The tyrosine kinase v‐Src modifies cytotoxicities of anticancer drugs targeting cell division

ABSTRACT v‐Src oncogene causes cell transformation through its strong tyrosine kinase activity. We have revealed that v‐Src‐mediated cell transformation occurs at a low frequency and it is attributed to mitotic abnormalities‐mediated chromosome instability. v‐Src directly phosphorylates Tyr‐15 of cyclin‐dependent kinase 1 (CDK1), thereby causing mitotic slippage and reduction in Eg5 inhibitor cytotoxicity. However, it is not clear whether v‐Src modifies cytotoxicities of the other anticancer drugs targeting cell division. In this study, we found that v‐Src restores cancer cell viability reduced by various microtubule‐targeting agents (MTAs), although v‐Src does not alter cytotoxicity of DNA‐damaging anticancer drugs. v‐Src causes mitotic slippage of MTAs‐treated cells, consequently generating proliferating tetraploid cells. We further demonstrate that v‐Src also restores cell viability reduced by a polo‐like kinase 1 (PLK1) inhibitor. Interestingly, treatment with Aurora kinase inhibitor strongly induces cell death when cells express v‐Src. These results suggest that the v‐Src modifies cytotoxicities of anticancer drugs targeting cell division. Highly activated Src‐induced resistance to MTAs through mitotic slippage might have a risk to enhance the malignancy of cancer cells through the increase in chromosome instability upon chemotherapy using MTAs.

with up-regulation of the CDK inhibitor p21. 7,8 Given that v-Src causes mitotic defects such as chromosome bridge formation and brings about colony formation at a low frequency, v-Src-mediated cell transformation may be attributed to stochastic results of chromosomal abnormalities. [7][8][9] Among the heterogeneous cell population generated by chromosomal abnormalities, survived cells from v-Src-caused inhibition of proliferation may contribute to proliferate to malignant cancers.
Recently, we found a new event of mitotic defects in v-Src-expressing cells 10 : v-Src induces mitotic slippage by directly phosphorylating CDK1 in mitosis, resulting in CDK1 inactivation and premature mitotic exit. Even when cells are treated with S-trityl-L-cysteine (STLC), an inhibitor of the kinesin motor protein Eg5, and arrested at mitosis, v-Src causes mitotic slippage and suppresses Eg5 inhibitor-induced cytotoxic effects. It is known that oncogenes confer resistance to various anticancer drugs through several mechanisms, such as activation of the compensational pathway and genetic alternations of drug target proteins. 11,12 Oncogene-induced mitotic slippage may be a way to develop resistance against anticancer drugs that target cell division in cancer cells.
In this study, we demonstrated that v-Src suppressed the cytotoxicity of several MTAs that cause mitotic arrest, but not anticancer drugs that generate DNA damages. By time-lapse imaging analysis, v-Src caused mitotic slippage in MTAs-treated mitotically arrested cells. Furthermore, we investigated whether v-Src alters the effects of the inhibitors of important mitotic kinases such as PLK1 and Aurora kinases. v-Src slightly suppressed the cytotoxicity of PLK1 inhibitor as well. Surprisingly, the Aurora kinase inhibitor drastically decreased the cell viability of v-Src-expressing cells.

| Antibodies
The primary antibodies used for immunofluorescence (IF), immu-

| Immunofluorescence microscopy
Immunofluorescence staining was performed as described previously. 14 In brief, cells were cultured on coverslips and fixed with 100% methanol at −30ºC for 5 minutes. After permeabilizing and blocking with PBS(−) containing 0.1% saponin and 3% bovine serum albumin for 30 minutes, the cells were incubated with the primary antibody for 1 hour and subsequently with the secondary antibody for 1 hour along with 1 µM Hoechst 33342 for DNA staining. The fluorescence images were obtained using an IX-83 fluorescence microscope (Olympus, Tokyo, Japan) equipped with a ×60 1.42 NA oilimmersion objective lens (Olympus). The optical system included a U-FUNA filter cube (360-370 nm excitation, 420-460 nm emission) and a U-FRFP filter cube (535-555 nm excitation, 570-625 nm emission) to observe Hoechst 33342 and Alexa Fluor 555 fluorescence, respectively. The captured images were edited using Photoshop CC and Illustrator CC software (Adobe).

| Time-lapse imaging
Time-lapse imaging was performed as described previously. 15,16 In brief, HeLa S3/v-Src and HCT116/v-Src cells were cultured with 2 ng/mL and 1 ng/mL Dox, respectively, for 11 hours, in the presence of 0.1 µM Hoechst 33342 to stain DNA during the last 1 hour.
Then, time-lapse imaging was performed using a high-content imaging system (Operetta, PerkinElmer Life Sciences, Waltham, MA) at 37°C in 5% CO 2 .
The cell lysates were subjected to SDS-polyacrylamide gel electrophoresis (PAGE) and electrotransferred onto polyvinylidene difluoride membranes (PVDF; Pall Corporation, Port Washington, NY). After blocking with Blocking One (Nacalai Tesque) at room temperature for 30 minutes, the membranes were incubated with the antibodies, which were diluted with Tween 20-containing Trisbuffered saline [20 mM Tris-HCl (pH 7.5), 137 mM NaCl and 0.1% Tween 20] containing 5% Blocking One. Clarity (Bio-Rad) was used as the chemiluminescence substrate. A ChemiDoc XRSplus image analyser (Bio-Rad) was used for the chemiluminescence detection and band intensity analysis.

| Statistics
Statistical differences between two datasets were analysed using Student's t test after analysis of variance by F test. Statistical differences among more than two datasets were analysed using one-way

| Suppressive effects of v-Src on the cytotoxicity of microtubule-targeting agents
Anti-mitotic drugs have been commonly used as anticancer drugs.
One class of popular anti-mitotic drugs are MTAs, which prolongs mitosis by activating the spindle assembly checkpoint (SAC) and causes mitotic cell death within prolonged mitosis. Recently, we showed that v-Src induces mitotic slippage by inactivating CDK1 in mitosis by directly phosphorylating CDK1 at Tyr-15. 10 The 1-day treatment of STLC, an inhibitor of the mitotic motor kinesin Eg5, sufficiently accumulated mitotic cells, which were detached from surrounding cells and rounded up. The following 2-day culture reduced the cell number, even though the drugs were washed out ( Figure 1A). The number of cells survived from STLC treatment was drastically increased by v-Src whose expression was induced by Dox treatment ( Figure 1B,C), as previously reported. 10 This indicates that v-Src reduces STLC cytotoxicity ( Figure 1C). SAC components are recruited to microtubule-unattached kinetochores through kinetochore-associated kinase MPS1-mediated phosphorylation, and SAC stops chromosome segregation while all kinetochores are not properly attached to microtubules. Therefore, MPS1 inhibition inactivates SAC, leading to premature mitotic progression. Considering that prolonged arrest in the prometaphase-like state through SAC activation is required for the STLC cytotoxicity, it is hypothesized that v-Src ended this arrest by inducing mitotic slippage and affected the STLC cytotoxicity. Based on this, it is expected that agents capable of inhibiting SAC and allowing cells to begin anaphase onset would reduce the cytotoxicity of STLC. Indeed, the MPS1 inhibitor AZ3146 treatment restored the cell viability reduced by STLC as v-Src did ( Figure 1D), confirming that v-Src reduces the STLC cytotoxicity through SAC inactivation.
Given that resolution of mitotic arrest by SAC inactivation may attenuate the cytotoxic effect of MTAs, 18 we here investigated whether v-Src suppressed sensitivity to other MTAs than STLC.
Paclitaxel and vincristine, both clinically used MTAs, reduced cell viability in the absence of v-Src expression ( Figure 1E). Expectedly, v-Src partially prevented paclitaxel-and vincristine-induced reduction in cell viability ( Figure 1F), suggesting that v-Src confers the resistance to MTAs not only to STLC.
Furthermore, to examine whether v-Src affected cell death caused by MTAs, we analysed the population of Sub-G1 cells, including dead DNA-fragmented cells, by flow cytometry. Although STLC or paclitaxel treatment strongly increased the ratios of Sub-G1 cells, these ratios were decreased by v-Src expression in F I G U R E 1 v-Src suppresses microtubule-targeting agents-induced reduction in cancer cell viability. A, (Upper left) Schematic depiction of drug treatment. HeLa S3/v-Src cells were cultured with or without 20 µM S-trityl-L-cysteine (STLC) for 24 hours. After washing STLC out, these cells were further cultured without any drug for 48 hours. (Right) Phase-contrast images were obtained just after STLC treatment (24 hours) and 48 hours after STLC removal (72 hours). Scale bar, 50 µm. B, HeLa S3/v-Src cells were cultured with or without 2 ng/mL doxycycline (Dox) for 24 hours. Whole cell lysates were analysed by Western blotting using anti-Src, anti-active Src (Src pY416) and anti-αtubulin antibodies. Full blots are shown in Figure S2A. C, HeLa S3/v-Src cells were cultured with or without 2 ng/mL Dox in the presence of 20 µM STLC, as shown in A. (Left) Phase-contrast images were obtained 48 hours after indicated drugs removal. Scale bar, 50 µm. D, HeLa S3/v-Src cells were cultured with or without 2.5 µM AZ3146 in the presence of 20 µM STLC, as shown in A. E, HeLa S3/v-Src cells were cultured with or without 0.1 µg/mL paclitaxel (PTX) (left) or 10 µM vincristine (VCR) (right), as shown in A. F, HeLa S3/v-Src cells were cultured with or without 2 ng/mL Dox in the presence of 0.1 µg/mL PTX (left) or 10 µM VCR (right), as shown in A. A, C-F, HeLa S3/v-Src cells were plated at a density of 8.0 × 10 3 cells/well of a 96-well plate. Cell viability was determined by WST-8 assay 48 hours after removal of indicated drugs. Graphs represent the mean ± SD of three independent experiments. Asterisks indicate significant differences (Student's t test, * P < .05; *** P < .001). G-J, HeLa S3/v-Src cells were cultured with or without 2 ng/mL Dox in the presence or absence of 20 µM STLC (G, H) or 0.1 µg/mL PTX (I, J) for 24 hours. Then, the cells were washed and further cultured without any drug for 72 hours. The cells were fixed, stained with propidium iodide (PI) (DNA staining) and subsequently analysed by flow cytometry. DNA histograms and the ratios of Sub-G1 cells are shown, and each plot represents 40,000 cells. The ratios of Sub-G1 cells are plotted, and graphs represent the mean ± SD of three independent experiments. Asterisks indicate significant differences (Games-Howell test, * P < .05) STLC-or paclitaxel-treated cells ( Figure 1G-J). These results suggest that v-Src suppresses the cytotoxicity of MTAs.

| Effect of v-Src on the cytotoxicity of adriamycin and bleomycin
Next, we examined whether v-Src affects the cytotoxicity of anticancer drugs other than MTAs, such as adriamycin and bleomycin.
Adriamycin inhibits topoisomerase II, thereby causing doublestrand breaks. Bleomycin induces free-radical-mediated DNA single-and double-strand breaks. Consequently, these drugs induce cell death. Actually, adriamycin and bleomycin treatment reduced cell viability in a concentration-dependent manner (Figure 2A,B).
In this experiment, the decreased number of cells (3.0 × 10 3 cells/ well in a 96 well plate) was seeded compared with that used in

| Induction of mitotic slippage by v-Src in microtubule-targeting agents-induced mitotic arrest
Our previous work showed that v-Src causes cytokinesis failure and mitotic slippage, generating tetraploid cells. 7,10 To investigate whether v-Src brought about mitotic slippage in STLC-treated mitotically arrested cells, we first determined the DNA contents and cyclin B1 levels by flow cytometry. The cyclin B1 protein level is gradually increased from G2 phase to M phase, leading to complex formation with CDK1 kinase and its activation, 19 which is necessary for the mitotic entry. For the mitotic exit, cyclin B1 should be degraded, thereby inactivating CDK1 activity. When cells undergo abnormal mitosis, such as cytokinesis failure and mitotic slippage, and resume the next cell cycle, these tetraploid cells exhibit a 4N DNA content with lower cyclin B1 level in G1 phase (4N-G1). As shown in Figure 3A, in controls cells (None), cyclin B1 level was increased in 4N cells compared with 2N cells. The ratio of 4N cells with lower cyclin B1 levels was increased upon v-Src induction (25%, v-Src), suggesting that tetraploid cells are increased as previously described. 7 The ratio of 4N cells with lower cyclin B1 levels was strongly in-   Figure 1A. Cell viability was determined by WST-8 assay 48 hours after removal of indicated drugs. Graphs represent the mean ± SD of three independent experiments. Asterisks indicate significant differences (Tukey's test, * P < .05; *** P < .001; N.S., not significant) | 1683 F I G U R E 3 v-Src-induced mitotic slippage in microtubule-targeting agents-treated mitotically arrested cells. A-C, HeLa S3/v-Src cells were cultured with or without 2 ng/mL Dox in the presence or absence of 20 µM STLC for 24 hours. Then, the cells were washed and further cultured without any drug for 72 hours. A, The cells were fixed, doubly stained with anti-cyclin B1 and PI, and subsequently analysed by flow cytometry. The bivariate dot plots of cyclin B1 protein level (y-axis, log scale) and DNA content (x-axis, linear scale) are shown together with DNA histograms. The ratios of cells are shown: 4N cells with lower (red solid line) and higher (red dotted line) cyclin B1 levels and 8N cells with higher cyclin B1 levels (black solid line). The ratios of 2N and 4N cells, sub-G1 cells, and polyploid cells (>4N) are indicated. Each plot represents 4,000 cells. B, C, The cells were fixed and doubly stained with anti-γ-tubulin and Hoechst 33342 (DNA staining). B, Representative images were indicated. Magnified images of the areas enclosed by dotted lines are shown. Scale bars, 10 μm. C, (Left) Ratios of living cells having more than two nuclei per single cell were calculated. (Right) Ratios of living cells having more than three γ-tubulin foci per mononucleated single cell were calculated. Graphs represent the mean ± SD of three independent experiments. Asterisks indicate significant differences (Tukey's test, * P < .05, ** P < .01, *** P < .001, N.S., not significant). D, E, HeLa S3/v-Src cells were cultured with or without 2 ng/mL Dox in the presence or absence of 20 µM STLC for 11 hours and then monitored for 9 hours by time-lapse imaging. D, Representative images of mitotic cells are shown: cells that exhibit normal mitosis (a, normal mitosis), the cleavage furrow regression after furrow ingression (b, binucleation) and the mitotic exit without chromosome segregation (c, slippage). Scale bars, 10 µm. E, The duration of each mitotic phase is shown: prophase/prometaphase/metaphase (P/PM/M: from nuclear envelope breakdown to chromosome alignment; red), anaphase/telophase (A/T: from anaphase onset to cleavage furrow ingression; blue), binucleation (Bi: binucleated interphase cells; yellow) and slippage (Sl: interphase cells that undergo mitotic slippage; green). In total, 23-50 mitotic cells were examined. Several cells were already arrested in mitosis at the beginning of time-lapse imaging: 24 cells (STLC) and 13 cells (STLC + v-Src). F, HCT116/v-Src cells were cultured with or without 1 ng/mL Dox for 24 hours. Whole cell lysates were analysed by Western blotting using anti-Src, anti-active Src (Src pY416) and anti-α-tubulin antibodies. Full blots are shown in Figure S2B Figure 1A. Cell viability was determined by WST-8 assay 48 hours after removal of the indicated drugs. Graphs represent the mean ± SD of three independent experiments. Asterisks indicate significant differences (Games-Howell test, * P < .05; ** P < .01; N.S., not significant). B, (Left) Phase-contrast images were obtained 48 hours after the indicated drugs removal. Scale bar, 50 µm. C, HeLa S3/v-Src cells were cultured with or without 2 ng/mL Dox in the presence or absence of 10 µM ZM for 24 hours. Then, the cells were washed and further cultured without any drug for 48 hours. The cells were stained with Annexin V and PI as described under 'MATERIALS AND METHODS' and analysed by flow cytometry. The bivariate dot plots of Annexin V level (y-axis, log scale) and PI level (x-axis, log scale) are shown. The regions designated by red dotted lines include Annexin V-positive / PIpositive cells, and the ratios within the region are shown. Each plot represents 25,000 cells mitotic slippage, condensed chromosomes were de-condensed without their alignment and segregation after prolonged mitosis ( Figure 3D, panel c). Compared with controls cells whose average duration of mitosis is 41 minutes (Figure 3E: None), v-Src induction prolonged mitosis (99 minutes) with binucleation and mitotic slippage ( Figure 3E: v-Src), as previously described. 10 Figure 4C). These results suggest that Aurora kinase inhibition exerts a cytotoxic effect against v-Src-expressing cancer cells.

| D ISCUSS I ON
Here, we show that v-Src restores cancer cell viability that is reduced by the microtubule-targeting drugs STLC, paclitaxel and vincristine.
Similarly, v-Src slightly restores cancer cell viability in cells treated with an inhibitor of PLK1, a core mitotic regulator. On the other hand, this does not occur in cells treated with adriamycin and bleomycin, which are not anti-mitotic drugs. Time-lapse imaging clearly shows that v-Src causes mitotic slippage upon MTAs-induced mitotic arrest, thereby increasing the ratio of mononucleated cells with excess centrosomes, but not that of multinucleated cells. Similar to v-Src induction, the MPS1 inhibitor restores cell viability reduced by STLC, suggesting that v-Src-mediated override of SAC leads to cell survival. Interestingly, an inhibitor of Aurora kinase, another core mitotic regulator, causes strong cell death and drastically reduces the cell viability of v-Src-expressing cells. Therefore, to effectively use anticancer drugs targeting mitosis, Src activity in cancer cells should be taken into consideration.
We previously reported that v-Src causes mitotic defects, such as chromosome bridge formation, cytokinesis failure and mitotic slippage. 7,9,10 26 Considering the present study of v-Src effect on MTAs, it is possible that highly activated c-Src may also cause mitotic slippage. We recently reported that knockdown of Csk, an inhibitory kinase of Src, slightly but significantly induces mitotic slippage. 10 Therefore, highly activated c-Src may contribute to the suppression of cytotoxic effects of MTAs, leading to cancer malignancy at the worst case.
v-Src directly phosphorylates Tyr-15 of CDK1, whose phosphorylation inhibits its kinase activity, leading to mitotic slippage through CDK1 inactivation after mitotic entry. 10 Here, we used the Eg5 inhibitor STLC, the microtubule stabilizer paclitaxel and the microtubule-depolymerizing drug vincristine. The common consequence among these agents is mitotic arrest due to SAC activation.
Prolonged mitosis gradually increases the apoptotic signal, thereby causing cell death. 27 Mitotic slippage, another fate during mitotic arrest, is thought to be an evasive way to survive from the mitotic arrest. 28 Indeed, our previous work showed that CDK1 inactivation induced mitotic slippage in cells mitotically arrested by STLC, thereby reducing the cytotoxicity of STLC. 10 Although MTAs caused cell death and reduced cell viability, v-Src suppressed that cytotoxicity ( Figure 1). Furthermore, time-lapse imaging analysis showed that MTAs cause strong mitotic arrest without chromosome alignment; however, v-Src induces mitotic slippage in more than half of cells after mitotic arrest within the analysis ( Figure 3E,G). Therefore, v-Src-mediated CDK1 inactivation contributes to mitotic slippage in MTAs-treated mitotically arrested cells, thus reducing MTA cytotoxicity and promoting cancer cell survival.
Recent reports showed that many cancer cells show tetraploid or near-tetraploid phenotypes, 29 and these characteristics generally link to tumour malignancy and unfavourable prognosis in patients. 30,31 Tetraploid cell formation is accompanied by centrosome duplication, and excess centrosome-mediated multipolarity gives rise to chromosome instability through asymmetrical chromosome segregation. 32 Tetraploid cells are generated by mitotic defects, such as mitotic slippage and cytokinesis failure; anti-mitotic drugs also contribute to generate those cells. 33 It is known that tetraploid cells undergo cell cycle arrest at G1 phase by the tetraploidy checkpoint, 34 which prohibits cells from entering the next cell division cycle. In tetraploidy checkpoint activation, excess centrosomes and cytoskeleton abnormalities trigger Rac activation, thereby activating the tumour suppressor LATS2, which is a Hippo pathway regulator.
Activated LATS2 sequestrates nuclear YAP in the cytoplasm and activates p53 by Mdm2 inactivation. 35 Our recent research showed that v-Src inactivates tetraploidy checkpoint, leading to nuclear accumulation of YAP in multinucleated cells. 36 In the present study, flow cytometry analysis showed that cells having 8N DNA content and higher cyclin B levels were observed when v-Src-expressing cells were treated with STLC, suggesting a possibility that v-Src-expressing tetraploid cells can evade from tetraploidy checkpoint and continue to proliferate. MTAs might induce chromosome instability and lead to tumour malignancy in cancer cells expressing Src with high activity.
We found that the Aurora kinase inhibitor induced severe cell death in v-Src-expressing cells ( Figure 4B,C). What is the mechanism of cell death promotion by v-Src in the Aurora kinase inhibitor-treated cells? The Aurora kinase inhibitor ZM447439 inhibits the kinase activities of Aurora A and Aurora B kinases at a higher concentration. 37 Aurora A localizes to the centrosome and exerts chromosome congression. 38 Kinetochore-localized Aurora B activates SAC while all kinetochores are not properly attached to microtubules, and midzone-localized Aurora B regulates cleavage furrow ingression to complete the cell division. 39 Although chromosome congression is attenuated in ZM447439-treated cells, mitosis progresses due to SAC inactivation. Furthermore, the cleavage furrow cannot be formed, resulting in cytokinesis failure and generation of binucleated cells. 37,40 Indeed, time-lapse imaging revealed that ZM447439-treated cells progressed mitosis without mitotic arrest and cleavage furrow ingression, resulting in binucleation ( Figure S1B). Given that v-Src causes Aurora B delocalization from the spindle midzone, 7 cytokinesis failure may be strongly induced by ZM447439 treatment of v-Src-expressing cells. It was reported that ZM447439 generates polyploid cells and causes apoptosis through the mitochondria pathway. 41 Furthermore, v-Src is known to generate mitochondria-dependent apoptotic signals, 42 and overexpression of mitochondria-localized c-Src reduces the cell viability through reduction of mitochondria activity by phosphorylation of mitochondrial single-stranded DNA-binding protein. 43 These studies suggest a possibility that v-Src induction together with ZM447439 treatment may accelerate apoptotic signalling in the mitochondria of polyploid cells, leading to severe cell death.
Recent reports showed that oncogenic events, such as KRAS mutation and MYC overexpression, causes mitotic defects, and oncogene-induced chromosome instability is thought to be attributed to those mitotic abnormalities. 44,45 Our reports regarding v-Src oncogene also supports the idea that there is a relationship between oncogenic mitotic abnormalities and genome instability. Oncogenes frequently confer resistance to anticancer drugs through several ways in cancer cells. Similar to v-Src that causes mitotic slippage to evade from mitotic arrest caused by various MTAs, it is possible that other oncogenes may induce mitotic slippage as a different way of acquiring resistance against anticancer drugs.
In conclusion, we show that v-Src oncogene causes mitotic slippage in MTAs-treated mitotically arrested cells, thereby suppressing the cytotoxicity of those agents. The resulting tetraploid cells with excess number of centrosomes have a potential to induce structural and numerical chromosome instability, increasing cancer cell malignancy through stochastic genetic alterations.

ACK N OWLED G EM ENTS
We are grateful to Dr. Hiroshi Ohnishi for the valuable plasmids.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used to support findings of the study are available from the corresponding author upon reasonable request.