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
), 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
). Furthermore, ASC reportedly collaborates with Cyropyrin and PYPAF-7 in inducing NF-κB activity, at least in transient transfection experiments in HEK293T cells (20
), 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
), 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
), 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
). 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
). 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.