Simian virus 40 large T antigen targets the microtubule-stabilizing protein TACC2 Free

Cellular transformation is essential for progression toward tumor development. The proteins involved in this process have been identified as oncogene products and tumor suppressors. Simian virus 40 (SV40), a polyomavirus of rhesus macaque origin, is a DNA tumor virus that can induce tumors in rodents and transform many types of cultured mammalian cells, including those of human origin (Ahuja et al., 2005; Fanning and Knippers, 1992; Poulin and DeCaprio, 2006). The SV40-encoded replication protein, designated large T antigen and referred to as T antigen, is a viral oncoprotein that modulates diverse cellular activities in genome integrity and cell-cycle regulation, thereby promoting the early steps of oncogenic transformation. Although many studies have reported that T antigen inactivates p53 and the RB family of tumor suppressors, it is suggested that other regulatory proteins are required for SV40-induced cellular transformation (Chang et al., 1997; Moens et al., 2007; Sachsenmeier and Pipas, 2001).

Polyomaviruses are highly specific to their host species but encode a similar T antigen (Ahuja et al., 2005; Chang et al., 1997; Fanning and Knippers, 1992; Moens et al., 2007; Poulin and DeCaprio, 2006; Sachsenmeier and Pipas, 2001). In humans, JC polyomavirus and BK polyomavirus were discovered to be responsible for progressive multifocal encephalopathy and renal nephropathy, respectively (Imperiale, 2000; Moens et al., 2007; Poulin and DeCaprio, 2006). Analogous to SV40, both viruses encode T antigens that transform the cells in culture and promote tumor formation in animals (Allander et al., 2007; Moens et al., 2007; Poulin and DeCaprio, 2006). Additionally, two human members, WU polyomavirus and KI polyomavirus, were found from respiratory tract specimens (Allander et al., 2007; Gaynor et al., 2007). More recently, Feng et al. described a new virus, Merkel cell polyomavirus, the genome of which is integrated into cellular DNA in the aggressive human skin cancers (Feng et al., 2008). These lines of evidence raise the essential question of how their oncoproteins function in the cells. In addition to the known targets of T antigen, this protein was recently reported to interact with Bub1, which has dual roles in spindle assembly checkpoint and chromosome congression (Cotsiki et al., 2004; Hein et al., 2009). Thus, SV40 T antigen might target as-yet unidentified cellular proteins to cause cell transformation and oncogenesis (Chang et al., 1997; Moens et al., 2007; Sachsenmeier and Pipas, 2001; Woods et al., 1994).

During our investigations, we found that T antigen interacts with transforming acidic coiled-coil (TACC) protein 2, designated TACC2. There are at least three TACC family proteins (TACC1, TACC2 and TACC3), all of which contain the conserved TACC domain and have been possibly implicated in tumorigenesis (Gergely, 2002; Raff, 2002). TACC1 and TACC3 were found to be overexpressed in human cancers, and overexpression of TACC1 induced the transformation of primary mouse cells in culture (Still et al., 1999a; Still et al., 1999b). By contrast, TACC2 was downregulated as breast tumors became more malignant (Chen et al., 2000). Overexpression of TACC2 in these cells reverted malignant phenotypes to benign phenotypes both in vivo and in vitro, suggesting that TACC2 might function as a potential tumor suppressor.

Here, we report that human TACC2 has a crucial role in microtubule stabilization in cultured cells, and that T antigen binds TACC2 for inducing mitotic defects and chromosomal instability. In addition to changes in chromosomes, T antigen, or the loss of TACC2 function, causes abnormalities in nuclear structural after cell division. Our findings suggest that TACC2 is a key target of T antigen for promoting mitotic defects leading to abnormal chromosomal and nuclear inheritance. These findings provide the first evidence that a viral oncoprotein can directly disrupt microtubule regulation, and shed light on the molecular basis of the initiation of cellular transformation.

To test the effect of T antigen on cellular function independently of p53 and RB, we introduced GFP-fused T antigen in HeLa cells that lacked these functional proteins but stably transmitted chromosomes during cell division (Fig. 1A). Abnormally nucleated cells, which had multiple nuclei and/or micronuclei, were increased by about fourfold at 72-96 hours after transfection, compared with control cells. This result suggests that T antigen might rapidly target other cellular factors rather than p53 and RB. In addition, we found the similar effects of T antigen in rodent CHO cells (supplementary material Fig. S1A, left). Abnormally nucleated CHO cells promptly augmented at 48 hours after transfection of T antigen, suggesting that these alterations induced by T antigen are unlikely to depend on cell type.

To identify novel factors that interact with T antigen, we performed yeast two-hybrid screening using the region containing amino acids 250 to 708 of T antigen as bait (Fig. 1B). From a screening of approximately 7×10⁶ independent transformants of 17-day-old mouse embryo cDNA libraries, we isolated four cDNA clones encoding the C-terminal TACC domain of TACC2 (amino acids 788-996; Fig. 1C). At least three members of the TACC protein family (TACC1, TACC2 and TACC3) have been identified in human, and are differently implicated in naturally occurring cancers (Gergely, 2002; Raff, 2002). To determine a direct interaction between T antigen and TACC2, we prepared glutathione S-transferase (GST)-fused T antigen(250-708) and His-tagged TACC domains of human TACC proteins, and subjected them to an in vitro pull-down analysis (Fig. 1D). The TACC domain of TACC2, but not those of TACC1 or TACC3, bound T antigen(250-708). The interaction of TACC2 with T antigen(250-708) was comparable to that of p53, which is known to bind the bipartite region of T antigen. To further visualize the subcellular locations of T antigen and TACC2, we carried out an immunofluorescence analysis in HeLa cells. As previously reported (Gergely, 2002; Gergely et al., 2003), TACC2 was diffusely distributed in the interphase nuclei and at the centrosomes (supplementary material Fig. S1B). During cell division, TACC2 was concentrated at the centrosomes and mitotic spindles. Simultaneous visualization showed that a portion of T antigen coexists with TACC2 in mitotic spindles (Fig. 1E) and possibly in interphase nuclei (supplementary material Fig. S1C). These data suggest that T antigen directly interacts with TACC2.

To address the function of TACC2 in human cells, we performed knockdown of endogenous TACC2 using specific short interfering RNAs (siRNAs; Fig. 2A). Western blot analysis revealed the effectiveness of TACC2 knockdown at 72 hours after transfection. In addition, immunofluorescence signals for this protein were mostly lost in both interphase and mitosis (Fig. 2A). TACC2 mRNA was undetectable at 24 hours after siRNA transfection, whereas TACC1 and TACC3 were not affected in the TACC2-depleted cells (data not shown). Similar to the observations in T-antigen-expressing cells (Fig. 1A), abnormally nucleated cells were significantly increased at 72 hours after TACC2 knockdown in HeLa (Fig. 2B) and CHO cells (supplementary material Fig. S1A, right), suggesting that loss of TACC2 induces mitotic defects. To test whether the occurrence of abnormally nucleated cells is due to mitotic spindle defects, we then detected centrosomes and mitotic chromosomes during cell division. TACC2 knockdown was frequently found to cause incompletely stretched bipolar and monopolar spindles (Fig. 2C, upper and lower left panels, respectively). Very similar results were obtained in T-antigen-expressing cells (Fig. 2C, right). To quantify these data, we measured the distances between two centrosomes in mitosis (Fig. 2D). Compared with control cells, the frequencies of short inter-centrosomal distances (<7 μm) were increased by more than twofold in TACC2 knockdown and T-antigen-expressing cells. These results suggest that the functional consequences of T antigen introduction or TACC2 depletion are correlated with defects of proper spindle formation.

There are at least two possible mechanisms for incompletely stretched bipolar and monopolar spindle formation. One is failure in the formation of stable microtubules, and the other is defects in the microtubule-associated motor proteins that drive the repulsive force between the poles (Sharp et al., 2000). To test the mitotic activity of TACC2, we used an inhibitor of kinesin motor protein Eg5, named monastrol, which blocks centrosome separation in mitosis without affecting microtubule formation (Kapoor et al., 2000). We then examined specific microtubule lengths based on the diameters of β-tubulin fluorescent zones around mono-asters. Measurement of the largest diameters of the monopolar spindles revealed that TACC2-depleted HeLa cells had shorter asters than control cells (Fig. 3A), suggesting that TACC2 is involved in stable microtubule formation. To further examine whether TACC2 enhances microtubule elongation, we performed a microtubule regrowth assay (Fig. 3B). Briefly, cells were chilled on ice to depolymerize the mitotic microtubules, returned to 37°C for 1 to 3 minutes to allow microtubule recovery, and then fixed for immunofluorescence analysis to assess the microtubule organization. Control cells produced robust microtubule asters in 1 minute, and nascent spindles were detected after 3 minutes, under the regrowth conditions (Fig. 3B, upper). By contrast, there was little aster formation in TACC2 knockdown cells, since only short and disorganized microtubules were detected even after 3 minutes under the regrowth conditions (Fig. 3B, lower). These results suggest that microtubule stabilization depends on TACC2 function.

To clarify the consequences of TACC2 knockdown, we visualized microtubules relative to centrosomes and mitotic chromosomes (Fig. 3C). Compared with control cells, the mitotic spindles (detected by staining for β-tubulin) and centrosome position relative to chromosomes were disorganized in TACC2 knockdown cells. We then checked whether TACC2 knockdown affects progression in mitosis (Fig. 3D). The cells were presynchronized, treated with nocodazole and then released into metaphase in the presence of a proteasome inhibitor, MG132, which inhibits the metaphase-anaphase transition. Both T-antigen-expressing cells and TACC2-knockdown cells progressed slowly and were delayed at prometaphase, while most control cells entered metaphase. Furthermore, expression of TACC2(864-989), which has an inhibitory effect on cellular TACC2, induced similar mitotic abnormalities (supplementary material Figs S2 and S3). Exogenously expressed full-length TACC2 or the TACC domain of this protein was previously reported to show polymer formation in the perinuclear cytoplasmic region (Gergely et al; 2000a; Lee et al., 2001). In agreement with these data, overexpressed TACC2(788-996) was aggregated in the perinuclear region (supplementary material Fig. S4, right). During our analysis of exogenously expressed TACC2, we found that FLAG-tagged TACC2(864-989), lacking a short stretch of the coiled-coil sequences, showed the same localization as endogenous TACC2 (supplementary material Fig. S3A). Rabbit polyclonal antibodies against TACC2, designated anti-TACC antibodies, recognized endogenous TACC2 and exogenously expressed TACC2(788-996), but not TACC2(864-989) (Gergely et al., 2000a) (supplementary material Fig. S2A,B). Interestingly, overexpression of FLAG-TACC2(864-989) disrupted the centrosomal localization of endogenous TACC2 (supplementary material Fig. S3B), suggesting that TACC2(864-989) has a dominant-negative effect on cellular TACC2. The localizations of TACC1 and TACC3 were not affected by TACC2(864-989) (data not shown).

To confirm the specificity of the mitotic defects induced by loss of TACC2 function, we expressed TACC2(864-989) and examined its effects on mitotic function in HeLa cells. Expression of TACC2(864-989) induced mitotic abnormalities (supplementary material Fig. S3C,D), similar to T-antigen-expressing or TACC2-knockdown cells (Fig. 1A, Fig. 2B). TACC2(864-989)-expressing cells exhibited a significantly higher incidence of abnormal nuclei (supplementary material Fig. S3C). Furthermore, we observed accumulation of prometaphase cells expressing TACC2(864-989) (supplementary material Fig. S3D). The proportion of prometaphase cells among mitotic cells increased by twofold, following functional inhibition of TACC2. Collectively, these results suggest that inactivation of TACC2 by T antigen or the dominant-negative form of TACC2 induces mitotic abnormalities.

TACC domain-containing proteins in other eukaryotic systems have been shown to recruit the microtubule-associated proteins chTOG (colonic and hepatic tumor-overexpressing gene) homologues to centrosomes and mitotic spindles (Kinoshita et al., 2005; Lee et al., 2001). In HeLa cell, endogenous chTOG was detected in the centrosomes and mitotic spindles (supplementary material Fig. S4A, left), and was recruited to exogenously expressed TACC domain of TACC2 protein (supplementary material Fig. S4A, right). However, the localization of chTOG was maintained even in TACC2-depleted unstable mitotic spindles (supplementary material Fig. S4B), suggesting that TACC2 stabilizes microtubule independently of chTOG (Cassimeris and Morabito, 2004; Gergely et al., 2003; Holmfeldt et al., 2004). In addition, TACC2 knockdown affected microtubule organization to some degree in interphase cells (supplementary material Fig. S2C), suggesting that TACC2 might stabilize microtubules throughout the cell cycle.

To prove that the interaction between T antigen and TACC2 is biologically important, we tested whether the effect of T antigen on microtubule formation is suppressed by overexpressing TACC2. As previously reported (Gergely et al., 2000; Lee et al., 2001), the exogenous expression of full-length TACC2 or TACC2(788-996) induced cytoplasmic aggregate formation (supplementary material Fig. S4A). Since FLAG-TACC2(864-989) had the proper localization and did not cause cytoplasmic aggregates (supplementary material Fig. S3A), the use of TACC2Δ(788-863), full-length TACC2 lacking this small region, enabled us to perform the rescue experiment. GFP-fused TACC2Δ(788-863) was localized to mitotic spindle, similar to endogenous TACC2 (Fig. 4A). To demonstrate a functional relationship between T antigen and TACC2, we performed a microtubule elongation assay using monastrol-treated HeLa cells. As shown in Fig. 3A, the ability of TACC2 to promote microtubule assembly can be quantitatively assessed by measuring the diameters of the monopolar spindles. T-antigen-expressing cells had shorter asters than control cells (Fig. 4B), as did the TACC2-depleted cells (Fig. 3A), providing the evidence that T antigen abrogates microtubule assembly. Importantly, the effect of T antigen on microtubule formation was suppressed by co-expressing TACC2Δ(788-863) to the control levels (Fig. 4B). The expression of TACC2Δ(788-863) alone did not increase the diameter of the spindles (data not shown). We further confirmed that TACC2Δ(788-863) interacted with T antigen in the cells, using an immunoprecipitation analysis (supplementary material Fig. S4C). Therefore, based on the interaction between T antigen and TACC2, these data suggest that TACC2Δ(788-863) rescues T antigen-induced short or unstable microtubules.

We then checked the direct binding of TACC2 to various T antigen mutants (supplementary material Fig. S5A). Maltose-binding protein (MBP) pull-down experiments revealed that ΔC1 (amino acids 1-487) and ΔC2 (amino acids 1-250) as well as full-length T antigen bound the TACC domain of TACC2. For this reason, we had difficulties in narrowing down the region of T antigen that interacts with TACC2 and in generating a mutant of T antigen that is specifically defective for TACC2 binding. Since mutation in the region containing amino acids 89-97 of T antigen caused the loss of binding to mitotic checkpoint protein Bub1 (Cotsiki et al., 2004), we used N-terminally deleted T antigen (amino acids 250-708), which can interact with TACC2 but not Bub1 (Fig. 5A). Approximately 20% of control HeLa cells were normally in prometaphase during mitosis, whereas expression of T antigen markedly increased the number of delayed prometaphase cells. Similar to the case for TACC2 knockdown, T antigen(250-708) comparably increased such prometaphase cells. In addition, we found that a part of T antigen(250-708)-expressing cells or TACC2 knockdown cells tended to undergo apoptosis. To determine whether the increase in delayed prometaphase cells is due to short or unstable microtubules, we checked the elongation of monastrol-induced monopolar spindles in T-antigen(250-708)-expressing cells. Expression of T-antigen(250-708) resulted in the formation of shorter spindles (supplementary material Fig. S5B), similar to the effect of TACC2 knockdown or T antigen expression (Fig. 3A, Fig. 4B). Thus, the T antigen(250-708) that preferentially interacts with TACC2 can induce mitotic abnormalities.

To further investigate the dynamics in cell-cycle progression, we carried out a time-lapse imaging analysis of HeLa cells stably expressing GFP-fused histone H2B (Fig. 5B). In control cells, the time between prometaphase onset and anaphase entry was approximately 31±9 minutes. By contrast, TACC2 knockdown cells and T-antigen(250-708)-expressing cells exhibited unusually slow progression of prometaphase (53±23 minutes and 50±21 minutes, respectively). Cells expressing full-length T antigen spent about 39±12 minutes on the transition from prometaphase to anaphase entry (data not shown), suggesting that full-length T antigen expression does not cause significant mitotic delays, compared with TACC2 knockdown or T antigen(250-708) expression, probably because of the failure of the mitotic checkpoint. Importantly, TACC2 knockdown or T antigen(250-708) expression induced chromosome missegregation, leading to the appearance of lagging chromosomes (white arrows in Fig. 5B) and bi-nucleated cells (red arrows in Fig. 5B); the latter probably arose because of incomplete cytokinesis. Analysis of more than 10 live images for each cell type (control cells, TACC2 knockdown cells, T antigen(250-708)-expressing cells and T antigen-expressing cells) indicated that T antigen(250-798) caused similar mitotic defects to those induced by TACC2 knockdown (data not shown). Characteristic movements of mitotic chromosomes were observed as follows: (1) condensed chromosomes were aligned on the metaphase plate, segregated once to opposite poles, then collapsed to one pole and repeatedly moved as a pendulum between the poles (Fig. 5B, middle panel); and (2) after alignment of most chromosomes on the metaphase plate, they were spread and aligned on the metaphase plate again, followed by their segregation in the presence of lagging chromosomes (Fig. 5B, lower panel). Furthermore, it has been reported that multiple chromosomal aberrations are observed in T-antigen-transformed cells (Chang et al., 1997; Ray et al., 1990; Stewart and Bacchetti, 1991; Woods et al., 1994), and that nuclear organization is modified in many cancers and transformed cells (Zaidi et al., 2007). To elucidate the consequence of the mitotic errors, we examined whether TACC2 knockdown and T antigen affect chromosome transmission during mitosis and nuclear organization in interphase (supplementary material Fig. S6). We prepared chromosome spreads from HeLa cells and counted the number of chromosomes per cell (supplementary material Fig. S6A). Control cells stably maintained chromosome numbers (2n), suggesting that chromosomal inheritance during mitosis is regulated. By contrast, TACC2 knockdown cells exhibited not only the normal complement of chromosomes (2n) but also some cells had 4n, together with loss or gain of a few chromosomes, indicating that TACC2-depleted cells have unstable chromosome transmission. In addition, T-antigen-expressing cells showed similar patterns of chromosomal changes, together with an increase in hypodiploid chromosomes (<2n). The effect of T antigen might be underestimated probably because of various expression levels of this protein in the cells studied. These data suggest that TACC2 dysfunction induces chromosomal instability.

Abnormally nucleated cells as a consequence of mitotic defects were characterized by the appearance of multiple nuclei and/or micronuclei (Fig. 1A, Fig. 2B and supplementary material Fig. S3C). Since nuclear organization is modified in many cancers and transformed cells (Zaidi et al., 2007), we examined whether T antigen affects nuclear organization in interphase (supplementary material Fig. S6B). Immunofluorescence analysis was performed using specific antibodies against various nuclear structural markers. Interestingly, in T-antigen-expressing cells the nuclear envelope became irregular and had multiple segments, compared with control cells. We further found that the nuclear envelope was discontinuous in asymmetrically divided micronuclei, suggesting that such incomplete nuclear envelope formation may cause loss of DNA content, thereby leading to aneuploid karyotypes. Similar results were found in TACC2 knockdown cells (data not shown). Taken together, interaction of T antigen with TACC2 may promote abnormal mitotic defects as a result of microtubule dysfunction, leading to chromosomal instability and nuclear structural disorganization.

To finally demonstrate complex formation by T antigen and TACC2 in vivo, we performed an immunoprecipitation analysis in HeLa cells (Fig. 5C). FLAG-tagged T antigen was expressed in the cells and immunoprecipitated by anti-TACC2 and anti-FLAG antibodies. Western blot analysis revealed that endogenous TACC2 was present in immunoprecipitates with FLAG-T antigen, but absent from the control. T antigen was also detected in immunoprecipitates with TACC2, suggesting that T antigen complexes with TACC2. Collectively, our data show that (1) T antigen directly interacts with TACC2, (2) T-antigen-induced microtubule destabilization is suppressed by overexpressing TACC2Δ(788-863), and (3) mitotic defects are caused by T antigen(250-708), which minimally interacts with TACC2, suggest the possibility that T antigen inhibits TACC2 function, leading to mitotic errors.

This study has identified the centrosome and mitotic spindle-localizing protein TACC2 as a novel T antigen target. Our data suggest that, besides known anti-tumor proteins such as p53 and RB, T antigen selectively interacts with TACC2 to disrupt mitotic spindle formation possibly toward cell transformation. These findings represent the first evidence that a viral oncoprotein can directly target the mitotic spindle apparatus to induce chromosome instability, and that TACC2 is involved in stabilizing microtubules in mitosis. Perturbation of p53 and RB was reported to be required, but not sufficient, for the transforming activity of T antigen (Sachsenmeier and Pipas, 2001). In addition, inhibition of RB and Bub1 by T antigen was reported to induce mitotic defects and chromosomal instability (Cotsiki et al., 2004; Quartin et al., 1994; Sotillo et al., 2007; Woods et al., 1994). Our study has further revealed that T antigen(250-708), which does not interact with these proteins, can induce mitotic defects including delayed progression through mitosis, suggesting that inhibition of TACC2 by T antigen causes mitotic defects independently of RB and Bub1. Thus, impairment of both TACC2 and Bub1 by T antigen might synergistically enhance mitotic defects and avoid mitotic checkpoint response.

The chromosomal abnormalities such as aneuploidy and polyploidy are common characteristics of naturally occurring human cancers, which result from incorrect cell division (Draviam et al., 2004; Kops et al., 2005; Weaver and Cleveland, 2006). Recently, several lines of evidence have indicated that chromosomal instability is a crucial early event during carcinogenesis and might be a cause, rather than a consequence, of cellular transformation (Weaver and Cleveland, 2006). Although this idea is still under debate, T antigens of polyomaviruses are likely to provide a clue in improving our understanding of chromosomal instability. The infected oncoviruses might actively disturb the mitotic spindle formation of the host cells, possibly leading to the increase of cell death. However, chromosomal instability induced by T antigen might produce some transformed cells from the small numbers of surviving cells as the benefit to further viral replication.

Human TACC family proteins have been implicated in cancer (Gergely, 2002; Raff, 2002). Although knockout of TACC2 in mice showed no obvious phenotypes (Schuendeln et al., 2004), this might not contradict our data that abrupt loss of TACC2 induced mitotic abnormalities in somatic cells. There could be developmental complements by other cellular factors in TACC2-deficient mice. Furthermore, TACC2 is likely to be the sole target for T antigen among the TACC family members. In addition, we have found functional differences between human TACC2 and TACC domain-containing proteins in other species. The TACC proteins in Caenorhabditis, Drosophila and Xenopus were reported to stabilize microtubules by recruiting ZYG9, Msps and XMAP215, respectively, which are homologues of human chTOG, to centrosomes and mitotic spindles (Gergely et al., 2003; Lee et al., 2001; Peset et al., 2005). XMAP215 can bind microtubules and enhance microtubule growth at the plus end (Kinoshita et al., 2005). However, human chTOG was able to localize to centrosomes and mitotic spindles in TACC2-depleted cells but failed to stabilize the microtubules (supplementary material Fig. S4B). In human cells, chTOG is likely to be recruited by redundant functions of TACC1 and TACC3. TACC2 might stabilize microtubule formation, independently of chTOG. Since T antigen forms a complex and coexists with TACC2 at mitotic spindles (Fig. 1), we speculate that TACC2 stabilizes microtubule formation together with as-yet unknown regulatory proteins, and that T antigen inhibits their cooperation. Our data in Fig. 1A and Fig. 2B showed that either the expression of T antigen or the knockdown of TACC2 induced abnormally nucleated cell phenotypes. In comparison with these conditions, the combination of T antigen expression and TACC2 knockdown tended to have a somewhat additive effect on the occurrence of the abnormal nucleation (supplementary material Fig. S7). In addition, the data in Fig. 4B showed that overexpression of TACC2Δ(788-863) rescued T antigen-induced short or unstable microtubules, although we have not found that T antigen-induced abnormal nucleation was suppressed by TACC2Δ(788-863) (data not shown), probably because of the involvement of multiple factors in the abnormally nucleated cell phenotype. Moreover, we have found that TACC2 contributes to microtubule stabilization throughout the cell cycle (supplementary material Fig. S4D). Recent studies have reported that TACC2 might be involved in transcriptional control, via associations with histone acetyltransferases and chromatin remodeling factors (Gangisetty et al., 2004). Our data do not exclude the possibility that T antigen affects the function of TACC2 in interphase as well as mitosis. Taken together, our findings provide insights into the molecular basis of the chromosomal instability and nuclear structural disorganization promoted by T antigen.

Yeast strain AH109 carrying pAS2-1-T antigen (amino acids 250-708) was transformed with mouse E17 whole embryo cDNA libraries constructed in pACT2 (Clontech). Plasmids harboring cDNA were recovered from both histidine- and adenine-positive colonies and subjected to DNA sequencing.

cDNAs encoding T antigen and TACC2 were cloned into pcDNA3 to create the following plasmids: pcDNA3-FLAG-T antigen, pcDNA3-FLAG-TACC2(788-996), pcDNA3-FLAG-TACC2(864-989), pcDNA3-EGFP-T antigen, pcDNA3-EGFP-T antigen(250-708), pcDNA3-EGFP-TACC2(788-996), pcDNA3-EGFP-TACC2(864-989) and pcDNA3-EGFP-TACC2Δ(788-863).

HeLa and CHO cells were cultured in a 1:1 mixture of Dulbecco's modified Eagle's minimum essential medium and Ham's F-12 nutrient medium (Sigma) supplemented with 10% (v/v) heat-inactivated fetal bovine serum.

Cells were transfected with plasmid DNAs using FuGENE 6 and FuGENE HD (Roche Applied Science). For siRNA experiments, cells were transfected with siRNA duplex oligonucleotides using RNAiMAX (Invitrogen) or co-transfected with siRNA duplex oligonucleotides and pcDNA3-EGFP using Lipofectamine 2000 (Invitrogen). Knockdown cells were analyzed 72 hours after each transfection. The siRNA duplexes used were designed to target the mRNAs encoding human and Chinese hamster TACC2. The sequences of the siRNAs (Japan Bioservice) were as follows: human TACC2, 5′-AGGACGCUGGAGCAGAAGA-3′ and Chinese hamster TACC2, 5′-AGAACGCUGGAGCAGAAGA-3′. The results were confirmed using more than two siRNA duplex oligonucleotides against different sequences. siRNA against firefly luciferase GL3 was used as a control, as described previously (Saitoh et al., 2006).

Human cDNAs encoding TACC1 (amino acids 605-805), TACC2 (amino acids 788-996) and TACC3 (amino acids 638-838) were cloned into pET28a (Novagen). The cDNA encoding TACC2(788-996) was also cloned into pMAL-c2X (New England Biolabs). A cDNA encoding T antigen(250-708) was cloned into pGEX4T-1 (Amersham Bioscience). Bacterial expressions of His-TACC2(788-996), MBP-TACC2(788-996), His-p53 and GST-T antigen(250-708) were similarly carried out.

Anti-human TACC2 polyclonal antibodies were generated by immunizing a rabbit with His-tagged TACC2(788-996), and affinity purified using TACC2 coupled to HiTrap Protein G HP (GE Healthcare). Rabbit anti-human TACC1, TACC2 and TACC3 antibodies, and rabbit anti-chTOG antibodies were described previously (Gergely et al., 2000; Kinoshita et al., 2005). Other antibodies were obtained from the following sources: mouse anti-T antigen (BD Biosciences PharMingen), mouse anti-FLAG (M5; Sigma), mouse anti-His (Qiagen), mouse anti-GST (DAKO), mouse anti-β-tubulin (Sigma), mouse anti-γ-tubulin (Sigma), rabbit anti-γ-tubulin (Santa Cruz Biotechnology), and mouse anti-lamin A/C and rabbit anti-RanBP2 antibodies (Saitoh et al., 2006).

For immunoprecipitation experiments, cells were lysed with RIPA buffer [50 mM Tris-HCl pH 7.5, 50 mM NaCl, 1.5 mM MgCl₂, 2.5 mM ZnCl₂, 0.5% Triton X-100, 10% (v/v) glycerol] supplemented with 1 mM Na₃VO₄ and protease inhibitors for 30 minutes on ice. After centrifugation for 30 minutes, the supernatant was collected, incubated with specific antibodies for 1 hour at 4°C, mixed with 20 μl of protein A/G-agarose beads (Amersham Bioscience) and further incubated for 1 hour. The beads were washed, and bound proteins were detected by western blot analysis. For GST pull-down assays, bacterially expressed GST and GST-fused T antigen (1 μg) were immobilized on glutathione-agarose beads and incubated with His-tagged TACC2 proteins (1 μg) and His-tagged p53 protein in a buffer consisting of 0.1% Triton X-100, 50 mM Hepes pH 7.5, 50 mM NaCl, 5% (v/v) glycerol, 1 mM dithiothreitol and protease inhibitors for 1 hour at 4°C. The beads were washed, and bound proteins were detected by western blot analysis. Maltose-binding protein (MBP) pull-down assays were carried out in a similar manner, except that the supernatant was incubated with bacterially expressed MBP-fusion proteins (1 μg) immobilized on Amylose resin (New England Biolabs).

Cells were fixed with methanol at –20°C for 3 minutes, washed with PBS, blocked with 0.5% bovine serum albumin in PBS, incubated with specific primary antibodies for 1 hour at room temperature, and then incubated with appropriate secondary antibodies for 1 hour. The following secondary antibodies were used: Alexa-Fluor-488-conjugated donkey anti-mouse IgG (Molecular Probes); Alexa-Fluor-488-conjugated donkey anti-rabbit IgG (Molecular Probes); Cy3-conjugated donkey anti-mouse IgG (Jackson ImmunoResearch); and Cy3-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch). Cells were counterstained with DAPI (1 mg/ml) before mounting. For immunofluorescence experiments analyzing nuclear structure, cells were fixed with 4% paraformaldehyde for 15 minutes at room temperature. Images were acquired with a Model IX71 microscope (Olympus) and analyzed using Lumina Vision (Mitani Corporation).

For microtubule regrowth assays, cells were synchronized at mitosis by double thymidine blocks and released, placed on ice for 30 minutes to depolymerize the microtubules and then incubated in warm medium (37°C) for 1, 2 or 3 minutes to allow microtubule recovery (Petretti et al., 2006). The cells were fixed with methanol at –20°C and stained with anti-β-tubulin antibodies. For monastrol (an inhibitor of kinesin motor protein Eg5) treatment, TACC2-depleted and T-antigen(250-708)-expressing cells were synchronized by double thymidine blocks. At 6 hours after release of the blocks, 100 μM monastrol was added and the cells were incubated for a further 6 hours.

HeLa cells stably expressing GFP-tagged histone H2B were grown on 35-mm glass-based dishes (IWAKI) (Marumoto et al., 2003). Fluorescence and DIC time-lapse analyses were performed using a microscope (Model XI70; Olympus). The camera, shutters and filter wheel were controlled by MetaMorph imaging software (Universal Imaging), and images were collected every 5 minutes with 50-msecond exposures. Through-focus z-series stacks consisting of three frames were acquired at each time point.