Mutational analysis of human NOD1 and NOD2 NACHT domains reveals different modes of activation

Abstract

Nucleotide-binding oligomerization domain-containing protein (NOD)1 and NOD2 are intracellular pattern recognition receptors (PRRs) of the nucleotide-binding domain and leucine-rich repeat containing (NLR) gene family involved in innate immune responses. Their centrally located NACHT domain displays ATPase activity and is necessary for activation and oligomerization leading to inflammatory signaling responses. Mutations affecting key residues of the ATPase domain of NOD2 are linked to severe auto-inflammatory diseases, such as Blau syndrome and early-onset sarcoidosis. By mutational dissection of the ATPase domain function, we show that the NLR-specific extended Walker B box (DGhDE) can functionally replace the canonical Walker B sequence (DDhWD) found in other ATPases. A requirement for an intact Walker A box and the magnesium-co-ordinating aspartate of the classical Walker B box suggest that an initial ATP hydrolysis step is necessary for activation of both NOD1 and NOD2. In contrast, a Blau-syndrome associated mutation located in the extended Walker B box of NOD2 that results in higher autoactivation and ligand-induced signaling does not affect NOD1 function. Moreover, mutation of a conserved histidine in the NACHT domain also has contrasting effects on NOD1 and NOD2 mediated NF-κB activation. We conclude that these two NLRs employ different modes of activation and propose distinct models for activation of NOD1 and NOD2.

Keywords

CARD15 STAND NF-κB localization EOS

Introduction

Nucleotide-binding oligomerization domain-containing protein (NOD)1 and NOD2 are intracellular pattern-recognition receptors (PRRs) of the nucleotide-binding domain and leucine-rich repeat containing (NLR) gene family that are involved in host defence against pathogens by recognition of so called microbe associated molecular patterns (MAMPs) (reviewed elsewhere¹^–³). The NLRs have a tripartite domain architecture similar to apoptotic protease activating factor-1 (APAF-1). Signal transduction of activated NLRs relies on the N-terminal effector domain, which adopts a death domain fold, and is either a caspase activation and recruitment domain (CARD) as in the case for NOD1 and NOD2, or a pyrin domain (PYD). However, some NLR members (such as NLRX1 and NLRC5) have no obvious CARD or PYD homolog N-terminal domains. The effector domain is followed by the centrally located NACHT-WH-SH domain mediating activation and oligomerization of the protein.⁴ Finally, the C-terminal part, a series of leucine-rich repeats (LRRs), constitutes the recognition motif for their cognate MAMPs.

NOD1 and NOD2 sense specific fragments of bacterial peptidoglycan (PGN) in the cytoplasm and their minimal elicitor motifs are L-Ala-γ-D-Glu-meso-diaminopimelic acid (triDAP)⁵ and muramyl dipeptide (MDP),⁶ respectively. Recognition of these elicitors induces NACHT-mediated oligomerization and formation of a signaling platform, that recruits the downstream CARD domain containing adaptor receptor-interacting serine/threonine protein kinase 2 (RIPK2; also called RICK or RIP2) through homotypic CARD–CARD interactions.^7,8 Thereby, RIPK2 is activated by induced proximity and subsequently triggers NF-κB and MAPK signaling cascades (reviewed by Kufer⁹). The NOD1 and NOD2 are tightly regulated molecular switches and mutations in their genes leading to a gain-of-function phenotype, have been linked with severe auto-inflammatory diseases.

Like all NLRs, NOD1 and NOD2 are members of the signal transduction ATPases with numerous domains (STAND) clade within the ATPases associated with various cellular activities (AAA+) ATPase superfamily. Experimental evidence obtained from several STAND ATPases, like the bacterial transcription regulator MalT and the human regulator of apoptosis APAF-1, led to the development of a model for the activation of AAA+ ATPases. It suggests cycling of the protein between a monomeric, inactive and a multimeric, active state.¹⁰ Elicitor recognition leads to conformational changes of the ADP-bound molecule, which finally allows nucleotide exchange to ATP. The ATP-bound NLR then oligomerizes to form an active platform that recruits downstream signaling adaptors. According to this model, only one ATP hydrolysis step is needed, namely to switch the signaling platform off and to return the molecule into a closed, ADP-bound, resting state. However, this model is based on experimental data of only a limited number of STAND ATPases characterized to date. Adenosine triphosphate binding and hydrolysis in AAA+ ATPases is mediated by conserved residues within the ATPase domain, most prominent the Walker A (P-loop) and Walker B box.^4,11 The Walker A box is crucial for ATP binding which is mediated by a conserved lysine residue (GKS/T; see Table 1). The Walker B box (DD/E) consists of two acidic residues: the first (aspartate) co-ordinates the Mg²⁺ cation and is required for ATP binding,¹² whereas the second (usually glutamate) primes a water molecule for ATP hydrolysis. Previous work revealed that all NLRs, with the exception of NAIP and NLRP11, contain a modified Walker B box, where the second acidic residue is missing (DG).⁴ Instead, they contain two conserved acidic residues distal to the Walker B box, which we term extended Walker B box (DGhDE), suggesting that NLRs might be activated differently than other AAA+ family members. This raises the question of how exactly nucleotide hydrolysis is carried out in NLR proteins and, in particular, if the extended Walker B box can replace the function of the classical Walker B box. In fact, there is indication of a pivotal function of the extended Walker B box in NOD2 signaling provided by the analysis of mutations in CARD15/NOD2, that are found in patients with Blau syndrome (BS) and early-onset sarcoidosis (EOS), two autosomal dominant genetic disorders characterized by granulomatous inflammation of multiple organs. At least one mutation of NOD2 that correlates with the development of BS and increased basal NOD2-mediated NF-κB activity – E383K – is located in the extended Walker B box.¹³ Additionally, the EOS-associated NOD2 D382E/A612T mutation, also affects a residue in the extended Walker B box,¹⁴ indicating a common molecular mechanism for the development of NLR-mediated auto-inflammatory diseases.

Table 1.

Overview of conserved motifs in the ATPase domain, their function and the analyzed residues in NOD1 (normal) and NOD2 (italic)

Motif	Proposed function	Mutation	Signaling^a	Membrane localization
Walker A	Nucleotide binding	K208A	–	–
		K305R	–	–
Walker B (first acidic residue)	Co-ordination of Mg²⁺	D284A D379A	– –	– –
Extended Walker B (first acidic residue)	Priming H₂O	D287A D382A	++ No autoactivation +	– –
Extended Walker B (second acidic residue)	Priming H₂O	E288A E383A	– +++ High autoactivation	– +
Sensor 1	Senses γ-phosphate of ATP	R333A R426A	– +	– –
GxP	Interacts with adenine of nucleotide	P391A P486A	– –	– –
Sensor 2 conserved His	AAA+: nucleotide binding/hydrolysis^b APAF–1: co-ordination of phosphate groups	H517A H603A	– ++	– +

Ability to activate NF-κB as determined in Figure 2A,B; (–, no activity [<10% of wild-type]; +, residual activity [>10% of wild-type]; ++, less active than wild-type; +++, activity comparable to wild-type).

Interacts with active site in neighbour molecule, where it is supposed to be involved in nucleotide binding.

Here, we characterized the mechanistic details underlying activation of human NOD1 and NOD2. Functional analysis of conserved residues in their NACHT domains revealed that the Walker A and B box, Sensor 1 and GxP motif are crucial for activation of both NOD1 and NOD2. Most interestingly, we elucidated a fundamental difference in NOD1 and NOD2 activation, as several gain-of-function mutations in the NOD2 NACHT domain are non-functional in NOD1.

Materials and methods

Plasmids

All FLAG-tagged NOD1 and NOD2 point mutations were generated by PCR mutagenesis according to the QuikChange site-directed mutagenesis procedure (Stratagene) using the recently described plasmids encoding FLAG-tagged NOD1 and NOD2.^15,16 Photo-activatable GFP-NOD1 (PA-GFP-NOD1) was cloned in pCMV-Tag2B (Invitrogen), where the FLAG-tag was exchanged by PA-GFP. Photo-activatable GFP was constructed by PCR from an eGFP template as described elsewhere.¹⁷ All constructs were subjected to full-length DNA sequencing.

Cell culture

HEK293T and HeLa cells were cultivated at 37°C in a 5% CO₂ atmosphere in Dulbecco’s modified Eagle’s medium (Biochrom AG) supplemented with 10% heat-inactivated fetal bovine serum (Biowest), penicillin and streptomycin (100 IU/ml and 100 µg/ml, respectively; Biochrom AG). Cells were continuously tested for absence of mycoplasma infection by PCR.

Indirect immunofluorescence microscopy

For indirect immunofluorescence microscopy, HeLa cells were seeded in 6-well plates on glass coverslips and transiently transfected with 1 µg of expression plasmids, as indicated using FuGene6 transfection reagent (Roche) according to the manufacturer’s instructions. After 24 h, cells were fixed with 3% paraformaldehyde (Roth) in PBS for 10 min and permeabilized with 0.5% Triton X-100 (Roth) in cold PBS for 5 min. Cells were blocked in 3% bovine serum albumin (BSA; Roth) in PBS for 20 min and incubated successively in primary and secondary antibodies. Primary antibody: mouse anti-FLAG M2 (1 : 20,000; Stratagene). Secondary antibody: goat anti-mouse AlexaFluor 546 (1 : 200; Invitrogen Molecular Probes). The DNA was stained with DAPI (5 µg/ml; Invitrogen Molecular Probes) and actin with Phalloidin-FITC (2.5 µg/ml; Sigma-Aldrich). Cells were mounted in ProLong Gold antifade reagent (Invitrogen Molecular Probes). Image acquisition of z-stacks was performed on an Olympus FV-1000 laser scanning microscope (objective: Olympus PlanApo, ×60/1.40 oil, 8/0.17) and processed using ImageJ software.¹⁸

For the analysis of PA-GFP, cells were co-transfected with PA-GFP-NOD1 and a vector expressing mOrange in order to identify transfected cells. PA-GFP was activated by a 405 nm light pulse and images of one focal plane were acquired every 20 s on an Olympus FV-1000 system equipped with a climate chamber set to 37°C and 5% CO₂.

Expression control of the mutants

HEK293T cells were seeded in 6 cm cell-culture dishes and transiently transfected with 1 µg plasmid as indicated using FuGene6 (Roche). The cells were washed with PBS containing 0.1 mm p-Aminoethylbenzenesulfonyl fluoride and lysed with NP40 buffer (150 mm NaCl, 50 mm Tris, pH 7.4, 1% NP40). Total protein concentration was measured (DC Protein Assay; Bio-Rad). Equal amounts of proteins were separated by Laemmli SDS-PAGE and subsequently transferred on nitrocellulose membrane (Bio-Rad) by semi-dry Western blotting. Proteins were detected by incubation of the membrane successively with primary and secondary antibody and a final incubation with SuperSignal West Pico Chemiluminescent Substrate or Femto Maximum Sensitivity Substrate (Pierce). Signals were recorded on an electronic imaging system (LAS4000, Fujifilm). Primary antibodies were rat anti-NOD1 7B10 (1 : 100¹⁶), mouse anti-FLAG M2 (1 : 2000, Stratagene) and rabbit anti-GAPDH (1 : 1000; Santa Cruz). Secondary antibodies were horseradish peroxidase (HRP)-conjugated goat anti-rat IgG (1 : 4000; Jackson ImmunoResearch Laboratories), HRP-conjugated goat anti-mouse IgG (1 : 4000; 170-6516, Bio-Rad) and HRP-conjugated goat anti-rabbit IgG (1 : 4000; 170-6515, Bio-Rad). All antibodies were diluted in 5% milk in PBS.

Luciferase reporter assay

Thirty thousand HEK293T cells per well were seeded in 96-well plates and transfected with 8.6 ng β-galactosidase, 13 ng luciferase NF-κB-reporter and NOD1 or NOD2 expression plasmids as indicated, added up with pcDNA to 51 ng total DNA using FuGene6 (Roche). Cells were directly stimulated with 500 nm triDAP, 50 nm MDP or 10 ng/ml TNF-α (all obtained from InvivoGen), as indicated. After 16 h, cells were lysed in luciferase lysis buffer (25 mm Tris, pH 8, 8 mm MgCl₂, 1% Triton, 15% glycerol, 1 mm DTT) and luciferase activity was measured using a standard plate luminometer. Standard deviation (SD) was calculated from triplets and luciferase activity was normalized as a ratio to β-galactosidase activity. All experiments were repeated independently at least three times.

Membrane fractionation

Cell fractionation was conducted as described previously.¹⁹ Briefly, HEK293T cells were seeded in 6 cm cell-culture dishes and transiently transfected with 1 µg plasmid as indicated using Lipofectamine2000 (Invitrogen). The cells were lysed in 300 µl Triton-buffer (1% Triton X-100, 100 mm NaCl, 10 mm HEPES, pH 5.6, 2 mm EDTA, 4 mm Na₃VO₄, and 40 mm NaF) supplemented with protease inhibitor cocktail (Complete, Roche Diagnostics). The cytosolic fraction was obtained by centrifugation at 10,000 g for 30 min at 4°C. Subsequently, the pellet was resuspended in 200 µl of lysis buffer containing 1% SDS and was sonicated for 120 s. After 5 min of centrifugation at 10,000 g, the resulting supernatant was collected. Equal amounts of proteins were subjected to SDS-PAGE and Western blot as described above.

Homology modeling NOD1 NACHT domain

The NOD1 protein sequence (1–598; GI:4760399) was submitted to profile sequence searches with the FFAS server (<http://ffas.ljcrf.edu>). The protein structure with highest scoring alignment was used as a template for homology modeling. Templates for NOD1 open and closed formation are as followed, APAF-1 (aa 1–581, pdb entry: 1z6t, seqID: 10%) and ribonuclease inhibitor (aa 21–426, PDB entry: 1bnh/1z7x, seqID 22%). Models were prepared with the SCWRL-Server (<http://www1.jcsg.org/scripts/prod/scwrl>) and program O. Model evaluation and visualization was done in Pymol (<http://www.pymol.org>; DeLano Scientific, Palo Alto, CA, USA).

Results

Characterization of active site residues critical for NOD1 and NOD2 signaling

In order to decipher the molecular details involved in NOD1 and NOD2 activation, we analyzed the role of conserved residues predicted to be involved in ATP-binding and hydrolysis. Based on sequence alignments of the human NLRs and APAF-1, we previously identified highly conserved residues in the NACHT domain which supposedly affect ATPase activity (Fig. 1 and Table 1).⁴ To determine their role in NOD1 and NOD2-mediated NF-κB activation, we mutated these residues and evaluated their effect on elicitor- and autoactivation-induced NF-κB activity in luciferase reporter gene assays in HEK293T cells. Overexpression of NOD1 and NOD2 is known to induce autoactivation; therefore, the amount of NOD1 and NOD2 used was carefully titrated in preliminary assays (data not shown) and three established DNA concentrations were used for each construct. Importantly, comparable expression of all constructs was obtained as shown by Western blot analysis (Fig. 2D). Mutation of the Walker A (P-loop) lysine (GKS/T), which is responsible for nucleotide binding,^4,11 to either alanine or arginine (NOD1, K208; NOD2, K305) resulted in unresponsiveness to elicitor stimulation and lack of autoactivation in both NOD1 and NOD2 (Fig. 2A,B and data not shown), confirming recent reports on the essential role of the Walker A box in NOD1 and NOD2 signaling.^7,8 Similarly, mutation of the magnesium-co-ordinating aspartate residue in the Walker B box of NOD1 (D284) and NOD2 (D379) resulted in complete loss of NF-κB activation (Fig. 2A,B), showing that both canonical ATPase features are essential for activation of NOD1 and NOD2. The NLR proteins, however, contain a modified Walker B box which lacks the second acidic residue (DG instead of DD) that is usually responsible for priming of a water molecule and required for ATP hydrolysis.⁴ However, in most NLR proteins, including NOD1 and NOD2, the Walker B box is followed by a highly conserved patch of acidic residues, the extended Walker B box (DGhDE), suggesting that these acidic residues can functionally compensate for the missing second aspartate of the canonical Walker B box. To establish the function of this modified Walker B box, we replaced the corresponding amino acids in NOD1 and NOD2 by alanine. Mutation of the first aspartic acid of the extended Walker B box (NOD1, D287; NOD2, D382) significantly reduced elicitor-mediated NF-κB activation and led to loss of autoactivation of both NOD1 and NOD2 (Fig. 2A,B). This suggested that this residue of the extended Walker B box can functionally complement the missing second acidic residue of a canonical Walker B box in NLRs. In agreement, also the second acidic residue of the extended Walker B box (E288) was found to be essential for NF-κB activation and displayed an even stronger phenotype in NOD1 (Fig. 2A). In sharp contrast, mutation of the corresponding amino acid in NOD2 (E383) led to significantly increased basal NF-κB activation (Fig. 2B). This unexpected finding revealed a fundamentally different function of the extended Walker B box residues in NOD1 and NOD2 signaling. Of note, this result fits very well with the fact that natural occurring mutations in the extended Walker B box residues lead to autoactivation of NOD2, whereas such mutations have not yet been reported for NOD1. Mutations in the NOD2 extended Walker B box are associated with early-onset sarcoidosis (EOS, D382E/A612T)¹⁴ and BS (E383K),¹³ two severe auto-inflammatory human diseases resulting in hyperactive NOD2. In our assay system, we could recapitulate this phenotype and observed that both NOD2 D382E and E383K led to enhanced autoactivation compared to the wild-type, whereas a somewhat lower response was observed with MDP (Fig. 2C). Of note, the EOS mutation D382E displayed less hyperactivity than the E383K mutation which we attributed to the fact that we used NOD2 D382E instead of a double mutant D382E/A612T usually found in patients.¹⁴ Since both glutamate and aspartate are acidic residues, we suggest that glutamate partially adopts the function of D382.

Figure 1.

The active site of the NACHT domain in NOD1. (A) Model of the NOD1 nucleotide-binding site with a bound ADP molecule and Mg²⁺ ion. Conserved sequence motifs Walker A, Walker B, extended Walker B, Sensor 1, GxP (C-domain) and winged-helix (WH) histidine are shown in sticks. (B) Schematic representation of the domain architecture of NOD1 and NOD2. The magnification of the NACHT domain indicates the position of amino acid residues analysed in this study.

Figure 2.

Functional analysis of conserved residues in NOD1- and NOD2-mediated NF-κB activation. Nuclear factor-κB activation of NOD1 (A) and NOD2 (B) wild-type and mutant proteins was assayed in HEK293T cells transiently transfected with 0.1, 0.5 or 1 ng of the indicated plasmid, together with a NF-κB luciferase reporter system. Cells were left unstimulated or stimulated for 16 h with 500 nm triDAP or 50 nm MDP for NOD1 and NOD2, respectively. Percentage of normalized luciferase activity relative to 1 ng FLAG-NOD1 or NOD2 WT is shown. Values are given as mean + SD. (C) NF-κB activation of NOD2 wild-type and the NOD2 early-onset sarcoidosis- (D382E) and BS-associated (E383K) mutants. Nuclear factor-κB luciferase assay was conducted as described for (A) and (B). Values are given as mean + SD. (D) Western blot analysis showing the expression levels of NOD1 and NOD2 constructs. GAPDH served as loading control (lower panels).

Next, we explored the contribution of the conserved ATPase-related features Sensor 1, the GxP motif, and the WH-histidine to NOD1 and NOD2 activation. The Sensor 1 motif is involved in co-ordinating the γ-phosphate of ATP, whereas the conserved proline residue in the GxP motif is thought to interact with the adenine moiety of ATP (for review, see Hanson and Whiteheart¹¹). Mutation of the Sensor 1 domain in NOD1 (R333A) resulted in complete loss of NF-κB activation (Fig. 2A), whereas the corresponding mutation in NOD2 (R426A) still showed residual NF-κB activation after MDP stimulation (Fig. 2B). Mutation of the conserved proline in the GxP motif abolished elicitor-dependent NF-κB activation in both NOD1 (P391A) and NOD2 (P486A; Fig. 2A,B). Furthermore, mutation of the WH-histidine completely blocked NF-κB activation by NOD1 (H517A). In contrast, NOD2 H603A was still able to respond to MDP, albeit at reduced levels compared to the wild-type and showed increased autoactivation-mediated NF-κB activity (Fig. 2A,B). The latter finding was similar to the results obtained when exchanging the glutamic acid in the extended Walker B box supporting our conclusion that NOD1 and NOD2 are differentially activated.

Figure 3.

Subcellular localization of NOD1 and NOD2 mutants. Indirect immunofluorescence micrographs of HeLa cells transfected with FLAG-NOD1 (A) or FLAG-NOD2 (B) plasmids. The NOD1 or NOD2 signal alone (upper panels) and merge of FLAG-NOD1 or FLAG-NOD2 (red) with actin (green) and DNA (blue) are shown. Bars: 10 µm.

Taken together, we conclude that a functional Walker A and B box as well as Sensor 1 and GxP motifs are crucial for NOD1 and NOD2 activation. Most importantly, we assign a function to the extended Walker B box in NLRs and demonstrate that mutations of the extended Walker B box and the WH-histidine have opposing effects on NOD1- and NOD2-mediated signaling, revealing different mechanisms for their activation and regulation.

The NOD1 and NOD2 activation status is reflected by their subcellular localization

Recently, we and others reported that exogenously expressed active NOD1 and NOD2 localize to the plasma membrane in different human cells which relates to autoactivation of the overexpressed NLRs.^15,19^–²¹ In order to complement the results from the NF-κB activation assays presented above in a further cell-line and read-out, we determined the subcellular localization of the described NOD1 and NOD2 constructs in HeLa cells by indirect immunofluorescence analysis. Figure 3A shows that only NOD1 wild-type localized to actin-rich regions at the plasma membrane, whereas all mutants of NOD1 that did not display autoactivation in the NF-κB reporter assays (NOD1 K208A, D284A, E288A, R333A, P391A, and H517A) were restricted to the cytoplasm. Accordingly, also NOD1 D287A, which showed only elictor-mediated activation failed to localize to the plasma membrane upon overexpression (Fig. 3A). Similar results were obtained for NOD2. Indirect immunofluorescence analysis of NOD2 revealed pronounced membrane localization of the wild-type, the highly auto-active NOD2 mutant E383A, and to a lesser extent NOD2 H603A (Fig. 3B). The two disease-associated NOD2 mutants D382E (EOS) and E383K (BS), which exhibited the highest autoactivation in the NF-κB assays showed also a prominent membrane recruitment, whereas non-signaling active NOD2 mutants (NOD2 K305R, D379A, D382A, R426A, and P486A) exhibited an explicitly cytoplasmic localization pattern (Fig. 3B). To analyze if the membrane-bound fraction of NOD1 is in exchange with the cytoplasmic pool, we used NOD1 fused to a photo-activatable (PA) version of GFP. A region in the cytosol of HeLa cells transiently transfected with this construct was activated by a 405 nm laser pulse. Subsequently, the subcellular distribution of the activated (green) NOD1 protein was monitored. This revealed that NOD1 was readily recruited to the plasma membrane, indicating that the pool of NOD1 (and likely also NOD2) at the membrane reflects a steady state equilibrium exchange of NOD1 between a membrane bound fraction and a cytosolic fraction (Fig. 4A).

Figure 4.

Plasma membrane recruitment of active NOD1 and NOD2. (A) Plasma membrane recruitment of PA-GFP-NOD1 in living cells. HeLa cells were transfected with PA-GFP-NOD1 and mOrange (red) as control for transfection. Images were recorded every 20 s after activation of PA-GFP (green) by a light pulse of 405 nm in the indicated region. (B) The indicated NOD1 or NOD2 constructs were expressed transiently in HEK293T cells. Subsequently, the cells were lysed and the cytosolic and membrane fraction were prepared and subjected to Western blot analysis to detect NOD1 and NOD2. Probing for the cytosolic protein GAPDH served as control.

To substantiate the different membrane association of the NOD1 and NOD2 mutants, we furthermore conducted membrane fractionations in HEK293T cells transfected with different versions of NOD1 and NOD2. In line with the data obtained in the immunofluorescence analysis in HeLa cells, wild-type NOD1 and wild-type NOD2 as well as the auto-activating mutants NOD2 E383A and NOD2 H603A were found to be present in the membrane fraction, whereas the non-autoactivating mutants NOD1 K208A, NOD1 D287A, NOD1 E288A and NOD1 H517A as well as NOD2 K305R and NOD2 D382A were only detectable in the cytosolic fractions (Fig. 4B). In summary, this showed that membrane recruitment of NOD1 and NOD2 reflects their activation status. It also confirmed the conclusions from the NF-κB activation assays, supporting different functions of the second extended Walker B residue and the conserved histidine in the WH motif in NOD1 versus NOD2 activation.

Discussion

Both, NOD1 and NOD2 are pivotal PRRs that mediate induction of innate and adaptive immune responses in the mammalian host and are involved in regulating tissue and immune homeostasis.^2,22,23 Studies in mice suggest that lack of either of these NLRs results in reduced immune competence of the affected animals. In contrast, certain polymorphisms in NOD2 increase its activity and are linked to severe inflammatory diseases, suggesting that NOD2 hyperactivation can be detrimental to the host. However, our picture of the molecular details underlying activation and control of NOD1 and NOD2 still remains fragmentary. In particular, despite the general perception that members of the NLR family function as nucleotide binding proteins, the role of ATP hydrolysis in NLR signaling is poorly understood. Since no structural data on full-length or LRR-deleted NLRs are available at present, the current model of NLR activation is deducted from their apoptotic relative APAF-1 and from in vitro studies on purified NLR proteins.^24,25 Sequence analysis and homology modeling revealed that most NLRs not only share similar domain architecture with APAF-1, but also display a high degree of conservation within catalytically important motifs. Furthermore, by using purified NLRP1, NLRP3, NLRP12, and NLRC4 (IPAF) several groups established that these NLR family members do bind and hydrolyze ATP.²⁴^–²⁷ Accordingly, to these studies the general assumption has been that nucleotide binding and hydrolysis are required for the function of these proteins, i.e. for the induction of higher molecular platforms that promote downstream signaling events. To date, it has been unclear how these proteins achieve nucleotide hydrolysis despite their obvious lack of the second acidic Walker B residue required for catalysis. In the present study, we explain the evident ATPase activity of NLR proteins by systematically analyzing the functional role of highly conserved residues in the NACHT domain of NOD1 and NOD2 and their impact on NF-κB activation. Our data indicate that ATP binding is essential for both NOD1 and NOD2 signaling as the Walker A box residue in NOD1 (K208) and NOD2 (K305) turned out to be crucial for activation. This finding is supported by previously published reports, which show that NOD1 and NOD2 require ATP binding for activation.^7,8 Furthermore, we could assign a functional role of the Walker B box in NOD1 (D284) and NOD2 (D379), which is essential for co-ordination of the magnesium ion required for catalyzing the nucleotide hydrolysis step. Mutations affecting this residue lead to complete loss of signaling. Therefore, we conclude that NOD1 and NOD2 use a nucleotide hydrolysis step for their activation. Hitherto, the fact that most NLRs contain a modified Walker B box, which lacks the second acidic residue required for hydrolysis (DG instead of DD) has not been addressed.⁴ This raises the question how exactly nucleotide hydrolysis is carried out in NLR proteins. In a recent study, Ye et al.²⁴ used an NLRP12 Walker A/B box mutant where the Walker A signature and the first acidic residues of the classic and extended Walker B box were mutated to alanine (GKS → AAA and DxxDE → AxxAE). This variant of NLRP12 displayed decreased ATP binding capacity and no ATPase activity. However, this study did not identify the specific function of the individual catalytic residues involved. Therefore, we explicitly aimed to identify the residues that complement for the function of the missing second acidic residue in the classic Walker B box. To this end, we introduced single point mutations in NOD1 and NOD2 affecting the corresponding residues of the extended Walker B box. While mutation of the first acidic residue of the extended Walker B box in NOD1 (D287A) dampened autoactivation and reduced elicitor-mediated NF-κB activation to about 50% compared to wild-type, mutation of the second acidic residue (E288A) almost completely blocked NF-κB activation. Remarkably, mutations of the corresponding residues in NOD2 had significantly different effects. Exchange of the first acidic residue of the extended Walker B box to alanine in NOD2 (D382A) almost completely abolished NF-κB activation. On the contrary, NOD2 E383A exhibited elicitor-mediated NF-κB activation and displayed significantly higher autoactivation compared to wild-type NOD2. This important finding shows that the extended Walker B in NLR proteins compensates for the second acidic Walker B residue usually found in ATPases. Notably, the EOS- and BS-associated mutations in NOD2 affect the two acidic residues of the extended Walker B box (D382E and E383K). Whereas NOD2 D382E was already known to be a gain-of-function mutation, we showed that also the BS-associated E383K mutation increased basal NF-κB activation. These data indicate that E383 is required for NOD2 deactivation. Mutation of this residue necessarily leads to autoactivation, most likely because the NOD2 signaling platform is trapped in an ‘on’ state incapable of returning into a dormant form. Analysis of auxiliary residues important for ATPase activity showed that mutations in the Sensor 1 motif, the GxP and the WH-histidine motif of NOD1 resulted in complete loss of NF-κB activation, providing evidence that these motifs are implicated in nucleotide binding and hydrolysis. Since no autoactivation for any of these NOD1 mutants was observed, we propose an important role for the respective motifs in co-ordinating ATP binding and or the first ATP hydrolysis step. Accordingly, mutation of the GxP motif in NOD2 also resulted in complete loss of NF-κB signaling and a mutated Sensor 1 led to significantly reduced NF-κB activation. In contrast, the NOD2 WH-histidine mutant (H603A) only showed reduced NF-κB activation (about 50% compared to wild-type), and exhibited even higher levels of autoactivation than wild-type NOD2. This further supports fundamental differences in the mode of activation used by NOD1 and NOD2. To substantiate the data obtained in HEK293T cells, we additionally analyzed the subcellular localization of the NOD1 and NOD2 mutants in HeLa cells. In accordance with recent reports showing that signaling active NOD1 and NOD2 are recruited to the plasma membrane,^15,19^–²¹ we observed a membrane association only for constructs displaying autoactivation in the NF-κB assay. This result was also confirmed by biochemical cellular fractionation in HEK293T cells (Fig. 4B).

Analysis of photo-activatable GFP-tagged NOD1 indicated that the membrane fraction of NOD1 was in a stead-state exchange with the cytoplasmic fraction (Fig. 4A). We, therefore, propose that NOD1 and NOD2 constantly switch to autoactivation in the cytosol and thereby get recruited to the plasma membrane where they are finally de-activated in a steady-state process. This process likely explains the presence of a small fraction of endogenous NOD1 and NOD2 found at the plasma membrane of unstimulated cells.^16,19,28,29

In summary, we show for the first time that the extended Walker B in NLR proteins compensates for the second acidic Walker B residue usually found in ATPases and that this domain and the WH histidine have a different functional implication in NOD1 and NOD2 signaling.

The current model for the activation mechanism of ‘nodosome’ formation comprises the following steps: (i) elicitor recognition; (ii) conformational changes leading to a semi-opened state; (iii) nucleotide exchange (ATP for ADP) to drive oligomerization; and (iv) deactivation of the signaling platform by ATP hydrolysis.¹⁰ According to this model, an ATP hydrolysis step is necessary only to stop signaling and, consequently, mutations of residues crucial for ATP hydrolysis should result in autoactivation. Our data do not entirely support this model for NOD1 and NOD2, as they clearly show that certain mutations in NOD1 and NOD2 that have a crucial function in ATP hydrolysis completely abolish NF-κB activation. We provide evidence that an initial ATP hydrolysis step is required for NOD1 and NOD2 activation and subsequent signaling. However, while NOD1 seems to require only one ATP hydrolysis step for activation, NOD2 uses an additional hydrolysis step for deactivation: Thus NOD2 requires: (i) a first ATP hydrolysis step essential for its activation; and (ii) a second ATP hydrolysis step mediated by the second acidic residue of the extended Walker B box (E383), that is crucial for proper deactivation of the NOD2 signaling platform.

The discrepancy between NOD1 and NOD2 deactivation could be explained by different oligomerization mechanisms. Based on the overall similarity of the NLR proteins with APAF-1 and cryo-EM analysis of recombinant NLRP1,²⁷ we assume that NOD2 forms wheel-like oligomers where E383 is necessary for the second ATP hydrolysis step which is crucial for NOD2 deactivation. Results obtained for NOD1, however, indicate that NOD1 might not form such multimeric signaling complexes, as the corresponding residue is not involved in its deactivation, but rather in the activation process. NOD1, therefore, might use a mechanism other than ATP hydrolysis for deactivation. Of note, the NOD1 CARD domain was recently reported to form dimers, which could be involved in the different formation of a signaling complex.³⁰

This model also predicts that NOD2 is more prone for autoactivation in vivo than NOD1. Indeed, to obtain comparable autoactivation in transient transfection assays less NOD2 is needed than NOD1 (data not shown) and numerous auxiliary control factors that negatively affect NOD2 activation have been reported. Contrary, only a limited number of factors have been proposed to function in controlling NOD1 (reviewed by Kufer⁹). In addition, NOD1 and NOD2 show different expression profiles and are linked to specific downstream signaling processes. Whereas both proteins can activate NF-κB and MAPK pathways by TAK1, NOD1 might also act on caspase-8 pathways.³¹ In contrast, NOD2 has also been implicated in the activation of the non-canonical NF-κB pathway and type I interferon responses.³²^–³⁴ Furthermore, autoactivation seems to be prevented in vivo by low protein expression. Of note, whereas NOD1 is expressed at basal levels in most tissues, NOD2 expression can be induced by pathogen stimuli.^32,35,36

Conclusions

This study provides a revised model for NOD1 and NOD2 activation and highlights so far underestimated differences in their mode of activation. Ultimately, this furthers our understanding of functional aspects of NLR-mediated inflammatory diseases and allows opening new avenues for the intervention of NLR dysfunctions.

Footnotes

Acknowledgements

Birte Zurek and Martina Proell contributed equally to this work. The authors thank Jörg Fritz (University of Toronto, Toronto, Canada) for helpful discussions, Stefan Riedl (Sanford-Burnham Medical Research Institute, La Jolla, USA) and Jane Parker (Max-Planck Institute for Plant Breeding Research, Cologne, Germany) for critical comments on the manuscript. BZ and TAK acknowledge support from the Deutsche Forschungsgemeinschaft (DFG) grant SFB670-NG01. This work was supported by MCEXT-033534 to RS. MP was supported by a DOC-FFORTE-fellowship of the Austrian Academy of Sciences.

References

Kanneganti

Lamkanfi

Nunez

. Intracellular NOD-like receptors in host defense and disease. Immunity 2007; 27: 549–559.

Fritz

Ferrero

Philpott

Girardin

. Nod-like proteins in immunity, inflammation and disease. Nat Immunol 2006; 7: 1250–1257.

Franchi

Warner

Viani

Nunez

. Function of Nod-like receptors in microbial recognition and host defense. Immunol Rev 2009; 227: 106–128.

Proell

Riedl

Fritz

Rojas

Schwarzenbacher

. The Nod-like receptor (NLR) family: a tale of similarities and differences. PLoS ONE 2008; 3: e2119–e2119.

Girardin

Boneca

Carneiro

, et al.. Nod1 detects a unique muropeptide from Gram-negative bacterial peptidoglycan. Science 2003; 300: 1584–1587.

Girardin

Boneca

Viala

, et al.. Nod2 is a general sensor of peptidoglycan through muramyl dipeptide (MDP) detection. J Biol Chem 2003; 278: 8869–8872.

Inohara

Koseki

del Peso

, et al.. Nod1, an Apaf-1-like activator of caspase-9 and nuclear factor-kappaB. J Biol Chem 1999; 274: 14560–14567.

Ogura

Inohara

Benito

Chen

Yamaoka

Nunez

. Nod2, a Nod1/Apaf-1 family member that is restricted to monocytes and activates NF-kappaB. J Biol Chem 2001; 276: 4812–48128.

Kufer

. Signal transduction pathways used by NLR-type innate immune receptors. Mol Biosyst 2008; 4: 380–386.

10.

Danot

Marquenet

Vidal-Ingigliardi

Richet

. Wheel of life, wheel of death: a mechanistic insight into signaling by STAND proteins. Structure 2009; 17: 172–182.

11.

Hanson

Whiteheart

. AAA+ proteins: have engine, will work. Nat Rev 2005; 6: 519–529.

12.

Iyer

Leipe

Koonin

Aravind

. Evolutionary history and higher order classification of AAA+ ATPases. J Struct Biol 2004; 146: 11–31.

13.

van Duist

Albrecht

Podswiadek

, et al.. A new CARD15 mutation in Blau syndrome. Eur J Hum Genet 2005; 13: 742–747.

14.

Kanazawa

Okafuji

Kambe

, et al.. Early-onset sarcoidosis and CARD15 mutations with constitutive nuclear factor-kappaB activation: common genetic etiology with Blau syndrome. Blood 2005; 105: 1195–1197.

15.

Kufer

Kremmer

Adam

Philpott

Sansonetti

. The pattern-recognition molecule Nod1 is localized at the plasma membrane at sites of bacterial interaction. Cell Microbiol 2008; 10: 477–486.

16.

Kufer

Kremmer

Banks

Philpott

. Role for erbin in bacterial activation of Nod2. Infect Immun 2006; 74: 3115–3124.

17.

Patterson

Lippincott-Schwartz

. A photoactivatable GFP for selective photolabeling of proteins and cells. Science 2002; 297: 1873–1877.

18.

Rasband

. ImageJ. Bethesda, MD: U.S. National Institutes of Health, 2007.

19.

Barnich

Aguirre

Reinecker

Xavier

Podolsky

. Membrane recruitment of NOD2 in intestinal epithelial cells is essential for nuclear factor-{kappa}B activation in muramyl dipeptide recognition. J Cell Biol 2005; 170: 21–26.

20.

Lecine

Esmiol

Metais

, et al.. The NOD2–RICK complex signals from the plasma membrane. J Biol Chem 2007; 282: 15197–15207.

21.

McDonald

Chen

Ollendorff

, et al.. A role for Erbin in the regulation of Nod2-dependent NF-kappaB signaling. J Biol Chem 2005; 280: 40301–40309.

22.

Bouskra

Brezillon

Berard

, et al.. Lymphoid tissue genesis induced by commensals through NOD1 regulates intestinal homeostasis. Nature 2008; 456: 507–510.

23.

Clarke

Davis

Lysenko

Zhou

Weiser

. Recognition of peptidoglycan from the microbiota by Nod1 enhances systemic innate immunity. Nat Med 2009; 16: 228–231.

24.

Lich

Moore

Duncan

Williams

Ting

. ATP binding by monarch-1/NLRP12 is critical for its inhibitory function. Mol Cell Biol 2008; 28: 1841–1850.

25.

Wang

Dorsch

Ocain

. Nucleotide binding to CARD12 and its role in CARD12-mediated caspase-1 activation. Biochem Biophys Res Commun 2005; 331: 1114–1119.

26.

Duncan

Bergstralh

Wang

, et al.. Cryopyrin/NALP3 binds ATP/dATP, is an ATPase, and requires ATP binding to mediate inflammatory signaling. Proc Natl Acad Sci USA 2007; 104: 8041–8046.

27.

Faustin

Lartigue

Bruey

, et al.. Reconstituted NALP1 inflammasome reveals two-step mechanism of caspase-1 activation. Mol Cell 2007; 25: 713–724.

28.

Legrand-Poels

Kustermans

Bex

Kremmer

Kufer

Piette

. Modulation of Nod2-dependent NF-kappaB signaling by the actin cytoskeleton. J Cell Sci 2007; 120: 1299–1310.

29.

Till

Rosenstiel

Brautigam

, et al.. A role for membrane-bound CD147 in NOD2-mediated recognition of bacterial cytoinvasion. J Cell Sci 2008; 121: 487–495.

30.

Srimathi

Robbins

Dubas

Hasegawa

Inohara

Park

. Monomer/dimer transition of the caspase-recruitment domain of human Nod1. Biochemistry 2008; 47: 1319–1325.

31.

da Silva Correia

Miranda

Leonard

Hsu

Ulevitch

. Regulation of Nod1-mediated signaling pathways. Cell Death Differ 2007; 14: 830–839.

32.

Sabbah

Chang

Harnack

, et al.. Activation of innate immune antiviral responses by Nod2. Nat Immunol 2009; 10: 1073–1080.

33.

Pandey

Yang

Jiang

, et al.. NOD2, RIP2 and IRF5 play a critical role in the type I interferon response to Mycobacterium tuberculosis. PLoS Pathog 2009; 5: e1000500.

34.

Pan

Kravchenko

Katz

, et al.. NF-kappa B-inducing kinase regulates selected gene expression in the Nod2 signaling pathway. Infect Immun 2006; 74: 2121–2127.

35.

Lee

Seo

, et al.. Induction and localization of NOD2 protein in human endothelial cells. Cell Immunol 2005; 237: 37–44.

36.

Rosenstiel

Fantini

Brautigam

, et al.. TNF-alpha and IFN-gamma regulate the expression of the NOD2 (CARD15) gene in human intestinal epithelial cells. Gastroenterology 2003; 124: 1001–1009.