Human T-lymphotropic Virus Type I Tax Activates I-κB Kinase by Inhibiting I-κB Kinase-associated Serine/Threonine Protein Phosphatase 2A*

I-κB kinase (IKK) is a serine/threonine kinase that phosphorylates I-κBα and I-κBβ and targets them for polyubiquitination and proteasome-mediated degradation. IKK consists of two highly related catalytic subunits, α and β, and a regulatory γ subunit, which becomes activated after serine phosphorylation of the activation loops of the catalytic domains. The human T-lymphotropic retrovirus type-I trans-activator, Tax, has been shown to interact directly with IKKγ and activates IKK via a mechanism not fully understood. Here we demonstrate that IKK binds serine/threonine protein phosphatase 2A (PP2A), and via a tripartite protein-protein interaction, Tax, IKKγ, and PP2A form a stable ternary complex.In vitro, PP2A down-regulates active IKK prepared from Tax-producing MT4 cells. In the presence of Tax, however, the ability of PP2A to inactivate IKK is diminished. Despite their interaction with IKKγ, PP2A-interaction-defective Tax mutants failed to activate NF-κB. Our data support the notion that IKKγ-associated PP2A is responsible for the rapid deactivation of IKK, and inhibition of PP2A by Tax in the context of IKK·PP2A·Tax ternary complex leads to constitutive IKK and NF-κB activation.

I-B kinase (IKK) is a serine/threonine kinase that phosphorylates I-B␣ and I-B␤ and targets them for polyubiquitination and proteasome-mediated degradation. IKK consists of two highly related catalytic subunits, ␣ and ␤, and a regulatory ␥ subunit, which becomes activated after serine phosphorylation of the activation loops of the catalytic domains. The human T-lymphotropic retrovirus type-I trans-activator, Tax, has been shown to interact directly with IKK␥ and activates IKK via a mechanism not fully understood. Here we demonstrate that IKK binds serine/threonine protein phosphatase 2A (PP2A), and via a tripartite proteinprotein interaction, Tax, IKK␥, and PP2A form a stable ternary complex. In vitro, PP2A down-regulates active IKK prepared from Tax-producing MT4 cells. In the presence of Tax, however, the ability of PP2A to inactivate IKK is diminished. Despite their interaction with IKK␥, PP2A-interaction-defective Tax mutants failed to activate NF-B. Our data support the notion that IKK␥associated PP2A is responsible for the rapid deactivation of IKK, and inhibition of PP2A by Tax in the context of IKK⅐PP2A⅐Tax ternary complex leads to constitutive IKK and NF-B activation.
NF-B/Rel family of transcription factors are controlled by inhibitory I-B proteins I-B␣ and I-B␤ and the I-B-like domains in NF-B1 and NF-B2 that sequester NF-B/Rel in the cytoplasm as multiprotein complexes (for reviews, see Ref. [1][2][3]. Upon activation by extracellular stimuli such as interleukin-1, tumor necrosis factor-␣, bacterial lipopolysaccharide, or by human T-lymphotropic virus type I (HTLV-I) 1 Tax, I-B␣ and I-B␤ become serine-phosphorylated and polyubiquinated and are rapidly degraded via proteasome-mediated proteolysis, resulting in heightened nuclear levels of NF-B and increased expression of a plethora of cellular genes under NF-B regulation, including the genes of many cytokines and their receptors, adhesion molecules, and immune modulators (1)(2)(3). Dysregulation and/or hyperactivation of the NF-B/I-B regulatory pathway as caused by chromosomal translocation (4), oncogene transduction (5), viral infection, or targeted gene disruption (6,7) leads to cancers of the hematopoietic cells or chronic inflammatory diseases.
The phosphorylation of I-B␣ and I-B␤ is mediated by a kinase called I-B kinase (IKK) (8,9). The core IKK enzyme consists of two highly homologous catalytic subunits ␣ and ␤ of 85 and 87 kDa in sizes, respectively, and a 48-kDa regulatory subunit, IKK-␥/NEMO (referred to as IKK-␥ herein) (9,10). Both IKK-␣ and IKK-␤ contain NH 2 -terminal kinase domains followed by leucine zippers and helix-loop-helix domains that mediate protein-protein interactions important for IKK oligomerization and kinase activity (8,9). Likewise, IKK-␥ also contains extensive helical regions and leucine zipper domains that engage in protein-protein interaction (9,10). In vivo the IKK holoenzyme exists as a large protein complex of at least 700 -900 kDa in size (8,9). It is not clear what other protein components are present in the holo-IKK enzyme complex in addition to IKK-␣, IKK-␤, and IKK-␥. Several members of the I-B and NF-B/Rel families of proteins and mitogen-activated protein kinase phosphatase-1 (MKP), MEK kinase, and NF-B inducing kinase (NIK) have been reported to interact with IKK, although these proteins have not been found to co-elute with the 900-kDa IKK complex chromatographically (9).
The transactivator/oncoprotein of HTLV-I, Tax, has been shown to activate IKK constitutively (11)(12)(13)(14)(15)(16). Tax-mediated IKK activation is due in part to a direct interaction between Tax and IKK-␥ (14,16,18). The molecular mechanism via which Tax affects IKK activation after its association with IKK-␥ remains incompletely understood, however. We have found recently that Tax can interact directly with the catalytic subunit of the major serine/threonine protein phosphatase 2A (PP2Ac) in vivo and in vitro. 2 Furthermore, Tax acts as a noncompetitive inhibitor of PP2A in in vitro assays where 32 P-labeled glycogen phosphorylase a is used as a PP2A substrate. 2 Consistent with the notion that Tax inhibits the activity of PP2A, MEK phosphorylation and to a lesser extent cAMP-response element-binding protein and ATF-1 phosphorylation in Tax-expressing HTLV-I-transformed cells and in Tax-transfected human embryonic kidney 293 cells are greatly elevated. 2 Most interestingly, PP2A interaction-defective mutants of Tax fail to stimulate MEK phosphorylation and are unable to activate NF-B, thus suggesting a link between the inhibition of PP2A by Tax and NF-B activation. 2 PP2A is a major serine/threonine protein phosphatase in all eukaryotic cells (for comprehensive reviews, see Refs. 19 -23). It is crucial for the negative regulation of multiple cellular processes. The holo-PP2A enzyme is a heterotrimer that consists of a core enzyme formed by the highly conserved A subunit (60 kDa, 2 human isoforms) and C (catalytic) subunit (36 kDa, 2 human isoforms) together with many different regulatory B-subunits that derive from multiple genes (three distinct families: B/B55/PR55 (3 genes), BЈ/B56/PR61 (Ͼ3 genes), and BЈЈ/ PR72/PR130 (1 gene)), and their alternatively spliced mRNAs (19 -21, 23, 24). The crystal structure of the A-subunit shows it to contain 15 tandem helical repeats (HEAT motifs) that assume a shape that resembles a horseshoe (25). It serves as a scaffold to which C-and B-subunits are attached (19 -21, 23). The heterodimer formed by A-and C-subunits and the heterotrimer containing all three subunits are termed core enzyme and holoenzyme, respectively (19 -21, 23). A role of PP2A in controlling the signal transduction pathway that leads to IKK/ NF-B activation has been reported previously (8). Okadaic acid, an inhibitor of PP2A, has been shown to activate NF-B in vivo (26,27). Furthermore, PP2A can inactivate the kinase activity of IKK in vitro (8). Here we provide evidence to show that via a tripartite interaction, Tax, PP2A, and IKK␥ form a ternary complex. In in vitro IKK assays, Tax reduces the ability of PP2A to inactivate IKK. Consistent with the notion that activation of IKK requires that Tax bind to both IKK␥ and PP2A, several Tax mutants that are abrogated or attenuated for PP2A binding fail to activate NF-B even though they continue to bind IKK␥. These results indicate that, in the context of the IKK⅐PP2A⅐Tax complex, PP2A activity is inhibited or diminished. In essence, IKK is activated by serine phosphorylation of its activation loop upon extracellular stimulation. In normal cells, phospho-IKK becomes rapidly inactivated by IKK␥-associated PP2A, returning IKK to a resting state. In HTLV-I infected or transformed cells, PP2A inhibition by IKK␥-bound Tax maintains IKK in the phosphorylated and active form, causing constitutive phosphorylation and degradation of I-B, which in turn leads to nuclear presence of NF-B/ Rel and potent activation of genes under NF-B/Rel control.

EXPERIMENTAL PROCEDURES
Reagents and Cell Culture-The anti-Tax monoclonal antibody 4C5 is of IgG2a subtype and reacts with amino acid residues 333-353 of Tax. 3 The anti-PP2A C-subunit monoclonal antibody (the COOH-terminal amino acid residues 295-309 of the catalytic subunit of human PP2A) and anti-IKK␥ (amino acids 1-419, representing full-length human IKK␥) monoclonal antibody were from Upstate Biotechnology, Inc. and Santa Cruz Biotechnology, Inc., respectively. The purified PP2A was derived from human red blood cells and consists of both A and C subunits (Upstate Biotechnology). Jurkat and the HTLV-I-transformed T-cell line MT-4 were cultured in RPMI medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and 100 units/ml penicillin (Invitrogen) in the presence of 10% CO 2 .
Immunoprecipitation and Western Blotting-Immunoprecipitations were carried out using extracts prepared from Jurkat and HTLV-I-transformed MT-4 cells. Ten million cells were harvested, washed 3 times, each with 10 ml of phosphate-buffer saline, and lysed by repeated passage through a 27.5-gauge syringe in 1 ml of lysis buffer (100 mM NaCl, 50 mM Tris-HCl, pH 8.0, 1% Nonidet P-40, 2 mM EDTA, 5 mM NaF, 1 mM Na 3 VO 4 , 0.2 mM phenylmethylsulfonyl fluoride) containing 50 g/ml each of the protease inhibitors pepstatin, leupeptin, bestatin, and aprotinin (Roche Molecular Biochemicals). Cell debris was removed by centrifugation at 12,000 rpm in a microcentrifuge for 10 min at 4°C. Immunoprecipitation was carried out with 200 g of proteins in 500 l of cell lysates. Briefly, cell lysates were precleared by incubation with 100 l of a 50% slurry of protein G-agarose (Invitrogen) and 2 g of normal mouse antiserum in 5 l for 30 min at 4°C followed by centrifugation at 12,000 rpm for 5 min. After preclearing, 2 g of a Tax monoclonal antibody (4C5), PP2A C-subunit monoclonal antibody, or IKK␥ monoclonal antibody were added to each sample. The reactions were preincubated at 4°C for 1 h. After the addition of 30 l of protein G-agarose to each reaction, the mixtures were incubated overnight. On the following day, immune complexes were pelleted by centrifugation at 3000 rpm for 5 min at 4°C, washed 3 times with 0.8 ml of lysis buffer, resuspended in 30 l of SDS-PAGE loading buffer, heated at 95°C for 3 min, and centrifuged, and then 15 l of sample were loaded and resolved on a 12% SDS-polyacrylamide gel containing a 4% stacking gel. Proteins resolved in the gels were transferred to nitrocellulose membranes (Schleicher & Schuell). For Western blotting, membranes were incubated with blocking buffer (50 mM Tris-HCl (pH 8.0), 100 mM NaCl, 0.05% Tween, 0.02% sodium azide, 5% nonfat dry milk) for 3 h and then probed in the same buffer for 2 h with mouse monoclonal antibodies against PP2Ac, IKK␥, or Tax. After washing, the blots were incubated for 1 h in the same buffer containing an anti-mouse horseradish peroxidase-conjugated, secondary antibody (diluted 1:1000, Santa Cruz Biotechnology), washed three times, developed using a chemiluminescent substrate (SuperSignal; Pierce), and exposed to x-ray films.
Glutathione S-Transferase Pull-down Assay-For GST pull-down experiments, ϳ500 ng each of purified GST, GST-Tax, or GST-IKK␥ was incubated with 300 ng of purified PP2A or PP2A and Tax in 30 l of 1ϫ binding buffer (25 mM HEPES (pH 7.9), 5 mM KCl, 0.5 mM MgCl 2 , 0.5 mM EDTA, 1 mg/ml bovine serum albumin, 10% glycerol, 0.15% Nonidet P-40, 0.25 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride) for 30 min at 30°C. Forty microliters of a 50% slurry of prewashed glutathione-Sepharose 4B (Amersham Biosciences) were added to each binding reaction, and samples were incubated for 1 h with agitation at 4°C. After incubation, the protein-bound Sepharose beads were washed 3 times with 800 l of 1ϫ binding buffer, pelleted by centrifugation at 1200 rpm for 5 min, resuspended in 30 l of SDS-PAGE loading buffer, and heated to 95°C, and 15 l from each reaction were resolved by 12% SDS-PAGE followed by immunoblotting, as described. To remove the GST moiety from GST-IKK␥, the fusion protein was proteolysed with 0.2 unit of thrombin (Sigma) overnight at 4°C, and the treated protein was then incubated with GST-Tax as described above. Under these conditions, GST-IKK␥ was converted completely to IKK␥ by thrombin, and the residual thrombin in the reaction mixture did not interfere significantly with the binding reactions.
In Vitro I-B Kinase Assay-To measure IKK activity, 10 7 MT-4 cells were harvested by centrifugation and extracted in 1 ml of a lysis buffer containing 20 mM Tris-HCl (pH 7.6), 20 mM glycerol phosphate, 250 mM NaCl, 3 mM EGTA, 3 mM EDTA, 0.5% Nonidet P-40, 0.1 mM sodium vanadate, 10 g/ml aprotinin, 2 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml of leupeptin. After a brief centrifugation, extracts were incubated for 2 h with agitation at 4°C with 0.5 g of an antibody against IKK␥ followed by protein G-Sepharose (Invitrogen) precipitation. Each IKK immune complex reaction was carried out using 1 g of recombinant GST protein fused to amino acid residues 1-54 of IB␣ (GST-IB␣-(1-54)) as the substrate and 10 M [␥-32 P]ATP in a final volume of 30 l of a reaction buffer containing 30 mM HEPES (pH 7.4), 10 mM MgCl 2 , and 1 mM dithiothreitol at 25°C for 30 min. The reactions were terminated by the addition of SDS-PAGE loading buffer. IKK activity as measured by the phosphorylation level of IB␣ was evaluated by SDS-PAGE followed by autoradiography. When applicable, 5 ng of PP2A and 0.1-0.6 g of Tax were added into a 30-l reaction.

RESULTS
IKK␥ Interacts with PP2A and Tax in Vivo-Several cellular kinases and regulatory proteins whose activities are regulated by serine/threonine phosphorylation form protein complexes with PP2A (29 -34). In this manner, PP2A is localized to its targets at once and can affect rapid on/off control of regulatory processes immediately after signaling events are initiated. Because Tax interacts with both PP2A and IKK␥, we wondered if PP2A also forms a complex with IKK. This might explain the rapidity with which the IKK/NF-B pathway is returned to the inactive state soon after its activation. Furthermore, because Tax constitutively activates IKK, we wondered if, in the context of the IKK-PP2A-Tax complex, the phosphatase activity of PP2A may be inhibited by Tax. Because IKK is activated through serine phosphorylation of its activation loop by upstream kinases, a block in its de-phosphorylation by Tax would maintain phospho-IKK in an active state and can explain the constitutive activation of IKK by Tax.
To determine whether IKK, PP2A, and Tax form a ternary complex in vivo, cell extracts were prepared from a Tax-expressing HTLV-I-transformed human T-cell line, MT-4, and a control human T-cell line, Jurkat, and subjected to immunoprecipitation using antibodies against Tax, the catalytic subunit of PP2A (PP2Ac), and IKK␥, respectively. As expected, mouse monoclonal antibody against IKK␥ precipitated IKK␥ easily (Fig. 1, lanes 3 and 4, upper panel). From Tax-positive MT-4 cell extracts, both the PP2A c-subunit (PP2Ac) and Tax were readily co-precipitated (lane 4, middle and lower panels, respectively). In Jurkat extracts, a low but detectable trace of PP2Ac was also present (lane 3, middle panel). Likewise, Tax monoclonal antibody 4C5 co-precipitated Tax, PP2Ac, and IKK␥ (lane 6, lower, middle, and upper panels). The 4C5 antibody is highly specific for Tax. It did not react with either PP2Ac or IKK␥, as indicated by immunoprecipitations done with Tax-null Jurkat extracts (lane 5). Finally, a mouse monoclonal antibody against the COOH-terminal region of PP2Ac co-precipitated IKK␥ from both Jurkat and MT-4 extracts. Interestingly and in agreement with the immunoprecipitation performed with anti-IKK␥ (lanes 3 and 4), the amount of IKK␥ that co-precipitated with PP2Ac is lower when Jurkat extracts were used (compare lanes 7 and 8, upper panel). The PP2Ac antibody failed to co-immunoprecipitate Tax, however (lanes 7 and 8, bottom panel). We think this is due to a disruption of the protein-protein interaction between Tax and PP2Ac by the PP2Ac antibody. The COOH-terminal region of PP2Ac is involved in extensive protein-protein interaction; the carboxyl leucine residue undergoes a methyl esterification (35-37) critical for C subunit assembly with A and B subunits (36,38,39) to form the core and holoenzymes. It is likely that this region is also important for binding Tax. Together, these data indicate that PP2A interacts with IKK directly. Furthermore, Tax interacts with IKK␥ and PP2A in vivo in a stable ternary complex.
IKK␥, Tax, and PP2A Form a Stable Multiprotein Complex in Vitro-To characterize further the interaction among Tax, IKK␥, and PP2A, GST pull-down experiments were performed using purified PP2A, consisting principally of both the catalytic C-subunit (PP2Ac) and the regulatory A-subunit, derived from human red blood cells (Upstate Biotechnology) in the presence or absence of Tax protein purified from an Escherichia coli expression system. Care was taken to ensure comparable levels of purified GST, GST-Tax, and GST-IKK␥ fusion proteins ( Fig.  2A) were used. In binding reactions containing PP2A only, GST-Tax, GST-IKK␥, and GST-IKK␥⌬C, the GST fusion of an IKK␥ mutant deleted for the COOH-terminal Tax binding domain (amino acid residues 307-419), all displayed efficient binding to PP2Ac (Fig. 2B, lower panel). Under these experimental conditions, ϳ2 and 4% of input PP2Ac became bound to GST-Tax and GST-IKK␥, respectively. The interactions appear specific since under the same binding condition, GST control did not bind PP2Ac. Interestingly, in reactions where both PP2A and Tax were added, the binding of PP2A to GST-IKK␥ became significantly enhanced (compare Fig. 2, B and C, GST-IKK␥ lanes, lower panel). By contrast, Tax did not increase PP2A binding to GST-IKK␥⌬C (Fig. 2C, compare the GST-IKK␥ and GST-IKK␥⌬C lanes in the lower panel) whose Tax binding site is deleted and is unable to bind Tax (upper panel), suggesting that when Tax is bound to IKK␥, it can further stabilize or enhance IKK␥-PP2A interaction. As anticipated, GST-IKK␥ interacted with Tax (Fig. 2C, GST-IKK␥ lane, upper panel). In agreement with earlier results, GST-Tax interacted with PP2A (GST-Tax lane, lower panel). Because Tax exists as homodimer, GST-Tax and Tax interaction was also seen (GST-Tax lane, upper panel). The exogenously added Tax (300 ng) did not significantly compete for PP2Ac binding to GST-Tax (500 ng), most likely because the interaction between Tax and PP2Ac was not saturating (only 2% input PP2Ac was pulled down by GST-Tax). No binding of Tax or PP2A to GST was detected. Together, these data support the immunoprecipitation results shown above and indicate that PP2A is associated with IKK. Furthermore, in the presence of Tax, a stable IKK⅐PP2A⅐Tax ternary complex can form. Finally, even though IKK␥⌬C mutant is defective in binding Tax (18,40), it is still able to associate with PP2A, much like the full-length IKK␥, suggesting that PP2A binds to a region in IKK␥ upstream of the COOH-terminal Tax binding site.
Tax-PP2A Interaction Is Necessary for NF-B Activation-If indeed the inhibition of PP2A by Tax in the context of the IKK⅐PP2A⅐Tax complex leads to a constitutive activation of IKK, then one might expect that both Tax-PP2A and Tax-IKK␥ interactions would be necessary for Tax-mediated activation of IKK and NF-B. Using yeast 2-hybrid analysis, we have previously mapped the domain critical for PP2Ac binding to the FIG. 3. Tax binding to both PP2A and IKK␥ is required for NF-B activation. A, domain organization of Tax. Arrows mark major trypsin sensitive sites of Tax at Lys-88 and Lys-346, or Arg-349. H41Q, H43Q, and K85N denote three mutants attenuated or abrogated for PP2A binding. M22 is defective in dimer formation, IKK␥ binding, and NF-B activation. M47 is defective in P/CAF binding and LTR transactivation. B, transactivation properties of Tax mutants. DNA transfection and luciferase assays were as previously reported (17). Typically, 0.5 g of CMV-Tax (or mutants) plasmid and 0.5 g of the reporter were used to transfect 3 ϫ 10 5 human embryonic kidney 293 cells in each well of a 12-well plate. DNA transfections were carried out in triplicate. The means of luminescence units from each set of three luciferase assays were then used to compute fold trans-activation of each Tax  NH 2 -terminal 100-amino acid residues of Tax (Fig. 3A). 2 This region is distinct from the domain responsible for IKK␥ binding previously localized to approximately amino acid residues 100 -150 previously (Fig. 3A) (40). Two Tax mutants, H43Q and K85N, shown to be defective in binding PP2A earlier, 2 and another mutant, H41Q (41), were examined for their interactions with PP2A and IKK␥. As reported previously and confirmed here, H43Q together with H41Q transactivate HTLV-I LTR reporter at a level comparable with that of the wild-type Tax but are considerably attenuated in activating the E-selectin enhancer/promoter driven by two NF-B binding motifs (Fig. 3B). K85N is a much more severe mutation that abrogated both LTR and NF-B activation of Tax (Fig. 3B). As shown in Fig. 3C, all three mutants are indeed defective in PP2A binding. Notably, H41Q and H43Q bound IKK␥ efficiently. Under the same conditions, K85N also showed reduced but detectable interaction with IKK␥. These results suggest that for IKK activation, Tax-IKK␥ interaction is necessary but not sufficient. Most likely, the binding and inhibition of PP2A by Tax is also needed. We also examined two other Tax mutants, M22 (T130A,L131S) and M47 (L319R,L320S), which have been used widely in HTLV-I Tax studies (42). M22 is a mutant impaired in dimerization (43) and IKK␥ binding (14,16,18). It is unable to activate NF-B but can transactivate HTLV-I LTR (Fig. 3B). By contrast, M47 is defective in LTR activation (because of a defect in P/CAF binding) (42,44) but shows considerable activity in promoting NF-B activation (Fig. 3B). Consistent with the notion that Tax interacts with both IKK and PP2A to activate IKK and NF-B, M22 showed little binding to either IKK or PP2A; by contrast, M47, like wild-type Tax, interacted with both IKK and PP2A (Fig. 3D). Together, these data support the functional importance of PP2A binding/inhibition by Tax in activating IKK and NF-B. Finally, although we think distinct domains of Tax interact with PP2A and IKK␥, M22 and K85N mutants appeared to be impaired in both IKK␥ binding and PP2A binding. This may be due to conformational alterations of Tax as a result of the amino acid substitutions. Indeed, K85N also failed to trans-activate HTLV-I LTR (Fig. 4B), and M22 has been shown previously to be impaired in dimerization (43).
Tax Prevents I-B Kinase Down-regulation by PP2A-To test the idea that Tax inhibits IKK-associated PP2A and by so doing maintains IKK in an active state, we performed in vitro IKK assays using IKK immunoprecipitated from MT-4 cell extracts and a GST fusion containing amino acid residues 1-54 of IB␣, GST-IB␣-(1-54), as a substrate (8). The form of IKK derived from MT-4 cells is activated (by Tax). As expected, it phosphorylated GST-IB␣-(1-54) efficiently but not the AA mutant in which the Ser-32 and Ser-36 of IB␣-(1-54) were substituted by alanine residues (compare Fig. 4A, lanes 6 and 7). Consistent with the notion that IKK activity is negatively regulated by PP2A, phosphorylation of GST-IB␣-(1-54) in vitro became significantly reduced in the presence of exogenously added PP2A (compare Fig. 4A, lanes 2 and 3). The PP2A inhibitor, okadaic acid (OA), effectively blocked this effect of PP2A (Fig.  4A, lane 5). Likewise, Tax also dampened the decrease in IKK activity caused by PP2A (Fig. 4A, lane 4) in a dose-dependent manner (Fig. 4A, lanes 8 -10). To show that IKK is inactivated by PP2A directly, we treated the IKK immunoprecipitate with PP2A in the presence or absence of OA for 1 h. The PP2Apretreated IKK was then incubated with the reaction substrates GST-IB␣-(1-54) and [␥-32 P]ATP for 30 min. As expected, IKK pre-incubated with PP2A lost its kinase activity, as reflected by a significant reduction in 32 P incorporation into  5 and 8 -10). Lanes 4,8,9, and 10 contained 0.1, 0.6, 0.3, and 0.1 g of Tax, respectively. Lane 5 contained OA at a concentration of 10 nM. The reactions were carried out at 25°C for 30 min and terminated by the addition of 6 l of 6 ϫ SDS gel-loading buffer, and the phosphorylation level of IB␣ was evaluated by SDS, 12% PAGE followed by autoradiography. B, the IKK kinase reactions were carried out as in A except that the IKK immunoprecipitates were incubated with PP2A for 1 h at 25°C for 1 h (lanes 1-7) with or without OA, GST-Tax, or GST-H43Q. In lane 2, OA was added together with the substrates after the PP2A preincubation. In lane 8, no PP2A was added. One of three repeats of the experiment is shown. IP, immunoprecipitates.
GST-IB␣-(1-54) (Fig. 4B, lanes 1, 2, and 4). IKK inactivated by PP2A pretreatment lost its activity irreversibly; the addition of OA or Tax after the PP2A preincubation could not reverse the inhibitory effect of PP2A (Fig. 4B, compare lanes 1 and 2). By contrast, in the reaction where PP2A is inhibited by the simultaneous addition of OA (Fig. 4B, lane 3) or GST-Tax (Fig.  4B, lanes 6) during the pretreatment, IKK retained significant activity. The addition of PP2A binding-defective GST-H43Q, however, could not effectively prevent IKK inactivation by PP2A (Fig. 4B, lane 7). Finally, the addition of GST-Tax in the absence of PP2A did not affect IKK kinase activity significantly (Fig. 4B, lane 8), consistent with the notion that Tax does not activate IKK directly. In aggregates, these data support the idea that IKK activation by Tax is most likely mediated by its binding and inhibition of PP2A. DISCUSSION In this study, we have demonstrated that the IKK␥ regulatory subunit, PP2A, and Tax form a stable multiprotein complex in HTLV-I-transformed MT-4 cells and in vitro. Furthermore, we show that Tax mutants defective in PP2A binding failed to activate NF-B despite wild-type levels of binding to IKK, indicating that interaction with both IKK␥ and PP2A is necessary for Tax to activate IKK-NF-B. Finally, evidence is provided to indicate that Tax prevents the active form of IKK isolated from MT-4 cells from inactivation by PP2A in vitro. Together, these results provide a mechanistic explanation for the constitutive activation of IKK by Tax; IKK-associated PP2A is responsible for rapid deactivation of IKK; in the context of the IKK⅐PP2A⅐Tax complex, PP2A is inhibited by Tax, causing constitutive phosphorylation/activation of IKK; the activated IKK, in turn, promotes phosphorylation and degradation of I-B, and potent NF-B activation.
Although interaction between IKK and PP2A can be detected in Tax-null Jurkat cells, their interaction appears weaker than that seen in Tax-positive MT-4 cells (Fig. 1). This is most likely due to the tripartite IKK-PP2A-Tax interaction and is supported by the increased stability of IKK and PP2A interaction in vitro when Tax is present (Fig. 2). Whether PP2A may also be recruited specifically to the activated form of IKK is not clear. Although our results indicate that Tax can prevent active IKK from dephosphorylation and inactivation by PP2A, they do not rule out the possibility that Tax may also prevent phospho-I-B␣ from dephosphorylation by PP2A. Further experiments are needed to determine whether the latter mechanism contributes to the potent NF-B activation brought on by Tax.
Two Tax mutants, H41Q and H43Q, are of particular interest. As previously reported by Semmes and Jeang (41) and confirmed here, H41Q and H43Q transactivate HTLV-I LTR reporter at a level comparable with that of the wild-type Tax but is considerably impaired in activating the E-selectin enhancer/promoter, whose activity is driven by two NF-B binding motifs, suggesting that these two amino acid substitutions impact on a very specific protein-protein interaction critical for NF-B activation. Our data indicate that the specific defect of H43Q and H41Q lies in an inability to interact with and, most likely, a failure to inhibit PP2A. We have also examined two other Tax mutants, M22 and M47. M22, a mutant that fails to activate NF-B but can transactivate via HTLV-I LTR, is defective in both IKK and PP2A binding. As expected, M47, a mutant that cannot transactivate LTR because of a defect in P/CAF binding but continues to promote NF-B activation binds both IKK and PP2A. These data add to the IKK interaction with M47 and the lack thereof with M22 previously reported and support the notion that for NF-B activation, Tax needs to interact with both IKK and PP2A.
The functional diversity of PP2A is mediated by the regula-tory subunits that control its subcellular localization and substrate specificity. Our data indicate that IKK␥ binds both PP2A and Tax. Yeast 2-hybrid analysis had shown that Tax and PP2A C-subunit interact directly. 2 Whether other PP2A or IKK subunits also participate in the tripartite interaction remains to be determined. Finally, IKK␥⌬C, which is deleted for the COOH-terminal Tax binding region, interact with PP2A like full-length IKK␥, indicating that Tax and PP2A binds to distinct domains of IKK␥.
Our recent data indicate that Tax binds the catalytic subunit of PP2A and acts as a noncompetitive inhibitor of PP2A. In reactions where 32 P-labeled glycogen phosphorylase a is used as a substrate, the k i (inhibitory constant) of Tax for PP2A is estimated to be ϳ300 nM. This contrasts with the most common inhibitor of PP2A, okadaic acid, a compound isolated from marine plankton, which has an IC 50 of 1 nM or lower for PP2A. How can the weaker interaction/inhibition of PP2A by Tax explain the potent activation of IKK and NF-B? We think the interaction between Tax and IKK is particularly relevant. Even though Tax is a weaker inhibitor for PP2A, the interaction between Tax and IKK would locate Tax in the immediate environment where its interaction with PP2A would be the most effective. In essence, through Tax-IKK interaction, the effective concentration of Tax around IKK is high, thus allowing strong inhibition of PP2A to occur. Indeed, we think Tax does not target all forms of PP2A. Rather, via additional protein-protein interactions such as binding to IKK, it interacts and inhibits sub-populations of PP2A involved in regulating specific signaling pathways. The fact that ϳ2-4% of PP2Ac co-immunoprecipitated with Tax 2 is consistent with this notion. Finally, it should be pointed out that IKK down-regulation involves not only phosphatases such as PP2A. The highly active IKK phosphorylated in the activation loop undergoes autophosphorylation at multiple sites in the COOH termini (9), which down-modulates its activity. Although it is not clear which form(s) of IKK and which phosphoamino acid residue(s) is the target of PP2A, based on the okadaic acid results and Tax-related studies it may be inferred that the phosphoserine residues in the activation loop are most likely the targets. In summary, the data reported here provide a novel, mechanistic explanation for the potent activation of IKK/NF-B-signaling pathway by Tax.