ABIN-1 Binds to NEMO/IKKγ and Co-operates with A20 in Inhibiting NF-κB*

Nuclear factor κB (NF-κB) plays a pivotal role in inflammation, immunity, stress responses, and protection from apoptosis. Canonical activation of NF-κB is dependent on the phosphorylation of the inhibitory subunit IκBα that is mediated by a multimeric, high molecular weight complex, called IκB kinase (IKK) complex. This is composed of two catalytic subunits, IKKα and IKKβ, and a regulatory subunit, NEMO/IKKγ. The latter protein is essential for the activation of IKKs and NF-κB, but its mechanism of action is not well understood. Here we identified ABIN-1 (A20 binding inhibitor of NF-κB) as a NEMO/IKKγ-interacting protein. ABIN-1 has been previously identified as an A20-binding protein and it has been proposed to mediate the NF-κB inhibiting effects of A20. We find that both ABIN-1 and A20 inhibit NF-κB at the level of the IKK complex and that A20 inhibits activation of NF-κB by de-ubiquitination of NEMO/IKKγ. Importantly, small interfering RNA targeting ABIN-1 abrogates A20-dependent de-ubiquitination of NEMO/IKKγ and RNA interference of A20 impairs the ability of ABIN-1 to inhibit NF-κB activation. Altogether our data indicate that ABIN-1 physically links A20 to NEMO/IKKγ and facilitates A20-mediated de-ubiquitination of NEMO/IKKγ, thus resulting in inhibition of NF-κB.

NF-B is a ubiquitously expressed family of transcription factors that controls the expression of numerous genes involved in immune and inflammatory responses (1). NF-B also plays an important role during cellular stress responses, due to its anti-apoptotic and proliferationpromoting functions (2). Aberrant activation of NF-B is a major hallmark of several inflammatory diseases such as arthritis (3,4), and a variety of human cancers (5,6). In resting cells, NF-B is sequestered in the cytoplasm in an inactive form by members of the inhibitory family of IB proteins (1). Various stimuli including pathogens, pathogen-related factors, and cytokines lead to phosphorylation of the inhibitory subunit IB␣ on specific serine residues (Ser 32 and Ser 36 ) (7) catalyzed by two IB kinases (IKKs), 3 namely IKK␣ and IKK␤ (8 -12). This step marks the IB protein for ubiquitination and subsequent degradation through a proteasome-dependent pathway (1). The active NF-B is then free for translocation to the nucleus, where it binds the B sequences present in the promoters of responsive genes.
IKK␣ and IKK␤ reside in a larger kinase complex (700 -900 kDa), called the IB kinase complex (IKK complex), that also contains the essential regulatory subunit NEMO (also known as IKK␥) (13,14). Genetic studies suggest that NEMO/IKK␥ is absolutely required for the activation of IKKs and NF-B in response to different stimuli (13,15). NEMO/IKK␥ contains several coiled-coil domains, a leucine zipper, and a C-terminal zinc finger domain. These motifs are required for the correct assembly of the IKK complex (13) and recruitment of upstream signaling mediators (16). Numerous proteins have been demonstrated to interact with NEMO/IKK␥, as the kinase RIP and the inhibitor of NF-B A20 (17), the viral trans-activator TAX (18 -20), and the adaptor proteins CIKS/Act-1, TANK, and CARMA (21)(22)(23). Therefore, NEMO/IKK␥ represents the point where most NF-B signaling pathways converge. Despite this information, the molecular mechanism regulating IKK complex function is not fully understood.
Ubiquitin conjugation has been most prominently associated with protein degradation through a proteasome-dependent pathway, but it is becoming increasingly evident that ubiquitination plays a key role in the signal transduction pathway leading to activation of NF-B (24,25). Recent reports show that lysine 63-linked ubiquitination of NEMO/ IKK␥ is an important step for the activation of IKKs and NF-B following various stimuli, such as TNF, lipopolysaccharide, and antigen receptor (26 -28). In contrast, the tumor suppressor CYLD is reported as a negative regulator of NF-B by specific de-ubiquitination of NF-B signaling molecules, such as TRAF2, TRAF6, and NEMO/IKK␥ (29). Also A20 functions as an inhibitor of the NF-B pathway by removing Lys 63linked ubiquitin chains from RIP, an essential mediator of the proximal TNF-Receptor-1 signaling complex. Then A20 targets RIP for Lys 48linked polyubiquitination and proteasomal degradation (30). Furthermore, A20 terminates Toll-like receptor-induced NF-B signaling, by cleaving ubiquitin chains from TRAF6 (27). The central role played by A20 in terminating NF-B activation is further demonstrated by the fact that A20 Ϫ/Ϫ mice develop severe inflammation and cachexia, are hypersensitive to both lipopolysaccharide and TNF, and die prematurely (27,31). Here we used NEMO/IKK␥ as bait in yeast two-hybrid screening, and identified ABIN-1 (A20 binding inhibitor of NF-B) as a NEMO/ IKK␥-interacting protein.
Yeast Two-hybrid Screening-The cDNA encoding the N-terminal part of mouse NEMO/IKK␥ (amino acids 1-311) was cloned in-frame into the GAL-4 DNA-binding domain vector pGBKT7 (Clontech). The resulting plasmid pGBKT7-NEMO/IKK␥ was used as bait in a yeast two-hybrid screen of a human liver cDNA library (Clontech) in Saccharomyces cerevisiae strain AH109. The NEMO/IKK␥ deletion mutants for two-hybrid mapping were made by conventional PCR and cloned into the pGBKT7 vector.
Gel Filtration of Cellular Extracts-Gel filtration procedures were performed as previously described (42). Fractions were analyzed by Western blotting for ABIN-1, NEMO/IKK␥, and IKK␤.
In Vitro Translation and GST Pull-down Assays-In vitro transcription and translation were carried out with 1 g of ABIN-1 constructs according to the TNT Quick Coupled Transcription/Translation System protocol (Promega) in the presence of [ 35 S]methionine.
GST-NEMO/IKK␥ fusion protein was produced and purified as described (33). GST pull-down assays were performed by incubating an aliquot of GST-NEMO/IKK␥ bound to glutathione-Sepharose beads (Amersham Biosciences) together with 10 l of in vitro translated ABIN-1 protein in phosphate-buffered saline, 1% Triton X-100 buffer (including Complete Protease Inhibitor mixture (Roche)) for 2 h at 4°C. Beads were then washed five times with the same buffer, resuspended in Laemmli buffer, and run on a SDS-polyacrylamide gel before autoradiography.
Transfection, Immunoprecipitation, and Luciferase Assay-Lipofectamine-mediated transfections were performed according to the manufacturer's instructions (Invitrogen). All transfections included supplemental empty vector to ensure that the total amount of transfected DNA was kept constant in each dish culture.
For immunoprecipitation of transfected proteins, HEK293 cells (3 ϫ 10 6 ) were transiently transfected and 24 h after transfection cells were lysed in Triton X-100 lysis buffer (20 mM Hepes, pH 7.4, 150 mM NaCl, 10% glycerol, 1% Triton X-100, and Complete Protease Inhibitor mixture). After an additional 15 min on ice, cell extracts were centrifuged for 10 min at 14,000 ϫ g at 4°C and supernatants were incubated for 4 h at 4°C with anti-FLAG antibodies bound to agarose beads (M2, Sigma). The immunoprecipitates were washed five times with Triton X-100 lysis buffer and subjected to SDS-PAGE.
For luciferase assay, HEK293 cells (4 ϫ 10 5 ) were seeded in 6-well plates. After 12 h cells were transfected with 0.5 g of Ig-B-luciferase reporter plasmid and various combinations of expression plasmids. 24 h after transfection, cells were stimulated with TNF-␣ for 3 h or left untreated. Cell extracts were prepared and reporter gene activity was determined via the luciferase assay system (Promega). Expression of the pRSV-␤-galactosidase vector (0.2 g) was used to normalize transfection efficiencies. In Vivo Ubiquitination and De-ubiquitination Assays-HEK293 cells (3 ϫ 10 6 ) were co-transfected with expression vectors containing epitope-tagged ubiquitin (1 mg) and NEMO/IKK␥ (200 ng), plus various constructs encoding A20 or ABIN-1 proteins. 24 h after transfection, cell lysates were prepared as above and analyzed for polyubiquitination of NEMO/IKK␥ either by Western blot anti-NEMO/IKK␥ (-FLAG) on total extracts or by immunoprecipitating FLAG-NEMO/IKK␥ with anti-FLAG beads followed by Western blot anti-HA-ubiquitin.

ABIN-1 Binds to NEMO/IKK␥-
The regulatory subunit of the IKK complex, NEMO/IKK␥, has an essential role in NF-B activation. To gain insights into how NEMO/IKK␥ modulates the activation of NF-B, we screened a human liver cDNA library for NEMO/IKK␥ interacting proteins, via the yeast two-hybrid system. 25 clones were identified that expressed NEMO/IKK␥-interacting proteins, including IKK␣ and CARMA (23). Three clones encoded for overlapping fragments of ABIN-1, a protein previously identified as an A20-binding protein that mimics the NF-B inhibiting effects of A20 (34).
To define the region of NEMO/IKK␥ that interacts with ABIN-1, we tested various deletion mutants of NEMO/IKK␥ for binding to the ABIN-1 fragment (amino acids 380 -636) in yeast. Data shown in Fig.  1A indicate that the region between amino acids 50 and 100 of NEMO/ IKK␥ is required for interaction with ABIN-1. The binding was confirmed in mammalian cells (Fig. 1B). HA-ABIN-1 was transiently coexpressed in HEK293 cells together with FLAG-NEMO/IKK␥ or a NEMO/IKK␥ mutant lacking the first 91 amino acids (FLAG-  NEMO⌬N91). Immunoprecipitates of FLAG-NEMO/IKK␥ contained HA-ABIN-1 only if both proteins were co-expressed (Fig. 1B, compare  lanes 3 and 4). In agreement with the data obtained in yeast, ABIN-1 did not co-immunoprecipitate with NEMO⌬N91 (lane 6, Fig. 1B). We were unable to detect the association between endogenous NEMO/IKK␥ and ABIN-1, probably because of the transient nature of the association and/or the high stringent conditions we used to perform co-immunoprecipitation experiments. However, gel filtration experiments showed that endogenous ABIN-1 was eluted from the column in the same fractions containing endogenous NEMO/IKK␥ and other components of IKK complex (Fig. 1C).
Mapping of the NEMO/IKK␥ and the A20 Binding Domains on ABIN-1-To define the domain of ABIN-1 required for its interaction with NEMO/IKK␥, we performed pull-down assays by using recombinant GST-NEMO/IKK␥ and in vitro translated [ 35 S]ABIN-1 ( Fig. 2A). ABIN-1 binds to GST-NEMO/IKK␥, indicating a direct interaction between the two proteins. Furthermore, amino acids 500 -588 of ABIN-1 represent the minimal region that binds to NEMO/IKK␥ ( Fig.  2A, upper panel). To confirm that the region between amino acids 500 and 588 of ABIN-1 was responsible for interaction with NEMO/IKK␥, we generated an internal deletion mutant of ABIN-1 (⌬500 -588) and evaluated its ability to interact with NEMO/IKK␥. As expected, the internal deletion of 89 amino acids from ABIN-1 abolished the interaction with NEMO/IKK␥ (Fig. 2B). Because ABIN-1 was identified as an A20-interacting protein (35), we confirmed that the region between amino acids 407 and 431 of ABIN-1 is responsible for interaction with A20 (Fig. 2C).
Both ABIN-1 and A20 Inhibit NF-B at the Level of the IKK Complex by Associating with NEMO/IKK␥-Both ABIN-1 and A20 are inhibitors of NF-B. It has been proposed that they interfere with a RIP and TRAF2mediated transactivation signal (34). The identification of the interaction between ABIN-1 and NEMO/IKK␥ prompted us to investigate if ABIN-1 was involved in controlling NF-B activation not only upstream but also at the level of the IKK complex. To this aim, we performed reporter assays by transfecting HEK293 cells with the Ig-B-luciferase plasmid in the presence of ABIN-1 and various activators of NF-B. ABIN-1 efficiently inhibited TNF-␣ and TRAF2-dependent activation of NF-B (Fig. 3A). ABIN-1 also blocked NF-B activation induced by proteins acting at the level of the IKK complex, such as CIKS and TAX (18 -21). In contrast, ABIN-1 was not able to inhibit the IKK␤-mediated activation of NF-B (Fig. 3A). Similarly, A20 inhibits NF-B activation mediated by TNF-␣ or ectopic expression of TRAF2, CIKS, and TAX but not IKK␤ (Fig. 3B). These results indicate that both ABIN-1 and A20 interfere with activation of NF-B at the level of the IKK complex, suggesting that the association of ABIN-1 with NEMO/IKK␥ could play an important role in the inhibition of NF-B.
Because ABIN-1 interacts with both NEMO/IKK␥ and A20, we tested whether the NEMO/IKK␥-and A20-binding domains of ABIN-1 were required for inhibition of NF-B. ABIN-1 deletion mutants lacking  Fig. 3. Lower panels in A-C show relative expression levels of each of the transfected proteins. D, ABIN-1 forms a complex with NEMO/IKK␥ and A20. HEK293 cells were transfected with constructs encoding NEMO/IKK␥, A20, and a deletion mutant of ABIN lacking the NEMO/IKK␥-binding domain (ABIN⌬500 -588). Cell extracts were immunoprecipitated with anti-FLAG antibodies (NEMO/IKK␥) and Western blotted (WB) anti-HA to reveal the co-precipitation of A20 and ABIN⌬500 -588. The presence of ϪHA and ϪFLAG proteins in total extracts is shown. E, ABIN-1 promotes association of A20 with NEMO/IKK␥. HEK293 cells were transiently transfected with constructs encoding FLAG-NEMO/IKK␥, HA-A20, and an increasing amount of His-ABIN. Cell extracts were immunoprecipitated with anti-FLAG antibodies (NEMO/IKK␥) and Western blotted anti-HA to reveal the co-precipitation of A20. The presence of ABIN, NEMO/IKK␥, and A20 in the whole cell lysate is shown.

ABIN-1 Binds to NEMO/IKK␥
either the NEMO/IKK␥ binding domain (ABIN⌬500 -588) or the A20 binding domain (ABIN⌬407-431) were still able to inhibit the activity of a NF-B-driven luciferase reporter following different stimuli (Fig. 4,  A and B). In contrast, a mutant of ABIN-1 in which both the NEMO/ IKK␥-and the A20-binding domains were deleted (ABIN⌬407-431/ ⌬500 -588) lost the ability to block activation of NF-B (Fig. 4C). These data were consistent with the hypothesis that ABIN-1 forms a complex with both NEMO/IKK␥ and A20. To address this hypothesis, we immunoprecipitated FLAG-NEMO/IKK␥ and monitored the co-precipitation of the ABIN-1 mutant lacking the NEMO/IKK␥-binding domain (HA-ABIN⌬500 -588) either in the presence or absence of A20 (Fig.  4D). ABIN⌬500 -588 co-immunoprecipitated with NEMO/IKK␥ only in the presence of A20 (Fig. 4D). To further support the idea that ABIN-1 promotes association of A20 with NEMO/IKK␥, we transfected A20 and NEMO/IKK␥ in the presence of an increasing amount of overexpressed ABIN-1. As expected, the amount of A20 co-immunoprecipi-tating with NEMO/IKK␥ increased in the presence of ABIN-1 (Fig. 4E). Taken together, these data indicated that ABIN-1 interferes with activation of NF-B at the level of the IKK complex, and support the idea that ABIN-1 promotes association of A20 with NEMO/IKK␥. A20 Inhibits NF-B by De-ubiquitinating NEMO/IKK␥-To explore the mechanism by which the interactions of both A20 and ABIN-1 with NEMO/IKK␥ down-regulate NF-B signaling, we assessed the effect of either A20 or ABIN-1 on NEMO/IKK␥ ubiquitination. Transfection of FLAG-NEMO/IKK␥ in the presence of HA-ubiquitin results in the polyubiquitination of NEMO/IKK␥ (Fig. 5A, lane 3). Co-transfection of A20 and NEMO/IKK␥ resulted in a dose-dependent disappearance of the ubiquitinated forms of NEMO/IKK␥ (Fig. 5A, lanes 4 and 5). In contrast, co-transfection of ABIN-1 did not affect NEMO/IKK␥ ubiquitination (Fig. 5A, lanes 6 and 7). We did not observe reduction in the overall level of ubiquitinated cellular proteins in the presence of A20, indicating that A20 does not have a global de-ubiquitinating activity in cultured cells (Fig. 5A). Importantly, A20 also blocks IB␣ degradation and NEMO/IKK␥ ubiquitination induced by TNF-␣ (Fig. 5B). To demonstrate that the de-ubiquitinating activity of A20 was required for the observed reduction in NEMO/IKK␥ ubiquitination, we generated two mutants in the OTU domain of A20, which is the domain responsible for the de-ubiquitinating activity of A20 (36). We replaced the cysteine residue of the DXXC motif with serine (C103S), and both the aspartic acid and the cysteine residues (D100A/C103S) with alanine and serine, respectively. The mutation C103S affected the ability of A20 to de-ubiquitinate NEMO/IKK␥ compared with wild type A20, whereas the double mutant D100A/C103S resulted in the complete loss of the de-ubiquitinating activity of A20 on NEMO/IKK␥ (Fig.  5C). As expected, the D100A/C103S mutant was not able to block the NF-B activity induced by different stimuli, such as TRAF2 ( Fig.  5D and data not shown).
These findings strongly suggest that NEMO/IKK␥ is a target of the de-ubiquitinating activity of A20 and confirmed that the ubiquitination of NEMO/IKK␥ is a crucial step in the mechanisms of NF-B activation.
ABIN-1 Mediates the De-ubiquitinating Activity of A20 on NEMO/IKK␥-Next, we explored whether ABIN-1 was involved in the A20-dependent de-ubiquitination of NEMO/IKK␥. To this purpose, we transfected HEK293 cells with a suboptimal amount of A20 and an increasing amount of ABIN-1 and checked for NEMO/IKK␥ ubiquitination. We found that ABIN-1 increases the ability of A20 to de-ubiquitinate NEMO/IKK␥ (Fig. 6A). To demonstrate a role for ABIN-1 in the A20-mediated de-ubiquitination of NEMO/IKK␥, we generated siRNA constructs targeting ABIN-1 (ABINi-370 and i-560). Fig. 6B shows that the construct i-370 knocked-down ABIN-1 expression, whereas the i-560 construct did not. Then, we evaluated whether interference of ABIN-1 impairs the de-ubiquitinating activity of A20 on NEMO/IKK␥. We co-transfected HEK293 cells with FLAG-NEMO/ IKK␥ and HA-ubiquitin and assessed the de-ubiquitinating activity of A20 alone or in the presence of either i-370 or i-560 constructs. The A20-dependent de-ubiquitination of NEMO/IKK␥ decreased only in the presence of the i-370 construct (Fig. 6C). The i-370 construct led to a 2-fold increase of both basal and induced (TRAF2 and CIKS) NF-B activity compared with the empty vector or the i-560 construct, which we used as controls (Fig. 6D). Accordingly with the data shown in Fig.  3A, interference of ABIN-1 did not influence the activation of the NF-Bdependent transfected IKK␤. Also in this case, NF-B activity correlated with the levels of NEMO/IKK␥ ubiquitination. In fact, transfected i-370 increased the ubiquitination of NEMO/IKK␥ with respect to both empty vector and i-560 (data not shown). From these experiments, we concluded that reduced levels of the ABIN-1 protein affect the ability of  Fig. 5D. E, RNA interference of A20 impairs the ABIN-1-mediated inhibition of NF-B. Relative reporter activity was evaluated in HEK293 cells co-transfected with the Ig-B-luciferase reporter, TRAF2, and either A20i or CTLi plasmids. 24 h after transfection cells were stimulated with TNF-␣ for 3 h or left untreated, as indicated. Analysis of luciferase activity was done as above.

ABIN-1 Binds to NEMO/IKK␥
A20 to de-ubiquitinate NEMO/IKK␥ and, consequently the A20-mediated inhibition of NF-B. To further support the functional interplay between ABIN-1 and A20, we knocked-down A20 (43) and evaluated the ability of ABIN-1 to interfere with NF-B activation. As shown in Fig. 6E, ABIN-1 requires A20 to efficiently block NF-B activation induced by TNF and TRAF2.

DISCUSSION
In the present study, we have performed experiments in yeast, in vitro and in transfected cells demonstrating that ABIN-1 physically associates with NEMO/IKK␥. The functional consequence of this interaction is that overexpression of ABIN-1 blocks not only activation of NF-B upstream of the IKK complex, but also NF-B activation mediated by proteins directly contacting the IKK complex, such as CIKS and TAX. Indeed, activation of NF-B mediated by overexpression of IKK␤, which may be considered functionally downstream of NEMO/IKK␥, is not affected by ABIN-1. ABIN-1 has been identified as an A20-binding protein and its overexpression mimics the inhibitory effect of A20, suggesting that ABIN-1, at least partially, mediates the inhibitory function of A20. A20 is a TNF-␣ responsive gene that inhibits NF-B-dependent gene transcription in response to TNF-␣ and other stimuli (37)(38)(39). Given that A20 expression itself is regulated by NF-B, it is possible that A20 and ABIN-1 participate in a negative feedback regulation of NF-B activation (40,41). However, the mechanism by which ABIN-1 participates in this process is not clear. We provide evidence that ABIN-1 and A20 co-operate to promote de-ubiquitination of NEMO/IKK␥, which results in functional inactivation of NF-B. In fact, recent studies have indicated ubiquitination as a key event in the regulation of NF-B activation. For example, A20 modifies the ubiquitination profile of RIP in a two-step model, by removing the Lys 63 -linked ubiquitin chain and by the subsequent ligation of the Lys 48 -linked ubiquitin chain (30). Similarly, A20 inhibits Toll-like receptor signaling by removing the Lys 63linked ubiquitin chain from TRAF6 (27). In this context our data demonstrate that NEMO/IKK␥ is an additional target of the de-ubiquitinating activity of A20. Currently, it is not clear if A20 directly de-ubiquitinates NEMO/IKK␥, or it affects the activity of other protein(s) regulating NEMO/IKK␥ ubiquitination. However, we have now shown that in cells knocked-down for ABIN-1, we observed a decrease in the ability of A20 to de-ubiquitinate NEMO/IKK␥, and that in cells knocked-down for A20, the inhibitory function of ABIN-1 is impaired. Altogether our data strongly suggest that ABIN-1 functionally connects A20 and NEMO/IKK␥.
In summary, we have identified a previously not reported association between ABIN-1 and NEMO/IKK␥ and we provide evidence that ABIN-1 co-operates with A20 in inhibiting NF-B at the level of the IKK complex. In addition, we propose that this association could target A20 on NEMO/IKK␥ and interfere with NEMO/IKK␥ ubiquitination, to negatively regulate NF-B activation.