The Kinase Activity of Rip2 Determines Its Stability and Consequently Nod1- and Nod2-mediated Immune Responses

Abstract

Rip2 (RICK, CARD3) has been identified as a key effector molecule downstream of the pattern recognition receptors, Nod1 and Nod2; however, its mechanism of action remains to be elucidated. In particular, it is unclear whether its kinase activity is required for signaling or for maintaining protein stability. We have investigated the expression level of different retrovirally expressed kinase-dead Rip2 mutants and the role of Rip2 kinase activity in the signaling events that follow Nod1 and Nod2 stimulation. We show that in primary cells expressing kinase-inactive Rip2, protein levels were severely compromised, and stability could not be reconstituted by the addition of a phospho-mimetic mutation in its autophosphorylation site. Consequently, inflammatory cytokine production in response to Nod1 and Nod2 ligands was abrogated both in vitro and in vivo in the absence of Rip2 kinase activity. Our results highlight the central role that Rip2 kinase activity plays in conferring stability to the protein and thus in the preservation of Nod1- and Nod2-mediated innate immune responses.

A key step in the initiation of effector immune responses is the recognition of highly conserved molecules expressed by microbial pathogens. The immune system has developed specific receptors that sense these so-called pathogen-associated molecular patterns and initiate appropriate immune responses. One key family of pattern recognition receptors is the Nod-like receptor (NLR)² family (1–3), of which two members, Nod1 and Nod2, have been implicated in the recognition of bacterial peptidoglycan derivatives released into the cytosol upon bacterial infection (4–6). Several studies have shown that Nod1 plays a role in host defense against invasive pathogens such as Helicobacter pylori and Escherichia coli (7, 8), and Nod2 mutations have been associated with a higher incidence of Crohn disease (9, 10), thus highlighting these NLRs as important regulators of inflammatory immune responses.

Rip2, also called CARD3, RICK, or CARDIAK, is a serine/threonine kinase, which was implicated in the induction of NF-κB activation and apoptosis (11–13). Rip2 has been described to be critical for responses against Toll-like receptor ligands such as LPS (14, 15), although findings from recent studies did not support this conclusion (16). Rip2 contains a caspase-recruitment domain (CARD), which mediates interaction with other CARD-containing proteins such as Nod1 and Nod2, in addition to an N-terminal kinase domain and an intermediate domain. Nod1 and Nod2 associate with Rip2 upon peptidoglycan ligation (17) leading to downstream signaling events that culminate in NF-κB and mitogen-activated protein kinase activation (15, 18–20). Recent reports have suggested that the mitogen-activated protein kinase kinase kinase family member TAK1 provides the link between Rip2 and NF-κB activation upon Nod1 and Nod2 stimulation (21–23). However, the exact role of Rip2 and in particular its kinase activity in mediating downstream effector activation in NLR signaling still remains unclear. Notably, in vitro investigations have suggested that Rip2 kinase activity may be dispensable for the induction of immune responses initiated by NLR-ligands (21, 24, 25) and that disruption of Rip2 kinase activity is associated with a loss in protein stability (23); however, such studies utilized protein overexpression in cell lines and are yet to be tested in primary cells or in vivo.

In the current investigation we sought to elucidate the role of Rip2 kinase activity in transducing inflammatory signals upon NLR stimulation in vitro and in vivo. To this end, we utilized both Rip2 knock-out (15) and Rip2 kinase-dead knock-in mice (24) in addition to Rip2 deficient primary cells that were retrovirally reconstituted with different kinase-inactive mutants. We show here that in the absence of intact kinase activity, Rip2 protein is not stable and that insertion of a phospho-mimetic mutation is not sufficient to restore stability. Moreover, pharmacological abrogation of Rip2 kinase activity in primary cells similarly leads to destabilization of the molecule. As a consequence, signaling downstream of Nod1 and Nod2 and inflammatory cytokine production is impaired both in vivo and in vitro. Our results highlight Rip2 kinase activity as a central regulator of protein stability and consequently innate immune responses triggered by Nod1 and Nod2 ligands.

EXPERIMENTAL PROCEDURES

Mice

C57BL/6 wild type mice were obtained from Charles River Breeding Laboratories. Rip2-deficient mice were kindly provided by Prof. R. Flavell (Yale University, New Haven, CT), and Rip2 K47A mice (24) were backcrossed >7 times onto a C57BL/6 background and bred in the BioSupport animal facility. The mice were maintained specific pathogen-free at BioSupport (Zürich, Switzerland) in isolated ventilated cages. The animals used in the experiments were between 8 and 10 weeks of age. All of the experiments were performed with permission from the Zürich Animal Ethics Committee.

Reagents

Cell culture LPS from E. coli 0111:B4 was purchased from Sigma-Aldrich. MDP and ultrapure LPS from E. coli 0111:B4 for in vivo injection were purchased from Invivogen. The synthetic Nod1 ligand FK565 (26) was supplied by Astellas Pharma Inc. (Osaka, Japan). The kinase inhibitors SB203580 and BIRB0786 were purchased from Axon Medchem BV.

Stimulation of BM-DCs and BM-DMs

Bone marrow-derived dendritic cells were generated from bone marrow cells cultured for 10 days in complete RPMI medium supplemented with granulocyte macrophage-colony-stimulating factor. At day 10, the cells were incubated with LPS, MDP or FK565. For intracellular cytokine staining, the cells were activated for 6 h, and 10 μg/ml brefeldin A was added to the cultures for the final 3 h. Alternatively, DCs were activated overnight, and supernatant was collected for ELISA. For kinase inhibition studies, BM-DCs were incubated for 24 h with 10 μm SB203580 or 0.1 μm BIRB0796 and then lysed for Western blot. Bone marrow-derived macrophages (BM-DMs) were prepared from bone marrow cells cultured for 7 days in complete RPMI medium supplemented with 10% L929 supernatant containing macrophage-stimulating factor.

Cytokine and Chemokine Detection

For detection of intracellular cytokine production, BM-DCs were stained with biotin-conjugated anti-CD11c mAb (BD Biosciences), PercP-labeled streptavidin (BD Bioscience) and, after fixation, with fluorescein isothiocyanate-labeled anti-tumor necrosis factor-α mAb, phosphatidylethanolamine-labeled anti-IL-6 mAb, and allophycocyanin-labeled anti-IL-12p40 mAb (all from eBioscience). The cells were washed and analyzed by flow cytometry (FACSCalibur®; Becton Dickinson) and FlowJo software (Tree Star, Inc.). For cytokine detection in supernatants from activated BM-DCs, alkaline phosphatase-dependent ELISA was performed. Anti-IL-6 and anti-IL-12p40 mAb were purchased from eBioscience. For chemokine and Rip2 mRNA detection, total RNA was isolated from activated BM-DMs and subjected to reverse transcription using SuperScript III RT (Invitrogen). Quantitative real time reverse transcription-PCR was performed using Brilliant SYBR Green (Stratagene) on an iCycler (Bio-Rad). Expression was normalized to the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase.

Endotoxic Shock

Naïve C57BL/6, Rip2 knock-out, and Rip2 K47A knock-in mice were primed by intraperitoneal injection of MDP (500 μg) and challenged by intraperitoneal LPS injection (250 μg) 24 h after priming and monitored for general condition and survival. Blood samples were collected in EDTA before MDP immunization and 1.5 h after LPS challenge. Serum was analyzed for cytokine levels by ELISA.

Western Blot

Monoclonal antibody against Rip2 (clone Nick-1) was purchased from Santa Cruz Biotechnology. Antibodies against the phosphorylated form of p38, ERK, JNK, and IκB-α were purchased from Cell Signaling. Polyclonal antibody against β-actin was purchased from Santa Cruz Biotechnology. Stimulated cells were lysed in Triton X-100 buffer containing protease inhibitors. The lysates were loaded on SDS-PAGE gels, transferred to nitrocellulose membranes, and immunoblotted with the primary antibodies followed by incubation with horseradish peroxidase-conjugated secondary antibodies (Southern Biotechnology). Proteins were detected by ECL.

Retrovirally Mediated Gene Complementation in Bone Marrow Cells

cDNA encoding wild type or mutated forms of Rip2 were cloned into pMY-IRES-GFP retroviral vectors (27), which were then used to transfect Phoenix packaging cells (28). Bone marrow cells were isolated from Rip2-deficient mice previously injected with 5′-fluorouracil (29) and infected with supernatant from the retrovirally transfected packaging cell line. Infected, stem cell-enriched bone marrow cells were then cultured in the presence of IL-6, IL-3, and stem cell factor. After 6 days of culture, infected GFP-expressing cells were sorted using a FACSVantage SE (Becton Dickinson).

RESULTS AND DISCUSSION

Rip2 Kinase Activity Is Required for Inflammatory Cytokine Production upon Nod1 and Nod2 Stimulation

Rip2 has been reported to regulate downstream signaling of NLRs; however, the importance of its kinase activity in primary cells and in vivo is unknown. We first sought to establish the requirements for Rip2 and its kinase activity in the signaling events following Nod1 and Nod2 stimulation in primary cells in vitro. Accordingly, we generated BM-DCs from naive C57BL/6, Rip2-deficient mice, and Rip2 kinase-inactive knock-in mice (Rip2 K47A). In the latter, the residue crucial for ATP binding, lysine 47, is mutated to alanine. Wild type, Rip2 knock-out, and knock-in dendritic cells were then stimulated with the Nod1 and Nod2 ligands, FK565 and MDP, respectively. Inflammatory cytokine production by the activated BM-DCs was determined after 6 h by intracellular staining and flow cytometry (Fig. 1A) or after 24 h in culture supernatant by ELISA (Fig. 1B, left panel). Although MDP and FK565 alone poorly induced cytokine secretion, cooperative induction of IL-6, IL-12p40, and tumor necrosis factor-α was observed in wild type cells when these ligands were administered together with LPS. Cells from Rip2-deficient mice did not respond to MDP or FK565 alone, nor in combination with LPS, confirming the importance of Rip2 in Nod1 and Nod2 responses. Notably, Rip2 kinase-inactive BM-DCs exhibited a similarly impaired cytokine response upon stimulation with MDP and FK565, indicating that an intact kinase activity was required for Nod1- and Nod2-induced cytokine responses.

FIGURE 1.

Inflammatory responses following Nod1 and Nod2 stimulation are abrogated in the absence of Rip2 kinase activity. A, BM-DCs were generated from C57BL/6, Rip2 knock-out, and Rip2 K47A knock-in mice and stimulated with 25 μg/ml MDP or 3 μg/ml FK565 alone or with the addition of 10 ng/ml LPS. As a control, the cells were also activated with 10 ng/ml LPS alone. After 6 h, tumor necrosis factor-α and IL-12p40 production was assessed by intracellular cytokine staining and flow cytometry. The numbers in FACS plots represent percentages of positive cells in each quadrant. The data are representative of 3–4 repeat experiments. B, BM-DCs and BM-DMs were stimulated with 25 μg/ml MDP or 3 μg/ml FK565 alone or with the addition of 10 ng/ml LPS. Left panel, IL-6 production was measured in the supernatant by ELISA after 24 h. The results represent cytokine production by BM-DCs from three different mice per group and are presented as the means ± S.D. Right panel, total RNA was isolated from BM-DMs after 6 h, and KC expression was analyzed by quantitative real time PCR. The results are represented as fold increase in chemokine expression compared with nonstimulated controls. The experiment was repeated 2–3 times. C, C57BL/6, Rip2 knock-out, and Rip2 K47A mice were primed intraperitoneally with 500 μg of MDP and challenged with ultrapure LPS (250 μg). Alternatively, the mice were primed with PBS. Serum was collected from naïve animals and 1.5 h after LPS challenge; IL-6 and IL-12p40 levels were measured by ELISA. *, p < 0.05; statistically significant differences. D, survival of wild type, Rip2 knock-out, and Rip2 K47A mice after LPS challenge. The experiment was repeated twice with eight mice/group.

In addition to inflammatory cytokines, microbial products also induce chemokine secretion. In particular, stimulation with a synthetic Nod1 ligand has been shown to induce chemokines such as KC (also known as CXCL1) in vitro and in vivo (30, 31). Our aim was to determine whether Rip2 kinase activity was involved in chemokine production upon NLR stimulation. We thus generated bone marrow-derived macrophages from C57BL/6, Rip2 knock-out, and Rip2 knock-in mice and stimulated them with MDP and FK565 alone or together with LPS. After 6 h, total RNA was isolated and reverse-transcribed, and KC mRNA expression was determined by quantitative real time PCR analysis (Fig. 1B, right panel). There was a 10–30-fold increase in KC mRNA expression when wild type cells were stimulated with FK565 or MDP together with LPS. Comparatively, stimulation of both Rip2 deficient and kinase-dead cells did not lead to KC mRNA transcription. These results are in line with the cytokine data and highlight a role for Rip2 kinase activity in inflammatory responses following stimulation with bacterial products. To test our in vitro results, we investigated whether the abrogation of Rip2 kinase activity also resulted in impaired NLR-mediated inflammation in vivo. It has previously been shown that MDP administration renders mice more susceptible to LPS-induced endotoxic shock (32). Accordingly, we sensitized wild type, Rip2 knock-out, and Rip2 kinase-inactive mice with MDP and subsequently challenged with LPS. Inflammatory cytokine levels in the serum were measured 1.5 h after LPS administration (Fig. 1C). A strong synergy between MDP and LPS was evident in wild type mice as shown by significantly higher levels of IL-6 and IL-12p40 as compared with stimulation with single ligands; however, this synergy was not observed in the serum of Rip2 kinase-inactive or knock-out mice, where only the effect of LPS was detected. Thus, also in vivo the absence of Rip2 kinase activity leads to impaired inflammatory cytokine production upon MDP stimulation. Notably, injection of ultrapure LPS alone induced similar levels of IL-6 and IL-12 in all groups of mice, further excluding a critical role for Rip2 in mediating Toll-like receptor 4 signaling.

The cytokine storm induced by LPS during an endotoxic shock can lead to organ dysfunction and eventually death (33). We therefore monitored the survival of mice injected with MDP or PBS and challenged with LPS (Fig. 1D). Wild type mice sensitized with PBS (right panel) showed a better survival rate than their MDP-immunized counterparts, all of which succumbed to LPS challenge by day 4 (left panel). Conversely, we did not observe any significant difference in survival between Rip2 knock-out or knock-in mice sensitized with either MDP or PBS, indicating that no immune response to this Nod2 ligand took place in the absence of an intact Rip2 kinase domain.

Rip2 Kinase Activity Is Required to Ensure Protein Stability and Efficient Signaling Downstream of Nod2

Our results indicate that Rip2 deficiency and kinase inactivation both lead to abrogated signaling following NLR stimulation. Notably, different reports showed that a mutation in the Rip2 kinase residue, lysine 47, leads to a decrease in protein expression (23, 24). To further investigate the role of Rip2 kinase activity on the stability of the protein, we assessed Rip2 expression under steady state conditions or after stimulation with LPS in bone marrow-derived macrophages. Wild type, Rip2 knock-in, or Rip2 knock-out cells were stimulated with LPS for 6 and 24 h, and protein expression was analyzed by Western blot. In line with previous reports, we found that protein levels of the kinase-dead Rip2 mutant were lower in BM-DMs as compared with its wild type counterparts (Fig. 2A). Moreover, we found that activation with LPS was not sufficient to restore Rip2 expression in knock-in cells as proposed elsewhere (24). We next sought to determine whether introduction of the K47A mutation would negatively regulate transcription or mRNA stability of the Rip2 gene and thus lead to decreased protein levels. To this purpose Rip2 mRNA expression was determined in wild type, Rip2 knock-in, and Rip2 knock-out BM-DMs by quantitative real time PCR. We did not observe any impairment in Rip2 gene transcription in the knock-in cells when compared with the wild type counterparts, both in the presence or in the absence of LPS (Fig. 2B). These results indicate that K47A mutation affects Rip2 protein levels as opposed to RNA; yet the possible causes for the instability of the kinase remain to be clarified.

FIGURE 2.

Kinase inactivation affects Rip2 protein stability and phosphorylation of effector signaling molecules downstream of Nod2. A, BM-DMs were generated from C57BL/6, Rip2 K47A knock-in, and Rip2 knock-out mice and stimulated with 100 ng/ml LPS. Total cell lysates were prepared and blotted against Rip2 and β-actin. B, alternatively, total RNA was isolated, and Rip2 expression analyzed by reverse transcription and quantitative real time PCR. Rip2 transcript expression was normalized to the expression of the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase. C, BM-DCs from C57BL/6 mice were incubated for 24 h in medium alone (control) or supplemented with the indicated kinase inhibitors. Total cell lysates were blotted against Rip2 and β-actin. D, retrovirally infected bone marrow cells expressing wild type (WT), K47R, or K47R/S176E Rip2 protein were lysed and blotted against Rip2 and β-actin. E, BM-DMs were generated as in A and stimulated with 25 μg/ml MDP. The cell lysates were blotted with antibodies against the phosphorylated forms of p38, ERK, JNK, and IκB-α. The same samples were blotted with polyclonal antibodies against β-actin. The proteins were detected by ECL.

Replacement of a lysine with the uncharged and hydrophobic amino acid alanine in the K47A kinase-dead mutation could potentially lead to misfolding of the Rip2 protein. Alternatively, Rip2 kinase activity might play a direct role in controlling the stability of the protein, so that its inactivation would affect protein expression levels. This latter hypothesis is supported by a previous investigation by Windheim et al. (23) indicating that the expression levels of wild type Rip2 in transfected cells could be decreased following treatment with the kinase inhibitor SB203580. We sought to further investigate the influence of Rip2 kinase inhibitors on endogenous protein expression in primary cells. Accordingly, BM-DCs were generated from wild type mice and incubated with SB203580, a p38 kinase inhibitor, which was previously shown to efficiently inhibit Rip2 kinase activity. Additionally, BM-DCs were treated with BIRB0796, which strongly inhibits p38 but has no effect on Rip2. Rip2 protein levels were then analyzed by Western blot before, as well as 24 h after treatment. In support of the previous results, we observed a decrease in Rip2 protein levels in response to the kinase inhibitor SB203580 as compared with both nontreated and BIRB0796-treated cells (Fig. 2C). These data further indicate that Rip2 kinase activity is directly involved in mediating the stability of the protein even under steady state conditions.

To rule out a potential misfolding of the protein caused by the introduction of the uncharged amino acid alanine, we generated a new mutation in which lysine 47 is replaced by the more similar amino acid arginine. Wild type and K47R Rip2 mutant gene sequences were cloned into retroviral pMY-IRES-GFP vectors, and the resulting retroviruses were used to infect Rip2-deficient bone-marrow cells. GFP-positive cells expressing wild type or K47R mutant Rip2 were FACS-sorted and lysed, and the level of Rip2 protein expression was analyzed by Western blot. As shown in Fig. 2D, protein expression of the Rip2 K47R mutant was still reduced as compared with the retrovirally expressed wild type Rip2, suggesting that protein instability is not due to a charge difference but to the inactivation of the enzymatic activity.

Because Rip2 has been shown to undergo autophosphorylation only in the presence of an intact kinase activity (34), it can be speculated that this is an important mechanism by which the protein stabilizes itself. Serine 176 has been mapped as a Rip2 autophosphorylation site, and mutation of this amino acid influences kinase activity (34). Because the protein must be stable to determine its role in the course of signaling events, we sought to generate a stable kinase-inactive Rip2 mutant by mimicking autophosphorylation; for this purpose serine 176 was replaced by the negatively charged phospho-mimetic glutamic acid. We cloned the S176E mutation into our retroviral plasmid expressing the Rip2 mutant K47R and subsequently infected Rip2-deficient bone marrow cells. Expression of the K47R/S176E Rip2 double mutant was compared with wild type and K47R Rip2 protein expression. We found that introduction of the S176E phospho-mimetic mutation did not lead to a stable kinase-inactive mutant, because protein expression was comparable with the Rip2 K47R mutant (Fig. 2D). However, it remains possible that other serine amino acids, such as in the intermediate or CARD domain are also involved in autophosphorylation and thus might support protein stability.

Our results indicate that Rip2 kinase activity is required to preserve inflammatory immune responses following Nod1 and Nod2 stimulation in vivo and in vitro. Specifically, we speculate that Rip2 kinase activity primarily functions to stabilize Rip2 expression levels such that it can participate in the signaling cascade downstream of Nod1 and Nod2. Nevertheless, a role for Rip2 kinase activity in directly phosphorylating downstream effector molecules in vivo remains to be elucidated and cannot be excluded. Rip2 has previously been shown to mediate phosphorylation of different signaling molecules in vitro (19, 35), as well as to induce IKK-γ/NEMO ubiquitinylation, thereby controlling NF-κB activation (25). Moreover, recent reports describe TAK1 kinase as the crucial effector molecule downstream of Rip2 in NLR signaling (21, 22), and Rip2 overexpression has been shown to mediate TAK1 phosphorylation and activation (23). Hasegawa et al. (21) additionally showed that Rip2 polyubiquitinylation is required for interaction with TAK1 and subsequent NF-κB activation. It is of note that the NF-κB activation was achieved by inducing protein overexpression and oligomerization and not by activation via Nod1 and Nod2 ligands. In our system we activated wild type, Rip2 knock-in, and Rip2 knock-out BM-DMs with MDP and analyzed phosphorylation of signaling molecules, including the NF-κB-inhibitor IκB-α. The results shown in Fig. 2E indicate that under these more physiological conditions, NF-κB and mitogen-activated protein kinase activation did not take place in the absence of Rip2 kinase activity.

Taken together, our in vitro and in vivo data show that the serine/threonine kinase, Rip2, is a central regulator of NLR-induced immune responses. Although the exact role of Rip2 kinase activity in the activation of signaling molecules still remains to be elucidated, our results highlight the importance of an intact kinase domain in preserving adequate Rip2 protein levels and thus in ensuring the efficient transmission of Nod1 and Nod2 signals. Accordingly, inhibition of Rip2 kinase activity abrogates signaling downstream of Nod1 and Nod2, and may thus prove to be a valid therapeutic strategy for the treatment of inflammatory diseases linked to the dysregulated activity of these intracellular receptors (36).