Research Article

Palmitoylation of NOD1 and NOD2 is required for bacterial sensing

See allHide authors and affiliations

Science  25 Oct 2019:
Vol. 366, Issue 6464, pp. 460-467
DOI: 10.1126/science.aau6391

NODs require S-palmitoylation to signal

The compartmentalization of proteins within the cell is essential for their function. The addition of lipid molecules redistributes proteins to the cell surface or to membrane-bound organelles. Working in transgenic mice and in tissue cultured cells, Lu et al. found that nucleotide oligomerization domain–like receptors 1 and 2 (NOD1 and NOD2), two proteins responsible for detecting bacterial products, required lipid modifications for their recruitment to the cell membrane and function. The specific modification, palmitoylation at a cysteine thiol, was mediated by the enzyme ZDHHC5. Loss of ZDHHC5 or removal of key modification residues in NOD1 and NOD2 abolished their function, compromising antibacterial responses. Human variants of NOD2 display altered palmitoylation, which could help to explain many inflammatory conditions, such as irritable bowel syndrome.

Science, this issue p. 460

Abstract

The nucleotide oligomerization domain (NOD)–like receptors 1 and 2 (NOD1/2) are intracellular pattern-recognition proteins that activate immune signaling pathways in response to peptidoglycans associated with microorganisms. Recruitment to bacteria-containing endosomes and other intracellular membranes is required for NOD1/2 signaling, and NOD1/2 mutations that disrupt membrane localization are associated with inflammatory bowel disease and other inflammatory conditions. However, little is known about this recruitment process. We found that NOD1/2 S-palmitoylation is required for membrane recruitment and immune signaling. ZDHHC5 was identified as the palmitoyltransferase responsible for this critical posttranslational modification, and several disease-associated mutations in NOD2 were found to be associated with defective S-palmitoylation. Thus, ZDHHC5-mediated S-palmitoylation of NOD1/2 is critical for their ability to respond to peptidoglycans and to mount an effective immune response.

The cytosolic pattern recognition receptors (PRRs) nucleotide oligomerization domain 1 (NOD1) and NOD2 play crucial roles in host defense and survival, primarily by conferring responsiveness to cytosolic bacterial peptidoglycans [γ-d-glutamyl-meso-diaminopimelic acid (iE-DAP) and muramyl dipetide (MDP)] shed by bacteria during infection (1, 2). Dysregulation of NOD1/2 function leads to severe immunologic and inflammatory diseases such as Crohn’s disease (CD) and Blau syndrome (1, 2). A NOD2 variant (3020insC frameshift mutation that leads to a truncated NOD2 protein) is implicated in the pathogenesis of CD (3, 4). Although soluble in the cytosol, NOD1/2 associate with the plasma membrane (PM) and endosomal compartments (58) for the surveillance of bacterial cell wall components and promote activation of the nuclear factor κB (NF-κB) and mitogen-activated protein kinase MAPK) signaling pathways from endosomal membranes by means of the RIP2 kinase (9). Although fully competent to bind MDP (10), membrane association is not observed in the NOD23020insC variant, suggesting that membrane localization of NOD1/2 is essential for their function (6). Lacking recognizable membrane-targeting domains (11), NOD1/2 have been suggested to be anchored to membranes indirectly, through cytoskeletal components or membrane-bound proteins (1216) or to endosomes by endosomal proteins such as SLC15A3 (6). However, these models do not fully explain the rapid redistribution of NODs from plasmalemma to endosomal compartments in response to bacterial invasion (17).

Soluble proteins can be targeted to cellular membrane structures by lipidation (18). Palmitoylation, unlike farnesylation or geranylgeranylation, does not require a consensus motif in target proteins and is widely implicated in the regulation of protein localization, trafficking, and stability (1820). S-palmitoylation, catalyzed by the ZDHHC domain–containing protein acyl-transferases (PATs) (21), also plays important roles in immune responses (22).

NOD1 and NOD2 are S-palmitoylated

Treatment with 2-bromopalmitate (2BP), an inhibitor of the ZDHHC PATs (21, 23), effected the redistribution of green fluorescent protein (GFP)–NOD1 and GFP-NOD2 from plasmalemmal and endosomal membranes (Fig. 1A and fig. S1A) to the cytosol in RAW264.7, HCT116 (Fig. 1B), and human embryonic kidney (HEK) 293 cells (fig. S1B), mimicking the phenotype reported for the CD mutant protein NOD23020insC (Fig. 1C and fig. S1C) (6). To quantify the effect of 2BP on NOD1/2 PM localization, we used the split superfolder GFP (sfGFP) system combined with dual-color flow cytometry (24). Bicistronic plasmids were created that coexpress seven copies of GFP11 tethered to the plasmalemma by the transmembrane domain of the PDGF receptor (HA-TMPDGF-GFP11x7) and the GFP1–10 fragment alone (as control) or fused to NOD1 or NOD2 at their N termini. Robust reconstituted GFP fluorescence (GFPcomp) was observed at the PM of 293 cells (fig. S1, D and E). 2BP treatment reduced the PM localization of NOD1/2 that prevented the formation of functional GFP to levels comparable with that of controls (Fig. 1, D and E). The subcellular distribution of endogenous NOD1 (fig. S1F) in primary mouse bone marrow–derived macrophages (mBMDMs) (Fig. 1F) and primary human monocyte–derived macrophages (hMDMs) (fig. S1G) was similarly altered by 2BP treatment (Fig. 1G and fig. S1H). 2BP treatment of primary mBMDMs (Fig. 1H and fig. S1I) or the colorectal epithelial cell line HCT116 (fig. S1, J and K) also substantially reduced the levels of endogenous NOD1 (fig. S1, L and M) or NOD2 (fig. S1, N and O) associated with the insoluble membrane fraction but did not affect total NOD1/2 protein levels (Fig. 1H and fig. S1, I to K). By contrast, the subcellular distribution of NOD1/2 was not affected by inhibition of farnesyltransferase or geranyltransferase activity by using FTI-276 or GGTI-2133, respectively (fig. S1P).

Fig. 1 NOD1/2 S-palmitoylation is required for membrane association and agonist-triggered signaling.

(A and B) Fluorescence microscopy of RAW264.7 and HCT116 cells expressing GFP-NOD1 or GFP-NOD2 in the presence of (A) dimethyl sulfoxide (DMSO) or (B) 50 μM 2BP. (C) Fluorescence microscopy of RAW264.7 cells expressing the GFP-NOD2 CD variant 3020insC. (D and E) Plasmalemmal GFP-fragment complementation assay (GFPcomp fluorescence) for (D) NOD1 and (E) NOD2. (F and G) Endogenous NOD1 [green; 4′,6-diamidino-2-phenylindole (DAPI) in blue] in mouse BMDM treated with (F) DMSO or (G) 2BP. (H) Total, cytosolic, and membrane fractions of BMDM pretreated with DMSO or 2BP (100 μM) were immunoblotted with antibodies directed against NOD1, NOD2, a cytosol marker (LDHA), or a PM (CD11b) marker. (I) S-palmitoylation levels of flag-NOD1/2 expressed in HEK293T cells in the presence of HAM or 2BP (100 μM). (J and K) Quantification of the 2BP treatment effect on (J) NOD1 and (K) NOD2 S-palmitoylation of immunoblots as in (I). (L and M) Covalent attachment of 17-ODYA to (L) flag-NOD1 or (M) flag-NOD2 expressed in HEK293T cells. (N and O) p65 and p38 phosphorylation levels in response to C12-iE-DAP- (250 ng ml−1) or MDP (500 ng ml−1) in RAW264.7 cells. Data in (H), (I), (N), and (O) are representative of three independent experiments, and data in (D), (E), and (J) to (M) represent the mean ± SEM of triplicate samples. ns, not significant; **P < 0.01, ***P < 0.001, ****P < 0.0001, Student’s t test; AU, arbitrary units. Scale bars, 5 μm.

To examine NOD1/2 palmitoylation, we performed an acyl-biotin exchange (ABE) assay (25). Ectopically expressed NOD1/2 were acylated (Fig. 1I), and their acylation levels were reduced by more than 75% upon treatment with 2BP (Fig. 1, J and K). Loss of signal upon omission of hydroxylamine (HAM) treatment demonstrated that NOD1/2 incorporate palmitate through a thioester linkage (Fig. 1I). Metabolic incorporation of the bioorthogonal fatty acid analog 17-octadecynoic acid (17-ODYA) (fig. S1R) was also monitored with click chemistry conjugation and fluorescent detection (Fig. 1, L and M) (26).

2BP impaired downstream signaling triggered by the PRRs, as indicated by reduced NF-κB activity in RAW267.4 and HCT116 cells treated with the iE-DAP analog C12-iE-DAP or MDP (Fig. 1, N and O, and fig. S1, S and T). These results agree with the ability of 2BP to impair the release of cytokines in C12-iE-DAP– and MDP-stimulated mBMDMs (fig. S1, U and V) (6). Furthermore, treatment with cerulenin (23), a drug that inhibits S-palmitoylation, yielded similar levels of NF-κB phospho-p65 and phospho-p38 inhibition (fig. S1, W and X) in RAW264.7 cells. Together, these data indicate that NOD1/2 are palmitoylated and that this posttranslational modification is required for optimal sensing of bacterial peptidoglycans and activation of NF-κB signaling.

Multiple cysteine residues in NOD1/2 are S-acylated

On the basis of evolutionary conservation, the crystal structure of the Oryctolagus cuniculus NOD2 protein (PDB:5IRL), and the loss of function in the NOD2C1033 truncation mutant protein, we created a series of NOD1 and NOD2 Cys to Ser mutants (NOD1C952S, NOD1C557S, NOD1C567S, NOD1C558,567,952S; NOD2C395S, NOD2C1033S, and NOD2C395,1033S) (fig. S2, A and B), and their S-palmitoylation levels were characterized by using confocal microscopy and ABE assays. Similar to that observed for the CD variant NOD23020insC (Fig. 1C and fig. S1C), all of these single- or multiple-cysteine GFP-NOD1/2 mutants displayed cytosolic localization in HEK293 cells and RAW264.7 macrophages (Fig. 2A and fig. S2, C to E) and decreased palmitoylation levels (Fig. 2, B and C, and fig. S2, F and G). Palmitoylation of the NOD1 triple mutant NOD1C558,567,952S (Fig. 2, B and C) and the NOD2 double-Cys mutant NOD2C395,1033S (fig. S2, F and G) was further reduced by more than 50% as compared with the single NOD1/2 Cys to Ser mutants. The lack of S-palmitoylation did not affect the stability of the mutant proteins (fig. S2H). Similar to that observed in 2BP-treated RAW264.7 and HCT116 cells expressing wild-type (WT) NOD1/2 (Fig. 1H and fig. S1, M to O), the NOD1C558,567,952S and NOD2C395,1033S variant proteins were depleted from the membrane fraction in untreated HCT116 cells (Fig. 2D and fig. S2I) and displayed reduced reconstituted GFP fluorescence (Fig. 2E and fig. S2J) and impaired agonist-induced NF-κB–dependent luciferase activity (Fig. 2F and fig. S2, K to M).

Fig. 2 Multiple cysteine residues in NOD1 can be acylated.

(A) Fluorescence microscopy of RAW264.7 cells expressing a GFP-tagged S-palmitoylation–deficient NOD1 mutant (NOD1C558,567,952S). (B) Streptavidin blot detection of NOD1 S-palmitoylation–deficient mutants. (C) Quantification of NOD1 S-palmitoylation–deficient mutants in (B). (D) Membrane fractions (or total lysate) of HCT116 cells expressing flag-WT or S-palmitoylation–deficient NOD1C558,567,952S were immunoblotted for NOD1 or the PM marker E-cadherin. (E) Reconstituted plasmalemma-associated GFPcomp fluorescence for NOD1C558,567,952S. (F) HEK293T cells were transfected with an NF-κB reporter β-Gal-SV40 expression plasmid and NOD1 overexpression plasmids. Luciferase assays were performed after stimulation with C12-iE-DAP (1 μg ml−1). (G) NF-κB nuclear translocation in HCT116 cells transiently transfected with GFP-WT NOD1, S-palmitoylation–deficient GFP-NOD1C558,567,952S or GAP43 (amino acids 1 to 11) tagged GFP-NOD1C558,567,952S and stimulated with 500 ng ml−1 C12-iE-DAP. Cells were stained with antibody to NF-κB p65 and DAPI. White asterisks indicate the nuclei location in transfected cells. (H) Quantification of NF-κB nuclear translocation, relative to total NF-κB p65 (n ≥ 30 cells for each condition). Data in (B) and (D) are representative of three independent experiments, and data in (C), (E), and (F) represent the mean ± SEM of triplicate samples. ns, not significant; **P < 0.01, ***P < 0.001, ****P < 0.0001, Student’s t test. Scale bars, 5 μm.

To demonstrate that membrane recruitment of NOD1/2 is required for activation of the NF-κB pathway, we made use of an S-palmitoylation–dependent PM targeting sequence of the neuronal growth cone protein GAP43 (amino acids 1 to 11, MLCCMRRTKQV; hereafter denoted GAP43PS) (27). Fluorescence microscopy of HCT116 cells expressing GFP-NOD1, GFP-NOD1C558,567,952S, and GAP43PS-GFP-NOD1C558,567,952S revealed that the GAP43 S-palmitoylation motif could rescue the localization of the S-palmitoylation–deficient NOD1 mutant (Fig. 2G and fig. S2N, bottom) and induced the nuclear translocation of the NF-κB p65 (RelA) subunit (Fig. 2H and fig. S2N, bottom) and consequently induced NOD1 agonist NF-κB–dependent luciferase activity (fig. S2O). Thus, S-palmitoylation of NOD1/2 is required for their proper membrane targeting and the activation of immune signaling pathways.

ZDHHC5 is necessary for NOD1 and NOD2 S-palmitoylation

BirA-tagged proteins were generated (fig. S3A) and used in a BioID screen (28) to identify NOD1- and NOD2-interacting proteins (tables S1 and S2). As expected, a number of plasmalemmal and endomembrane proteins were identified as high-confidence NOD1- and NOD2-proximity interactors (fig. S3, B and C). Membrane proteins and membrane-associated polypeptides [as determined with Gene Ontology (GO) enrichment analysis: pantherdb.org GO:0016020] represented >85% (111 of 129) of the NOD1 and >80% (58 of 72) of the NOD2 high-confidence proximity interactors. Because NOD1/2 are membrane localized in the absence of agonists in 293 cells, the BioID screen was conducted in resting cells. Thus, known interactors of NOD1/2 such as tumor necrosis factor (TNF) receptor–associated factor 2 (TRAF2) or RIP2 were not found in our dataset. Among the NOD1/2 high-confidence proximity interactors was ZDHHC5 (fig. S3B) (19), a ubiquitously expressed integral membrane protein of the DHHC palmitoyltransferase (PAT) family. Similar to the NODs themselves, ZDHHC5 also localized to the PM and vesicular structures (fig. S3D). We thus hypothesized that ZDHHC5 could form a transient complex with NOD1/2. ZDHHC5 was detected in immunoprecipitates of both NOD1 and NOD2 (fig. S3, E and F), and NOD1/2 were detected in ZDHHC5 immunoprecipitates (fig. S3, G and H). ZDHHC5 did not interact with NLRP3, showing a degree of specificity for its interaction with NOD1/2 (fig. S3I). GFP-NOD1 and mCherry-ZDHHC5 displayed strong colocalization in RAW264.7 and HEK293 cells (fig. S3, J and K), and endogenous NOD1 colocalized with ZDHHC5 in hMDMs (fig. S3L). By contrast, GFP-NOD1 was largely cytosolic when coexpressed with a catalytically inactive ZDHHC5 mutant, ZDHHC5C134S (21, 29) (fig. S3M). Similar results were obtained for GFP-NOD2 (fig. S3, N and O).

Silencing the expression of ZDHHC5 in RAW264.7 cells by means of small interfering RNA (siRNA) [validated with quantitative polymerase chain reaction (PCR)] (fig. S4A) or in HEK293 cells with a short hairpin RNA (shRNA) (fig. S4B) rendered NOD1/2 largely cytosolic (Fig. 3A and fig. S4, C and D), resembling the distribution observed after 2BP treatment (Fig. 1B and fig. S1B). Expression of the shRNA targeting ZDHHC5 also caused profound defects in NOD1 and NOD2 palmitoylation (Fig. 3, B to D), as assessed with the ABE assay. By contrast, the related PAT ZDHHC20 did not associate with NOD1/2 (fig. S5, A and B), and silencing of its expression did not affect NOD1/2 S-palmitoylation (fig. S5, D to G). Silencing ZDHHC5 expression in RAW264.7 cells also abrogated C12-iE-DAP–induced NF-κB activation as assessed with phospo-p65 and phospho-p38 immunoblotting (fig. S4E), similar to that observed in HEK293T cells (fig. S4, F and G).

Fig. 3 The palmitoyl acyltransferase ZDHHC5 regulates NOD1 and NOD2 S-palmitoylation.

(A) Representative images of GFP-NOD1 localization in RAW264.7 cells transfected with (left) nonspecific scramble RNA or (right) siRNAs targeting mouse ZDHHC5. (B) NOD1 and NOD2 S-palmitoylation levels in WT and ZDHHC5 KO cells. (C and D) Densitometric analysis of immunoblots as in (B). (E) Total, cytosolic, and membrane fractions of BMDM cells from LysM-Cre/ZDHHC5+/+ and LysM-Cre/ZDHHC5fl/fl mice were immunoblotted with antibodies directed against NOD1/2, a cytosol marker (LDHA), and the PM marker CD11b. (F) Confocal microscopic analysis of BMDM cells derived from LysM-Cre/ZDHHC5+/+ or LysM-Cre/ZDHHC5fl/fl animals. Endogenous NOD1, green; DAPI, blue. (G and H) Effect of ZDHHC5 KO on C12-iE-DAP–induced (250 ng ml−1) or MDP-induced (500 ng ml−1) p65 and p38 kinase phosphorylation in mouse BMDM cells from LysM-Cre/ZDHHC5+/+ and LysM-Cre/ZDHHC5fl/fl mice. Data in (B), (E), (G), and (H) are representative of three independent experiments, and data in (C) and (D) represent the mean ± SEM of triplicate samples. ****P < 0.0001, Student’s t test. Scale bars, 10 μm.

A myeloid cell lineage–specific conditional ZDHHC5 knockout (KO) animal model was generated by crossing mice carrying a floxed-ZDHHC5 allele (ZDHHC5fl/fl) with myelomonocytic cell–specific Lysozyme M (LysM)–Cre Recombinase mice (fig. S4, H and I). Backcrossing of LysM-Cre/ZDHHC5fl/+ mice led to homozygous ablation of ZDHHC5 in the mouse myeloid cell lineage (LysM-Cre/ZDHHC5fl/fl). Genotyping by means of PCR confirmed the presence of LoxP and LysM-Cre in LysM-Cre/ZDHHC5fl/fl mice (fig. S4J). Immunoblot analysis and immunofluorescence microscopy confirmed the depletion of ZDHHC5 protein expression in LysM-Cre/ZDHHC5fl/fl BMDM (fig. S4, K and L). These mice exhibited no obvious phenotype under standard laboratory conditions. Cellular fractionation of WT and ZDHHC5 KO mBMDMs confirmed that NOD1/2 membrane association was substantially reduced in the absence of ZDHHC5 (Fig. 3E and fig. S4M). In contrast to that observed in WT mice, endogenous NOD1 was predominantly cytosolic in ZDHHC5 KO mBMDMs (Fig. 3F), and ZDHHC5 KO mBMDMs displayed an impaired NOD1/2–dependent activation of NF-κB and p38 MAPK signaling in response to C12-iE-DAP (Fig. 3G) or MDP (Fig. 3H), despite similar NOD1/2 protein expression levels (Fig. 3E).

NOD1/2 S-palmitoylation levels increased by more than twofold upon stimulation with C12-iE-DAP or MDP (fig. S6, A to C) and correlated with additional recruitment of NOD1/2 proteins to peripheral membranes (fig. S6, D to J). Thus, subcellular fractionation of HCT116 cells and primary mBMDMs stimulated under similar conditions yielded similar results (fig. S6, D to J). This agonist-mediated increase in S-palmitoylation was dependent on ZDHHC5 because cells failed to respond to NOD agonists in its absence (fig. S6, K to M). Thus, ZDHHC5 mediates NOD1/2 S-palmitoylation, and this modification is required for membrane association and downstream signaling in a physiologically relevant context.

ZDHHC5-mediated palmitoylation regulates NOD1/2 function

GFP-NOD1 and GFP-NOD2 are localized to Salmonella-containing vacuoles (SCV) in Salmonella typhimurium–infected cells (Fig. 4, A and B, and fig. S7, A and B) (6), and SCV localization was markedly reduced with 2BP treatment (fig. S7, C and D) or ZDHHC5 knockdown (Fig. 4, C and D). Consistent with this observation, palmitoylation-deficient NOD1/2 mutants failed to localize to SCVs (Fig. 4, E and F, and fig. S7, E and F), and levels of flag-NOD2 protein associated with phagosomes containing MDP-coated beads were substantially decreased by 2BP treatment (fig. S7G). Consistent with these observations, endogenous NOD1 was associated with phagosomes in WT mBMDMs (Fig. 4G) but not KO mBMDMs (Fig. 4H).

Fig. 4 ZDHHC5-mediated S-palmitoylation is indispensable for NOD1/2 recruitment to bacteria containing phagosomes.

(A to F) Confocal microscopy of NOD1 and NOD2 in Salmonella-infected HEK293 cells. Cells transiently transfected with [(A) and (C)] GFP-NOD1, [(B) and (D)] GFP-NOD2, or [(C) and (D)] shZDHHC5::tagBFP (shown in the blue inset) and [(A) to (D)] infected with red fluorescent protein (RFP)–expressing S. typhimurium. [(E) and (F)] Confocal microscopy of Salmonella-infected HEK293 cells transiently expressing the indicated (E) S-palmitoylation deficient GFP-NOD1or (F) GFP-NOD2 mutants. Colocalization of NOD1/2 and S. typhimurium was described by the Pearson’s correlation coefficient (PCC). (G and H) Confocal microscopy of Salmonella-infected ZDHHC5-deficient (LysM-Cre/ZDHHC5fl/fl) or WT (LysM-Cre/ZDHHC5+/+) BMDMs immunostained with α-NOD1 antibody (green). Scale bars, 10 μm and (insets) 5 μm.

Similar to NOD1/2 (Fig. 4, A and B), overexpressed mCherry-ZDHHC5 (Fig. 5A), GFP-ZDHHC5 (fig. S8A), and endogenous ZDHHC5 were recruited to large phagosomes (fig. S8B) and SCVs in infected WT mBMDMSs and hMDMs (Fig. 5B and fig. S8C) (1, 6). These observations were in line with the NODs and ZDHHC5 co-fractionating with early and late endosomal markers. Consistently, endogenous ZDHHC5 and NOD1 were found on C12-iE-DAP–coated bead–containing phagosomes isolated from mBMDMs after asynchronous phagocytosis (Fig. 5C). By contrast, C12-iE-DAP bead phagosomes isolated from LysM-Cre/ZDHHC5fl/fl mBMDMs were devoid of NOD1 but were positive for EEA1 and Rab7 (Fig. 5C). Furthermore, overexpressed ZDHHC5 was associated with LAMP1-positive phagosomes generated by the internalization of C12-iE-DAP– or MDP-coated beads (Fig. 5D and fig. S8D). The recruitment of ZDHHC5 to phagosomes did not require its transferase activity (Fig. 5D). However, overexpression of catalytically inactive ZDHHC5 prevented the recruitment of NOD2 to MDP-coated bead–containing phagosomes (Fig. 5D), similar to cells treated with 2BP (fig. S7G).

Fig. 5 ZDHHC5 is recruited to bacteria-containing phagosomes.

(A) Confocal microscopy of RAW264.7 cells expressing mCherry-ZDHHC5 challenged with red blood cells labeled with Cy5-conjugated IgG (blue). (B) Confocal microscopy of wild-type (LysM-Cre/ZDHHC5+/+) and ZDHHC5-deficient (LysM-Cre/ZDHHC5fl/fl) BMDMs immunostained with α-ZDHHC5 antibody (green) and infected with RFP–S. typhimurium, as in Fig. 4G. (C) Western blot analysis of C12-iE-DAP–coated magnetic beads containing phagosomes isolated from ZDHHC5-deficient (LysM-Cre/ZDHHC5fl/fl) or WT (LysM-Cre/ZDHHC5+/+) BMDMs, respectively. (D) Western blot analysis of MDP-coated magnetic beads containing phagosomes isolated from HEK293T cells coexpressing flag-NOD2 with 3xHA-ZDHHC5 or catalytically inactive 3xHA-ZDHHC5C134S. (E to H) ZDHHC5-deficient (LysM-Cre/ZDHHC5fl/fl) or wild-type (LysM-Cre/ZDHHC5+/+) BMDMs were incubated with C12-iE-DAP– or MDP-coated beads. CXCL-1 and IL-6 in medium were quantified by means of enzyme-linked immunosorbent assay. Data in (C) and (D) are representative of three independent experiments, and data in (E) to (H) represent the mean ± SEM of triplicate samples. ns, not significant; **P < 0.01, ***P < 0.001, ****P < 0.0001, Student’s t test; AU, arbitrary units. Scale bars, 10 μm.

Consistent with previous findings (6), when WT macrophages were incubated with C12-iE-DAP (Fig. 5, E and F) or MDP beads (Fig. 5, G and H), CXCL-1 (Fig. 5, E and G), and interleukin-6 (IL-6) (Fig. 5, F and H) were produced. However, CXCL-1 and IL-6 production by ZDHHC5-deficient mBMDMs was markedly lower than that in littermate control cells, despite internalizing more beads (fig. S8E). Together, these data indicate that ZDHHC5 is constitutively localized to phagosomes and that its catalytic activity is required for NOD1/2 phagosomal recruitment and NOD1/2–dependent signaling.

NOD2 S-palmitoylation and disease relevance

NOD1 and NOD2 polymorphisms are associated with CD, ulcerative colitis (UC), Blau syndrome, Behcet’s syndrome, early-onset sarcoidosis (EOS), and atopic diseases (1, 2). CD-associated NOD2 loss-of-function mutations result in reduced NF-κB activity upon stimulation with MDP (6), whereas EOS and Blau syndrome are associated with apparent NOD2 gain-of-function mutations (30, 31). The three most common nonsynonymous NOD2 variants (R702W, G908R, and 3020insC) (Fig. 6A and fig. S9A) account for roughly 80% of all NOD2-associated variants found in CD cases (1) and retain the ability to bind to MDP with nanomolar affinity (10).

Fig. 6 NOD2 coding variants manifest aberrant S-palmitoylation.

(A) Ribbon representation of CD (3020insC, orange; R702, G908, L248, A612, and A755, red) and EOS (C495Y, blue) mutations on a hNOD2 3D structure model. (Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. In the mutants, other amino acids were substituted at certain locations; for example, R702W indicates that arginine at position 702 was replaced by tryptophan.) (B) S-palmitoylation levels of the indicated NOD2 variants (3020insC, R702, G908, L248, A612, and A755) expressed in HEK293T cells with or without HAM. (C) Quantification of S-palmitoylation of NOD2 variants of immunoblots as in (B). (D) Effect of 2BP (50 μM) on S-palmitoylation of the NOD2 EOS variant NOD2C495Y. (E) Quantification of S-palmitoylation level of NOD2C495Y in (D). (F) NF-κB–dependent luciferase reporter gene activity in HEK293 cells expressing GFP-NOD2C495Y treated with MDP (2 μg ml−1) and DMSO or 2BP (50 μM). Luciferase activity normalized as in Fig. 1G. Data in (C), (E) and (F) are the mean ± SEM of at least triplicate samples. ns, not significant; *P < 0.05, ***P < 0.001, ****P < 0.0001, Student’s t test;. RLU, relative luminescence units; AU, arbitrary units.

We found that a number of CD-associated NOD2 mutant proteins (3020insC, R702W, L248R, A612T, A755V, and R1019stop) displayed a ≈70 to 90% reduction in S-palmitoylation levels (Fig. 6, B and C, and fig. S9, B and C). As expected, decreased palmitoylation strongly correlates with a more pronounced cytosolic distribution (Fig. 1C and fig. S9D). We thus surmise that a reduction in the S-palmitoylation of these NOD2 variants leads to membrane dissociation and hence an inability to respond to MDP, despite their nanomolar affinity for this compound (10). By contrast, the EOS-associated NOD2 gain-of-function mutant C495Y displayed a twofold increase in S-palmitoylation compared with that of WT NOD2 (Fig. 6, D and E). NOD2 C495Y palmitoylation and associated NF-κB hyperactivity could be counterbalanced by the use of 2BP (Fig. 6F). These data indicate a strong correlation between defects in NOD2 (and potentially NOD1) palmitoylation and several different human pathological conditions.

Discussion

We have established that S-palmitoylation of both NOD1 and NOD2 is required for optimal membrane targeting and hence for proper signaling upon cognate peptidoglycan detection. Membrane association of NOD1 and NOD2 is vital for their function (1, 7, 8, 16), as clearly illustrated by the CD-associated NOD2 loss-of-function variant 3020insC, which fails to associate with membranes and, as a consequence, is unable to detect bacterial intrusion (6). We provide evidence that NOD1 and NOD2 S-palmitoylation is required not only for steady-state membrane association but also for ligand-induced signaling. Accordingly, S-palmitoylation–deficient mutants of NOD1 and NOD2 are mislocalized and lose their ability to induce NF-κB signaling in response to C12-iE-DAP or MDP. Furthermore, we found that ZDHHC5 is required for S-palmitoylation of NOD1 and NOD2, and both the presence of ZDHHC5 at the site of bacterial entry and its enzymatic activity are necessary for proper recruitment of NOD1/2 to the bacterial entry site and phagosomes. How ZDHHC5 is recruited to the site of bacterial entry remains an open question.

We conclude that a palmitoylation-dependent local accumulation of NODs, and their subsequent exposure to bacterial by-products, ensures a compartmentalized and efficient signaling response. Consequently, cells can respond to very low concentrations of bacterial products. We propose that NOD1 and NOD2, together with ZDHHC5 and various transporters (such as SLC15A3), form specialized platforms for pathogen sensing. We anticipate that our findings will advance the understanding and ultimately the treatment of NOD-driven inflammatory diseases, including NOD-dependent autoimmune diseases and chronic bacterial infections.

Supplementary Materials

science.sciencemag.org/content/366/6464/460/suppl/DC1

Materials and Methods

Figs. S1 to S9

Tables S1 and S2

References (3247)

References and Notes

Acknowledgments: The authors thank S. Grinstein for advice with experimental design, valuable discussions, and critically reading the manuscript as well as P. Bilan for assistance with development of experimental methods and tools. We are also thankful for the technical support by the Core Facilities, Zhejiang University School of Medicine. Funding: This study was supported by grants from the Natural Science Foundation of China (NSFC) (31770938 and 91854113) and the Key Program of Zhejiang Provincial Natural Science Foundation of China (LZ16C050001) to D.N.; Q.S. is supported by grants from NSFC (31771525 and 91754113) and Ministry of Science and Technology of the People’s Republic of China (2017YFA0503402). G.D.F is supported by Canadian Institutes of Health Research (CIHR) (MOP-133656 and PJT-166010) and the J. P. Bickell Foundation. J.H.B. is supported by operating grants from the CIHR (FDN#154329) and holds the Pitblado Chair in Cell Biology. A.K. was supported by a Foundation Grant from CIHR (FND#143203). A.M.M. is funded by the CIHR and the Leona M. and Harry B. Helmsley Charitable Trust. Author contributions: D.N., Q.S., B.R., E.C., and G.D.F. designed the research; Y.Lu, Y.Z., E.C., C.Z., Y.Y., X.W., J.S.C, Y.M., and X.C. performed the experiments; D.N., Q.S., B.R., and G.D.F. analyzed the results; D.N., G.D.F., and Q.S. wrote the paper; Y.Li. generated hDMCs. B.R., J.H.B., S.E.G., G.D.F., A.K., N.W., and H.H. edited and commented on the manuscript. Competing interests: The authors declare no competing financial interests. Data and materials availability: Proteomic data have been deposited at massive.ucsd.edu (ID MSV000082079). All DNA constructs and the ZDHHC5 KO mouse model described in this study are available upon request. All other data needed to evaluate the conclusions in the paper are present in the paper or the supplementary materials.

Stay Connected to Science

Navigate This Article