The Tax Oncoprotein of Human T-cell Leukemia Virus Type 1 Associates with and Persistently Activates IκB Kinases Containing IKKα and IKKβ*

The Tax oncoprotein of human T-cell leukemia virus type 1 (HTLV1) chronically activates transcription factor NF-κB by a mechanism involving degradation of IκBα, an NF-κB-associated cytoplasmic inhibitor. Tax-induced breakdown of IκBα requires phosphorylation of the inhibitor at Ser-32 and Ser-36, which is also a prerequisite for the transient activation of NF-κB in cytokine-treated T lymphocytes. However, it remained unclear how Tax interfaces with the cellular NF-κB/IκB signaling machinery to generate a chronic rather than a transient NF-κB response. We now demonstrate that Tax associates with cytokine-inducible IκB kinase (IKK) complexes containing catalytic subunits IKKα and IKKβ, which mediate phosphorylation of IκBα at Ser-32 and Ser-36. Unlike their transiently activated counterparts in cytokine-treated cells, Tax-associated forms of IKK are constitutively active in either Tax transfectants or HTLV1-infected T lymphocytes. Moreover, point mutations in Tax that ablate its IKK-binding function also prevent Tax-mediated activation of IKK and NF-κB. Together, these findings suggest that the persistent activation of NF-κB in HTLV1-infected T-cells is mediated by a direct Tax/IKK coupling mechanism.

Human T-cell leukemia virus type 1 (HTLV1) is the etiologic agent of an aggressive malignancy of activated CD4 ϩ T lymphocytes (16). The HTLV1 provirus encodes a 40-kDa oncoprotein, termed Tax, which potently induces the constitutive nuclear expression of NF-B (17)(18)(19). Studies with Tax-transgenic mice suggest that this viral/host interaction is required to maintain the transformed phenotype of HTLV1-infected cells (20). We demonstrated recently that Tax converts IB␣ into a labile proteasome substrate by a mechanism involving phosphorylation of the inhibitor at Ser-32 and Ser-36 (7,21,22). These findings suggested that Tax induces a chronic NF-B response by acting upstream of one or more IKKs (7). However, the precise mechanism by which Tax accesses the host NF-B signaling pathway remained unknown.
We demonstrate here that cytokine-inducible IB kinases containing IKK␣ and IKK␤, which normally function in a transient manner (12)(13)(14), are expressed as constitutively active signal transducers in HTLV1-infected T lymphocytes. These activated forms of IKK associate stably with Tax when the oncoprotein is expressed in either transiently transfected or virally infected cells. Point mutations in Tax that ablate its NF-B-inducing activity also prevent the formation of functional IKK/Tax complexes. Our findings suggest that Tax persistently activates IKK proteins via a direct binding mechanism, resulting in the sustained degradation of IB␣ and chronic nuclear expression of NF-B in HTLV1-infected cells.

EXPERIMENTAL PROCEDURES
Reagents-The anti-IKK␣ antiserum was generated by immunizing rabbits with a synthetic peptide corresponding to amino acids 1-29 of human IKK␣, which was covalently coupled to keyhole limpet hemocyanin. 2 Purified polyclonal antibodies specific for IKK␤ were provided by Frank Mercurio (Signal Pharmaceuticals, Inc., San Diego, CA). Polyclonal (23) and monoclonal (Tab-170) anti-Tax antibodies were obtained from Bryan Cullen (Duke University) and John Brady (NCI, Bethesda, MD), respectively. Complementary DNAs encoding wild-type and mutant forms of (23) were cloned into the expression vector pCMV4 (24). Recombinant glutathione S-transferase (GST)-IB␣ fusion proteins were overexpressed and purified by affinity chromatography as described (12).
Immunoprecipitation and Immunoblotting-Cytoplasmic extracts were prepared from cultured cells by detergent lysis (30) in the presence of phosphatase (1 mM Na 4 P 2 O 7 , 1 mM NaF, 10 M Na 2 MoO 4 , 1 mM Na 3 VO 4 ) and protease (31) inhibitors. Lysates were adjusted to a final concentration of 50 mM HEPES (pH 7.0), 250 mM NaCl, and 5 mM EDTA before the addition of specific antisera. Immunoprecipitations were typically carried out in 300-l reaction mixtures containing 200 g of total cytoplasmic protein and 15 l of protein A-Sepharose beads (Zymed). Immunoprecipitates were washed with ELB buffer (7) in the presence of phosphatase inhibitors, resolved by SDS-PAGE, and transferred to polyvinylidine difluoride membranes. Immunoreactive polypeptides were detected using an enhanced chemiluminescence system (Amersham Pharmacia Biotech).

Cytoplasmic IKKs Are Constitutively Active in HTLV1-infected T-cells-Whereas
NF-B is normally sequestered in the cytoplasm by IB proteins, this transcription factor is persistently expressed in the nuclei of HTLV1-infected T-cells (17)(18)(19). We demonstrated previously that Tax stimulates phosphorylation of the NF-B inhibitor IB␣ at Ser-32 and Ser-36, which targets it to the ubiquitin-proteasome pathway (7,21,22). Site-specific phosphorylation of IB␣ is also required for the transient activation of NF-B in response to the cytokine TNF (7-10). As such, we reasoned that Tax might persistently activate an IKK, resulting in the sustained rather than transient nuclear action of NF-B. In this regard, recent studies have identified a multisubunit IKK complex containing two interactive catalytic components, termed IKK␣ and IKK␤, which mediate site-specific phosphorylation of IB␣ at Ser-32 and Ser-36 in TNF-treated cells (11)(12)(13)(14)(15). However, the relationship between these cytokine-inducible IKKs and Tax remained unknown.
To explore this relationship, cytoplasmic and nuclear extracts were prepared from three HTLV1-transformed T-cell lines, including MT-2, SLB-1, and C8166. Whereas the MT-2 and SLB-1 lines produce replication-competent virions (25,26), C8166 cells harbor a defective provirus that selectively expresses Tax (27). Subcellular extracts were also prepared from Jurkat T-cells, which are TNF-responsive (32) and transformed by an HTLV1-independent mechanism. As demonstrated in gel shift assays, treatment of Jurkat T-cells with TNF led to the rapid induction of NF-B DNA binding activity in the nuclear compartment (Fig. 1A, lanes 1 and 2, top panel). However, NF-B was constitutively activated in each of the three HTLV1-infected lines (Fig. 1A, lanes 3-5, top panel). This constitutive pattern of nuclear NF-B activity correlated with the expression of Tax protein in the cytoplasmic compartment of MT-2, SLB-1, and C8166 cells (Fig. 1A, lanes 3-5, bottom  panel).
To monitor IKK activity in HTLV1-infected cells, in vitro phosphorylation assays were performed using a GST-IB␣ (amino acids 1-54) fusion protein as substrate (12). Endogenous IKKs were isolated by immunoprecipitation with either IKK␣ or IKK␤ peptide-specific antisera and incubated with the GST-IB␣ substrate and [␥-32 P]ATP. Consistent with prior studies (11)(12)(13)(14)(15), IB kinase activity directed by either IKK␣ or IKK␤ in Jurkat T-cells was modestly stimulated by treatment with TNF for 10 min (Fig. 1B, lanes 1 and 2). However, in HTLV1-infected T-cells, IKK␣-and IKK␤-associated kinase activities were constitutively expressed (Fig. 1B, lanes 3-5). These IKK activities were specific for Ser-32 and Ser-36 of IB␣, because replacement of both sites with alanine in the GST-IB␣ substrate eliminated phosphoryl group transfer (Fig. 1B, lanes 6 -10). As demonstrated by immunoblotting (Fig. 1C), steady-state levels of the IKK␣ protein were reduced in HTLV1-infected cells relative to unstimulated Jurkat T-cells (top panel), whereas IKK␤ protein expression was modestly elevated (bottom panel). These studies suggest that Tax persistently activates endogenous IKKs by a post-translational mechanism. Because the predominant form of IKK appears to be dependent on the activities of both IKK␣ and IKK␤ (13), we were prepared as described previously (30). Nuclear extracts (5 g of protein) were added to reaction mixtures containing a 32 P-labeled B probe and nucleoprotein complexes were resolved by electrophoresis on a native 5% polyacrylamide gel (top panel) (31). Cytoplasmic extracts were immunoprecipitated with a peptide-specific antiserum for HTLV1 Tax (23) infer that both of these interactive catalytic subunits are affected in cells harboring HTLV1.
HTLV1 Tax Associates Physically with Cytoplasmic IKKs-To explore the possibility that Tax persistently activates IKKs via specific protein/protein interactions, Tax was immunoprecipitated from the cytosol of HTLV1-infected T-cells and assayed for the presence of IB␣ kinase activity. As shown in Fig. 2A, GST-IB␣ phosphorylating activity was detected in Tax immunoprecipitates isolated from virally infected lymphocyte lines (lanes 2-4), but not in control immunoprecipitates derived from Jurkat T-cells (lane 1). The IB kinase activity associated with Tax exhibited the appropriate sequence specificity, because disruption of Ser-32 and Ser-36 in IB␣ eliminated phosphoryl group transfer ( Fig. 2A, lanes 5-8). These results suggest that Tax forms stable cytoplasmic complexes with active IKKs in HTLV1-infected T-cells.
To determine whether either IKK␣ or IKK␤ mediates this Tax-associated kinase activity, monoclonal anti-Tax antibodies were used to isolate cytoplasmic Tax from MT-2, SLB-1, and C8166 cells. Tax immunocomplexes were then fractionated by SDS-PAGE and probed on immunoblots with a mixture of IKK␣ and IKK␤ peptide-specific antibodies. As shown in Fig.  2B (lanes 2-4), endogenous IKK␣ was detected in association with Tax in each of the HTLV1-infected cell lines tested, whereas a Tax-bound form of the IKK␤ protein was also evident in SLB-1 cells. To extend these findings, we used peptidespecific antibodies to immunoprecipitate either IKK␣ or IKK␤ from SLB-1 cells and then assayed for the presence of Tax in the resultant IKK immunocomplexes. Tax was readily detected in complexes containing either of these IKKs, albeit at higher levels in the IKK␣ immunoprecipitates (Fig. 2C, lanes 2 and 3).
These results provide direct biochemical evidence that Tax associates stably with IKKs in HTLV1-infected cells.
Persistent Activation of IKK Proteins by Tax Requires the Formation of Tax-IKK Complexes-In light of these findings, we reasoned that the association of Tax with IKKs might be a prerequisite for IKK activation. To test this hypothesis, 293T cells were transfected with cDNA expression vectors for wildtype Tax (Tax-WT) and forms of the oncoprotein containing missense mutations that selectively disrupt its ability to access either the CREB/ATF (Tax-M47) or the NF-B/Rel (Tax-M22) transcription factor pathway (23). Subcellular extracts were then prepared and subjected to gel shift, co-immunoprecipitation, and in vitro phosphorylation analyses. As shown in Fig.  3A (top panel), Tax-M22 failed to induce the nuclear expression of NF-B DNA binding activity in transfected cells (lane 3), whereas wild-type Tax and the Tax-M47 mutant were fully competent to execute this function (lanes 2 and 4). All three of the Tax constructs were expressed at comparable levels in the cytosol of 293T transfectants (Fig. 3A, bottom panel).
To assess the status of endogenous IKK activity in these Tax-expressing cells, cytoplasmic extracts were subjected to immunoprecipitation with antibodies for either Tax, IKK␣, or IKK␤ and the resultant immune complexes were monitored for GST-IB␣ phosphorylating activity. As shown in Fig. 3B (top  panel), immune complexes containing either wild-type Tax or Tax-M47 possessed a significant level of IKK activity (lanes 2  and 4). In contrast, immunoprecipitates derived from cells transfected with either a blank expression vector or the Tax-M22 cDNA failed to affect GST-IB␣ phosphorylation (lanes 1  and 3). Similar qualitative results were obtained in experiments conducted with IKK␣ (Fig. 3B, middle panel) and IKK␤ (Fig. 3B, lower panel) immunoprecipitates, confirming that the mutation in Tax-M22 disrupted the functional interplay between Tax, IKK, and NF-B.
To determine whether the mutation in Tax-M22 affected its FIG. 2. Constitutively active IKKs associate stably with HTLV1 Tax. A, cytoplasmic extracts derived from the indicated cell lines were subjected to immunoprecipitation with a Tax-specific antiserum. Resultant immune complexes were assayed for either IB kinase activity with the indicated substrates (upper panel) or probed for the presence of Tax protein on immunoblots with monoclonal antibodies (lower panel). B, cytoplasmic extracts from the indicated cell lines were subjected to immunoprecipitation with monoclonal anti-Tax antibodies. Resultant immune complexes were fractionated by SDS-PAGE, transferred to polyvinylidine difluoride membranes, and immunoblotted with a combination of IKK␣and IKK␤-specific antibodies. C, cytoplasmic extracts from SLB-1 cells were subjected to immunoprecipitation with normal rabbit serum (lane 1) or the indicated IKK-specific antibodies (lanes 2 and 3), and resultant immune complexes were probed with monoclonal anti-Tax antibodies on immunoblots as described in B.

FIG. 3. Activation of IKK and NF-B by HTLV1
Tax correlates with its IKK-binding function. Nuclear and cytoplasmic extracts were prepared from 293T cells (1 ϫ 10 6 ) following transient transfection with the indicated expression vectors (7.5 g each). A, nuclear extracts (5 g) were added to reaction mixtures containing a 32 P-labeled B probe. Resultant DNA-protein complexes were resolved by electrophoresis on a native 5% polyacrylamide gel and visualized by autoradiography (top panel) (31). Cytoplasmic extracts were subjected to immunoprecipitation with a Tax-specific antiserum, and resultant immune complexes were probed for HTLV1 Tax on immunoblots using monoclonal anti-Tax antibodies (bottom panel). B, cytoplasmic extracts were subjected to immunoprecipitation with peptide-specific antisera for either Tax (top panel), IKK␣ (middle panel), or IKK␤ (bottom panel). Immune complexes were assayed for IB kinase activity in reaction mixtures containing [␥-32 P]ATP and the GST-IB␣ substrate. Reaction products were resolved by SDS-PAGE, and phosphoproteins were visualized by autoradiography. C, cytoplasmic extracts were subjected to immunoprecipitation with an IKK␣-specific antiserum. Resultant immune complexes were fractionated by SDS-PAGE and probed on immunoblots with antisera for either IKK␣ (top panel) or HTLV1 Tax (bottom panel). Positions of molecular size markers (in kilodaltons) and immunoglobulin heavy chains (IgH) are indicated.
IKK-binding function, cytoplasmic extracts from these transfectants were subjected to immunoprecipitation with an IKK␣specific antiserum and resultant immune complexes were analyzed for the presence of either IKK␣ or Tax. Comparable amounts of endogenous IKK␣ protein were immunoprecipitated from cells overexpressing either wild-type Tax, Tax-M22, or Tax-M47 (Fig. 3C, lanes 2-4, top panel). Consistent with their capacity to activate both IKK and NF-B, wild-type Tax and Tax-M47 were detected in association with IKK␣ (Fig. 3C,  lanes 2 and 4, bottom panel). However, formation of Tax/IKK complexes was not apparent in cells overexpressing Tax-M22, which fails to activate IKK and NF-B (Fig. 3C, lane 3, bottom  panel). Taken together, these findings provide a strong correlation between the IKK-binding function of Tax and the ability of this oncoprotein to persistently activate IKK and NF-B.

DISCUSSION
The Tax protein of HTLV1 and the cytokine TNF both activate NF-B by a mechanism involving signal-dependent phosphorylation of IB␣ at Ser-32 and Ser-36 (7)(8)(9)(10). In turn, hyperphosphorylated IB␣ is targeted for destruction by the ubiquitin-proteasome pathway (21,33). The IB␣ gene is under NF-B control, which provides a negative-feedback loop to ensure the transient action of NF-B in TNF-treated cells (34,35). In contrast, NF-B appears to escape from feedback inhibition by newly synthesized IB␣ in cells expressing Tax (22). We have recently shown that treatment of HTLV1-infected T lymphocytes with proteasome inhibitors leads to the rapid accumulation of hyperphosphorylated IB␣ (22). However, the precise mechanism by which HTLV1 Tax interfaces with the host NF-B signaling machinery to elicit a persistent pattern of nuclear NF-B expression remained unclear.
In the present report, we provide a unifying explanation for these prior experimental observations that links the constitutive activities of Tax, NF-B, and IB kinases known to mediate TNF-inducible phosphorylation of IB␣ at Ser-32 and Ser-36 (11)(12)(13)(14)(15). Four new lines of biochemical and genetic evidence indicate that these cytokine-responsive IKKs are primary cellular targets of HTLV1 Tax in the host NF-B signaling pathway. First, IKK complexes containing either IKK␣ or IKK␤ are constitutively activated in HTLV1-infected T-cells, but not in Tax-deficient Jurkat T-cells. Second, Tax is stably associated with these IKKs in HTLV1-infected cells. Third, overexpression of Tax in mammalian cells is sufficient to activate endogenous IKKs. Fourth, mutations in Tax that ablate its IKK-inducing function also prevent Tax-mediated activation of NF-B. We conclude that newly synthesized IB␣ in HTLV1-infected cells is rapidly phosphorylated at Ser-32 and Ser-36 by constitutively active IKKs, resulting in continuous IB␣ turnover and persistent NF-B expression (7,21,22).
How does the Tax transforming protein of HTLV1 convert cytokine-inducible IKKs into constitutively active kinases? One clue to this mechanistic question is our finding that Tax associates with IKK complexes in HTLV1-infected T-cells. Moreover, mutations in Tax that disrupt its IKK-binding function also interfere with Tax-mediated activation of IKK, suggesting that the formation of Tax-IKK complexes is required for IKK activation. Accordingly, Tax might mimic a positive regulatory subunit that recruits an activated upstream signal transducer to the IKK complex. Two attractive candidates for this upstream signal transducer are the NF-B-inducing kinase (NIK) and the related mitogen-activated protein kinase kinase kinase MEKK1, both of which have been shown to activate NF-B and IKK proteins when overexpressed in vivo (36 -39). Alternatively, Tax could function to displace an inhibitory subunit within the IKK complex. Indeed, indirect evidence for the existence of cellular IKK inhibitors has been reported previously (12,40).