Role of Adapter Function in Oncoprotein-mediated Activation of NF-κB

Mechanisms by which the human T-cell leukemia virus type I Tax oncoprotein activates NF-κB remain incompletely understood. Although others have described an interaction between Tax and a holo-IκB kinase (IKK) complex, the exact details of protein-protein contact are not fully defined. Here we show that Tax binds to neither IKK-α nor IKK-β but instead complexes directly with IKK-γ, a newly characterized component of the IKK complex. This direct interaction with IKK-γ correlates with Tax-induced IκB-α phosphorylation and NF-κB activation. Thus, our findings establish IKK-γ as a key molecule for adapting an oncoprotein-specific signaling to IKK-α and IKK-β.

HTLV-I 1 Tax is the etiological oncoprotein associated with adult T-cell leukemia (1). Tax is a 40-kDa nuclear phosphoprotein whose expression sufficiently immortalizes T-lymphocytes (2,3). Although its mechanisms for immortalization are not fully understood, Tax has been shown to dysregulate cell cycle progression (4 -6) and to subvert host DNA damage surveillance pathways (7)(8)(9)(10)(11). Tax is also a well characterized transcriptional activator of the HTLV-I long terminal repeats as well as many cellular promoters (12,13) with abilities to activate cAMP-responsive, NF-B-responsive, and serum-respon-sive promoters (14 -16). Despite its pleiotropic effects, specificity of Tax action has been shown to occur through direct contacts with cellular proteins (e.g. Refs. 10 and 17-19).
Because of potential implications for cellular transformation, it is of interest to understand how transforming viruses activate NF-B. HTLV-I represents an attractive model; its oncoprotein, Tax, has been suggested to target both IKK-␣ (34 -36) and IKK-␤ (34 -37), presumably through direct protein-protein contacts. Mechanistically, how IKK-␣ and IKK-␤ are impinged upon by Tax has not been defined. Here we show that Tax binds to neither IKK-␣ nor IKK-␤ but instead contacts directly IKK-␥. We propose that IKK-␥ functionally adapts oncoprotein signaling to IKK-␣/IKK-␤.
GST Pull-down Assay-Expression and purification of GST, GST-IKK-␥, and MBP-Tax from Escherichia coli were performed using protocols from Amersham Pharmacia Biotech and New England BioLabs. Protein affinity chromatography was performed as described previously (10).
To address this issue, we co-expressed Tax (pIEX) and HAtagged human IKK-␥ (pHA␥) in HeLa cells and performed reciprocal co-immunoprecipitations (Fig. 1A). HeLa cells transfected singularly with pHA␥ abundantly expressed HA-IKK-␥, which was detected easily by direct immunoblotting with a mouse anti-HA antibody (Fig. 1A, lane 1, ␣-HA (m)) and by immunoprecipitation with a rabbit anti-HA antiserum (Fig. 1A, lane 2, ␣-HA (r)) followed by immunoblotting with mouse anti-HA antibody. In HeLa cells co-transfected with pHA␥ and pIEX (Fig. 1A, lanes 5 and 8), HA-IKK-␥ was observed in the mouse anti-Tax precipitate (lane 5), and Tax was found in the rabbit ␣-HA antiserum precipitate (lane 8). These reciprocal findings are consistent with an intracellular Tax-IKK-␥  complex. IKK-␥ is one component of a larger holo-IKK complex (31,33). In view of this, the observed co-immunoprecipitation of IKK-␥ and Tax does not exclude the possibility that IKK-␥ is simply a passenger protein recovered as a consequence of Tax interaction with another member of the IKK complex. To challenge this possibility, we performed in vitro pull-down assays with GST-IKK-␥ and MBP-Tax purified from E. coli. In agreement with a direct contact between Tax and IKK-␥, Fig. 1B verified that MBP-Tax bound to GST-IKK-␥ (lane 4) but not to GST alone (lane 2).
To confirm the association between Tax and IKK-␥ within a eukaryotic cell, we checked for interactions in yeast (Fig. 1C). When BD-Tax was co-expressed in yeast with AD-IKK-␥, we observed that this pair (Fig. 1C, column 4) conferred Ͼ20-fold stimulation over the background ␤-galactosidase activity induced by each plasmid singularly (Fig. 1C, columns 1 and 2). The interaction measured for Tax and IKK-␥ (Fig. 1C, column 4) is comparable with the previously characterized binding between Tax and HsMAD1 (Fig. 1C, column 6, and Ref. 10). Because no IKK homologs have been identified in the complete sequence of the yeast genome, we are reassured that the observed results are unlikely to be consequences of bridging fortuitously supplied by yeast IKKs. Hence, the GST pull-down assays and the yeast two-hybrid results collectively support a direct interaction between Tax and IKK-␥.
Tax and IKK-␥ Interaction Correlates with IB-␣ Phosphorylation and NF-B Activation-Tax activates NF-B (14, 34 -36). Although the involvement of MEKK-1, IKK-␣, IKK-␤, and NIK have been suggested (34 -37), the exact mechanisms for this activation are not fully understood. Based on the above observation, we next asked whether binding of IKK-␥ to Tax correlates functionally with Tax activation of NF-B.
Previously, we have constructed and characterized 47 Tax point mutants (16). The phenotypes of two mutants, Tax S258A and Tax L320G, distinguish clearly between activity through the cellular CREB/ATF versus the NF-B pathways. Thus, Tax S258A activates CREB/ATF but not NF-B; whereas Tax L320G activates NF-B but not CREB/ATF (3,16). We used these two mutants to clarify the significance of Tax-IKK-␥ interaction for NF-B activation.
HeLa cells were separately co-transfected with pHA␥ and pIE vector (Fig. 2, lane 1), pHA␥ and pIEX (Tax wild type; lane 2), pHA␥ and pIEX S258A (lane 3), or pHA␥ and pIEX L320G (lane 4). The equivalent expression of Tax and Tax mutants was verified by Western blotting (Fig. 2A). In parallel, nuclear NF-B DNA binding activity and cytoplasmic IKK activity associated with Tax were assessed by EMSA (Fig. 2B) and in vitro kinase assay (IKK act.; Fig. 2C), respectively. Consistent with previous findings (16), Tax L320G activated NF-B and IKK in a manner similar to wild type Tax (Fig. 2, B and C,  compare lanes 4 with lanes 2). By contrast, Tax S258A was defective for this activation (Fig. 2, B and C, compare lanes 3 to  lanes 1 and 2). Concordantly, anti-Tax serum (␣-Tax; Fig. 2D) co-immunoprecipitated IKK-␤ (as detected by immunoblotting with ␣-IKK-␤ serum) from cells that expressed Tax or Tax L320G (Fig. 2D, lanes 2 and 4), but not from cells that expressed Tax S258A (Fig. 2D, lane 3). In the same immunoprecipitations, we also found that Tax or Tax L320G but not Tax S258A associated with HA-IKK-␥ (Fig. 2E). Considered together, these findings correlated Tax/Tax mutant binding to IKK-␥ ( Fig. 2E) with nuclear NF-B activity (Fig. 2B), IB-␣ phosphorylation (Fig. 2C), and co-immunoprecipitation with  1 and 2) and GST-IKK-␥ (lanes 3 and 4) proteins were bound to Sepharose beads. Beads were incubated with MBP-Tax, and bound proteins were then eluted. Flowthrough (FT, lanes 1 and 3) and eluates (lanes 2 and 4) were analyzed by Western blotting with rabbit anti-Tax. Anti-Tax-reactive proteins are marked by an arrow and an asterisk. The band marked by an asterisk likely represents a degradation product. C, yeast two-hybrid assay. Yeast SFY526 was transformed with plasmids expressing the indicated proteins. Stable transformants were selected and assayed for relative ␤-galactosidase activity in chlorophenol red-␤-D-galactopyranoside (CPRG) units (38). Results are representative of three independent experiments. IKK-␤ (Fig. 2D).
Evidence Supports Direct Contact of Tax with IKK-␥ but Not with IKK-␣ nor IKK-␤-Data presented here and elsewhere are compatible with the following: (a) direct binding of Tax to IKK-␥ (Fig. 1); (b) direct/indirect association of Tax with IKK-␣ and/or IKK-␤ ( Fig. 2D and Ref. 34); and (c) a Tax-induced increase in IKK activity ( Fig. 2C and Refs. 34 -36). Others have shown that IKK-␥ can bind IKK-␤ directly (31) and associates with IKK-␣ in vivo (33). Based on these observations, one cannot formally distinguish between a model in which Tax contacts all three IKK-proteins (IKK-␣, IKK-␤, and IKK-␥; Fig.  3A) equally versus another model in which Tax contacts only IKK-␥, which then intermediates a signal to IKK-␣ and IKK-␤ (Fig. 3B).
To define better the details of protein-protein contact, we performed modified protein hybrid assays in yeast (Fig. 3C). First we sought to determine whether there was a direct contact of Tax with either IKK-␣ or IKK-␤. We co-expressed either BD-Tax and AD-IKK-␣ (Fig. 3C, column 2) or BD-Tax and AD-IKK-␤ (Fig. 3C, column 3) in yeast. Neither pair showed any increased reporter activity over the background level observed with BD-Tax alone (Fig. 3C, column 1). By contrast, yeast transformants that co-expressed BD-Tax and AD-IKK-␥ showed an activity Ͼ10-fold over background (Fig. 3C, compare  column 4 with column 1). These results are most simply interpreted by a direct Tax interaction with IKK-␥ but not with IKK-␣ nor IKK-␤ (Fig. 3B).
The model deduced from results in yeast can be further tested in mammalian cells. Indeed, if IKK-␥ represents an upstream adapter whereas IKK-␣/IKK-␤ represent downstream effectors, then one prediction is that a dominant negative (DN) form of the latter would repress activity from the former. When overexpressed in HeLa cells, IKK-␥ conferred a 4-fold activation of an NF-B-dependent reporter (Fig. 3D, compare column 2 with column 1). Notably, this activation was abrogated by co-expression of DN mutants of either IKK-␣ or IKK-␤ (Fig. 3D, columns 3-6). We observed that compared with DN IKK-␤ (Fig. 3C, columns 5 and 6), a higher dosage of DN IKK-␣ is required for the inhibition of IKK-␥-mediated activation (columns 3 and 4). This might be explained by the intrinsic differences in specific kinase-activity for IKK-␣ and IKK-␤ (27).
IKK-␥ Is a Mediator for Tax Activation of NF-B-Previously, a C-terminal truncated form of IKK-␥ (IKK-␥⌬C) has been described as a DN inhibitor of IKK-␥ function (33). We used a similarly constructed IKK-␥⌬C mutant to explore the role of IKK-␥ in Tax activation of NF-B. In HeLa and Jurkat cells, expression of wild type IKK-␥ enhanced Tax activation of an NF-B-dependent reporter (Fig. 4, compare group 3 with group 2). In contrast, activation by Tax was progressively diminished with increased expression of IKK-␥⌬C (Fig. 4, groups 4 -6, compared with group 2), consistent with a DN inhibition of function. This finding agrees with the demonstration that a genetic defect in IKK-␥ led to loss of responsiveness to several NF-B-activating stimuli including Tax (31).
Two salient points emerge from the current work. The first is the unexpected finding of IKK-␥ as a direct Tax-binding factor. Previously, several studies have shown an association of Tax with IKK complex (34 -37). Although contact with IKK-␣ and IKK-␤ was suggested (34), the exact details were not defined. We now show that the functional ability of Tax to activate IKK-␣/IKK-␤ is unlikely to stem from immediate contact but occurs indirectly through binding between Tax and IKK-␥.
The second instructive point from this work is that IKK-␥, previously shown to be essential for IKK-␣/IKK-␤ activation of NF-B (31,33), is located functionally and physically upstream of its kinase counterparts. Although our work does not directly address how Tax activates IKK-␣/IKK-␤, the finding that Tax binds IKK-␥ but not IKK-␣/IKK-␤ ascribes physical recruitment of Tax into the holo-IKK complex to the functional adapter function suggested for IKK-␥. IKK-␥-tethered Tax protein is further expected to bring Tax-associated kinase(s) (e.g. MEKK-1 and/or NIK (35)(36)(37)) to the local proximity of IKK-␣/ IKK-␤ for phosphorylation-mediated activation.
An obvious question raised by our results is why should oncoproteins interact via an adapter rather than directly with the IKK-␣/IKK-␤ kinases? Although this is not completely understood, one suggestive explanation is provoked by our recent finding that the human gene for IKK-␥ localizes on the X chromosome (40). If IKK-␥ is indeed essential (as has been shown in Refs. 31 and 33) for a multitude of signals that activate NF-B, then some hints of gender-linked NF-Bbased diseases might have been expected. In the absence of evidence for such, a reasonable deduction is that many yet to be described IKK-associated proteins (another example is IKAP; Ref. 32) might also provide adapter function. The predicted existence of many adapters could explain specificity and redundancy of signaling that cannot otherwise be easily reconciled if each of the numerously different NF-Bactivating signals all contacted directly IKK-␣/IKK-␤. Future studies are needed to clarify the identities of other adapters and how they function in dictating specificity of IKK activation.