HTLV-1 Tax mutants that do not induce G1 arrest are disabled in activating the anaphase promoting complex

HTLV-1 Tax is a potent activator of viral transcription and NF-κB. Recent data indicate that Tax activates the anaphase promoting complex/cyclosome (APC/C) ahead of schedule, causing premature degradation of cyclin A, cyclin B1, securin, and Skp2. Premature loss of these mitotic regulators is accompanied by mitotic aberrations and leads to rapid senescence and cell cycle arrest in HeLa and S. cerevisiae cells. Tax-induced rapid senescence (tax-IRS) of HeLa cells is mediated primarily by a dramatic stabilization of p27KIP and is also accompanied by a great surge in the level of p21CIP1mRNA and protein. Deficiencies in p27KIP prevent Tax-IRS. A collection of tax point mutants that permit normal growth of S. cerevisiae have been isolated. Like wild-type tax, many of them (C23W, A108T, L159F, and L235F) transactivate both the HTLV-LTR and the NF-κB reporters. One of them, V19M, preferentially activates NF-κB, but is attenuated for LTR activation. None of the mutants significantly elevated the levels of p21CIP1and p27KIP1, indicating that the dramatic surge in p21CIP1/WAF1and p27KIP 1induced by Tax is brought about by a mechanism distinct from NF-κB or LTR activation. Importantly, the ability of these mutants to activate APC/C is attenuated or abrogated. These data indicate that Tax-induced rapid senescence is causally associated with APC/C activation.

Indeed, in tax-expressing HeLa, MT4, and S. cerevisiae cells, the levels of cyclin A, cyclin B and the anaphase inhibitor: securin/Pds1p (precocious dissociation of sister chromatids) were found to be significantly reduced [30]. We have found that the loss of cell cycle regulators and the mitotic defects induced by Tax may be causally linked and are associated with premature activation of the anaphase promoting complex/cyclosome (referred to as APC/C henceforth), a multiprotein E3-ubiquitin ligase that controls the onset of anaphase and mitotic exit by targeting mitotic cyclins and other cell cycle regulators for degradation [36]. More recently, we have shown that the cell cycle dysregulation induced by tax does not end with mitotic abnormalities. Tax-transduced HeLa cells, after passage through a faulty cell division cycle, immediately entered into a senescence-like G 1 arrest termed tax-induced rapid senescence, tax-IRS [37]. These cells expressed high levels of Cdk2 inhibitors: p21 CIP1/WAF1 and p27 KIP 1 as a consequence of Tax-mediated activation of p21 CIP1/WAF1 mRNA transcription, and increased stabilization of p21 CIP1/ WAF1 and p27 KIP 1 proteins. Consistent with these findings, Tripp et al have also reported that expression of tax can cause CD34+ hematopoietic cells to cease proliferation [38].
During normal cell cycle progression, p21 CIP1/WAF1 and p27 KIP 1 transiently accumulate during G 1 , but become degraded in S. The destruction of p21 CIP1/WAF1 and p27 KIP 1 during S phase is regulated by the multisubunit E3 ubiquitin ligase, SCF (Skp-Cullin-F box), together with its substrate-targeting subunit, Skp2 [39][40][41][42][43][44] and the cell cycle regulatory protein, Cks1 [39,44,45]. Recent evidence indicates that Skp2 and Cks1 are both substrates of the Cdh1associated APC/C (APC Cdh1 ). They become polyubiquitinated and degraded in late M and early G 1 when APC Cdh1 is highly active. This renders SCF SKP2 inactive and allows p21 CIP1/WAF1 and p27 KIP 1 levels to build up in G 1 . When tax is expressed, APC/C becomes prematurely activated. This causes Skp2 to be polyubiquitinated and degraded starting in S, through G 2 /M and in subsequent G 1 . The drastic reduction in Skp2 and possibly Cks1, apparently inactivated SCF SKP2 , profoundly stabilized p21 CIP1/WAF1 and p27 KIP1 , thereby committing cells to senescence. The stabilization and surge of p21 CIP1/WAF1 and p27 KIP 1 in taxexpressing cells, therefore, is temporally and causally linked to premature APC/C activation. In essence, Tax activates the cellular program for mitotic exit far ahead of schedule, thereby driving cells into a state of permanent arrest. Interestingly and as might be predicted, we have found that HTLV-1 transformed T-cells invariably express lower levels of p27 KIP1 . Indeed, a loss of p27 KIP 1 function allows cells to evade tax-IRS [37].
Our earlier results have indicated that expression of Tax in S. cerevisiae also leads to unscheduled, APC-mediated degradation of Clb2p and Pds1p, G 2 /M delay, chromosome aneuploidy, growth arrest, and loss of cell viability [30]. Considering the highly conserved nature of the cellular machineries that control mitosis in eukaryotes, this is probably not surprising. The powerful genetics available for S. cerevisiae provides an opportunity to dissect the mechanism by which Tax dysregulates APC/C and mitosis, which is otherwise difficult to perform in human cells.
Here we report the isolation of a collection of 26 tax point mutants whose expression in S. cerevisiae did not lead to growth arrest. Five mutants (V19M, C23W, A108T, L159F, and L235F) -with amino acid substitutions that span the majority of Tax protein sequence -were chosen for indepth analyses. C23W, A108T, L159F, and L235F transactivated both the HTLV-LTR and the NF-κB reporters. One mutant, V19M, preferentially activated NF-κB, but was attenuated in LTR activation. All became impaired or abrogated in their ability (i) to activate APC, (ii) to increase the levels of p21 CIP1/WAF1 and p27 KIP1 , and (iii) to cause tax-IRS. These data strongly suggest that tax-IRS, with the associated mitotic aberrations and the accompanying rise in p21 CIP1/WAF1 and p27 KIP 1 levels, is coupled to APC/C activation, and is mechanistically unrelated to the CREB/ATF-CBP/p300 or IKK-NF-κB pathway.

Isolation of tax mutants that do not cause growth arrest in S. cerevisiae
In the course of a yeast 2-hybrid screen using Tax as bait, we noticed that the yeast strain expressing the lexA-tax fusion grew significantly slower than the lexA control. This prompted us to examine more closely the effect Tax exerts on the growth and proliferation of S. cerevisiae. To this end, W303a, a standard laboratory yeast strain, was transformed with pRS315-Gal10-Tax, a CEN plasmid carrying the tax gene under the control of a galactose inducible promoter [30]. As reported previously [30], expression of tax after galactose induction lead to a cessation of cell growth and proliferation. Upon Tax expression, the W303a/Gal10-Tax cells initially suffered a delay in S/G 2 / M progression [30]. They then became arrested at G 1 phase of the cell cycle. The growth-arrested cells became greatly enlarged in size, but were without buds and displayed severe DNA aneuploidy [30]. Their viability was also significantly decreased. These results immediately suggest that tax mutants that do not cause growth and proliferation arrest may be readily isolated in S. cerevisiae and these mutants may have similar or identical properties in human cells. To isolate tax mutants impaired in causing growth arrest, we mutated pRS315-Gal10-Tax by hydroxylamine (36). W303a cells were then transformed with the pool of chemically mutated plasmid preparation, and plated to select for galactose-resistant transformants. The colonies on galactose plates were then screened by colony dot blots for tax expression using a mouse hybridoma Tax antibody, 4C5. A total of 26 non-growth-arrest tax mutants were identified, for which the respective pRS315 vectors were isolated. We next confirmed loss of the growth arrest phenotype by reintroducing plasmids carrying the mutant tax alleles into W303a. As expected, all W303a expressing mutant tax alleles readily grew on both glucose and galactose plates (Fig. 1B), while W303a expressing the wild-type tax failed to grow on galactose plates as previously described [30].
DNA sequence analyses of the tax coding sequence revealed that each of the 26 tax mutants contained a single amino acid substitution that resulted from a G to A or C to T transition, as might be expected for hydroxylamine mutagenesis. The altered amino acid residues in the Tax protein sequence are listed in Fig. 1A. Many of the amino acid substitutions are clustered in the NH 2 -terminal half of Tax (20/26). Consistent with the notion that the amino acid substitutions had occurred in important regions of Tax, we noticed that the T130I substitution overlap with the dual amino acid substitutions -T130A L131S -in a well characterized tax mutant known as M22, which is partially defective in dimerization and is severely impaired in IKKγ/NEMO-binding and NF-κB activation. Two distinct mutations (G61E and G61R) and (A108T and A108V) were isolated for each of the amino acid residues 61 and 108, suggesting the importance of these residues in protein-protein interactions that mediate Tax functions. Finally, the expression levels of all mutants in S. cerevisiae were comparable as judged by immunoblotting (data not shown).

Tax mutants selected in W303a are functional in HTLV-1 LTR and NF-κB trans-activation
Next, we investigated the biological activities of tax mutants in mammalian cells. Mutant tax alleles were cloned into a lentiviral vector, HR'CMV-SV40-puro. This vector allowed tax to be expressed transiently from the CMV immediate early promoter after DNA transfection or stably after lentivirus vector-mediated gene transduction. We first examined the ability of the Tax mutants to transcriptionally activate luciferase reporters driven respectively by the HTLV-1-LTR (LTR-Luc) and the NF-κBinducible E-selectin-promoter (E-selec-Luc) [46]. Twenty one mutants were analyzed by luciferase reporter assays (Fig. 2). The other five mutants (V24E, C36Y, G61R, P92L, and L183F) were excluded from the reporter assays because of either the drastic amino acid alterations caused by the mutations or the existence of alternative amino acid substitution in the same position. Approximately half of the mutants analyzed (S32F, A47T, H52Y, G61E, L75F, T145I, W147L, P169L, A285T, and S300F) were greatly impaired in both transactivation functions of Tax. Many of these mutations are in the highly conserved NH 2terminus of Tax. Because of their severe defects, no attempts have been made to determine if their expression in 293T cells may be altered by the respective amino acid substitutions. The levels of expression of these mutants in S. cerevisiae were normal, however. By contrast, several mutants (C23W, P102L, A108T, A108V, H127T, L159F, 235F, G245D, and D264Y) continued to transactivate both LTR and NF-κB reporters to levels (greater than 50%) comparable to those of the wild-type Tax. Of note, V19M was specifically impaired in LTR activation but remained a potent NF-κB activator, while T130I was defective in NF-κB activation, but exhibited significant LTR activation capability, reminiscent of similar properties of the M22 (T130A L131S) mutation mentioned above. As indicated below, at least five of these mutants (V19M, C23W, A108T, L159F, and L235F) expressed at levels comparable to that of the wild-type Tax in HeLa cells (see below). These results indicate that mutations that impaired the ability of Tax to arrest growth of W303a cells did not necessarily affect LTR or NF-κB transactivation. Finally, although the reporter assays for some of the mutants varied more than others, the variations occurred mostly due to strong transactivation; and importantly, the LTR and NF-κB transactivations by the 5 mutants that were analyzed in depth have been confirmed by independent methods (Fig. 3 and see below). We infer from these data that the growth arrest phenotype of Tax most likely involves interactions with a cellular process distinct from the CREB/CBP/p300 and the IKK/NF-κB pathways.

Expression of tax mutants in HeLa cells
We next selected 5 mutants (V19M, C23W, A108T, L159F, and L235F) that retained the ability to transactivate LTR and/or NF-κB for further analysis. Lentivirus vectors (LV) capable of transducing the mutant tax alleles were generated by co-transfection of the respective HR'-CMV-tax-SV40-puro vectors together with packaging plasmids that encode HIV structural proteins and VSV G protein as previously reported [37,47]. A stable HeLa cell line, HeLa-18 × 21-DsRed, which expresses DsRed under the control of a Tax-inducible enhancer/promoter cassette containing 18 copies of the 21-bp repeat upstream of a minimal HTLV-1 promoter [48], was used as the cellular background for introducing the tax alleles. As the expression of DsRed in HeLa-18 × 21-DsRed is strictly Tax-dependent, cells that express Tax after gene transduction can be readily detected by fluorescence microscopy (Fig. 3). HeLa-18 × 21-DsRed cells were infected with LV carrying the wildtype, V19M, C23W, A108T, L159F, L235F tax alleles, or the EGFP gene. The LV-transduced cells were then selected in media containing 1 μg/ml puromycin for 2-3 days. Drug-resistant colonies were then grown in puromycinfree medium for 1 day and observed under a fluorescence microscope for DsRed expression. In agreement with the LTR-Luc reporter activities described above (Fig. 2), C23W, A108T, L159F, and L235F, but not V19M activated DsRed expression (Fig. 3B). As expected, HeLa-18 × 21-DsRed transduced with the LV-EGFP control did not express DsRed. Previously, we have demonstrated that Tax expression in HeLa cells greatly elevated the levels of p21 CIP1/WAF1 and p27 KIP 1 cyclin-dependent kinase inhibitors, thereby causing HeLa cells to enter into a senescencelike G 1 arrest termed Tax-induced rapid senescence (Tax-IRS) [37]. The HeLa cells in Tax-IRS are flat, enlarged, vacuolated, often binucleated, and stained positive for the senescence associated β-galactosidase. Indeed, in agreement with previous results, microscopic examination of the HeLa-18 × 21-DsRed cell line transduced with LV-Tax (wild-type) revealed a prevalence of enlarged and binucleated cells, consistent with the notion that they were in the state of Tax-IRS (Fig. 3A). By contrast, the morphology of cells transduced with mutant tax alleles, with the exception of A108T, resembled those of control cells transduced with the EGFP gene. Finally, we noted that despite some similarity of A108T cells to Tax (wild-type) cells, the extent of arrest and morphological changes in A108T cells appeared to be attenuated (Fig. 3A).

Tax mutants whose expression is permissible in S. cerevisiae do not cause, or are attenuated in inducing cell cycle arrest in HeLa cells, but remain functional in activating I-κB degradation and p100 processing
To characterize the various tax mutants further, we analyzed the LV-mutant-tax-transduced cells by flow cytometry. Three days after puromycin selection, asynchronously grown LV-transduced cells were transferred to puromycinfree medium for 24 h and harvested for analyses. As anticipated from the cell morphology in Fig. 3A, most cells that expressed the wild-type Tax (75%) appeared in the G 1 phase of the cell cycle (Fig. 4). In contrast, G 1 populations for cells transduced with the various tax mutant alleles were significantly lower, albeit somewhat higher than that of the EGFP-transduced control (Fig. 4). These results support the notion that those tax mutants that failed to cause growth arrest in S. cerevisiae are also significantly disabled or attenuated in inducing senescence/cell cycle arrest in mammalian cells, albeit with varying degrees of attenuation.
The phenotypes of the tax mutants were not due to variations in the levels of Tax protein expression as indicated by immunoblotting (Fig. 5). In accordance with the overall cell morphology and flow cytometry analyses, the levels of p21 CIP1/WAF1 and p27 KIP 1 in the various taxtransduced cells correlated with their extent of growth arrest or lack thereof, with wild-type tax greatly increasing (A) A summary of amino acid substitutions in HTLV-1 tax point mutants whose expression did not cause growth arrest in S. cerevisiae the levels of p21 CIP1/WAF1 and p27 KIP1 , followed by the A108T mutant, and with the remaining mutants having only moderate to no effect (C23W, L159F, L235F, V19M). As might be expected, the levels of cyclin B1 in the respective cell lines inversely correlated with the growth characteristics of the respective cells. Likewise, the levels of Skp2 in the transduced cells also correlated with their respective cyclin B1 levels. Finally, consistent with the ability of the tax mutants to transactivate the E-seletin-Luc NF-κB reporter (Fig. 2), the levels of I-κBα in the tax-transduced cells were reduced, while those of p52, the mature NF-κB2, were increased. Here again, in general agreement with the reporter assays, the A108T mutant is equivalent or possibly better than the wild-type tax in inducing I-κBα degradation and p52 NF-κB processing, second by V19M and C23W, followed lastly by L235F and L159 mutants. While for some mutants (A108T, L159F, and L235F) there appears to be some correlation between the severity of cell cycle arrest/senescence phenotype and the degree of NF-κB activation, for others such as V19M and C23W, that are strong NF-κB activators, the senescence phenotype was significantly attenuated. These results support the notion that the Tax-induced cell cycle arrest/rapid senescence (Tax-IRS) and increase in p21 CIP1/WAF1 and p27 KIP 1 levels are causally related and do not involve directly either the CREB/CBP/p300 or the IKK/NF-κB pathway. Whether the IKK/NF-κB pathway may share a common Tax-targeted regulatory factor with the cell cycle/APC pathway remains to be seen.

S. cerevisiae-viable tax mutants are attenuated in APC/C activation
We have shown previously that the mitotic abnormalities and rapid senescence that Tax induces in S. cerevisiae and HeLa cells are associated with unscheduled activation of the anaphase promoting complex and the premature degradation of mitotic/cell cycle regulators including cyclin A, Clb2/cyclin B, Pds1/securin, and Skp2 [30]. The levels of cyclin B1, Skp2, p21 CIP1/WAF1 and p27 KIP 1 in the HeLa cells expressing the various tax alleles suggest that the S. cerevisiae-viable tax mutants are impaired in APC/C activation. To determine the effect of the tax mutants on APC/C directly, we introduced them into a yeast strain, KY630, which contains a chromosomally integrated HA-CLB2 at the CLB2 locus. As anticipated, upon induction of tax expression for 2 h, a reduction of Clb2p in cells expressing wild-type tax was observed compared to the tax-null control. By contrast, the S. cerevisiae-viable V19M, C23W, A108T, L159F, and L235F tax mutants were attenuated in causing Clb2p reduction/degradation (Fig. 6A). We have shown previously that Tax activates the anaphase promoting complex in S. cerevisiae, HeLa, 293T and HTLV-1 trans- formed T cells [36]. For the ease of transfection, the wildtype and mutant tax alleles were individually co-transfected with HA-tagged ubiquitin into 293T cells. Indeed, the five tax mutants were also found to be attenuated in inducing cyclin B1 polyubiquitination when compared to the wild-type control (Fig. 6B). The extents of attenuation of the five tax mutants in S. crevevisiae versus 293T cells were not exactly identical (compare Fig. 6A and 6B). This may reflect subtle structural differences between S. cerevisiae and human anaphase promoting complex. Finally, the degrees of the mutants to cause cyclin B1 polyubiquitination (A108T ≥ C23W > L159F, L235F, and V19M) cor-related largely with the levels of p21 CIP1/WAF1 and p27 KIP 1 increase in the transduced cells (A108T > C23W > L159F, L235F, and V19M). Taken together, these results support the idea that unscheduled activation of the anaphase promoting complex is responsible for the Tax-induced rapid senescence/cessation of cell proliferation in both human and S. cerevisiae cells.

Discussion
In this study, we have described a collection of 26 tax single-point mutants that are disabled in causing cell cycle arrest in S. cerevisiae. A total of 21 tax alleles were analyzed further by luciferase reporter assays for LTR and NF-κB transactivation. Approximately half of the amino acid substitutions likely have impacted on critical regions of Tax so as to render it severely defective. Special attention was directed to five mutants (V19M, C23W, A108T, L159F, and L235F) that remained strong LTR and NF-κB transactivators. Their phenotypes in HeLa cells were largely consistent with those seen in S. cerevisiae -all were attenuated or significantly impaired in causing p21 CIP1/ WAF1 and p27 KIP 1 accumulation, but remained able to induce I-κB degradation and p100 NF-κB2 processing. Whereas the majority of HeLa cells transduced with wildtype tax entered into tax-IRS, cells transduced with each of the 5 mutant tax alleles continued to proliferate, albeit at rates that varied dependent on the levels of p21 CIP1/ WAF1 and p27 KIP 1 expressed. These results validated the utility of S. cerevisiae as a model for dissecting the mitotic abnormalities and rapid senescence/cell cycle arrest induced by Tax.
The levels of p21 CIP1/WAF1 and p27 KIP 1 are regulated through transcription, phosphorylation (by cyclinE/ Cdk2), subcellular localization, ubiquitination, and proteasome-mediated degradation [40,49,50]. The E3 ubiquitin ligase, SCF, together with its substrate-recognition subunit, Skp2, mediates the ubiquitination and degradation of p21 CIP1/WAF1 and p27 KIP 1 [39,44,45]. The level of Skp2 oscillates in a cell cycle-dependent manner. Recent data have indicated that Skp2 and another SCF subunit, Cks1, are substrates of the Cdh1-associated APC/C (APC Cdh1 ) [39,44]. Both become ubiquitinated and degraded in late M and early G 1 when APC Cdh1 is highly active. This inactivates SCF and allows p21 CIP1/WAF1 and p27 KIP 1 to accumulate transiently in G 1 . We have shown recently that in HeLa cells transduced with tax, early APC/ C activation sets in motion premature loss of cyclin A, cyclin B, securin, and Skp2, and causes a dramatic build-up of p21 CIP1/WAF1 and p27 KIP 1 during S phase. After an aberrant mitotic division cycle, the great surge in p21 CIP1/ WAF1 and p27 KIP 1 in tax-expressing cells then commits cells into a state of irreversible cell cycle arrest. Results from the present analysis, i.e. mutations in tax that abrogated induction/stabilization of p21 CIP1/WAF1 and p27 KIP 1 also disabled APC/C activation, are in agreement with that conclusion.
Activation of p21 CIP1/WAF1 mRNA transcription by Tax has been reported previously [23,[51][52][53]. Because a Tax mutant, M47, which is deficient in LTR activation, became disabled in activating p21 CIP1/WAF1 promoter, some of these earlier studies have suggested that Tax-induced increase in p21 CIP1/WAF1 resulted from transactivation via the CREB/ATF-CBP/p300 pathway [23,53]. Since four of the five mutants analyzed here activated LTR-luciferase reporter at levels (70-80%) comparable to that of the wild-type tax, yet were substantially impaired in causing p21 CIP1/WAF1 and p27 KIP 1 increase, we conclude that the CREB/ATF-CBP/p300 pathway is most likely not the principal determinant in the accumulation of p21 CIP1/WAF1 and p27 KIP1 . Likewise, several of the mutants -A108T, V19M and C23W, in particular -are potent activators of IKK-NF-κB as indicated by luciferase reporter assays and the extent of I-κB degradation and p100 processing. These mutants are nevertheless impaired in elevating p21 CIP1/WAF1 and p27 KIP 1 levels. These results support the notion that the NF-κB pathway is not directly responsible for the Taxinduced increase in p21 CIP1/WAF1 and p27 KIP1 . Our earlier data suggest that a major factor for Tax-induced p21 CIP1/ WAF1 increase is correlated with premature APC/C activation [37]. Therefore, promoter transactivation by Tax may only contribute moderately to the overall build-up of p21 CIP1/WAF1 protein. This would explain the discrepancy between the data reported here and the earlier studies which relied heavily on p21 CIP1/WAF1 promoter-luciferase reporter assays. Finally, we have shown that via a tripartite interaction, Tax, PP2A and IKKγ form a stable ternary complex wherein PP2A activity is inhibited or diminished S. cerevisiae-viable tax mutants are attenuated in inducing cell cycle arrest/senescence in HeLa cells by Tax. In essence, PP2A inhibition by IKKγ-bound Tax maintains IKK in an active, phosphorylated state, causing constitutive phosphorylation and degradation of I-κB, which, in turn, leads to potent activation of genes under NF-κB/Rel control. Since PP2A regulates many critical cellular processes, it is conceivable that APC/C activation by Tax is also mediated through an inhibition of PP2A. In this sense, a tax mutant deficient in PP2A interaction will be disabled for both NF-κB and APC/C activation, but may continue to transactivate LTR. Several tax mutants, H43Q, K85N, and M22 (T130A, L131S) have been shown previously to be disabled for PP2A binding [18]. Both H43Q and M22 are deficient in NF-κB but not in LTR transactivation, while K85N is defective for both. Preliminary analyses suggest that these three mutants are also impaired in inducing cell cycle arrest. Mutants like H43Q and M22, with the possible exception of T130I, however, are not highly represented in the current collection. This may be due, in part, to the importance of the PP2A-binding domain of Tax in mediating other critical protein-protein interaction.
The biological/virological relevance for the profound cell cycle arrest induced by Tax remains unclear. It is possible that HTLV-1-infected T-cells that are arrested in a senescence-like state may persist longer in vivo and may, in this condition, be co-opted to devote significant cellular resources to virus replication. Alternatively, the dramatic morphological changes associated with the senescencelike arrest may facilitate virus assembly and/or transmission. Many of the tax mutants characterized here that remain functional in transactivating viral LTR and NF-κB can be incorporated into an infectious molecular clone of HTLV-1 to address this question.

Tax mutagenesis and selection for tax mutants
The CEN plasmid, pRS315 Gal10-Tax, that contains tax under the control of a galactose-inducible promoter derived from Gal10 gene has been previously described [30]. To introduce mutations into tax sequence, pRS315 Gal10-Tax plasmid DNA was exposed to hydroxylamine (1 mg/ml) overnight at 37°C [54]. The plasmid was then purified using a PCR purification kit (Qiagen), and used (1 μg) to transform S. cerevisiae. The transformants were plated on leucine-dropout plates that contained galactose as the sole carbon source. Colonies that appeared on galactose plates were then picked and seeded in grids on galactose plates, transferred to nitrocellulose filters, lysed, and screened for Tax expression using a monoclonal antibody (4C5) against Tax. Only positive clones were chosen for plasmid extraction/isolation. The plasmids extracted from W303-1a were than used to transform competent E. coli for DNA preparation and sequence analysis.

Cloning mutant tax alleles into a lentiviral vector
The lentivirus vector, LV-Tax-SV-Puro, which contains the wild-type tax gene under the control of the CMV enhancer promoter and the puromycin-resistance gene expressed from the SV40 enhancer/promoter (SV-Puro), has been previously reported [37]. A mutant tax allele, M47, which carries a diagnostic BglII restriction site in the tax coding sequence, was cloned into LV-Tax-SV-Puro via the BamHI (located immediately upstream of the translational initiation codon) and SmaI (downstream of the M47 mutations) restriction sites to generate LV-M47-SV-Puro. Most mutant tax alleles were cloned into LV-M47-SV-puro similarly except that an internal MluI site and an XmaI site, located at the aforementioned SmaI site, were used. The recombinants were identified by a loss of the diagnostic BglII site from the recombinant, and confirmed by DNA sequence analysis. For NH 2 -terminal mutations that lie upstream of the MluI site, DNA fragments harboring the mutations were generated by PCR and cloned into LV-Tax-SV-puro via the BamHI and MluI sites. Primers used to amplify the NH 2 terminal coding region of Tax are 5'TaxBamHI 5'-CGCGGATCCGCCACCATGGCCCACT-TCCCAGGGTT-3' (with the translational start site under-Immunoblot analyses of HeLa cells transduced with wild-type or mutant tax alleles Figure 5 Immunoblot analyses of HeLa cells transduced with wild-type or mutant tax alleles. HeLa cells transduced with the wild-type or the respective mutant tax allele were harvested. Cell lysates were prepared, resolved by SDS-12% PAGE, and probed with antibodies against Tax, cyclin B1, I-κBα, NF-κB, p52, p27 KIP1 , p21 CIP1/WAF1 , Skp2, and actin, respectively. S. cerevisiae-viable tax mutants are attenuated in APC/C activation Figure 6 S. cerevisiae-viable tax mutants are attenuated in APC/C activation. (A) KY630 cells carrying a gene encoding CLB2-3XHA integrated at the CLB2 locus were transformed with either Gal10-wild-type tax or Gal10-mutant tax. KY630/Gal10-tax cells were cultured in SC medium containing 2% raffinose at 30°C overnight for 12 h, then diluted to 0.3 A 600 , grown to midlog phase, and induced for tax expression by the addition of 2% galactose. Cell lysates were prepared after 120 minutes. Immunoblots were carried out with anti-HA and 4C5 monoclonal antibodies. (B) 293T cells were transfected with a plasmid encoding HA-tagged human ubiquitin together with an expression plasmid for wild-type tax or mutant tax. The cells were arrested at the G 1 /S border by a single thymidine treatment, released into complete DMEM containing 10 μM MG132 for 5 h, immunoprecipitated with cyclin B1 antibody and immunoblotted with an HA antibody to detect polyubiquitinated cyclin B1.

Construction of the Tax-inducible reporter, 18 × 21-DsRed and derivation of the 18 × 21-DsRed indicator cell line
A highly Tax-inducible enhancer/promoter cassette that contains 18 copies of the Tax responsive 21 bp repeat element upstream of a minimal HTLV-1 promoter (18 × 21) has been reported previously [48]. A blunt-ended BamHI fragment containing the 18 × 21 cassette was inserted upstream of the DsRed reporter gene in the pDsRed2-C1 (Stratagene) plasmid (blunt-ended at AseI and AgeI sites) to make p18 × 21DsRed-Neo. A HeLa reporter cell line for Tax was derived by transfecting cells with the p18 × 21-DsRed-Neo plasmid, followed by G418 (1 μg/ml, Invitrogen) selection. G418-resistant clones were expanded and transduced with the HR'CMV-Tax-puro lentiviral vector and observed for DsRed expression. One clone that had low basal expression but high DsRed expression in the presence of Tax was chosen for this study.

DNA transfection and luciferase reporter assay
Approximately 10 5 293T cells/well in a 12-well plate were transfected with the expression construct for each of the tax alleles (0.5 μg) together with either HTLV-LTR-Luc (0.1 μg) or E-selectin-Luc reporter plasmid (0.5 μg) using a calcium phosphate transfection kit (Invitrogen). Forty eight hours after transfection, cells were lysed using 250 μl of reporter lysis butter and 20 μl lysate from each transfection was used for the luciferase assay. After injection of 100 ul luciferase substrate (Promega), the luciferase activity was measured by a MLX microtiter plate luminometer. Transactivation functions of V19M, C23W, A108T, L159F, and L235F mutants were further confirmed by including in the transfection mixture 0.5 μg of a control plasmid, pRL-TK, that contains the renilla luciferase reporter gene driven by the herpesvirus thymidine kinase promoter.

Lentiviral vector production and gene transduction
Lentiviral vectors (LV) were produced as previously described after transfection of 293T cells [37]. Culture supernatants were harvested at 24, 48 and 72 h after transfection, pooled, filtered, aliquoted, and stored at -80 °C. Viral titers were measured by adding serially diluted LV stocks to 2 × 10 5 (HeLa 18 × 21-DsRed-Neo) cells were seeded in a 24-well plate. Polybrene (8 μg/ml, Sigma) was added to the medium together with the vector stocks to facilitate infection. Three days post-transduction, the number of RFP-positive cells in each well was counted as a measure of viral titer. To transduce tax alleles, 2 × 10 5 HeLa cells were first plated in a 6-well plate similar as in [37]. They were then transduced with LV (m.o.i. = 2) the next day. After 24 h, the medium was removed and fresh DMEM containing puromycin (1 μg/ml, Sigma) was added. The selection medium and cell debris were removed 48 h after selection by 1× PBS washes, and replaced with fresh puromycin-free DMEM.

Cell cycle analysis of HeLa cells transduced with tax alleles
HeLa cells transduced with LV containing either the wildtype or each of the mutants tax alleles were selected with puromycin and maintained as described above. Cells were harvested 4-5 days post-transduction for flow cytometry as previously described [36].

Detection of polyubiquitinated cyclin B
HEK 293T cells were co-transfected with HA-tagged ubiquitin and the various tax alleles as above. Cells are washed the next day and harvested 48 h later for immunoprecipitation using the cyclin B1 antibody (Santa Cruz Biotechnology) as previously described [36] except that RIPA buffer (Upstate protocol Tris-HCl, NP40 1%, Na-deoxycholate 0.25%, NaCl 150 mM, EDTA 1 mM, PMSF 1 mM, protease inhibitor cocktail (1 μg/ml), Na 3 VO 4 1 mM, and NaF 1 mM) was used instead of the lysis buffer previously described. Immunoblots were carried out with anti-HA (Santa Cruz), anti-cyclin B1 (Santa Cruz Biotechnology) and 4C5 (Tax) monoclonal antibodies.

Authors' contributions
RM and C-ZG were responsible for the design of the study and the draft of the manuscript. RM performed most of the experiments. CC isolated the tax mutants and assisted in the reporter assays. SH assisted in the analyses of S. cerevisiae-derived tax mutants. LZ and ML contributed the 18x21-EGFP reporter construct. YLK assisted in the cell cycle analyses. All authors read and approved the final manuscript.