Muramyl dipeptide activation of nucleotide-binding oligomerization domain 2 protects mice from experimental colitis (original) (raw)
Administration of MDP protects mice from TNBS colitis. We previously reported that simultaneous stimulation of murine APCs with PGN and MDP led to a reduction of IL-12 production induced by PGN stimulation of APCs via TLR2 (11) and that this inhibitory effect of NOD2 on TLR2 signaling plays an important role in the prevention of colonic inflammation driven by microbial antigens (18). Furthermore, we recently found that NOD2-transgenic mice expressing NOD2 under the control of class II promoter or administered a plasmid expressing NOD2 manifest decreased susceptibility to TNBS colitis (21). These findings led us in the present study to determine whether administration of MDP protects mice from experimental colitis.
In the first form of experimental colitis we studied, hapten-induced colitis induced by administration of TNBS, C57BL/10 mice were administered MDP (100 μg i.p.) or PBS (i.p.) for 3 consecutive days (days –3 to –1) prior to intrarectal injection of 3.75 mg of TNBS in 45% ethanol (day 0). As shown by the body weight curves depicted in Figure 1A, MDP administration prior to the TNBS challenge protected mice from the loss of weight normally seen during the development of TNBS colitis. Moreover, as shown in Figure 1B, whereas mice that were administered PBS prior to intrarectal TNBS instillation exhibited destruction of crypt architecture and infiltration of mononuclear cells in the colonic lamina propria (LP) in tissue examined on day 4 after challenge with TNBS, mice that were administered MDP prior to intrarectal TNBS instillation exhibited little epithelial damage or cellular infiltration. As shown in Figure 1C, this was confirmed by the colitis scores. Thus, these data provide strong evidence that MDP administration prior to TNBS challenge inhibits the development of TNBS colitis.
Systemic administration of MDP prevents the development of TNBS colitis. C57BL/10 mice were administered MDP or PBS (i.p.; see Methods) on days –3, –2, and –1 and then challenged with intrarectal TNBS on day 0. (A) Changes in body weight in mice treated with PBS or MDP (n = 10) and challenged with intrarectal administration of TNBS. **P < 0.01 compared with mice treated with PBS. (B) H&E-stained colonic tissue of mice harvested on day 4. Histology of PBS-treated mice showed massive infiltration of mononuclear cells as well as destruction of crypt architecture (top row); histology of MDP-treated mice showed almost normal colonic tissue with minimal infiltration of mononuclear cells (bottom row). Original magnification, ×100. (C) Histology score of the colonic tissue of the mice harvested on day 4.
We next determined whether this prevention of colitis development was associated with reduced inflammatory cytokine responses to microbial antigens, i.e., TLR ligands and nucleotide-binding oligomerization domain–like receptor (NLR) ligands. For this purpose, mesenteric LNs (MLNs) and LP cells isolated from mice pretreated with MDP or PBS before TNBS challenge were stimulated with a broad range of TLR ligands. As shown in Figure 2, A and B, MLN cells and colonic LP cells obtained from mice administered MDP and then stimulated with a broad range of TLR ligands displayed markedly reduced production of IL-12p40, IL-12p70, IL-6, and TNF upon stimulation with TLR2, TLR3, TLR4, TLR5, and TLR9 ligands as compared with cells obtained from mice administered PBS before TNBS challenge. Thus, prior activation of NOD2 by MDP administration caused suppression of multiple TLR pathways, not just TLR2. Furthermore, these reduced innate cytokine responses were associated with reduced adaptive Th1 responses since, as shown in Figure 2C, MLN and colonic LP cells from MDP-treated mice exhibited markedly reduced IFN-γ production upon anti-CD3 stimulation. We next determined whether reduced cytokine responses were due to the downregulation of NF-κB activation. Accordingly, we isolated nuclear proteins from MLN cells and assessed such activation by both EMSA and a semiquantitative method based on the binding of the extract to NF-κB consensus sequences, followed by detection of bound components with subunit-specific Abs (NF-κB ELISA study) (11). As shown in Figure 2, D and E, NF-κB activation by LPS or PGN stimulation was suppressed in MLN cells from MDP-treated mice as compared with those from PBS-treated mice. Taken together, these data suggest that MDP treatment protects against TNBS colitis by inhibiting responses to a wide range of TLR ligands.
MDP administration reduces the TLR-induced cytokine responses of MLN and colonic LP cells from mice with TNBS colitis. (A) Colon LP lymphocytes (1 × 106/ml) isolated from the mice on day 3 were stimulated with PGN, Pam3CSK4, LPS, and CpG in the presence of IFN-γ (20 ng/ml); cultured supernatants were collected at 48 hours and analyzed for cytokine production. *P < 0.05; **P < 0.01 when supernatants from MDP-treated mice were compared with supernatants from PBS-treated mice. (B) MLN cells (1 × 106/ml) isolated from mice with TNBS colitis on day 3 were stimulated with anti-CD3 (1 μg/ml) and a broad range of TLR ligands. Cultured supernatants were collected at 48 hours and analyzed for cytokine production. *P < 0.05; **P < 0.01 when supernatants of MDP-treated mice were compared with supernatants of PBS-treated mice. (C) MLN cells and colon LP lymphocytes isolated from the mice on day 3 were stimulated with anti-CD3 (1 μg/ml); cultured supernatants were collected at 48 hours and analyzed for IFN-γ production. **P < 0.01 when supernatants from MDP-treated mice were compared with supernatants from PBS-treated mice. (D) Evaluation of NF-κB activation in MLN cells isolated from the mice on day 3 and stimulated with LPS or PGN. Nuclear extracts were prepared from MLN cells isolated from PBS- or MDP-treated mice and stimulated with LPS or PGN for 2 hours and then subjected to EMSA. (E) Nuclear extracts obtained in D assayed for p65 and c-Rel activation using NF-κB transcription factor ELISA. *P < 0.05; **P < 0.01 compared with nuclear extracts from PBS-treated mice.
Administration of MDP protects mice from DSS colitis. In studies of a second experimental colitis, DSS colitis, control NOD2-intact or NOD2-deficient mice treated with 4% DSS in the drinking water from day 0 to day 5 to induce colitis were administered either MDP (100 μg i.p.) or PBS (i.p.) for 3 consecutive days at the early phase of the colitis (days 0, 1, 2). As shown in Figure 3A, NOD2-intact mice with DSS colitis treated with PBS but not those treated with MDP exhibited significant body weight loss during the observation period. Furthermore, this protective effect was mediated by NOD2 activation, since NOD2-deficient mice with DSS colitis exhibited comparable body weight loss whether they were treated with PBS or MDP. As shown in Figure 3B, these body weight data correlated with serum amyloid A levels as well as with the colitis scores of the NOD2-intact mice treated with MDP and PBS, and again, no difference was seen in these parameters in NOD2-deficient mice treated with MDP or PBS. Thus, these studies provide strong evidence supporting the view that MDP administration protects mice from the development of DSS colitis and complement the data above on TNBS colitis.
MDP administration prevents DSS colitis. NOD2-intact (NOD2+/+) and NOD2-deficient (NOD2–/–) mice were treated with drinking water containing 4% DSS for 6 days (days 0–5). At the early phase of colitis induction (days 0, 1, 2), mice were administered MDP (i.p.) or PBS every day. (A) Changes of body weight in PBS- or MDP-injected mice. *P < 0.05; **P < 0.01, time point values of NOD2-intact mice administered MDP compared with NOD2-intact mice administered PBS. (B) Serum amyloid A (SAA) levels and colitis scores of mice (see Methods) on day 7. SAA levels were determined by ELISA. *P < 0.05, NOD2-intact mice administered MDP compared with mice administered PBS.
In studies of the mechanism of such protection, we analyzed the profiles of cytokine production of mice described above that received MDP administration at the onset of DSS challenge. For this purpose, MLN cells obtained from mice on day 5 after the initial challenge with DSS were stimulated with several TLR ligands. As shown in Figure 4A, MDP administration was associated with a reduction in TLR2-, TLR3-, and TLR4-mediated IL-12p40 and IFN-γ production by MLN cells obtained from NOD2-intact but not NOD2-deficient mice. However, this reduction of IL-12p40 or IFN-γ production seen in NOD2-intact mice treated with MDP was not observed in MLN cells from NOD2-deficient mice treated with MDP. Finally, as shown in EMSA and NF-κB ELISA studies in Figure 4, B and C, respectively, PGN or LPS-mediated NF-κB activation was greatly suppressed in MLN cells from NOD2-intact mice treated with MDP as compared with those not so treated. Thus, as in the case of TNBS colitis, MDP pretreatment protects mice from DSS-induced colitis through downregulation of proinflammatory cytokine responses evoked by a number of TLR ligands, not just TLR2.
Cytokine production by MLN cells in mice treated with DSS. (A) MLN cells (1 × 106/ml) isolated from NOD2+/+ and NOD2–/– mice on day 5 were stimulated with a broad range of TLR ligands. Cultured supernatants were collected at 48 hours and analyzed for cytokine production by ELISA. *P < 0.05 when supernatants were compared with supernatants from mice treated with PBS (white bars). (B) Activation of NF-κB in MLN cells isolated from NOD2+/+ and NOD2–/– mice on day 5 following stimulation with LPS or PGN. Nuclear extracts were prepared from MLN cells isolated from PBS- or MDP-treated mice and stimulated with LPS or PGN for 2 hours and then subjected to EMSA. (C) Nuclear extracts obtained in B assayed for p65 and c-Rel activation using NF-κB transcription factor ELISA. *P < 0.05 compared with nuclear extracts from PBS-treated mice.
MDP administration does not protect from DSS colitis in mice with abnormal NOD2 arising from a CARD15 frame-shift mutation. The data above show that MDP activation of NOD2 protects mice from TNBS or DSS colitis by downregulating multiple TLR pathways. To relate these findings to Crohn disease, we next determined whether NOD2 arising from a CARD15 mutation associated with Crohn disease has a similar protective function. To this end, we determined the ability of MDP to protect NOD2-deficient mice reconstituted with plasmids expressing murine intact CARD15, frame-shift CARD15 (L980fs) equivalent to human 3020insC, or control “empty” plasmid from the development of DSS colitis. Accordingly, NOD2-deficient mice were treated with 5.5% DSS in drinking water from day 0 to day 5 to induce DSS colitis. As in previous studies, the mice were administered MDP (100 μg i.p.) for 3 consecutive days beginning at the time of colitis induction (days 0, 1, 2), but in this case each MDP dose was accompanied by i.p. administration of either intact CARD15, frameshift CARD15, or control empty vector encapsulated in hemagglutinating virus of Japan (HVJ) for efficient in vivo delivery (21). As shown by the weight curves in Figure 5A and the histologic data in Figure 5B, while MDP injection protected NOD2-deficient mice reconstituted with intact CARD15 plasmid, it did not protect NOD2-deficient mice reconstituted with frame-shift CARD15 plasmid or with control empty plasmid. Colitis scores correlated with the tissue histology shown in Figure 5B (data not shown). Thus, these data suggest that abnormal NOD2 arising from the Crohn disease–associated CARD15 frame-shift mutation lacks the ability to control colonic inflammation upon systemic administration of MDP.
DSS colitis in MDP-treated NOD2-deficient mice reconstituted with intact or frameshift NOD2. NOD2-deficient (NOD2–/–) mice were treated with drinking water containing 5.5% DSS for 6 days (days 0–5). At an early phase of colitis induction (days 0, 1, 2), mice were administered MDP and HVJ-encapsulated plasmid (see Methods). (A) Changes in body weight in MDP-administered NOD2-deficient mice reconstituted with intact NOD2, frameshift NOD2, or control empty vector. Weights of MDP-administered NOD2-deficient mice given DSS are shown as a control. **P < 0.01, time point values of intact-NOD2–reconstituted mice compared with control empty vector–reconstituted mice. (B) H&E-stained colonic tissue of the mice harvested on day 7. Original magnification, ×50.
Human DCs subjected to preactivation of NOD2 by MDP exhibit reduced proinflammatory cytokine production upon subsequent stimulation with TLR ligands. Since MDP administration inhibits the inflammation occurring in murine models of colitis, it was of interest to determine the conditions under which MDP stimulation could also inhibit inflammatory cytokine responses to multiple TLR ligands in vitro.
In the relevant studies, we determined the effect of MDP prestimulation of cells on the assumption that such prestimulation recapitulated the pretreatment of mice in the above colitis models. In initial studies, we preincubated human monocyte–derived DCs with medium (absence of NOD2 prestimulation) or with MDP (presence of NOD2 prestimulation) for 24 hours prior to stimulation with TLR ligands alone or stimulation with TLR ligands plus MDP (MDP costimulation). As shown in Figure 6A, in keeping with prior results, in the absence of NOD2 prestimulation, PGN-mediated IL-12p40 and IL-6 production was inhibited by NOD2 costimulation, whereas double-stranded RNA (dsRNA), LPS, or CpG-mediated IL-12p40, IL-6, and TNF production was variably enhanced by NOD2 costimulation. In contrast, in the presence of MDP prestimulation, production of proinflammatory cytokines and chemokines such as IL-12p40, IL-6, and CXCL10 stimulated with a wide range of TLR ligands was inhibited and there was reversal of enhancement by NOD2 costimulation. The effect of NOD2 prestimulation and costimulation on TNF production was somewhat different, since here NOD2 costimulation in the absence of NOD2 prestimulation was either associated with no inhibition or enhancement of TLR stimulation; nevertheless, in this case as well, such enhancement was usually reversed by NOD2 prestimulation. Thus, consistent with results obtained from studies of in vivo colitis models, MDP pretreatment has a remarkable inhibitory effect on multiple types of TLR responses.
Human and mouse DCs prestimulated with MDP exhibit reduced cytokine and chemokine production when stimulated with TLR ligands. (A) Human monocyte–derived DCs (1 × 106/ml) from 6 healthy donors were preincubated with MDP or medium for 24 hours and then stimulated with a broad range of TLR ligands alone or in combination with MDP for an additional 24 hours. Cultured supernatants were collected at 24 hours and analyzed for cytokine and chemokine production by ELISA. *P < 0.05; **P < 0.01 compared with supernatants from DCs preincubated with medium and stimulated with TLR ligands alone (light blue bars). (B) CD11c+ DCs (1 × 106/ml) derived from bone marrow cells from NOD2-intact (NOD2+/+) and NOD2-deficient (NOD2–/–) mice were preincubated with MDP (50 μg/ml) or medium alone for 24 hours and stimulated with a broad range of TLR ligands. Cultured supernatants were collected at 24 hours and analyzed for cytokine production by ELISA. *P < 0.05; **P < 0.01 when supernatants were compared with NOD2-intact DCs preincubated with medium and stimulated with TLR ligands (light blue bars). (C) OVA323-339 peptide–specific CD4+ T cells (OT-II) were purified from the spleens of OT-II transgenic mice; OT-II cells (1 × 106/ml) were cocultured with NOD2-intact or NOD2-deficient BMDCs (2 × 106/ml) in the presence of a broad range of TLR ligands and OVA peptide (0.5 μM); cultured supernatants were collected at 72 hours and analyzed for IFN-γ production by ELISA. *P < 0.05; **P < 0.01 compared with supernatants from NOD2-intact DCs preincubated with medium and stimulated with TLR ligands (light blue bars).
It is unlikely that the above inhibition of TLR responses by NOD2 prestimulation is due to induction of either cell “exhaustion” as defined previously (22) or simply to cell death. Thus, as shown in Figure 6A, NOD2 pretreatment had no inhibitory effect on either PGN- or LPS-induced TNF production and, as shown in Supplemental Figure 1A (supplemental material available online with this article; doi:10.1172/JCI33145DS1), no effect on CXCL8 (IL-8) production. In addition, as shown in Supplemental Figure 1B, pretreatment had no effect on CD40 ligand–induced IL-12p40 or IL-6 production. As shown in Supplemental Figure 1C, pretreatment led to enhanced expression of cell-activation markers without a marked increase in apoptotic cells. Finally, this reduction of proinflammatory cytokine production by NOD2 prestimulation was unlikely to be due to counterregulation by IL-10 since, as shown in Supplemental Figure 1A, IL-10 production was also reduced in DCs subjected to NOD2 prestimulation.
Taken together, these data show that NOD2 prestimulation reduces subsequent TLR-mediated induction of proinflammatory cytokines and chemokines even in situations where simultaneous TLR and NOD2 stimulation are associated with enhancing effects of NOD2 costimulation. In further proof of this conclusion, as shown in Supplemental Figure 2, DCs stimulated by killed E. coli organisms and by multiple TLR ligands simultaneously also exhibited reduced IL-12p40 responses upon NOD2 prestimulation.
Murine bone marrow–derived DCs also exhibit reduced TLR-mediated cytokine responses upon NOD2 preactivation. To both confirm and expand the above findings, we next determined the effect of NOD2 prestimulation on bone marrow–derived DCs (BMDCs) generated from NOD2-deficient mice. In these studies, BMDCs preincubated with MDP for 24 hours were subjected to stimulation with a broad range of TLR ligands. As shown in Figure 6B, NOD2 prestimulation led to a substantial reduction of IL-12p40 and IL-6 production by NOD2-intact BMDCs stimulated with TLR2, TLR4, TLR5, and TLR9 ligands. In contrast, this inhibitory effect was not observed with NOD2-deficient BMDCs. As shown in Supplemental Figure 3A, a similar effect of MDP pretreatment was observed with IL-12p70 and TNF production. Finally, as shown in Supplemental Figure 3B, MDP stimulation did not affect BMDC expression of costimulatory molecules or MHC class II expression. Similarly, MDP treatment did not change BMDC expression of TLR2 and TLR4 (data not shown).
In further studies, we asked whether this reduction of proinflammatory cytokine responses by NOD2 prestimulation results in decreased cytokine responses by naive OVA-specific CD4+ T cells. For this purpose, we determined the ability of DCs subjected to NOD2 prestimulation to induce OVA-specific T cells to produce IFN-γ upon coculture with OVA peptide and naive T cells from OVA–T cell receptor (OT-II) transgenic mice. As shown in Figure 6C, OVA peptide presentation by NOD2-intact NOD2-prestimulated BMDCs as compared with nonprestimulated BMDCs led to greatly reduced IFN-γ production by CD4+ T cells isolated from the spleens of OT-II transgenic mice in the presence of TLR2, TLR4, TLR5, and TLR9 ligands. In contrast, this reduced response was not observed with MDP-prestimulated NOD2-deficient BMDCs. These studies of mouse DCs thus confirm the inhibitory effect of NOD2 prestimulation observed with human DCs and provide evidence that the effect is in fact mediated by NOD2.
Inhibition of TLR cytokine responses by NOD2 prestimulation is associated with upregulation of IRF4. The fact that we could inhibit multiple TLR responses with NOD2 prestimulation allowed us to explore the molecular basis of the inhibitory effect. Initially, we assessed the effect of MDP prestimulation on subsequent TLR activation of NF-κB in human DCs by performing gel-shift assays on nuclear extracts isolated from the stimulated DCs. As shown in Figure 7A, nuclear extracts isolated from MDP-prestimulated DCs subsequently stimulated with LPS, PGN, or flagellin gave rise to bands of reduced intensity when incubated with 32P-labeled oligo probes specific to NF-κB as compared with extracts from cells not subjected to prestimulation. These data suggest that NOD2 prestimulation reduces subsequent activation of NF-κB by TLR ligands.
NOD2 stimulation is associated with upregulation of IRF4 expression. (A) NF-κB activation in human monocyte–derived DCs. DCs were preincubated with MDP or medium for 24 hours and then stimulated with LPS, PGN, or flagellin for 2 hours; nuclear extracts of the cells were then obtained and subjected to gel-shift assays; results shown are representative of those obtained with 2 healthy donors. (B) Upregulation of IRF4 in MDP-stimulated human monocyte–derived DCs. Whole-cell extracts prepared from DCs incubated with MDP, LPS, or medium for 24 hours were immunoblotted with Abs against the indicated components; results shown are representative of those obtained in 3 healthy donors. (C) IRF4 expression in monocyte-derived DCs transfected with IRF4 siRNA. DCs were transfected with 2 μg of IRF4 siRNA, IRAK-M siRNA, or control siRNA using the Amaxa nucleofection method; 16 hours after transfection, DCs were stimulated with MDP, LPS, or medium for 24 hours, at which point whole-cell extracts were prepared and subjected to immunoblotting with Abs against the indicated components. (D) Effects of IRF4 or IRAK-M siRNA transfection on cytokine production by human monocyte–derived DCs. DCs (2 × 107/ml) from 6 healthy donors were transfected with IRF4 siRNA, IRAK-M siRNA, or control siRNA, as described in C. After 24 hours of culture with MDP, LPS, or medium, DCs were stimulated with PGN, Pam3CSK4, or LPS for another 24 hours; cultured supernatants were assayed for IL-12p40 by ELISA. *P < 0.05; **P < 0.01 compared with DCs transfected with control siRNA and preincubated with medium (white bars).
In further studies, we performed immunoblot analyses to determine the expression of signaling molecules involved in the TLR signaling pathway or those previously shown to be negative regulators of that pathway (19, 20, 23) in cells incubated in medium versus cells incubated with MDP or LPS. Because of the results of the gel-shift assay described above, we focused on those molecules involved in the activation of NF-κB. As shown in Figure 7B, there was no difference in the expression of TLR signaling molecules such as MyD88, TAK1, TRAF6, IRF3, IRF5, or IKK-γ in cells without and with either NOD2 (MDP) or TLR4 (LPS) stimulation. In contrast, MDP-stimulated cells expressed increased amounts of the negative regulator IRF4 at 24 hours after stimulation (and, as shown in Supplemental Figure 4A, at 36 hours as well) but not other negative regulators such as IL-1 receptor–associated kinase M (IRAK-M) or SOCS1. Consistent with previous reports (19, 20, 23, 24), cells stimulated by LPS expressed increased amounts of both IRF4 and IRAK-M but not SOCS1.
Inhibition of TLR cytokine responses by NOD2 prestimulation requires the expression of IRF4. To determine whether NOD2 prestimulation is not only associated with IRF4 expression but actually requires IRF4 expression, we initially asked whether gene silencing of IRF4 or IRAK-M expression by siRNA specific for these molecules affects NOD2- or TLR4-mediated inhibitory effects. As shown in Figure 7C, transfection of human DCs with a mixture of IRF4 siRNA substantially reduced expression of IRF4 at the protein level in MDP- and LPS-stimulated human DCs. Similarly, transfection of IRAK-M siRNA reduced expression of IRAK-M in LPS-stimulated cells. As shown in Figure 7D, transfection of IRF4 siRNA but not IRAK-M siRNA led to increased IL-12p40 production in MDP-prestimulated DCs subsequently stimulated with PGN, N-palmitoyl(S)-[2,3-bis(palmitoyloxy)-(2Rs)-propyl]Cys-Ser-Lys4 (Pam3CSK4), and LPS. In addition, as shown in Figure 7D, transfection of IRAK-M siRNA but not IRF4 siRNA led to increased IL-12p40 production in LPS-preincubated DCs stimulated with PGN, Pam3CSK4, and LPS. In a final study along these lines, we determined whether individual IRF4-specific siRNAs could abolish inhibition of NOD2 prestimulation to rule out off-target effects of the siRNA mixture. As seen in the studies shown in Supplemental Figure 4B, addition of 2 different siRNAs again led to loss of the inhibition of NOD2 prestimulation in both PGN- and LPS-stimulated cell cultures.
In a second approach to studying the role of IRF4 in inhibition following NOD2 prestimulation, we examined the capacity of such prestimulation to inhibit responses in cells naturally lacking IRF4, such as human monocytes (25). In these studies, we determined whether NOD2 prestimulation of THP1 cells, a monocytic cell line, affects their subsequent TLR-induced cytokine production. As shown previously, THP1 cells express NOD2 (26, 27) but, as shown in Figure 8A, do not express IRF4 before or after stimulation by MDP or LPS; in contrast, they express IRAK-M, especially after stimulation with LPS. As shown in Figure 8B, in keeping with the data shown above, MDP prestimulation of THP1 cells did not result in reduced IL-12p40 or TNF responses following subsequent stimulation with PGN, Pam3CSK4, or LPS, whereas LPS prestimulation did lead to reduced IL-12p40 or TNF production by THP1 cells subsequently stimulated with PGN or LPS. Finally, as shown in Supplemental Figure 5, transfection of THP1 cells with an IRF4-expressing vector led to downregulation of PGN, Pam3CSK4, and LPS responses, which was particularly evident after MDP prestimulation; thus, the lack of response to MDP prestimulation could in fact be shown to be due to lack of IRF4 expression.
Mechanism of NOD2-induced IRF4 inhibition of TLR signaling. (A) Whole-cell extracts were prepared from THP1 cells stimulated with MDP or LPS for 24 hours and then immunoblotted with Abs against IRF4, IRAK-M, and actin. (B) THP1 cells (5 × 105/ml) were prestimulated with MDP, LPS, or medium for 24 hours and stimulated with TLR ligands; cultured supernatants were collected at 24 hours and analyzed for cytokine production by ELISA. *P < 0.05 compared with the concentrations of cytokines by cells preincubated with medium and stimulated with TLR ligands (white bars). (C) Physical interactions between IRF4 and RICK, MyD88, and TRAF6. Whole-cell extracts of HEK293 cells transfected with vectors (2 μg) expressing FLAG-tagged human IRF4 and HA-tagged human MyD88 or with untagged RICK, TRAF6, or TRAF2 were immunoprecipitated with anti-FLAG–conjugated beads and then immunoblotted with anti-HA Abs or with anti-RICK, -TRAF6, or -TRAF2. (D) Negative regulation of NF-κB by IRF4. HT-29 cells (1 × 105/96-well plate) transfected with pNF-κB–Luc (50 ng) and pSV–β-galactosidase (10 ng) were cotransfected with vectors expressing human RICK (200 ng), human MyD88 (200 ng), or human TRAF6 (200 ng) with or without an IRF4-expressing vector (50 ng, 200 ng, 1000 ng). *P < 0.05; **P < 0.01 compared with cells without IRF4 transfection (white bars). (E) Physical interaction of IRF4 and RICK in MDP-prestimulated human DCs. DCs were cultured with MDP or medium for 24 hours and then stimulated with Pam3CSK4 for an additional hour; whole-cell extracts were prepared and then immunoprecipitated with anti-IRF4 Abs and immunoblotted with anti-RICK Abs.
Along similar lines, we determined whether NOD2 prestimulation affected TLR-induced CXCL8 or CXCL10 production in HT-29 cells (i.e., a highly differentiated epithelial cell line), which, like THP1 cells, do not express IRF4. Here again, NOD2 prestimulation did not downregulate responses to subsequent stimulation with dsRNA or flagellin stimulation (data not shown).
The above in vitro studies involving gene silencing of IRF4 and the study of cells naturally lacking IRF4 provide strong support for the view that the inhibitory effects of NOD2 prestimulation depend on IRF4. Further support for this view comes from in vivo studies described below. It should be noted, however, that while IRF4 is necessary for NOD2-mediated inhibition, it is not sufficient for such inhibition. This follows from the fact that, as we have seen, LPS induces IRF4 expression but IRF4 does not mediate LPS-mediated inhibition. A further confirmation of this point came from studies depicted in Supplemental Figure 6A, showing that human DC stimulation with the NOD1 ligand γ-d-glutamyl-diaminopimelic acid (γDGDAP) also induced IRF4 expression (albeit less than that induced by the NOD2 ligand), yet, as shown in Supplemental Figure 6B, prestimulation of cells with the NOD1 ligand did not lead to inhibition of subsequent TLR responses.
Sensitivity of PGN and LPS signaling to IRF4 inhibition. The studies described above provide data that suggest that PGN and LPS responses differed in their sensitivity to IRF4-mediated inhibition; NOD2 prestimulation produced more profound inhibition of PGN (TLR2) responses than LPS (TLR4) responses, and suppression of PGN-mediated IL-12p40 production was more resistant to IRF4 siRNA–induced reversal than that of LPS-mediated IL-12p40 production (Figure 7D). To examine this possibility more directly, we measured PGN and LPS responses in human DCs transfected with increasing doses of FLAG-tagged IRF4 cDNA (28). As shown in Supplemental Figure 7, transfection of FLAG-tagged IRF4 induces protein expression in a dose-dependent manner. In addition, as also shown in Supplemental Figure 7, low expression of transfected IRF4 (0.2 μg) is sufficient to greatly inhibit PGN-induced production of IL-12p40 and IL-6, whereas in contrast, high expression of IRF4 is necessary to comparably inhibit LPS-induced production of these cytokines. Thus, PGN-mediated TLR2 signaling is more sensitive to negative regulation by IRF4 than LPS-mediated TLR4 signaling. Since IRF4 levels can be assumed to be low in the absence of prestimulation, i.e., when cells are stimulated by TLR ligands and MDP simultaneously, these findings suggest that the reason simultaneous stimulation of TLR ligands and MDP results in reduced PGN-induced production of IL-12p40 while it has little effect on IL-12p40 production by other TLR ligands is that with simultaneous stimulation the level of IRF4 is sufficient to suppress PGN responses but not responses to other TLR ligands (11). However, further in vivo studies will be necessary to substantiate this possibility.
Mechanisms of NOD2-induced IRF4 inhibition of TLR signaling. In further studies, we explored the mechanisms underlying NOD2-induced inhibition of TLR signaling. In initial studies, we determined possible physical interactions between IRF4 and various components of the TLR signaling pathway. To this end, we performed immunoblots on extracts of HEK293 cells cotransfected with FLAG-tagged human IRF4 cDNA (28) together with either MyD88, TRAF6, or RICK cDNA, all relevant components of the TLR and NOD2 signaling pathways. In addition, we cotransfected the FLAG-tagged human IRF4 cDNA with TRAF2 cDNA, a component of the TNF signaling pathway, as a negative control. As shown in Figure 8C, we found that each of these components except for TRAF2 does indeed bind to IRF4. In further studies, we cotransfected plasmids expressing MyD88, TRAF6, RICK, and IRF4 into HT-29 colon epithelial cells expressing an NF-κB luciferase reporter to determine the capacity of IRF4 to inhibit the capacity of each of the components to activate the reporter and generate luciferase. As shown in Figure 8D, cotransfection of IRF4 led to a dose-dependent reduction in NF-κB activation by each of the components, whereas IRF4 had no effect on the ability of TNF, a non–TLR/NLR-related NF-κB activator to activate NF-κB. This correlated with the fact that TNF signals mainly through TRAF2, shown above not to interact with IRF4. In additional studies along these lines, we buttressed these overexpression studies with a study of IRF4 binding to RICK in stimulated human DCs (i.e., cells in which IRF4 and signaling components were not overexpressed). As shown in Figure 8E, an immunoblot of extracts of Pam3CSK4-stimulated cells prestimulated with MDP or medium alone provided evidence that in MDP-prestimulated cells, IRF4 binds to RICK. These data offer evidence that IRF4 inhibition of TLR responses may involve, in part, binding to a key component of the TLR signaling pathway, in line with previous studies of IRF4 inhibition (19, 20). However, further details of the inhibitory mechanism await more complete studies of the functional effects of such binding on NF-κB activation.
Systemic administration of MDP prevents the development of TNBS or DSS colitis by upregulating IRF4 expression. The in vitro studies described above provided considerable evidence that the suppression of multiple TLR pathways mediated by MDP activation of NOD2 depends upon the expression of IRF4. To verify this hypothesis in vivo, we returned to the studies of the effect of MDP pretreatment on the development of experimental colitis to determine whether IRF4 also mediated inhibitory MDP effects in these models.
We first asked whether systemic injection of MDP induces upregulation of IRF4 as in the case of human DCs. For this, CD11b+ myeloid cells were isolated from MLNs and spleens of mice that had been administered either MDP or PBS. As shown in Figure 9A, IRF4 protein was barely detectable in cells from PBS-treated mice but was easily detected in cells from MDP-treated mice. We next determined the effect of IRF4 gene silencing on the capacity of MDP administration to protect mice from TNBS colitis. In these studies, mice administered MDP (i.p.) according to the schedule described in Figure 1 were also administered 100 μg of either siRNA targeting murine IRF4 or control siRNA encapsulated in HVJ by the intrarectal route on day –2 to day 1 (with respect to TNBS administration). As shown in Figure 9A, induction of IRF4 expression was not seen in CD11b+ MLN cells examined on day 0 obtained from mice administered both MDP and HVJ-IRF4 siRNA, whereas IRF4 expression was clearly seen in cells from mice administered MDP and HVJ-control siRNA. As shown in Figure 9B, while mice administered MDP and control siRNA did not exhibit body weight loss following TNBS challenge, mice administered MDP and IRF4 siRNA exhibited a body weight loss similar to that of mice administered PBS. As shown in Figure 9C, these weight loss data correlated with pathology: mice administered MDP and IRF4 siRNA exhibited severe colonic epithelial cell damage and massive infiltration of inflammatory cells in the LP equivalent to the changes seen in mice treated with PBS. Finally, as shown in Figure 9D, the weight loss data correlated with cytokine production findings in that the marked reduction in TLR-mediated IL-12p40, IL-6, and TNF accompanying MDP administration was almost completely restored by IRF4 siRNA administration. Taken together, these data show that MDP administration protects mice from TNBS colitis via its ability to induce NOD2 induction of IRF4 and subsequent IRF4 inhibition of TLR-mediated inflammatory cytokine production.
Systemic administration of MDP prevents the development of TNBS colitis by upregulating IRF4 expression. Mice administered intrarectal TNBS on day 0 were injected with MDP or PBS i.p. on days –3, –2, and –1 and also administered 100 μg of HVJ-encapsulated control siRNA or IRF4 siRNA by intrarectal instillation on days –2, –1, 0, and 1. (A) IRF4 expression in CD11b+ myeloid cells from MLNs and spleens from mice on day 0 (top); IRF4 expression in whole-cell extracts of CD11b+ myeloid cells from MLNs isolated from mice treated with IRF4 siRNA on day 0 (bottom). (B) Changes of body weight in mice treated with MDP and siRNAs. **P < 0.01 compared with body weight of mice treated with PBS. (C) H&E-stained colonic tissue of mice harvested on day 4. Histology of PBS-treated mice and IRF4 siRNA–treated mice showed massive infiltration of mononuclear cells as well as destruction of crypt architecture; histology of control siRNA-treated mice showed almost normal colon tissue with minimal infiltration of mononuclear cells. Original magnification, ×100. (D) MLN cells (1 × 106/ml) isolated from mice on day 4 were stimulated with a broad range of TLR ligands; cultured supernatants were collected at 48 hours and analyzed for cytokine production by ELISA. *P < 0.05; **P < 0.01 compared with the concentrations of cytokines from PBS-treated mice (white bars).
Finally, we asked whether the protection from DSS colitis in mice treated with MDP depends upon IRF4, in this case utilizing a well-described gene-targeted mouse with a background strain susceptible to DSS colitis (29). Indeed, immunoblot studies of extracts of spleen cells of IRF4-deficient mice with DSS colitis did not reveal an IRF4 band, whereas IRF4-intact mice did reveal a band (data not shown). As shown in Figure 10A, as judged by body weight loss, MDP administration as described above did not inhibit the development of DSS colitis in IRF4-deficient mice, whereas it did inhibit such colitis in IRF4-intact mice. In addition, as shown in Figure 10B, the colitis scores of MDP-treated and untreated IRF4-deficient mice were not significantly different, whereas those of MDP-treated IRF4-intact mice were significantly reduced as compared with untreated IRF4-intact mice. Finally, as shown in Figure 10C, whereas MLN cells from MDP-treated IRF4-intact mice exhibited a clear decrease in both IL-12p40 and IFN-γ secretion upon TLR ligand stimulation, comparable cells from IRF4-deficient mice exhibited no such decrease.
IRF4 signaling is necessary for the suppression of DSS colitis. IRF4-intact (IRF4+/+) and IRF4-deficient (IRF4–/–) mice were treated with 5% DSS in drinking water for 6 days (days 0–5). At an early phase of colitis induction (days 0, 1, 2), mice were administered MDP or PBS (i.p.). (A) Weight curves of IRF4–/– or IRF+/+ mice administered MDP or PBS. **P < 0.01 compared with PBS-injected IRF4+/+ mice. (B) Histology score of IRF4–/– mice treated with PBS or MDP on day 7. (C) MLN cells (1 × 106/ml) isolated from IRF4+/+ and IRF4–/– mice on day 7 were stimulated with PGN, LPS, or CpG; cultured supernatants were collected at 48 hours and analyzed for cytokine production by ELISA.