Domain-specific Interaction with the IκB Kinase (IKK) Regulatory Subunit IKKγ Is an Essential Step in Tax-mediated Activation of IKK*

The human T-cell leukemia virus type 1 Tax oncoprotein deregulates the NF-κB signaling pathway by persistently stimulating a key signal transducer, the IκB kinase (IKK). Tax physically associates with the IKK regulatory subunit, IKKγ, although the underlying biochemical mechanism and functional significance remain unclear. We show that the Tax-IKKγ interaction requires two homologous leucine zipper domains located within IKKγ. These leucine zipper domains are unique for the presence of a conserved upstream region that is essential for Tax binding. Site-directed mutagenesis analysis revealed that a leucine-repeat region of Tax is important for IKKγ binding. Interestingly, all the Tax mutants defective in IKKγ binding failed to engage the IKK complex or stimulate IKK activity, and these functional defects can be rescued by fusing the Tax mutants to IKKγ. These results provide mechanistic insights into how Tax specifically targets and functionally activates the cellular kinase IKK.

Unlike CREB/ATF, which is constitutively expressed in the nucleus, NF-B is normally sequestered in the cytoplasmic compartment by physical interaction with inhibitors, including IB␣ and related proteins (22,23). NF-B nuclear expression can be transiently induced upon cellular stimulation by a variety of agents, such as T-cell mitogens and proinflammatory cytokines (21,24,25). These agents stimulate the activity of a multisubunit IB kinase (IKK), which phosphorylates IB␣, targeting this inhibitor for degradation through the ubiquitinproteasome pathway (26). In Tax-expressing cells, IKK is chronically activated (Refs. 27-29, reviewed in Ref. 30), which is associated with constitutive IB␣ phosphorylation and NF-B nuclear expression (31)(32)(33).
Given the persistent nature of Tax-induced IKK activation, this viral specific pathway may be mediated by a specific mechanism. Interestingly, Tax is stably associated with the IKK complex in HTLV-I infected T cells (49). These findings raise the possibility that constitutive association of Tax with the IKK complex may be responsible for persistent IKK activation in Tax-expressing cells. More recently, several independent studies have demonstrated that the IKK␥ subunit of IKK physically interacts with Tax and promotes the association of Tax with the IKK catalytic subunits (50 -52). However, how Tax recognizes IKK␥ and whether the Tax-IKK␥ interaction serves as an essential step in Tax-mediated IKK activation is not currently known. In this paper, we present data demonstrating that the Tax-IKK␥ interaction requires two homologous LZs present in IKK␥ and a leucine-rich region present in Tax. This molecular interaction is essential for Tax-mediated stimulation of IKK and subsequent activation of NF-B.

MATERIALS AND METHODS
Plasmid Constructs and Antibodies-The B-luc is a luciferase reporter driven by the human immunodeficiency virus (HIV)-1 B enhancer and a TATA box (53). HTLV-I LTR-luc contains the luciferase gene driven by the HTLV-I LTR (31). Expression vectors encoding wild-type Tax, Tax M22, IKK␣, IKK␤, and murine IKK␥ have been described previously (52). IKK␥ truncation mutants were generated by restriction digestion or PCR amplification of a murine cDNA (35) and subsequent cloning of the DNA fragments into the pcDNA mammalian expression vector (Invitrogen) downstream of an epitope tag (Myc for IKK␥-(47-412) and IKK␥-(69 -412); HA for all other mutants). The IKK␥ mutants harboring internal deletions were generated by sitedirected mutagenesis (Stratagene) using the wild-type HA-tagged murine IKK␥ as template. The IKK␥ ⌬LRc mutant harbors a deletion of the four leucine repeats of LZc (amino acids 315-336); ⌬LRn has a deletion of the first leucine repeat of LZn (amino acids 101-115); ⌬URc and ⌬URn have deletions of the UR of LZc and LZn, respectively. Tax mutants harboring amino acid substitutions were generated with the same strategy using wild-type Tax expression vector as template (see Fig. 2B). IKK␥⅐Tax chimeras were constructed by inserting the IKK␥ cDNA upstream of wild-type or mutant forms of Tax. The anti-HA and anti-IKK␥ monoclonal antibodies were from Roche Molecular Biochemicals and Imgenex Corporation, respectively. The anti-Tax monoclonal antibody was prepared from a hybridoma (168B17-46-34) provided by the AIDS Research and Reference Program, NIAID, National Institutes of Health. All other antibodies were purchased from Santa Cruz Biotechnology, Inc.
Immunoprecipitation (IP) and Immunoblotting Assays-Human 293 kidney carcinoma cells were seeded in 0.1% gelatin-treated 6-well plates (1 ϫ 10 5 cells/well) and transfected using DEAE-dextran with the indicated cDNA expression vectors. The DNA amounts used for the transfections were normalized based on the expression efficiency of each of the expression vectors: 100 ng for IKK␣ and IKK␤, 50 ng for IKK␥ and its mutants, 0.5 g for Tax and its mutants. After 40 h, recipient cells were lysed in radioimmune precipitation buffer (50 mM Tris-HCl, pH7.4, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 1:100 (v/v) of a protease inhibitor mixture (54)). Whole cell lysates were subjected to IP in the radioimmune precipitation buffer as described previously (54), and the precipitated proteins were analyzed by SDS-polyacrylamide gel electrophoresis followed by immunoblotting (54). For direct immunoblotting analyses of proteins, cell lysates were subjected to SDS-polyacrylamide gel electrophoresis followed by immunoblotting.
Luciferase Reporter Gene Assays-Jurkat T cells were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and antibiotics. The JM4.5.2 is a mutant Jurkat cell line lacking expression of IKK␥, which was generated by somatic mutagenesis (42). The parental and mutant Jurkat cells (5 ϫ 10 6 ) were transfected using DEAE-dextran (55) with 200 ng of luciferase reporter plasmids together with the indicated effector cDNA expression vectors. At 40-h post-transfection, the recipient cells were subjected to extraction using a reporter lysis buffer (luciferase reagent, Promega) at about 100 l/10 6 cells. Luciferase activity was detected by mixing 5 l of extract with 25 l of luciferase substrate (Promega) and measured with a luminometer. For reporter gene assays performed with 293 cells, the cells were seeded in 24-well plates (2.5 ϫ 10 4 cells/well) and transfected using DEAE-dextran with 20 ng of B-luc and the indicated cDNA expression vectors encoding wild-type or mutant forms of Tax (50 ng).

IKK␥-Tax Interaction Requires Two Homologous LZ Domains with Conserved Upstream
Motifs within IKK␥-To delineate the sequences of IKK␥ required for Tax binding, we performed IKK␥ truncations to remove its individual coiled-coil domains (Fig. 1A). The generated IKK␥ mutants were subjected to coimmunoprecipitation (co-IP) assays for analyzing their Tax binding activity in mammalian cells. Removal of the N-terminal region, covering a major part of the first coiled-coil domain (Fig. 1A, CC1) and its upstream sequences, did not affect the interaction of IKK␥ with Tax ( Fig. 1B, upper, lanes 3  and 4). However, further deletion to remove the second coiledcoil domain (Fig. 1A, CC2) completely abolished the binding of IKK␥ to Tax (lane 5). As we previously reported (52), a Cterminal deletion removing the fifth coiled-coil domain (CC5) and its downstream sequences significantly, although not completely, inhibited the Tax-IKK␥ binding (Fig. 1A, lane 6). Parallel immunoblot analyses revealed that the IKK␥ mutants and Tax were expressed at comparable levels (Fig. 1B, middle and lower panels). These results indicate that both the second and the fifth coiled-coil domains might participate in Tax binding.
Of note, the fifth coiled-coil domain forms a typical LZ, which has been proposed to interact with upstream IKK activators (36). Interestingly, by sequence alignment, we found that the second coiled coil of IKK␥ also contained a putative LZ domain (designated LZn) (Fig. 1C) exhibiting striking sequence similarities with the C-terminal LZ (Fig. 1C, LZc). Although one of the conserved leucines in the heptad leucine repeats of LZn is substituted by alanine, this singular alteration is unlikely to affect LZ dimerization (58), especially because the LZn is located within an extended coiled-coil dimerization sequence. An interesting feature of the two LZ domains is the presence of a homologous upstream region (UR); this region is composed of six amino acids, four of which are conserved (Fig. 1C). For simplicity, we considered the UR a part of the LZ domains and designated the downstream coiled-coil motifs as leucine repeats (LR). To examine the role of IKK␥ LZ domains in mediating Tax binding, IKK␥ mutants lacking each of the LZs were subjected to co-IP. Deletion of the LR of LZn (LRn) largely abolished the IKK␥-Tax association (Fig. 1D, lane 4), suggesting that this coiled-coil dimerization domain is indeed essential for Tax binding. Remarkably, when the UR of LZn (URn) was removed without altering the downstream coiled-coil domain, the Tax binding activity of IKK␥ was affected even more dramatically (Fig. 1D, lane 6). Deletions introduced to the LZc domain of IKK␥ also revealed a more critical role for the UR than the LR. Whereas deletion of the LRc only caused a partial inhibition of Tax-IKK␥ interaction (Fig. 1D, lane 3), removal of the URc resulted in a much more profound inhibitory effect (Fig. 1D, lane 5). Thus, the conserved UR motifs present in the upstream regions of LZn and LZc of IKK␥ play an essential role in Tax binding, although their downstream coiled-coil dimerization domains appear to be important as well. These proteinprotein interaction studies also demonstrate that the LZn plays a more critical role than LZc in mediating Tax binding, although the optimal IKK␥-Tax association requires both LZ motifs.
A Leucine-rich Region of Tax Is Important for IKK␥ Binding-Previous studies have shown that a Tax mutant, M22 (59), is unable to interact with IKK␥ (52) or stimulate the catalytic activity of IKK (27)(28)(29). Notably, the M22 mutant harbors a two-amino acid substitution in a region of Tax that contains heptad leucine repeats ( Fig. 2A, LR). Computer analysis using the COILS program (60) revealed that this leucinerepeat region (LRR) does not likely form a LZ because no coiled-coil structure was predicted (data not shown). However, secondary structure prediction using the PREDATOR program (61) revealed a short ␣-helix located in the UR of the Tax LRR (data not shown), which is homologous to the UR of IKK␥ ( Fig.  2A, UR). A similar motif is also present in the Tax protein encoded by HTLV-II ( Fig. 2A, Tax2). We examined the role of the Tax LRR in IKK␥ binding by introducing amino acid substitutions ( Fig. 2B) followed by examining the ability of the Tax mutants to bind IKK␥ (Fig. 2C). Wild-type Tax strongly interacted with IKK␥ ( Fig. 2C, upper, lane 2), whereas the M22 mutant failed to mediate this interaction (lane 3). Remarkably, single substitution of each of the conserved leucines (Leu-124, Leu-131, Leu-138) in the LRR of Tax completely prevented Tax from binding IKK␥ (Fig. 2C, upper, lanes 4 -6). Mutation of asparagine 117 to glycine (N117G) and arginine 116 to serine (R116S) also resulted in a significant inhibitory effect (Fig. 2C, lanes 9 and 12). On the other hand, mutation of two amino acids located downstream of the LRR (Thr-145 and Val-152) only moderately affected the Tax-IKK␥ interaction (Fig. 2C,  lanes 7 and 8). Single substitution of the conserved amino acids present in the UR of Tax each caused a partial inhibition of Tax-IKK␥ interaction (Fig. 2C, lanes 13 and 14). Double and quadruple substitutions of these conserved amino acids completely disrupted the IKK␥ binding activity of Tax (Fig. 2C,  lanes 10 and 11). Parallel immunoblot analyses revealed that all the Tax mutants, as well as the cotransfected IKK␥, were stably expressed in the cells (Fig. 2C, middle and lower panels).
These results indicate that the LRR of Tax is important for Tax binding to IKK␥. Of course, it is unlikely that the LRR is the only region of Tax required for Tax-IKK␥ interaction, because some Tax variants mutated outside of this region are also defective in NF-B activation (59,62) or IKK␥ binding (51).
Tax/IKK␥ Interaction Is Required for Recruiting Tax to IKK␣/IKK␤-Prior studies have shown that Tax does not ap-preciably interact with the IKK catalytic subunits; however, this oncoprotein forms stable complexes with both IKK␣ and IKK␤ in the presence of IKK␥ (50 -52). To understand the mechanism by which IKK␥ promotes the association of Tax with the IKK catalytic subunits, we determined whether this "docking" function of IKK␥ required its physical interaction with Tax. For these studies, we examined the ability of the various IKK␥ mutants to promote Tax association with IKK␣/ IKK␤ by co-IP. As expected (50 -52), Tax failed to associate with IKK␤ when these two proteins were coexpressed in 293 cells (Fig. 3A, lane 2). However, the two proteins became stably associated when wild-type IKK␥ was also expressed in the cells (lane 3). More importantly, the IKK␥⌬LRc, which retained significant Tax binding activity (see Fig. 1D, lane 3), was able to stimulate strong association between Tax and IKK␤ (lane 4). In contrast, consistent with their poor Tax binding activity, the IKK␥⌬URc and IKK␥⌬LRn exhibited a weak docking activity (lanes 5 and 6). Moreover, the IKK␥ mutant harboring a deletion of the URn was completely unable to promote the Tax-IKK␤ association (Fig. 3A, lane 7); this result was in agreement with the Tax-binding deficiency of this IKK␥ mutant (see Fig.  1D, lane 6). We also examined the effect of these IKK␥ mutants on Tax-IKK␣ association and obtained similar results (data not shown). Thus, the function of IKK␥ in promoting the binding of Tax to IKK␣/IKK␤ is correlated with its ability to physically interact with Tax.
To further assess the role of Tax/IKK␥ interaction in IKK␥mediated recruitment of Tax to the IKK catalytic subunits, we examined the function of various Tax mutants in associating with IKK␣/IKK␤. Wild-type Tax formed a stable complex with IKK␤ when these two proteins were coexpressed with IKK␥ (Fig. 3B, lane 2). Similarly, two Tax mutants (T145A and V152A) competent in IKK␥ binding also associated with IKK␤ Cell lysates were subjected to IP with anti-HA followed by IB using anti-Tax (upper). The expression level of IKK␥ mutants and Tax were analyzed by IB using anti-HA (middle) and anti-Tax (lower), respectively. The multiple bands of IKK␥ resulted from its constitutive phosphorylation, as they could be converted to a single band following in vitro phosphatase treatment (data not shown). (Fig. 3B, lanes 7 and 8). In sharp contrast, the Tax mutants incapable of IKK␥ binding failed to engage IKK␤ (Fig. 3B, lanes 3-6 and 9 -11). Parallel co-IP assays to detect the IKK␥-dependent Tax/IKK␣ association revealed a similar result (data not shown). Together, these data suggest that the Tax-IKK␥ interaction is required for recruiting Tax to the IKK catalytic subunits.
Tax Mutants Defective in IKK␥ Binding Fail to Stimulate IKK Kinase Activity and NF-B Transcriptional Function-We next determined the functional significance of Tax-IKK␥ interaction in Tax activation of this cellular kinase and its target transcription factor NF-B. Immunocomplex kinase assays were performed to analyze the ability of the various Tax mutants to stimulate IKK catalytic activity in 293 cells. In the absence of exogenously transfected IKK␥, Tax only weakly induced the transfected IKK catalytic subunits (IKK␣ and IKK␤) in 293 cells (data not shown, also see Fig. 5B, lane 2). This was probably caused by the relatively low level of endogenous IKK␥ expressed in these cells (see Fig. 5D). However, together with transfected IKK␥, the wild-type Tax potently activated the IKK catalytic subunits (Fig. 4A, lane 2), as demonstrated by the phosphorylation of the substrate IB␣, (lower, IB␣-P) and IKK␣/␤ autophosphorylation (upper, IKK␣/␤-P). More importantly, Tax M22 and the other Tax mutants defective in IKK␥ binding were inactive in IKK activation (lanes 3-6 and 9 -11). In contrast, those Tax mutants competent in IKK␥ binding (T145A and V152A) significantly stimulated IKK catalytic activity (lanes 7 and 8). We extended these functional studies by performing kinase assays using Jurkat Tag cells. As previously reported (28), Tax was able to stimulate significant activity of IKK in these T cells in the absence of transfected IKK␥ (Fig. 4B, upper, lane 2); this result was consistent with the presence of high amounts of endogenous IKK␥ in Jurkat Tag cells (Fig. 5D, lane 5). More importantly, under these conditions, the function of Tax mutants in IKK activation was also well correlated with their ability to physically interact with IKK␥ (lanes 3-11). We next performed reporter gene assays to further evaluate the function of Tax mutants in stimulating NF-B signaling. Each of the Tax mutants were transfected into 293 cells (Fig. 4C) or Jurkat T cells (Fig. 4D) along with a luciferase reporter driven by the NF-B target enhancer, B (B-luc). Wild-type Tax potently stimulated the B enhancer, resulting in marked induction of luciferase activity (Fig. 4, C and D, bar 2). A significant B stimulatory activity was also obtained with two Tax mutants (T145A and V152A) that were able to bind IKK␥ (bars 7 and 8). However, none of the Tax mutants defective in IKK␥ binding could activate the B enhancer (bars 3-6 and 9 -11). Thus, the Tax/IKK␥ physical interaction is required for Tax-mediated stimulation of IKK and subsequent activation of NF-B.

Fusion of Interaction-defective Tax Mutants to IKK␥ Rescues Their Functional Defect in NF-B Signaling-
The observation that Tax mutants defective in IKK␥ binding fail to activate IKK provides strong support for the idea that the Tax-IKK␥ interaction plays an essential role in Tax activation of IKK. We reasoned that fusion of the interaction-deficient Tax mutants to IKK␥ might rescue their functional defect in NF-B signaling. To test this idea, we generated fusion proteins composed of IKK␥ and Tax mutants and determined their function in stimulating IKK catalytic activity and NF-B transcriptional func-

FIG. 2. A leucine-repeat region of Tax is important for IKK␥ binding. A, sequence showing the heptad leucine repeats (LR) and the upstream region (UR) of the Tax proteins encoded by HTLV-I (Tax) and
HTLV-II (Tax2). The amino acid residues conserved between the UR of Tax LRR and IKK␥ LZ (shown in Fig. 1C) are bold. The leucines present in the heptad leucine repeats are underlined. B, amino acid substitutions within and downstream of the Tax LRR. M22 is a mutant previously generated by random mutations (59). The new Tax mutants were designated based on the positions of the mutated amino acids and the replacing residues (for example, Q107G harbors a substitution of glutamine 107 by glycine). C, interaction of IKK␥ with Tax and Tax mutants. 293 cells were transfected with Tax or its mutants either alone (lane 1) or together with HA-tagged murine IKK␥ (lanes 2-14). The cell lysates were subjected to IP using anti-HA followed by IB using anti-Tax (upper). The level of expression of HA⅐IKK␥ and Tax mutants were analyzed by direct IB using anti-HA (middle) and anti-Tax (lower), respectively.

FIG. 3. Tax-IKK␥ interaction is essential for recruiting Tax to IKK␣ and IKK␤.
A, promotion of Tax-IKK␤ interaction by IKK␥ and IKK␥ mutants. 293 cells were transfected with the indicated HA-tagged IKK␥ constructs together with expression vectors for HA⅐IKK␤ and Tax. Cell lysates were subjected to IP using anti-Tax followed by IB with anti-HA (upper). The HA⅐IKK␤ coprecipitated with Tax is indicated. The expression level of the HA⅐IKK␤ and HA⅐IKK␥ mutants was detected by IB (with anti-HA) and is shown in the middle and lower panels, respectively. Tax expression level was similar to that shown in Fig. 1 (data not shown). A nonspecific band is indicated by ns. B, IKK␥-dependent binding of IKK␤ with Tax mutants. 293 cells were transfected with the wild-type Tax (TaxWT) or the indicated Tax mutants together with HA⅐IKK␤ and HA⅐IKK␥. Cell lysates were subjected to IP using anti-Tax followed by IB with anti-HA. The coprecipitated IKK␤ is indicated (upper). A band below the HA⅐IKK␤ band in lane 7 was nonspecific because it was not detected in other experiments (data not shown). The expression level of the Tax mutants was analyzed by direct IB and is shown in the lower panel. The expression level of IKK␤ and IKK␥ was similar to that shown in A.
tion. When linked to the C terminus of IKK␥, all the Tax mutants were efficiently recruited to IKK␤ (Fig. 5A, lanes 5-11) and IKK␣ (data not shown). Immunocomplex kinase assays were performed to examine whether fusion of the Tax mutants with IKK␥ restored their function in stimulating IKK catalytic activity. As expected, wild-type Tax potently stimulated IKK activity when coexpressed with IKK␥ in 293 cells (Fig. 5B, lane 4), whereas M22 or the other Tax mutants defective in IKK␥ binding failed to activate IKK (Figs. 5B, lane 5 and 4A). Remarkably, when these interaction-deficient Tax mutants were fused to IKK␥, their IKK-stimulatory function was efficiently restored (lanes 6 -12). A parallel kinase assay performed with Jurkat Tag cells revealed that the IKK␥⅐Tax chimeras also stimulated IKK activity in these T cells (Fig. 5C).
We further validated the finding presented above by examining the ability of the IKK␥⅐Tax chimeras to stimulate NF-B transcription activity. We took advantage of an IKK␥-deficient Jurkat cell line, JM4.5.2, recently isolated in our laboratory (42). In these mutant T cells, Tax is unable to activate the B enhancer unless exogenous IKK␥ is provided (42). As expected, a marked luciferase activity was detected when IKK␥ was coexpressed with the wild-type Tax (Fig. 6A, bar 3), but no reporter gene expression was detected when IKK␥ was coexpressed with the Tax mutants defective in IKK␥ binding (bars 4 -10). More importantly, consistent with the kinase assays, fusion of these Tax mutants to IKK␥ efficiently restored their NF-B activation function (bars [11][12][13][14][15][16][17]. Similar results were obtained in a parallel reporter gene assay performed with wild-type Jurkat cells (Fig. 6B). In these cells, Tax efficiently induced B-Luc reporter when transfected at an adequate amount (0.4 g, bar 4) and weakly induced the reporter at a low dose (80 ng, bar 3). In contrast, none of the Tax mutants tested was able to activate the B-Luc when transfected at either low or high doses (Figs. 6B, bars 5-11 and 4D). However, all the IKK␥⅐Tax chimeras significantly induced the luciferase activity (bars 12-18). We found that the level of B activation by the chimeras was lower in Jurkat cells than in the IKK␥-deficient JM4.5.2 cells (compare Fig. 6, A versus B). Although the amounts of DNA used for transfecting the wild-type Jurkat cells were lower, increasing the DNA amounts did not further enhance the level of reporter activation (data not shown). It is likely that the lower efficiency of NF-B activation in wild-type Jurkat cells is caused by the association of IKK␣/IKK␤ with the endogenous IKK␥, which would interfere with binding of the IKK catalytic subunits to the transfected IKK␥⅐Tax chimeras. Nevertheless, a functional rescue, as a result of fusion between IKK␥ and Tax mutants, was clearly detected in the wild-type Jurkat cells. Taken together, these results demonstrate that the IKK␥-Tax interaction serves as an essential step in Taxtriggered NF-B signaling.

DISCUSSION
The multisubunit kinase IKK responds to a large variety of cellular stimuli as well as the signal triggered by the HTLV Tax protein (26,30,63). Although precisely how IKK responds to the diverse signals remains unclear, recent genetic evidence suggests that the regulatory subunit, IKK␥, plays an essential role in IKK activation by both cellular signals and Tax (35,42,43). An interesting structural feature of IKK␥ is the presence of several stretches of coiled-coil motifs, including a C-terminal LZ (36). This LZ has been proposed to mediate interaction with upstream IKK activators (34). In the present study, we have identified an N-terminal LZ (LZn), which shares striking overall sequence similarities with the C-terminal LZ (termed LZc in this study). A unique feature of the IKK␥ LZs is the presence of a highly conserved UR motif. We have shown that the LZn is particularly important for IKK␥ binding to Tax; deletion of either the heptad leucine repeats (LRn) or the URn severely crippled the Tax binding activity of IKK␥ (Fig. 1D). The LZc seems to be also involved in the IKK␥-Tax interaction; deletion of URc significantly diminished the Tax binding activity of IKK␥, although deletion of the leucine-repeat region of this LZ (LRc) did not produce a significant effect.
Previous studies suggest that the region of Tax containing the M22 mutation is important for Tax dimerization and activation of NF-B (64). Interestingly, this region contains three continuous heptad leucine repeats ( Fig. 2A). Single substitution of the conserved leucines by alanines results in Tax mutants incapable of IKK␥ binding, suggesting the importance of these hydrophobic residues in the Tax-IKK␥ interaction. It is

FIG. 5. Fusion of interaction-defective Tax mutants to IKK␥ rescues their functional defect in IKK stimulation and NF-B activation.
A physical interaction of IKK␥⅐Tax chimeras with IKK␤. 293 cells were transfected with an empty vector, Tax, IKK␥ plus Tax, IKK␥ plus Tax M22, or the indicated IKK␥⅐Tax chimeras, along with the HA⅐IKK␤. Cell lysates were subjected to IP with anti-Tax followed by immunoblotting with anti-HA (upper). The HA⅐IKK␤ coprecipitated with Tax or IKK␥⅐Tax chimeras as indicated. The free forms of Tax and IKK␥⅐Tax fusion proteins in the cell lysates were detected by immunoblotting using anti-Tax (middle and lower). B, IKK activation by Tax and IKK␥⅐Tax fusion proteins in 293 cells. 293 cells were transfected in 6-well plates with cDNA expression vectors encoding either the indicated free forms of Tax and IKK␥ proteins or IKK␥⅐Tax fusion proteins along with HA⅐IKK␣ and HA⅐IKK␤ (0.5 g for free forms of Tax, 10 ng for IKK␥, 10 ng for HA⅐IKK␣, 5 ng for HA⅐IKK␤, 0.2 g for IKK␥⅐Tax chimeras). Cell extracts were analyzed by an immunocomplex kinase assay using a GST⅐IB␣-(1-54) as substrate. The phosphorylated GST⅐IB␣-(1-54) (IB␣-P) in the kinase assay (KA, top) is indicated. HA⅐IKK␣ and HA⅐IKK␤ were detected by IB with anti-HA (lower). The expression level for the free and chimera forms of Tax proteins was similar to that shown in A. C, IKK activation by Tax and IKK␥⅐Tax fusion proteins in Jurkat Tag cells. Jurkat-Tag cells were transfected with cDNA expression vectors encoding either the indicated free forms of Tax and IKK␥ proteins or IKK␥⅐Tax fusion proteins along with HA⅐IKK␣ and HA⅐IKK␤ (1 g for free forms of Tax, 10 ng for IKK␥, 10 ng for HA⅐IKK␣, 5 ng for HA⅐IKK␤, 0.4 g for IKK␥⅐Tax chimeras). Cell extracts were analyzed by KA and IB as described in B. The results from transfection of vector and Tax wild-type alone are the same as those presented in Fig. 4B (lanes 1 and 2). D, IKK␥ expression levels in untransfected and transfected cells. 293 or Jurkat Tag cells were transfected with the indicated expression vectors, as described in B or C, followed by immunoblotting analysis using anti-IKK␥. were transfected with expression vectors encoding the indicated free forms of Tax or Tax mutants (1 g) together with IKK␥ (0.2 g) or the IKK␥⅐Tax chimera constructs (0.4 g). All the cells were also transfected with the B-Luc reporter (0.2 g). Luciferase activity was determined and presented as in Fig. 4. B, wild-type Jurkat T cells (5 ϫ 10 6 ) were transfected with expression vectors encoding the indicated free forms of Tax (indicated amounts) or mutants (80 ng) together with IKK␥ (80 ng) or the IKK␥⅐Tax chimera constructs (20 ng). All the cells were also transfected with the B-Luc reporter (0.2 g). Luciferase activity was determined and presented as in Fig. 4. currently unclear whether the LRR of Tax forms any type of protein-protein interaction domain. Computer analysis did not reveal a significant potential of this region for the formation of a coiled-coil structure, a characteristic of LZ motifs. Thus, it is unlikely that the Tax-IKK␥ binding is mediated through mutual LZ interaction. Interestingly, we have found that the upstream region of the Tax LRR contains a sequence element that is homologous to the UR motifs of IKK␥ LZs. When analyzed by the PREDATOR program (61), the UR motif predicts an ␣-helix (data not shown). At least in Tax, the UR motif appears to be well exposed, because it is located at one of the previously determined protease-hypersensitive sites of Tax (17). Compared with the leucine repeats, the UR motif of Tax seems to be less sensitive to mutations; single substitution of the conserved residues in this motif only partially inhibited the Tax/IKK␥ interaction (Fig. 2C). One hypothesis would be that the leucine repeats of Tax are critical for homodimerization or maintaining a proper conformation of Tax required for its interaction with IKK␥, while the UR motif forms an interaction surface. However, it is clear that a fundamental understanding of the biochemical mechanism mediating the Tax/IKK␥ interaction will rely on structural analysis of these proteins.
Prior studies suggest that Tax activation of the NF-B and CREB/ATF cellular transcription factor pathways is mediated by different mechanisms (19). Tax activation of CREB/ATF involves physical interaction of Tax with the bZIP motifs of the CREB/ATF factors (12)(13)(14)66), whereas Tax activation of NF-B occurs indirectly by stimulating the IKK signaling machinery (27)(28)(29)49). Our finding that Tax interacts with IKK␥ via LZ motifs in IKK␥ suggests that the initiation of the two Tax-specific cellular pathways may be fundamentally similar, both involving LZ-mediated molecular interaction. This idea does not contradict the previous finding that certain Tax mutants can functionally segregate the NF-B and CREB/ATF pathways (19), because the development of these two pathways does involve many different mechanisms. In this regard, the Tax M22 mutant, known to have a specific defect in the NF-B pathway, has recently been shown to be partially defective in bZIP binding and thus exhibits reduced activity in CREB/ATF activation (65). As a result of the studies described in this work, it is apparent that the M22 mutant harbors a mutation within the LRR of Tax and is defective in IKK␥ binding. We have also shown that several other Tax mutants harboring LRR mutations exhibit a defect in both the NF-B and CREB/ATF pathways (data not shown). These results raise the possibility that the LRR of Tax is important for Tax binding to both IKK␥ and the CREB/ATF bZIP proteins. Of course, as discussed above, the LRR may not form the direct interaction site but rather mediate Tax homodimerization, which in turn may be required for Tax binding to CREB and IKK␥. In this regard, the N terminus of Tax has been suggested to interact with CREB (67). Our data indicate that the UR of Tax may possibly form an interaction surface for binding to IKK␥.
By generating a number of Tax and IKK␥ mutants and IKK␥⅐Tax fusion proteins, we have provided functional data demonstrating that binding of Tax to IKK␥ serves as an essential step in Tax-stimulated IKK activation. We have shown that Tax mutants defective in IKK␥ binding fail to activate IKK. Interestingly, this functional defect of the Tax mutants can be rescued by their fusion to IKK␥. We have found that the NF-B stimulatory function of the IKK␥⅐Tax chimeras is more efficient in an IKK␥-deficient Jurkat cell line (42) than in parental Jurkat cells (Fig. 6). This result may reflect the fact that the IKK␥⅐Tax fusion proteins have to compete with endogenous IKK␥ for engaging IKK␣/IKK␤ in the parental Jurkat cells, whereas such competition would not occur in the IKK␥-defi-cient cells. Nevertheless, the functional rescue of Tax mutants in the context of IKK␥ fusions can also be readily detected in parental Jurkat cells (Fig. 6B). Interestingly, when fused to IKK␥, Tax becomes completely inactive in transactivating the HTLV-I LTR (data not shown). Our preliminary studies suggest that this functional inactivation is likely caused by the cytoplasmic sequestration of the IKK␥⅐Tax chimeras (data not shown). This finding supports the previous report that cytoplasmic forms of Tax retain their NF-B-inducing function (64), and also suggests that the linkage of Tax to IKK␥ specifically targets Tax to the NF-B signaling pathway.