|Home | About | Journals | Submit | Contact Us | Français|
The innate immune system is the first line of defense against microorganisms and is conserved in both plants and animals. The NLR protein family is a recent addition to the members of innate immunity effector molecules. These proteins are characterized by a central oligomerization domain termed NACHT (or NBD/NOD) and a protein interaction domain, LRRs (Leucine rich repeats) at the C-terminus. It has been shown that NLR proteins are localized to the cytoplasm and recognize microbial products. To date, it is known that Nod1 and Nod2 detect bacterial cell wall components, whereas IPAF and NAIP detect bacterial flagellin and NALP1 has been shown to detect anthrax lethal toxin. NLR proteins comprise a diverse protein family (over 20 in humans), indicating that NLRs have evolved to acquire specificity to various pathogenic microorganisms, thereby controlling host-pathogen interactions. Activation of NLR proteins results in inflammatory responses mediated either by NF-κB, MAPK or Caspase-1 activation, accompanied by subsequent secretion of pro-inflammatory cytokines. Mutations in several members of the NLR protein family have been linked to inflammatory diseases, suggesting these molecules play important roles in maintaining host-pathogen interaction and inflammatory responses. Therefore, understanding NLR signaling is important for the therapeutic intervention of various infectious and inflammatory diseases.
The relationships among organisms have evolved into four different outcomes: symbiosis, mutualism, commensalism and parasitism. Parasitism is a burden to the host organism and have forced the host to evolve methods to detect and eliminate parasitic organisms. Higher organisms are equipped with a basic defense mechanism, called innate immunity. The innate immune system is phylogenetically ancient compared to the more evolved form of immunity, the adaptive immune system, which exists only in vertebrates. Unlike the adaptive immune system, innate immunity does not require gene rearrangement and its diversity is dependent on a number of innate immunity genes and their splice variants encoded within the host genome. Therefore, innate immunity is able to rapidly respond against pathogens, and serve as the first line of defense. Molecules recognized by host innate immunity are specific to microbial organisms and are called PAMPs (pathogen associated molecular patterns) . PAMPs are recognized by Pattern-recognition receptors (PRRs) of the innate immune system .
To detect microorganisms, plants and animals have evolved to have two large gene families. One is a family of cell surface receptors with Leucine-rich repeats, LRRs (called Toll-like receptors or TLRs in humans and eLRR in plants), and the other is a family of cytoplasmic proteins composed of NBD (nucleotide binding domain or NBS) and LRRs [1-6]. In both plants and animals, the most diverse family of genes in the innate immune system is the cytoplamic NBD-LRR family [4, 6], having 3-30 fold more members than the cell surface LRR receptors, suggesting that NBD-LRRs can detect a large array of microorganisms in order to protect the host organism.
The critical role of NBD-LRRs in host-pathogen interactions was first discovered in plants . These molecules, that respond to microbial virulence factors, have the structure of an NBS (nucleotide binding site) followed by LRRs and are referred to as disease resistance genes (R genes) . NBS-LRRs in plants are a diverse protein family and mediate some of the most important mechanisms of host defense against infection in plants. It has been recently shown that some of the R gene-virulence factor interactions are not direct and the R genes seem to recognize enzymatic activities of the virulence proteins on an intermediate host protein [9, 10]. The NLR protein family in animals is a newly emerging class of innate immune molecules. Their similarity to plant R genes was first suggested by the cloning of CARD4/NOD1 [11, 12] and it was postulated early on that CARD4/NOD1 may act as a cytoplasmic sensor of microbial products . Today, it is known that NLRs are a diverse family of pattern recognition receptors in the cytoplasm.
Numerous research articles have attempted to elucidate the mechanism of NLR proteins in inflammatory responses and diseases. The field of innate immunity, with focus on NBD-LRRs is rapidly expanding. We will attempt to provide an overview of select aspects concerning NBD-LRRs such as nomenclature/structure, bacterial sensing, and mode of activation/signaling as well as their roles in inflammatory diseases and usage in therapeutic interventions.
The NLR family has been referred to by different research groups as the NOD, NALP or CATERPILLER protein family [3, 13, 14]. Recently, the HUGO Gene Nomenclature Committee approved the usage of the common nomenclature for this family as the NLR or Nuceleotide-binding domain (NBD), leucine rich repeat (LRR) containing family. Other abbreviation for NLRs, such as NACHT-LRR and NOD-LRR are still in use.
The NLR proteins are characterized by a tripartite domain structure  (Fig. 1). They have a centrally located NACHT (domain present in neuronal apoptosis inhibitor protein (NAIP), the major histocompatibility complex (MHC) class II transactivator (CIITA), HET-E and TP1) domain [14, 16], C-terminal Leucine-rich repeats (LRRs) and an N-terminal effector binding domain, which can consist of either CARD (Caspase recruitment domain), pyrin domain (also known as PAAD, PYD or DAPIN) [17-23] or BIR (baculovirus inhibitor of apoptosis repeat) domains (Fig. 1) .
The NACHT domain (also known as NBD or NOD domain) belongs to a large superfamily of the NTPase domains, which hydrolyze either ATP or GTP [16, 25]. Among NTPase superfamily, the NACHT domain in the NLRs is phylogenetically close to another NTPase domain termed NB-ARC, which exists in plant disease resistant genes and the apoptotic gene, Apaf-1 . Like other NTPase domains, the NACHT domain is proposed to oligomerize upon activation in the presence of ATP. The NACHT domain seems to be the main module mediating activation of the molecule, since mutations in the ATP binding region (Walker’s A box) or the magnesium binding region (Walker B box) abolish the signaling from NLRs [26, 27]. LRRs, on the other hand are very diverse and found in many molecules. They are typically known as a protein-protein interaction domain that can interact with many other molecules. LRRs in NLR proteins are believed to be a ligand recognition domain, which defines the specificity of the NLR protein to a particular ligand. However, it is not clear if LRRs in NLR proteins can interact with microbial products directly or need intermediate signaling molecules.
The effector binding domain at the N-terminus is believed to recruit downstream effector molecules to activated NLR proteins, thus it is critical to transduce signal. The NLR protein family can be classified into subfamilies by effector domains at the N-terminus. There are two major subfamilies: the CARD subfamily (called NLRC, or NACHT, LRR and CARD domains-containing protein) and the Pyrin subfamily (called NLRP, or NACHT, LRR and PYD domains-containing protein) (Fig. 1). The CARD and pyrin effector domains belong to the subfamily of death domains, comprised of four member domains, the CARD, Pyrin domain, DD and DED [17, 19, 20]. Typically they have a six alpha helical structure and signal via homophilic interactions i.e., CARD–CARD and pyrin-pyrin.
There are approximately 20 NLRs in humans and only a few have been matched up with putative ligands/elicitors. A number of putative elicitors of NLRs have been reported but for several there has been no consensus in the research community (Table 2). Therefore, the field of NLR ligand identification remains open to a large number of possibilities. It is unclear whether NLRs are able to recognize ligands directly through their LRRs. In fact, neither the TLRs nor the NLRs have been shown to interact directly with their putative ligands. Therefore, it is possible that the NLR-microbial recognition process is more complicated than the sensing that will be discussed in this section.
The first NLRs reported to function as intracellular microbial recognition molecules by responding to specific PAMP stimuli were Nod1 (CARD4) and Nod2 (CARD15). Both are members of the CARD subfamily of NLRs, where Nod1 contains a single CARD domain and Nod2 contains two at the N-terminus (Fig. 1 and Table 1). These two proteins were shown to recognize moieties of the bacterial cell wall component, peptidoglycan. Peptidoglycan has a structure as a crystal lattice formed from linear chains of two alternating amino sugars, N-acetyl glucosamine (GlcNAc or NAG) and N-acetyl muramic acid (MurNAc or NAM) attached to a short amino acid chain (Fig. 2). Many bacteria are classified by their cell wall structure and thickness of the peptidoglycan layer as either Gram positive, having thick peptidoglycan, or Gram negative, having a thin layer of peptidoglycan. Nod2 has been shown to respond to muramyl dipeptide (MDP), consisting of NAM-L-Ala-D-Glu [28, 29]. This moiety is conserved in both Gram positive and Gram negative organisms, suggesting that Nod2 may confer resistance to a wide variety of bacteria. Indeed, Nod2 has been proposed to sense peptidoglycan from a variety of Gram positive and Gram negative heat killed bacteria and bacterial extracts . However, it is interesting that not all bacterial species containing large amounts of peptidoglycan have strong Nod2 stimulatory activity . This may suggest that there is a specificity of Nod2 towards certain organisms, but the nature of this specificity, whether it is the ligand itself or the mode of ligand presentation, is unclear. Nod1, on the other hand, has been shown to detect GM-tripeptide containing meso-DAP as well as simpler peptide versions containing meso-DAP (Table 2) [15, 31]. The sensing of peptidoglycan moieties with meso-DAP, found predominantly in Gram negative bacteria, suggests that Nod1 restricts the growth of certain types of bacteria. Studies with heat killed bacteria and their extracts also indicated that there is vast variation in Nod1 stimulatory activity from organisms that contain the minimal iE-DAP structure . Interestingly, in those studies, Bacillus species had the greatest Nod1 stimulatory activity and a strong Nod2 stimulatory activity. Both Nod1 and Nod2 have been implicated in restricting growth of a number of specific bacteria, such as Listeria monocytogenes and Shigella flexneri [30, 32-36]. Although Hasegawa et al. show that live Listeria infection and infection with the listeriolysin O mutant, which does not allow Listeria to escape into the cytosol, are capable of activating NF-κB in a Nod1 dependent manner, the same study shows that heat killed bacterial cells and cell supernatants from overnight cultures failed to induce Nod1 dependent NF-κB activation . This suggests that the putative extracellular Nod1 stimulatory molecules are linked to Listeria viability. It is then possible that Nod1 may have additional stimulatory molecules that may not be peptidoglycan moieties or that bacterial and not host enzymes are required to generate Nod1 ligands. Overall, whether the nature of the specific growth restrictions is a direct result of sensing those specific bacteria or a secondary result of regulation of antimicrobial peptide production due to broad-range sensing, remains to be determined.
A re-occurring question is the mode of MDP delivery into the host cell for recognition by Nod2. Currently, there are two hypotheses: the first suggests that MDP is delivered into the cytosol by an MDP specific transporter on the plasma membrane, PepT1, although the role of PepT1 in mediating MDP recognition by Nod2 is unclear [37, 38]. Human PepT1 (hPepT1) functions as a transporter of small peptides as part of the normal absorption function of the gut epithelial cells. It has been shown to be expressed in the small intestine but not the colon . However, chronically inflamed colonic epithelial cells do express PepT1 . In addition, hPepT1 is also expressed in normal human macrophages and human monocytic KG-1 cells, where it has been shown to mediate transport of di- and tripeptides . Therefore, PepT1 remains as a possibility in terms of MDP transport. The second possibility is that during endo/phagocytosis of the bacteria, the MDP moiety or a related muropeptide is generated in the phagolysosome by lysosomal enzymes . It may subsequently be transported out where it either diffuses into the cytosol or remains associated with the phagolysosomal membrane . It is possible that in vivo, both models may play roles in delivering MDP into the cytosol for recognition by Nod2, since both the host and the bacterium have enzymes that degrade peptidoglycan. Hosts and bacteria have muramidases, or lytic transglycosylases in bacteria, that are able to degrade peptidoglycan to GM-tripeptide, containing meso-DAP, and GM-dipeptide without meso-DAP (Fig. 2) [31, 43, 44]. Nod2 is able to sense GM-dipeptide as well as MDP, however, to date, only bacteria seem to have the enzymes necessary to generate MDP [31, 45]. One example is the endopeptidase secreted by Listeria monocytogenes, which cleaves the bond between D-Glu and meso-DAP (Fig. 2) . Nevertheless, there is evidence that the phagolysosome is required for Nod2 dependent signaling [42, 46]. These two hypotheses are still being actively investigated.
Another set of proteins of the NLR family that have been shown to recognize microbial structures are Ipaf and Naip. Ipaf belongs to the CARD subfamily, whereas Naip has three BIR domains at the N-terminus. Both of these have been shown to respond to flagellin, the main component of the bacterial flagellum [47-51]. Experimental evidence using flagellin mutants in bacteria such as Salmonella and Legionella have indicated that those strains were unable to signal through Ipaf or Naip5/Birc1e. This showed that both proteins detect flagellin from Salmonella and Legionella, suggesting redundancy between Ipaf and Naip5. The recognition of flagellin by these two NLRs raises some interesting questions: what is the region of similarity in Salmonella and Legionella flagellin that allows for redundant recognition by Ipaf and Naip5 and do Ipaf and Naip5 specifically recognize certain types of flagellin? In a recent study, researchers complemented the Legionella pneumophila flagellin (FlaA) mutant with flagellin from Salmonella (FliC) suggesting that Naip5 is able to recognize both forms of flagellin . Interestingly, they were not able to complement the FlaA mutant with FliC from Shigella flexneri or E. coli. This may suggest that Naip5 only recognizes flagellins from specific bacteria. In addition, another study involving Naip5 found that purified flagellin from Salmonella typhimurium, Legionella pneumophila and Bacillus subtilis was able to presumably activate Naip5, although the data does not specifically suggest that recognition is mediated by Naip5 or Ipaf . Further studies to determine the minimum units of flagellin being recognized as well as the differences between different flagellin peptides that may influence specificity would be of importance in further understanding the function of Naip5. No such differences have yet been identified with Ipaf. On an interesting side note, both Salmonella and Legionella are intracellular pathogens that reside in the phagosomal compartment by suppressing defense responses and phago-lysosomal fusion. This suppression of host defenses and changes in vesicular traffic are mediated by specialized secretion systems that deliver virulence factors into the host to promote disease. The type III and type IV secretion systems are examples of membrane pore-forming and virulence factor delivery mechanisms [52, 53]. Since Salmonella and Legionella are not able to escape the phagosome into the cytosol, studies have suggested that flagellin delivery into the cytosol for recognition by Ipaf and Naip, is mediated by the type III and type IV secretion system generated pores [48, 49].
Perhaps the most controversial NLR in terms of which ligands/elicitors are involved in its activation is Nalp3/Cryopyrin of the Pyrin subfamily of NLRs, containing an N-terminal pyrin domain. Unlike the NLRs discussed above, Nalp3 has been shown to respond to several elicitors such as bacterial RNA, uric acid crystals, ATP and pore forming toxins (Table 2) [54-56]. Studies showing that Nalp3 may sense bacterial RNA and uric acid crystals independently or partially independently of TLRs suggest that there may be two distinct pathways leading to the release of IL-1β [54, 56]. The active cytokine is generated by cleavage of its pro-IL-1β form by Caspase-1. Caspase-1 is produced as an inactive zymogen which also needs to be cleaved into an active form. NLR family proteins such as Ipaf and Nalp3 have been implicated in regulating Caspase-1 activation. However, additional studies have shown that IL-1β release from macrophages requires two stimuli, one from TLR signaling, where pro-IL-1β is generated and the other from a stimulus such as ATP that presumably induces oligomerization and inflammasome assembly (Fig. 4B) [55, 57, 58]. Studies with Nalp3 deficient mouse macrophages showed that they are unable to produce IL-1β when stimulated with exogenous ATP that is sensed through the P2X7 receptor in complex with Pannexin-1, the latter being a hemichannel protein that is required for Caspase-1 activation and IL-1β secretion [55, 57, 59, 60]. Recent studies showed that heat-killed bacteria and TLR ligands such as LPS (ligand for TLR4) can induce Caspase-1 activation in the presence of ATP in macrophages deficient in TLR4, MyD88 or TRIF (downstream adaptors for TLRs), suggesting that bacterial products may activate Caspase-1 together with ATP by TLR independent manner . Additionally, studies suggest that Nalp3 does not directly sense ATP but rather intracellular potassium depletion resulting from ATP signaling [55, 60]. Pathogen toxins which insert themselves into host membranes, have been proposed to alter intracellular potassium in a Nalp3 dependent manner . Such toxins include: nigericin, a potassium ionophore; maitotoxin, primarily a Ca2+ channel, but also a transporter of other cations; and listeriolysin O. Therefore, it is likely that exogenous ATP may serve as a “danger signal” leading to potassium depletion. This suggests that some NLRs may also sense non-microbial signals and molecules involved in cell defense.
Nalp1 cannot neatly fit into a single subfamily of NLRs as it contains both an N-terminal pyrin domain and a C-terminal CARD domain, suggesting that Nalp1 may have multiple signaling roles. The elicitors for Nalp1 have been suggested to be MDP and the anthrax lethal factor (LF) of the Bacillus anthracis toxin (LeTx) [61-63]. Studies using MDP included considerable biochemical and kinetic analysis between MDP and Nalp1 . In vitro reconstitution experiments with those molecules showed that Nalp1 was activated and signaled efficiently for IL-1β production . The activation of purified Nalp1 with pure MDP in controlled reconstitution experiments raises the possibility that Nalp1 may not need the aid of another protein to recognize MDP, suggesting that interaction may be direct, although further studies are necessary to confirm this. One potential problem of the Nalp1 reconstitution experiments is that they were performed in vitro using purified protein. An in vivo study with a Nalp1 knockout is necessary to confirm MDP as a true ligand. As mentioned previously, Nalp1 was also shown to respond to anthrax lethal factor . LF is a Zn2+ dependent endoprotease that cleaves the N-terminus of MAPKKs, altering signaling pathways and leading to apoptosis. It is not known precisely how Nalp1 responds to LF, but this could be the first NLR shown to respond directly to a disease causing virulence factor secreted by a bacterium. In plant systems, the Resistance genes described above recognize specific microbial virulence factors delivered into the host, but this recognition has been shown in some cases to be indirect, where the R genes actually recognize the virulence protein’s enzymatic activities on another host protein [9, 10]. It is possible that the recognition of LF by Nalp1 may be through a similar mechanism.
One bacterium for which there is data suggesting that it may be acting via intracellular sensors such as NLRs is a Gram-negative coccobacillus, Francisella tularensis. F. tularensis is a facultative intracellular pathogen that causes the disease tularemia (also known as “rabbit fever”). Francisella has been shown to induce the activation of Caspase-1 in an ASC dependent manner in a multi-protein complex known as the inflammasome, discussed below . This indicates that perhaps one of NLR proteins recognize Francisella tularensis, leading to the activation of Caspase-1 and IL-1β secretion.
As a sensor of microorganisms, NLR proteins are programmed to activate host defense mechanisms. Two major signaling pathways have been described. One is NF-κB and MAPK activation through a serine/threonine kinase Rip2 (Rick/Cardiak/Ripk2) (Fig. 4A) [65-67]. A typical example is signaling through Nod1 or Nod2, which detect active moieties of bacterial peptidoglycan. The other is activation of Caspase-1 through the activation of several NLRs, including Nalp1, Nalp3, Ipaf and Naip. This pathway leads to IL-1β secretion and programmed cell death (Fig.4A).
It has been proposed that oligomerization of NLR proteins and activation of recruited effector molecules by their proximity to each other play a major role in the activation of NLR signaling . This model is based on the structural similarity of the NLR proteins to an apoptosis inducing protein, Apaf-1, which is able to bind Cytochrome-c and folds into an oligomer with a 7-fold symmetry [69, 70]. During apoptosis via the mitochondrial pathway, Cytochrome-c is released from the mitochondria and binds to the WD40 repeats of Apaf-1 (Fig. 4B). This triggers activation of Apaf-1 resulting in the formation of a multimeric protein complex, an apoptosome, which consists of Apaf-1 and a recruited downstream effector molecule, Caspase-9 (Fig. 4, B and C) [71, 72]. After recruitment, Caspase-9 zymogens autoactivate by cross activation due to proximity to each other [71-74].
The following is a proposed model of NLR activation, although much experimental data is needed to test the model. Inactive NLR proteins may rest in an autoinhibited conformation through intramolecular inhibition of the NACHT domain by LRRs (Fig. 4B) [75, 76]. Analogous to Apaf-1 activation, ligand recognition first may cause a conformational change of NLR proteins (Fig. 4B). Nucleotide triphosphate binding to the p-loop of the NACHT (NBD) domain further changes its conformation, which leads to oligomerization of the molecules. Oligomerization of NLRs subsequently recruits downstream effector molecules (Fig. 4, B and C). In the case of Nod1 and Nod2, they recruit the downstream kinase, Rip2 which may cause autophosphorylation (Fig. 4, A and B) . Rip2 activates the downstream signaling cascades including MAP kinases and NF-κB (Fig. 4, A and B). In the case of Nalp3, it recruits pro-Caspase-1 via the adaptor ASC (Fig. 4, A and B) . The pro-caspase-1 proteins cause autocleavage and activate Caspase-1 itself (Fig. 4B) . This model was supported by a recent report of observation of the Nalp1 oligomeric complex described below . It is still unknown how many NLR molecules are involved in the final oligomeric protein complex.
Nod1 and Nod2 can detect active moieties of peptidoglycan [28, 29, 31, 79]. It has been shown that a kinase Rip2 is required for the downstream signaling of Nod1 and Nod2 . Nod1 and Nod2 associate with Rip2 via CARD-CARD homophilic interactions and the deletion of CARD in Nod1 or Nod2 abolishes the activation of NF-κB [11, 12, 75]. Therefore the N-terminal CARD is a signaling effector domain of Nod1 and Nod2. Once activated, Rip2 turns on downstream signaling cascades, including NF-κB through the IKK (inhibitor of NF-κB kinase) complex and MAP kinase cascades, resulting in the production of pro-inflammatory cytokines/chemokines, such as IL-6, TNF-α, IL-12 or IL-8 [35, 80, 81]. Several proteins are proposed to regulate the activation of Nod2 signaling, which includes a nuclear protein GRIM19, a PDZ domain and LRR containing protein, Erbin, a kinase Tak1 (Transforming growth factor-beta-activated kinase 1) and a GTPase activating protein Centaurin beta1 [34, 82-85]. Although the mechanisms of these proteins in Nod2 signaling have not been elucidated, it has been shown that Erbin associates with the CARD domains of Nod2 and inhibits NF-κB activation upon MDP stimulation of Nod2 transfected cell lines. Therefore, a negative regulatory role of Erbin in Nod2 activation has been proposed [34, 83]. Although it was shown by in vitro studies that the Rip2 pathway activates Caspase-1 [67, 86], the studies using mutant mice deficient for Nod2 or Rip2 failed to show involvement in IL-1β secretion [35, 80, 87].
Some members of the NLR family, including Nalp1, Nalp3, Ipaf and Naip have been shown to activate Caspase-1 via an adaptor, ASC. ASC (TMS1/Pycard/CARD5) (Fig. 1) is a dual adaptor molecule which has a pyrin and a CARD domain [19, 88, 89]. Since Caspase-1 has a CARD at its N-terminus, ASC has been proposed as a bridging molecule for the association between some NLRs and Caspase-1 [77, 90]. The Pyrin subfamily of NLR proteins, such as Nalp3 have been shown to interact with Caspase-1 via ASC . In the case of Ipaf, since it has a CARD, structurally it may not require ASC for the association with Caspase-1. However, the phenotype of ASC deficient mice is similar to Ipaf deficient mice upon bacterial infection . Flagellin-induced Caspase-1 activation was abolished in both Ipaf and ASC deficient cells , therefore, ASC may be involved in Ipaf signaling to activate Caspase-1. Indeed ASC associates with Ipaf via CARD-CARD interactions in in vitro studies . Therefore, the exact protein structure of the Ipaf protein complex involving ASC remains an intriguing question.
Another NLR, Naip (also called Birc1) has also been shown to activate Caspase-1 [49, 51]. Naip does not have either a CARD or a pyrin domain at its N-terminus, but has a protein interaction domain called BIR [3, 24]. It has been shown that one of the Naip isoforms in mice, Naip5 (Birc1e) is critical for the resistance against Legionella pneumophilla infection [92, 93]. The downstream Caspase-1 activation by Naip5 is somewhat controversial. The macrophage cell death induced by Legionella infection has been shown dramatically reduced in a B6 congenic mouse deficient for Naip5 . In direct measurements using a fluorescent dye with high binding affinity to Caspase-1, FAM-YVAD, another study showed that the congenic mouse macrophages deficient for Naip5, B6.A-Chr13, was significantly reduced in Caspase-1 activation . However, independent studies recently showed using anti-Caspase 1 antibodies that in the same Naip5 deficient mouse background, Legionella was able to induce cleavage to mature Caspase-1 . Therefore, the activation of Caspase-1 downstream of Naip5 remains to be clarified. It is an interesting question as to how BIR containing Naip activates Caspase-1. Interestingly, ASC deficient macrophages restrict Legionella growth normally whereas AJ macrophages are permissive to Legionella growth , suggesting that Naip5 may have multiple downstream effectors.
An activated oligomerized complex is frequently called the “inflammasome”. This term has been developed analogous to the apoptosome that is generated by Apaf-1 during apoptosis mediated via the mitochondrial pathway . Tschopp and colleagues proposed that an inflammasome composed of three NLRs (NLR1, NLR2 and NLR3) and two adaptors (ASC and Cardinal, another CARD containing protein), which may be generated by stimulation of LPS or MDP [77, 95, 96]. However, the exact composition and stoichiometry of the inflammasome are still elusive. Using purified recombinant proteins, Reed and colleagues have reconstituted the Nalp1 inflammasome, which was induced by purified recombinant Nalp1 upon MDP stimulation . Oligomerization was induced by a two-step mechanism, initial ligand stimulation and subsequent ATP binding, as observed in the Apaf-1 activation by Cytochrome-c and ATP .
Caspase-1 activation defends the host by two distinct mechanisms. One is secretion of IL-1β and the other is the induction of programmed cell death. IL-1β is synthesized as pro-IL-1β and it requires cleavage by an active Caspase-1 to generate the mature form of IL-1β. IL-1β can act as a powerful pro-inflammatory cytokine to help non-infected cells and the host to fight against pathogenic organisms. The other two IL-1 family cytokines, IL-18, a potent inducer of IFN-γ, and IL-33, involved in Th2 responses, are also generated by active Caspase-1 [97, 98]. Caspase-1 can also induce programmed cell death of infected cells, which can serve as a host defense mechanism by eliminating infected cells, thus killing parasitic microorganisms.
Although called Nalp, Nalp10 (Pynod) has only a pyrin and a NACHT domain but lacks LRR, thus it is not a typical NLR protein. Since it does not have LRRs, it is unlikely that Nalp10 detects microbial products. Suda and colleagues proposed that Nalp10 negatively regulates Caspase-1 activation by binding to ASC, thus Nalp10 may be a regulator of inflammation . Pyrin is an adaptor protein containing a pyrin domain, two B-boxes, with a coiled-coil domain (BBCC) and a SPRY domain (also called as B30.2 or RFP domain). Pyrin associates with ASC via pyrin-pyrin domain homophilic interactions  This interaction seems to negatively regulate the signaling pathway of ASC and the activation of Caspase-1 [101, 102]. It has also been reported recently that the SPRY domain of Pyrin also associates with Nalp3 and pro-Caspase-1, suggesting that Pyrin may inhibit Caspase-1 activation in an ASC independent manner. [102, 103]. It is noteworthy to mention that both mutations in the Pyrin gene are inherited in a recessive manner. Mutations in Nalp3 with a gain-of-function phenotype result in hereditary inflammatory diseases characterized with high IL-1β production [104-106].
Crohn’s disease is one of two major forms of inflammatory bowel diseases characterized by chronic inflammation. The clinical features include abdominal pain, diarrhea, fever and complications such as anemia, toxic megacolon, stenosis and fistulae. Familial clustering and studies of monozygotic twins suggested a genetic contribution to the development of Crohn’s disease although diverse environmental factors are also likely to play a role in the development of this disease. Linkage analysis in affected families has revealed several genetic loci termed IBD1 through IBD9 that show a significant association with Crohn’s disease [107, 108]. Mapping of the IBD1 locus had lead to the identification of NOD2 as the first gene to be strongly associated with Crohn’s Disease susceptibility in North American and European populations [109, 110]. The Card15 gene is located on the human chromosome 16q12 that encodes NOD2. There are three main NOD2 sequence variants that are associated with Crohn’s Disease susceptibility. All three variants alter the C-terminal portion of NOD2 that are within or close to the region of the LRRs. The frameshift mutation in the LRRs of NOD2 (L1007fsinsC), which results in a partial truncation of the LRRs and other single nucleotide polymorphisms (SNPs) within the LRRs (R702W and G908R) are associated with the development of Crohn’s disease (Fig. 5A and B and Table 3) .
Three models have been proposed for the mechanism by which NOD2 mutations cause Crohn’s disease. The major issue is to understand the mechanism of pathogenesis by which human NOD2 mutations result in either a “loss-of-function” or a “gain-of-function”. The first proposed model is diminished defensin expression. Crohn’s disease-associated mutations within the LRRs of NOD2 have been reported to abolish the ability to sense MDP and activate NF-κB . Crohn’s disease-associated NOD2 mutations predispose to ileal involvement [111, 113-115] that corresponds to the location of Paneth cells. Crohn’s disease patients have been shown to have decreased expression of human Paneth cell α-defensins, HD-5 and HD-6, in the small intestine with ileal involment [116-118]. We have demonstrated that mice deficient in Nod2 were defective in innate mucosal immune defense against oral infection with L. monocytogenes and had diminished expression of Paneth cell-derived antimicrobial peptides, Defcr4 and Defcr-rs10 . The pathogenesis of Crohn’s disease seems to involve inappropriate immune responses and impaired epithelial barrier function. Disease associated mutations lead to diminished NF-κB activation upon MDP stimulation. Impairment of Nod2 function may facilitate the entry of bacteria into epithelial cells because of defective regulation of defensin expression that results in impaired bactericidal capacity (“loss of function”) (Fig. 5B).
The second proposed model for the development of Crohn’s disease involves altered TLR2 signaling by NOD2 mutations. Splenic macrophages from mice deficient in Nod2 produced increased levels of IL-12 upon TLR2 stimulation, suggesting a negative regulatory role of NOD2 in a TLR2 agonist-mediated production of IL-12. IL-12 promotes IFN-γ production by T and NK cells and promotes growth and differentiation towards T helper type 1 (Th1) effector cells, which have been proposed to be important in the pathogenesis of Crohn’s disease . Therefore, a negative regulatory role of Nod2 in TLR2-mediated Th1 responses has been proposed [120, 121]. However, a major concern about this proposed model for Crohn’s disease development is the reproducibility of these observations as our group observed a synergistic effect, not the reverse, of lipopeptide Pam3CS(K)4 (TLR2 ligand) and MDP stimulation for the production of pro-inflammatory cytokines, IL-6 and IL-12p40 . Other groups have also shown a synergistic effect of MDP with synthetic ligands for TLRs 2 and 4 for the production of IL-8 by epithelial cells . Furthermore, MDP and Pam3CS(K)4 stimulation of blood mononuclear cells from Crohn’s disease patients resulted in a synergistic production of TNF- by cells isolated from patients that harbor either wildtype and heterozygous mutation, but not those that were homozygous for the mutation L1007fsinsC of Nod2 . These results indicate that the negative regulatory role of NOD2 on TLR2 agonist mediated responses is not a universal phenomenon and could be dependent on the specific genetic background or different cell types.
The third model involves a “gain-of-function” mutation of NOD2 that results in increased sensitivity to MDP and increased inflammatory responses. A mouse model has been developed that harbors a similar mutation, Nod22939iC, to the human Crohn’s disease associated NOD23020insC frameshift mutation. Macrophages isolated from these mutant mice were hyper-responsive to MDP which resulted in increased NF-κB and IL-1β secretion when compared to wildtype mice. Also, when these mice were treated with dextran sodium sulfate, mutant mice displayed increased IL-1β and bacterial dependent inflammation when compared to wildtype treated mice suggesting a “gain-of-function” phenotype . This “gain-of-function” mutation is at odds with the “loss-of-function” mutation of Nod2 in humans [28, 29, 123, 125, 126]. Peripheral blood mononuclear cells from individuals that are homozygous for the L1007fsinsC NOD2 mutation but do not have Crohn’s disease are defective in responding to MDP stimulation [29, 81]. Also, dendritic cells from Crohn’s disease patients homozygous for the L1007fsinsC NOD2 mutation fail to produce cytokines and up-regulate co-stimulatory molecules CD80 and CD86 upon MDP stimulation . Finally, Blau syndrome (discussed in the following section) is a systemic inflammatory disease caused by a “gain-of-function” mutation of NOD2 but lacks intestinal inflammation [128-132]. These indicate that the molecular nature of the Nod22939iC mice may differ from that of human NOD2 mutations associated with Crohn’s disease.
Blau syndrome (BS) is an autosomal dominant disorder characterized by early-onset granulomatous inflammation (arthritis, uveitis, skin rash with camptodactyly) . CARD15 mutations were found in 4 French and German families with BS and found that two families shared a missense mutation (R334Q); other families had missense mutations (L469F and R334W) (Table 3) . BS mutations are in close vicinity to the Mg2+-binding sites of the NACHT domain and cause basal activation of NF-κB (“gain of function”) (Fig. 5C). EOS shares with BS distinct inflammatory conditions that involve the skin, joints and eyes. The majority of analyzed cases of EOS had heterozygous mutations in the NOD2 gene. EOS shares with BS a common genetic etiology of CARD15 mutations that cause constitutive NF-κB activation [133, 134].
The CIAS1 (also known as NALP3 and PYPAF1) gene is located on human chromosome 1q44 that encodes Cryopyrin. Mutations within the CIAS gene results in three autosomal dominant autoinflammatory diseases, Muckle-Wells syndrome (MWS), Familial cold urticaria (FCU) and NOMID (neonatal-onset multi-system inflammatory disease, also referred to as CINCA (chronic infantile neurologic cuteaneous) syndrome (Table 3) . These diseases have similar clinical manifestations and are collectively called as a cryopryrin-associated periodic syndrome (CAPS). Patients with CIAS-1 associated diseases are classified with having autoinflammatory disorders that are characterized by recurrent episodes of systemic inflammation . Symptoms of patients with FCU include fever, arthralgia, urticaria-like rash and conjunctivitis  whereas patients with MWS presents symptoms of fever, limb pain, urticaria-like rash and some patients experience progressive neurosensory hearing loss and systemic amyloidosis . Patients with FCU have recurrent episodes following generalized cold exposures, while MWS patients often have attacks not triggered by cold exposures . Patients with NOMID suffer from meningitis, seizures, developmental delay, visual and hearing impairment and some have deforming overgrowth of the distal femur . Cryopyrin is expressed primarily in neutrophils, monocytes and chondrocytes . Although the function of Cryopyrin is not clear, Cryopyrin may have roles in cell signaling by regulating cytokine responses at the post-translational level through Caspase-1 and IL-1β processing. Mutations in the NACHT domain of Cryopyrin result in the augmented Caspase-1 signaling (gain-of-function) and lead to overproduction of IL-1β (Fig. 5C) [137, 140]. This could explain many of the systemic and tissue inflammatory symptoms that are seen in patients.
Common generalized vitiligo is a chronic skin disorder characterized by the disappearance of pigment cells (melanocytes) from the epidermis that cause well defined irregular white patches. Vitiligo affects approximately 1% of the world population. The actual cause is not known, however, genetic, autoimmune and environmental factors have been considered. Vitiligo patients have an increased frequency of other autoimmune disorders, particularly autoimmune thyroid disease, latent autoimmune diabetes in adult, rheumatoid arthritis, psoriasis, pernicious anemia, systemic lupus erythematosus and Addison’s disease [141, 142]. Jin et. al. recently reported that mutations in the NALP1 gene, which is located on chromosome 17p13, is associated with generalized vitiligo  . Particularly a point mutation of L155H, which is located between the N-terminal pyrin and NACHT domain, has a strong association both with vitiligo alone and with other autoinflammatory diseases .
The MHC2TA gene is found on the human chromosome 16p13 that encodes the master transcription factor of the MHC class II genes, known as CIITA [13, 145]. Defects in MHC2TA results in the autosomal recessive hereditary immunodeficiency, bare lymphocyte syndrome (BLS) . A majority of BLS affected families are from northern Africa, although it is also found in families from Spain and Turkey . BLS patients are extremely vulnerable to bacterial, viral, protozoan and fungal infections because of defects in cellular and humoral immunity . BLS is a collection of MHC class II deficiencies that are defined by four genetic complementation groups. Changes in the MHC2TA gene represent a defect in complementation group A. The structure of CIITA consists of an N-terminus transcriptional activation domain, a central NACHT domain and LRRs. There are four isoforms of CIITA. The type II isoform is expressed in low levels in humans and is not expressed in mice. The three other isoform have cell-specific expression. Specifically, dendritic cells constitutively express all three isoforms including type I CIITA which has an N-terminal CARD . CIITA regulates the transcription of all MHC class II genes. It has been shown that lack of CIITA expression in both humans and mice results in a severe reduction of MHC class II expression and have reduced numbers of peripheral CD4 T cells . CIITA is not a DNA binding protein but interacts with the basal transcription factors, as well as histone-modifying acetylases and methylases. The current model is that CIITA coordinates the assembly of acetylases and methylases with other basal transcription factors on MHC class II promoters [150, 151]. Therefore, the lack of CIITA results in a “bare” promoter that lacks associated DNA binding proteins.
PYRIN is encoded by MEFV (Mediterranean fever) gene which is on the locus of human chromosome 16p13.3 and when it is mutated results in familial Mediterranean fever (FMF). FMF is a recessively inherited systemic autoinflammatory disease that is common among certain ethnic groups such as Jews, Turks, Arabs and Armenians . Attacks of FMF consist of 1-3 day episodes of fever with severe abdominal or chest pain, monoarticular arthritis or an erysipeloid rash. Histologically, there is a massive influx of polymorphonuclear leukocytes into affected regions, neutrophilia and a rapid actute phase response, but autoantibodies and antigen-specific T cells are generally not found . PYRIN is expressed at high levels in myeloid/monocytic cells [104, 105, 152]. Full-length PYRIN has been shown to co-localize with microtubules and the actin cytoskeleton ; however, PYRIN has also been found in the nucleus . Mutations in the MEFV gene have been identified and include four-different disease-associated conservative missense mutations in the C-terminal protein interacting SPRY domain (also known as B30.2 or RFP) (Table 3). These mutations and one additional mutation, E148Q located in exon2, accounts for the majority of FMF mutations . The N-terminal pyrin domain binds the adaptor protein ASC and through this interaction PYRIN has been shown to regulate IL-1β processing, NF-κB activation and apoptosis [77, 90, 103]. Recent studies have shown that the SPRY domain of PYRIN can interact not only with Nalp3 but also with Caspase-1 and pro-IL-1β and inhibit Caspase-1 activation and IL-1β secretion . Therefore, mutation in PYRIN probably leads to uncontrolled production of IL-1β and NF-κB activation resulting in excessive inflammation.
Since the discovery of MDP as the Nod2 ligand, Crohn’s disease has been thought to result from an inappropriate host response to normal intestinal bacterial flora [119, 156]. Studying MDP induced signaling leading to inflammatory responses and the mechanisms of its recognition has become an important aspect in understanding the disease. On the other hand, MDP has been used in immunotherapy of cancer as an immunoadjuvant for years [157-160]. In clinical studies, peptidoglycan moieties and MDP derivatives have been shown to increase the efficacy of cancer therapeutic agents resulting in increased overall survival [161-164]. Specifically, in Meyers et al. they showed that in combination with the drug ifosfamide, muramyl tripeptide, when delivered into cells by liposomes is capable of increasing event-free survival in osteosarcoma patients .
In addition, MDP has been tested in liver metastasis, human breast carcinoma and upon injection of human colon carcinomas [162, 165, 166]. In all cases, MDP contributed to inhibition of metastatic liver growth upon Kupffer cell activation with MDP as well as inhibition of metastatic growth upon tumor inoculation of melanomas and carcinomas. Therefore, in order to better understand how we can improve the use of MDP in cancer therapy, understanding the mechanisms of MDP signaling is crucial.
In addition to Nod1 and Nod2 peptidoglycan moiety ligands, studies with flagellin have suggested that it may be used as an immunoadjuvant in treating bacterial infections [167-169]. However, the mechanism of this flagellin signaling, whether through TLR5 or Ipaf and Naip has not been fully clarified as studies have generally focused on TLR5.
Recent rapid progress of the research in innate immunity has succeeded in illustrating the significance of NLRs in host-pathogen interactions and immunological disorders. Several strains of mutant mice deficient for NLRs, including Nod1, Nod2, Ipaf and Nalp3, have been used to demonstrate that these molecules are critical to control pathogenic microorganisms. Several inflammatory diseases, such as Crohn’s disease, Blau syndrome and Muckel-Wells syndrome, are caused by mutations in NLR proteins. However, many questions still remain unanswered in this family of proteins, such as, why NLR genes have evolved. Mammals have cell surface receptors, TLRs, which are highly sensitive innate immune receptors. Therefore, it is unclear why the host requires a second set of microbial sensors that are located in the cytoplasm. In addition, the Rip2 signaling pathway downstream of Nod1 and Nod2 seems redundant to the TLR response because both pathways result in similar outcomes such as secretion of IL-6, IL-12, TNF-α via NF-κB and MAPK. This redundancy cannot explain the requirement for additional cytoplasmic innate immune effectors. One explanation may be that the host might require different detection mechanisms depending on the route of infection or route of delivery of microbial products. TLRs may be important for sensing bacteria attached to the cell surface and Nod1/Nod2 may play a role in sensing bacteria in the phagosome, as discussed earlier. For the activation of Caspase-1, it is unclear why NLR but not TLR activation results in the strong activation of Caspase-1 and subsequent IL-1β production. Although, several NLRs such as Nalp3, Ipaf, Nalp1 and Naip can activate Caspase-1, it remains to be elucidated why microbial sensing in the cytoplasm but not on the cell surface has been chosen to activate Caspase-1.
Second, although NLR genes in animals have a similar domain structure to NBS-LRR disease resistant genes (R genes) in plants, there is no evidence these two structurally related gene families have originated from a common ancestor. Indeed searching databases in Drosophila and C. elegans failed to identify any NLR homologues . Therefore, NLR and plant NBS-LRRs might have evolved in a convergent rather than divergent manner. Third, it is still unclear how cytoplasmic NLR proteins can detect microorganisms which are not internalized into the host cell. Several possibilities have been discussed in former sections. It is possible that different NLRs utilize different mechanisms to detect microbial products in the cytoplasm.
Fourth, outcomes of NLR activation are still under investigation. Adjuvant activity of Nod1 and Nod2 ligands was investigated three decades ago and data was confirmed recently by using Nod1 and Nod2 deficient mice [35, 171-173]. However, it is poorly known how activation of these molecules activates adaptive immune responses. Although, Nod2 has been shown to be critical for the expression of alpha defensins in both mice and humans [35, 117, 174], it is unknown how Nod2 is able to regulate the transcription of alpha defensin genes in intestinal Paneth cells. Finally, we do not know most of the NLR ligands. Bacterial peptidoglycan (ligand for Nod1 and Nod2), Flagellin (for Ipaf and Naip) and anthrax toxin (for Nalp1) have been identified to activate NLR proteins. Considering the diversity of this family, it is likely that more ligands from microbes will be uncovered.
The authors thank Y. S. Koh and T. B. Meissner for helpful discussions. This work was supported by grants from the NIH (K.S.K. DK074738-01A1) and the Crohn’s and Colitis Foundation of America (K.S.K.). K.S.K is a recipient of the Investigator Award from the Cancer Research Institute and the Claudia Adams Barr Award, and T.P.O. is a recipient of the Benacerraf Memorial Fellowship.