Pattern recognition receptors are central to the responsiveness of various eukaryotic cell types when they encounter pathogen-associated molecular patterns. IFN-γ is a cytokine that is elevated in humans and other animals with bacterial infection and enhances the LPS-induced production of antibacterial mediators by macrophages. Mice lacking the pattern recognition receptor, TLR4, respond very poorly to stimulation by LPS, but administration of IFN-γ has been described as restoring apparent sensitivity to this stimulatory ligand. In this study, we show that IFN-γ primes murine macrophages stimulated by crude LPS preparations to produce the antibacterial mediator NO, a proportion of which is independent of TLRs 2 and 4. This response is lost in tlr4−/− IFN-γ-primed murine macrophages when the LPS preparation is highly purified. NO is also induced if chemically synthesized muramyl dipeptide, an intermediate in the biosynthesis of peptidoglycan, is used to stimulate macrophages primed with IFN-γ. This is absolutely dependent on the presence of a functional nucleotide oligomerization domain-2 (NOD-2) protein. IFN-γ increases NOD-2 expression and dissociates this protein from the actin cytoskeleton within the cell. IFN-γ priming of macrophages therefore reveals a key proinflammatory role for NOD-2. This study also shows that the effect of IFN-γ in restoring inflammatory responses to Gram-negative bacteria or bacterial products in mice with defective TLR4 signaling is likely to be due to a response to peptidoglycan, not LPS.
Pattern recognition receptors (PRRs)2 are central to the induction of innate and acquired immunity via their roles in the recognition of pathogen-associated molecular patterns (PAMPs). Important classes of PRRs include the TLR and the nucleotide oligomerization domain (NOD) receptors (1, 2). TLRs are transmembrane proteins that respond to widely different PAMPs. TLR4 responds to LPS during infection with many different species of Gram-negative bacteria (1). TLR2 responds to a range of Gram-positive bacterial ligands such as lipoteichoic acid and bacterial lipoproteins (1, 2). NODs are primarily intracellular and only respond to peptidoglycan breakdown products, such as muramyl dipeptides (MDP) and muramyl tripeptides (MTP), and these receptors are important in the pathogenesis of intracellular pathogens (such as Shigella flexneri and Salmonella enterica) and in modulating Crohn’s disease (2).
Stimulation of macrophages via TLR4 results in the release of a wide range of cytokines and mediators. These include TNF-α through a signaling pathway linked to the adapter protein MyD88 (3) and NO through MyD88-dependent and MyD88-independent Toll/IL-1 receptor domain-containing adaptor inducing IFN-β (TRIF)/TRIF-related adaptor molecule-dependent signaling (4, 5). Stimulation via TLR2 induces MyD88-dependent signaling only, therefore generating a more limited inflammatory response, with TNF-α but not NO being induced (3, 4). Activation of NOD receptors also activates NF-κB-signaling pathways through the recruitment of receptor-interacting protein 2, caspase recruitment domain-containing serine/threonine kinase (also known as RICK, CARDIAK, CCK, and RipK2), but cytokine production in response to activation of these receptors is less clearly defined than for TLR activation (6, 7). Chaimaillard et al. (8) used synthetic g-d-glutamyl-meso-diaminopimelic acid to induce IL-6 and TNF-α by a NOD-1-dependent mechanism from bone marrow-derived macrophages (BMM). Published data showing significant detection of inflammatory mediators from BMM stimulated with compounds like MDP, however, is sparse, although this ligand stimulates bone marrow-derived dendritic cells to produce IL-6 and TNF-α (9).
C3H/HeJ mice have a mutated TLR4-signaling domain (10, 11, 12) and monocytes/macrophages from these mice are resistant to stimulation by LPS, but they can release TNF-α in response to crude LPS preparations if they are primed with the Th type 1 cytokine IFN-γ (13). IFN-γ enhances LPS-induced TNF-α and NO production from macrophages in vitro. Crude LPS preparations are now known to contain other PAMPs such as bacterial lipoproteins and peptidoglycan (14), and the IFN-γ-primed TNF-α release from C3H/HeJ cells has been assumed to be due to stimulation of TLR2 by these contaminating ligands. IFN-γ is produced during infection and can enhance the responses induced by many different PRRs, thus profoundly influencing the course and clinical outcome of infectious disease (15, 16). The mechanism underlying the phenomenon of IFN-γ priming has been a subject of contention for some time. Possible mechanisms to explain this phenomenon include increased surface expression of PRRs, an increased total number of PRRs, up-regulation of the PRR-related signaling proteins, cross-talk between the IFN-γ and PRR signal transduction pathways to enhance efficacy (17), or the production of an autocrine mediator in response to IFN-γ (18, 19). IFN-γ enhances gene transcription and surface expression of TLR4 (16), but also sensitizes the downstream signaling pathways, for example, enhancing STAT-1 signaling during macrophage activation by LPS (20, 21, 22). The effect of IFN-γ on other PRRs is less clearly defined although expression of many, including NODs, can be enhanced by this cytokine (23, 24).
The study of this enhancement phenomenon, and of PRRs in general, has been complicated by the presence of contaminating material in bacterial ligand preparations. A consequence of this has been the false definition of PRRs such as TLR2 and NOD-2 as LPS receptors (25, 26). In this study, we show that IFN-γ-primed BMM from wild-type, tlr4−/−, or tlr2−/− mice produce NO in response to MDP. In addition, a proportion of the NO produced in IFN-γ-primed BMM stimulated with commercial preparations of LPS is independent of TLR4 or TLR2 and this is lost with highly purified preparations of LPS. This TLR4-independent production of NO in IFN-γ-primed BMM is dependent on NOD-2 activation of the cells by the constituent units of peptidoglycan. This is the first demonstration of discrete production of the mediator NO, that has previously been shown to be produced primarily by activation of TLR4, from IFN-γ-primed macrophages in response to stimulation of NOD-2.
Materials and Methods
The tlr2−/−, tlr4−/−, and myd88−/− and their congenic wild-type control mice on a C57BL/6 background were a gift from Prof. S. Akira (Research Institute for Microbial Diseases, Osaka University, Osaka, Japan). C3H/HeN and C3H/HeJ mice were obtained from Harlan U.K. The generation and characterization of the Card15−/− (nod-2−/−) and the congenic wild-type control mice is described by Pauleau and Murray (27). These nod-2−/− mice are phenotypically similar to those generated and described by Kobayashi et al. (9). Primary BMM were isolated from the femur and tibia of mice killed by cervical dislocation and cultured in RPMI 1640, 10% FCS, 5% horse serum, 2 mM l-glutamine, 1 mM sodium pyruvate, 10 μg/ml gentamicin, and 20% supernatant taken from L929 cells (28). Before the experiments, cells were plated onto 96-well plates at a plating density of 2 × 105/well. RAW cells were grown in RPMI 1640 containing 10% FCS, 2 mM lEscherichia coli 0157; Sigma-Aldrich), highly purified LPS (29
Synthesis of peptidoglycan intermediates
The NOD ligands used here were synthesized in two stages: uridine 5′diphosphoryl N-acetyl muramyl (UDPmurNac)-l-alanyl-d-glutamate and UDPmurNac-l-alanyl-d-glutamyl-meso 2,6-diaminopimelate were synthesized from UDP-N-acetyl glucosamine, using recombinant Pseudomonas aeruginosa MurA, MurB, MurC, and MurD enzymes for the former peptide and MurA, MurB, MurC, MurD, and MurE for the latter diaminopimelate-containing peptide and purified essentially as described before (30, 31). The purity of the final products were assessed by analytical FPLC, by anion exchange using MonoQ, and by enzymatic analysis with MurE in the case of UDPmurNac-l-alanyl-d-glutamate and MurF in the case of UDPmurNac-l-alanyl-d-glutamyl-meso-2,6-diaminopimelate. By both criteria, the purity of UDPmurNac-l-alanyl-d-glutamate and UDPmurNac-l-alanyl-d-glutamyl-meso-2,6-diaminopimelate were (fast protein liquid chromatography analysis, enzymatic analysis) 99.0, 98.0, and 94.3, 98.3%, respectively. Electrospray mass spectral analysis (negative ion) of the two UDPmurNac peptides provided confirmation of synthesis. The observed (expected) m/z for doubly charged UDPmurNac l-alanyl-d-glutamate was 438.58 (438.59) and for doubly charged UDPmurNac-l-alanyl-d-glutamyl-meso 2,6-diaminopimelate was 524.63 (524.60).
In the second stage, the NOD ligands UDPmurNac l-alanyl-d-glutamate and UDPmurNac-l-alanyl-d-glutamyl-meso 2,6-diaminopimelate were released by acid hydrolysis from their corresponding UDP precursors in 0.1 M HCl at 100°C for 30 min (32). Hydrolyzed products were shown to be entirely free of the corresponding UDPmurNac peptide precursor by coupled enzymatic analysis for UDPmurNac l-alanyl-d-glutamate with MurE and for UDPmurNac-l-alanyl-d-glutamyl-meso 2,6-diaminopimelate with P. aeruginosa MurF. Further, negative ion electrospray mass spectrometric analysis of the hydrolyzed UDP murNac peptides revealed no remaining UDP murNac species (data not shown). Peptides were then analyzed by positive ion electrospray/mass spectral yielding (expected/observed for singly charged cation) 516.50/516.18 and 666.26/666.28 for murNac l-alanyl d-glutamate (+Na+) and the murNac l-alanyl-d-glutamyl-2,6 diaminopimelate, respectively. The other products of the hydrolysis (UMP, UDP, and inorganic phosphate) were assayed spectrophotometrically in coupled enzymatic reactions, furnishing control solutions to be used in experiments with the murNac peptides. To determine whether there was any LPS contamination of the peptidoglycan ligand, a Limulus amebocyte lysate assay (Charles River Laboratories) was performed following the manufacturer’s instructions. The levels of LPS were MTP control solution <0.0183 ng/ml; MTP <0.0025 ng/ml; UDP-MTP <0.0008 ng/ml; MDP control solution <0.0008 ng/ml; MDP <0.0008 ng/ml; UDP-MDP <0.0008 ng/ml. None of these concentrations of LPS have any effect on our primary BMM in the presence or absence of IFN-γ.
Assays for NO and TNF-α
To determine NO synthase activity the supernatant of the cultured cells was removed 24 h after addition of ligand and assayed for nitrite accumulation by the Griess reaction as an indication of inducible NO synthase (iNOS) activity (33
Immunostaining for fluorescence microscopy
RAW cells were grown as a monolayer on sterile coverslips within 6-well plates, at a concentration of 1 × 106 cells/well. One hundred milliliters of blocking solution (10% normal goat serum (DakoCytomation) in TBS (pH 7.6) containing 0.02% saponin; Sigma-Aldrich) was applied to each 22 × 22-mm coverslip for 10 min at room temperature. These were then drained and 100 ml of the NOD-2 polyclonal Ab (34) (Caymen Chemical) was added to the cells at a dilution of 1/100 in blocking solution. This was left to incubate overnight at 4°C. Two 30-min washes of the cells were conducted by filling each well with a large volume of TBS. The TBS was removed and 100 ml of Pacific Blue-conjugated goat anti-rabbit Ab (Molecular Probes P-10994) were added at a dilution of 1/100 in blocking solution and left to incubate at room temperature for 30 min. After two further 15-min washes, Phalloidin-Alexa (9) (Molecular Probes A-12381) was then added at a concentration of 1/200 in TBS, at room temperature for 20 min, to stain actin filaments. After two 15-min washes, the coverslips were removed from the wells, drained, and placed cell side down onto 20 ml of Vectashield with 4′,6′-diamidino-2-phenylindole (Dapi; Vector Laboratories). Images were taken on a Leica DM6000B Fluorescent microscope running FW4000 acquisition software and processed using Adobe Photoshop 6.0 software.
PRR activation by peptidoglycan contaminants is an important component of IFN-γ enhancement of LPS-induced NO production
We compared the stimulatory effects of commercially available LPS with those of highly purified LPS (29) in IFN-γ-primed BMM isolated from wild-type, tlr4−/−, and tlr2−/− C57BL/6 mice by measuring the production of TNF-α and NO. IFN-γ enhanced TNF-α and NO responses to both unpurified and highly purified LPS when wild-type BMM were stimulated. Unpurified LPS induced TNF-α in tlr4−/− BMM and both, TNF-α and NO, in IFN-γ-primed TLR4−/− BMM. The production TNF-α and NO were both lost when tlr4−/− BMM were stimulated with highly purified LPS (Fig. 1⇓a). These results were also confirmed in BMM from C3H/HeN (TLR4 wild-type) and C3H/HeJ (TLR4P712H) mice (Fig. 1⇓b). These data indicate the presence of a TLR2 ligand, probably bacterial lipoprotein, which stimulates TNF-α production via TLR2 in the tlr4−/− BMM, contaminating commercial LPS. These results also show the presence of another biologically significant contaminating ligand in these preparations that induces NO but not TNF-α, which may be peptidoglycan.
Synthetic MDP stimulates NO but not TNF-α production from IFN-γ-stimulated primary BMMs
Using a panel of synthetic peptidoglycan intermediates (0.1–10 μg/ml; for structures, see Fig. 2⇓a), we stimulated primary BMM cells isolated from C57BL/6 mice and measured TNF-α and NO production with or without priming of the cells with IFN-γ. Production of TNF-α was very low in response to all of the peptidoglycan intermediates whether or not the cells were primed with IFN-γ. Concentrations of the peptidoglycan intermediates above 10 μg/ml did not result in TNF-α production. Administration of MDP and MTP (UDP-modified and unmodified) to the cells (in LipofectAMINE transfection reagent or in solution) induced NO production from IFN-γ-primed BMM, whereas peptide mixtures of the synthetic precursors of the dipeptides and tripeptides had no effect on these cells (Fig. 2⇓b). The ligands used are synthetic and highly purified and therefore do not contain minimal levels of contaminating bacterial products. TNF-α was not induced in response to any of the peptide ligands. Activation of TLRs leads to the production of TNF-α. As we could measure NO, but not TNF-α, production in response to the peptide ligands, this suggests that the induction of NO is independent of TLRs. To confirm this observation BMMs isolated from tlr4−/−, tlr2−/−, and cd11b−/− mutant mice were stimulated with these ligands and the production of TNF-α and NO was determined. None of the experiments resulted in TNF-α production. NO production was independent of TLR4, TLR2 (Fig. 3⇓), and CD11b (data not shown), a protein that acts in concert with TLR4 and CD14 to modify PAMP signaling (35). These results show that specific synthetic peptidoglycan intermediates stimulate NO, but not TNF-α, production in IFN-γ-primed BMMs independently of TLR signaling.
The production of NO from IFN-γ-stimulated primary murine BMMs is dependent on NOD-2
MTP-diaminopimelate is produced by Gram-negative bacteria whereas MDP is synthesized by all bacteria (36). NOD-2 is expressed in macrophages and responds to both MDP and MTP (37). We stimulated IFN-γ-primed BMM from mutant mice lacking NOD-2 with MDP and measured NO production. All IFN-γ-primed NO production was lost in BMM from card15−/− mice indicating a requirement for NOD-2 in this process (Fig. 4⇓). Interestingly, the response of BMM lacking NOD-2 primed with IFN-γ to commercial LPS was significantly reduced (Fig. 4⇓). This confirms that the peptidoglycan contaminant, by activating NOD-2, is an important inducer of NO production in commercial LPS preparations. Pauleau and Murray (27) showed that NOD-2 does not play an essential role in IFN-γ signaling per se and, as this receptor has only been previously described primarily as an intracellular protein, it is unlikely that IFN-γ stimulation will induce expression of NOD-2 on the macrophage cell surface. The concentrations of IFN-γ we used here did not induce NO production when administered alone (Fig. 4⇓) and our use of synthetic ligands precludes the problems of contamination with other PRR ligands. Disruption of the actin cytoskeleton up-regulates cytokine-induced NO synthase (iNOS) expression (38) and it is possible that alteration of the localization of NOD-2 within the cell maybe a mechanism by which IFN-γ primes the cells to produce NO in response to MDP. To determine the effect of IFN-γ on NOD-2 localization, we performed immunohistochemistry studies on RAW 264.7 murine macrophage-like cells which respond in the same way as BMMs to IFN-γ priming and NOD-2 ligands (data not shown).
In untreated cells, the expression of NOD-2 was limited to a perinuclear localization with limited cytosolic distribution. This distribution is very similar to NOD-2 localization in epithelial cells (39). In cells treated with IFN-γ, NOD-2 expression was enhanced and redistributed throughout the cytoplasm with limited foci of high intensity staining (Fig. 5⇓), possibly indicative of receptor clustering. NOD-2 colocalizes with actin in the perinuclear region of the cell, but when IFN-γ is added, NOD-2 is dissociated from the actin cytoskeleton. When both IFN-γ and MDP are added to the cells, actin clusters form, colocalized with NOD-2. These data suggest that IFN-γ primes cells by causing redistribution of NOD-2 receptors within the cell, and that when MDP is added to the cells, NOD-2 again redistributes to associate with actin clusters. By altering the association of NOD-2 with the cytoskeleton its ability to associate with signaling proteins may also be altered and this may be permissive for MDP activation of iNOS expression. The combination of the IFN-γ sensitization of iNOS transcription pathways and the altered cytoskeletal association of NOD-2 may explain how MDP is able to stimulate NO production under these conditions.
NO is an important part of the host’s defense against bacterial infection. Previously, enhanced NO production by macrophages has been uniquely associated with the activation of TLR4 by LPS, although a TLR-independent component of the macrophage response to infection with live Mycobacterium tuberculosis (through IFN-α and IFN-β production and STAT1 activation) induces iNOS by an unknown mechanism (40). The research reported here shows that NO can be produced by macrophages in response to MDP, but only after stimulation of the cells by IFN-γ and that this is absolutely dependent on NOD-2. Activation of NOD-2 results in its redistribution within the cell and this may contribute to the mechanism by which this receptor acts. This is the first evidence linking a receptor such as NOD-2 with production of NO and it is likely that NOD-2 activity is important in modulating antibacterial responses. Evidence is now emerging that suggests that both NOD-1 and NOD-2 are required for the induction of intracellular bacterial killing and the data presented here, given the postulated bactericidal role for NO, suggest a possible mechanism for this process.
MDP-induced production of NO was only observed when cells were primed with IFN-γ. IFN-γ has previously been linked to enhancing both NOD-2 expression and TLR4-dependent production of NO, but until this study there has been no evidence to link IFN-γ and NOD-2 signaling to the induction of inflammatory mediators. This study also provides a possible mechanism for an earlier observation whereby NO was produced from J774 macrophage-like cells in response to MDP and IFN-γ (41). NOD receptors associate with receptor-interacting protein 2, a protein that also links to TLR4 signaling (42), to activate IκB kinase γ leading to phosphorylation of IκBα and hence activation of NF-κB (6, 7). Induction of NO requires activation of other transcription factors in addition to NF-κB, and activation of IFN regulatory factor 1 by IFN-γ increases iNOS promoter activity and expression in response to LPS (43). In our experiments with BMM, NO was not produced in response to IFN-γ and our use of synthetic ligands reduces any problems of contamination with other PRR ligands such as LPS. It is possible, therefore, that in the presence of IFN-γ priming, MDP-induced NOD-2 activation of RIP2 is sufficient to induce NO production. The mechanism by which IFN-γ modifies ligand activation of NOD-2 is further complicated by the intracellular location of this family of receptors (44). It is unclear exactly how NOD-2 will bind to MDP and in our study, in common with other investigators, we used high concentrations of MDP to initiate activation after priming with IFN-γ. Pauleau and Murray (27) showed that NOD-2 does not play an essential role in IFN-γ signaling per se and, as this receptor is an intracellular protein, it is unlikely that IFN-γ will induce expression of NOD-2 on the macrophage cell surface. Other investigators have noted that IFN-γ enhances the expression of NOD-2 (24), an observation we also see in our experiments, but this does not necessarily explain how MDP can induce NO through this receptor. Our immunohistochemistry study showed striking changes in the cellular localization of NOD-2 after IFN-γ priming of cells which were enhanced if MDP and IFN-γ were added together. Our data show that IFN-γ induces the actin cytoskeleton and NOD-2 to dissociate, conditions permissive for the induction of iNOS (38). NOD-2 associates with, and its activity is regulated by, Erbin, a protein that modifies cytoskeletal functions such as cell polarization (45). There is, therefore, evidence to support the concept that changes in the cellular distribution of NOD-2 will affect its proinflammatory signaling. The combination of IFN-γ sensitization of iNOS transcription pathways and altered cytoskeletal association of NOD-2 may explain how MDP is able to stimulate NO production under these conditions.
When we started this work, we expected that stimulation of BMM with MDP would probably induce TNF-α production as this is a mediator most commonly associated with PRR activation. IFN-γ priming of cells enhances the production of a number of inflammatory mediators including TNF-α (20), and we therefore expected that IFN-γ priming of BMMs would induce TNF-α production in response to MDP. We were therefore surprised that we did not observe TNF-α production in response to MDP by cells primed with IFN-γ. TNF-α production has been observed in other studies in response to MDP, but only when this synthetic compound is given in combination with TLR ligands, suggesting that TLRs and NOD-2 can act in synergy (9). In comparison to wild-type mice, NOD-2-deficient mice primed with MDP are resistant when subsequently challenged with highly purified LPS suggesting that the priming effects of MDP in vivo are lost in Card15−/− mice (9). NOD-2-deficient mice are also partially resistant to challenge in vivo with impure LPS (probably contaminated with cell wall fragments including MDP) such that the synergy effect is lost in the absence of NOD-2 function (27). NOD-2 can also be a negative regulator of TLR2-driven TH-1 responses (34) and, given these contradictory data, it may be, particularly in vivo, that NOD-2 acts as a complex modulator of the immune response. The true functional role of this receptor may therefore be to modulate the infectious disease process after TLR4 has been activated to produce cytokines such as IFN-γ. The production of IFN-γ will enhance PRR-driven production of antibacterial mediators such as NO (46) to clear bacterial infection. Indeed, evidence exists to suggest that NOD-2 has a direct, but as yet unexplained, role in intracellular killing of Salmonella (47). NOD-2 may, therefore, fulfil a similar role to that hypothesized for NOD-1.
We thank Prof. S. Akira for the TLR2−/− and TLR4−/− mice, Professor U. Zähringer for the ultra-pure LPS preparation, and Profs. R. C. Levesque and A. E. Zoeiby for the overexpression vectors required to synthesize the Mur enzymes used in ligand synthesis in this paper.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 Address correspondence and reprint requests to Dr. Clare Bryant, Department of Veterinary Medicine, University of Cambridge, Madingley Road, Cambridge CB3 0ES, U.K. E-mail address:
↵2 Abbreviations used in this paper: PRR, pattern recognition receptor; PAMP, pathogen-associated molecular pattern; NOD, nucleotide oligomerization domain; MDP, muramyl dipeptide; MTP, muramyl tripeptide; BMM, bone marrow-derived macrophage; Dapi, 4′,6′-diamidino-2-phenylindole; iNOS, inducible NO synthase.
- Received September 15, 2005.
- Accepted January 31, 2006.
- Copyright © 2006 by The American Association of Immunologists