Apoptosis-associated speck-like protein containing a Caspase recruitment domain (ASC) belongs to a large family of proteins that contain a Pyrin, AIM, ASC, and death domain-like (PAAD) domain (also known as PYRIN, DAPIN, Pyk). Recent data have suggested that ASC functions as an adaptor protein linking various PAAD-family proteins to pathways involved in nuclear factor (NF)-κB and pro-Caspase-1 activation. We present evidence here that the role of ASC in modulating NF-κB activation pathways is much broader than previously suspected, as it can either inhibit or activate NF-κB, depending on cellular context. While coexpression of ASC with certain PAAD-family proteins such as Pyrin and Cryopyrin increases NF-κB activity, ASC has an inhibitory influence on NF-κB activation by various proinflammatory stimuli, including tumor necrosis factor (TNF)α, interleukin 1β, and lipopolysaccharide (LPS). Elevations in ASC protein levels or of the PAAD domain of ASC suppressed activation of IκB kinases in cells exposed to pro-inflammatory stimuli. Conversely, reducing endogenous levels of ASC using siRNA enhanced TNF- and LPS-induced degradation of the IKK substrate, IκBα. Our findings suggest that ASC modulates diverse NF-κB induction pathways by acting upon the IKK complex, implying a broad role for this and similar proteins containing PAAD domains in regulation of inflammatory responses.
Proteins containing the death domain fold (DDF)* play pivotal roles in apoptosis and inflammatory responses. The DDF represents a protein interaction motif consisting of a bundle of (usually) six antiparallel α-helices. This core structure comprises four families of evolutionarily conserved and closely related domain families, including the death domains (DDs), death effector domains (DEDs), caspase recruitment domains (CARDs), and Pyrin, AIM, Apoptosis-associated speck-like protein containing a Caspase recruitment domain (ASC), and death domain like (PAAD; also known as PYRIN, DAPIN, Pyk) domains (1–7).
Several DDF proteins are also known to participate in activation of transcription factor nuclear factor (NF)-κB, a family of dimeric transcription factors containing the Rel-homology domain. In mammals, NF-κB family members play critical roles in regulating expression of genes involved in inflammatory and immune responses, including certain cytokines, lymphokines, immunoglobulins, and leukocyte adhesion proteins (for a review, see reference 8). NF-κBs exist in the cytoplasm in inactive form sequestered by inhibitory proteins called inhibitor of NF-κB (IκB). Proteasome-dependent degradation of IκBs is linked to their phosphorylation by the IκB kinase (IKK) complex, which consists of two related kinases, IKKα and IKKβ, and a scaffold subunit IKKγ (Nemo; for a review, see reference 9). Phosphorylation of IκBs triggers their poly-ubiquitination and degradation, thus freeing NF-κB-family transcription factors to enter the nucleus and transactivate promoters of various target genes.
PAADs are found in diverse proteins implicated in apoptosis, inflammation, and cancer, though their molecular mechanisms of action are largely unknown. The founding member of the PAAD-family proteins, Pyrin, is mutated in families with Familial Mediterranean Fever, a hereditary hyperinflammatory response syndrome (10). Mutant alleles of a gene encoding another PAAD-family protein, Cryopyrin (PYPAF-1, NALP3) have been associated with familial cold auto-inflammatory syndrome, Muckle-Wells syndrome, and Chronic infantile neurological cutaneous and articular syndrome providing further hints of a role for PAADs in control of inflammatory responses (11, 12). A role for some PAAD-containing proteins in regulation of apoptosis has also been suggested (13–16).
ASC consists of a NH2-terminal PAAD followed by a COOH-terminal CARD, representing one of only two genes in the human genome that encodes proteins combining these two protein interaction domains. ASC derives its name, “apoptosis-associated speck-like protein containing a CARD” from its reported ability to trigger apoptosis when overexpressed in some tumor cell lines and from its localization to punctuate cytosolic structures (specks; reference 15). ASC is also known as TMS1 (target of methylation-induced silencing), and becomes inactive in ∼40% of breast cancers (16). Expression of ASC is found predominantly in monocytes and mucosal epithelial cells (17). Recently, it has been reported that, in transient transfection overexpression assays, coexpression of ASC with certain other PAAD-family proteins (Cryopyrin, PYPAF-7) activates NF-κB. However, the role of endogenous ASC has heretofore not been explored. We report here that the PAAD of ASC associates with the IKK complex, modulating activation of IKKα and IKKβ by cytokines and LPS. Moreover, ASC has dual properties as either an enhancer or suppressor of NF-κB, depending on which pathways for NF-κB activation are stimulated.
Materials And Methods
The complete open reading frame of ASC cDNA and segments corresponding to ASC-PAAD or -CARD were amplified by RT-PCR (Stratagene) from HL-60 cells and subcloned in sense and antisense orientation into pcDNA3 vector (Invitrogen) containing a NH2-terminal Myc epitope tag. Green fluorescent protein (GFP) fusion-protein vectors were generated by subcloning into pEGFP (CLONTECH Laboratories, Inc.). Authenticity of all constructs was confirmed by DNA sequencing. Plasmids encoding IKKα, IKKα (K44M), IKKβ, IKKβ (K44A), and IKKγ were gifts of Michael Karin (University of California, San Diego, CA), while IKKi, and TBK1 were gifts from Shizuo Akira (Osaka University, Osaka, Japan).
HEK293N Neo- or ASC-PAAD stable expressing cells were treated with 20 ng ml−1 TNFα for 4 h, lysed in Trizol reagent (GIBCO BRL), and total RNA was isolated according to the instructions provided by the manufacturer. DNase I–treated RNA (1 μg) was transcribed into single-stranded cDNA using Superscript II (GIBCO BRL) and amplified for 30 cycles using Amplitaq (CLONTECH Laboratories, Inc.) with either TRAF1 or GAPDH specific primers.
Cell Culture, Transfection, and Reporter Gene Assays.
HEK293N, HEK293T, and MCF7 cells were cultured in DMEM, while THP-1 cells were cultured in RPMI 1640 medium, supplemented with 10% heat-inactivated FBS. When indicated, cells were treated with 10 to 20 ng ml−1 TNFα or IL-1β, or 600 ng ml−1 LPS for various times. Transfection of HEK293 cells was accomplished using Superfect (QIAGEN), while THP-1 (107) cells were transfected with Lipofectamine Plus (Life Technologies), holding total DNA content constant. At 48 h after transfection, stable THP-1 cells or 293N cells were selected in 650 or 800 μg ml−1 G418 (Calbiochem), respectively. MCF7 and THP-1 cells were transfected with double-strand siRNAs on two consecutive days using Oligofectamine (Life Technologies). siRNA1: 5′-UCAUCCUGAAUCUGAUCUUdTdT-3′, 5′-AAGAUCAGAUUCAGGAUGAdTdT-3′, siRNA2: 5′-GAUGCGGAAGCUCUUCAGUdTdT-3′, 5′-ACUGAAGAGCUUCCGCAUCdTdT-3′, siRNAcntr: 5′-CAAGUAUUUGACGACCGAGdTdT-3′, 5-CUCGGUCGUCAAAUACUUGdTdT-3′.
For NF-κB reporter gene assays, typically, 105 cells cultured in 24-well plates in 5% serum were transfected with a total of 1 μg plasmid DNA (normalized for total DNA), including 100 ng of pNF-κB-LUC, pAP1-LUC, or p53-LUC (CLONTECH Laboratories, Inc.) and 6 ng of a Renilla luciferase gene driven by a constitutive TK promoter (pRL-TK; Promega). Lysates were analyzed using the Dual Luciferase kit (Promega).
For immunoprecipitations, cells were lysed in isotonic lysis buffer (150 or 500 mM NaCl, 20 mM Tris/HCl [pH 7.4], 0.2% NP-40, 12.5 mM β-glycerophosphate, 2 mM NaF, 200 μM to 1 mM Na3VO4, 1 mM PMSF, and 1× protease inhibitor mix [Roche]), using 2–8 × 107 cells for endogenous proteins. Clarified lysates were subjected to immunoprecipitation using agarose-conjugated anti-c-Myc (Santa Cruz Biotechnology, Inc.), or protein-G–conjugated anti-IKKβ (Santa Cruz Biotechnology, Inc.), anti-IKKα (BD Biosciences), or anti-ASC antibodies (17). After incubation at 4°C for 4–12 h, immune-complexes were washed three times in lysis buffer, separated by SDS/PAGE, and analyzed by immunoblotting using various antibodies as above in conjunction with ECL detection system (Amersham Biosciences). Alternatively, lysates were directly analyzed by immunoblotting after normalization for total protein content. Anti-Tubulin and anti-β-Actin antibodies were purchased from Sigma-Aldrich, and anti–ICAM-1 and anti-GFP antibodies were purchased from Santa Cruz Biotechnology, Inc.
IKKα or IKKβ were immunoprecipitated from cell lysates, using 5 × 105 cells for IKK transfectants and 106 cells for endogenous IKKs. Immune-complexes were washed twice in lysis buffer (as above), once in lysis buffer containing 2 M urea followed by two washes in kinase buffer (20 mM Hepes [pH 7.6], 50 mM NaCl, 20 mM β-glycerophosphate, 1 mM Na3VO4, 0.5 mM DTT), equilibrated for 5 min in kinase buffer, adjusted to 10 mM MgCl2 and 1 mM DTT, and finally incubated in 20 μl kinase buffer supplemented with 35 μM ATP, 5 μCi γ[32P] ATP and 1 μg glutathionine-S-transferase (GST)-IκBα (Santa Cruz Biotechnology, Inc.) at 30°C for 30 min (18).
NF-κB DNA-binding Activity Assays.
Electromobility gel-shift assays (EMSA) were used to measure NF-κB DNA-binding activity, essentially as described (19). Briefly, 106 cells, either untreated or treated with TNFα for 20 min were lysed in buffer A (10 mM Hepes, pH 8.0, 0.5% NP-40, 1.5 mM MgCl2, 10 mM KCl, 0.5 mM DTT, and 200 mM sucrose), washed twice in buffer A, and pelleted nuclei were incubated in 1× packed cell volume of buffer B (20 mM Hepes, pH 7.9, 1.5 mM MgCl2, 420 mM NaCl, 0.2 mM EDTA, and 1 mM DTT) overnight, clarified supernatants diluted 1:1 in buffer C (20 mM Hepes, pH 7.9, 100 mM KCl, 0.2 mM EDTA, 20% glycerol, and 1 mM DTT). Protease and phosphatase inhibitors were added to all buffers. Nuclear extracts (2 μg) were incubated with 10 fmole of a 32P-end-labeled double-strand consensus NF-κB oligonucleotide (Promega) probe with or without 2 μg of anti-p65 antibody or control IgG (Santa Cruz Biotechnology, Inc.). For competition assays, a 50× molar excess of unlabeled oligonucleotide was added. DNA–protein complexes were separated by nondenaturing PAGE, and analyzed by autoradiography.
Cells were transferred to 4-well polylysine-coated chamber slides (LabTec), fixed in 4% paraformaldehyde, stained with 0.4 μg ml−1 of the indicated antibodies (Santa Cruz Biotechnology, Inc.), followed by 4 μg ml−1 FITC and TRITC labeled secondary antibodies (DakoCytomation/Molecular Probes). Both secondary antibodies were combined and used for each well in 0.1% BSA and 1% serum. Cells were analyzed by confocal laser-scanning microscopy (Bio-Rad Laboratories).
ASC Differentially Modulates NF-κB Activity, Depending on the Stimulus.
Recently, it was reported that coexpression of ASC with Cryopyrin (PYPAF-1/NALP3) or PYPAF-7 (PAN6) induces NF-κB activity in transient transfection reporter gene assays performed in HEK293T cells (20, 21). These cells contain essentially no detectable ASC (unpublished data), thus avoiding contributions of the endogenous ASC protein. Similar to previous reports, we observed 20–70-fold inductions in NF-κB activity, when ASC was coexpressed with Cryopyrin or Pyrin in HEK293T cells, as measured by reporter gene assays in HEK293T cells (Fig. 1 A). In contrast, neither ASC, nor Pyrin or Cryopyrin alone induced significant NF-κB activity. The ability of ASC to collaborate with other PAAD-containing proteins in NF-κB induction was selective, as coexpression with NAC (NALP1, DEFCAP, CARD7), PAN1 (PYPAF-2, NALP2, NBS1), or PAN2 (PYPAF-4, NALP4) did not result in significant NF-κB activity.
During studies of the effects of ASC on NF-κB induction, we noticed that expression of this protein reduced NF-κB activity in HEK293N and HEK293T (Fig. 1) and other cell lines (unpublished data) after stimulation with cytokine TNFα. For example, TNFα-induced increases in NF-κB activity were reduced by approximately half in ASC-transfected compared with control transfected HEK293T cells (Fig. 1 B). Not only ASC, but all proteins which contain PAAD-domains were found to have an inhibitory influence on TNFα-mediated induction of NF-κB activity in these reporter gene assay experiments, including NAC, PAN1, PAN2, and to a lesser extent Pyrin, and Cryopyrin. When ASC was coexpressed with Pyrin or Cryopyrin (but not NAC, PAN1, or PAN2), the net levels of NF-κB activity returned to control levels seen in TNFα-stimulated cells (Fig. 1 B).
To extend these studies to other types of stimuli known to induce NF-κB activity, we performed similar reporter gene assays, stimulating cells with the cytokine IL-1β or transfecting cells with NF-κB–inducing proteins such as Bcl-10 (which has been implicated in NF-κB induction by B cell and T cell antigen receptors; reference 22) and Nod-1 (a putative intracellular sensor of LPS; reference 23). In every case examined, transient overexpression of ASC suppressed NF-κB activity (Fig. 1, C–E). Similar results were obtained for several cell lines, including HEK293T, HEK293N, HeLa, Cos7, and HT1080 cells (unpublished data). The effects of ASC on NF-κB induction were specific, in as much as overexpression of ASC did not interfere with transcriptional activation of p53, AP-1, β-Catenin/Tcf, as measured by reporter gene assays (Fig. 1, F and G; and unpublished data). Immunoblotting experiments confirmed production of all plasmid-derived proteins, excluding a general effect of ASC on protein synthesis or stability (see below).
ASC Inhibits NF-κB Induction via Its PAAD.
To map the domain in ASC responsible for modulation of NF-κB activity, we compared the effects of full-length ASC to truncation mutants containing only the PAAD or CARD domains. Neither the PAAD nor the CARD of ASC induced NF-κB activity when expressed by transient transfection in cell lines such as HEK293N (Fig. 2 A), consistent with a previous report (24). In TNFα-stimulated cells, both full-length ASC and the PAAD of ASC profoundly suppressed NF-κB activity, while the CARD of ASC did not (Fig. 2 B). Inhibition of TNFα-induced NF-κB activity by the PAAD of ASC was dose-dependent (Fig. 2 C). NF-κB DNA binding activity was also reduced by overexpression of the PAAD of ASC, as measured by EMSAs using a DNA probe containing NF-κB binding sites (Fig. 2 D), and correlated with reduced translocation of the p65 subunit of NF-κB into the nuclei of TNFα-stimulated cells, as visualized by immunofluorescence microscopy analysis (unpublished data).
To further explore the role of the PAAD of ASC in modulating NF-κB activity, we generated stable transfectants of HEK293N cell expressing the ASC-PAAD protein. The levels of ASC-PAAD protein produced in the stable transfectants were determined to be comparable to the levels of ASC found endogenously in some types of cells, such as THP-1 monocytic cells in which we have observed that LPS stimulates expression of endogenous ASC mRNA and protein (Fig. 2 E). Clones of HEK293N stably expressing ASC-PAAD demonstrated ∼80% reduced activation of a transfected NF-κB reporter gene in response to TNFα (Fig. 2 E), consistent with expectations from transient transfection experiments. Moreover, this effect on NF-κB activity in ASC-PAAD expressing HEK293N cells was confirmed to be ASC-dependent by introducing of an ASC-antisense plasmid, in which a cDNA fragment corresponding to ASC-PAAD was subcloned in reverse orientation downstream of a GFP open reading frame (Fig. 2 F). Transfection of various amounts of this ASC-antisense plasmid induced concentration-dependent decreases in the amount of Myc-ASC-PAAD protein, correlating with dose-dependent restoration of NF-κB activity after TNFα stimulation.
Next, we evaluated whether stable expression of ASC-PAAD interfered with TNFα-induced expression of an endogenous NF-κB target gene, TRAF1, which contains several NF-κB binding sites in its promoter (25). While TNFα increased TRAF1 protein and mRNA, this response was markedly blunted in ASC-PAAD–expressing cells (Fig. 2, G and H). Analysis of control proteins (TRAF2, Tubulin) and control mRNA (GAPDH) demonstrated the specificity of these results.
To extend these studies yet to another cell line, THP-1 monocytic cells were stably transfected with plasmids encoding either GFP or GFP fused to ASC-PAAD. LPS induced expression of endogenous NF-κB–inducible gene, ICAM-1 (26), in control GFP-expressing THP-1 cells, producing a >20-fold increase in ICAM-1 protein within 10 h (Fig. 2 I). In contrast, ICAM-1 was markedly reduced in ASC-PAAD–expressing cells. Comparable expression of GFP-ASC-PAAD and GFP was confirmed by immunoblotting with anti-GFP antibody (Fig. 2 I). Taken together, these data indicate that ASC is capable of suppressing TNFα- and LPS-induced expression of endogenous NF-κB target genes through its PAAD.
ASC-PAAD Modulates NF-κB Induction at the Level of the IKK Complex.
To map where the PAAD of ASC affects the NF-κB activation pathway, NF-κB activity was induced by transient transfection of plasmids encoding various intracellular signal-transducers that operate within cytokine receptor pathways leading to phosphorylation of IκB, a key event required for NF-κB release. Coexpression of ASC blocked induction of NF-κB activity by the adaptor proteins TRAF2 and TRAF6, the TRAF-binding kinases TBK1 and NIK, the IKK complex constituents IKKα and IKKβ, and the related kinase IKKi (Fig. 3 A). In contrast, coexpression of ASC did not suppress reporter gene activation induced by NF-κB-p65. Thus, ASC blocks upstream of NF-κB, apparently at the level of the IKK complex.
To directly evaluate the effects of ASC on IKK activity, in vitro kinase assays were performed. For initial experiments, either HA-epitope tagged IKKα (Fig. 3 B) or IKKβ (Fig. 3 C) was expressed alone or in combination with ASC or ASC-PAAD. Then HA-IKKα or HA-IKKβ was immunoprecipitated from cell lysates, and in vitro phosphorylation of GST-IκBα substrate was measured. As negative control, kinase-dead IKKs were used. TNFα induced ∼5–10-fold increases in IKKα and IKKβ activity, while ASC or ASC-PAAD expression suppressed IKKα and IKKβ activity to essentially baseline levels (Fig. 3, B and C). This inhibitory effect of ASC and ASC-PAAD on IKKα and IKKβ activity was not attributable to a difference in the total levels of the IKKα or IKKβ proteins, as determined by immunoblot analysis, thus confirming that ASC suppresses activation of the IKK complex. ASC also suppressed autophosphorylation of IKKα (Fig. 3 B) and IKKβ (unpublished data), as well as phosphorylation of associated endogenous IκBα, in these kinase assays. Similar results were obtained when IKKα or IKKβ was activated in cells by transient transfection of plasmids encoding intracellular signaling proteins such as TRAF2 and TRAF6 (Fig. 3 D, and unpublished data).
To extend these studies involving expression of epitope-tagged proteins by transient transfection, the activity of endogenous IKKα was evaluated using HEK293N cells stably transfected with either control or Myc-ASC-PAAD (Fig. 3 E). ASC-PAAD potently suppressed TNFα-induced activation of endogenous IKKα in these cells, as determined by kinase assays where immunoprecipitated IKKα was tested for ability to phosphorylate GST-IκBα substrate in vitro. Autophosphorylation of IKKα was also suppressed in ASC-PAAD–expressing cells. These differences in IKKα activity were not due to differences in the total levels of IKKα protein, as determined by immunoblotting (Fig. 3 E).
ASC Associates with IKKα and IKKβ.
Having mapped the site of action of ASC to the IKK complex, we performed experiments to explore whether ASC associated with these protein kinases. In the course of our studies of ASC, we observed that expression of ASC is induced in myeloid-lineage hematopoietic cells such as THP-1 monocytic or HL-60 monomyelocytic leukemia cell lines by LPS and TNFα (Fig. 2 E, and unpublished data). We therefore asked whether endogenous ASC protein could be found associated with endogenous IKK complex components after LPS- or TNFα-stimulation in these cells. Co-IP experiments provided evidence of association of ASC with both IKKα and IKKβ in LPS-stimulated THP-1 and TNFα-treated HL-60 cells (Fig. 4 A, and unpublished data). These protein interactions were reciprocally demonstrable, regardless of whether immune-complexes were prepared using anti-IKKα and anti-IKKβ antibodies (followed by immunoblotting with anti-ASC antiserum) or using anti-ASC antiserum (followed by immunoblotting with anti-IKKα or anti-IKKβ antibodies; Fig. 4 A, and unpublished data). We also analyzed the HEK293N-ASC-PAAD stable transfectants to determine whether the PAAD is sufficient for association with the endogenous IKK complex. Again, coimmunoprecipitation (coIP) experiments demonstrated specific interaction of ASC-PAAD with the endogenous IKK complex, as indicated by the association of ASC with IKKα, IKKβ, as well as IKKγ (Fig. 4 B).
In addition, immunofluorescence microscopy was used to examine the intracellular location of IKKα and IKKβ in HEK293-Neo and -ASC-PAAD cells (Fig. 4, C and D). In HEK293-Neo cells, endogenous IKKα and IKKβ were diffusely distributed throughout the cytosol. In contrast, IKKα and IKKβ relocated to filament-like structures in HEK293N-ASC-PAAD cells, colocalizing with the ASC-PAAD proteins. Treatment of these cells with TNFα did not affect the colocalization of ASC-PAAD with IKKs, nor did it affect the location of IKKs in control cells lacking ASC-PAAD (unpublished data). We conclude therefore that the PAAD of ASC associates with endogenous IKKα- and IKKβ-containing protein complexes, altering their intracellular location and suppressing cytokine- and LPS-mediated activation of these protein kinases involved in NF-κB induction.
Suppression of Endogenous ASC Expression Enhances IκBα Degradation.
As IKK activity induces phosphorylation and subsequent degradation of IκB-family proteins, we evaluated the effects of the PAAD of ASC on levels of endogenous IκBα in these stably transfected cells, before and at various times after TNFα stimulation. Immunoblot analysis of lysates from HEK293N-Neo cells using anti-IκBα antibody demonstrated the appearance of a doublet band (indicative of phosphorylation of IκBα) within 5 min after TNFα simulation, followed by disappearance of IκBα protein. In contrast, IκBα protein levels were sustained at detectable levels in ASC-PAAD–expressing HEK293N cells despite TNFα treatment. Furthermore, the IκBα doublet band indicative of phosphorylation was not observed until much later, at 15–30 min after stimulation (Fig. 5 A), thus demonstrating a marked delay relative to control cells.
Though the expression of the PAAD domain of ASC in cells interfered with IKK activation (Fig. 3) and IκBα degradation (Fig. 5 A) in response to proinflammatory stimuli, these results could reflect a dominant-negative effect of this protein domain. Consequently, the full-length ASC protein might operate as an agonist rather than antagonist of IKK activation. To explore the role of endogenous ASC, we used the technique of small interfering RNA (siRNA) to reduce ASC expression in cells, assessing the impact on TNFα- and LPS-mediated degradation of endogenous IκBα by immunoblotting (for a review, see reference 27). For these experiments, either a 19 bp double-strand RNA corresponding to ASC nucleotides 721–741, or to nucleotides 474–494 was introduced into either MCF7 epithelial cells or THP-1 monocytic cells. Both siRNAs directed against ASC mRNA substantially reduced ASC protein levels in MCF7 cells and siRNA (474–494) also to a lesser extent in differentiated THP-1 cells, while a control siRNA molecule (c) did not (Fig. 5, B and C). These experiments revealed that an average siRNA-mediated reduction in endogenous ASC protein levels of ∼60% was associated with enhanced IκBα degradation in cells stimulated with TNFα or LPS averaging ∼23% compared with control siRNA transfected cells. Thus, reductions of ASC were associated with enhanced signaling by TNFα and LPS, allowing us to conclude that endogenous ASC functions as a suppressor of the NF-κB pathway, at least in some cellular contexts.
PAAD (PYRIN, Dapin) domains are found in multiple genes within the human genome, including several implicated in hereditary hyperinflammation syndromes, interferon responses, cancer suppression, and apoptosis induction (3). Certain PAADs are capable of homotypic interactions with themselves or other members of the PAAD family (20, 28, 29), suggesting opportunities for creating protein interaction networks that link various signaling pathways and permit fine-tuning of responses. Recently, the PAAD of ASC has been reported to bind the corresponding PAADs of Pyrin and Cryopyrin, which are encoded by the causative genes involved in Familial Mediterranean Fever and Familial Cold Autoinflammatory Syndrome, Muckle-Wells syndrome, and Chronic infantile neurological cutaneous and articular syndrome, respectively (20, 28). Furthermore, ASC reportedly collaborates with Cyropyrin and PYPAF-7 in inducing NF-κB activity, at least in transient transfection experiments in HEK293T cells (20, 21), requiring overexpression of both ASC and these other PAAD-family proteins. However, as shown here, ASC does not collaborate with all PAAD-family proteins in inducing NF-κB activity, and its overexpression is associated with suppression rather than enhancement of NF-κB activity in cells stimulated with proinflammatory cytokines (TNFα, IL-1β) or LPS. Given our observation that the PAAD of ASC associates with and suppresses components of the IKK complex, it is possible that Cryopyrin interaction with ASC dislodges ASC from the IKKs, relieving endogenous suppression of these kinases, and permitting NF-κB activation. Alternatively, PAAD-containing proteins such as Cryopyrin, which are thought to self-oligomerize via a nucleotide-binding NACHT domain (30), might employ ASC as an adaptor for bridging to the IKK complex, achieving kinase activation through an induced proximity mechanism (31). The ultimate impact of ASC interactions with other PAAD-family proteins may therefore depend on their relative ratios, where ASC can function as either an inhibitor or activator of IKK, depending on cell context and on the stimulus used to engage pathways leading to the IKK complex.
We propose therefore that ASC is a dual modulator of NF-κB activation, which by virtue of its association with IKKs, acts at a point of convergence of multiple pathways leading to NF-κB induction. The ability of ASC to either enhance or inhibit NF-κB induction, depending presumably on the ratio of its levels relative to other ASC-binding proteins, is reminiscent of proteins such as c-FLIPL, which can function as either a pro-Caspase-8 activator or inhibitor, dependent on cell context (32). Similarly, some IAP-family proteins can either enhance or inhibit NF-κB induction by TNFα, depending apparently on whether they induce degradation of certain associated proteins (33, 34), but a combination of both stimulatory and inhibitory properties has not been attributed thus far to a single protein (e.g., cIAP1 inhibits; cIAP2 enhances). Thus, ASC may represent the first identified protein that has dual properties as both an inhibitor and enhancer of NF-κB induction. Though other interpretations are possible, this two-sided nature of ASC is entirely consistent with its hypothesized role as a molecular bridge involved in assembly of multiprotein complexes (molecular machines), in which the correct stoichiometry of components would be necessary for activity and where either insufficiency or excess of ASC could interfere with complex assembly.
Evidence is presented here that ASC associates with components of the IKK complex, the kinase complex responsible for phosphorylation of the IκB family proteins that sequester NF-κB in the cytosol (for a review, see reference 35). In pilot experiments, we could not demonstrate interaction of ASC with constituents of the IKK complex by ectopic overexpression of ASC with epitope-tagged IKKα, IKKβ, or IKKγ individually, as determined by coIP experiments (unpublished data). Thus, we favor the idea that ASC associates directly or indirectly with the assembled IKK complex. However, ectopic overexpression of single components of the IKK complex might disrupt the complex because of changes in protein stoichometry, thus preventing ASC binding. Interestingly, Chen et al. recently demonstrated the existence of several additional proteins associated with endogenous IKK complexes (36). It remains to be determined whether ASC associates with IKK complexes through one of these proteins. Also, the molecular events that govern physical and functional interactions of ASC with the IKK complex remain to be clarified, including the possible role of posttranslational modifications of ASC or other associated proteins. Thus, it is unclear at present how the PAAD of ASC suppresses IKK activation in response to proinflammatory stimuli.
Recently, we determined that another PAAD-family protein, PAN2, can also associate with IKKα and suppress IKKα activation by TNFα (19). Interestingly, however, PAN2 does not appear to associate with IKKβ or IKKγ, suggesting the possibility of differences in interactions with IKK complex components compared with ASC, which could be coimmunoprecipitated with either anti-IKKα or anti-IKKβ antibodies. Similar to ASC, however, the PAAD of PAN2 is sufficient for interactions with and suppression of IKKα (19).
Though we used the PAAD domain of ASC as a probe to demonstrate the ability of this region of ASC to functionally and physically interact with IKK complex components, it should be noted that the human genome contains at least two genes predicted to encode PAAD-only proteins (reference 3, and unpublished data), which are analogous to the ASC-PAAD protein employed here in our studies. Furthermore, we have observed that these proteins function very similar to the PAAD of ASC in their effects on IKK and NF-κB induction (unpublished data). Some poxviruses also contain potential ORFs encoding proteins with significant sequence similarity to cellular PAADs, such as the rabbit myxoma virus. Thus, endogenous and viral proteins consisting of only the PAAD domain may operate as negative regulators of IKK activation, analogous to our studies of a fragment of ASC comprising only the PAAD domain. Just as the mechanism by which ASC suppresses IKK activation induced by proinflammatory stimuli is presently unknown, similarly, it remains unclear how ASC enhances NF-κB induction when coexpressed with PAAD-family proteins such as Pyrin, (this paper), Cyropyrin (20), and PYPAF-7 (21). ASC has been reported to recruit Pyrin, Cryopyrin, and PYPAF-7 into cytosolic specks (20, 21, 28), suggesting a role for these intracellular bodies in the process of NF-κB induction, but the location of IKK complex proteins under these circumstances has not been assessed. Future studies should therefore address the consequences of Pyrin, Cryopyrin, and PYPAF-7 protein interactions with ASC with regards to association with and regulation of the IKK complex.
Interestingly, ASC associates with uncharacterized structures in the cytosol of cells, forming specks. The formation of speck-like structures is not merely an artifact of protein overexpression, because they can be identified by immunohistochemical techniques in normal tissue (17), and because certain treatments of cultured cells can induce speck formation by endogenous ASC (15). Indeed, the endogenous ASC protein was first discovered because of its association with Triton X-100–insoluble aggregates in HL-60 cells pretreated with retinoic acid (15). The targeting of ASC to these locations requires the combination of PAAD and CARD (unpublished data), and truncation mutants of ASC containing only the PAAD form filament-like structures in the cytosol of cells, but fail to produce the speck-like morphology for which this protein was named. In ASC-PAAD–expressing cells, IKK components colocalized with these filaments, which form in a manner reminiscent of previously identified NF-κB regulators, such as TRADD, RIP, and Bcl-10 (37, 38). Given recent suggestions that the ASC-binding protein Pyrin associates with cytoskeletal proteins, it is tempting to speculate the ASC may associate with or coordinate formation of a specialized site on the cytoskeleton (39). The fate of proteins recruited to these uncharacterized complexes where ASC localizes is unknown. We have seen no evidence that ASC overexpression results in degradation of ASC-interacting proteins. Possibly speck-like structures targeted by ASC are sites for posttranslation protein modifications or simply providing a location for sequestering certain proteins.
ASC was originally reported to induce apoptosis when overexpressed in certain tumor lines (15, 24). However, at the doses of ASC-encoding plasmid employed and levels of ASC expression attained in our experiments, we did not observe apoptosis. Based on our findings, we propose that ASC might modulate apoptosis under some circumstances where NF-κB is important for avoiding cell death, given that NF-κB can regulate expression of apoptosis-relevant genes such as A20, Bcl-XL, Bfl-1, cIAP2, and others (for a review, see reference 40). In this regard, inhibition of IKKs is known to sensitize cells to apoptosis induction by TNF-family death ligands (41). These effects of ASC on apoptosis might account for the observation that this gene is commonly silenced in breast cancers by gene methylation (16).
Because expression of ASC is initially low but inducible by LPS and TNFα in THP-1 monocytic cells, we speculate that ASC may be involved in a negative feedback suppression of pathways that induce NF-κB. This inducible expression in response to a variety of proinflammatory stimuli was also recently demonstrated in neutrophils (42). In this way, ASC could play a role in terminating inflammatory responses, thus ensuring that only a short burst of NF-κB activity activation occurs. This scenario is consistent with our siRNA results where reductions in ASC were correlated with enhanced IκBα degradation. An alternative but not mutually exclusive possibility is that antagonism of the TNFα, IL-1β, and LPS pathways for NF-κB induction by ASC reflects a competition between alternative pathways for access to IKK or IKK-associated proteins. In this regard, the physiological or pathogenic stimuli that normally engage ASC and other PAAD-family proteins in NF-κB induction are unknown, but at least 14 members of the PAAD family have been identified that contain leucine rich repeats (LRRs) similar to those found in the extracellular domains of Toll-related receptors (TLRs) and in the CARD-family proteins Nod1 and Nod2, which are known to bind bacterial LPS or other molecules made by microbial pathogens. Thus, while speculative, many PAAD-family proteins may participate in a pathway that senses intracellular bacteria. Cross talk between this pathway and other NF-κB activation pathways, such as those triggered by TNFα, IL-1β, TLRs, and antigen receptors (TCR, BCR), may play important roles in steering innate and acquired immune responses toward different ultimate outcomes. Future studies, including targeted ablation of the gene encoding ASC in mice, will reveal the overall importance of ASC in inflammation and innate immunity.
We thank Drs. H. Chan, J.S. Damiano, Y. Kim, and J.M. Zapata for discussions, A. Stassinopoulos for excellent technical support, D. Chaplin, M. Karin, J. Tschopp, and J. Yuan for valuable reagents, and acknowledge J. Valois and R. Cornell for manuscript preparation.
C. Stehlik and L. Fiorentino are recipients of fellowships from the Austrian Science Foundation (FWF, J1809-Gen/J1990-Gen), Department of Defense (DOD) Breast Cancer Research Program (DAMD17-01-1-0171), and the American-Italian Cancer Foundation, respectively.
C. Stehlik and L. Fiorentino contributed equally to this work.
Abbreviations used in this paper: ASC, apoptosis-associated speck-like protein containing a Caspase recruitment domain; coIP, coimmunoprecipitation; CARD, Caspase-recruitment domain; DDF, death domain fold; EMSA, electromobility gel-shift assays; GFP, green fluorescent protein, GST, glutathionine-S-transferase; IκB, inhibitor of NF-κB; IKK, IκB kinase; NF, nuclear factor; PAAD, Pyrin, AIM, ASC, and death domain like; PAN, PAAD and NACHT containing protein.