In this study, we directly analyzed the role of Tax nuclear bodies and Tax SUMOylation on NF-κB activation.
Previous studies have described that Tax forms nuclear spots called Tax nuclear bodies or Tax speckled structures [16, 17]. A Tax speckled structure localization signal (TSLS) positioned between residues 50 and 75 was subsequently mapped in the Tax sequence . In order to identify new functional motifs of Tax, we selected the PxQxT motif at position 79–83 because it fits with a TRAF-binding motif and is adjacent to the TSLS. We found that this motif does not control Tax interaction with the TRAF since its mutation alters neither Tax ubiquitination (this study) nor the co-precipitation of Tax with either TRAF2 or TRAF5 (data not shown). In contrast, mutations of the P79 and Q81 residues dramatically reduce the formation of Tax nuclear bodies, confirming the role of the TSLS and showing that this sequence includes the PxQ motif.
We and others have demonstrated that Tax ubiquitination, especially conjugation to K63-linked ubiquitin chains, permits Tax binding to NEMO [24, 30, 32]. Here we report that Tax-P79AQ81A, which is conjugated to either total ubiquitin or K63-linked ubiquitin chains at the same level than wt Tax, binds to NEMO and activates IKK like wt Tax. Moreover, analysis of a series of Tax mutants allowed us to show that the amount of endogenous phospho-IKKα/β coprecipitated with Tax correlates with the level of Tax ubiquitination but not SUMOylation. These findings confirm the critical role of Tax ubiquitination in NEMO binding and IKK activation and also demonstrate that Tax SUMOylation is dispensable for these processes. Tax conjugation to K63-linked ubiquitin chains was also shown to relocalize Tax and NEMO to perinuclear spots [32, 33]. However, recent findings showed that NEMO targeting to perinuclear spots is also impaired upon siRNA-mediated SUMO silencing . Using both microscopy analysis and cell fractionation, we found that endogenous NEMO is recruited in perinuclear spots at the same level in cells producing either wt Tax or Tax-P79AQ81A. Hence, lowering Tax SUMOylation has no effect on NEMO targeting to cytoplasmic spots. The effect of SUMO silencing could be explained by the role of a minor fraction of SUMOylated Tax or of a SUMOylated unknown substrate in the cytoplasmic targeting of NEMO.
Tax nuclear bodies were previously reported to contain RelA and NEMO and therefore identified as transcriptionally active structures [16, 19, 20]. Surprisingly, we found that lack of nuclear body formation by Tax-P79AQ81A does not prevent this mutant from activating both the cytoplasmic and nuclear steps of the NF-κB pathway. Importantly, such proper NF-κB promoter activation was observed not only in adherent cells but also in T cells, notably CD4+ primary T cells. The composition of Tax nuclear bodies has essentially been studied in transfected adherent cells [15–17], and we were indeed able to detect these structures in all transfected cells, including T cells. However, using the same staining procedure, we observed that Tax nuclear bodies are nearly absent in HTLV-1-infected T cells. Hence, Tax nuclear bodies appear to be visible when Tax is transiently produced but not in an endogenous situation. This may suggest that nuclear bodies represent a storage compartment rather than transcriptionally active structures. However, it cannot be excluded that small clusters of Tax, undetectable by confocal microscopy, are indeed formed in HTLV-1-infected T cells. Further investigations are therefore needed to clarify the pattern and role of nuclear Tax, in particular in HTLV-1-infected CD4+ T cells.
Our work also allowed us to further explore the relationship between Tax ubiquitination and Tax SUMOylation. Indeed, as mentioned above, Tax-P79AQ81A mutant whose endogenous SUMOylation is reduced by around 80% is ubiquitinated at the same level as wt Tax, both in terms of total ubiquitination and specific conjugation to K63-linked ubiquitin chains. This strongly suggests that Tax SUMOylation is dispensable for Tax ubiquitination. That Tax SUMOylation may represent a signal for Tax ubiquitination was indeed recently proposed based on findings showing that RNF4, a SUMO-targeted ubiquitin ligase (STUbL), was able to modulate Tax ubiquitination . It was shown that RNF4 induced the ubiquitination of a SUMO-1 fused recombinant Tax protein in vitro and that siRNA-mediated depletion of RNF4 abolished Tax ubiquitination. However, we found here that the SUMO-1 fused form of Tax was ubiquitinated at comparable level as non-fused Tax in HeLa cells. Moreover, we show that in contrast to RNF4 depletion, low Tax SUMOylation does not prevent Tax ubiquitination in cells. Of note, a GFP-tagged Tax was used in the RNF4 study  while our experiments were performed using a Tax-6his construct, which could lead to difference in Tax modifications and/or localization. In addition, it cannot be excluded that the low residual level of SUMOylation of Tax-P79AQ81A could be still sufficient to promote Tax ubiquitination. However, this would likely have been associated to a certain degree of reduction of Tax ubiquitination, as observed in RNF4-depleted cells . Along with these findings, our data suggest therefore that RNF4 may not directly modulate wild-type Tax ubiquitination, but acts in an indirect manner by interfering with ubiquitination machineries or with direct regulators of Tax ubiquitination.
We previously concluded that ubiquitination and SUMOylation were both required for optimal NF-κB activation by Tax through analysis of lysine mutants and SUMO-1-fused proteins. In this study, we revisited the role of Tax SUMOylation through a direct approach based on an ubiquitinated but intrinsically weakly SUMOylated Tax mutant. We found that Tax-P79AQ81A retains most of the NF-κB activity of wt Tax, while in the same conditions, very little NF-κB activity was measured for a mutant defective for both ubiquitination and SUMOylation (Tax-K4-8R). Potent NF-κB activation by Tax-P79AQ81A was not only observed in 293 T cells and HeLa cells but also in T cells. Indeed, Tax-P79AQ81A is as active as wt Tax in CEM T cells and more importantly, in primary CD4+ T lymphocytes. Of note, study of another mutant, K7R, confirms that weak SUMOylation does not preclude NF-κB activation. Finally, we were able to document that the abilities to activate a NF-κB promoter of a series of Tax mutants correlate with their levels of ubiquitination but not of SUMOylation. It could be argue that low level of SUMOylation would be sufficient to regulate some Tax activities, even in absence of Tax nuclear bodies. However, our findings provide strong evidence that Tax SUMOylation is not a key determinant for Tax-induced NF-κB activation.
That low SUMOylation does not alter Tax-induced NF-κB activation appears to contradict our previous findings showing that fusion of SUMO-1 to lysine mutants restored their NF-κB activities. However, we believe that our current and earlier data can be reconciled in light of recent findings from other groups. Indeed, we and others previously noticed that the SUMO-1 fusion restores the NF-κB activity of some but not all lysine Tax mutants: i.e., it restores the NF-κB activities of Tax-K6-8R or Tax-K7-8R (also referred to as Tax-R4-6 K), which retains partial ubiquitination, but not that of Tax-K4-8R, which is no longer ubiquitinated [20, 24]. Moreover, we documented that fusion of SUMO-1 does not increase the NF-κB activity of either wt Tax  or Tax-P79AQ81A (data not shown), which are both fully ubiquitinated. Fusion of SUMO-1 appears, therefore, to only enhance the NF-κB activity of partially but not fully ubiquitinated Tax proteins. As mentioned above, RNF4 was recently shown to induce the ubiquitination of a recombinant GFP-Tax protein fused to SUMO-1 in vitro. Hence, artificial fusion of SUMO-1 to the Tax mutants may similarly enhance their ubiquitination by facilitating the interaction with RNF4. Strikingly, it was previously reported that fusion of SUMO-1 increases the endogenous ubiquitination of Tax-K7-8R/Tax-R4-6 K but not Tax-K4-8R . Furthermore, we show here that fusion of SUMO-1 stabilizes both Tax and Tax-P79AQ81A in the cytosol but only significantly increases ubiquitination of Tax-P79AQ81A. This latter effect is reminiscent of the cytoplasmic relocalization and enhanced ubiquitination of the GFP-tagged Tax construct upon RNF4 overexpression . Why RNF4 would only increase the ubiquitination of Tax mutants but not wt Tax remains to be elucidated. Because these proteins differ by the presence or absence of SUMOylation, it could be speculated that the natural SUMO chains of Tax may somehow prevent the interaction of wt Tax or Tax-SUMO-1 with RNF4. Whatever the exact mechanism involved, these findings support the view that the effect previously attributed to the fusion of SUMO-1 to lysine Tax mutants was actually linked to Tax stabilization and/or facilitation of Tax ubiquitination.