Human T Cell Leukemia Virus Type 2 Tax-Mediated NF-κB Activation Involves a Mechanism Independent of Tax Conjugation to Ubiquitin and SUMO

Abstract

Permanent activation of the NF-κB pathway by the human T cell leukemia virus type 1 (HTLV-1) Tax (Tax1) viral transactivator is a key event in the process of HTLV-1-induced T lymphocyte immortalization and leukemogenesis. Although encoding a Tax transactivator (Tax2) that activates the canonical NF-κB pathway, HTLV-2 does not cause leukemia. These distinct pathological outcomes might be related, at least in part, to distinct NF-κB activation mechanisms. Tax1 has been shown to be both ubiquitinated and SUMOylated, and these two modifications were originally proposed to be required for Tax1-mediated NF-κB activation. Tax1 ubiquitination allows recruitment of the IKK-γ/NEMO regulatory subunit of the IKK complex together with Tax1 into centrosome/Golgi-associated cytoplasmic structures, followed by activation of the IKK complex and RelA/p65 nuclear translocation. Herein, we compared the ubiquitination, SUMOylation, and acetylation patterns of Tax2 and Tax1. We show that, in contrast to Tax1, Tax2 conjugation to endogenous ubiquitin and SUMO is barely detectable while both proteins are acetylated. Importantly, Tax2 is neither polyubiquitinated on lysine residues nor ubiquitinated on its N-terminal residue. Consistent with these observations, Tax2 conjugation to ubiquitin and Tax2-mediated NF-κB activation is not affected by overexpression of the E2 conjugating enzyme Ubc13. We further demonstrate that a nonubiquitinable, non-SUMOylable, and nonacetylable Tax2 mutant retains a significant ability to activate transcription from a NF-κB-dependent promoter after partial activation of the IKK complex and induction of RelA/p65 nuclear translocation. Finally, we also show that Tax2 does not interact with TRAF6, a protein that was shown to positively regulate Tax1-mediated activation of the NF-κB pathway.

INTRODUCTION

Human T cell leukemia virus type 1 (HTLV-1) and type 2 (HTLV-2) share a similar genomic organization with the presence, in the 3′ region of the genome, of a pX region encoding the viral Tax transactivator (Tax1 and Tax2, respectively) (reviewed in reference 1). However, while HTLV-1 is the agent of adult T cell leukemia/lymphoma (ATLL), a fatal disorder that affects 1 to 5% of infected individuals (2, 3), and of tropical spastic paraparesis/HTLV-1-associated myelopathy (TSP/HAM), a chronic neurological inflammatory disease (4, 5), HTLV-2 infection has only been associated with cases of TSP/HAM-like diseases (6, 7) and never with a malignant hematological disease. In addition to its ability to recruit the cellular polymerase II (PolII) machinery onto the viral promoter (8), Tax1 also alters the expression and/or function of a number of cellular factors involved in the control of the cell cycle or apoptosis, which ultimately leads to cell transformation (for reviews, see references 9 and 10). Identification of the distinctive features of Tax1 and Tax2 functions that would contribute to differences in pathogenesis has therefore been an active area of research.

Key distinctions have been identified between Tax1 and Tax2 in connection with the transformation abilities of HTLV-1 versus HTLV-2 (for a review, see reference 11). Critical functions were assigned to a PDZ-binding motif (PBM) located in the carboxy-terminal part of Tax1 that is absent in Tax2. Tax1 PBM was first involved in the induction of DNA damage-indicative micronuclei, a feature that is not shared by Tax2 (12). This motif was also shown to be responsible for Tax1-induced perturbation of hematopoietic CD34⁺ precursor maturation in vitro, a property that is not shared by Tax2 and that could be linked to viral persistence in hematopoietic precursor reservoirs in vivo (13, 14). Tax1 interaction with PDZ domain-containing cellular proteins such as Dlg was also suggested to be involved in Tax1 transforming activity (15).

In addition, Tax1 and Tax2 have been shown to bear distinct distributions in cells. Tax1 forms nuclear speckles (16–20) and is detected in Golgi/centrosome-associated cytoplasmic domains (21–24). Although it is also detectable in the nucleus, Tax2 is predominantly found in the cytoplasm of the cells (25–28) (for a review, see reference 29). Both proteins possess a nuclear export sequence located between amino acids (aa) 188 and 202 (30, 31). However, we previously showed that a 10-amino-acid sequence located in the N-terminal part of the proteins (aa 90 to 100) dictates their divergent localization (25).

Interestingly, cytoplasmic localization of Tax1 was shown to be critical for the constitutive activation of NF-κB observed in cells that express the viral protein (32, 33). Moreover, by analyzing Tax1/Tax2 chimeric proteins, we previously showed that their efficiency as NF-κB activators is correlated with their cytoplasmic to nuclear localization ratio (25). NF-κB is a family of transcription factors involved in the control of cell growth and survival (34, 35). Depending on the nature of the inducing signal, NF-κB may become activated either through the canonical or the noncanonical pathway. In contrast to Tax2, which activates only the canonical pathway, Tax1 constitutively activates both pathways (25, 27, 36–41). The molecular determinant of the ability to activate the noncanonical pathway has been mapped to Tax1 aa 225 to 232 (41).

The IκB kinase (IKK) complex is a central regulator of NF-κB. It consists of two catalytic subunits, IKK-α and IKK-β, and of a regulatory subunit, IKK-γ/NEMO. In the canonical pathway, activation of the IKK complex leads to the ubiquitination and degradation of IκBα and to the translocation of RelA/p65-p50 heterodimers into the nucleus (42). Although Tax1 directly interacts with IKK-γ/NEMO (43), the exact mechanism that leads to canonical NF-κB activation is still a matter of investigation (36, 44–47) (for a review, see reference 48). Tax1 undergoes posttranslational modifications, including phosphorylation, acetylation, ubiquitination, and SUMOylation. A critical role was notably assigned to Tax1 K63-linked polyubiquitination in the cytoplasmic steps of the NF-κB pathway. This includes binding to and activation of IKK (21–23, 49–53) (for recent reviews, see references 54 and 55). Tax SUMOylation was initially proposed to regulate the nuclear steps of the pathway by facilitating NF-κB promoter activation within Tax nuclear speckles (23, 52). However, this conclusion was challenged by the recent finding that a Tax1 mutant that supports only low levels of SUMOylation still activates an NF-κB promoter (56), further highlighting the key role played by Tax1 ubiquitination.

The Tax1 sequence contains 10 lysine residues. Ubiquitination occurs on lysines K4 to K8, while SUMOylation is restricted to lysines K7 to K8 (23, 49, 52). Ubc13, an E2-conjugating enzyme that functions together with members of the tumor necrosis factor (TNF) receptor-associated factor (TRAF) family, is required for Tax1 ubiquitination (57). RNF4, a SUMO-targeted ubiquitin ligase, also allows the ubiquitination of a SUMO-1-Tax fusion protein in vitro and enhances Tax1 transcriptional activity on an NF-κB-dependent promoter (58). On the other hand, the E3 ubiquitin ligase PDLIM2 mediates Tax1 K48-linked polyubiquitination, thereby targeting Tax1 to proteasomal degradation (59). Finally, another recent work demonstrated that USP20 deubiquitinates Tax1 and suppresses NF-κB activation (47).

In contrast, very few data are available concerning the mechanism by which Tax2 activates the canonical NF-κB pathway. We previously reported that Tax2 coimmunoprecipitates with IKKγ/NEMO in transfected cells (25), a result that was further confirmed by colocalization assays (36).

Tax1 lysines K4 to K8, which are targeted by ubiquitination, are conserved in the Tax2 sequence. A previous report suggested that, as Tax1, Tax2 was both ubiquitinated and SUMOylated (27). However, in this report, analyses of Tax1 and Tax2 modifications were performed in cells overexpressing ubiquitin or SUMO proteins, which may lead to false-positive results (60). We have therefore performed all our experiments with endogenous ubiquitin, SUMO, and, when possible, Tax proteins. We first show that Tax2 colocalizes with IKKγ/NEMO in infected cells. We then demonstrate that Tax2, in contrast to Tax1, is barely conjugated to endogenous ubiquitin, even when Ubc13 is overexpressed, or to SUMO. However, we also show that, like Tax1, Tax2 is acetylated. We further show that a Tax2 mutant that can neither be ubiquitinated nor SUMOylated or acetylated still significantly activates the NF-κB pathway. These results provide the notion that Tax2 activates the NF-κB pathway by a mechanism independent of its ubiquitination, SUMOylation, and acetylation. We then demonstrate that the carboxy-terminal region that surrounds Tax2 K4 to K8 is responsible for the lack of ubiquitination and SUMOylation of the protein. Finally, we show that, contrary to Tax1, Tax2 does not bind to TRAF6. Together, our results demonstrate that Tax1 and Tax2 harbor distinctive patterns of conjugation to ubiquitin and SUMO and that activation of the canonical NF-κB pathway by Tax1 or Tax2 does not have the same requirements toward Tax ubiquitination.

MATERIALS AND METHODS

Cell culture.

HeLa and 293T cells were grown in Dulbecco's modified Eagle's medium (DMEM). HTLV-2-transformed C19 cells, HTLV-uninfected CEM cells, and Jurkat T cells were grown in Rosewell Park Memorial Institute (RPMI) 1640 medium. DMEM and RPMI medium were supplemented with 10% fetal bovine serum (20% for C19 cells) and antibiotics (penicillin, 100 U/ml; streptomycin, 100 μg/ml). Cell lines were maintained at 37°C in 5% CO₂.

Constructs.

pSG5M backbone vector, pSG5M-Tax1, pSG5M-Tax1-His, pSG5M-Tax1-K0-His, pcDNA4/HisMax-Tax2 (referred to as Tag₅₀-His-Tax2), HA-SUMO1, NF-κB-luciferase, and HTLV-1-LTR-luciferase plasmids were described previously (25, 49, 52, 61). The Ubc13-encoding vector was generously provided by E. Harhaj (62). The Flag-TRAF6 construct (described in reference 63) was kindly provided by B. Carter (Vanderbilt University, Nashville, TN). To generate the pSG5M-Tax2-His vector, Tax2 cDNA was amplified from the pcDNA4/HisMax-Tax2 vector (25) using primers for the carboxy-terminal fusion of a His₆ tag (forward primer, CTCCCGGAATTCCACCATGGCCCATTTCCCAGGATTTG; reverse primer, GAGCCCAAGCTTTCAGTGATGGTGGTGATGGTGCTTGGGATTGTTTGTGTGAGACGGT). PCR was performed according to the manufacturer's recommendations (Phusion; Finnzymes). The Tax2 sequence was then inserted between the EcoRI and HindIII sites in the pSG5M plasmid. Similarly, the His-tagged chimeric proteins were amplified from the previously described pSG5M plasmids (25). pSG5M-Tax2-K0-His (i.e., where all lysine residues were replaced by arginine), pSG5M-Tax2-K5-11R-His (i.e., where lysine residues 5 to 11 were replaced by arginine), and pSG5M-Tax2-K5-13R-His (i.e., where lysine residues 5 to 13 were replaced by arginine) were obtained using the QuikChange II site-directed mutagenesis kit (Stratagene).

Nickel pulldown.

Forty-eight hours posttransfection (Polyfect; Qiagen), HeLa cells were harvested in cold phosphate-buffered saline (PBS) and lysed in a highly denaturing lysis buffer and Ni-nitrilotriacetic acid (NTA) pulldown was performed as described previously (49). Bound proteins were eluted in Laemmli buffer and processed for Western immunoblot analyses. For whole-cell lysates, cells were washed twice with PBS and lysed (50 mM Tris-HCl [pH 7.4], 120 mM NaCl, 5 mM EDTA, 0.5% NP-40, 0.2 mM Na₃VO₄, 1 mM dithiothreitol [DTT], 1 mM phenylmethylsulfonyl fluoride [PMSF]) in the presence of protease inhibitors (Complete; Roche Diagnostics) and cell debris was pelleted by centrifugation.

Western immunoblot analyses were performed as described previously (25). Analyses were performed with the following antibodies: anti-His₆ (sc-804; Santa Cruz), anti-ubiquitin (sc-8017; Santa Cruz), anti-SUMO2/3 (ab3742; Abcam), anti-HA (HA.11 clone 16B12; Eurogentec), and antiacetylated lysine (Ac-K-103, product no. 9681; Cell Signaling). The total signal intensity of each lane was measured using ImageJ software (Wayne Rasband, NIH). Values were then normalized to signal intensities in the corresponding anti-His blots. The signal intensity for Tax1-His modified forms (Fig. 1D) is expressed as a percentage of the total Tax1-His signal intensity (unconjugated Tax plus modified forms). Signal intensities for the modified forms of Tax1, Tax2 (Fig. 1E and F), and Tax1/2 chimeras (see Fig. 6C) are expressed as a percentage of wild-type Tax1 modified forms.

Fig 1.

Tax2 activates the NF-κB pathway despite low levels of conjugation to endogenous ubiquitin and SUMO. (A) HTLV-2-transformed C19 T lymphocytes cells were mixed with CEM-uninfected (HTLV-negative) lymphocytes at a 1:10 ratio. Cells were stained with an anti-Tax2 antibody (green signal) and an anti-IKKγ/NEMO antibody (red signal). Nuclei were stained using DAPI (blue signal). Cells were observed as described in Materials and Methods. Scale bar, 10 μm. (B) HeLa cells were transfected with a NF-κB-luc construct together with either an empty vector or a Tax1-His- or Tax2-His-encoding vector. Eighteen hours later, luciferase activity was measured and normalized. Results are shown as mean relative luciferase units (RLU) ± standard errors of the means (SEM) of one experiment performed in duplicate and representative of three independent experiments. (C) Schematic representation of Tax1 and Tax2, highlighting the positions of lysine residues in both proteins. Ten lysine residues are present in Tax1 (aa positions 85, 88, 111, 189, 197, 263, 280, 284, 324, and 346) and 13 in Tax2 (aa positions 88, 100, 111, 149, 185, 189, 197, 263, 280, 284, 323, 347, and 356). Tax1 lysines targeted by ubiquitination and SUMOylation are shown. (D) HeLa cells were transfected with 2 μg of an empty vector, an untagged Tax1-encoding vector (lrr.), a Tax1-His-encoding vector, or a Tax2-His-encoding vector. Histidine-tagged Tax proteins and their modified forms were retained on Ni-NTA beads and processed for Western blot analysis using anti-ubiquitin (a) or anti-SUMO2/3 (b) antibodies. (c) Levels of purification of Tax1-His and Tax2-His were determined by Western blot analysis using an anti-His antibody. See Materials and Methods for the quantification method. (E) 293T cells were transfected with 2 μg of an empty vector, a Tax1-His-encoding vector, or a Tax2-His-encoding vector and 10 μg of a HA-SUMO1-encoding vector. (a) Histidine-tagged Tax proteins and their modified forms were retained on Ni-NTA beads and processed for Western blot analysis using an anti-HA antibody. (b) Levels of purification of Tax1-His and Tax2-His were determined by Western blot analysis using an anti-His antibody. See Materials and Methods for the quantification method. (F) HeLa cells were transfected with 1 μg of a Tax1-His-encoding vector or a Tax2-His-encoding vector, together with 5 μg of an Ubc13-encoding vector (+) or an empty vector (−), as specified. (a) Histidine-tagged Tax proteins and their modified forms were retained on Ni-NTA beads and processed for Western blot analysis using an anti-ubiquitin antibody. (b) Levels of expression of Tax1-His and Tax2-His in input were determined by Western blot using an anti-His antibody. See Materials and Methods for the quantification method. (G) HeLa cells were transfected with a NF-κB-luc construct together with either an empty vector, a Tax1-His-encoding vector, or Tax2-His-encoding vector, in the absence (+ empty vector) or presence (+ Ubc13) of a Ubc13-encoding vector (1 μg), as indicated. Eighteen hours later, luciferase activity was measured and normalized. Results are shown as mean RLU ± SEM of one experiment performed in duplicate and representative of two independent experiments. Results of t tests are shown (**, P < 0.01; ns, nonsignificant).

Fig 6.

Domains encompassing K4 to K10 and K6 to K10 are sufficient for Tax1 ubiquitination and SUMOylation, respectively. (A) Schematic representation of the Tax1/Tax2 chimeras used in panels B, C, and D. The table summarizes the results of panel C: Tax ubiquitination (Ub column) and SUMOylation (SUMO2/3 column) status is indicated by a plus (>100% relative intensity compared to Tax1) or minus (<10% relative intensity compared to Tax1) sign. (B) HeLa cells were transfected with 400 ng of the indicated plasmids. Cells were stained with anti-His (green signal) and anti-GM130 (red signal) antibodies. Nuclei were stained using DAPI (blue signal). Cells were observed as described in Materials and Methods. Arrows point to the cytoplasmic fraction of Tax1. Scale bar, 10 μm. The percentage of cells containing cytoplasmic Tax (C), Tax nuclear speckles (NS), or both cytoplasmic Tax and Tax nuclear speckles (C+NS) is indicated. At least 100 cells were analyzed under each condition. (C) HeLa cells were transfected with 2 μg of the indicated vectors. Histidine-tagged Tax proteins and their modified forms were retained on Ni-NTA beads and processed for Western blot analysis using an anti-ubiquitin (a) or an anti-SUMO2/3 (b) antibody, as indicated. (c) Levels of purification of each Tax construct were determined by Western blot analysis using an anti-His antibody. See Materials and Methods for the quantification method. (D) HeLa cells were transfected with a NF-κB-luc construct together with either an empty vector or the indicated chimeric Tax1/2-His-encoding vectors or the Tax1-K0-encoding vector. Eighteen hours later, luciferase activity was measured and normalized. Results are shown as mean RLU ± SEM of one experiment performed in duplicate and representative of three independent experiments. (E) HeLa cells were transfected with an HTLV1-LTR-luc construct together with either an empty vector or the indicated chimeric Tax1/2-His-encoding vectors or the Tax1-K0-encoding vector. Eighteen hours later, luciferase activity was measured and normalized. Results are shown as mean RLU ± SEM of one experiment performed in duplicate and representative of three independent experiments.

Indirect immunofluorescence analyses.

Twenty-four hours posttransfection (Effectene; Qiagen), HeLa cells were processed for immunofluorescence analyses as described previously (25). Analyses were performed with the following antibodies: anti-His₆ (sc-804 [Santa Cruz] or ab5000 [Abcam]), anti-GM130 (610823; BD Transduction Laboratories), anti-γ-tubulin (sc-7396; Santa Cruz), anticalreticulin (PA3-900; Affinity BioReagents), anti-IKK-γ/NEMO (611306; BD Transduction Laboratories), anti-RelA (sc-372; Santa Cruz), fluorescein- or Texas-Red-conjugated goat anti-rabbit (Vector), horse anti-mouse (Vector), and donkey anti-goat (Jackson ImmunoResearch). Coverglasses were washed, mounted in DAPI (4′,6-diamidino-2-phenylindole)-containing Fluoromount-G (Southern Biotech), and examined under a Leica SP5 spectral confocal microscope. A similar procedure was performed for C19 cells using the anti-Tax2 antibody (GP3738) (25). For RelA/p65 nuclear translocation quantification, total cells and nuclei were delineated using ImageJ software and ratios of the mean brightness of the nuclear/total RelA/p65 staining were calculated. Values obtained for nontransfected cells were then subtracted from those for Tax-expressing cells in the same field.

Coimmunoprecipitation assay.

Twenty-four hours posttransfection (Lipofectamine; Invitrogen), HeLa cells were lysed (120 mM NaCl, 20 mM Tris-HCl [pH 8], 0.2 mM NaF, 0.2 mM EGTA, 0.2% sodium deoxycholate, 0.5% NP-40, protease, and phosphatase inhibitors [Roche]). Cell lysates were incubated overnight with an anti-His antibody (27-4710-01; GE Healthcare) or an anti-Flag M2 affinity gel (A2220; Sigma) at 4°C. When necessary, antibody complexes were captured on protein G-Sepharose beads (GE Healthcare) for 1 h at 4°C. Beads were then washed 5 times in lysis buffer before elution in Laemmli buffer. Samples were loaded into 10% Bis-Tris gels, subjected to electrophoresis, and transferred onto a nitrocellulose (Optitran BA-S83; Whatman) or polyvinylidene difluoride (PVDF) (Immobilon-P; Millipore) membrane. Membranes were analyzed using the following antibodies: anti-His (27-4710-01 [GE Healthcare] or sc-804 [Santa Cruz]), anti-IKK-γ/NEMO (sc-8330; Santa Cruz), anti-phospho-IKK-α/β (C84E11; Cell Signaling Technology), and anti-Flag (M2; Sigma). Signal intensities for Tax1-K0- and Tax2-K0-coimmunoprecipitated partners (Fig. 3B) are expressed as percentages of the signal intensities obtained for wild-type Tax1 and Tax2, respectively, normalized to the input intensities.

Fig 3.

Tax2 activation of the IKK complex involves a mechanism independent of Tax conjugation to ubiquitin and SUMO. (A) HeLa cells were transfected with 200 ng of wild-type or 400 ng of lysineless Tax1-His- or Tax2-His-encoding vectors. Cells were stained with anti-His (green signal) and anti-IKKγ/NEMO (red signal) antibodies. Nuclei were stained using DAPI (blue signal). Cells were observed as described in Materials and Methods. Scale bar, 10 μm. Enlargements are shown in the “Merge” panel. The intensity of fluorescence for each staining along the line drawn on the enlarged merged images is plotted in the diagrams on the right. (B) HeLa cells were transfected with 5 μg of Tax1-His-, Tax2-His-, or Tax2-K0-His-encoding vectors or 10 μg of a Tax1-K0-His-encoding vector, as indicated. Lysates were immunoprecipitated with an anti-His antibody, and a Western blot analysis was performed using anti-IKKγ/NEMO (a), anti-His (b), and anti-phospho-IKK-α/β (c) antibodies. Whole-cell lysates were also blotted with the same antibodies (d to f). See Materials and Methods for the quantification method.

Luciferase assays.

HeLa or Jurkat cells were transfected (Effectene and Superfect, respectively; Qiagen) with NF-κB-luc or an HTLV-1-LTR-luc reporter plasmid (50 ng) together with either the pSG5M empty vector or the His-tagged Tax1/2-encoding plasmid (100 ng for wild-type and chimeric Tax-His plasmids, 200 ng for K0, K5-11R, and K5-13R plasmids, and 600 ng for the Tag₅₀-His-Tax2 plasmid) and the indicated amount of the Ubc13-encoding plasmid. Amounts of transfected Tax plasmids were chosen as the lowest dose, inducing significant NF-κB activity. Transfections were carried out in the presence of a Renilla luciferase vector (phRG-TK, 4 ng) in order to normalize the transfection efficiency. Luciferase activity was assayed 18 h posttransfection using the Dual-Luciferase reporter assay system (Promega). When indicated, statistical unpaired t tests were performed using Prism software, and the two-tailed P value was calculated.

RESULTS

Tax2 activates the NF-κB pathway despite low levels of conjugation to endogenous ubiquitin and SUMO.

Data previously obtained in transfected cells suggest that the mechanism of Tax2-induced NF-κB activation involves Tax2 interaction with IKKγ/NEMO (25, 36). In order to better characterize this mechanism, we sought to confirm these results in HTLV-2 chronically infected T lymphocytes (C19 cell line). Immunofluorescence studies were performed in order to investigate the HTLV-2-dependent relocalization of IKKγ/NEMO to Tax2-positive cytoplasmic domains. As published previously (25), C19 cells were mixed with uninfected CEM T lymphocytes at a 1:10 ratio, allowing us to have an internal negative control, and the coverslip was stained for Tax2 and IKK-γ/NEMO. IKK-γ/NEMO spots were visible only in Tax2-positive cells where both proteins colocalized and formed cytoplasmic spots (Fig. 1A). These results demonstrate that Tax2 also recruits IKK-γ/NEMO in HTLV-2-infected T lymphocytes and further support the notion that the mechanism of Tax2-induced NF-κB activation involves Tax2 interaction with IKKγ/NEMO.

Because Tax1 interaction with IKKγ/NEMO requires Tax1 ubiquitination, we set up an assay that compares Tax1 and Tax2 ubiquitination patterns in transfected HeLa cells. A short histidine tag was added to the carboxy-terminal end of Tax1 and Tax2 proteins, allowing Tax purification on Ni-NTA beads. We first performed an NF-κB reporter gene assay in HeLa cells to verify that the addition of the tag did not alter Tax function (Fig. 1B). Tax1 and Tax2 expression induced an approximately 10-fold induction of NF-κB activity. As a control, an HTLV1-LTR-luciferase test was also performed and demonstrated similar activation of the reporter construct by the two Tax proteins (data not shown). These results demonstrate that histidine-tagged Tax1 and Tax2 are both potent NF-κB activators in HeLa cells.

We then investigated the ubiquitination patterns of Tax1 and Tax2. Tax1 lysines K4 to K8, which are targeted by ubiquitination, are conserved in the Tax2 sequence (Fig. 1C). In order to avoid any false-positive results that could be due to ubiquitin overexpression, we compared the levels of Tax1 and Tax2 conjugation to endogenously expressed ubiquitin [Fig. 1D(a)]. Tax pulldown was performed in a highly denaturant guanidine-containing buffer to prevent copurification of any noncovalently bound partners, while allowing purification of posttranslationally modified Tax proteins (49). Guanidine also inhibits all cellular deubiquitinase and SUMO protease activities that would otherwise decrease the amount of modified Tax proteins that are retrieved. Surprisingly, no signal was observed for Tax2 when blotting the purified extracts with an anti-Ub antibody [Fig. 1D(a), lane 4]. Consistent with previous reports (21–23, 49, 52), a ladder of Tax1 products was detected after blotting with the same anti-Ub antibody [Fig. 1D(a), lane 3], showing that Tax1 was properly conjugated to endogenous ubiquitin under the same experimental conditions.

We then investigated whether Tax2 was or was not SUMOylated in cells expressing endogenous levels of SUMO proteins. Following pulldown, endogenously SUMOylated Tax proteins were revealed using a SUMO2/3 specific antibody [Fig. 1D(b)]. No Tax2 signal was detected [Fig. 1D(b), lane 4] while, as expected from earlier studies (23, 52), SUMO-modified Tax1 forms were present [Fig. 1D(b), lane 3]. An analysis of pulldown fractions indicated that Tax1-His and Tax2-His were bound equivalently to nickel beads [Fig. 1D(c)]. The quantification of the ladder of Tax1 high-molecular-weight products [Fig. 1D(c), lane 3] indicated that the level of modified Tax1 (ubiquitinated or SUMOylated) reached 25% of total Tax1. The quantification of Tax2 signals showed only the background level of modified products (<1%) (lane 4). An analysis of Tax modification under conditions where SUMO1 was overexpressed was also undertaken (Fig. 1E). SUMO1 is an isoform that is unable to form chains and can only mono-SUMOylate proteins or terminate chains of poly-SUMO2/3. SUMO1-modified Tax2 could barely be detected (5% of the signal intensity obtained for Tax1, which is equivalent to the background level), while Tax1 SUMOylation was easily detected. Interestingly, both mono- and poly-SUMOylation of Tax1 by SUMO1 species were present, consistent with the presence of Tax1 products conjugated to a unique SUMO1 monomer on one or more lysines and/or to SUMO2/3 chains terminated by a SUMO1 monomer. Subsequent experiments were only performed with SUMO2/3.

Together, these results demonstrate that, at steady state and in contrast to Tax1, ubiquitinated and SUMOylated forms of Tax2 are barely detectable in HeLa cells expressing endogenous levels of ubiquitin and SUMO proteins. Similar results were obtained in 293T cells (data not shown).

Ubc13 is an E2 ubiquitin-conjugating enzyme that catalyzes K63-linked polyubiquitination and is known to support Tax1 polyubiquitination (57). In order to confirm that Tax2 and Tax1 behave distinctly toward ubiquitination, we investigated the effect of overexpressing Ubc13 on Tax2 ubiquitination (Fig. 1F). As expected, Ubc13 overexpression resulted in a 5-fold increase of Tax1 ubiquitination levels [Fig. 1F(a), lane 2 versus lane 1, note that a short exposure time is shown in order to limit saturation of the signal in lane 2 and to allow quantification]. In contrast, even under those conditions, Tax2 endogenous ubiquitination remained barely detectable [Fig. 1F(a), lane 4 versus lane 3, <10% of the signal intensity obtained for Tax1, which is equivalent to the background level]. We then performed NF-κB reporter gene assays (Fig. 1G). In accordance with the results from the ubiquitination assay and contrary to Tax1, for which overexpression of Ubc13 induced a significant increase in NF-κB promoter activation (unpaired t test, P = 0.0063), NF-κB activation by Tax2 was not significantly changed upon overexpression of Ubc13 (P = 0.09). Together, these results indicate that, in contrast to Tax1, Tax2 is a potent NF-κB activator despite very low levels of endogenous ubiquitination and SUMOylation.

Tax2 retains its ability to activate the canonical NF-κB pathway in the strict absence of Tax2 ubiquitination and SUMOylation.

Levels of Tax2 posttranslational modifications that are undetectable under our experimental conditions might suffice to allow efficient activation of NF-κB. Therefore, we studied the properties of a Tax2 mutant in which all 13 lysine codons were substituted by arginine codons and that can no longer be modified by either ubiquitin or SUMO (Fig. 2A, Tax2-K0). Of note, this mutant should support neither branched nor linear polyubiquitination, a recently described modification that also targets lysine residues. As expected, the lysineless Tax2-K0 mutant underwent neither ubiquitination nor SUMOylation [Fig. 2B(a) and (b), lane 5]. As controls, strong signals were obtained in cells expressing Tax1 (lane 2), while the lysineless Tax1 construct (Tax1-K0-His) was neither ubiquitinated nor SUMOylated (lane 3). Wild-type and K0 Tax1 and Tax2 were equivalently bound to nickel beads [Fig. 2B(c)].

Fig 2.

Tax2-mediated NF-κB activation involves a mechanism independent of Tax conjugation to ubiquitin and SUMO. (A) Schematic representation of wild-type Tax2 and the Tax2-K0 mutant, in which all 13 lysine residues were mutated into arginine. (B) HeLa cells were transfected with 2 μg of wild-type or 4 μg (to ensure equivalent expression levels) of lysineless Tax1- or Tax2-encoding vectors. Histidine-tagged Tax proteins and their modified forms were retained on Ni-NTA beads and processed for Western blot analysis using an anti-ubiquitin (a) or an anti-SUMO2/3 (b) antibody. (c) Levels of purification of each Tax construct in purified extracts were determined by Western blot analysis using an anti-His antibody. (C) HeLa or Jurkat cells were transfected with a NF-κB-luc or HTLV-1-LTR-luc construct together with either an empty vector or with wild-type or lysineless Tax1-His- or Tax2-His-encoding vectors, as indicated. Eighteen hours later, luciferase activity was measured and normalized. Results are shown as mean ± SEM of one experiment performed in triplicate and representative of three independent experiments. Results of t tests are shown (**, P < 0.01; *, P < 0.05; ns, nonsignificant). Levels of expression of each Tax construct in HeLa cells were determined by Western blot analysis using an anti-His antibody. (D) HeLa cells were transfected with 1 μg of C-terminally tagged wild-type Tax1 or Tax2, 2 μg of C-terminally tagged lysineless Tax2, or 6 μg of N-terminally tagged (Tag₅₀-His) Tax2-encoding vectors (to ensure equivalent expression levels). (a) Histidine-tagged Tax proteins and their modified forms were retained on Ni-NTA beads and processed for Western blot analysis using an anti-ubiquitin antibody. (b) Levels of purification of each Tax construct were determined by Western blot analysis using an anti-His antibody. (E) HeLa cells were transfected with an NF-κB-luc construct together with an empty vector, with wild-type or lysineless Tax2-His-encoding vectors, or with a Tag₅₀-His-Tax2-encoding vector, as indicated. Eighteen hours later, luciferase activity was measured and normalized. Results are shown as mean RLU ± SEM of one experiment performed in triplicate and representative of two independent experiments. Results of t tests are shown (ns, nonsignificant). Levels of expression of each Tax construct were determined by Western blot analysis using an anti-His antibody. (F) Schematic representation of the Tax1, Tax2, and Tax2 mutants (top). HeLa cells were transfected with 1 μg of wild-type Tax2 or 2 μg of the indicated mutant Tax2-encoding vectors (to ensure equivalent expression levels) (bottom). (a) Histidine-tagged Tax proteins and their modified forms were retained on Ni-NTA beads and processed for Western blot analysis using an anti-acetyl-lysine antibody. (b) Levels of purification of each Tax construct were determined by Western blot analysis using an anti-His antibody. (G) HeLa cells were transfected with a NF-κB-luc construct together with either an empty vector or with wild-type or mutant Tax2-His-encoding vectors, as indicated. Eighteen hours later, luciferase activity was measured and normalized. Results are shown as mean RLU ± SEM of one experiment performed in triplicate and representative of two independent experiments. Results of t tests are shown (ns, nonsignificant). Levels of expression of each Tax construct were determined by Western blot analysis using an anti-His antibody.

We then performed NF-κB reporter gene assays (Fig. 2C). Strikingly, in HeLa cells, expression of Tax2-K0 still triggered significant NF-κB activation (Fig. 2C, upper panel; unpaired t test, P = 0.0153 compared to the empty vector) that reached about 60% of that of wild-type Tax2. In contrast, and consistent with previously published results (22, 23, 52), Tax1-K0 was impaired in its ability to activate the NF-κB signaling pathway (values equal to or smaller than the basal values). In Jurkat T cells (Fig. 2C), expression of wild-type Tax2 led to an approximately 30-fold increase in NF-κB activity compared to basal activity. Importantly, Tax2-K0 also led to a significant 18-fold induction of NF-κB activity (unpaired t test, P = 0.0032 compared to the empty vector). As a control, the ability of the lysineless Tax1 and Tax2 mutants to activate the transcription of an HTLV1-LTR-luc reporter gene was also assessed in HeLa cells (Fig. 2C, bottom panel). While the activity of Tax1-K0 was abolished, as reported previously (values equal to or smaller than the basal values) (22, 23, 52), Tax2-K0 retained significant activity (P = 0.0265 compared to the empty vector). Taken together, these results demonstrate that, in sharp contrast to Tax1, Tax2 is able to efficiently activate the NF-κB pathway in the strict absence of lysine modification by either ubiquitin or SUMO.

Because the Tax2-K0 mutant might be ubiquitinated at low but functional levels on its N-terminal alpha-amino group, we used a 50-amino-acid-long N-terminally tagged Tax2 construct (Tag₅₀-His-Tax2) (25). According to the literature, adding such a long tag is the only way to prevent N-terminal ubiquitination of a protein in vivo (for example, see reference 64). Ni-NTA pulldown experiments did not show any Tax2 ubiquitination (Fig. 2D, lane 5). Luciferase experiments then demonstrated that Tag₅₀-His-Tax2 retains the ability to activate an NF-κB-dependent promoter (Fig. 2E). Together, these results indicate that Tax2 activates NF-κB independently of an N-terminal ubiquitination.

Lodewick et al. recently showed that both Tax1 and Tax2 are acetylated (65). Since acetylation occurs on lysine residues, it is expected that Tax2-K0 would be defective for acetylation, which might in turn impact its transcriptional activity. Since Tax1 acetylation occurs on K10, which is located in the C-terminal part of the protein, we hypothesized that Tax2 acetylation might also occur on C-terminal lysines (K12 and K13). We therefore constructed two Tax2 mutants in which the central lysine residues only (Tax2-K5-11R) or in combination with the C-terminal lysine residues (Tax2-K5-13R) were mutated (see Fig. 2F). We predicted that only the latter mutant would be affected in its acetylation. The acetylation profile of these mutants was then determined by Ni-NTA assays. As expected, both wild-type and K5-11R Tax2 were acetylated (Fig. 2F, lanes 1 and 3), while Tax2-K0 and Tax2-K5-13R were not (lanes 2 and 4). Luciferase assays showed that both Tax2-K5-11R and Tax2-K5-13R mutants were able to activate NF-κB at a level similar to wild-type Tax2 (Fig. 2G). These data demonstrate that acetylation is not affecting Tax2-dependent NF-κB activation on a nonintegrated promoter and are consistent with those of Lodewick et al. for Tax1. These results also indicate that the partial loss of function of the Tax2-K0 mutant is not due to impaired acetylation but rather is a consequence of mutating a high number of lysine residues.

We then investigated the mechanism underlying the activation of NF-κB by Tax2 and Tax2-K0. Tax1-induced NF-κB activation is triggered by Tax1 binding to and clustering of IKK-γ/NEMO in perinuclear structures, followed by the phosphorylation of the IKK kinases IKK-α and IKK-β (21, 52). We therefore analyzed whether Tax2 and Tax2-K0 were able to induce perinuclear relocalization of IKK-γ/NEMO (Fig. 3A). Both Tax1 and Tax2 induced IKK-γ/NEMO relocalization in perinuclear spots where colocalization of both Tax proteins with IKKγ/NEMO is found [Fig. 3A(a) and (c) and quantification on the right]. Although the absence of lysines drastically altered Tax2 localization [compare Tax2 signals in Fig. 3A(c) and (d)], perinuclear spots containing IKK-γ/NEMO were still observed [Fig. 3A(d) and quantification on the right]. Consistent with previous data (22, 23), Tax1-K0 appeared as a diffuse staining and did not induce any IKK-γ/NEMO relocalization [Fig. 3A(b)]. Thus, Tax2-induced IKK-γ/NEMO perinuclear relocalization occurs despite the lack of Tax2 ubiquitination and SUMOylation. We next studied the ability of Tax2 and Tax2-K0 to bind IKK-γ/NEMO (Fig. 3B). IKK-γ/NEMO coprecipitated with both Tax1 and Tax2 [Fig. 3B(a), lanes 2 and 4], while it did not with Tax1-K0 (lane 3). Importantly, IKK-γ/NEMO also coprecipitated with Tax2-K0 (lane 5), indicating that Tax2 interaction with IKK-γ/NEMO occurs in the absence of Tax2 ubiquitination and SUMOylation. Furthermore, contrasting with Tax1-K0, Tax2-K0 was still able to precipitate phospho-IKK-α/β [Fig. 3B(c), lane 5]. Although the amount of precipitated phospho-IKK-α/β found for Tax2-K0 reached only about 40% of that of wild-type Tax2, these results indicate that, in contrast to Tax1, the strict absence of Tax2 ubiquitination and SUMOylation still allows significant activation of the IKK complex. Control Western blot analyses showed that equivalent amounts of Tax-K0 proteins were expressed and/or immunoprecipitated compared to their wild-type counterparts [Fig. 3B(b) and (e)]. Control Western blot analyses were also performed for IKK-γ/NEMO and phospho-IKK-α/β [Fig. 3B(d) and (f)].

We next studied the nuclear translocation of the NF-κB RelA/p65 subunit, which is a well-established indicator of NF-κB activation (Fig. 4). Both Tax1 and Tax2 induced RelA/p65 nuclear translocation [Fig. 4A(a) and (c), RelA/p65 panel, compare Tax-expressing cells (arrows) to nontransfected cells (arrowheads)], while Tax1-K0 did not [Fig. 4A(b)]. Interestingly, Tax2-K0 also induced a significant nuclear translocation of RelA/p65 [Fig. 4A(d), RelA/p65 panel, compare Tax-expressing cells (arrows) to nontransfected cells (arrowheads)]. Quantification of a nuclear RelA/p65 signal increase (Fig. 4B) allowed us to confirm that, unlike Tax1-K0, Tax2-K0 induced an efficient (80% of wild-type Tax2-induced translocation) RelA/p65 nuclear translocation. These results indicate that, in contrast to Tax1, Tax2 is able to induce RelA/p65 nuclear translocation in the strict absence of Tax2 ubiquitination and SUMOylation.

Fig 4.

Tax2-induced nuclear translocation of RelA/p65 involves a mechanism independent of Tax conjugation to ubiquitin and SUMO. (A) HeLa cells were transfected with 200 ng of wild-type or 400 ng of lysineless Tax1-His- or Tax2-His-encoding vectors, as indicated. Twenty-four hours later, cells were fixed and stained with anti-His (green signal) and anti-RelA/p65 (red signal) antibodies. Nuclei were stained using DAPI (blue signal). Cells were observed as described in Materials and Methods. Scale bar, 10 μm. Arrows indicate Tax-expressing cells. Arrowheads indicate nontransfected, Tax-negative cells. (B) Total cells and nuclei as shown in panel A were delineated using Image J software, and the mean brightness ratio of the nuclear RelA/p65 staining was calculated. Values obtained for nontransfected cells were then subtracted from those for Tax-expressing cells in the same field. Tax-induced increase in RelA/p65 nuclear signal intensity is plotted.

Taken together, these results indicate that Tax2 is able to activate NF-κB in the absence of prior conjugation to ubiquitin and SUMO by a mechanism involving binding to and activation of the IKK complex followed by RelA/p65 nuclear translocation.

One explanation for the fact that Tax2 and Tax2-K0 are still able to activate NF-κB would be that the lack of Tax2-associated Ub chains is compensated for by ubiquitinated chains bound to Tax2 cellular partners. One potential candidate for this function is TRAF6, an E3 ligase working with Ubc13 that triggers its own K63 ubiquitination and that of other effectors of the NF-κB pathway. A coimmunoprecipitation experiment indicated that, while Tax1 or its K0 mutant interacts with TRAF6 (Fig. 5, lanes 3 and 4), this was not the case for Tax2 or its K0 mutant (lanes 5 and 6), ruling out a direct role of ubiquitinated TRAF6 in NF-κB activation by Tax2.

Fig 5.

Tax2 interaction with TRAF6 is not necessary for efficient NF-κB activation. HeLa cells were transfected with 1.5 μg of Tax1-His- or Tax2-His-encoding vectors or 3 μg of Tax1-K0-His- or Tax2-K0-His-encoding vectors, together with 3 μg of a Flag-TRAF6-encoding vector, as indicated. Lysates were immunoprecipitated with anti-Flag-coupled beads, and Western blot analysis was performed using anti-Flag and anti-His antibodies. Whole-cell lysates were also blotted with the anti-His antibody.

Ubiquitination and SUMOylation are determined by distinct domains of Tax.

The clearly distinct Tax1 versus Tax2 posttranslational modification patterns prompted us to investigate the determinants of such differences. Since lysines K4 to K8 are conserved in Tax2 (Fig. 1C), we reasoned that the differential ubiquitination and SUMOylation statuses of Tax1 and Tax2 are not linked to the presence or absence of the targeted lysines but rather to specific domains that surround these lysines. Alternatively, distinct Tax1 versus Tax2 posttranslational modification patterns could be a consequence of distinct subcellular localizations that could be critical for interaction with the modification machineries. Indeed, as mentioned above, Tax1 is both nuclear and cytoplasmic while Tax2 is predominantly cytoplasmic (25, 26, 28).

We therefore investigated the relationship between Tax1 and Tax2 sequence, localization, and posttranslational modification profiles using a series of Tax1/2 chimeric proteins and Tax1 or Tax2 as controls (Fig. 6A). As previously described (16–24), Tax1 displayed a mixed nuclear and cytoplasmic localization (Fig. 6B, see pattern frequencies on the right of the panel). In the nucleus, Tax1 formed nuclear speckles whereas Tax1 cytoplasmic fraction (marked by the white arrow) colocalized with GM130 (Golgi apparatus marker). As previously reported (25), Tax2 displayed a predominantly perinuclear localization with a low number of cells displaying nuclear speckles (see pattern frequencies on the right). Contrary to Tax1, an overlay of the Tax2 cytoplasmic signal with GM130 could not be observed. Thus, Tax2 cytoplasmic localization sharply differs from that of the cytoplasmic fraction of Tax1. Consistent with our previous report, the chimeric Tax1_1–90-2_91–113-1_114–353 protein localization was similar to that of Tax2 (Fig. 6B) (25), confirming that the domain encompassing Tax aa 91 to 113 is critical for Tax localization. We thus investigated the posttranslational modification pattern of Tax1_1–90-2_91–113-1_114–353, which only contains 22 Tax2 amino acids in its sequence (Fig. 6C). Interestingly, this chimeric protein was ubiquitinated and SUMOylated at levels similar to Tax1 [Fig. 6C (a) and (b), lane 3 versus lane 2]. These results indicate that subcellular localization alone does not account for low levels of Tax2 ubiquitination and SUMOylation.

We then assessed the localization and posttranslational modification profiles of two other constructs: Tax2_1–113-1_114–353 and Tax2_1–220-1_221–353. Consistent with the fact that aa 90 to 113 control Tax subcellular localization, and similarly to wild-type Tax2, Tax2_1-113-1_114-353 and Tax2_1-220-1_221-353 were most predominantly found in the cytoplasm and did not colocalize with GM130 (Fig. 6B, see pattern frequencies on the right of the panel). However, ubiquitination of Tax2_1-220-1_221-353 was significantly reduced compared to that of Tax1 [Fig. 6C (a), lane 5 versus lane 2], while ubiquitination of Tax2_1-113-1_114-353 was similar to that of Tax1 (lane 4 versus lane 2). These results demonstrate that aa 114 to 353 of Tax1 are sufficient for Tax ubiquitination and that aa 114 to 220 of Tax2 prevent Tax1 ubiquitination.

Conversely, SUMOylation of both Tax2_1-113-1_114-353 and Tax2_1-220-1_221-353 was similar to that of Tax1 [Fig. 6C (b), lanes 4 and 5 versus lane 1], indicating that the domain spanning aa 221 to 353 of Tax1 is sufficient for Tax SUMOylation and that the same domain of Tax2 prevents Tax1 SUMOylation. Control Western blot analyses demonstrated similar levels of purification for all constructs [Fig. 6C (c)]. Control NF-κB-luc and HTLV-1-LTR-luc reporter gene assays performed in similar experimental settings confirmed the functionality of all chimeras (Fig. 6D and E). Efficient transcription from HTLV-1-LTR by Tax2 and chimeras that are similarly localized mainly in the cytoplasm (Tax1_1-90-2_91-113-1_114-353 and Tax2_1-113-1_114-353) might seem surprising. However, more than 15% of the cells still display a strong nuclear Tax signal in nuclear bodies (see Fig. 6B, frequencies on the right) and it is likely that more cells have a diffuse nuclear signal that is difficult to quantify and that could allow such an efficient transcription.

Together, these results show that (i) subcellular localization alone does not account for low levels of Tax2 ubiquitination and SUMOylation, (ii) the Tax1 domain encompassing lysines K4 to K10 is sufficient to mediate Tax ubiquitination, and (iii) the Tax1 domain encompassing lysines K6 to K10 is sufficient to mediate Tax SUMOylation. This demonstrates that the amino acid context of the targeted lysine residues critically determines the ubiquitination and SUMOylation statuses of each Tax protein.

DISCUSSION

Ubiquitination and SUMOylation of Tax1 and their impact on Tax1-mediated NF-κB activity and Tax1 localization have been extensively studied (21–23, 51, 52). Here we report that, in sharp contrast to Tax1, Tax2 is barely conjugated to endogenous ubiquitin and SUMO in cellulo. More importantly, a lysineless Tax2 mutant that is intrinsically nonubiquitinable and non-SUMOylable still significantly activates the NF-κB pathway. These findings demonstrate that Tax2-mediated NF-κB activation involves a mechanism independent of Tax conjugation to either ubiquitin or SUMO.

A previous report suggested, however, that, like Tax1, Tax2 is both ubiquitinated and SUMOylated (27). It is worth noting that those experiments were performed under conditions where ubiquitin or SUMO proteins were overexpressed. Nevertheless, differences in the pattern of Tax1 versus Tax2 ubiquitination and SUMOylation were still observed. In particular, lower levels of short modifications (monoubiquitin, diubiquitin, and di- and tri-SUMO chains) were retrieved for Tax2 than for Tax1 (27). Experiments using endogenous ubiquitin and SUMO proteins, which could thus prevent false-positive results, were not performed in this study, and the functional roles of those posttranslational modifications were not assessed. Consistent with the literature (60), we hypothesize that overexpressing ubiquitin leads to nonspecific, uncontrolled biases. This is the reason why we performed all our experiments using endogenous levels of both ubiquitin and SUMO proteins and, when possible, endogenous levels of Tax. In addition, all our experiments included Tax1 as a control, and our Tax1 results were consistent with the literature.

Tax1 and Tax2 share approximately 78% of amino acid sequence identity. In particular, the lysine residues targeted by ubiquitination and SUMOylation (K4 to K8 in Tax1) are conserved between the two proteins. This strongly suggests that the lack of detectable Tax2 ubiquitination and SUMOylation is not linked to the lack of targeted lysines but rather to the absence of other determinants, allowing the interaction with the modification machineries. This conclusion is supported by our data, showing that the Tax1 region encompassing the conserved ubiquitinated lysines (K4 to K10) confers ubiquitination to Tax2, while the Tax1 region encompassing the conserved SUMOylated lysines (K6 to K10) confers SUMOylation to Tax2. Hence, Tax1/Tax2 amino acid differences within these domains are likely to be responsible for the low level of Tax2 modification. Indeed, the aa 114-to-353 region of Tax1 and Tax2 displays 73 nonconserved amino acids (30%), while the aa 221-to-353 region displays 46 nonconserved amino acids (35%). Consistent with this model, we show here that overexpression of Ubc13 does not enhance Tax2 ubiquitination, contrary to Tax1, indicating that the effect of Ubc13 distinguishes between Tax1 and Tax2. Interestingly, the aa 221-to-353 domain of Tax1 has previously been shown to contain key motifs that distinguish between Tax1 and Tax2 in the context of the activation of the noncanonical NF-κB pathway (41), supporting the concept that cellular partners can differentiate Tax1 from Tax2. Whether observations from Shoji et al. are linked to Tax posttranslational modifications should be further investigated.

Several E3 ubiquitin ligase families have been identified, including proteins that recognize motifs containing phosphorylated residues (66). Interestingly, the aa 114-to-353 region of Tax1 that confers ubiquitination to Tax2 contains a series of serine and threonine residues that are not conserved in the Tax2 sequence. This notably includes S336 that is phosphorylated in Tax1 (67). In addition, the [SLTTGAL] (aa 199 to 205) sequence found in Tax1 but not in Tax2 fits with the consensus motif for phosphorylation by casein kinase 1 (CK1), a kinase shown to allow phosphorylation-dependent ubiquitination of substrates (68). Further investigations are needed to assess the role of these phosphorylated residues and other motifs of Tax in differential Tax1 and Tax2 ubiquitination.

SUMOylation was reported to mainly occur on consensus motifs such as ψKxD/E (where ψ is a large hydrophobic residue) (69). Such a consensus motif is not present in Tax1, although a PKD motif is located at aa positions 262 to 264, encompassing K6. Interestingly, this motif is changed into PKA in Tax2. Further experiments are needed to determine whether this slight change in amino acid sequence accounts for the differences observed between Tax1 and Tax2 SUMOylation.

Regulation of ubiquitination and SUMOylation was reported to be mediated by cross talks between posttranslational modifications. As an example, ubiquitination and SUMOylation may act sequentially (69). We describe here one chimera that is efficiently SUMOylated but weakly ubiquitinated. This result indicates that ubiquitination of Tax might not be a prerequisite for Tax SUMOylation. Whether SUMOylation is a prerequisite for Tax1 ubiquitination needs to be clarified.

Our results, based on endogenously expressed ubiquitin and SUMO proteins, clearly demonstrate that the functions of each Tax protein are regulated through distinct molecular mechanisms. Our results show that Tax2-mediated activation of NF-κB involves a mechanism independent of the lysine residues of Tax2 and therefore independent of Tax2 conjugation to ubiquitin and SUMO. Of note, we recently demonstrated that Tax1 SUMOylation, initially believed to be required for Tax1-mediated NF-κB activation, is indeed dispensable for this process (56). In this respect, Tax2 is similar to Tax1. From our results and those of Bonnet et al., it can therefore now be suggested that SUMOylation is not a key determinant for Tax-induced NF-κB activation. In contrast, Tax1 and Tax2 strongly differ by their requirement for ubiquitination, which is a key modification for Tax1 but is dispensable for Tax2. It is believed that Tax1 contains at least one IKKγ/NEMO-binding domain, i.e., the leucine-rich region located between aa 121 to 141 (70). However, Tax1 ubiquitination is also required for Tax binding to IKKγ/NEMO (21, 52). Our results show that a lysineless Tax2 mutant is still able to interact with IKKγ/NEMO. Hence, Tax2/IKKγ/NEMO interaction might either be direct, indicating that, in contrast to Tax1, the ubiquitin-independent Tax2 interaction with IKKγ/NEMO is sufficient, or be indirect. In the latter case, this could suggest that Tax2 recruits an ubiquitinated partner that then allows recruitment of IKKγ/NEMO. Our results show that TRAF6 is not one of those Tax2-binding proteins. The recent characterization of the Tax2 interactome (71) and of its ubiquitinated partners will be of major interest in this matter and might allow us to determine how Tax2-mediated activation of the NF-κB pathway is regulated.

NF-κB constitutive activation is a hallmark of ATL-transformed cells and of a number of cancer cells. Our results, showing that Tax2 does not require posttranslational modification to efficiently activate NF-κB, are therefore surprising. Whether the lack of requirement of Tax ubiquitination and SUMOylation for NF-κB activation functionally impacts the “quality” of the NF-κB response (in terms of gene activation profile) should also be investigated and might eventually lead to new insights into HTLV-1 versus HTLV-2 distinct pathological outcomes.

In conclusion, our study reveals significant new differences between Tax1 and Tax2 and favors the notion that the molecular determinants involved in the activation of the NF-κB pathway by Tax1 and Tax2 are not identical.