|Home | About | Journals | Submit | Contact Us | Français|
Dysregulation of NOD2 signaling is implicated in the pathology of various inflammatory diseases, including Crohn's disease, asthma, and sarcoidosis, making signaling proteins downstream of NOD2 potential therapeutic targets. Inhibitor-of-apoptosis (IAP) proteins, particularly cIAP1, are essential mediators of NOD2 signaling, and in this work, we describe a molecular mechanism for cIAP1's regulation in the NOD2 signaling pathway. While cIAP1 promotes RIP2's tyrosine phosphorylation and subsequent NOD2 signaling, this positive regulation is countered by another E3 ubiquitin ligase, ITCH, through direct ubiquitination of cIAP1. This ITCH-mediated ubiquitination leads to cIAP1's lysosomal degradation. Pharmacologic inhibition of cIAP1 expression in ITCH−/− macrophages attenuates heightened ITCH−/− macrophage muramyl dipeptide-induced responses. Transcriptome analysis, combined with pharmacologic inhibition of cIAP1, further defines specific pathways within the NOD2 signaling pathway that are targeted by cIAP1. This information provides genetic signatures that may be useful in repurposing cIAP1-targeted therapies to correct NOD2-hyperactive states and identifies a ubiquitin-regulated signaling network centered on ITCH and cIAP1 that controls the strength of NOD2 signaling.
Recognition of microorganisms is mediated through various extracellular and intracellular pattern recognition receptors, mainly the Toll-like receptors (1), NOD-like receptors, and RIG-I-like receptors. However, this recognition and response are a double-edged sword. Too weak a response causes immunodeficiency, while too strong a response causes autoimmune or inflammatory syndromes. In no protein is this dichotomy more evident than in the NOD-like receptor NOD2. The NOD2 protein responds to muramyl dipeptide (MDP), a breakdown product of peptidoglycan found in both Gram-positive and Gram-negative bacteria (1, 2). Signaling through the NOD2 pathway culminates in the activation of inflammatory signaling cascades, stimulation of autophagy and antigen presentation, and release of antimicrobial peptides (3–9). These functions not only help in the direct control of invading microorganisms but also promote the appropriate cytokine and inflammatory milieu for directing the adaptive immune response to eradicate the offending pathogen. NOD2 has a critical role in immunologic homeostasis, as both loss-of-function polymorphisms and gain-of-function mutations in NOD2 precipitate inflammatory disease (2, 10–17). Though most notably recognized for its involvement in the pathogenesis of Crohn's disease through a mechanism involving a loss of downstream NOD2 signaling, it is becoming more and more appreciated that overactivation of NOD2 signaling is an underlying theme in a number of disorders such as early-onset sarcoidosis, Blau syndrome, allergy and asthma, and autoimmunity (13, 14, 16–23). Therefore, deeper insights into the molecular mechanisms underlying not only NOD2 activation but also NOD2 downregulation are critical for the understanding of NOD2-driven inflammatory disease, as well as for the design and testing of various NOD2-directed pharmacological agents.
Over the past decade, it has become apparent that the activation and deactivation of NOD2 signaling are tightly regulated by posttranslational modifications, most specifically, ubiquitination and phosphorylation (24, 25). This is highlighted by the fact that the obligate kinase in the NOD2 signaling pathway, the dual-specificity kinase RIP2, undergoes both autophosphorylation and ubiquitination in a site-specific and time-specific manner to affect the activation and subsequent deactivation of NOD2 signaling. We have shown that RIP2 undergoes tyrosine autophosphorylation at Y474, which correlates with the activation of downstream NOD2 signaling. Importantly, this phosphorylation event also serves as a mechanism by which activated RIP2 can be recognized and downregulated by the ITCH/A20 ubiquitin-editing complex (25). This coupled activation-deactivation mechanism allows a specificity in signal transduction such that NOD2 signaling is appropriately tuned and neither immunodeficiency nor autoinflammatory diseases result.
Opposing the downregulatory effects of the E3 ligase ITCH are the inhibitor-of-apoptosis (IAP) proteins cIAP1, cIAP2, and XIAP. All three IAPs have been demonstrated to bind to and promote the ubiquitination of RIP2 (26–29). cIAP1 and cIAP2 have been shown to positively influence NOD2 signaling through promoting K63 polyubiquitination of RIP2 at lysine 209 (K209), leading to the activation of downstream signaling pathways (26, 30, 31). XIAP has also recently been reported to recruit the linear ubiquitin chain assembly complex to NOD2 and ubiquitinate RIP2 to allow for efficient NF-κB activation and cytokine secretion in response to stimulation with MDP (28). Thus, the IAP proteins appear to promote ubiquitination that centers on RIP2 activation and appears to center on the ubiquitination of lysine 209 (K209) on RIP2. In contrast, the ITCH/A20 ubiquitin-editing complex recognizes tyrosine-phosphorylated, activated RIP2 and specifically downregulates NOD2-RIP2 signaling through the ubiquitination of alternative RIP2 lysines. This multiubiquitination of RIP2 by ITCH occurs at sites distinct from K209 (31), and while ITCH- and cIAP1-induced RIP2 polyubiquitination appears to drive the activation and deactivation of this pathway, respectively, the interplay between this ITCH/A20 ubiquitin-editing complex and the IAP proteins in the context of NOD2 signaling has not been studied.
While evidence from multiple independent groups has shown that overexpression of cIAP1 leads to K63-linked polyubiquitination of RIP2 (26–29), data obtained with cIAP1−/− mice is complicated by the recent unexpected realization that these mice are also deficient in caspase 11 because of a passenger mutation in the originating 129S mouse strain (32). Caspase 11's proximity to caspase 1 makes these mice potentially also deficient in caspase 1. Given that caspase 11 is also an upstream activator of caspases 1 and 3 and given that the NOD2-RIP2 pathway influences inflammasome activity and cell death, data generated with the cIAP1-deficient mouse must be reevaluated in light of these findings. For these reasons, the use of pharmacologic cIAP1 inhibitors, validated with siRNA knockdown of cIAP1 expression, might be the more desirable approach to separation of the function of the IAPs in the NOD2 signaling pathway until the two mutations are genetically separated.
Despite reported overlapping functions of cIAP1 and cIAP2, clear differences exist. For example, cIAP1 but not cIAP2 regulates XIAP levels by promoting its degradation (33). There is also a differential requirement for each IAP in mice doubly deficient for a combination of cIAP1, cIAP2, and XIAP for survival (34). More importantly, there is preferential binding of RIP2 to cIAP1 over cIAP2, and in vitro ubiquitination assays have shown greater ubiquitination of RIP2 by cIAP1 than by cIAP2 (26, 27). Thus, integration of these data from independent laboratories suggests that two separate E3 ubiquitin ligases, cIAP1 and ITCH, are the main E3 ligases mediating the activation and subsequent deactivation of the NOD2-RIP2 complex. cIAP1 appears to target K209 on RIP2 through K63-linked polyubiquitination, while ITCH appears to recognize and multiubiquitinate only activated RIP2 at a distinct site(s).
In this report, we describe novel molecular mechanisms by which NOD2 signaling is positively regulated by cIAP1 and negatively regulated by ITCH. First, cIAP1 promotes the activation of RIP2's kinase activity, resulting in a tyrosine autophosphorylation of RIP2 that ultimately targets RIP2 for ITCH-mediated downregulation. Second, ITCH ubiquitinates cIAP1 to promote its lysosomal degradation. Third, we demonstrate how these basic biochemical findings can be translated into a potential therapy for NOD2-hyperactive states. We show that ITCH−/− cells, which have increased NOD2 activity, can be pharmacologically targeted with the cIAP1 antagonist (−)-N-[(2S,3R)-3-amino-2-hydroxy-4-phenyl-butyryl]-l-leucine methyl ester (MEBS) to correct excessive NOD2 responses. By next-generation RNA sequencing, we identify NOD2-specific, cIAP1-specific pathways, functions, transcription factors (TFs), and target genes. This gene signature not only confirms the interrelationship between cIAP1 and ITCH as cIAP1-affected genes are upregulated in ITCH−/− macrophages exposed to MDP and downregulated in response to cIAP1 inhibition but, given the fact that IAP inhibitors are in clinical development as anticancer agents, allows a transcriptional profile to be monitored if repositioning these agents for use in the downregulation of unrestrained NOD2 activation.
HEK293, RAW264.7, and HT29 cells were obtained from ATCC and grown in 10% fetal bovine serum (FBS) plus Dulbecco's modified Eagle's medium (DMEM). Hemagglutinin (HA)-tagged NOD2, HA-ubiquitin, HA-RIP2, FLAG-ITCH, FLAG-ITCH C830A, and Omni RIP2 were used as previously described (25). Myc-tagged cIAP1 was a gift from Jon Ashwell (NIH). All point mutants were generated by QuikChange site-directed mutagenesis (Stratagene) and confirmed by sequencing. Stable cell lines were generated by subcloning into pBABE or pMXneo. Viral particles were generated in HEK293 cells by transfection of pBABE/pMXneo along with VsVg and GagPol plasmids before the infection of target cell lines in the presence of 4 mg/ml Polybrene (Millipore) and selection (puromycin or G418; Invivogen) as previously described (35). RIP2 (rabbit), ITCH (rabbit), actin (mouse), and myc (mouse, clone 9E10) antibodies were obtained from Santa Cruz. cIAP1 (rabbit), myc (rabbit), phospho-IκBα (mouse), total IκBα (rabbit), phospho-p38 (rabbit), total p38 (mouse), phospho-JNK (rabbit), phospho-ERK (mouse), and NF-κB2 (rabbit) were all from Cell Signaling. Anti-cIAP1 (goat) was from R&D. Anti-HA antibody (clone HA.11) was obtained from Covance. MDP was obtained from Bachem. Cycloheximide (CHX) and chloroquine were obtained from Sigma. MEBS was obtained from EMD Millipore.
Transfections were performed by using the calcium phosphate method for HEK293 cells or using a NEON electroporator for HT29 cells. Small interfering RNA (siRNA) duplexes (ITCH, FlexiTube set GS83737; cIAP1, FlexiTube set GS329) were obtained from Qiagen and used at a concentration of 5 nM. Cells were harvested 24 h after transfection and washed once in 1× phosphate-buffered saline before lysis. Lysis buffer consisted of 50 mM Tris, 150 mM NaCl, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM sodium EDTA, 1 mM EGDA, 1 mM β-glycerophosphate, 5 mM iodoacetamide, and 5 mM N-ethylmaleimide adjusted to pH 7.5, with calyculin, protease inhibitor cocktail, 1 mM phenylmethylsulfonyl fluoride, and 1 mM sodium orthovanadate added fresh before use. For experiments with RIP2 tyrosine phosphorylation, cells were treated with 100 mM pervanadate 5 min before being harvested. Immunoprecipitation (IP) was performed by incubating lysates with immunoprecipitating antibody for 1 h before the addition of 10 ml protein G-Sepharose (Invitrogen) overnight. Immunoprecipitates were washed three to five times in lysis buffer before the aspiration of buffer and boiling Sepharose in 2× sample buffer. Depending on the molecular weight of the protein assayed, the gels used were 8 to 10% acrylamide. Ubiquitination experiments were handled identically, with the exception of high-stringency washes (1 M NaCl, 1% SDS) during IP.
Bone marrow-derived macrophages (BMDMs) from WT or Itchy mice (MRC Harwell; backcrossed for at least 8 generations to C57BL/6 mice from Jackson Laboratories) were generated by culturing bone marrow for 7 days in 10% DMEM with 25% conditioned Ladmac medium (gift from Clifford Harding, Case Western Reserve University [CWRU]). Cells were rested in 10% FBS–DMEM overnight before use. BMDMs were pretreated with 30 mM MEBS or the vehicle (dimethyl sulfoxide) for 30 min before stimulation with MDP for 4 h. Cells were then harvested, and RNA was extracted with a Qiagen RNeasy kit in accordance with the manufacturer's instructions. The experiment and RNA sequencing (RNA-Seq) analysis were done independently in duplicate.
RNA was sent to the Oklahoma Medical Research Foundation for sequencing. The 100-bp paired-end Illumina reads were processed to remove the 3′ bases with Phred quality scores of <20. Reads that were less than 20 bases after quality trimming were removed from further analysis. The reads from each replicate of each sample were mapped to mouse genome release mm9 by using the TopHat v1.4.1 program (36) before guided assembly by using the Cufflinks v1.3.0 program (37) with mouse genome annotation from the RefSeq database and the lincRNA annotation from the Ensembl database. Differential expression of transcripts was analyzed with the two-tailed Student t test with Benjamini-and-Hochberg correction of the false-discovery rate (FDR) (38). An FDR-corrected P value of 0.05 was set as the cutoff of statistical significance. Canonical and biofunctional pathway analysis was carried out with Ingenuity IPA software comparing MDP-treated to MEBS-plus-MDP-treated BMDMs. Canonical pathway charts and TF tables were generated by Core analysis and may reflect selected subsets of data for clarity of presentation (for example, nonmacrophage, nonimmune signaling pathways were excluded for the canonical pathway chart).
RNA was reverse transcribed with a Quantitect RT kit (Qiagen). The primers used for amplification were mCXCL10-F (5′ TCCTTGTCCTCCCTAGCTCA 3′), mCXCL10-R (5′ ATAACCCCTTGGGAAGATGG 3′), mCXCL11-F (5′ CCAAAGCCCAGGCAGAGAGCTG 3′), mCXCL11-R (5′ CCGGGGCCGATGCAAAGACA 3′), mCD40-F (5′ TCTCGCCCTGCGATGGTGTCT 3′), mCD40-R (5′ CGGCTTCCTGGCTGGCACAA 3′), mCD69-F (5′ TGGCCCAACGCTCTTGTTCTGA 3′), mCD69-R (5′ CGCACCTCCCAGACCCCGTC 3′), mFAS-F (5′ AAAGCTGAGGAGGCGGGTTCGT 3′), mFAS-R (5′ GTGCATGGGGCGCAGGTTGG 3′), mCCL5-F (5′ ATATGGCTCGGACACCACTC 3′), mCCL5-R (5′ GTGACAAACACGACTGCAAGA 3′), mNOS2-F (5′ TGGCTCGCTTTGCCACGGAC 3′), mNOS2-R (5′ GCTGCGACAGCAGGAAGGCA 3′), mGBP5-F (5′ TCCCGTTTGAAGCAGTGTTCT 3′), mGBP5-R (5′ TGCTCCCAATGAGGCACAAG 3′), mJAG1-F (5′ ACAGGGCTGCAGTCCCAAGCA 3′), mJAG1-R (5′ GTGCGGGATGCACTTGTCGC 3′), mGAPDH-F (5′ AGGCCGGTGCTGAGTATGTC 3′), and mGAPDH-R (5′ TGCCTGCTTCACCACCTTCT 3′). Sybr green was obtained from Bio-Rad, and the real-time PCRs were carried out with a CFX96 C1000 Real-Time Thermal Cycler from Bio-Rad.
RNA sequencing was performed twice (biological replicates), and differential expression of transcripts was analyzed with a two-tailed Student t test with Benjamini-and-Hochberg correction of the FDR (38). An FDR-corrected P value of 0.05 was set as the cutoff of statistical significance. Pathway analysis of RNA sequencing data was performed with IPA software (Ingenuity Systems). Canonical pathway analysis identified the pathways from the IPA library of canonical pathways that were most significant for the data set. A subset of these pathways was selected for clarity of presentation (nonmacrophage, nonimmune signaling pathways were excluded). The significance of the association between the data set and the canonical pathway was measured in two ways. (i) A ratio of the number of molecules from the data set that map to the pathway divided by the total number of molecules that map to the canonical pathway was displayed, and (ii) Fisher's exact test was used to calculate a P value determining the probability that the association between the genes in the data set and the canonical pathway is explained by chance alone. RT-PCR data are presented as the mean ± the standard error of the mean. RT-PCR experiments were performed in duplicate and repeated twice. Results of representative experiments are shown. The significance of the comparisons shown was assessed by Student's two-tailed t test with the cutoff for significance set at P = 0.05.
In order to determine specific pathways, functions, and TFs affected by cIAP1 during NOD2 signaling without the confounding effects of caspase 11 deficiency, we performed RNA-Seq analysis with the compound MEBS, which specifically causes degradation of cIAP1 but not cIAP2 or XIAP (39). MEBS was added to BMDMs 30 min prior to MDP stimulation for 4 h. As a control for MDP-independent effects of MEBS, RNA-Seq analysis of cells treated with MEBS alone was also performed. Two independent RNA-Seq analysis experiments were performed with macrophages derived from age- and sex-matched mice. NOD2-specific, cIAP1-specific responses were determined by comparing the differential expression of transcripts between vehicle-plus-MDP treatment and MEBS-plus-MDP treatment (MDP affected, MEBS affected). This resulted in 811 transcripts differentially expressed, 19 of which were altered by MEBS in the absence of MDP and which were therefore excluded (genes affected by the drug alone). The list of transcripts for MDP-affected genes altered by cIAP1 inhibition and their kilobase of exon per million fragments mapped and P values are listed in Table S1 in the supplemental material. Pathway analysis of these 792 MDP-affected, MEBS-affected transcripts was then performed with Ingenuity IPA software. For clarity, only macrophage-specific, immune-related canonical pathways that meet the statistical significance cutoff (P < 0.05) are shown (Fig. 1A; see Table S2 in the supplemental material). The complete list of MDP-affected, cIAP1-affected pathways and the corresponding P values and genes within each pathway are listed in Table S3 in the supplemental material. As expected, inhibition of cIAP1 affected well-described pathways downstream of NOD2 activation such as NF-κB and mitogen-activated protein kinase (MAPK), as well as protein ubiquitination pathways and pathways affecting autophagy (mTOR) (Fig. 1A). According to this analysis, the pathway most affected by cIAP1 inhibition was lymphotoxin beta receptor (LTβR) signaling, which is involved in apoptosis, inflammation, and organogenesis depending on classical or alternative NF-κB activation (Fig. 1A). LTβR signaling in a number of cell types (including macrophages) has been shown to be critical in the control of inflammation in various colitis models (40–43). Interestingly, NOD2-deficient mice are reported to have defective function and development of Peyer's patches (which are known to require LTβR signaling during organogenesis) (44). To our knowledge, this is the first study suggesting that NOD2 activation itself affects LTβR signaling and that cIAP1 may be critical for the induction of this pathway. Other unexpected findings include the involvement of cIAP1 in endocytic trafficking and cytoskeletal signaling (integrin-linked kinase signaling, actin signaling) (Fig. 1A). Although there has been some evidence that suggests that NOD2 is involved in the activation of antiviral pathways, in this study, no effect of cIAP1 inhibition on these pathways was observed.
The 792 differentially expressed genes were then reanalyzed to determine biological functions (Fig. 1B) and TFs (Fig. 1C) affected by cIAP1 in the context of NOD2 activation. Ingenuity IPA software is able to predict the activation state of the function or TF based on the z score (with values of <2 predicted to be decreased and values of >2 predicted to be activated). Functional and TF analyses showed that inhibition of cIAP1, in large part, resulted in decreased macrophage chemotaxis and phagocytosis and in inhibition of the STAT1 and IRF1 TFs, suggesting that cIAP1 affects these functions, TFs, and their downstream genes disproportionately. Shown for clarity are the top 10 biological functions (Fig. 1B) and top 15 TFs affected (ranked by z score) (Fig. 1C). The z score is affected by both the up- or downregulation of TF expression and the identities of the TF target genes induced. While the NF-κB pathway is a clear target of cIAP1 inhibition (Fig. 1A), the NF-κB TFs score neutral by z score because of limited transcriptional upregulation of the actual NF-κB TFs. This likely reflects the fact that cIAP1 affects mainly acute NF-κB signaling and has a lesser effect on secondary (e.g., expression induction of p50/p65) NF-κB signaling. Complete lists of the biological functions and TFs affected by cIAP1 in the context of NOD2 activation are provided in Tables S4 and S5 in the supplemental material, respectively.
To validate the transcriptomic data, MDP-affected, MEBS-affected (NOD2-specific, cIAP1-specific) genes were selected and their expression was verified by quantitative RT-PCR (qRT-PCR). BMDMs from wild-type (WT) C57BL/6 mice showed a significant decrease in the expression of multiple NOD2-specific, cIAP1-specific transcripts with MEBS treatment (Fig. 2A). Given that not all of the MDP-induced genes identified by RNA-Seq analysis were affected by MEBS, as an additional control, qRT-PCR of MDP-responsive genes predicted to be nonresponsive to MEBS, such as FAS and CCL5, showed MDP inducibility without an effect of cIAP1 inhibition (Fig. 2A), again supporting the results in Fig. 1. To determine that this inhibition was not due to off-target effects of MEBS, we confirmed these results with RT-PCR in combination with siRNA knockdown of cIAP1 (Fig. 2B) or through the use of a cell-permeating smac mimetic, smac n7 (Fig. 2C). While the effect was weaker with siRNA knockdown because of the complete degradation of cIAP1 with the pharmacologic agent versus the partial inhibition of cIAP1 expression in primary macrophages with siRNA, the two transcripts tested were significantly inhibited. Likewise, treatment with the second cIAP1 antagonist smac n7 also caused significant downregulation of CXCL10 and CXCL11 (for confirmation of the degradation of cIAP1 in BMDMs with smac n7, see the last lane of Fig. 7B). Altogether, combination of pharmaceutical downregulation of cIAP1 with next-generation sequencing technology and pathway analysis resulted in an informative picture of cIAP1-specific effects in the NOD2 signaling pathway. We have described the involvement of cIAP1 in novel pathways such as LTβR signaling, endocytic trafficking, and cytoskeletal signaling and confirmed the importance of cIAP1 in macrophage chemotaxis and phagocytosis in the context of NOD2 activation. Validated targets include but are not limited to the chemokines CXCL10 (IP-10, chemoattractive for various immune cells) and CXCL11 (I-TAC, chemoattractive for T cells), which are downstream targets of the STAT1 TF, as well as the cell surface molecule CD40 and enzyme NOS2 (inducible nitric oxide synthase), which are downstream targets of both STAT1 and IRF1.
The specific biochemical events by which cIAP1 regulates NOD2 signaling and how cIAP1 is regulated in the context of NOD2 signaling are unknown. To address this issue, we examined the effect of cIAP1 on the proximal NOD2 effector, the dual-specificity kinase RIP2. cIAP1 has previously been reported to promote the nondegradative K63-linked polyubiquitination of RIP2, causing activation of NOD2 signaling (27, 30). As our laboratory has demonstrated that tyrosine-autophosphorylated RIP2 is a proximal marker for activation of the NOD2 pathway (25), we tested the effect of cIAP1 on RIP2 tyrosine autophosphorylation. WT or kinase-inactive RIP2 (RIP2 K47A) was transfected into HEK293 cells with or without ligase-sufficient cIAP1 or ligase-defective cIAP1 (cIAP1 H588A). RIP2 was immunoprecipitated, and Western blot assays were performed. Transfection of cIAP1 with WT RIP2 but not RIP2 K47A led to RIP2 tyrosine autophosphorylation (Fig. 3A, lane 3), suggesting that cIAP1 was promoting the activation of RIP2. This effect was lost when ligase-defective cIAP1 was used (Fig. 3A, lane 5), indicating that the ligase activity of cIAP1 is important in promoting RIP2 activation. We then transfected in increasing amounts of cIAP1 to see what effect this would have on RIP2 tyrosine phosphorylation. As expected, we found a dose-dependent increase in RIP2 tyrosine phosphorylation (Fig. 3B). Conversely, knockdown of cIAP1 with two different siRNAs decreases RIP2 tyrosine phosphorylation to basal levels (Fig. 3C). Since other reports have shown a positive correlation between cIAP1-directed RIP2 ubiquitination and activation of downstream NOD2 signaling, we hypothesized that the ubiquitin ligase activity of cIAP1 would be necessary to observe the effect on RIP2 tyrosine phosphorylation and that ligase-deficient cIAP1 (cIAP1 H588A) would act in a dominant negative manner. Titration of increasing amounts of cIAP1 H588A decreased NOD2-induced RIP2 tyrosine autophosphorylation in a dose-dependent fashion (Fig. 3D). Collectively, these results suggest that cIAP1 promotes the tyrosine autophosphorylation of RIP2.
To determine then if cIAP1-mediated ubiquitination of RIP2 is required prior to RIP2 tyrosine autophosphorylation (activation), we utilized a mutant form of RIP2 in which the lysine residue to which cIAP1 conjugates ubiquitin chains has been mutated to an arginine (K209R). Transfection of cIAP1 promoted ubiquitination and tyrosine phosphorylation of RIP2. Both of these processes were abolished when the K209R mutant was used (Fig. 3E), indicating that cIAP1-mediated ubiquitination of RIP2 is the initial step before the activation and tyrosine autophosphorylation of RIP2.
Multiple E3 ligases have been reported to ubiquitinate RIP2. The two major ones are cIAP1, which positively influences NOD2 signaling through site-specific ubiquitination of RIP2, and ITCH, which causes multiubiquitination of RIP2 to negatively regulate NOD2 signaling (27, 30, 31). Given their opposing effects on the NOD2 pathway, we wanted to determine what effect having both of these ubiquitin ligases active would have on RIP2 tyrosine phosphorylation. To test this, we transiently transfected RIP2 into HEK293 cells with each of the ligases singly or in combination. We then immunoprecipitated RIP2 and assessed the influence of each of the ligases on RIP2 tyrosine phosphorylation and ubiquitination. We observed that although both cIAP1 and ITCH promoted the ubiquitination of RIP2, only cIAP1 promoted the tyrosine phosphorylation of RIP2 (Fig. 3F, lane 3 versus lane 5). This cIAP1-induced RIP2 tyrosine phosphorylation was abolished when both ubiquitin ligases were active (Fig. 3F, lane 3 versus lane 7) but not when ITCH was ligase defective (Fig. 3F, lane 8). Furthermore, regardless of the activity of ITCH, the presence of ligase-deficient cIAP1 resulted in no inducible RIP2 tyrosine phosphorylation (Fig. 3F, lanes 9 and 10). These results suggest that when both ligases are present, the effects of the E3 ligase ITCH on RIP2 tyrosine phosphorylation predominate.
Altogether, we have shown that cIAP1 promotes the tyrosine autophosphorylation of RIP2. We have shown that the ligase activity of cIAP1 is necessary for this effect and that cIAP1-mediated RIP2 ubiquitination is the initial step before RIP2 activation. In addition, we have demonstrated that in the presence of the two opposing E3 ligases cIAP1 and ITCH, the effects of ITCH on RIP2 tyrosine phosphorylation predominate.
Given our previous finding that ITCH predominates over cIAP1 when both ligases are present, we hypothesized that ITCH was influencing cIAP1 directly. Transient transfection of ubiquitin and ITCH with ligase-defective cIAP1 H588A (which removes the confounding effects of cIAP1 autoubiquitination) resulted in ITCH-mediated cIAP1 ubiquitination (Fig. 4A, lane 5). In the presence of ligase-defective ITCH C830A, ubiquitination of cIAP1 was abolished (Fig. 4A, compare lanes 5 and 6). This effect of E3 ligase regulation was not bidirectional, as cIAP1 did not cause ubiquitination of ligase-defective ITCH C830A (Fig. 4B, lane 5; RIP2 ubiquitination is shown as a positive control). To confirm that the observed ITCH-mediated cIAP1 ubiquitination was not an artifact of overexpression, we then stimulated two cell lines, N2 HT29s, a human intestinal epithelial cell line stably expressing NOD2, and RAW264.7, a mouse macrophage cell line, with the NOD2 agonist MDP and found that cIAP1 ubiquitination occurs endogenously in an agonist-dependent manner in both of these cell lines (Fig. 4C and andDD).
Ubiquitination mediates both degradative and nondegradative functions. Conjugation of proteins with ubiquitin via K48 linkages has been shown to mediate the more typical proteasomal degradation, though K11, K29, and K63 linkages have also been shown to do so with less frequency (45). Lys63 linkages have been associated with nondegradative functions such as protein sorting (46), DNA repair (47), and inflammation (48) but also with lysosomal targeting and degradative functions (49–51). ITCH has been shown to induce the formation of K29 (52), K48 (53, 54), and K63 linkages (31, 55, 56). As ITCH appeared to have a negative effect on cIAP1 (negating the effects of cIAP1 when both ligases were present) and as ITCH is found primarily in endosomes/lysosomes (57), we tested if ITCH-mediated cIAP1 ubiquitination was leading to cIAP1 K63 ubiquitination and lysosomal degradation. First, we used a K63-ubiquitin linkage-specific antibody (D7A11 from Cell Signaling) to detect ITCH-mediated cIAP1 ubiquitination. Indeed, ligase-sufficient, but not ligase-defective, ITCH promoted K63 polyubiquitination of cIAP1 (Fig. 5A, compare lanes 3 and 4). This was further confirmed with three different mutant forms of ubiquitin that allow conjugation of ubiquitin only through a specific linkage (i.e., are mutated on all except one lysine on ubiquitin)—K29 only, K48 only, or K63 only. We observed that ITCH-mediated cIAP1 ubiquitination occurs only in the presence of WT ubiquitin or K63 linkage-specific ubiquitin and not in that of K29 or K48 linkage-specific ubiquitin mutants (Fig. 5B). To determine next whether this ITCH-mediated K63 polyubiquitination of cIAP1 was promoting its degradation, we used CHX, an inhibitor of protein synthesis, and assessed the difference in the half-life of cIAP1 in the absence of ITCH and in the presence of either ligase-active or ligase-deficient ITCH C830A (Fig. 5C). For each condition, cIAP1 levels were normalized to starting quantities (0 h CHX time point). Calculation of protein half-life was based on a one-phase exponential decay, setting the plateau to 0. With these parameters, there was an appreciable reduction in the half-life of cIAP1 H588A when ITCH was present (Fig. 5C, from 2.8 h to 1.7 h), and this was prolonged when ligase-deficient ITCH was used (Fig. 5C, from 2.8 h to 4.7 h). Because ITCH has been reported to localize to endocytic compartments that are continuous with the lysosome, we next wanted to determine if degradation of cIAP1 was mediated through lysosomal degradation. We compared the effects of the proteasomal inhibitor MG132 and the lysosomal inhibitor chloroquine on ITCH-induced cIAP1 degradation and found that cIAP1 was rescued in the presence of chloroquine but not MG132 (Fig. 5D). Following this line of reasoning, loss of ITCH should lead to increased levels of cIAP1. To test this, we used primary BMDMs generated from ITCH−/− mice or WT littermate control mice to compare levels of cIAP1. BMDMs generated from ITCH−/− mice show cIAP1 levels higher than those of BMDMs from WT mice (Fig. 5E). Altogether, these data suggest that ITCH-mediated cIAP1 ubiquitination occurs through K63 linkages and promotes cIAP1 lysosomal degradation and that loss of ITCH leads to increased levels of cIAP1.
The above data, coupled with our previously published data (25, 31), suggest a sequential ubiquitination of RIP2 whereby cIAP1 binds and ubiquitinates RIP2. RIP2 then autophosphorylates to both drive downstream NOD2 signaling pathways and also allow ITCH to recognize and downregulate activated RIP2 (25). Given this potential sequence of events, we tested whether ITCH expression or ITCH exposure to a preformed cIAP1-RIP2 complex could cause complex dissociation. Consistent with findings reported by other groups (7, 26, 27), RIP2 binds cIAP1 (Fig. 6A and andB,B, lanes 4). Upon ITCH overexpression, this binding was lost (Fig. 6A and andB,B, lanes 6). To determine then if recombinant, purified ITCH could inhibit a preformed cIAP1-RIP2 complex, cIAP1 and RIP2 were coexpressed and purified. Bacterially produced glutathione S-transferase (GST)-ITCH (90% pure) (31) was then added to the purified cIAP1-RIP2 complex in increasing amounts. Recombinant protein was equalized through the use of purified GST. Purified GST-ITCH could dissociate preformed cIAP1-RIP2 complexes (Fig. 6C). Collectively, these data suggest a model of the activation and deactivation of NOD2 signaling in which activation of NOD2 upon exposure to MDP results in the recruitment of RIP2 and binding of cIAP1 for ubiquitination of RIP2 at lysine 209 (K209). Such ubiquitination results in the activation of RIP2, as evidenced by RIP2 tyrosine autophosphorylation. This then allows downstream signal transduction events to occur. Deactivation of NOD2 signaling is mediated by the recognition of activated, tyrosine-phosphorylated RIP2 by ITCH, which competes with cIAP1 for binding to RIP2. While in this complex, ITCH promotes the ubiquitination of both RIP2 and cIAP1, leading to cIAP1 lysosomal degradation and, ultimately, downregulation of NOD2 responses (the model is presented in Fig. 6D).
If cIAP1 positively influences NOD2 signaling by ubiquitinating RIP2 and promoting its tyrosine phosphorylation and if ITCH negatively regulates NOD2 signaling by mediating ubiquitination and lysosomal degradation of cIAP1, then it should follow that loss of ITCH would lead to increased levels of cIAP1 and increased tyrosine phosphorylation of RIP2. To determine this, we used four siRNAs targeting ITCH to inhibit ITCH expression. This ITCH inhibition resulted in an increase in both NOD2-induced RIP2 tyrosine phosphorylation and cIAP1 levels (Fig. 7A). Conversely, in ITCH-null cells, inhibition of cIAP1 expression might reverse the NOD2 hyperactivity in ITCH−/− macrophages (31). Administration of MEBS downregulated endogenous cIAP1 in BMDMs in a dose-dependent fashion (Fig. 7B). A second cell-permeating smac mimetic (smac n7) was also demonstrated to downregulate endogenous cIAP1 (Fig. 7B, last lane). With the NOD2-induced, cIAP1-specific genes from the transcriptome analysis (Fig. 1 and and2),2), we observed that the expression of various inflammatory cell surface molecules, chemokines, and enzymes that were overexpressed in the setting of ITCH deficiency was reduced to WT levels by MEBS administration (Fig. 7C; also compare Fig. 2A, as the experiments were performed concurrently). In addition, knockdown of cIAP1 in ITCH−/− BMDMs with two different siRNAs showed significant downregulation of CXCL10, verifying that the effects of MEBS were specific to cIAP1 (Fig. 7D). Altogether, the data demonstrate that a molecular and biochemical understanding of NOD2 signaling can lead to a potential targeted therapy for the NOD2 overactivation present in ITCH deficiency.
Involvement of hyperactive NOD2 signaling in asthma, sarcoidosis, and other inflammatory diseases has made NOD2 and components of the NOD2 signaling pathway attractive pharmaceutical targets to modulate the immune response in the setting of disease. The most obvious and appealing target proteins are NOD2 itself and its interacting kinase RIP2. At present, it is unclear whether global inhibition of NOD2 signaling or inhibition of more specific downstream NOD2 effectors is the more desirable route for drug design in settings of NOD2 hyperactivity. One could speculate that certain downstream NOD2 signaling pathways are affected differentially in different overactive NOD2 states or might function aberrantly in different cell types, resulting in the varied pathology observed. Therefore, there is a need for the ability to target NOD2 signaling at different stages of NOD2 activation, both proximally and distally. Understanding the specific defect in a particular NOD2-overactive state may also allow a more personalized treatment that may leave certain aspects of NOD2-mediated microbial recognition and response intact so as to prevent complete immunodeficiency. One such example is the setting of ITCH deficiency.
In humans, loss of ITCH results in systemic autoimmunity and widespread inflammation in the lungs, liver, and gastrointestinal tract (58), which are also reflected in the Itchy mouse (59). Although it is recognized that ITCH has multiple downstream targets, our work suggests that in the context of NOD2 overactivation, part of the defect lies in the failure to ubiquitinate and degrade cIAP1, resulting in its overexpression and continuous activation of RIP2 (this report; 31). Insufficient clearance of activated RIP2 by ITCH could exacerbate this inflammatory cascade (31). Recognition of these defects allowed us to repurpose the cIAP1 antagonist MEBS, originally developed as an anticancer agent, for downregulation of NOD2 responses in the setting of ITCH deficiency. This approach was successful in lowering the production of NOD2-induced inflammatory mediators in Itchy mice to the levels seen in WT animals. Importantly, such treatment did not result in complete ablation of the NOD2 response (Fig. 7C). Additionally, although we describe the setting in which we downregulate the NOD2 response through inhibition of cIAP1, preventing the ITCH-mediated cIAP1 ubiquitination event can potentially result in increased cIAP1 stability and increased NOD2 signaling. If the specific ITCH-cIAP1 interaction is disrupted, this finding may be exploited in settings where NOD2 signaling is lacking because of an upstream defect in the pathway (NOD2 L1007insC) and possibly restore the responsiveness of this pathway.
The identified MDP-specific, cIAP1-specific transcriptional signature may also be useful in evaluating cIAP-directed pharmaceutical agents for use in the treatment of NOD2-driven inflammatory disease. Currently, there is movement toward repurposing existing drugs or “rescuing” compounds for use in novel disease settings. Such strategies both reduce the cost of drug development and accelerate the rate at which compounds are used in clinical settings. Critical to this goal is the need for reliable and robust biomarkers as a measure of both the efficacy and the safety of drugs. In this work, we describe potential markers, verified by multiple means, as a guide for assessing the effectiveness of using cIAP1 antagonists to downregulate NOD2 responses. Also, recognizing the fact that cIAP1 is involved in various other cell death and innate immune pathways, we hope that the data generated in this work may guide the development of compounds that are more specific to NOD2 signaling.
Compounds for the inhibition of NOD2-induced signaling have already been identified and are currently in preclinical testing (60–66; T. A. Kufer, H.-G. Schmalz; M. Krönke, J. Velder, A. Saiar, and H. Bielig, European patent application EP2353597A1, 13 January 2010). The current challenges we face are demonstration of the actual efficacy of these agents in vivo in various NOD2-hyperactive states and comparison of the advantages and disadvantages of using one inhibitor rather than another. Understanding the molecular events that are perturbed in the different settings of NOD2 overactivation may yield valuable insights and help to answer these questions. In this work, we have described a novel molecular mechanism for cIAP1-mediated positive regulation of NOD2 signaling and a mechanism for ITCH-mediated negative regulation of NOD2 signaling. We delineate a NOD2-specific, cIAP1-specific gene signature that can be exploited in drug discovery or repurposing. Lastly, we demonstrate that the repurposing of cIAP1 antagonists such as MEBS is effective in the inhibition of NOD2 signaling in the setting of ITCH deficiency.
We thank Sylvia Kertesy for technical assistance with mouse husbandry and members of the Abbott, Pizarro (CWRU), Ley (La Jolla Institute for Allergy & Immunology, La Jolla, CA), and Cominelli (CWRU) labs for helpful discussions in the course of their program project grant (P01DK091222) meetings. We thank Peter Scacheri (CWRU Department of Genetics) for helpful advice on the interpretation and presentation of the RNA-Seq analysis data.
This work was supported by NIH research grants R01GM86550-01 and P01DK091222, a Burroughs Wellcome Career Award for Biomedical Scientists (10061206.01 to D.W.A.), and an American Cancer Society postdoctoral fellowship (120209-PF-11-058-01-MPC to J.T.T.-A.). Bioinformatic support was made possible by a core utilization grant from the Clinical and Translational Science Collaborative of Cleveland to D.W.A. and J.T.T.-A., UL1TR000439 from the National Center for Advancing Translational Sciences component of the National Institutes of Health, and the NIH Roadmap for Medical Research.
Justine T. Tigno-Aranjuez designed, performed, and interpreted the experiments. Derek Abbott designed, performed, and helped to interpret the experiments. Xiaodong Bai performed the bioinformatic analysis of the transcriptome data. Justine T. Tigno-Aranjuez and Derek Abbott wrote the manuscript.
We have no conflict of interest.
Published ahead of print 29 October 2012
Supplemental material for this article may be found at http://dx.doi.org/10.1128/MCB.01049-12.