Several years ago it was discovered that homozygous (or compound heterozygous) mutations of CARD15, a gene encoding NOD2 (nucleotide oligomerisation domain 2), are a major susceptibility factor underlying a significant subgroup of patients with Crohn’s disease in Western countries.^1,2,3,4 More recently it has been shown that mutations of this gene are also associated with diseases that are immunologically related to Crohn’s disease, such as early onset sarcoidosis.⁵ Thus it has become increasingly clear that these mutations offer an extraordinary window on the function of the mucosal immune system and how defects in such function lead to disease.

The road to understanding the consequences of NOD2 mutations begins with knowledge of how NOD2 functions in normal individuals. NOD2 is now known to be a member of a large and somewhat heterogeneous protein family known as the NOD-LRR (leucine-rich repeat) protein family that are grouped together because they more or less share the same structural regions: an LRR domain, a NOD domain, and a CARD (caspase recruitment domain) (as in the case of NOD2) or PYRIN domain.⁶ The LRR domain is a cognate of a similar domain found in all members of the Toll-like receptor (TLR) family and that is used by such TLRs to recognise an array of microbial components. It is in fact through such recognition that TLRs facilitate “innate” (rapid and non-specific) host defence responses to potential pathogens. From this one can infer that while the microbial components recognised by the LRR domain of most NOD-LRR proteins are still to be defined, these proteins are also microbial component sensing molecules that play a role in innate immunity. The NOD domain is a central region common to most, if not all, NOD-LRR family members that, on binding of a ligand to the LRR, facilitates a conformational change in the molecule and thus its activation. This conformational change also involves the third region of the molecule, the CARD or PYRIN domains, in that it facilitates the interaction of these effector domains with a “downstream” protein capable of initiating an inflammatory programme.⁷ It should be noted however that the CARD and PYRIN domains interact with different downstream molecules and thus lead to different types of inflammatory responses; thus, while activation of the CARD domain results in activation of nuclear factor kappa B (NFκB), activation of the PYRIN domain leads to activation of procaspase 1 and, ultimately, interleukin (IL)-1β.^6,7 Two other points concerning NOD-LRR proteins are firstly that they are intracellular (cytoplasmic) proteins and thus differ from TLRs that are either on the surface of cells or in phagolysomal compartments, and secondly, while the signalling cascades of TLRs and NODs differ very considerably, points of intersection do occur; this introduces the possibility that NOD-LRR family members regulate TLR responses. As we shall see, this is important in the case of NOD2.

A major step forward in the understanding of NOD2 and the structurally similar NOD-LRR family member, NOD1, was the discovery that these molecules bind (via their LRR domains) specific microbial peptides derived from peptidoglycan (PGN), a glycoprotein found in the wall of virtually all bacteria, including those comprising the commensal microbial microflora of the gut. These consist of muramyl dipeptide (MDP) in the case of NOD2 and γ-D-glutamyl meso diaminopimelic acid in the case of NOD1.^8,9 As PGN is also a ligand for TLR2,¹⁰ this finding implies that PGN signals the cell in three ways: at the cell surface as an intact molecule interacting with TLR2 and in the cell cytoplasm as a peptide breakdown product interacting with NOD2 or NOD1. Recently, it has been shown that MDP is also recognised by cryopyrin,¹¹ another NOD-LRR family member that is similar to NOD2 and NOD1, except that it has a PYRIN region instead of a CARD region; this brings to four the number of ways PGN can signal the cell. Interestingly, as in the case of NOD2, mutations in cryopyrin are also associated with inflammatory diseases.¹² Finally, studies of the cellular distribution of NOD2 show that it is expressed in antigen presenting cells (APCs) and in a subset of epithelial cells at the base of the intestinal crypts, known as Paneth cells.¹³ Thus it is apparent that NOD2 is well positioned to influence responses to both commensal and pathogenic organisms in the gut. NOD1, on the other hand, is expressed in all columnar epithelial cells and in APCs.

MDP induced NOD2 signalling results in NFκB activation and subsequent NFκB dependent cytokine responses. However, these are much weaker than those elicited by PGN induced TLR2 signalling and may thus be a relatively unimportant aspect of NOD2 function.^14,15 This is in fact supported by recent studies that show that NOD2 exerts strong regulatory influence on TLR signalling, both positive and negative. An important example of negative regulation that relates to the possible role of NOD2 mutations in Crohn’s disease has emerged from studies of NOD2 effects on PGN induced IL-12 and IL-18 production, two cytokines whose synthesis is dependent on activation of NFκB.¹⁶ These studies have provided evidence that MDP activation of NOD2 induces only low levels of IL-12 (p70 heterodimer) synthesis and, perhaps more importantly, markedly inhibits the capacity of PGN to induce this cytokine.¹⁶ This negative effect of NOD2 signalling is brought into bold relief by study of cells from NOD2 knockout mice that lack NOD2: such cells produce markedly increased amounts of IL-12. The molecular mechanism of this NOD2 regulatory effect appears to involve subtle but important differences in activation of NFκB by PGN in the presence and absence of NOD2 signalling. In the presence of NOD2, PGN signalling results in relatively low amounts of c-Rel translocation to the nucleus, a key event in induction of IL-12 by NFκB.¹⁷ In contrast, in the absence of NOD2, PGN signalling results in relatively high amount of c-Rel translocation and thus greater amounts of IL-12 synthesis.¹⁶

As IL-12 is one of the main initiators of the Th1 inflammatory response and higher amounts of IL-12 production are reflected in higher interferon γ (IFN-γ) production, this regulatory effect of NOD2 on PGN responses could be the basis of the fact that NOD2 mutations (and accompanying deficiency in NOD2 function) is a susceptibility factor in Crohn’s disease. One could postulate here that the “immunostat” for innate immune responses relating to PGN derived from commensal organisms in the gastrointestinal tract is set to a lower level by the NOD2 mutations so that the mucosal environment in patients with NOD2 mutations produce a higher basal level of IL-12 and thus support Th1 responses by any number of antigens derived from ordinarily non-pathogenic organisms in the intestinal microflora. However, while this thesis is compelling in the context of animal models of mucosal inflammation and in mice that are NOD2 deficient, it has yet to be verified in humans.

A second type of regulation of TLR responses exerted by NOD2 signalling is the obverse of that just discussed and involves enhancement of TLR responses. Thus evidence is also at hand showing that NOD2 signalling in human monocytic cell lines has an enhancing effect on IL-8 and tumour necrosis factor (TNF) production induced in these cells by TLR2, TLR4, and TLR9 ligands.^18,19 In addition, as discussed below, NOD2 signalling also has an enhancing effect on TLR9 responses in fresh peripheral blood human cells. These studies indicate that NOD2 modulation of TLR responses can be either positive or negative depending on the types of TLR stimuli.

What molecular mechanisms account for these very different forms of NOD2 regulation? The answer is not yet known but there is some evidence that it involves the way NOD2 and TLRs interact with RICK (RIP2), an adaptor molecule containing a CARD region that links NOD2 and TLR signalling to NFκB activation. As mentioned above, activation of NOD2 by ligand (MDP) binding to its LRR results in a change in conformation of its CARD region. This in turn leads to binding and activation of RICK via a CARD/CARD interaction. Activated RICK then ubiquitinylates the IKKγ subunit of the NFκB IKK complex and thus activates IKKβ and downstream NFκB components.²⁰ Interestingly, PGN signalling via TLR2 also leads to activation of RICK by an independent route involving MyD88 (as discussed below) so that the NOD2 and TLR2 signalling pathways intersect at RICK.²¹ This opens the door to the possibility that NOD2 signalling negatively regulates TLR2 signalling with respect to induction of IL-12 by imposing a form of ubiquitinylation on the IKKγ subunit of NFκB (different from that induced by TLR2) that leads to decreased c-Rel activation, as discussed above. Within this framework we can also explain enhancement of TLR responses by NOD2 signalling if we assume that some TLR responses, such as the TLR9 response to CpG stimulation, leads to activation of NFκB through activation of a canonical TLR pathway that goes through MyD88, IRAK, and TRAF6 and does not involve RICK, as indicated by the fact that RICK deficient macrophages show normal cytokine responses following CpG stimulation.^10,21 In addition, even alternative TLR signalling response that moves from MyD88 directly to RICK may not lead to downregulation of cytokine responses except in the case of IL-12 which is uniquely sensitive to c-Rel activation. On the contrary, here one may see enhancement of TLR responses due to the fact that NOD2 activation upregulates MyD88 expression.²²

In this issue of the Gut, van Heel and colleagues²³ report that in conformity with previous studies mentioned above, whereas MDP stimulation of peripheral blood mononuclear cells (PBMCs) from normal individuals results in 2-3-fold enhancement of CpG DNA stimulation of PBMC production of TNF-α and IL-8, such enhancement is not seen in PBMCs from patients with Crohn’s disease bearing NOD2 mutations (see page 1553). Concomitant studies of IL-12 secretion were not reported in this study or in a previous study of normal individuals and Crohn’s disease patients with NOD2 mutations,²⁴ perhaps because secretion of IL-12 by PBMCs is low and thus difficult to assess. Thus we do not know if the loss of the enhancing effect of NOD2 is counterbalanced by loss of an inhibitory effect. We also do not know if the same or similar findings would be obtained if the authors had studied intestinal cells that differ considerable from peripheral cells in response to various stimuli.²⁵ In any case, based on these findings, van Heel and colleagues²³ propose that synergistic cytokine response between TLR9 and NOD2 might be beneficial in maintaining intestinal homeostasis and the lack of such synergism is a cause of Crohn’s disease.

Intestinal or mucosal homeostasis by any definition is the end result of effector cell responses that protect the individual from infections of the mucosa by pathogens and regulatory responses that protect the individual from excessive responses induced either by pathogens or by commensal organisms in the microflora.²⁶ It is doubtful that disruption of mucosal homeostasis by affecting the effector cell part of the equation is the cause of Crohn’s disease as the latter is well known to be associated with an excessive Th1 immune response, especially one involving TNF-α or IL-12, both cytokines targeted for reduction in the treatment of patients with anti-TNF-α or anti-IL-12 antibodies.^27,28 Nevertheless, this possibility is worth considering in view of recent information concerning the relation of NOD2 signalling in Paneth cells to the production of α-defensins.

Paneth cells are specialised epithelial cells at the base of the intestinal crypts that are characterised by dense granules containing α-defensins, peptides with bactericidal properties.²⁹ As such, they fulfil an important host defence function in the gut, as shown by the fact that transgenic mice that express human α-defensin 5 are resistant to Salmonella typhimurium infection and, conversely, mice lacking the ability to process α-defensins into a mature active form are more susceptible to such infection.³⁰ Of interest to the present discussion, Paneth cells express TLR9 and, as mentioned earlier, they also express NOD2^13,31 In addition, they can be triggered by CpGs and by MDP to secrete α-defensins. Thus it is reasonable to propose that the presence of NOD2 mutations renders epithelial cells unable to produce maximal amounts of α-defensins and, as a result, organisms colonise the intestinal surface that induce the inflammation in Crohn’s disease. Consistent with this notion, there is some as yet incomplete evidence that patients with Crohn’s disease, particularly those with NOD2 mutations, express decreased amounts of α-defensin 5 in the small intestinal mucosa³² and one line of mice lacking NOD2 exhibit reduced α-defensin mRNA expression in Paneth cells and are more susceptible to oral (but not intravenous) L monocytogenes infection of the liver.³³ It should be noted however that these data supportive of the concept that NOD2 mutations produce a kind of immunodeficiency state that predisposes to a type of bacterial colonisation of the gut are counterbalanced by other data that rule against this concept. Perhaps the most persuasive is that mice that lack MMP-7 (matrilysin) and that thus cannot convert α-defensin precursors into mature and active forms of α-defensins³⁴ or, indeed, mice that lack Paneth cells³⁵ altogether, do not spontaneously develop chronic inflammation indicative of a Crohn’s disease-like condition, at least under the pathogen free conditions that are able to support other causes of chronic mucosal inflammation. In addition, convincing data have yet to emerge that the intestinal surface of patients with Crohn’s disease or in spontaneous experimental models of Crohn’s disease is colonised with organisms that cause Crohn’s disease; on the contrary, there is good evidence that commensal organisms have ready access to the internal milieu of the gut,³⁶ despite the presence of an intact α-defensin system, and thus have the potential to induce Crohn’s disease in patients with normal defensin production.

We come then to the second way in which intestinal homeostasis could be disturbed by lack of enhancement of TLR9 responses to CpG by NOD2 responses to MDP—namely, by the failure of the mucosal immune system in this situation to mount adequate regulatory cell (that is, tolerogenic) responses. The basis of this possibility comes from the recent observation that various forms of experimental colitis in mice are paradoxically prevented by provision of CpGs either in the form of immunostimulatory oligonucleotides or as components of so-called probiotic organisms.^37,38 Furthermore, this protective effect can be traced to TLR9 stimulation as it is not seen in mice lacking TLR9 by gene targeting or in mice deficient in the signalling of type I IFN,³⁹ the cytokine class thought to be produced by plasmacytoid dendritic cells implicated in the induction of regulatory T cells.⁴⁰ Thus these data suggest that impairment in optimal responses to CpGs via TLR9, such as that caused by lack of enhancement by NOD2 signalling, could lead to deficient regulatory T cell development and the development of Crohn’s disease due to failure to maintain tolerance to commensal organisms.

This is an interesting possibility but one that is far from being proven. Firstly, while there is ample evidence that experimental colitis can be due to deficient regulatory cell responses, there are few data that this is the case in human Crohn’s disease.²⁶ Secondly, while van Heel and colleagues²³ have shown that NOD2 signalling enhances TLR9 signalling,²³ there is no evidence that TLR9 signalling by itself can induce regulatory T cells. Finally, and perhaps most importantly, there is a large body of evidence that under most circumstances TLR9 signalling induces Th1 responses (that is, the very type of cytokine response that underlies Crohn’s disease).⁴¹ This may explain the fact that delivery of CpG to mice with already established inflammation leads to intensified inflammation.^42,43 The fact is that we still do not know enough about TLR9 signalling and its relation to induction of effector versus regulatory T cells to predict whether enhancement of TLR9 responses would exacerbate or abate Crohn’s disease.

The above discussion shows very clearly that we have gained a great deal of knowledge concerning the role of NOD2 in normal individuals in the past several years. Nevertheless, much remains to be learned concerning its role in disease. We believe that ultimately NOD2 mutations will prove to be the cause of Crohn’s disease, primarily because they lead to loss of negative regulatory effects on TLR signalling. However, additional studies in patients with mutations such as the one reported in this issue of Gut will be necessary to elucidate the issue.