT and homozygous Nod22939iC mice (8 to 12 weeks old) were given 3% DSS in drinking water for 6 days and monitored for weight loss, a characteristic of severe intestinal inflammation. After 8 days, body weight loss was greater in Nod22939iC mice relative to WT mice (Fig. 3A). Nod22939iC mice also exhibited increased mortality relative to WT mice (37.5% versus 0%) (fig. S4). Surviving mice of both genotypes regained body weight after day 11 and returned to normal 30 days after DSS administration (12). Histological analyses revealed that the severity and extent of inflammatory lesions in the colons of Nod22939iC mice were significantly (P < 0.05) greater than in WT controls, with larger areas of ulceration and increased infiltration of F4/80-positive macrophages (Fig. 3B, fig. S5).

Fig. 3.

Enhanced NF-κB activation and inflammation in DSS-treated Nod22939iC mice. (A) Increased body weight loss in DSS-exposed Nod22939iC mice. Mice of either genotype were given 3% DSS in drinking water for 6 days and weighed daily. Data are means ± SEM. Significant differences, *P < 0.05. (B) Typical colon appearance (upper panels) and histology (bottom panels) 11 days after initiation of DSS administration. Nod22939iC mice exhibit more inflammation and ulceration. Arrowheads, borders of ulcers. Magnification, 100×. (C) Induction of inflammation-associated genes in colons of DSS-treated mice. Colonic RNA isolated 11 days after initiation of DSS treatment was analyzed by real-time PCR. Results are means ± SEM of fold increase in normalized (relative to glyceraldehyde-3-phosphate dehydrogenase mRNA) mRNA amounts in DSS-treated mice over untreated mice of the same genotype (n = 4 per group). (D) Elevated IL-1β and IL-6 in colons of DSS-treated Nod22939iC mice. The indicated cytokines were measured in colonic extracts prepared 0 or 11 days after DSS exposure. Results are means ± SD (n = 4 to 8). Significant difference, *P < 0.05. (E) Immunohistochemical detection of IL-6 and Cox-2. Colon sections prepared 11 days after initiation of DSS treatment were analyzed by indirect immunoperoxidase staining for IL-6 and Cox-2. Magnification, 100×. (F) Colonic NF-κB and IKK activities. Nuclear and cytosolic extracts of colonic mucosa prepared 0 and 11 days after initiation of DSS administration were analyzed for NF-κB DNA binding and IKK kinase (KA) activities. Protein recovery in nuclear extracts was determined by immunoblotting with antibody against histone deacetylase (HDAC), and IK recovery was determined with antibody against IKK (IKKα).

After DSS exposure, Nod22939iC homozygotes expressed greater amounts of mRNAs encoding proinflammatory cytokines and chemokines in their colons relative to WT mice (Fig. 3C). The amounts of IL-1β, IL-6, and cyclooxygenase-2 (Cox-2) protein were significantly higher in colons of DSS-treated Nod22939iC mice relative to WT counterparts (Fig. 3D). IL-6 and Cox-2 were predominantly expressed in F4/80-positive macrophages within inflammatory lesions (Fig. 3E, fig. S6) (12). IKK and NF-κB activities and RelA(p65) nuclear staining were also higher in colons of Nod22939iC mice than in the WT (Fig. 3F, fig. S7). MAPK activation, however, was only marginally affected by the genotype (fig. S8).

The intestinal inflammatory response to DSS is dramatically reduced by oral antibiotics, which supports involvement of enteric bacteria (14). When given a high dose of DSS (6%) without oral antibiotics, WT and Nod22939iC mice died within 9 days after DSS administration (12), but mice that received oral antibiotics survived and developed mild inflammation and weight loss, without any genotype-linked differences (fig. S9). Thus, enteric bacteria elicit the inflammatory response to DSS, and without bacterial exposure, Nod22939iC mice have the same reaction as WT counterparts.

Exposure of macrophages to bacteria activates inflammatory and apoptotic caspases (15). More apoptotic cells, most of which were positive for the F4/80 macrophage marker, were found in the lamina propria of DSS-treated Nod22939iC mice than in WT counterparts (Fig. 4, A and B). Increased macrophage apoptosis is associated with activation of caspase-1 (16), an enzyme required for secretion of mature IL-1β (17, 18). Congruently, only background levels of secreted IL-1β were present in colons of untreated mice, but IL-1β concentrations were elevated after DSS treatment, particularly in Nod22939iC mice (Fig. 3D). Macrophage activation with LPS induces pro–IL-1β, but its processing and release require activation of caspase-1 by a different signal (16). LPS did not induce secretion of mature IL-1β in either Nod22939iC or WT macrophages, although it stimulated TNFα release (Fig. 2, D and E). In contrast, MDP stimulated release of mature IL-1β but not TNFα by Nod22939iC macrophages. To determine whether IL-1β secretion may be involved in the increased inflammatory response to DSS in Nod22939iC mice, mice were injected once daily with IL-1 receptor antagonist (IL-1-RA) from the start of DSS exposure. Average body weight loss and histological score were improved in IL-1-RA–treated mice, and differences in weight loss (Fig. 4C) and inflammatory score (Fig. 4D, fig. S10) between the genotypes were abolished.

Fig. 4.

IL-1β is an important contributor to elevated colonic inflammation in Nod22939iC mice. (A and B) Increased macrophage apoptosis in Nod22939iC (m/m) mice treated with DSS. Tissue specimens prepared 0 and 11 days after initiation of DSS administration were analyzed by TUNEL staining (A) or by TUNEL (green) plus immunoperoxidase staining for F4/80 (red) (B). Magnification: (A), 200×; (B), 400×. (C) Increased body weight loss in DSS-exposed Nod22939iC mice is IL-1β dependent. Mice of either genotype were given 3% DSS for 6 days with or without concomitant treatment with IL-1RA (100 mg/kg per day). Mice were weighed daily. Data are means ± SEM. Asterisks indicate significant differences (WT versus WT IL-1-RA, m/m versus m/m IL-1-RA; P < 0.05). (D) Histological inflammation and tissue damage scores were determined 11 days after initiation of DSS treatment in the mice from (C). Results are means ± SEM. Significant differences, *P < 0.05.

By contrast to the Nod22939iC mutation, deletion of Ikkb in hematopoietic and myeloid cells reduced the inflammatory response to DSS (fig. S11). However, its deletion in enterocytes increased the inflammatory response to DSS (19).

Collectively, our results suggest that Nod22939iC is a gain-of-function allele, whose product induces elevated IKK and caspase-1 activation in response to MDP. Although NOD2 was suggested to be a negative regulator of TLR2 (20), we found no effect of the Nod22939iC mutation on signaling by TLR2, as coincubation of macrophages with MDP plus a TLR2 agonist (PGN) did not reduce the response to PGN (Fig. 2D). The inhibitory function hypothesis is also inconsistent with in vivo findings in Nod2 knockout mice, which did not show increased inflammation (11). The gain-of-function hypothesis is consistent with clinical observations made in CD patients (8, 21).

The NF-κB signaling pathway induces many proinflammatory genes coding for cytokines and chemokines, including IL-1β, TNFα, and IL-6 (22, 23), and may therefore be an important pathogenic factor in CD (8). Although increased transcription of many NF-κB targets was observed, the results with IL-1β were unique, as it was the only proinflammatory cytokine whose secretion in response to MDP was markedly elevated in Nod22939iC macrophages relative to WT counterparts. Our results suggest that IL-1β is indeed an important contributor to the increased colonic inflammation in Nod22939iC mice, as previously suggested for CD patients (2).

Although NF-κB was thought to be the major effector for NOD2, it should be noted that NF-κB is more effectively activated by bacterial products through TLRs (see Fig. 2). Thus NF-κB activation is not unique to NOD2, and its loss may not compromise NF-κB signaling in response to bacterial infection. Recently, TLR signaling and a certain amount of enteric bacteria were shown to be critical for maintenance of the intestinal barrier function (24), a function that was suggested to deteriorate in CD patients (2). However, maintenance of barrier function is unlikely to involve NOD2. By contrast, a unique function of NOD2, not provided by TLRs, is induction of IL-1β processing and release. This function may be mediated through the N-terminal CARD domains of NOD2, which may directly interact with caspase-1 or upstream caspases. Given the importance of IL-1β for the pathology of DSS-induced colitis in Nod22939iC mice and the imbalance between IL-1β and IL-1RA in CD patients (2), it would be of interest to critically evaluate its role in CD pathogenesis.

Supporting Online Material

Materials and Methods

Figs. S1 to S11

References and Notes

References and Notes

View Abstract

Navigate This Article