NOD1 is an intracellular pattern-recognition receptor specific for Gram-negative peptidoglycan that is important in host response to infections (e.g. Helicobacter pylori and Shigella flexneri). Genetic variation in NOD1 predisposes to asthma and inflammatory bowel disease. Functional responses have not previously been studied in primary human cells. NOD1 activation by low nanomolar concentrations of the specific muropeptide ligand M-TriDAP induced minimal human peripheral blood mononuclear cell TNF-α, IL-1β or IL-10 secretion, but synergistically increased Toll-like receptor (TLR)-induced responses. Synergistic responses were seen across multiple ligands (to TLR1/2, 2/6, 4, 5, 7/8) and a broad range of cytokine secretion (TNF-α, IL-1α, IL-1β, IL-4, IL-6, IL-10, GM-CSF). Synergy was also observed in the allogeneic mixed lymphocyte reaction. These responses were similar in cells homozygous for Crohn's disease-associated NOD2 mutations. In contrast to cell lines, primary human peripheral blood mononuclear cells respond to NOD1 muropeptides at ∼ 100-fold lower concentrations. Cross-talk between cytosolic NOD1 and membrane-bound TLR enhances responses to the multiple antigens simultaneously presented by a microbe.
Innate immune responses to bacteria rely on sensing conserved pathogen-associated molecular patterns. Membrane-bound Toll-like receptors (TLR) are the most extensively studied. More recently the family of intracellular NOD proteins has also been shown to recognise bacterial products within the cytosol. The widely expressed NOD1 protein (also known as CARD4) senses a naturally occurring muropeptide breakdown product of bacterial peptidoglycan, GlcNAc-MurNAc-L-Ala-γ-D-Glu-meso-diaminopimelic acid (GM-TriDAP), found mostly in Gram-negative bacteria 1, 2. Studies using synthetic peptides have shown the minimal motif to be D-Glu-meso-DAP (referred to also as iE-DAP) with an exposed terminal DAP stem, neither amino acid being present in eukaryotes 2, 3. The NOD1 and NOD2 proteins are thought to be the sole sensors of peptidoglycan, earlier reports suggesting TLR2 might sense peptidoglycan were probably due to impure preparations 4.
NOD1 mediates epithelial cell signal transduction following infection with Shigella flexneri and enteroinvasive Escherichia coli5, 6, and in vivo NOD1-deficient mice show increased susceptibility to cag pathogenicity island positive Helicobacter pylori infection 7. Genetic variation in the NOD1 gene is associated with susceptibility to asthma and inflammatory bowel disease 8, 9. Studies using plasmid-transfected epithelial cell lines have suggested that NOD1 activates the pro-inflammatory transcription factor NF-κB via RIP2 and IκBα degradation 10. In mouse macrophages, NOD1 ligand stimulation was reported to induce TNF-α and IL-6 secretion, and NOD1 ligand combined with lipopolysaccharide (LPS) generated a larger IL-6 response than LPS alone, suggesting an additive effect 2. In cell lines, NOD1 overexpression has been reported to enhance caspase-mediated apoptosis, although this is controversial 10, 11, and enhance pro-IL-1β processing 11. How NOD1 might function in primary human cells is unknown: use of such cells enables study of signalling systems without artefact due to cell immortalisation, in systems expressing other relevant interacting and downstream signalling proteins and in the presence of other immunologically important pathways. Furthermore, data derived from mouse studies of the NOD protein family may not be applicable to humans: as suggested by the extreme polymorphism of the mouse NOD2 protein 12, species difference in response to NOD2 ligands 13, lack of intestinal phenotype in the NOD2 knockout mouse, and opposing results of NOD2 stimulation in different types of NOD2 gene-targeted mice 14–17.
Low M-TriDAP concentrations do not induce PBMC cytokine secretion, but strongly enhance responses to TLR1/2 and TLR4 ligands
Initial dose-response studies (Fig. 1) with nanomolar concentrations of the specific purified NOD1 ligand M-TriDAP induced no or minimal PBMC pro-inflammatory (TNF-α, IL-1β) or anti-inflammatory (IL-10) cytokine secretion as compared to Pam3CysLys4 or LPS (TLR1/2 and TLR4 ligands respectively, at sub-optimal cytokine inducing concentrations, optimisation data not shown). However, M-TriDAP markedly increased cytokine responses induced by Pam3CysLys4 and LPS (for IL-1β, 3.5- to 9.5- and 2.1- to 8.9-fold increase, respectively; for TNF-α, 1.1- to 2.0- and 1.7- to 5.7-fold increase; for IL-10, 1.4- to 2.1- and 1.5- to 4.2-fold increase, dose dependent and mostly maximal at 100 nM M-TriDAP, n = 3 healthy individuals).
M-TriDAP enhancement of TLR responses is observed at the level of multiple cytokines
Protein array analysis of PBMC responses to 20 nM M-TriDAP showed the strongly synergistic effect with Pam3CysLys4 or LPS on cytokine production, also occurred with IL-1α, IL-4, IL-6 and GM-CSF (Fig. 2). Both M-TriDAP and TLR ligands induced strong IL-8 secretion alone, without apparent synergy at the doses studied.
M-TriDAP does not synergize with NOD2 ligands, or act via NOD2
Contamination of the synthetic M-TriDAP preparation with other TLR ligands was considered unlikely due to the lack of pro-inflammatory cytokine production in PBMC and negative Limulus amebocyte lysate endotoxin assay. To exclude contamination by muramyl dipeptide-based NOD2 ligands, we studied NOD2 double mutant PBMC (from Crohn's disease patients). IL-1β responses to LPS alone (Fig. 3) were similar in both wild-type and NOD2 double-mutant cells (p = 0.2, unpaired t-test). In wild-type PBMC, addition of NOD1 and/or NOD2 ligands significantly increased IL-1β responses to LPS (ANOVA p < 0.0001, post hocp < 0.01 M-TriDAP, p < 0.01 MDP, p < 0.01 M-TriDAP + MDP). In NOD2 double-mutant PBMC, synergy with LPS was only observed in the presence of NOD1 ligand (ANOVA p = 0.0001, post hocp < 0.01 M-TriDAP, p > 0.05 MDP, p < 0.05 M-TriDAP + MDP). Similar results in NOD2 mutant cells were observed for Pam3CysLys4/ M-TriDAP synergy (data not shown). These data excluded NOD2 activation by MDP contamination of the M-TriDAP preparation as the source of the observed synergy. Furthermore, no evidence for a combined synergistic effect of NOD1 and NOD2 ligands with TLR ligands was observed in wild-type cells.
M-TriDAP enhances secretion of cytokines induced by multiple TLR ligands
To elucidate whether NOD1 activation was specific to certain TLR homo- or heterodimers we studied the effect of M-TriDAP combined with ligands specific to a broad range of TLR (Fig. 4). Responses to 20 nM M-TriDAP alone were again minimal (typically below assay detection limits) and indistinguishable from unstimulated cells. Marked synergistic increases in IL-1β production were seen for all five TLR ligands tested (p = 0.001 for each ligand), suggesting a broad synergistic effect of NOD1 on TLR induced cytokine production.
Synergy between LPS and M-TriDAP in the mixed lymphocyte reaction
We pre-stimulated PBMC with NOD1 and TLR ligands, washed out ligands, and observed a significant effect (ANOVA p < 0.0001) of cell pre-stimulation in the allogeneic mixed lymphocyte reaction (Fig. 5). Post-hoc analysis revealed increased proliferation with M-TriDAP (p < 0.05), higher with combined M-TriDAP + LPS (p < 0.01), although not significant (p > 0.05) with ultra-pure low-dose LPS alone (in contrast to the stronger responses observed with crude LPS typically used for MLR studies). Analysis of individual responses to LPS alone (corrected for media) versus M-TriDAP + LPS (corrected for M-TriDAP) suggested synergy (p = 0.03). These data suggest synergy between NOD1 and TLR responses extends to activation of antigen presentation.
Microbial peptidoglycan is sensed in human cells through the NOD1 and NOD2 pattern recognition receptor system 4. Muramyl dipeptide, a key component of Freund's adjuvant and a naturally occurring peptidoglycan breakdown product, has long been studied in immunology, although the specific receptor (NOD2) has only recently been identified 18. A synergistic interaction between muramyl dipeptide and lipopolysaccharide induced responses has been widely reported 19–21. Because NOD2 and NOD1 are structurally related, and with the characterisation of specific NOD1 ligands 2, 3, we aimed to study the possibility that interactions between TLR and NOD1 might also occur.
Using primary human cells, we observed strong enhancement of TLR induced cytokine responses by NOD1 activation. Although a small effect was detectable at 1 nM muropeptide concentrations, manifest synergy occurred around 10 nM (∼7 ng/mL M-TriDAP). Interestingly, these low doses of muropeptide did not stimulate cytokine production alone (with the exception of the chemokine IL-8). Synergistic effects were observed at the level of multiple cytokines and with multiple TLR ligands, suggesting a generalised effect of NOD1 activation on TLR/cytokine signalling. During preparation of the current manuscript, Uehara et al. 22 reported a synergistic effect of micromolar concentrations of NOD1-specific muropeptides on TLR-induced IL-8 production from THP1 monocyte cell lines. In contrast, we did not observe synergy in IL-8 production in primary human PBMC – possibly due to the high IL-8 production by unstimulated PBMC and saturation of PBMC IL-8 responses at low concentrations of TLR ligands. In our analysis of primary PBMC responses, synergy between NOD1 and TLR ligands was also observed when pre-stimulated cells were analysed in the mixed lymphocyte reaction. These data suggest that a principal effect of NOD1 stimulation, occurring at low ligand concentrations, may be to enhance innate immune responses through the TLR pathway.
Studies in transfected cell lines have suggested that NOD1 stimulation activates the NF-κB pro-inflammatory transcription factor system. Although we have not specifically studied NOD1 signalling in this report, it is interesting that in primary human PBMC low nanomolar concentrations of M-TriDAP did not have an effect on cytokine production alone but did enhance TLR-induced responses. RIP2 may provide a potential mechanism for the observed synergistic interactions, as it mediates both TLR and NOD1/2 protein signalling 23, 24. Recent data suggests that activation of the homologous NOD2 protein leads to ubiquitinylation of NEMO, a key component of the NF-κB signalling complex 25. However, NOD proteins have also been shown to interact with other proteins including TAK 26, CLAN 27, caspase-1 11 and stimulate the NALP3/Cryopyrin inflammasome 28. Our current understanding of the (perhaps multiple) signalling mechanisms of NOD1 is limited and further research is necessary, in particular the response to different ligand concentrations and TLR co-activation should be an aim of future studies.
It is interesting that all other studies of NOD1 function have used either micromolar concentrations of muropeptides or stimulated with muropeptides simultaneously with cell permeabilisation for plasmid transfection. The current data suggest that either NOD1 in primary PBMC is responding to a much lower concentration of intracellular ligand or that there is active intracellular muropeptide transport in these cells. Such transport mechanisms have been described in epithelial cells 29; these cells, as well as haematopoietic cells, express NOD1 and although we presume a similar interaction with TLR ligands occurs, confirmatory studies are necessary.
In vivo, microbial interactions with the innate immune system presumably involve triggering of multiple pattern recognition receptors, including the NOD and TLR families, by multiple antigens simultaneously presented by an entire organism. Our data suggest that – in humans - NOD1 may function to synergistically enhance innate immune responses triggered by TLR ligands rather than having a direct effect to induce an inflammatory response. Responses of individual target cells may be fine tuned by control of NOD1 and TLR expression levels before and after microbial exposure. Cross-talk between pattern recognition pathways may permit enhanced innate immune response against microorganisms.
Materials and methods
Peripheral blood mononuclear cells were prepared by density gradient centrifugation (Lymphoprep, Nycomed, Oslo, Norway) of heparinised venous blood. Blood samples were collected after written informed consent was given and with local research ethics committee approval. Cytokine studies were performed using 2 × 105 cryo-preserved PBMC cultured for 22 h in 250 μL media per well in 96-well plates. Cryo-preservation and liquid nitrogen storage of PBMC was according to a method yielding similar results in overnight assays to fresh cells 30 to permit parallel culture followed by ELISA assays of multiple conditions. In the allogeneic mixed lymphocyte reaction, 2 × 105 each of irradiated (60 Gy) stimulator and responder PBMC (both fresh) were cultured for 5 days in 200 μL media in 96 well plates (in triplicate for each condition) with addition of 2 μCi [3H]dThd 16 h prior to harvesting and beta-counting. Serum-free X-VIVO 15 media with gentamicin (50 μg/mL) was used for all experiments (Cambrex BioScience, Wokingham, UK).
Pattern recognition receptor ligands
Ligands were used in cell culture at the concentrations indicated: synthetic Pam3CysLys4 (Alexis Biochemicals, Nottingham, UK), TLR1/2 ligand, 10 ng/mL; 1 ng/mL; synthetic MALP-2 (Alexis), TLR2/6 ligand, 1 ng/mL; synthetic R-848 (resiquimod, Invivogen, San Diego, CA), TLR7/8 ligand, 500 ng/mL; purified flagellin from Salmonella typhimurium (Alexis), TLR5 ligand, and 100 ng/mL. In order to minimise the potential problem of peptidoglycan contamination of LPS, phenol and further gel-filtration chromatography purified LPS from E. coli 055:B5 (Sigma, Poole, UK L2637) was used at 1 ng/mL. A previous study had shown this LPS preparation to be a TLR4 ligand without NOD2-type peptidoglycan activity 31. Ultra-pure M-TriDAP, NOD1 ligand, was synthesised as described 1 and used at 1–100 nM. Synthetic pharmaceutical grade muramyl dipeptide (MDP-Lys18, Nopia/romurtide), NOD2 ligand, was used at 20 nM.
Cell culture supernatants were used at 1 part in 4 dilution (TNF-α, IL-1β or IL-10) in 96-well plate sandwich ELISA and protein array ELISA (Proteoplex 16-well human cytokine array kit, Merck, Darmstadt, Germany, according to manufacturer's instructions). TNF-α ELISA was performed as per manufacturer's instructions (Bender MedSystems, Vienna, Austria). IL-1β and IL-10 ELISA were performed using antibodies at 1/400 (AL-Immunotools, Friesoythe, Germany), Streptavidin-HRP (R&D Systems, Abingdon, UK) and TMB-H202 (BD Bioscience, Oxford, UK). Plates were read at 450 nm with 620 nm correction.
Data were analysed with paired t-tests or for multiple comparisons repeated measures ANOVA (with post-hoc Dunnett’s-test if appropriate). Wilcoxon signed rank test was used for analysis of the IL-1β response to multiple TLR ligands (Fig. 4) due to the presence of outlying data points. Prism 4 software was used for all analyses (GraphPad, San Diego, CA).
We thank all individuals who kindly contributed blood samples; Daiichi Pharmaceutical Company, Japan for the gift of MDP-Lys18; Profs. D. P. Jewell, A. Forbes and C. Mathew for recruitment and genotyping of Crohn's disease patients; Dr. D. Philpott for advice and endotoxin assays of M-TriDAP. The study was funded by a Wellcome Trust Clinician Scientist Fellowship (to DAvH) and the Hammersmith Hospitals Trustees Research Committee.