Oncogene (2014) 33, 4242–4252 & 2014 Macmillan Publishers Limited All rights reserved 0950-9232/14 www.nature.com/onc

ORIGINAL ARTICLE A small compound targeting TACC3 revealed its different spatiotemporal contributions for spindle assembly in cells

RYao1, Y Kondoh2, Y Natsume1, H Yamanaka1, M Inoue1,3,HToki3, R Takagi2, T Shimizu2, T Yamori4, H Osada2 and T Noda1,3

The mitotic spindle is assembled by the coordinated action of and kinetochore microtubules. An evolutionally conserved family, transforming acidic coiled-coil (TACC), has been shown to be involved in this process. In humans, TACC3 is aberrantly expressed in a variety of human , but its biological significance remains to be elucidated. Here, using a novel compound targeting TACC3, spindlactone (SPL), we show that the perturbation of TACC3 selectively inhibited the nucleation of microtubules in ovarian cancer cells. In contrast to centrosome microtubules, the kinetochore microtubules were robustly assembled, forming ectopic spindle poles that resulted in multipolar spindles. Interestingly, the extensive inhibition of TACC3 partially suppressed the nucleation of kinetochore microtubules. These dose-dependent effects of SPL were consistent with the results observed by the depletion of TACC3 and its binding partner, colonic and hepatic tumor overexpressed protein (TOGp). Although these both have roles in the assembly of centrosome and kinetochore microtubules, their contributions were spatiotemporally different. Notably, SPL did not affect spindle assembly in normal cells. Furthermore, the oral administration of SPL significantly suppressed tumor growth in vivo. The unique mechanism of action of SPL not only enables it to be used as a tool to dissect the molecular basis of spindle assembly but also to provide a rationale for the use of TACC3 as a molecular target for cancer treatment. This rationale offers an opportunity to develop new strategies for cancer chemotherapy that overcome the limitations of microtubule toxins and expand their scope and clinical efficacy.

Oncogene (2014) 33, 4242–4252; doi:10.1038/onc.2013.382; published online 30 September 2013 Keywords: TACC3; TOGp; spindle assembly

INTRODUCTION one of the sister kinetochores on a .7 Although this The mitotic spindle is a well-established molecular target for ‘search and capture’ mechanism appears to be the predominant cancer chemotherapy, and clinically, drugs targeting microtubules means by which become attached and aligned on are used to treat hematological and solid malignancies. In tumor the spindle, mathematical modeling studies have indicated that cells, the anti-mitotic effects of these drugs are thought to result unbiased search and capture alone is not sufficient to align all the from the suppression of the dynamic instability of microtubules, chromosomes on the mitotic spindle within a normal timeframe.8 which is required for spindle assembly during . However, Growing evidence indicates that in addition to centrosomes, these drugs also inhibit microtubule function in non-dividing cells, chromosomes also drive spindle assembly. This chromosome- which results in serious side effects, such as peripheral neuro- driven mitotic spindle formation was originally thought to function pathy. In addition, resistance to microtubule-targeted drugs is only in meiosis, a process in which centrosomes are not present. acquired by the overexpression of specific tubulin isoforms.1 To However, subsequent studies have indicated that this process overcome these problems, new therapeutic agents are being works cooperatively with centrosome microtubules to assemble developed to target the mitotic proteins involved in spindle bipolar mitotic spindles in somatic cells.9–11 In these systems, assembly rather than the tubulin itself.2 Among these mitotic kinetochore microtubules support effective spindle assembly by proteins, the motor protein KIF-11, which is also called Eg5; the guiding centrosome microtubule growth toward the kinetochores. Aurora kinases; and the Polo-like kinases have been extensively However, the direct evaluation of kinetochore microtubules in studied as potential drug targets. Although the antitumor somatic cells is hampered by the mass of centrosome microtubules activities of these agents have been demonstrated both in vitro that rapidly grow after nuclear envelope breakdown (NEB). Thus, and in xenograft models,3–5 the scope of their clinical efficacy the relative contributions of centrosome and kinetochore appears to be limited.6 Thus, novel therapeutic targets that microtubules to spindle assembly remain to be elucidated. selectively perturb spindle microtubules without affecting Human transforming acidic coiled-coil (TACC3) was originally microtubule functions in post-mitotic cells are beneficial to the identified within a translocation breakpoint region that was development of effective therapeutics for cancer treatment. associated with multiple myelomas,12 and subsequent studies In vertebrate somatic cells, spindle microtubule nucleation occurs have indicated that it is aberrantly expressed in various cancers.13–16 at the centrosome, which contains plus ends that dynamically grow TACC proteins are evolutionally conserved, from yeasts to humans. and shrink, probing space until they are captured and stabilized by In Drosophila, a microtubule-associated protein, termed msps, was

1Department of Cell Biology, Cancer Institute, The Japanese Foundation for Cancer Research, Tokyo, Japan; 2Chemical Biology Core Facility, RIKEN Advanced Science Institute, Saitama, Japan; 3Team for the Advanced Development and Evaluation of Human Disease Models, Bioresource Center, RIKEN, Tsukuba, Japan and 4Division of Molecular Pharmacology, Cancer Chemotherapy Center, The Japanese Foundation for Cancer Research, Tokyo, Japan. Correspondence: Dr R Yao, Department of Cell Biology, Cancer Institute, The Japanese Foundation for Cancer Research, 3-8-31, Ariake, Koto-ku, Tokyo 135-8550, Japan. E-mail: [email protected] Received 11 March 2013; revised 17 July 2013; accepted 9 August 2013; published online 30 September 2013 Spindle aberration by the TACC3 inhibitor R Yao et al 4243 identified as a binding partner of the D-TACC protein, and their the investigation of the molecular mechanisms involved in spindle interaction is thought to be critical for the function of TACC proteins assembly and for the development of new cancer chemotherapy in promoting microtubule assembly.17,18 A Xenopus homolog of strategies. msps, XMAP215, regulates microtubule plus-end assembly and is a processive microtubule polymerase.19 On the basis of these findings, the current model suggests that TACC proteins recruit msps/XMAP215 to the centrosome, enhancing microtubule RESULTS polymerization.20 Interestingly, tumor overexpressed gene protein The Identification of small compounds that inhibit TACC3 (TOGp), the human homolog of msps/XMAP215, which was To explore the role of TACC3 in spindle assembly and to develop a originally identified as a gene that was overexpressed in colon novel cancer chemotherapeutic targeting TACC3, we searched for and hepatic cancers,21 has been shown to physically interact with small compounds that were able to inhibit TACC3 function. We human TACC3.18 Furthermore, TOGp depletion has been shown to utilized a chemical array onto which small compounds obtained induce multipolar spindles, but the molecular basis of this action from the Natural Products Depository (NPDepo) chemical library remains under debate.22–24 The depletion of TACC3 leads to a range had been immobilized (Figure 1a).28 Using a cell lysate that of phenotypes, including a minor decrease in microtubule expressed a TACC3-DsRed fusion protein as a probe, we screened polymerization,22 chromosome misalignment leading to caspase- 6800 compounds and identified 70 as candidate TACC3- dependent apoptosis25 and declustered centrosomes in cancer cells interacting molecules. with supernumerary centrosomes.26 Using conditional knockout To determine the effects of the candidate compounds on mice, we previously demonstrated that disrupting Tacc3 in mitosis, we performed a visual screen using single cell-based lymphoma cells induced the formation of multipolar spindles.27 analysis. We used ovarian cancer cells because previous studies Interestingly, an in vivo analysis demonstrated that the thymic have indicated that TACC3 is aberrantly expressed in this type of lymphoma tissues underwent massive apoptosis, leading to tumor cancer.14,29 Cells expressing enhanced green fluorescent protein regression, whereas the normal thymus exhibited no overt (EGFP)-a-tubulin were synchronized using a double thymidine abnormalities. These observations indicate that Tacc3 is an block, and the candidate compounds were added to the medium important component of spindle assembly in cancer cells and is a at the release from the second block. Fluorescence images were potential target for cancer chemotherapy. In this study, we collected, and the image sequences were analyzed to determine developed a novel chemical compound that perturbed TACC3 the duration of mitosis. function. The pharmacological inhibition of TACC3 revealed Among the 70 candidates, four small molecules, NPD3574, different spatiotemporal contributions of TACC3 in spindle NPD2568, NPD4089 and NPD2448, induced mitotic arrest in a dose- formation. This unique mechanism of action provides a tool for dependent manner (Supplementary Figures S1a, b). Interestingly,

Figure 1. The identification of small compounds that bind to and inhibit the function of the TACC3–TOGp complex. (a) Design of the screen for chemical compounds. (b) Structures of SPL-A and SPL-B. (c) Dose-dependent mitotic arrest induced by SPL treatment. SKOV-3 cells expressing EGFP-a-tubulin were synchronized using a double thymidine block, and the indicated concentrations of the compounds were added at the second release. Time-lapse imaging was then used to determine the fates of single cells. The duration of mitosis was determined from the image sequences of at least 50 cells. (d) Interactions between SPLs and the TACC3–TOGp complex. SPL-A and SPL-B were immobilized on agarose beads and were incubated with the SKOV-3 cell lysates. The indicated proteins bound to the beads were detected using immunoblot analysis. GAPDH was used as a loading control.

& 2014 Macmillan Publishers Limited Oncogene (2014) 4242 – 4252 Spindle aberration by the TACC3 inhibitor R Yao et al 4244 these compounds had similar chemical structures. Each com- The spindle aberrations in ovarian cancer cells induced by pound had a dicoumarin structure and a hydroxyl group at the C7 SPL-A treatment were much stronger than the findings previously position. The hydroxyl group was indispensable for the activity of reported in TACC3-depleted HeLa cells. Therefore, we examined the compounds, and its propylation completely abolished their the effects of SPL-A in HeLa cells and found that, in contrast activity (Supplementary Figures S1a, b). On the basis of these with ovarian cancer cells, no overt spindle abnormalities were results, we then synthesized a series of chemical compounds, and induced (Supplementary Figure S5). However, a close examination structure-activity relationship studies identified two novel com- of chromosomes revealed that they were not properly aligned pounds that inhibited the progression of mitosis at different at the plates, and the percentages of cells containing effective doses. The compounds, spindlactone (SPL)-A and SPL-B, misaligned chromosome were increased in a dose-dependent both contained an N,N-disubstituted aminomethyl group at the C8 manner. This phenotype highly resembled to that reported in position (Figure 1b), and SPL-B showed more robust activity TACC3-depleted HeLa cells. These observations demonstrated that (Figure 1c). the effect of SPL-A depends on the cell types. The ability of the SPLs to interact with TACC3 was confirmed To address the mechanisms underlying the formation of using a pull-down assay (Figure 1d). The SPLs were immobilized multipolar spindles induced by SPL-A treatment, EGFP-a-tubulin on agarose beads with a photoaffinity linker, and SKOV-3 cell and mCherry-CENP-A were expressed in ovarian cancer cells, lysates were incubated with these beads. Immunoblot analysis and the dynamics of spindle assembly in these cells were revealed the presence of the co-precipitated TACC3 protein on the examined using time-lapse imaging. In the control cells, two SPL beads, indicating the binding of the SPLs to TACC3. intense EGFP-a-tubulin signals associated with centrosomes The intensity of the TACC3 signal was consistently stronger with were evident at the onset of NEB (Figure 2d, control). These the SPL-B beads compared with the SPL-A beads. The binding of microtubules grew rapidly and formed bipolar spindles. When the SPL-A and SPL-B to TACC3 was not nonspecific because other cells were treated with SPL-A, two centrosome-associated micro- microtubule-binding proteins, such as NDC80, MCAK and TPX2 as tubules were apparent at NEB (Figure 2d, multipolar). However, well as GAPDH, were barely detectable in the compound-loaded the intensities of these two microtubules were decreased beads. Interestingly, both types of SPL beads also pulled down the following NEB, which indicated that SPL-A perturbed the nucleation TOGp protein, suggesting that the TACC3–TOGp complex binds to and growth of centrosome microtubules. Intriguingly, the nuclea- SPL-A. Further analysis of TACC3-interacting molecules indicated tion and polymerization of a-tubulin was evident near centromere that Aurora-A and b-tubulin were also pulled down by the protein A (CENP-A) at the onset of NEB. An immunofluorescence compound-loaded beads. These observations suggested that SPL- analysis indicated that in contrast to the centrosome microtubules, A and SPL-B bind directly or indirectly to TACC3 and its interacting these spindles contained microtubules that were resistant to proteins. calcium treatment and were attached to kinetochores, suggesting To further investigate the interaction between SPL-A and TACC3 that they were chromosome-driven microtubules (Supplementary proteins, we performed the pull-down experiment after the Figure S6). These microtubules were nucleated and grew rapidly depletion of TACC3 and found that TACC3 depletion significantly in SPL-A-treated cells. The centrosomes were integrated into the reduced the binding of TOGp protein to the compound-loaded kinetochore microtubules during late prometaphase (Figure 2d, beads (Supplementary Figure S2). The residual binding may be multipolar, 12 min) but did not coalesce at the ectopic spindle potentially due to the incomplete deletion of TACC3 protein. In poles, thereby resulting in multipolar spindles. These observations contrast, TOGp depletion did not alter the binding of TACC3 demonstrate that SPL-A severely perturbed the nucleation protein to SPL-A, although the total amount of TACC3 protein was and elongation of centrosome microtubules, whereas the chromo- increased, which was most likely due to the mitotic arrest induced some-driven microtubules were assembled. In control cells, the by TOGp depletion. These observations suggested that the chromosome-driven microtubules were rapidly integrated into the interaction between TOGp with the compound was mediated, at mass of centrosome microtubules; therefore, the evaluation of least in part, by TACC3. However, the depletion of TACC3 led to SPL-A effects on the chromosome-driven microtubule nucleation robust alteration in the spindle structure, and the reduced binding was difficult. of TOGp may be an indirect effect caused by TACC3 depletion. The In addition to multipolar spindles, SPL-A also induced dis- precise mode of interaction needs to be further elucidated. organized spindles. These spindles consisted of two poles but the intensities of a-tubulin were lower than those in control cells. In addition, the chromosomes were not properly aligned at the SPLs induce aberrant spindles in ovarian cancer cells in a dose- metaphase plate (Figure 2b). The disorganized spindles were most dependent manner frequently observed following SPL-A treatment at a concentration Next, we examined the effects of the SPLs on spindle assembly in of 5 mg/ml in SKOV-3 cells (56.7%) and 2.5 mg/ml in OVCAR-3 cells SKOV-3 cells. Immunofluorescence analysis revealed that SPL-A (57.3%) (Figure 2a). Time-lapse imaging revealed that the treatment induced the formation of multipolar spindles. At a centrosome microtubules failed to grow after NEB (Figure 2d, concentration of 1.25 mg/ml, 53.3% of the SKOV-3 cells contained disorganized), a result that is similar to those observed in the multipolar spindles (Figure 2a). SPL-B essentially induced identical multipolar spindle formation. In addition, the nucleation and phenotypes, albeit with different dosage effects (Supplementary polymerization of kinetochore microtubules was significantly Figure S3). Another ovarian cancer cell line, OVCAR-3, exhibited delayed compared with those in multipolar spindle formation. similar spindle aberrations when treated with SPL-A, which The centrosomes were captured and integrated into the suggested that the spindle aberrations induced by SPL-A treatment kinetochore microtubules before they assembled ectopic spindle were not limited to SKOV-3 cells. g-Tubulin was localized to each poles, resulting in bipolar spindles. These observations demon- spindle pole, whereas centrin 2 was detected at only two (Figure 2b). strated that SPLs inhibited the nucleation of the centrosome and These observations indicated that the multipolar spindles were not kinetochore microtubules in a dose-dependent manner. induced by the overproduction of centrosomes or by cytokinesis At higher concentrations, SPL-A induced monopolar spindles failure following the previous round of mitosis. Notably, SPL-A (Figure 2a). OVCAR-3 cells were more sensitive to SPL-A treatment treatment did not alter the localization of TACC3 or TOGp (Figure 2c). compared with SKOV-3 cells. Treatment with 5 mg/ml of SPL-A Consistent with this finding, SPL-A did not prevent the phosphoryla- resulted in monopolar spindles in 10.6% of the SKOV-3 cells and in tion of TACC3 at Ser558, which was shown to be phosphorylated by 45.5% of the OVCAR-3 cells. Time-lapse imaging of OVCAR-3 cells Aurora-A and to mediate the localization of TACC3 to mitotic treated with SPL-A demonstrated that in addition to the spindles (Supplementary Figure S4). significant delay in the nucleation of centrosome and kinetochore

Oncogene (2014) 4242 – 4252 & 2014 Macmillan Publishers Limited Spindle aberration by the TACC3 inhibitor R Yao et al 4245 abDMSO SPL-A SKOV-3 bipolar multipolar disorgnized monopolar

γ-tubulin α-tubulin DAPI

centrin2 α-tubulin DAPI

OVCAR-3 c DMSO SPL-A bipolar multipolar disorgnized monopolar

TACC3 γ-tubulin DAPI

TOGp γ-tubulin DAPI d

control

multipolar

disorganized

Figure 2. SPL-A induces aberrant spindles in a dose-dependent manner by perturbing the nucleation of centrosomes and chromosome microtubules. (a) The quantification of cells containing aberrant spindles. The cells were treated with the indicated concentrations of SPL-A for 24 h, fixed and stained with antibodies against a- and g-tubulin. The numbers of cells used to determine the percentage of the aberrant spindles are indicated (N). The data represent the mean values and standard deviations (s.d.’s) of three independent experiments. *Po0.05 and **Po0.01 versus corresponding phenotypes in untreated cells (t-test). (b) Aberrant spindles induced by SPL treatment in SKOV-3 cells. Cells were treated with SPL-A at the concentration of 1.25 mg/ml (multipolar), 5 mg/ml (disorganized) or 10 mg/ml (monopolar) for 24 h; fixed; and stained with anti-a-tubulin (green), anti-g-tubulin or anti-centrin2 antibodies (red) and 4,6-diamidino-2-phenylindole (DAPI) (blue). Bar ¼ 10 mm. (c) Localization of TACC3 and TOGp in SPL-treated cells. The cells were treated as described in b, fixed and stained for g-tubulin (green), TACC3 or TOGp (red) and DAPI (blue). Bar ¼ 10 mm. (d) Time-lapse imaging of SKOV-3 cells expressing EGFP-a-tubulin (green) and mCherry-CENP-A (red). Cells were left untreated (control) or treated with SPL-A at 1.25 and 5.0 mg/ml to induce multipolar and disorganized spindles, respectively. Images were then collected. The numbers indicate the time in minutes, and 0 defines the point of NEB. The upper panels show composite images, and the lower panels show images of a-tubulin.

& 2014 Macmillan Publishers Limited Oncogene (2014) 4242 – 4252 Spindle aberration by the TACC3 inhibitor R Yao et al 4246 microtubules, two centrosomes were moved toward the center TOGp depletion induced monopolar spindles that recapitulated of the cell and eventually became monopolar spindles the spindles observed at high concentrations of SPL-A (Supplementary Figure S7). These observations indicated that The spindle aberrations induced by SPL-A treatment recapitulated the severe perturbation of microtubule nucleation by high the phenotypes observed in TACC3-depleted cells, suggesting that concentrations of SPL-A led to a reduction of centrosome ejection SPL-A induced aberrant spindles by targeting TACC3. However, force. monopolar spindles were not observed in TACC3-depleted cells. Monopolar spindles were only observed at high doses of SPL-A; therefore, they may be induced via TACC3-independent mechan- Centrosome microtubule assembly requires TACC3 function isms. Alternatively, because cells transfected with siRNA targeting The depletion of TACC3 function in mammalian cells has been TACC3 still expressed a significant amount of TACC3 protein shown to lead to a range of spindle defects that are distinct from (Figure 3a), its depletion may not have been sufficient to induce the findings observed in SPL-A-treated cells, with the exception of monopolar spindles. To address this issue, we depleted TOGp in the occurrence of multipolar spindles in TACC3-depleted lym- ovarian cancer cells because TOGp was a critical protein involved phoma cells.22,25–27 Next, we investigated the role of TACC3 in TACC3-mediated nucleation of microtubules.20 In addition, our in ovarian cancer cells. When cells were transfected with pull-down experiments in ovarian cancer cells indicated that TACC3-specific small interfering RNA (siRNA) oligonucleotides, TACC3–TOGp as a complex binds to SPL-A (Figure 1d). Transfection the TACC3 protein levels were reduced to 18.2 and 15.2% in the with TOGp-specific siRNA oligonucleotides reduced the level of SKOV-3 and OVCAR-3 cells, respectively, compared with the cells TOGp protein to 16.8 and 6.2% of its levels in SKOV-3cells and transfected with control siRNA oligonucleotides (Figure 3a). An OVCAR-3 cells, respectively, compared with the cells transfected immunofluorescence analysis revealed the accumulation of multi- with control siRNA oligonucleotides (Figure 4a). polar spindles in the cells depleted of TACC3 (Figures 3b, c). Sixty- Consistent with previous reports,22–24,30 we found that 30% of percent of the SKOV-3 cells and 69% of the OVCAR-3 cells the TOGp-depleted SKOV-3 cells contained multipolar spindles contained multipolar spindles 24 h after transfection with TACC3 (Figure 4b). Similar to the observations in TACC3-depleted cells, siRNA. Each spindle pole contained g-tubulin but only two of these each spindle pole contained g-tubulin, whereas centrin 2 was poles contained centrin 2 (Figures 3b–d). Furthermore, time-lapse localized to only two spindle poles (Figure 4c). In addition to the imaging analysis revealed that TACC3 deprivation selectively observed multipolar spindles, 36% of the cells contained inhibited the nucleation of centrosome microtubules, whereas disorganized bipolar spindles (Figure 4b). These spindles consisted kinetochore microtubules rapidly grew and polymerized, leading of two poles, but their a-tubulin intensities were weaker to ectopic spindle poles (Figure 3e, multipolar). Thus, the compared with normal bipolar spindles, and the chromosomes depletion of TACC3 led to the formation of multipolar spindles were not properly aligned at the metaphase plate (Figure 4c). by selectively inhibiting the nucleation of centrosome micro- These observations demonstrated that TOGp had prominent roles tubules. These observations indicated that SPL-A treatment in the nucleation of kinetochore microtubules in addition to recapitulated the spindle aberrations induced by TACC3 depletion, centrosome microtubules. Interestingly, in OVCAR-3 cells, 69% of supporting the notion that SPL-A perturbed spindle assembly by the cells exhibited monopolar spindles in response to TOGp targeting TACC3 function. depletion (Figure 4b). An immunofluorescence analysis revealed Although multipolar spindles were the major aberration of the that these spindles contained two centrosomes and centrioles and mitotic spindle in response to TACC3 depletion, a small but were similar to the findings observed in SPL-A-treated cells significant number of cells contained disorganized spindles. This (Figure 4d). These observations support the idea that SPL-A finding was observed in 17 and 15% of SKOV-3 cells and OVCAR-3 induced monopolar spindles by perturbing TOGp-mediated cells, respectively (Figures 3b, c). Although these spindles spindle assembly. In addition, monopolar spindles were only contained two centrosomes, the intensity of microtubules was observed in OVCAR-3 cells, and this finding was consistent with lower than the intensity observed in normal bipolar spindles, and the result demonstrating that OVCAR-3 cells were more sensitive the chromosomes were not correctly aligned at the metaphase to the SPL-A treatment compared with SKOV-3 cells (Figure 2a). plates (Figure 3b). Time-lapse imaging revealed that the disorganized spindles resulted from the inhibition of both the kinetochore microtubules and the centrosome microtubules SPL selectively inhibits mitosis in cancer cells and suppresses (Figure 3e). These observations indicate that although TACC3 in vivo tumor growth had a major role in the nucleation of centrosome microtubules, it The results described above demonstrated that in the ovarian also contributed to the nucleation of kinetochore microtubules. cancer cells that expressed a high level of TACC3, TACC3 was The observations described above indicated that TACC3 indispensable for spindle assembly. Therefore, we examined contributed to spindle assembly at spatially distinct sites; there- TACC3 function in normal cells to explore the potential of TACC3 fore, we investigated the subcellular localization of TACC3 during as a molecular target for cancer treatment. Consistent with the mitosis. TACC3-mCherry was expressed in SKOV-3 cells with EGFP- results obtained by siRNA transfection, the depletion of TACC3 a-tubulin, and its localization was examined using time-lapse using lentivirus-mediated small hairpin RNA expression signifi- imaging (Supplementary Figure S8). At the onset of NEB, TACC3 cantly induced mitotic arrest in SKOV-3 cells and in OVCAR-3 cells accumulated rapidly around the centrosomes, suggesting that it (Figures 5a, b). In contrast, despite efficient reductions of TACC3 has a role in the nucleation of centrosome microtubules. At this protein levels, three normal cell lines, hTERT-RPE, MCF10A and time, TACC3 was not detected in kinetochore microtubules. 184B5, exhibited no overt alterations in mitosis. Next, we found However, 5 min after NEB, TACC3 was recruited to kinetochore that SPL-A specifically induced mitotic arrest in cancer cells microtubules that were in close proximity to centrosomes and (Figure 5c). Interestingly, the specific induction of mitotic arrest in that were about to be integrated into centrosome microtubules cancer cells was not observed with microtubule toxins as well as (Supplementary Figure S8, 5 min, arrowheads in inset). These with small compounds that inhibit mitotic proteins, such as Aurora observations are consistent with the idea that TACC3 has and Polo-like kinases and the motor protein, Eg5 (Supplementary dominant roles in the nucleation of centrosome microtubules Figure S9). These observations demonstrated that the unique at NEB and additional roles involving kinetochore microtubules mechanism of action of SPL conferred selective perturbation of at later stages. These results may explain the disorganized spindle assembly in ovarian cancer cells. spindles that were present in the cells treated with higher doses To evaluate the role of TACC3 in tumor proliferation in vivo,we of SPL-A. examined the ability of SPL-B to suppress the growth of SKOV-3

Oncogene (2014) 4242 – 4252 & 2014 Macmillan Publishers Limited Spindle aberration by the TACC3 inhibitor R Yao et al 4247 SKOV-3 OVCAR-3 abcontrol si TACC3 si control si TACC3 si SKOV-3 OVCAR3 bipolar multipolar disorganized bipolar multipolar disorganized CT CT

TACC3 TACC3 α-tubulin DNA GAPDH

γ-tubulin α-tubulin DNA

c SKOV-3 OVCAR-3 d control si TACC3 si control si TACC3 si bipolar multipolar bipolar multipolar

TOGp γ-tubulin DNA

centrin2 γ-tubulin DNA

e

multipolar

disorganized

Figure 3. Depletion of TACC3 leads to spindle aberrations in ovarian cancer cells. (a) The expression levels of TACC3 protein in cells transfected with control (C) or TACC3 (T) siRNA were determined using immunoblot analysis. GAPDH served as the loading control. (b) Induction of multipolar and disorganized spindles in TACC3-depleted cells. Cells were transfected with the indicated siRNA. After 48 h, cells were fixed and stained with the indicated antibodies (green and red) and DAPI (blue). Bar ¼ 10 mm. (c) Quantification of the presence of aberrant spindles. Cells were transfected with the indicated siRNA and analyzed with immunofluorescence analysis using antibodies against a- and g-tubulin. The numbers of cells used to determine the percentage of the aberrant spindles were indicated (N). The data represent the mean values and the s.d.’s of three independent experiments. *Po0.05 and **Po0.01 versus corresponding phenotypes in untreated cells (t-test). (d) The localization of centrosomal proteins in TACC3-depleted cells. Cells were transfected, fixed and stained with the indicated antibodies (green and red) and DAPI (blue), as described in b. Bar ¼ 10 mm. (e) Time-lapse imaging of SKOV-3 cells expressing EGFP-a-tubulin (green) and mCherry-CENP-A (red). Cells were transfected with the TACC3 siRNA, and the images were collected beginning 24 h after transfection. The numbers indicate the time in minutes, and 0 defines the point of NEB. The upper panels show composite images, and the lower panels show images of a-tubulin. cell xenografts in nonobese diabetic/severe combined immuno- bioavailability, and 6 h after the oral administration of 20 mg/kg deficient mice. Similar to findings observed with SPL-A, SPL-B of SPL-B, its plasma concentration was 270 ng/ml, whereas the selectively induced mitotic arrest in ovarian cancer cells concentration of intraperitoneally administered SPL-B was 90 ng/ (Supplementary Figure S10). SPL-B exhibited excellent oral ml. Cells were injected subcutaneously into animals, which were

& 2014 Macmillan Publishers Limited Oncogene (2014) 4242 – 4252 Spindle aberration by the TACC3 inhibitor R Yao et al 4248 b a SKOV-3 OVCAR-3 CCTT

TOGp

GAPDH

c SKOV-3 d OVCAR-3 control si TOGp si control si TOGp si bipolar multipolar disorganized bipolar monopolar

γ-tubulin γ-tubulin α-tubulin α-tubulin DNA DNA

centrin 2 α centrin 2 -tubulin α-tubulin DNA DNA

TACC3 TACC3 γ-tubulin γ-tubulin DNA DNA

Figure 4. The depletion of TOGp induces spindle aberrations in ovarian cancer cells. (a) The expression levels of the TOGp protein in cells transfected with control (C) or TOGp (T) siRNA were determined using immunoblot analysis. GAPDH served as the loading control. (b) The quantification of aberrant spindles. The cells were transfected with the indicated siRNA and stained with antibodies against a- and g-tubulin. The numbers of cells counted for determining the percentage of the aberrant spindles are indicated (N). The data represent the mean values and s.d.’s of three independent experiments. *Po0.05 and **Po0.01 versus corresponding phenotypes in untreated cells (t-test). (c) Aberrant spindles induced by TOGp depletion in SKOV-3 cells. Cells were transfected with control or TOGp siRNA, fixed and stained with the indicated antibodies (green and red) and DAPI (blue). Bar ¼ 10 mm. (d) Monopolar spindles induced by TOGp depletion in OVCAR-3 cells. Cells were transfected with control or TOGp siRNA, fixed and stained with the indicated antibodies (green and red) and DAPI (blue). Bar ¼ 10 mm.

subsequently treated with SPL-B via oral administration. In mice by the lack of an in vitro measure of the biochemical activity of treated with a single administration of SPL-B at a concentration of TACC3. In this study, using chemical array screening followed by 10 mg/kg, the percentage of cells positive for histone H3 visual screening, we identified several novel chemical compounds, phosphorylation was significantly increased (1.2% and 2.1% in SPLs. Although the pull-down experiments indicated that SPLs dimethyl sulfoxide and SPL-B-administered mice, respectively; bound to TACC3, we do not know whether this binding occurred Po0.001) (Figure 5d), indicating that SPL-B induced mitotic arrest directly or indirectly. However, we showed that the phenotypes in vivo. induced by low concentrations of SPLs are highly similar to the The tumor volumes of mice treated twice a day for 14 days with phenotypes induced by TACC3 depletion, and the dose- 10 mg/kg SPL-B were reduced by 48% (Po0.002) in comparison dependent effects of SPL on spindle assembly were well with the control group (Figure 5e). Tumor growth reduction was correlated with TACC3 and TOGp depletion. These observations dose-dependent and was significant at a dosage of 2.5 mg/kg provide evidence that SPLs induce aberrant spindles by (69%, Po0.05). SPL-B was well tolerated at a dose of 10 mg/kg, perturbing the nucleation of microtubules involved in TACC3 and no significant decrease in the body weight of the mice was function. To formally prove that the inhibition of TACC3 is observed (Figure 5f). These observations demonstrated the responsible for the spindle aberrations observed, it will be potential utility of SPL-B as a chemotherapeutic agent for cancer necessary to determine the structure of the SPL–TACC3 complex treatment. and/or to identify mutations in TACC3 that render it resistant to SPLs.31 We have shown that the depletion of TACC3 selectively DISCUSSION inhibited the nucleation of centrosome microtubules. Notably, TACC3 is aberrantly expressed in human cancers and is considered the robust nucleation of chromosome-driven microtubules was to be a potential molecular target for cancer chemotherapy. Using observed in the absence of TACC3 (Figure 3e). Although several conditional knockout mice, we previously demonstrated that the lines of evidence indicate that kinetochore microtubules operate disruption of TACC3 led to tumor regression, whereas normal in mammalian somatic cells after recovery from microtubule tissues exhibited no overt abnormalities.27 Therefore, small toxins and cold treatment,9,32,33 their kinetics and dynamics under compounds that inhibit TACC3 function may be beneficial in physiological conditions and at the onset of NEB remain largely developing new strategies for cancer chemotherapy. However, the unknown. Centrosome-independent mitotic spindle formation development of these types of compounds has been hampered was previously described in green monkey fibroblasts in which

Oncogene (2014) 4242 – 4252 & 2014 Macmillan Publishers Limited Spindle aberration by the TACC3 inhibitor R Yao et al 4249

Figure 5. SPL-B suppresses in vivo tumor growth. (a) The expression levels of TACC3 protein in cells infected with lentiviruses-expressing control (C) or TACC3 (T) small hairpin RNA (shRNA) were determined using immunoblot analysis. GAPDH served as the loading control. (b) Depletion of TACC3 in ovarian cancer cells induced the accumulation of cells in the G2/M phase. Depletion of TACC3 in normal cells did not lead to the accumulation of cells in the G2/M phase. Cells were infected with lentiviruses-expressing control (black line) or TACC3 small hairpin RNA (red line). At 24 h after infection, puromycin was added, and the cells were incubated for another 24 h. The cells were then analyzed using flow cytometry after staining with propidium iodide. (c) SPL-A selectively induced mitotic arrest in ovarian cancer cells. Cells were treated with the indicated concentrations of SPL-A for 24 h and analyzed using flow cytometry after staining with propidium iodide. The percentage of cells in the G2/M phase is indicated. (d) SKOV-3 cells were subcutaneously injected into nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice, and the mice were treated with SPL-B at a concentration of 10 mg/kg via oral administration. Tumors were collected and fixed 2 days after treatment. The paraffin-embedded sections were stained with phosphorylated histone H3 (red), human HLA-DR antigen (green) and DAPI (blue). (e) SKOV-3 cells were subcutaneously injected into both flanks of 9 NOD/SCID mice, with 2–3 mice in each group. The mice were treated with the indicated amounts of SPL-B every 2 days, and the tumor volumes were scored every 2 days. The data represent the mean values and s.d.’s. (f) The NOD/SCID mice bearing SKOV-3 cells were treated with SPL-B at a concentration of 10 mg/kg, as described e, and their body weights were scored every 2 days. the centrosomes had been destroyed with laser microsurgery.11 It dynamics of kinetochore spindle assembly and its molecular should be noted that this laser treatment destroyed centrosomes mechanisms. as organelles, and the centrosomal proteins were distributed In HeLa cells, the depletion of TACC3 does not induce overt elsewhere in the cytoplasm, whereas our approach specifically spindle aberrations, whereas TOGp depletion induces multipolar depleted TACC3 without affecting other centrosomal components. spindles.22,24 However, the molecular basis of these differences This difference creates difficulty in directly comparing remains to be clear. In this study, we demonstrated that a major centrosome-independent microtubule assembly in TACC3- phenotype of TACC3 perturbation is the inhibition of the depleted cells versus cells in which the centrosomes had been centrosome microtubule nucleation. We also showed that TOGp destroyed. The kinetochore microtubules were observed in the has more profound roles in the assembly of kinetochore presence of centrosome microtubules (Supplementary Figure S8), microtubules. Interestingly, SPLs inhibit spindle assembly in a confirming a previous report.10 We noticed that in the absence of dose-dependent manner. For example, at low concentrations, TACC3, the nucleation and polymerization of kinetochore SPLs only perturbed the nucleation of centrosome microtubules, microtubules was strikingly more extensive than in the presence whereas at high concentrations, SPLs also inhibited kinetochore of TACC3 activity. This difference may indicate that the mass of microtubule assembly. One possible explanation for these centrosome microtubules rapidly incorporates kinetochore observations is that the nucleation of centrosome microtubules microtubules, which makes their visualization difficult. It is also requires higher microtubule polymerization activity compared possible that tubulin proteins are abundantly present in the with kinetochore microtubules, suggesting that the TACC3–TOGp absence of centrosome microtubules, leading to the robust complex has spatiotemporally different contributions to spindle nucleation and polymerization of kinetochore microtubules. Our assembly. These observations support the current model that system of visualizing kinetochore microtubule assembly provides TACC3 facilitates spindle assembly by stimulating TOGp micro- a unique and valuable tool for investigating the kinetics and tubule polymerization activity; therefore, the depletion of TOGp

& 2014 Macmillan Publishers Limited Oncogene (2014) 4242 – 4252 Spindle aberration by the TACC3 inhibitor R Yao et al 4250

exhibited a more drastic phenotype than TACC3 depletion. In this chromatography (CHCl3: MeOH ¼ 20: 1 and CHCl3: MeOH ¼ 10: 1) to study, we did not observe any alterations in the cellular generate SPL-A (10.2 mg, 76%) as a yellow oil. 1 localization of TOGp in response to TACC3 depletion or vice versa. H Nuclear magnetic resonance (NMR) (500 MHz, CDCl3) d ¼ 1.32 (m, 2H), Similarly, SPLs did not change the cellular localization of TACC3 or 1.53 (m, 3H), 1.79 (brd, J ¼ 12.6 Hz, 2H), 2.24 (brs, 2H), 3.03 (brs, 2H), 3.70 TOGp. Further analysis regarding the mechanism by which SPLs (t, J ¼ 6.9 Hz, 2H), 4.04 (s, 2H), 6.22 (s, 1H), 6.67 (d, J ¼ 9.2 Hz, 1H), 7.15 (d, J 9.2 Hz, 1H), 7.36 (brdd, J 7.4, 7.4 Hz, 1H), 7.42 (brd, J 8.6 Hz, 1H), inhibit TACC3 function and perturb the nucleation and polymer- ¼ ¼ ¼ 7.57 (dd, J ¼ 7.4, 1.7 Hz, 1H), 7.63 (ddd, J ¼ 8.6, 7.4, 1.7 Hz, 1H), 7.82 (s, 1H). ization of microtubules will provide clues for a better under- 13 C NMR (125 MHz, CDCl3) d ¼ 32.0, 39.0, 53.3, 54.3, 60.3, 108.1, 110.2, standing of spindle assembly. 114.0, 117.0, 118.4, 124.3, 125.0, 126.2, 128.5, 133.0, 143.1, 150.1, 152.9, Our present work demonstrates that TACC3 depletion induces 154.3, 159.0, 160.7 and 168.8. þ mitotic arrest in ovarian cancer cells but not in normal cells and is HRMS-FAB: m/z [M þ H] expected for C26H26NO6: 448.1760; observed: consistent with previous studies showing that oncogenic changes 448.1764. may drive TACC3 dependency for spindle assembly.27,34 We demonstrated that the robust nucleation and elongation of Spindlactone B. kinetochore microtubules induced ectopic spindle poles, which led to multipolar spindles. These observations suggested that in cancer cells, the elevated polymerization activity of kinetochore microtubules as well as their pole-focusing activity might contribute to the cellular context-dependent sensitivity to TACC3 depletion. In fact, several of the kinetochore components, including the Ndc80 complex, Zwint and Aurora B, are aberrantly expressed in various cancers.35–37 Recently, NuMA, which localizes microtubules at the poles and tethers them to centrosomes,38 was shown to be highly expressed in ovarian cancers.39 These molecules may also contribute to the selective inhibition of cancer cells by SPLs. Overall, the unique mechanism of the action of SPLs not only aids in the dissection of the molecular basis of spindle assembly but also provides a rationale

for the use of TACC3 as a molecular target for cancer treatment. Formaldehyde (37% in H2O) (458 ml, 6.21 mmol) and 4-(ethylamino- This finding offers an opportunity to develop new strategies for methyl)pyridine (267 mg, 1.96 mmol) were added to a stirred solution of cancer chemotherapy with expanded scope and clinical efficacy. dicoumarin (100 mg, 0.327 mmol) in acetonitrile (15 ml) at room tempera- ture. The stirred reaction mixture was heated under reflux for 24 h. The mixture was cooled to room temperature, concentrated and diluted with MATERIALS AND METHODS CHCl3. The organic layer was washed with H2O and brine, dried (Na2SO4) Chemical array screening and concentrated. The resulting yellow oil was purified twice with preparative thin-layer chromatography (CHCl3: MeOH ¼ 10: 1 and CHCl3: Human TACC3 complementary DNA was cloned into the pDsRed-Express- MeOH ¼ 50: 1) to produce SPL-B (73.7 mg, 49%) as a yellow solid. N1 vector, which was transfected into 293T cells using Lipofectamine 1 H NMR (500 MHz, CDCl3) d ¼ 1.21 (t, J ¼ 6.9 Hz, 3H), 2.69 (q, J ¼ 6.9 Hz, 2000 (Invitrogen, Carlsbad, CA, USA). At 48 h after transfection, the cell 2H), 3.73 (brs, 2H), 4.16 (s, 2H), 6.28 (s, 1H), 6.74 (d, J ¼ 9.2 Hz, 1H), 7.20 (d, pellet was collected and suspended in phosphate-buffered saline. Cell J ¼ 9.2 Hz, 1H), 7.28 (dd, J ¼ 4.6, 1.7 Hz, 2H), 7.38 (ddd, J ¼ 7.4, 7.4, 1.1 Hz, lysates were prepared by sonication and by centrifugation to remove any 1H), 7.43 (brd, J ¼ 8.6 Hz, 1H), 7.61 (dd, J ¼ 7.4, 1.7 Hz, 1H), 7.66 (ddd, J ¼ 8.6, insoluble materials. The expression of the TACC3-DsRed fusion protein was 7.4, 1.7 Hz, 1H), 7.87 (s, 1H), 8.60 (dd, J ¼ 4.6, 1.7 Hz, 2H). detected by immunoblot analysis using antibodies against TACC3 and 13C NMR (125 MHz, CDCl ) d ¼ 11.0, 47.6, 49.9, 57.0, 108.4, 110.5, 112.3, DsRed (Clontech, Mountain View, CA, USA). The cell lysates were incubated 3 28 113.8, 116.9, 118.3, 124.0, 124.1, 125.1, 126.5, 128.5, 133.1, 143.3, 145.6, with the chemical array slides for 1 h at 4 1C in a hybridization chamber. 150.1, 150.2, 152.9, 154.2, 159.0, 160.6 and 162.8. The cell lysates obtained from the DsRed-expressing cells were used as a High resolution mass spectrum-fast atom bombardment (HRMS-FAB): control. After three washes with ice-cold phosphate-buffered saline, the þ m/z [M þ H] expected for C27H23N2O5: 455.1607; observed: 455.1589. slides were dried and scanned on a GenePix microarray scanner at an [M þ Na] þ expected for C H N NaO : 477.1426; observed: 477.1423. excitation wavelength of 532 nm and an emission wavelength of 575 nm. 27 22 2 5 Binding of SPLs to TACC3 and TOGp SPL preparation SPLs immobilized on agarose beads were prepared as previously Spindlactone A. described.40 SKOV-3 cell lysates were prepared by sonication in binding buffer (10 mM Tris-HCl, pH 7.5, 50 mM KCl, 5 mM MgCl2,1mM EDTA and a protease inhibitor cocktail (Roche, Basel, Switzerland)). After the insoluble materials were removed by centrifugation, the supernatants (500 mgof protein) were incubated with 30 ml of beads for 3 h at 4 1C. The bound beads were washed four times with binding buffer containing 0.025% NP- 40, and the bound proteins were eluted in SDS–polyacrylamide gel electrophoresis sample buffer. Next, the samples were analyzed by immunoblot analysis. An anti-NADPH antibody was used as a control.

Time-lapse imaging Cells were transfected with siRNA oligonucleotides and plated on 60-mm glass-bottom plates (IWAKI, Tokyo, Japan). At 16 h after transfection, 150 ml of a CellMatrix gel (Nitta Gelatin, Osaka, Japan) was overlaid to minimize Formaldehyde (37% in H2O) (17.7 ml, 240 mmol) and 4-piperidineethanol the movement of cells during mitosis, and 150 ml of Roswell Park Memorial (9.7 mg, 75.0 mmol) were added to a stirred solution of dicoumarin (9.2 mg, Institute medium 1640 medium supplemented with 10% fetal bovine 30.0 mmol) in acetonitrile (3.5 ml) at room temperature. The stirred reaction serum was added. Fluorescence images were acquired every 2 min using a mixture was heated under reflux for 27 h. The mixture was cooled to room Leica AF6000 fluorescence microscope with a 63 Â /1.40–0.70 Plan APO temperature, concentrated and diluted with CHCl3. The organic layer was objective lens. Z-projections are shown in Figures 2 and 3. washed with H2O and brine, dried (Na2SO4) and concentrated. The For the analysis of cell fates, the cells were seeded onto 96-well optical- resulting yellow oil was purified twice with preparative thin-layer bottom plates (Nunc, Roskilde, Denmark) and treated with 2.5 mM

Oncogene (2014) 4242 – 4252 & 2014 Macmillan Publishers Limited Spindle aberration by the TACC3 inhibitor R Yao et al 4251 thymidine for 16 h, followed by incubation in thymidine-free fresh medium The lentiviral particles were produced by the co-transfection of 293T cells for 8 h. During this period, the cells were transfected with siRNA with the pLKO.1 constructs and the pMD2.G and psPAX2packaging oligonucleotides as described above. The cells were then treated again plasmids. The transfections were conducted using Lipofectamine 2000 with 2.5 mM thymidine for 16 h. Images were collected using a Leica (Invitrogen), and the virus was harvested at 48 and 72 h after transfection. AF6000 with a 20 Â /0.5 Plan Fluotar objective lens every 10 min for 3 days The cells were incubated with the lentiviral supernatants in the presence of in phenol red-free Roswell Park Memorial Institute medium 1640 medium 6 mg/ml polybrene for 24 h, and the infected cells were selected with 1– supplemented with 5% fetal bovine serum. The image sequences were 8 mg/ml puromycin. At 48 h after infection, the cells were collected, fixed in viewed using ImageJ 1.42a software (National Institute of Health, Bethesda, ethanol and analyzed using flow cytometry after staining with propidium MD, USA), and the cell fates were determined manually. iodide.

Cell lines and transfection of siRNA Animal experiments and immunohistological analysis SKOV-3 cells and OVCAR-3 cells were cultured in Roswell Park Memorial All animal studies and procedures were approved by the Japanese Institute medium 1640 medium supplemented with 5% fetal bovine Foundation for Cancer Research (JFCR) Cancer Institute Animal Committee serum.41 hTERT-RPE cells were cultured in Dulbecco’s modified Eagle (Japan). For the xenograft tumor models, 1  107 SKOV-3 cells were medium supplemented with 10% fetal bovine serum, and MCF10A and injected subcutaneously into each flank of nonobese diabetic/severe 184B5 cells were cultured in MEGM supplemented with 100 ng/ml and combined immunodeficient mice. The mice were treated with 50 mlof 1 ng/ml cholera toxin, respectively. Stable clones that constitutively SPL-B dissolved in dimethyl sulfoxide by oral administration every 2 days, expressed EGFP-a-tubulin (Clontech) were obtained via lentiviral and the tumor volumes were scored. Dimethyl sulfoxide was administered transduction using the pLenti4/V5-DEST vector (Invitrogen). Infected cells as a control using the same procedure. were selected using Zeocin, followed by limiting dilution cloning. Cell lines Tissues were collected, fixed in 10% buffered formalin (Wako, Osaka, that expressed mCherry-CENP-A or mCherry-TACC3 were obtained by Japan) and paraffin-embedded. Next, 1-, 4- or 6-mm sections were infecting EGFP-a-tubulin-expressing cells with pLenti6/V6-DEST prepared. Immunohistochemical analysis was performed as previously (Invitrogen) and were selected using blasticidin S HCl (Invitrogen). described.27 The control siRNA and siRNAs specific for TACC3 and TOGp were obtained from Invitrogen. Cells were transfected with these siRNAs using Lipofectamine RNAiMAX in accordance with the manufacturer’s instruc- CONFLICT OF INTEREST tions. Three sequences were tested for each target, and at least two sequences were required to exhibit similar effects on gene silencing and The authors declare no conflict of interest. mitosis for the target to be considered valid. ACKNOWLEDGEMENTS Immunoblotting We thank T Hirota for providing the mCherry-CENP-A construct, K Yamazaki for The antibody against TACC3 was generated by immunizing rabbits with a preparing the cells for the JFCR cancer panels, K Honda for excellent technical peptide corresponding to the amino acids 213–224 of human TACC3. The support with the chemical screening and T Saito for supplying compounds from rabbit anti-TOGp antibody was raised against the C-terminal 111 amino NPDepo. This study was performed as a research program of the Project for acids of human TOGp fused to glutathione S-transferase. Affinity-purified Development of Innovative Research on Cancer Therapeutics (P-Direct), Ministry of antibodies were used in this study. The anti-GADPH antibodies (sc-25778) Education, Culture, Sports, Science and Technology of Japan, and was supported by a were purchased from Santa Cruz Biotechnology (Dallas, TX, USA). The Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, phospho-TACC3 (Ser558) (D8H10) and anti-Aurora-A (1G4) antibodies were Science and Technology of Japan. purchased from Cell Signaling Technology (Danvers, MA, USA). The anti-b- tubulin (ab6046) antibody and NDC80 (9G3.23) were purchased from Abcam (Cambridge, UK) and Novus Biologicals (Cambridge, UK), REFERENCES respectively. Total cell lysates were prepared in 1% NP-40 buffer (1% NP-40, 50 mM 1 Kavallaris M. Microtubules and resistance to tubulin-binding agents. Nat Rev Tris-HCl, pH 8.0, 150 mM NaCl and 10% glycerol). Equal amounts of proteins Cancer 2010; 10: 194–204. were separated by SDS–polyacrylamide gel electrophoresis and transferred 2 Sudakin V, Yen TJ. Targeting mitosis for anti-cancer therapy. BioDrugs 2007; 21: to nitrocellulose membranes. The blots were incubated with primary 225–233. antibodies, followed by the incubation with horseradish peroxidase- 3 Jackson JR, Patrick DR, Dar MM, Huang PS. Targeted anti-mitotic therapies: can we conjugated secondary antibodies and visualization using an enhanced improve on tubulin agents? Nat Rev Cancer 2007; 7: 107–117. chemiluminescence detection system (GE Healthcare, Buckinghamshire, 4 Keen N, Taylor S. Aurora-kinase inhibitors as anticancer agents. Nat Rev Cancer UK). For quantitation, the blots were scanned, and the total pixel intensity 2004; 4: 927–936. was calculated using ImageJ 1.42a software and normalized to the total 5 Strebhardt K, Ullrich A. Targeting polo-like kinase 1 for cancer therapy. Nat Rev pixel intensity of the GADPH control. Cancer 2006; 6: 321–330. 6 Harrison MR, Holen KD, Liu G. Beyond taxanes: a review of novel agents that target mitotic tubulin and microtubules, kinases, and kinesins. Clin Adv Hematol Immunofluorescence Oncol 2009; 7: 54–64. The cells were cultured on chamber slides and fixed in 3.7% PFA, 100% 7 Kirschner M, Mitchison T. Beyond self-assembly: from microtubules to morpho- methanol or 10% TCA, as previously described.42,43 They were then genesis. Cell 1986; 45: 329–342. incubated with primary antibodies. Monoclonal antibodies against 8 Wollman R, Cytrynbaum EN, Jones JT, Meyer T, Scholey JM, Mogilner A. Efficient a-tubulin (DM1A) and g-tubulin (DQ19 and GTU88) were obtained from chromosome capture requires a bias in the ’search-and-capture’ process during Sigma. Anti-pericentrin and anti-Aurora-A antibodies were purchased from mitotic-spindle assembly. Curr Biol 2005; 15: 828–832. Covance (Princeton, NJ, USA) and TransGenic (Kumamoto, Japan), 9 Khodjakov A, Copenagle L, Gordon MB, Compton DA, Kapoor TM. Minus-end respectively. Anti-centrin 2 (sc-27793) was obtained from Santa Cruz capture of preformed kinetochore fibers contributes to spindle morphogenesis. Biotechnology. Fluorescein-conjugated goat anti-mouse IgG (Cappel, J Cell Biol 2003; 160: 671–683. Solon, OH, USA) and Cy3-conjugated goat anti-rabbit IgG (Millipore, 10 Maiato H, Rieder CL, Khodjakov A. Kinetochore-driven formation of kinetochore Temecula, CA, USA) were used as secondary antibodies. DNA was detected fibers contributes to spindle assembly during animal mitosis. J Cell Biol 2004; 167: using 4,6-diamidino-2-phenylindole. Imaging was performed using a Leica 831–840. DM6000B upright microscope equipped with a  100/1.40–0.70 Plan APO 11 Khodjakov A, Cole RW, Oakley BR, Rieder CL. Centrosome-independent mitotic objective lens. Z-projections are shown in Figures 2–4. spindle formation in vertebrates. Curr Biol 2000; 10: 59–67. 12 Still IH, Vince P, Cowell JK. The third member of the transforming acidic coiled coil-containing gene family, TACC3, maps in 4p16, close to translocation break- Lentiviral infection and FACS analysis points in multiple myeloma, and is upregulated in various cancer cell lines. Cells were transduced with pLKO.1-puro lentiviral small hairpin RNA Genomics 1999; 58: 165–170. targeting TACC3 (TRCN000062023–7, Thermo Scientific, Hudson, NH, USA). 13 Ulisse S, Baldini E, Toller M, Delcros JG, Gueho A, Curcio F et al. Transforming The representative results using TRCN000062026 are presented in Figure 5. acidic coiled-coil 3 and Aurora-A interact in human thyrocytes and their

& 2014 Macmillan Publishers Limited Oncogene (2014) 4242 – 4252 Spindle aberration by the TACC3 inhibitor R Yao et al 4252 expression is deregulated in thyroid cancer tissues. Endocr Relat Cancer 2007; 14: 30 Holmfeldt P, Stenmark S, Gullberg M. Differential functional interplay of TOGp/ 827–837. XMAP215 and the KinI kinesin MCAK during interphase and mitosis. Embo J 2004; 14 Peters DG, Kudla DM, Deloia JA, Chu TJ, Fairfull L, Edwards RP et al. Comparative 23: 627–637. analysis of ovarian carcinoma and normal ovarian epithelium by 31 Wacker SA, Houghtaling BR, Elemento O, Kapoor TM. Using transcriptome serial analysis of gene expression. Cancer Epidemiol Biomarkers Prev 2005; 14: sequencing to identify mechanisms of drug action and resistance. Nat Chem Biol 1717–1723. 2012; 8: 235–237. 15 Ma XJ, Salunga R, Tuggle JT, Gaudet J, Enright E, McQuary P et al. Gene expression 32 Tulu US, Fagerstrom C, Ferenz NP, Wadsworth P. Molecular requirements for profiles of human breast cancer progression. Proc Natl Acad Sci USA 2003; 100: kinetochore-associated microtubule formation in mammalian cells. Curr Biol 2006; 5974–5979. 16: 536–541. 16 Kiemeney LA, Sulem P, Besenbacher S, Vermeulen SH, Sigurdsson A, Thorleifsson 33 De Brabander M, Geuens G, De Mey J, Joniau M. Nucleated assembly of mitotic G et al. A sequence variant at 4p16.3 confers susceptibility to urinary bladder microtubules in living PTK2 cells after release from nocodazole treatment. Cell cancer. Nat Genet 2010; 42: 415–419. Motil 1981; 1: 469–483. 17 Cullen CF, Ohkura H. Msps protein is localized to acentrosomal poles to ensure 34 Cappell KM, Sinnott R, Taus P, Maxfield K, Scarbrough M, Whitehurst AW. Multiple bipolarity of Drosophila meiotic spindles. Nat Cell Biol 2001; 3: 637–642. cancer testis antigens function to support tumor cell mitotic fidelity. Mol Cell Biol 18 Lee MJ, Gergely F, Jeffers K, Peak-Chew SY, Raff JW. Msps/XMAP215 interacts with 2012; 32: 4131–4140. the centrosomal protein D-TACC to regulate microtubule behaviour. Nat Cell Biol 35 Lens SM, Voest EE, Medema RH. Shared and separate functions of polo-like 2001; 3: 643–649. kinases and aurora kinases in cancer. Nat Rev Cancer 2010; 10: 825–841. 19 Brouhard GJ, Stear JH, Noetzel TL, Al-Bassam J, Kinoshita K, Harrison SC et al. 36 Urbanucci A, Sahu B, Seppala J, Larjo A, Latonen LM, Waltering KK et al. Over- XMAP215 is a processive microtubule polymerase. Cell 2008; 132: 79–88. expression of androgen receptor enhances the binding of the receptor to the 20 Peset I, Vernos I. The TACC proteins: TACC-ling microtubule dynamics and cen- chromatin in prostate cancer. Oncogene 2011; 31: 2153–2163. trosome function. Trends Cell Biol 2008; 18: 379–388. 37 Hayama S, Daigo Y, Kato T, Ishikawa N, Yamabuki T, Miyamoto M et al. Activation 21 Charrasse S, Mazel M, Taviaux S, Berta P, Chow T, Larroque C. Characterization of of CDCA1-KNTC2, members of centromere protein complex, involved in pul- the cDNA and pattern of expression of a new gene over-expressed in human monary carcinogenesis. Cancer Res 2006; 66: 10339–10348. hepatomas and colonic tumors. Eur J Biochem 1995; 234: 406–413. 38 Radulescu AE, Cleveland DW. NuMA after 30 years: the matrix revisited. Trends Cell 22 Gergely F, Draviam VM, Raff JW. The ch-TOG/XMAP215 protein is essential for Biol 2010; 20: 214–222. spindle pole organization in human somatic cells. Dev 2003; 17: 336–341. 39 Bruning-Richardson A, Bond J, Alsiary R, Richardson J, Cairns DA, McCormac L 23 Cassimeris L, Morabito J. TOGp, the human homolog of XMAP215/Dis1, is et al. NuMA overexpression in epithelial ovarian cancer. PLoS One 2012; 7: required for centrosome integrity, spindle pole organization, and bipolar spindle e38945. assembly. Mol Biol Cell 2004; 15: 1580–1590. 40 Kawatani M, Okumura H, Honda K, Kanoh N, Muroi M, Dohmae N et al. The 24 Barr AR, Gergely F. MCAK-independent functions of ch-Tog/XMAP215 in micro- identification of an osteoclastogenesis inhibitor through the inhibition of glyox- tubule plus-end dynamics. Mol Cell Biol 2008; 28: 7199–7211. alase I. Proc Natl Acad Sci USA 2008; 105: 11691–11696. 25 Schneider L, Essmann F, Kletke A, Rio P, Hanenberg H, Wetzel W et al. The 41 Mashima T, Oh-hara T, Sato S, Mochizuki M, Sugimoto Y, Yamazaki K et al. p53- transforming acidic coiled coil 3 protein is essential for spindle-dependent defective tumors with a functional apoptosome-mediated pathway: a new ther- chromosome alignment and mitotic survival. JBiolChem2007; 282: 29273–29283. apeutic target. J Natl Cancer Inst 2005; 97: 765–777. 26 Fielding AB, Lim S, Montgomery K, Dobreva I, Dedhar S. A critical role of integrin- 42 Yao R, Natsume Y, Noda T. TACC3 is required for the proper mitosis of sclerotome linked kinase, ch-TOG and TACC3 in centrosome clustering in cancer cells. mesenchymal cells during formation of the axial skeleton. Cancer Sci 2007; 98: Oncogene 2010; 30: 521–534. 555–562. 27 Yao R, Natsume Y, Saiki Y, Shioya H, Takeuchi K, Yamori T et al. Disruption of 43 Hayashi K, Yonemura S, Matsui T, Tsukita S. Immunofluorescence detection TACC3 function leads to in vivo tumor regression. Oncogene 2012; 31: 135–148. of ezrin/radixin/moesin (ERM) proteins with their carboxyl-terminal threonine 28 Miyazaki I, Simizu S, Osada H. The application of the chemical array for biological phosphorylated in cultured cells and tissues. J Cell Sci 1999; 112(Pt 8): study. Methods Mol Biol 2010; 669:95–107. 1149–1158. 29 Lauffart B, Vaughan MM, Eddy R, Chervinsky D, DiCioccio RA, Black JD et al. 44 Shamma A, Takegami Y, Miki T, Kitajima S, Noda M, Obara T et al. Rb Regulates Aberrations of TACC1 and TACC3 are associated with ovarian cancer. BMC DNA damage response and cellular senescence through E2F-dependent sup- Womens Health 2005; 5:8. pression of N-ras isoprenylation. Cancer Cell 2009; 15: 255–269.

Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)

Oncogene (2014) 4242 – 4252 & 2014 Macmillan Publishers Limited