A20-binding Inhibitor of Nuclear Factor-κB (NF-κB)-2 (ABIN-2) Is an Activator of Inhibitor of NF-κB (IκB) Kinase α (IKKα)-mediated NF-κB Transcriptional Activity*

NF-κB transcription factors are pivotal players in controlling inflammatory and immune responses, as well as cell proliferation and apoptosis. Aberrant regulation of NF-κB and the signaling pathways that regulate its activity have been involved in various pathologies, particularly cancers, as well as inflammatory and autoimmune diseases. NF-κB activation is tightly regulated by the IκB kinase (IKK) complex, which is composed of two catalytic subunits IKKα and IKKβ, and a regulatory subunit IKKγ/NEMO. Although IKKα and IKKβ share structural similarities, IKKα has been shown to have distinct biological functions. However, the molecular mechanisms that modulate IKKα activity have not yet been fully elucidated. To understand better the regulation of IKKα activity, we purified IKKα-associated proteins and identified ABIN-2. Here, we demonstrate that IKKα and IKKβ both interact with ABIN-2 and impair its constitutive degradation by the proteasome. Nonetheless, ABIN-2 enhances IKKα- but not IKKβ-mediated NF-κB activation by specifically inducing IKKα autophosphorylation and kinase activity. Furthermore, we found that ABIN-2 serine 146 is critical for the ABIN-2-dependent IKKα transcriptional up-regulation of specific NF-κB target genes. These results imply that ABIN-2 acts as a positive regulator of NF-κB-dependent transcription by activating IKKα.

NF-B transcription factors are pivotal players in controlling inflammatory and immune responses, as well as cell proliferation and apoptosis. Aberrant regulation of NF-B and the signaling pathways that regulate its activity have been involved in various pathologies, particularly cancers, as well as inflammatory and autoimmune diseases. NF-B activation is tightly regulated by the IB kinase (IKK) complex, which is composed of two catalytic subunits IKK␣ and IKK␤, and a regulatory subunit IKK␥/⌵⌭⌴⌷. Although IKK␣ and IKK␤ share structural similarities, IKK␣ has been shown to have distinct biological functions. However, the molecular mechanisms that modulate IKK␣ activity have not yet been fully elucidated. To understand better the regulation of IKK␣ activity, we purified IKK␣-associated proteins and identified ABIN-2. Here, we demonstrate that IKK␣ and IKK␤ both interact with ABIN-2 and impair its constitutive degradation by the proteasome. Nonetheless, ABIN-2 enhances IKK␣-but not IKK␤-mediated NF-B activation by specifically inducing IKK␣ autophosphorylation and kinase activity. Furthermore, we found that ABIN-2 serine 146 is critical for the ABIN-2-dependent IKK␣ transcriptional up-regulation of specific NF-B target genes. These results imply that ABIN-2 acts as a positive regulator of NF-B-dependent transcription by activating IKK␣.
IKK is composed of two catalytic subunits, IKK␣ and IKK␤, and a regulatory subunit IKK␥/NEMO (10,17,18). Although IKK␣ and IKK␤ share a high degree of sequence similarity, they have largely distinct functions, due to their different substrate specificities and modes of regulation. IKK␤ (and IKK␥) is essential for rapid and transient NF-B induction by proinflammatory signaling cascades, such as those triggered by TNF␣ or LPS, via the "classical NF-B pathway" that mainly relies on IB␣ degradation (19 -21). In contrast, IKK␣ is required for the late and sustained activation of NF-B in response to a specific subset of TNF family members (e.g. lymphotoxin ␤, B cell activating factor (BAFF), CD40 ligand) via the "alternative NF-B pathway" that relies on the inducible proteolysis of p100 (22,23). Nonetheless, there is evidence indicating that IKK␣ contributes to cytokine-induced IB␣ phosphorylation and subsequent degradation, thus participating in the classical NF-B activation pathway as well. For instance, IKK␣ is essential for efficient TNF␣-induced IB␣ phosphorylation in HeLa cells (24), and it has been reported that IKK␣ is also required for receptor activator of nuclear factor -B ligand (RANKL) mediated classical NF-B activation in mammary epithelial cells (25). Furthermore, IKK␣ exerts nuclear NF-B transcriptional activating functions through the control of histone phosphorylation (26,27) and can also act in specific cell types as a negative mediator of NF-B activation by attenuating IKK␤-driven NF-B activation or modulating RelA Ser-536 phosphorylation (28,29). IKK␣ is also involved in epidermal differentiation, but independently of its kinase activity (30 -32).
IKK␣ and IKK␤ activation is dependent on phosphorylation of specific serine residues located within the activation loop (T loop), either by upstream kinases or by trans-autophosphorylation (33)(34)(35)(36). In addition, conformational changes by protein interactions may also be a mechanism to stimulate IKK activity. For example, Cdc37 and Hsp90 form a chaperone complex with IKK␣/␤ that is required for IKK activation in response to TNF␣ and DNA damage (37,38), and NIK and IKK␤ binding protein (NIBP) is an enhancer of TNF␣-induced NF-B activity by interacting with IKK␤ and increasing IKK kinase activity (39).
Because IKK␣ is involved in many ways in the regulation of NF-B activity, the possibility that IKK␣-interacting proteins might specifically regulate its activity is a crucial issue with profound implications toward specific NF-B response. We have purified IKK␣-associated proteins in mouse embryonic fibroblasts (MEFs), and mass spectrometry analysis of the copurified proteins revealed the presence of A20-binding inhibitor of NF-B (ABIN-2) (40,41). In the study presented here, we show that IKK␣ and IKK␤ both interact with ABIN-2 and impair its constitutive degradation by the proteasome. Nonetheless, ABIN-2 increases IKK␣ but not IKK␤-mediated NF-B activation by specifically increasing IKK␣ autophosphorylation and kinase activity. In addition, we found that ABIN-2 serine 146 is critical for ABIN-2-dependent IKK␣ transcriptional up-regulation of specific NF-B target genes.
Lentiviral Production and Transduction-Production of infectious recombinant lentiviruses was performed by transient transfection of 293T cells as described in Ref. 42. For infections, 10 5 cells in 35-mm dishes were transduced with 5000 ng/ml p24 (HIV-1 capsid protein). 48 h later, cells were washed, and fresh medium was added. The culture was then continued as described above.
In Vivo Biotinylation Approach-The approach was adapted from the method described by de Boer et al. based on efficient biotinylation in vivo and single-step purification of tagged transcription factors in mammalian cells (44). Briefly, we first generated a lentiviral vector named pTRIP-BirA by subcloning the bacterial protein-biotin ligase BirA cDNA into the lentiviral vector pTRIP-⌬U3-EF1␣-IRES GFP (42) and then subcloned the cDNA encoding full-length IKK␣ tagged at the N terminus by a peptide of 23 amino acids that can be biotinylated by BirA in vivo (IKK␣-BP). IKK␣-deficient MEFs were then stably transduced with either pTRIP-BirA empty vector or pTRIP-BirA expressing IKK␣-BP. Following large scale cell culture, cytoplasmic and nuclear extracts were incubated with streptavidin beads. Eluted proteins were separated by SDS-PAGE and then stained with Colloidal blue. Each entire lane was divided into 2-3-mm gel slices (at least 20 gel plugs/lane) that were each treated by an in-gel-digested method using modified trypsin and further analyzed by nanoflow liquid chromatographytandem coupled to a Q-TOF system. Data base searches were performed using MASCOT and Profound.
Indirect Immunofluorescence Microscopy-Cells seeded on coverslips were transfected using Lipofectamine. 24 h later, cells were fixed in 4% (w/v) paraformaldehyde in PBS for 25 min, washed twice in PBS, and permeabilized for 15 min in 0.25% Triton X-100 in PBS at room temperature. After blocking with 10% FBS in PBS for 10 min at room temperature, cells were then incubated for 1 h at room temperature with the primary antibody in 3% FBS in PBS, washed twice in PBS, and further incubated with the fluorophore-conjugated secondary antibody for 1 h. After two washes in PBS, the coverslips were mounted with Vectashield mounting medium containing 1.5 g/ml DAPI (Vector Laboratories). Fluorescence imaging was performed using a Leica DMI6000 inverted microscope and a MicroMAX-1300Y/HS (Princeton Instruments) camera using the Metamorph v.7 software (Molecular Devices).
Coimmunoprecipitation and Immunoblotting-Coimmunoprecipitation and immunoblotting were performed as reported in Ref. 45.
EMSAs-EMSAs were realized as described in Ref. 45. For supershift assays, nuclear extracts were incubated with specific antibodies for 30 min on ice before incubation with the labeled probe.
Kinase Assays-For IKK immunocomplex kinase assays, transfected cells were collected in whole cell lysis buffer (300 mM NaCl, 25 mM Hepes, pH 7.7, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5% Triton X-100 supplemented with 10 mM p-nitrophenyl phosphate disodium salt, 20 mM ␤-glycerol phosphate, 100 M Na 3 VO 4 , 1 mM PMSF, 1 ϫ Complete protease inhibitor mixture (Roche Diagnostics)). HA-IKK␣ or HA-IKK␤ was immunoprecipitated from 100 g of cell lysates with HA antibody for 2 h or overnight at 4°C, after which protein A/G-agarose beads were added and incubation continued for 90 min at 4°C. The immunoprecipitates were collected, washed three times in lysis buffer and once in kinase buffer (20 mM Hepes, pH 7.6, 10 mM MgCl 2 ), and then incubated at 30°C for 25 min in 30 l of kinase reaction mixture containing 1 Ci of [␥-32 P]ATP, 1.5 g of either bacterial expressed GST-IB␣(1-54) or GST-ABIN-2 fulllength and mutant forms as substrate. The reaction was stopped by addition of an equal volume of Laemmli buffer and heat denaturation for 5 min at 90°C. Proteins were separated on 10% SDS-polyacrylamide gel and transferred to nitrocellulose membranes, stained with Ponceau S, subjected to autoradiography to visualize phosphorylated GST fusion proteins, and finally analyzed by immunoblotting for normalization on amount of immunoprecipitated IKKs.
Calf Intestinal Phosphatase Assays-Whole cells extracts prepared in ubiquitination assay buffer as described above were incubated for 30 min at 37°C with 1 unit of calf intestinal phosphatase (New England Biolabs) either untreated or heat-inactivated for 10 min at 95°C and analyzed by immunoblotting.
Luciferase Reporter Assays-Firefly and Renilla luciferase activities were determined using the Dual Luciferase Assay System (Promega) according to the manufacturer's instructions. Experiments were conducted three times in triplicate.
RT-qPCR-Total RNA extraction and RT were performed as described previously (45). Real-time PCR analysis was carried out with LightCycler FastStart DNA Master plus SYBR Green I on a Light Cycler 1.5 (Roche Applied Science). All values were normalized to the level of hypoxanthine-guanine phosphoribosyltransferase (HPRT) mRNA. Primer sequences are available upon request.

RESULTS
ABIN-2 Interacts with IKK␣ and IKK␤-Because IKK␣ plays pivotal and complex roles in the regulation of NF-B activation, we undertook to identify new IKK␣-interacting partners that could participate in the activation of NF-B transcriptional activity. We applied a biochemical approach based on specific biotinylation in vivo using IKK␣ as a bait (see "Materials and Methods"), taking advantage of the very high affinity of avidin/ streptavidin for biotinylated templates and efficient single-step purification and identified ABIN-2 as a new IKK␣-interacting partner candidate.
To examine further the significance of this ABIN-2-IKK␣ interaction, we first examined the subcellular localization of ABIN-2 when coexpressed with IKK␣/␤ in HeLa cells. As shown in Fig. 1A, when expressed alone, ABIN-2 formed punctate structures of various sizes in the cytoplasm. However, when coexpressed with IKK␣ or IKK␤, ABIN-2 relocalizes uniformly all over the cytoplasm, indicating that IKK␣/␤ have an effect on the subcellular localization of ABIN-2. We next performed coimmunoprecipitation experiments using whole cell extracts from HEK293 cells cotransfected with E tag-ABIN-2 along with either FLAG-IKK␣ or FLAG-IKK␤. As shown in Fig. 1B, reciprocal experiments with anti-FLAG-and anti-E tag specific antibodies showed that ABIN-2 coimmunoprecipitates with both IKK␣ and IKK␤ at similar levels. Finally, we confirmed the association between ABIN-2 and IKK␣/␤ on endogenous proteins in HEK293 and HeLa cells (Fig. 1C).

IKK␣ and IKK␤ Impair ABIN-2 Constitutive Degradation by the Proteasome through Deubiquitination of Lys-48-linked
Polyubiquitin Chains-In the course of our experiments, we noted that when expressed alone in HEK293 cells, ABIN-2 protein was hardly detectable (Fig. 1B), although the total mRNA level of ABIN-2 mRNA presented a 500-fold increase over basal level of untransfected cells (Fig. 2B). Remarkably, ABIN-2 protein expression is also extremely low in wild-type MEFs stably transduced with a lentivirus expressing E tag-ABIN-2 ( Fig. 2A, left, lane 1). These observations suggest that ABIN-2 might be constitutively targeted for degradation. Treatment of E tag-ABIN-2 expressing MEFs ( Fig. 2A, left), as well as HEK293 cells ( Fig. 2A, right), with the proteasome inhibitor MG132 led to a strong increase in ABIN-2 protein levels, indicating that ABIN-2 is constitutively degraded in a proteasome-dependent manner. Coexpression of IKK␣/␤ with ABIN-2 in HEK293 cells resulted in a strong increase of ABIN-2 protein expression levels in an IKK kinase activity-dependent manner (Fig. 2C), whereas it had only a slight effect on ABIN-2 mRNA level (Fig.  2B). Taken together, these observations led us to speculate that IKK␣/␤ might induce a reduction in the extent of conjugation of ABIN-2 to the Lys-48-linked polyubiquitin chains which provide a tag for recognition by the 26 S proteasome (46). To test this hypothesis, we first examined whether ABIN-2 can be ubiquitinated via a Lys-48 linkage by cotransfection of HEK293 cells of E tag-ABIN-2 with an HA-tagged ubiquitin Lys-48 mutant containing only one lysine at position 48 (Lys-48-Ub). After immunoprecipitation of ABIN-2 (E tag antibody), Lys-48linked polyubiquitination was analyzed by immunoblotting with an HA antibody. As shown in Fig. 2D, left, ABIN-2 is efficiently ubiquitinated with ubiquitin conjugated through Lys-48. Most importantly, coexpression of either HA-IKK␣ or HA-IKK␤ led to a marked reduction in Lys-48-linked ubiquitination of ABIN-2. Significantly, coexpression of IKK␣/␤ with wild-type ubiquitin also decreased ABIN-2 polyubiquitination (Fig. 2D, center), whereas it does not affect Lys-63-linked ubiquitination (Fig. 2D, right). Together, these results indicate that ABIN-2 is constitutively degraded by the proteasome and that IKK␣/␤ can increase ABIN-2 stability, possibly via the induction of deubiquitination of Lys-48-linked polyubiquitin chains of ABIN-2.

ABIN-2 Enhances IKK␣-mediated NF-B Activation-Given
the ability of ABIN-2 to interact with and to be stabilized by IKK␣ and IKK␤, we wanted to see whether ABIN-2 could affect IKK␣/␤-mediated NF-B activation. To our surprise, the expression of ABIN-2 led to a strong increase in IKK␣-mediated NF-B DNA binding activity in an ABIN-2 dose-dependent manner (Fig. 3A). In contrast, although the basal level of NF-B DNA binding complexes was slightly higher when IKK␤ was expressed alone compared with that seen with IKK␣ alone (lane 3 versus lane 8), ABIN-2 expression had little effect on IKK␤-mediated NF-B DNA binding activity (Fig. 3A). We next examined the subunit composition of the NF-B DNA-binding complexes. Supershift analysis of nuclear extracts from HEK293 cells transfected with HA-IKK␣ and E tag-ABIN-2 (extracts from Fig. 3A, lane 7) revealed that B-binding complexes were mainly composed of RelA-p50 and cRel-p50 dimers (Fig. 3B). Not surprisingly, expression of ABIN-2 had no effect on catalytically inactive K44M IKK␣ and IKK␤ mutants (KD) (Fig. 3C). Most importantly, coexpression of IKK␣ with ABIN-1, another ABIN family member (41), had no effect on the overall NF-B DNA binding activity, indicating that ABIN-2 exerts specific function in enhancing IKK␣-mediated NF-B activation (Fig. 3D). To address whether ABIN-2-activating function could also be seen on IKK␣-dependent NF-B  were incubated with the indicated antibodies before incubation with the labeled probe. Complex I, RelA/p50; complex II, cRel/p50. C, IKK␣ kinase activity is required for the ABIN-2-dependent IKK␣-mediated NF-B DNA binding activity. Nuclear extracts from HEK293 cells transfected with E tag-ABIN-2 along with either WT IKK␣/␤ or KD mutants were analyzed for NF-B activity by EMSA. Protein expression levels were analyzed by immunoblotting for the indicated proteins. D, ABIN-2 but not ABIN-1 synergizes with IKK␣ to induce NF-B DNA binding activity. Nuclear extracts from HEK293 cells transfected with either E tag-ABIN-1 or E tag-ABIN-2 alone or with either wild-type or KD mutant of HA-IKK␣ were analyzed for NF-B activity by EMSA. Protein expression levels were analyzed by immunoblotting for the indicated proteins. E, ABIN-2 enhances IKK␣-mediated NF-B transcriptional activity. HEK293 cells were transfected with 3ϫNF-B-luciferase reporter along with the indicated expression plasmids, and luciferase activity was determined. transcriptional activity, we compared the ability of ABIN-2 to potentiate IKK␣ and IKK␤ activation of an NF-B-dependent reporter gene in transiently transfected HEK293 cells. Whereas IKK␣ and IKK␤ alone led to similar activation of the NF-B reporter, overexpression of ABIN-2 greatly enhanced IKK␣induced NF-B activation in a dose-dependent manner but had little or no effect on IKK␤-mediated activation (Fig. 3E).
Because ABIN-2 interacts with IKK␣ and specifically activates IKK␣-dependent NF-B activation, we wondered whether ABIN-2 might increase IKK␣ activity. We first measured IKK␣ activity in an in vitro kinase assay. Whereas in the absence of ABIN-2, only a weak IKK␣ kinase activity toward IB␣ could be detected (Fig. 3F), this activity was enhanced markedly when cells were cotransfected with E tag-ABIN-2 (Fig. 3F, lane 2 versus lane 3). Interestingly, ABIN-2 also greatly enhanced IKK␣ autophosphorylation. We next examined IKK␣ by immunoblotting (Fig. 3G) and similarly observed that ABIN-2 also markedly increased IB␣ phosphorylation on serine 32 and serine 36 in vivo as well as IKK␣ phosphorylation.
Further in vivo evidence for a role for ABIN-2 in regulating IKK␣ activity was obtained by comparing IKK/NF-B activity in WT and ABIN-2-deficient primary MEFs upon prolonged stimulation with TNF␣. As shown in Fig. 3H, TNF␣ treatment of WT MEFs resulted in the characteristic biphasic time course of NF-B DNA binding activity. In contrast, the late phase of NF-B activity that occurs at 4 h and persisted for at least 8 h in WT MEFs was strongly decreased in ABIN-2-deficient MEFs.

JOURNAL OF BIOLOGICAL CHEMISTRY 32283
Because temporal control of IKK activity has been reported previously to be crucial for dynamic of NF-B activation (47), we thus compared the IKK␣ activation profile in WT and ABIN-2-deficient MEFs at late time points of TNF␣ treatment. Examining IKK␣ kinase activity in WT MEFs revealed that IKK␣ was highly active at early time points, dropped to low levels after about 45 min, and was reactivated at 4 h of TNF␣ stimulation and persisted over 6 h in WT MEFs (Fig. 3I, upper). In contrast, the late phase of TNF␣-induced IKK␣ activity was almost completely abolished in absence of ABIN-2. Similarly, IKK␣ phosphorylation induced by prolonged stimulation by TNF␣ was markedly reduced in ABIN-2-deficient MEFs compared with that seen in WT MEFs (Fig. 3I, lower). Together, these results indicate that ABIN-2 induces IKK␣ autophosphorylation and kinase activity in vivo, thus resulting in increased IKK␣-mediated NF-B transcriptional activity.
IKK␣ Induces the Phosphorylation of ABIN-2-During the course of ABIN-2 immunoblotting assays, we noticed that more slowly migrating ABIN-2 species could be detected when ABIN-2 was coexpressed with HA-IKK␣ or HA-IKK␤ (Fig. 1B, Inputs, and Fig. 2C). Importantly, highly resolutive SDS-polyacrylamide gels indicated that coexpression of ABIN-2 with IKK␣ resulted in the appearance of an even more shifted band compared with the one observed upon coexpression of ABIN-2 with IKK␤ (Fig. 2C, lane 2 versus lane 4). Because coexpression of ABIN-2 with IKK␣ led to a strong increase in IKK␣ kinase activity and subsequent IKK␣-mediated NF-B activation, these results raise the intriguing possibility that IKK␣ may also directly phosphorylate ABIN-2. Thus, we decided to investigate whether ABIN-2 might be phosphorylated by IKK␣ and whether such a phosphorylation might be, at least in part, responsible for the enhancement of IKK␣-dependent NF-B target gene expression.
Interestingly, the slower migrating form of ABIN-2 was converted to its basal form upon treatment with calf intestinal phosphatase, revealing phosphorylation as the nature of ABIN-2 modification (Fig. 4A). In a second step, to determine whether IKK␣ directly phosphorylates ABIN-2, we generated a GST-ABIN-2 full-length fusion protein and performed in vitro IKK␣ immunocomplex kinase assays using IKK␣ that was immunoprecipitated from HEK293 cells overexpressing HA-IKK␣, and the GST-ABIN-2 protein as a substrate. The GST-ABIN-2 protein was significantly phosphorylated by IKK␣, the relative level of phosphorylation being slightly lower than that of IKK␣ autophosphorylation (Fig. 4B, right). Moreover, the ABIN-2 phosphorylation was specific, because IKK␣ did not phosphorylate the GST protein alone (Fig. 4B, right). To localize the region(s) of ABIN-2 that is (are) phosphorylated by IKK␣, five GST-ABIN-2 deletion mutants, each containing a subdomain of the ABIN-2 coding region, as described in Fig. 4C (⌬1-⌬5 mutants), were used as substrates in IKK␣ kinase assays. As shown in Fig. 4D, only mutants ⌬1 and ⌬2 were phosphorylated efficiently by IKK␣. Note that the band observed with mutant ⌬3 is nonspecific, because its size does not correspond to that of the substrate detected by Ponceau S staining (Fig. 4D). Mutant ⌬1 and mutant ⌬2 contained three and four serines, respectively. Because IKK␣ is known to be a serine/threonine kinase (10), one or more of these serines are likely involved in ABIN-2 phosphorylation. We have mutated each of the serine residues to alanine into the ⌬1 and ⌬2 GST-ABIN-2 deletion mutants, and in vitro kinase assays revealed that a point mutation at serine 62 and serine 146 abolished phosphorylation of the ⌬1 and the ⌬2 ABIN-2 deletion mutant, respectively (Fig. 4E). Together, these results suggest that ABIN-2 serine 62 and/or serine 146 phosphorylation by IKK␣ might be required for enhancement of IKK␣-mediated NF-B activity.
ABIN-2 Serine 146 Is Critical to Enhance IKK␣-dependent NF-B Target Gene Expression-We therefore analyzed the functional consequence of ABIN-2 serine 62 or serine 146 mutations on endogenous NF-B target gene expression. Using real-time PCR, we compared the expression of 15 known NF-B target genes, including inflammatory chemokines and cytokines as well as antiapoptotic genes, in HEK293 cells transfected with HA-IKK␣ along with either wild-type ABIN-2 or S62A or S146A mutants. Remarkably, whereas the expression of RANTES and monocyte chemoattractant protein-1 was strongly increased in HEK293 cells transfected with IKK␣ and either wild-type ABIN-2 or the S62A ABIN-2 mutant (Fig. 5A, upper panels), coexpression of IKK␣ with the S146A mutant resulted in a marked reduction in RANTES and monocyte chemoattractant protein-1 mRNA levels. The decrease is specific, because it was not observed for TNF␣ and cIAP2, two other NF-B target genes for which the expression level is also strongly induced in cells coexpressing IKK␣ and wild-type ABIN-2 (Fig. 5A, middle panels). Moreover, analysis of wildtype and mutant ABIN-2 expression showed that their mRNA levels are similar when coexpressed with IKK␣ (Fig. 5A, lower  left panel). Similarly, IKK␣ expression is not affected by coexpression with either form of ABIN-2 (Fig. 5A, lower right panel). Taken together, our results show that ABIN-2 exerts a serine 146-dependent selective activating function serving to enhance IKK␣-mediated NF-B activity.

DISCUSSION
In this report, we reveal an unexpected and intimate relationship between IKK␣ and ABIN-2. We show that ABIN-2 interacts with and is stabilized by IKK␣. Importantly, ABIN-2 activates IKK␣ kinase activity, thus enhancing IKK␣-mediated NF-B transcriptional activity. Furthermore, ABIN-2 is phosphorylated by IKK␣ at serine 146, a serine that appears to be critical for ABIN-2-dependent IKK␣ transcriptional up-regulation of specific NF-B target genes. We propose that ABIN-2 serves as an adaptor protein to activate IKK␣ which in turn can also phosphorylate ABIN-2. This cross-regulation between ABIN-2 and IKK␣ seems to be essential for activation of selective NF-B target gene expression (Fig. 5B). However, it is not yet clear whether ABIN-2 stabilization is absolutely required for an ABIN-2-dependent increase in IKK␣ activity.
In contrast to our findings that ABIN-2 interacts with IKK␣ and IKK␤ both in overexpressing conditions and when the interactions of the endogenous proteins were examined, a previous report showed that ABIN-2 binds to IKK␥/NEMO, but not to IKK␣ and IKK␤ in transfected HEK293 cells (48). The reasons for these discrepant results are not clear but could be due to differences in conditions employed for the coimmuno-precipitation experiments. Nevertheless, because all three IKK subunits are known to form a complex in vivo (49), it is likely that ABIN-2 could be part of the classical tripartite IKK complex (Fig. 5B). Nevertheless it cannot be excluded that ABIN-2 might also bind to the alternative IKK␣-containing complex independently of IKK␤ and IKK␥ (50).
Based on its ability to bind A20 and block inflammatory cytokine-induced NF-B activation upon overexpression, ABIN-2 was originally described as a negative regulator of NF-B-dependent gene expression (40). Such an inhibition was also observed in response to TPA and EGF, thus reinforcing the concept that ABIN-2 overexpression leads to inhibition of NF-B (40,51). Interestingly, ABIN-2 interferes with NF-B activation induced by RIP1, TRAF2, TRAF6, and IRAK1, but not IKK␤, suggesting that the inhibitory effect of ABIN-2 occurs upstream of IKK␤ (40). In the present report, we demonstrate that ABIN-2 can act as an activator of NF-B through its interaction with IKK␣ and subsequent enhancement of IKK␣-mediated NF-B transcriptional activity. It thus appears that ABIN-2 may exert dual functions depending on its targets within the NF-B activation pathways. Interestingly, IKK␣ was shown to associate with RelA in the nucleus and activate the FIGURE 4. IKK␣ induces the phosphorylation of ABIN-2. A, differential migration of ABIN-2 upon ectopic expression of IKK␣ is due to phosphorylation. HEK293 cells were transfected with the indicated expression plasmids. Whole cell extracts prepared as in Fig. 2D were incubated with calf intestinal phosphatase, either untreated (CIP) or heat-inactivated (CIP*), and analyzed by immunoblotting for the indicated proteins. B, IKK␣ phosphorylates ABIN-2 in vitro. Whole cell extracts from HEK293 cells transiently transfected with HA-IKK␣ were subjected to immunoprecipitation with anti-HA, and kinase assays were performed using either GST alone or GST-ABIN-2 full-length as substrates. Immunoprecipitated HA-IKK␣ protein was detected by immunoblotting with anti-HA. Arrows indicate phosphorylated substrate and IKK␣ autophosphorylation, and * indicates nonspecific bands. C, schematic representation of GST-ABIN-2 deletion mutants. Numbers refer to amino acid positions within human ABIN-2.  expression of NF-B-responsive genes through direct modulation of histone function (26,27). Although we were unable to detect nuclear ABIN-2 by immunofluorescence microscopy experiments, there is evidence to suggest that ABIN-2 may function as a transcriptional coactivator. First, tandem affinity purification indicated that ABIN-2 forms a constitutive complex with several NF-B subunits (RelA, p50, p52) (52); second, a fusion protein of ABIN-2 and the Gal4-DNA binding domain leads to the induction of a Gal4-dependent reporter gene (53); and third, the C-terminal domain of ABIN-2 can enter the nucleus and exert transactivating activity (53). In the near future, chromatin immunoprecipitation experiments designed to determine whether ABIN-2, potentially in conjunction with IKK␣ and/or RelA, is recruited to the promoter regions of NF-B-responsive genes will provide a direct test of this hypothesis.
Analysis of ABIN-2-deficient mice revealed that ABIN-2 is required for ERK activation in response to TNF␣ and LPS in macrophages as well as upon CD40 ligation in B cells by stabilization of the upstream kinase TPL-2 (54). In contrast, LPSand TNF␣-induced activation of ERK was unaffected by ABIN-2 deficiency in B cells and dendritic cells, respectively (54). Therefore, it is likely that the ABIN-2-activating function in the ERK signaling pathway is both stimulus-and cell typespecific, reflecting the alternate use of TPL-2 and other MEK kinases to trigger ERK activation. No defect in NF-B activation triggered by LPS was observed in primary B cells isolated from ABIN-2-deficient mice (54). Similarly, activation of NF-B induced by TNF␣ and LPS was unaffected in ABIN-2-deficient macrophages (54). However, because overexpression experiments point to a role of ABIN-2 in the control of NF-B transcriptional activity, it is tempting to hypothesize that ABIN-2, just as it does for the ERK pathway, plays a role in the regulation of NF-B activation in vivo in a cell type-and stimulus-specific manner. In support of this idea, a recent report demonstrates the existence of recurrent ABIN-2 somatic mutations in diffuse large B cell lymphoma that altered the capacity of ABIN-2 to inhibit NF-B (55).
In an effort to address the physiological relevance of the activation of NF-B activity by ABIN-2, we have performed realtime PCR on several known NF-B target genes with roles in inflammation and cell survival. We observed that coexpression of ABIN-2 and IKK␣ leads to a strong induction of several inflammatory cytokines and chemokines compared with what was seen upon expression of either IKK␣ or ABIN-2 alone, suggesting a proinflammatory role for ABIN-2, most likely due to its NF-B-activating properties described in this report. Interestingly, serine 146, located within the N-terminal domain of ABIN-2, seems to be a critical phosphorylation site for ABIN-2-dependent IKK␣ transcriptional up-regulation of specific NF-B target genes. Because ABIN-2 serine 146 is phosphorylated by IKK␣ in vitro, our data suggest an intimate interplay between ABIN-2 and IKK␣ that is crucial for activation of selective NF-B target gene expression. However, further experiments are required to determine whether IKK␣ is responsible for ABIN-2 phosphorylation in vivo and whether other kinases might be involved.
In conclusion, our study has highlighted a new and unexpected NF-B-activating function for ABIN-2 and defined a previously unrecognized mechanism of cross-regulation between ABIN-2 and IKK␣ that seems to be an essential regulator of NF-B-dependent gene expression.