HTLV-1 Tax Is a Critical Lipid Raft Modulator That Hijacks IκB Kinases to the Microdomains for Persistent Activation of NF-κB*

Upon T cell activation, IκB kinases (IKKs) are transiently recruited to the plasma membrane-associated lipid raft microdomains for activation of NF-κB in promoting T cell proliferation. Retroviral Tax proteins from human T cell leukemia virus type 1 and type 2 (HTLV-1 and -2) are capable of activating IKK, yet only HTLV-1 infection causes T cell leukemia, which correlates with persistent activation of NF-κB induced by Tax1. Here, we show that the Tax proteins exhibit differential modes of IKK activation. The subunits of IKK are constitutively present in lipid rafts in activated forms in HTLV-1-infected T cells that express Tax. Disruption of lipid rafts impairs IκB kinase activation by Tax1. We also show that the cytoplasmic Tax1 protein persistently resides in the Golgi-associated lipid raft microdomains. Tax1 directs lipid raft translocation of IKK through selective interaction with IKKγ and accordingly, depletion of IKKγ impairs Tax1-directed lipid raft recruitment of IKKα and IKKβ. In contrast, Tax2 activates NF-κB in a manner independent of lipid raft recruitment of IKK. These findings indicate that Tax1 actively recruits IKK to the lipid raft microdomains for persistent activation of NF-κB, thereby contributing to HTLV-1 oncogenesis.

Human T cell leukemia virus type 1 (HTLV-1) 3 is an oncogenic retrovirus that causes human adult T cell leukemia. The viral oncoprotein Tax1, encoded by the HTLV-1 genome, is the molecular determinant for transformation of T lymphocytes (1). Tax1 modulates the activity of the transcriptional factor NF-B to promote T cell proliferation, and it also inactivates tumor suppressor p53 (2,3). These molecular bases are thought to be crucial for Tax1-induced transformation of T lymphocytes. Moreover, vast amounts of experimental data have shown that Tax1 regulates a wide variety of cellular factors in addition to NF-B and p53, which may play distinct roles at various stages of oncogenesis (4). Persistent activation of NF-B is one of the characteristic features of adult T cell leukemia. Tax1 is essential to establish proliferative growth of T cells at least in the early stage of infection, and Tax1 immortalizes primary human T cells in a manner highly dependent on its ability to activate NF-B (2). Tax2, the Tax protein from HTLV-2, shares a significant amino acid sequence homology with Tax1 and is able to activate NF-B and to transactivate the long terminal repeat of both HTLV-1 and HTLV-2 by enhancing the activity of CREB (5). Although the viral genome can be isolated from some hairy cell leukemia patients, HTLV-2 infection is not causally linked to T cell leukemia (5). Furthermore, unlike Tax1, Tax2 rarely transforms primary cells although it can immortalize primary human T cells. The underlying mechanism for the differential effect of Tax1 and Tax2 in transformation of T cell remains poorly understood.
The precise mechanism of Tax activation of NF-B remains elusive. It has been demonstrated that Tax1 induces activation of NF-B via several distinct mechanisms. First, Tax1 modulates the IB kinase complex, the key regulator of NF-B signaling, by directly stimulating the activity of the catalytic subunits, IKK␣ and IKK␤, through interaction with the non-catalytic subunit, IKK␥ or NEMO (6 -11). Indeed, Tax1 fails to activate IB kinases in NEMO/IKK␥-deficient cells (12), indicating that IKK␥ is one of the key targets for Tax1-mediated activation of IKK. Second, Tax1 was shown to modulate the activity of upstream kinases of IKK, which include mitogen-activated protein kinase/extracellular signal-regulated kinase kinase 1 (MEKK1) and NF-B-inducing kinase (13,14). It has been recently shown that Tax1 stimulates the activity of transforming growth factor ␤-activated kinase 1 (Tak1), an upstream kinase of IKK, and promotes interaction of Tak1 with IKK␥ (15). Depletion of Tak1 abrogates the activation of IKK induced by Tax1, suggesting a critical role of Tak1 in this activation process. Furthermore, Tax1 was shown to interfere with the noncanonical pathway of NF-B signaling by inducing processing of NF-B2/p100 through up-regulation of IKK␣ in T cells (16). These evidences implicate that Tax1-mediated activation of NF-B involves a complex process. The mechanism of the persistent activity of NF-B by Tax1 is yet to be addressed.
In T lymphocytes, activation of TCR elicits tyrosine phosphorylation cascades and induces membrane lipid raft recruitment and activation of signaling molecules such as ZAP70, phosphatidylinositol 3-kinase, and PKC (17,18). The lipid rafts are glycosphingolipid-and cholesterol-enriched, detergent-resistant microdomains that play crucial roles in signaling transduction (19 -20). The lipid rafts are assembled in the Golgi and can be recycled between the plasma membrane and Golgi (21). Carma1, PKC, and Bcl10 are the key players that direct the plasma membrane lipid raft recruitment of the IKK complex, leading to transient activation of IKK and nuclear translocation of NF-B upon T cell activation (22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33). Aside from TCR-directed signaling events, tumor necrosis factor ␣ (TNF␣) also induces lipid raft translocation of IKK, together with the IKK-associated chaperone protein, Hsp90 (34). It remains largely unclear whether Tax1-mediated activation of NF-B is involved in lipid raft translocation of IB kinases. It has been shown that cytoplasmic Tax1 mediates activation of IB kinases. Indeed, a portion of Tax1 was found to reside in the Golgi and to recruit the IKK complex to this subcellular location (35)(36)(37). In the present study, we show that the Tax1 protein accumulates in the Golgi-associated lipid rafts in directing translocation of IKK to the microdomains by primarily targeting IKK␥. This process is crucial for Tax1 activation of NF-B in both T cells and non-lymphoid cells. In contrast, Tax2 activates NF-B in a manner independent of the lipid raft recruitment of IKK as seen in Tax2-immortalized T cells. These differential modes of IKK activation by the Tax proteins may have implications on the pathogenesis of T cell leukemia.

EXPERIMENTAL PROCEDURES
Cell Lines, Antibodies, and Reagents-Human T cell lines including MT1, MT2, MT4, SupT1, and Jurkat were cultured in RMPI1640 medium supplemented with 10% fetal bovine serum plus antibiotics at 37°C, 5% CO 2 . SupT1 and Jurkat E6-1 cell lines were obtained from ATCC (Manassas, VA), MT1 and MT4 were kindly provided by Drs. Atsushi Koito and Takeo Ohsugi (Center for AIDS Research and Institute of Resource Development and Analysis, Kumamoto University, Japan). MT2 cells were obtained from Dr. Douglas Richman (AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, National Institutes of Health). Antibodies for IB␣, Tak1, Hsp90, and HA epitope were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-IKK␣, IKK␤, IKK␥, and serine-phosphorylated IB␣ were purchased from IMGENEX (San Diego, CA). Anti-LAT was from Upstate Biotechnology (Charlottesville, VA). Anti-␤-actin, anti-FLAG M2 monoclonal antibodies, horseradish peroxidase-conjugated cholera toxin ␤ subunit, methyl-␤cyclodextrin (M␤CD), protease, and phosphatase inhibitor mixtures were obtained from Sigma. The proteasome inhibitor MG-132 was purchased from Calbiochem, and anti-Tax1 antibody was acquired from the AIDS Research and Reference Reagent Program.
Mammalian Expression Plasmids, DNA Transfection, and GST Pulldown-The expression plasmids for GST-tagged IKK␣, IKK␤, and IKK␥ as well as HA-tagged Tax1 were described previously (38). To generate Tax1 mutants, PCRbased site-directed mutagenesis was performed to construct the fragments with mutations at the positions illustrated in Fig. 3A. The fragments of Tax1 mutants were inserted into expression vector pCEF with a C-terminal HA or GST tag and verified with DNA sequencing. Tax2 was constructed in the same vector as Tax1. To construct lipid raft-targeted IKK␥, a myristoylation signal from Lck was attached to the full-length of IKK␥ to generate the myristoylated IKK␥ fusion fragment (Myr-IKK␥), which was inserted into the lentivirus vector. GFP-GalT (galactosyltransferase) was purchased from Addgene Inc. (MA). To construct RFP-tagged caveolin-1, the full-length of the caveolin-1 (CAV1) cDNA was amplified from a human cDNA library, and C-terminal tagged with the RFP (monomeric red fluorescence protein) fragment. The CAV1-RFP fusion fragment was inserted into vector pLCEF8 and CAV1 was sequence-validated.
To examine the interaction of IKK␥ with Tax1 and its mutants, the FLAG-IKK␥ expression plasmid was co-transfected with GST-tagged wild type Tax1, the Tax1 mutants, or IKK␤ into HEK cells using SuperFect transfection reagent (Qiagen, Alencia, CA). 24 h post-transfection, the cells were lysed in the Buffer A containing 1% Triton X-100, 40 mM Tris-Cl (pH 7.5), 150 mM NaCl, 2 mM MgCl 2 , 0.5 mM dithiothreitol and protease inhibitor mixture at 15°C for 30 min. Glutathione-Sepharose beads were added into the soluble supernatants and incubation was at room temperature for 2 h. The beads were then washed three times with the lysis buffer and subjected for SDS-PAGE plus Western blot analysis using anti-FLAG M2 monoclonal antibody to detect FLAG-IKK␥. To evaluate the interaction of endogenously expressed Tak1 with IB kinases, GST-tagged IKK␣, IKK␤, or IKK␥ was transfected into HEK cells in the presence of either the empty vector or Tax1-HA. GST pulldown assay and immunoblot analysis were performed as described above.
Lentivirus Vector and Transduction-The full-length fragment of enhanced green fluorescence protein was fused with IKK␤ and IKK␥ to generate GFP-IKK␤ and GFP-IKK␥ fusion genes, which were constructed in the lentivirus vector pLCEF8, a modified vector of pLL3.7 (39) in which human elongation factor 1␣ promoter replaced U6 promoter. The Tax1-GFP fusion fragment was also constructed in pLCEF8. The lentivirus production and transduction in T cell lines were performed as described previously (38), and ϳ10 multiplicity of infection was used for transduction of T cells. Transduction efficiency was verified with fluorescence imaging and immunoblot.
Generation of Tax2 Immortalized T Cell Lines-Primary CD4ϩ T lymphocytes from healthy donors were isolated with Dynal beads conjugated with anti-CD4 antibody (Invitrogen). The CD4ϩ cells were activated with phytohemagglutinin (1 g/ml) and recombinant interleukin-2 (100 units/ml) for 5 to 7 days prior to transduction with the lentivirus expressing Tax2-GFP fusion protein. Following one month of in vitro culture of the transduced cells in the media supplemented with interleukin-2 (50 units/ml), virtually all of the viable cells were green fluorescence positive and the cells were maintained in culture for more than 6 months.
Western Blot Analysis-Cells were collected and lysed in lysis buffer B containing 40 mM Tris-Cl (pH 7.6), 1% Triton X-100, 1% deoxycholate, 150 mM NaCl plus protease and phosphatase inhibitor mixtures at 4°C for 30 min. Equal amounts of cellular proteins were analyzed by SDS-PAGE, followed by immunoblot. Anti-␤-actin blot was used for the protein loading control.
In Vitro Kinase Assay and NF-B Reporter Assay-In vitro kinase assay to detect the activity of IB kinases and the NF-B reporter assay were performed as previously reported (38).
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts were prepared from various T cell lines with or without TNF␣ stimulation (10 ng/ml for 10 min at 37°C), using NE-PER nuclear and cytoplasmic extraction reagents (Pierce). The sequence of the oligonucleotide corresponding to the B element from the interleukin-2R␣ gene was 5Ј-gatcCGGCAGGG-GAATCTCCCTCTC-3Ј. The underlined sequence is the B cis-element. The oligonucleotide was 5Ј-end labeled with biotin (Integrated DNA Technologies, Coralville, IA) and annealed to its complementary strand. The NF-B binding activity to the B element was examined by EMSA using the LightShift Chemiluminescent EMSA Kit (Pierce). In brief, 5 g of the nuclear extracts were preincubated in a 20-l total volume containing 100 mM Tris (pH 7.5), 500 mM KCl, 10 mM dithiothreitol, 200 mM EDTA (pH 8.0), 50% glycerol, and 1 g of polydeoxyinosinic-deoxycytidylic acid and 2 l of biotin-labeled probe (20 fmol) for 20 min at room temperature. The reactions were mixed with 5 l of 5ϫ loading buffer and run on a 6% non-denaturing polyacrylamide gel in 0.5ϫ TBE (1ϫ TBE: 89 mM Tris borate, 2 mM EDTA, pH 8.3) for 90 min at 100 V on ice, and transferred to nylon membranes (Amersham Biosciences) at 380 mA for 1 h in 0.5ϫ TBE on ice. The membrane was optimally UV light cross-linked. Biotin-labeled DNA was detected by streptavidin-horseradish peroxidase, and followed by chemiluminescence.
Lipid Raft Fractionation by OptiPrep Density Gradient Ultracentrifugation-Cells (4 ϫ 10 7 for T cells and 1 ϫ 10 7 cells for HEK cells) were lysed in 2 ml of extraction buffer (20 mM Tris-Cl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100 plus protease inhibitor mixture). Lysates were sheared by 20 passages through a 22-gauge needle, incubated for 20 min on ice before mixing with the OptiPrep density gradient medium (Iodixanol solution, final concentration, 40% v/v; AXIS-SHIELD PoC AS, Oslo, Norway), and placed at the bottom of a 12-ml tube. By overlaying 4 ml of 30% and 4 ml of 5% of Opti-Prep medium, a discontinuous OptiPrep gradient was formed. Ultracentrifugation was performed at 100,000 ϫ g for 4 h at 4°C in an SW41 rotor. 1 ml of each fraction from the top to bottom was collected and subjected to Western blot analysis. Depletion of plasma and intracellular membrane cholesterol by M␤CD in T cells was performed by pretreatment of the cells (4 ϫ 10 7 cells/each sample) with 10 mM M␤CD for 45 min at 37°C in Hanks' balanced salt solution. Following this step, the cells were subjected to density gradient ultracentrifugation for lipid raft fractionation analysis.
Fluorescence Imaging-MT2 cells were transduced with GFP, Tax1-GFP, GFP-IKK␤, or GFP-IKK␥ using the lentivirus-mediated gene delivery system. Lipid raft was labeled with the Alexa Fluor 594 Lipid Raft Labeling Kit (Molecular Probes, Eugene, OR). Cells were centrifuged and gently resuspended in chilled, complete culture media. Following centrifugation, the cell pellets were gently resuspended in 500 l of Alexa 594conjugated cholera toxin B (6 g/ml) at 4°C for 20 min. After this incubation, the cells were washed twice with chilled 1ϫ phosphate-buffered saline and resuspended in 500 l of the chilled anti-cholera toxin B antibody (100-fold dilution working solution) for 15 min at 4°C. The cells were washed twice with chilled 1ϫ phosphate-buffered saline, and transferred to poly-L-lysine-coated dishes and fixed in 4% paraformaldehyde for 15 min. The Golgi-lipid raft association of Tax1 was assessed by co-transfection of Tax1-GFP with the Golgi lipid raft marker caveolin-1 tagged with RFP. The cells were then incubated with 300 nM 4Ј,6-diamidino-2-phenylindole (Sigma) to stain the nuclei. Microscopy was performed using a Leica TCS SP2 AOBS confocal microscope.
Alkaline Phosphatase Activity Assay-p-Nitrophenyl phosphate phosphatase assay was carried out for 12 h at 37°C in an assay mixture (100 l) consisting of 100 mM Tris-Cl (pH 9.5), 100 mM NaCl, 5 mM MgCl 2 , 1 mg/ml p-nitrophenyl phosphate (Sigma). The reaction was started by adding 10 l of each fraction, and terminated by adding 100 l of 1 N NaOH. Alkaline phosphatase activity was measured by absorbance at 405 nm.

IB Kinases Are Constitutively Present in the Lipid Rafts in HTLV-1-infected T Cells That Express Tax-Five T cell lines, including non-HTLV-1-infected T cells (Jurkat and SupT1)
and HTLV-1-infected T cells (MT1, MT2, and MT4) were assessed for Tax1 expression. MT2 cells, but not other T cell lines, produced a detectable level of Tax1 (Fig. 1A). A hyperphosphorylation of IB␣, a substrate for activated IB kinase, was detected in MT2 cells when the cells were pretreated with the proteasome inhibitor MG-132 (Fig. 1B). Accordingly, MT2 cells exhibited a hyperactivity of the B DNA binding, and TNF␣ stimulation did not further increase the B binding activity (Fig. 1C), suggesting that NF-B was maximally activated in these cells even in the absence of extracellular stimuli. IB␣ was weakly phosphorylated in MT1 cells with a detectable basal activity of NF-B, and this activity was further enhanced upon TNF␣ treatment (Fig. 1, B and C). In non-HTLV-1 infected, Jurkat T cells, the phosphorylation of IB␣ was not detected and the basal activity of NF-B activity was not observed, whereas the NF-B activity was induced by TNF␣ (Fig. 1, B and C). These results implicate that a constitutive NF-B activation is present in HTLV-1-infected T cells, and the expression of Tax1 leads to induction of full-scale activation of IB kinases.
To determine whether the activation of IB kinase by Tax1 requires a process that involves lipid raft translocation of IKK, density gradient ultracentrifugation was applied for lipid raft fractionation analysis. As shown in Fig. 1D, portions of three subunits of the IKK complex, including IKK␣, IKK␤, and IKK␥, were constantly present in the lipid raft fractions in MT2 cells, corresponding to lipid raft biomarkers LAT and GM1 (fractions 4 and 5), whereas ERK1 exclusively resided in the soluble fractions. In contrast, in the HTLV-1-infected, non-Tax1 expressing cell line MT1, IKKs constantly remained in the cytoplasmic, soluble fractions ( Fig. 1E). We further tested another non-Tax1 expressing, HTLV-1-infected T cell line, TL-Om1, and found that IB kinases were in the soluble fractions (data not shown). To verify if the lipid raft presence of the IKK complex correlates with Tax1 expression, we examined another HTLV-1-infected T cell line, SLB-1, that expresses Tax1 (40). The activity of NF-B in SLB-1 cells was comparable with that in MT2 cells (Fig. 1F). Indeed, the IB kinases, not ERK1, were found in the lipid raft fractions in SLB-1 cells, corresponding to GM1 fractions (Fig. 1G). The amounts of the IB kinase proteins in lipid rafts were abundant, which were detected in both fractions 4 and 5.
To verify whether the IB kinases are associated with lipid rafts in HTLV-1-infected, Tax1-expressing T cells, we treated MT2 cells with M␤CD (dissolved in water), a selective cholesterol inhibitor that impairs formation of lipid rafts. Both IKK␤ and IKK␥ were exclusively found in the soluble, non-lipid raft fractions, whereas the level of IKK␣ in lipid rafts was significantly reduced following M␤CD treatment ( Fig. 2A). In MT2 cells, IKK␣ was persistently phosphorylated in both the lipid raft fraction (highest intensity of the IKK␣ phosphorylation in the fraction 5) and non-lipid raft fractions (Fig. 2B). Under the identical immunoblot conditions, phosphorylation of IKK␣ was rarely detected in MT1 cells, primarily due to low affinity of the phospho-spe-cific antibody, which was unable to detect weak phosphorylation at short exposure. These data indicate that Tax1 induces hyperphosphorylation and activation of IB kinases, which in turn drives a full range of activation of NF-B in T cells. Indeed, disruption of lipid rafts by M␤CD inhibited NF-B activity in MT2 cells (Fig. 2C), and prevented Jurkat cells from TNF␣induced activity of NF-B (Fig.  2D). These results suggest that the lipid raft-translocated IKKs are in activated forms in Tax1-expressing T cells and that Tax1 is likely to be the causative factor to promote lipid raft translocation of IB kinases.
Tax1 Directs Lipid Raft Translocation of IKK Correlating with Its Ability to Activate IKK-To evaluate structural recruitments of Tax1 in activating IB kinases, we constructed several mutants of Tax1. M22 contains two amino acid substitutions at amino acids 130 -131 and abrogates the ability of Tax to activate IKK but retains a full capacity to induce viral gene transcription through HTLV-1 LTR. As depicted in Fig. 3A, two additional mutants of Tax1, ⌬GG and ⌬PXXP, with deletions at amino acids 33-34 and 73-79, respectively, were generated, as these two amino acid sequences are highly conserved among the Tax proteins from HTLV-1 and HTLV-2. We assessed the ability of wild type (WT) Tax1 and Tax1 mutants in inducing the activity of IKK␤ and found that the WT Tax1, but not any of the Tax1 mutants, stimulated the kinase activity of IKK␤ (Fig. 3B). Accordingly, Tax1, but not any of the mutants induced NF-B driven ␤-galactosidase activity (Fig. 3C). Similar to the WT Tax1, M22 maintained the ability to activate HTLV-1 LTR, whereas the ⌬GG and ⌬PXXP mutants lost such ability (data not shown). Because the ⌬GG and ⌬PXXP mutants abolished the abilities in inducing the activities of NF-B and CREB that transactivates HTLV-1 LTR, it is likely that these two mutants may be improperly folded.
Tax1 activation of IB kinases does not require lymphocyte-specific factors. To determine whether the lipid raft translocation of the IKK complex correlates with the ability of Tax1 to activate IKK, we transfected WT Tax1 and its mutants into a non-lymphoid cell line, HEK. The WT Tax1 induced the lipid raft translocation of IKK␣, IKK␤, and IKK␥, corresponding to the GM1 fraction, whereas ERK1 remained in the soluble fractions (Fig. 3D, fraction 5). In contrast, all three Tax1 mutants lost the ability to drive the translocation of IB kinases to lipid rafts (Fig. 3, E-G). More- over, the IKK-associated chaperone protein Hsp90, an essential modulator of the IKK complex, and Tak1, an upstream kinase of IKK, were also able to translocate into the lipid raft fractions in Tax1-expressing HEK cells, but not in cells transfected with the Tax1 mutants (Fig. 3, H and I). In HEK cells transfected with Grb2 or in parental HEK cells, the subunits of IKK remained in the soluble fractions (data not shown). Together, these results suggest that Tax1 is the molecular determinant that directs lipid raft translocation of IKK, correlating with its ability to activate IB kinases.

FIGURE 1. Constitutive presence of IB kinases in lipid rafts in Tax1-expressing T cells. A, Tax1 expression in various HTLV-1-infected T cell lines (MT1, MT2, and MT4) and non-HTLV-1-infected T cell lines (Jurkat and
Tax1 Is a Virally Derived, Lipid Raft Modulator-To investigate the possibility that Tax1 is associated with lipid rafts, we evaluated the lipid raft presence of Tax1. As shown in Fig.  4A, a portion of the Tax1 protein was constitutively present in the lipid raft fractions in HEK cells transfected with HAtagged Tax1 (upper panel). The Tax1 mutants, M22, ⌬GG, and ⌬PXXP, were also found in this structure, whereas the control protein, HA-tagged Grb2, resided in the soluble fractions (Fig. 4A). To rule out the possibility of HA tag interference, we tested GFP-tagged Tax1 and found that Tax1-GFP was distributed predominantly in the lipid raft fraction (Fig.  4B, fraction 5). Accordingly, in Tax1-GFP transfected HEK cells, the subunits of IKK were shifted to the lipid raft fractions (Fig. 4B). Moreover, fraction 5 of the Tax1-GFP-transfected HEK cells exhibited a peak absorbance at A 600 nm , a peak of protein concentration and the alkaline phosphatase activity (Fig. 4, C-E, respectively), supporting the notion that fraction 5 is the lipid raft fraction in which Tax1 and the  IB kinases are associated. In contrast, in GFP-transfected HEK cells, the GFP protein and IB kinases resided exclusively in the soluble fractions (Fig. 4F). Furthermore, in HTLV-1-transformed MT2 T cells, Tax1 was expressed in two distinct forms, a major form of the Env-Tax fusion protein (p68) and a minor form of the full-length of Tax1 (p40) (40). Only p40 Tax1 was expressed in SLB-1 cells (40). As shown in Fig. 4G, significant portions of both p68 and p40 forms of the Tax1 proteins in MT2 and SLB-1 cells were found in the lipid raft fractions, supporting the notion that Tax1 is able to associate with lipid rafts in lymphoid and non-lymphoid cells.
To further evaluate the lipid raft presence of Tax1, we applied the confocal fluorescence imaging technique. To avoid potential nonspecific immunostaining from primary antibodies, we expressed GFP-tagged fusion proteins including Tax1-GFP, GFP-IKK␤, and GFP-IKK␥ as well as the control protein GFP into MT2 cells using the lentivirus transduction method. These cells were co-stained with cholera toxin B labeled with red fluorescence dye, which binds specifically to the sphingolipid-enriched microdomains. A yellow cast is generated when green and red fluorescence signals overlap. As shown in Fig. 5A, GFP-tagged Tax1, IKK␤, or IKK␥ was mostly expressed in the cytoplasm in MT2 cells and significant portions of these proteins overlapped with GM1, emitting a yellow cast in the merged pictures. GFP alone did not show apparent co-localization with GM1 as the merged picture emitted separate green and red fluorescence signals. These results indicate that Tax1 and IB kinases co-localize in the lipid rafts. Next, as it was reported that the cytoplasmic Tax1 is localized in the Golgi, we examined the subcellular localization of Tax1 in HEK cells. We co-transfected HEK cells with the Golgi marker protein galactosyl-transferase tagged with GFP (GFP-GalT) and the Golgi-lipid raft marker protein caveolin-1 (CAV1) tagged with RFP (CAV1-RFP). Two-color confocal imaging analysis showed that GFP-GalT and CAV1-RFP co-localized to form perinuclear clusters, a typical pattern of the Golgi (Fig. 5B, upper  panel). Furthermore, Tax1-GFP and CAV1-RFP also co-localized in the perinuclear Golgi clusters, emitting a yellow cast (Fig. 5B, lower panel). Lipid rafts are synthesized in the Golgi and can be recycled to the plasma membrane. Taken together, these results indicate that Tax1 targets the Golgilipid raft microdomains in recruiting IKK to these structures for activation.
Tax1 Directs Lipid Raft Translocation of IKK by Targeting IKK␥-Because it has been shown that Tax1 primarily interacts with IKK␥ to stimulate the kinase activity of the catalytic subunits of IKKs, we then investigated the structural requirements of Tax1 for the interaction with IKK␥. As shown in Fig. 6A, WT Tax1, but not any of the Tax1 mutants, co-precipitated with IKK␥. These results implicate that the interaction of Tax1 with IKK␥ correlates with Tax1-induced IB kinase activation and Tax1-directed lipid raft translocation of IKK. In addition, we found that Tak1 co-precipitated with IKK␥, but not with IKK␣ or IKK␤, regardless of the presence of Tax1 (Fig. 6B). Moreover, by expressing Tax1-GFP in the IKK␥-deficient Jurkat T cells, the depletion of IKK␥ apparently impaired Tax1-mediated lipid raft translocation of IKK␣ and IKK␤ (Fig. 6C). Therefore, these results validated the role of Tax1 as a viral lipid raft protein, which actively hijacks the IKK complex and Tak1 to the Golgiassociated lipid rafts through selective interaction with IKK␥.
To determine whether lipid raft targeting of IKK␥ alone is sufficient to mediate lipid raft translocation of the catalytic subunits of IKKs and to activate NF-B, we generated myristoylated IKK␥ (Myr-IKK␥). Expression of Myr-IKK␥ in HEK cells resulted in exclusive distribution of Myr-IKK␥ in the lipid raft fractions, yet the endogenous IKK␣, IKK␤, and Hsp90 remained in the soluble fractions (Fig. 6D). In addition, Myr-IKK␥ did not induce activation of NF-B in transfected HEK cells (Fig. 6E), suggesting that lipid raft targeting of IKK␥ alone is not sufficient for mediating lipid raft translocation and activating the catalytic IKKs.
Activation of NF-B by Tax2 Is Independent of the Lipid Raft Recruitment of IKK-We tested the ability of Tax1 and Tax2 in inducing the activity of NF-B in transfected HEK cells and found that both Tax proteins activated NF-B efficiently (Fig. 7A). Tax1 and Tax2 interacted with IKK␥ equally well in co-transfected HEK cells (Fig. 7B). The lipid raft fractionation analysis showed that Tax2 did not direct the lipid raft translocation of the IKK complex, although a portion of Tax2 was localized in the lipid raft fraction (Fig. 7C).
To further evaluate the differential effect of Tax1 and Tax2 in induction of the lipid raft translocation of IKK, we generated Tax2-immortalized, primary human CD4ϩ T cell lines. As shown in Fig. 7D, in Tax2-immortalized T cells, IKKs were mostly in the soluble fractions. Examination of two additional Tax2-immortalized T cells yielded similar results (data not shown). These results indicate that Tax1 and Tax2 exhibit differential modes of activation of IB kinases.
Although both Tax1 and Tax2 can interact with IKK␥, it is likely these homologous viral proteins may differentially regulate the catalytic subunits of IKKs. To investigate this possibility, we first assessed the interactions of the Tax proteins with IKK␣ and IKK␤. In consistency with the previous reports, transient co-transfection of HA-tagged Tax and GST-tagged IKKs demonstrated that Tax1 co-precipitated with either IKK␣ or IKK␤ (Fig. 7E). In contrast, at the identical conditions, Tax2 failed to co-precipitate with IKK␣ or IKK␤. Next, we generated IKK␣-specific knockdown HEK cells (Fig. 7F). Expression of Tax1 in the IKK␣-depleted HEK cells failed to activate NF-B (data not shown), and abrogated the ability of Tax1 to promote lipid raft translocation of the IKK complex, although Tax1 was still able to translocate into the lipid rafts (Fig. 7G). Thus, these results suggest that the interaction of Tax1 and IKK␥ is necessary but not sufficient to promote the translocation of IKK to lipid rafts. A preconditioning event of the catalytic subunits of IKKs such as IKK␣ by Tax1 is critical to execute the ability of Tax in hijacking IKKs to lipid rafts for persistent activation.

DISCUSSION
The present study illustrated a novel mechanism that the Tax protein derived from HTLV-1 actively recruits IKKs to the Golgi-associated lipid raft microdomains as one of the critical processes for the persistent activation of IB kinases. The ability of Tax1 to drive the IKK complex to lipid rafts in both T cells and non-lymphoid cells implicates that such a process is independent of lymphocyte-specific factors such as Carma1, PKC, and Bcl10, which are otherwise critical in TCR-directed membrane lipid raft recruitment and activation of the IB kinases. This process is dependent on the lipid raft targeting of Tax1 and the interaction of Tax1 with IKK␥, thereby promoting lipid raft translocation of IB kinases.
The Tax protein of HTLV-1 exhibits both cytoplasmic and nuclear distribution patterns as it consists of nuclear import and export signal peptides, allowing Tax1 shuttling from the nucleus to the cytoplasm (41)(42)(43). These unique distribution patterns are in accordance with two major activities: activation of IB kinases by the cytoplasmic Tax1 and transactivation of HTLV-1 LTR through induction of the activity of CREB by nuclear Tax1 (44). The percentage of Tax1 protein distributed in the cytoplasm and nucleus varies in HTLV-1infected T cell lines. Predominant cytoplasmic distribution of Tax1 was found in MT2 cells (45). In this study, we utilized several approaches to assess the subcellular distribution of Tax1 in MT2 cells and non-lymphoid cells. Our results confirmed that Tax1 is mostly distributed in the cytoplasm in MT2 cells by confocal imaging analysis. One concern about the subcellular localization of Tax1 is that an active viral replication of HTLV-1 occurs in MT2 cells, which requires the nuclear Tax1. It is likely that the nuclear Tax, even at a low level, is sufficient to promote viral transcription. Alternatively, Tax1 may be rapidly exported from the nucleus to the cytoplasm. The presence of nuclear import and export signal peptides in Tax1 supports this notion, suggesting that Tax1 has a dynamic cellular distribution in HTLV-1-infected T cells.
Our data demonstrated that Tax1 is a viral lipid raft protein as evidenced by lipid raft fractionation analysis and confocal imaging. A significant portion of the Tax1 protein is constitutively present in the lipid rafts. Through selective interaction with IKK␥, Tax1 hijacks the entire IKK complex to the lipid rafts. The Tax1 mutants that fail to interact with IKK␥ are unable to bring the IKK complex to the microdomains. Depletion of IKK␥ abrogates Tax1-directed translo-cation of IKKs to lipid rafts and activation of NF-B. However, our data also show that lipid raft targeting of IKK␥ alone is not sufficient for lipid raft translocation of the catalytic subunits of IKK and activation of NF-B. Modification of IKK␥ or a catalytic activity of IKKs prior to lipid raft translocation of IKKs may be critical. Nevertheless, these findings are in strong support of the process involved in Tax1-dependent, lipid raft translocation of IB kinases. This scenario is necessary for induction of IB kinase activation by Tax1. Aside from Tax1, the viral transactivator Tat of HIV-1 has been shown to induce the NF-B activity in a manner highly dependent of the lipid raft protein, Lck (46,47). In another scenario, the latent infection membrane protein 1 of the Epstein-Barr virus resides in lipid rafts and mediates engagement with tumor necrosis factor receptor-associated TRAF to activate NF-B via the NF-Binducing kinase/IKK␣-dependent pathway (48). Indeed, the lipid raft microdomains are critical mediators for supporting NF-B activation induced by certain types of viral proteins and upon TCR activation.
The association of viral proteins with lipid rafts plays crucial roles in dysregulation of cell function and viral budding. For instance, HIV-1 Nef protein is a myristoylated protein that associates with T cell-specific Src kinase Lck within lipid rafts (49 -54), which directs a signaling cascade to enhance viral infectivity (52). The lipid rafts are critical for viral budding of HIV-1 and HTLV-1 (55)(56)(57)(58). Viral assemblies of these viruses occur within lipid rafts through interaction of the viral envelope, capsid, and host factors. It is therefore plausible that Tax1 may have a potential role in facilitating viral budding. This possibility is yet to be determined. Generation of a Tax mutant with defective lipid raft targeting but with reservation of CREB activation should be able to prove the point. Tax1 does not possess a defined lipid raft-targeting leader. However, it has been known that certain cellular proteins that lack well defined lipid raft targeting signals, such as stress-induced Hsp70 (59), are capable of relocating to the microdomains. Our data confirmed the lipid raft presence of Tax1, and it is plausible that Tax1 targets to this structure directly with an undefined signal or through interaction with a cellular lipid raft protein. Although it is still not clear about the mechanism of activation of IB kinases in lipid rafts, the experimental data have indicated that the membrane translocation of IKK is crucial for activation of IKK upon engagement of TCR. Similar to the lipid raft recruitment of signaling molecules such as ZAP70 and phosphatidylinositol 3-kinase, translocation of IKK and its upstream kinase Tak1 to the lipid raft microdomain enriches these molecules in this structure, providing an optimal microenvironment necessary for the activation event. In contrast to the physiological activation of TCR in which IKK is transiently translocated to lipid rafts in response to antigen stimulation, Tax1 induces a persistent lipid raft presence of IKK, correlating with the constitutive activity of NF-B.
Similar to Tax1, Tax2, a homologous Tax protein from non-pathogenic HTLV-2, interacts with IKK␥ and activates NF-B in a comparable potency to that of Tax1. Moreover, like Tax1, Tax2 can translocate to lipid rafts. In sharp contrast, Tax2 does not possess the ability to direct the lipid raft translocation of IKK in transfected HEK cells and in Tax2immortalized primary T cells. Such differential effects of the Tax proteins may be due to their distinct abilities to modulate the catalytic subunits of IKKs, IKK␣, and IKK␤. Tax1, but not Tax2, is able to interact strongly with both IKK␣ and IKK␤, a crucial step to precondition the IKK complex prior to their lipid raft translocation. Thus, it is clear that Tax1, which acts as a viral lipid raft modulator, holds a unique ability in hijacking IB kinases to lipid rafts, thereby contrib-uting to the persistent activation of NF-B and HTLV-1 oncogenesis.