Pathogen Sensing by Nucleotide-binding Oligomerization Domain-containing Protein 2 (NOD2) Is Mediated by Direct Binding to Muramyl Dipeptide and ATP^*

Construction of NOD2 Expression Plasmids

Standard techniques of DNA manipulation were utilized (33). The NOD2-encoding cDNA was amplified with high fidelity Pfx DNA polymerase (Invitrogen) with primers that generated flanking PmeI and SgfI sites (for primer sequences, see supplemental Table 1). The resulting DNA fragment was digested by PmeI and SgfI and ligated with pFN21A vector (Promega), creating a gene encoding a NOD2 fusion protein with an N-terminal HaloTag. We designated this plasmid as pFN21A-NOD2. The following primers were used to amplify the HaloTag-NOD2 DNA fragment by PCR: 5′-TCGAATCTAGAATGGCAGAAATCGGTACTGG-3′ and 5′-AAGCAAGAGTCTGGTGTCCCT-3′. HaloTag-NOD2-His₆ DNA fragment was then ligated with pFastBac/CT-TOPO vector (Invitrogen), which generated a C-terminal hexahistidine tag. DNA sequencing was performed to confirm that there was not mutation in the Halo-NOD2-His coding sequence. pFastBac-Halo-NOD2 plasmid was used to transform DH10Bac competent cell to generate recombinant bacmid virus DNA. The NOD2-encoding cDNA and truncations of the NOD2-encoding cDNA were inserted in the mammalian expression vector, pcDNA^TM3.1/V5-His TOPO® (Invitrogen) or pFN21A (Promega) for HaloTag fusion proteins, using the manufacturers' recommendations. As described above, DNA fragments were amplified with high fidelity Pfx DNA polymerase (Invitrogen), and all plasmids and cloning intermediates were confirmed using both restriction digest and sequencing. The primer sequences used to generate these mammalian expression plasmids are noted in supplemental Table 1.

Baculovirus and Insect Cell Culture

Recombinant baculovirus expressing Halo-NOD2-His₆ was generated using the Bac to Bac system (Invitrogen). Insect cell culture and baculovirus infection were performed according to the manufacturer's protocols.

Purification of Recombinant NOD2, NOD2 CARDs, and GST

Sf9 cells infected with Halo-NOD2-His₆-expressing baculovirus were collected by centrifugation and washed with phosphate-buffered saline (PBS). The cell pellet was then resuspended in binding buffer (50 mm sodium phosphate, 300 mm NaCl, pH 8.0, supplemented with 0.1% CHAPS and Complete^TM proteinase inhibitor mixture (Roche Applied Science)), and homogenized using one pass through a French press. The solubilized lysate was prepared by centrifugation at 15,000 × g for 30 min and applied to an immobilized nickel column (Sepharose 6 Fast Flow, GE Healthcare). The column was washed with 5 column volumes of binding buffer, and the recombinant protein was eluted with binding buffer supplemented with 500 mm imidazole. The eluted recombinant protein was applied to HaloLink (Promega), washed, and eluted with His₆-tagged tobacco etch virus (TEV) protease. The proteins eluted from HaloLink resin were subsequently applied to immobilized nickel resin to remove His₆-tagged TEV protease, and the flow-through was collected.

NOD2 CARDs (residues Met²⁸–Arg²²⁷) were inserted using a Gateway cloning system into the vector pDest-HisMBP. The recombinant protein was expressed in Rosetta 2 Escherichia coli cells as an N-terminal His-MBP and C-terminal FLAG fusion. The protein was purified via nickel-NTA affinity chromatography, followed by removal of the N-terminal fusion tags with TEV protease. The protein was passed over a 1-ml HisTrap column (GE Healthcare) to remove the cleaved tag, residual full-length protein, and His-tagged TEV protease.

GST was expressed in BL21 E. coli cells from the plasmid pGEX-4T1(GE Healthcare). The protein was purified using glutathione-Sepharose 4B (GE Healthcare) as per the manufacturer's instructions.

ATPγS Binding and ATP Hydrolysis Assays

Nucleotide binding was assayed by a nitrocellulose filter binding assay (34). Nucleotide hydrolysis assays were carried out utilizing a modified acidified charcoal precipitation (35).

Cell Culture and Transfection

THP-1 and HEK293 (ATCC) were cultured under the recommended conditions. Fugene 6 (Promega) was used for transfection of HEK293 cells according to the manufacturer's protocol.

Immunoprecipitation and Immunoblot Analysis

Immunoprecipitations were carried out in 50 mm Tris-HCl (pH 8.0), 150 mm NaCl, 1 mm MgCl₂, and 0.5% Nonidet P-40 supplemented with Complete^TM protease inhibitor mixture (Roche Applied Science) and 1 mm ATP. Lysates from 5 × 10⁷ cells were incubated for 30 min at 30 °C prior to immunoprecipitation with agarose-conjugated anti-FLAG (M2) (Sigma-Aldrich) followed by SDS-PAGE. Anti-V5 antibody was from Invitrogen. Anti-hexahistidine antibody (GenScript, Piscataway, NJ) and anti-NOD2 (NOD2-15) were from Biolegend (San Diego, CA). Electrophoresis was carried out using the NuPAGE gel system (Invitrogen), and separated proteins were transferred to nitrocellulose using the iBlot dry blotting system (Invitrogen). Immunoblots were imaged with horseradish peroxidase-conjugated secondary antibodies (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and SuperSignal West chemiluminescent substrate products (Pierce) using the FluorChem E CCD-based imaging system (Protein Simple).

Biotinylated MDP Binding Assay

Purified NOD2 was incubated with biotinylated MDP at 24 °C for 1 h in 50 mm Tris-HCl (pH 7.0), 150 mm NaCl, and 0.1% CHAPS in the absence or presence of 10 μm ATP or ATPγS. 100 μl of Streptavidin microbeads (Miltenyi Biotec, Auburn, CA) was added to the reaction mix after a 1-h incubation. To assess MDP binding by NOD2 and truncated NOD2 proteins in cellular extracts, HEK293T cells were plated in 12-well culture plates and transfected with vectors expressing the indicated V5-tagged NOD2 proteins. After 24 h, cells were washed with PBS and lysed with radioimmune precipitation assay buffer (50 mm Tris-HCl (pH 8.0), 150 mm NaCl, 1% Igepal Co-630, 0.5% deoxycholate, and 0.1% SDS) containing proteinase inhibitors. Lysates were centrifuged to remove cell debris. Biotinylated MDP was added to lysates and rotated at 4 °C overnight. After 1-h or overnight incubation, 100 μl of Streptavidin microbeads was mixed with the proteins. MDP-bound protein was isolated by magnetic isolation using μ columns (Miltenyi Biotec, Auburn, CA) according to the manufacturer's protocol. The eluate was subjected to SDS-PAGE followed by immunoblot blot analysis using anti-NOD2 antibody or anti-V5 antibody, as indicated.

Surface Plasmon Resonance

Surface plasmon resonance experiments were performed on a Biacore T100 at 25 °C. Full-length NOD2 was diluted 10-fold in pH 4.0 acetate buffer and immobilized onto the surface of a CM5 sensor chip to a level of ∼3,500 response units. Recombinant NOD2 CARDs and GST proteins were buffer-exchanged into 10 mm sodium phosphate, pH 6.0, 150 mm NaCl, 50 μm EDTA, 0.01% Tween 20 and passed over the chip surface at a flow rate of 15 μl/min and concentrations of 0, 6.25, 12.5, 25, 50, and 100 μm. Background subtraction from a protein-free reference surface was performed, and the data were analyzed using BiaEvaluation software and GraphPad Prism 5.0.

Recombinant NOD2 Can Be Purified to Homogeneity by Sequential Affinity Chromatography

Biochemical characterization of the NLR protein functions remains largely unstudied more than 10 years after the discovery of these proteins. Generation of functional purified protein from recombinant sources has been a major challenge in this field. Portions of both NLRP12 and NLRC4 containing the central nucleotide-binding domain have been expressed, purified, and characterized (27, 28). To date, only NLRP1, NOD1, and NLRP3 have been characterized as full-length proteins (25, 26, 30). We have previously expressed and purified a dually tagged recombinant NLRP3 with an N-terminal hexahistidine tag and a C-terminal FLAG^TM tag expressed using recombinant baculovirus in insect cells (26). This strategy did not yield purified recombinant NOD2 in a quantify sufficient for biochemical analysis (data not shown). We evaluated numerous additional expression systems and fusion protein partners in an attempt to isolate purified NOD2. Ultimately, we generated a recombinant baculovirus expressing an N-terminal HaloTag^TM and C-terminal hexahistidine-tagged form of NOD2 in which both tags could be removed using recombinant TEV protease (Fig. 1A). The majority of NOD2 expressed in Sf9 cells infected with this recombinant baculovirus can be solubilized in CHAPS detergent (Fig. 1B). However, most of this solubilized protein fails to bind to immobilized nickel. Immunoblot analysis with antibodies directed to the hexahistidine tag indicates that the NOD2 protein that fails to bind nickel-containing chromatography resins does indeed contain the hexahistidine tag. This is probably the result of the protein folding (or misfolding) in a way that prevents access of this tag to the resin. The recombinant NOD2 protein that can be eluted from immobilized nickel can be purified to near homogeneity using HaloTag-ligand chromatography, removal of the tags using hexahistidine-tagged TEV protease, and subsequent reapplication to immobilized nickel (Fig. 1C). The final protein is immunoreactive with a monoclonal antibody directed against NOD2 and is not reactive with antibodies against the HaloTag^TM or hexahistidine tag (Fig. 1D). This procedure yields between 100 and 300 μg of purified NOD2 for each liter of Sf9 culture.

NOD2 Binds Non-hydrolyzable ATP Analogs and Exhibits ATPase Activity

Purified NOD2 protein exhibited binding activity toward ATPγS, a non-hydrolyzable ATP analog. NLR proteins contain conserved putative nucleotide-binding domains that fall into the structural superfamily known as AAA-ATPases (2). The Walker A motif, which contains the conserved sequence, GK(S/T), is involved in coordinating the γ-phosphate of bound ATP in many proteins with this motif (36). The Walker B motif, which contains the sequence ψψψD, is predicted to be a critical element for nucleotide binding and γ-phosphate hydrolysis. A mutant NOD2-encoding cDNA was constructed in which the Walker A motif amino acids Gly³⁰⁴, Lys³⁰⁵, and Ser³⁰⁶ (GKS → AAA) and Walker B motif amino acids Thr³⁷⁷, Phe³⁷⁸, and Asp³⁷⁹ (TFD → AAA), were mutated to alanine. This mutant NOD2 designated (NOD2 WAB) protein was purified to near homogeneity, as described for the wild type protein, but contained an additional protein contaminant (Fig. 2, A and B). WT and WAB mutant NOD2 were assayed for ATPγS binding; the mutant NOD2 had diminished binding compared with WT protein (Fig. 2C). Because contaminating proteins were more abundant in the WAB mutant NOD2 protein preparation than in the WT NOD2 protein preparation, these data indicate that the ATP binding in the preparation is attributable to the recombinant NOD2 and not other contaminants in the preparations. To further demonstrate that NOD2 was responsible for ATPγS binding activity in the protein preparations, the preparations were incubated with a nonhydrolyzable, biotin-labeled, photoactivatable ATP analog (8-azido-ATPγ-biotinpentylamine) and exposed to UV radiation. Analysis of ATP analog-linked proteins using SDS-PAGE and detection of biotin with streptavidin HRP indicated that virtually all of the detectable ATP-bound protein appears to be NOD2 (Fig. 2D). Purified NOD2 also exhibited ATPase activity, as shown by the release of [³²P]PO₄ from [γ-³²P]ATP incubated with purified NOD2 (Fig. 2E). Partially purified NOD2 WAB had diminished ATPase activity when compared with the wild type protein (Fig. 2E). The NOD2 WAB protein was found to label with 8-azido-ATPγ-biotinpentylamine, although with less intensity than WT NOD2 (supplemental Fig. S1A). Consistent with markedly diminished, but not completely extinguished, ATP binding and hydrolysis in vitro, cells transfected with cDNA encoding the NOD2 WAB mutant had minimally increased NF-κB-dependent reporter gene activity that was significantly less than that observed in cells transfected with cDNA encoding WT NOD2 and was not responsive to MDP stimulation (supplemental Fig. S1, B–D). Hence, the NOD2 WAB mutant protein retains some residual ATPγS binding activity, ATPase activity, and biological activity. Thus highly purified, recombinant NOD2 has both ATP binding and hydrolysis activities that depend on intact Walker A and Walker B motifs for full activity.

The rates of ATPγS binding and dissociation were also measured in preparations of wild type NOD2. The association of ATPγS was measured over time and demonstrated a rapid phase with a T½ of ∼4 min and a second prolonged phase. The second phase did not reach saturation in a measurable time frame (Fig. 3A). The rate of dissociation of ATPγS was also measured by loading the protein with [³⁵S]ATPγS and then incubating in a 200-fold excess of unlabeled ATPγS and measuring the decay of bound radiolabel. The disassociation rate of nucleotide has not been documented for other NLR proteins. As with ATPγS association, nucleotide disassociation appeared to occur in two phases, one with a T½ of <1 min and the second with a T½ of ∼8 min (Fig. 3B). Approximately half of the bound counts disassociated in the first phase and half in the second. Both disassociation rates were surprisingly rapid when compared with guanine nucleotide binding signaling proteins, which bind almost irreversibly to non-hydrolyzable GTP analogs.

A competition assay for NOD2 ATPγS binding activity was used to assess the nucleotide binding specificity of NOD2. Unlabeled ATP and dATP competed for [³⁵S]ATPγS binding by NOD2 with an IC₅₀ of 57 and 277 nm, respectively. The IC₅₀ for ADP competition with ATPγS was 2.8 μm (Fig. 4A). These data show NOD2 has a strong binding preference for triphosphate containing adenine nucleotides. Competition with non-adenine nucleotides did not fit one-site competition models well; however, estimates of the IC₅₀ were generated using this model. The calculated IC₅₀ for UTP, CTP, or GTP ranged from 47 μm to 24 mm, indicating at least a 1000-fold higher affinity for adenine-containing triphosphate nucleotides (Fig. 4B). Combined, these studies conclusively demonstrate that NOD2 is an adenine nucleotide-binding protein.

Purified Recombinant NOD2 Binds to MDP

Although NOD2 signaling is clearly activated by MDP in vivo and NOD2 is commonly referred to as a “receptor” for this bacterial PAMP, binding of this or other bacterial ligands to NOD2 has not been demonstrated. We sought to determine whether NOD2 was able to directly associate with MDP. Purified NOD2 was incubated with biotinylated MDP or biotin (as a negative control). The biotin-associated material was captured using streptavidin-carrying paramagnetic beads, and the associated proteins were analyzed by immunoblot. Purified NOD2 was found to associate with biotinylated MDP but not biotin alone (Fig. 5A). In this assay, only a small portion (∼3%) of the purified NOD2 was recovered from the streptavidin beads, which may result form poor affinity of NOD2 for the biotinylated MDP or steric hindrance limiting simultaneous binding of NOD2 and streptavidin to biotinylated MDP. Further experiments using isotope-labeled MDP would allow us to further address this, but these reagents are not commercially available at this time. The binding of MDP by NOD2 was also confirmed using lysates from cells transfected with cDNA encoding NOD (Fig. 5B). Lysates prepared from the transfected cells were exposed to biotinylated MDP. Subsequently, the MDP-associated proteins were isolated using streptavidin paramagnetic isolation. NOD2 was found to associate with biotinylated MDP, whereas other control proteins expressed after transfection, β-galactosidase, NLRC4, and NOD1, did not (Fig. 5B). To determine the domain(s) of NOD2 that is required for MDP binding, a series of NOD2 deletion mutants were expressed in HEK293 cells (Fig. 5C). Interestingly, all NOD2-derived proteins containing the central NBD were found to associate with MDP (Fig. 5C). Because the leucine-rich repeat-containing C terminus of NOD2 had previously been implicated in MDP-responsive signaling, we sought to further test whether this domain might also associate with MDP (37). Because the LRR domain failed to express as an isolated V5-tagged fusion protein at reasonable levels, full-length NOD2 and the LRR domain of NOD2 were expressed as HaloTag fusion proteins in transfected HEK293 cells. Full-length NOD2 fused to HaloTag associated with biotinylated MDP, as had been observed with other recombinant full-length NOD2 proteins. However, the HaloTag-NOD2 LRR fusion protein did not associate with MDP (Fig. 5C). Combined, these data suggest that NOD2 is a bona fide intracellular MDP receptor and that the nucleotide-binding domain containing amino acids 216–821 is sufficient to associate with MDP.

Recombinant NOD2 Homo-oligomerizes and Binds RIP2K

The activation of NOD2 signaling is thought to depend on the homo-oligomerization of the protein that facilitates the formation of a signaling complex that includes the kinase RIP2K (12, 13). We sought to determine whether recombinant purified NOD2 was able to form this signaling complex. Recombinant Halo-NOD2-His₆, enriched by immobilized nickel chromatography, was immobilized by incubation with HaloLink resin. Lysates from cells transfected with plasmids that express either V5-tagged NOD2 or RIP2K were incubated with immobilized NOD2 or control resin in which NOD2 was removed using TEV protease. The association of NOD2 or RIP2K from the transfected cell lysates with immobilized Halo-NOD2 and control resin was assayed by immunoblot analysis. Both NOD2 and RIP2K were found to associate with Halo-NOD2-His₆, whereas V5-tagged β-galactosidase did not (Fig. 6, A and B). Weak association of NOD2 and RIP2K with control resin was noted and may be the result of incomplete removal of NOD2 or nonspecific interactions with the resin (Fig. 6, A and B). Yeast two-hybrid studies have previously shown the CARDs of NOD2 to undergo homotypic interactions (38). To further assess the functionality of the full-length NOD2, we sought to repeat these observations using surface plasmon resonance. Full-length NOD2 was immobilized, and increasing concentrations of recombinant NOD2 CARDs or GST were used as analyte. Specific binding was only observed to the NOD2 CARDs (Fig. 6C) with a measurable K_d of 12.37 ± 0.58 μm. Non-linear regression using a specific binding model with Hill slope produced a value of the Hill slope of 1.6 ± 0.13, indicative of multiple binding sites and potential cooperativity consistent with a tandem CARD interaction process (Fig. 6C). These data suggest that recombinant purified NOD2 not only binds ATP and MDP but also is capable of supporting protein-protein interactions important in in vivo NOD2 signaling.

ATP Binding Modulates MDP Binding and Homo-oligomerization by Recombinant NOD2

We sought to determine whether ATP binding by NOD2 could alter biologically relevant activities of the recombinant protein. The association of purified NOD2 with MDP was assessed as described above after NOD2 was incubated with ATP, ATPγS, or no nucleotide. MDP binding by NOD2 was enhanced by incubation with non-hydrolyzable ATP analog but not with ATP (Fig. 7, A and B). Interestingly, the NOD2 WAB protein binding to biotinylated MDP was dramatically reduced when compared with WT NOD2 (Fig. 7C), further supporting the concept that MDP binding by NOD2 is modulated by association of the protein with ATP. The homo-oligomerization of NOD2 was also assessed by assaying interaction between V5-tagged NOD2 in transfected cell lysates and HaloLink, resin-bound, recombinant Halo-NOD2-His₆ incubated with ATP, ATPγS, or no nucleotide. As with association between NOD2 and MDP, NOD2 homo-oligomerization was increased by binding to ATPγS (Fig. 7D).