Oncogene (2008) 27, 976–984 & 2008 Nature Publishing Group All rights reserved 0950-9232/08 $30.00 www.nature.com/onc ORIGINAL ARTICLE Ectopic Tbx2 expression results in polyploidy and cisplatin resistance

E Davis1, H Teng2, B Bilican3, MI Parker2, B Liu3, S Carriera3, CR Goding3 and S Prince1

1Division of Cell Biology, Institute of Infectious Disease and Molecular Medicine, Faculty of Health Sciences, University of Cape Town, Cape Town, Western Province, South Africa; 2Division of Medical Biochemistry, Faculty of Health Sciences, Institute of Infectious Disease and Molecular Medicine, University of Cape Town, Cape Town, Western Province, South Africa and 3Signalling and Development Laboratory, Marie Curie Research Institute, Oxted, Surrey, UK

T-box factors play critical roles in embryonic development development and mutations affecting their function can and have been implicated in cell cycle regulation and cancer. result in human disorders with dramatic phenotypes For example, Tbx2 can suppress senescence through a (Naiche et al., 2005). In addition, T-box factors have mechanism involving the repression of the cyclin-depen- been implicated in cell cycle regulation and in the genesis dent kinase inhibitors, p19ARF and p21WAF1/CIP1/SDII, and the of cancer. Both Tbx2 and Tbx3, for example, can Tbx2 is deregulated in melanoma, breast and prevent senescence in mouse embryonic fibroblasts and pancreatic cancers. In this study, several transformed ST.HdhQ111 striatal cells through a mechanism involving human lung fibroblast cell lines were shown to down- their ability to repress the cyclin-dependent kinase regulate Tbx2. To further investigate the role of Tbx2 in inhibitors p19ARF (Jacobs et al., 2000; Carlson et al., oncogenesis we therefore stably reexpressed Tbx2 in one 2001; Brummelkamp et al., 2002) and p21WAF1/CIP1/SDII such cell line. Compared to their parental cells, the (referred to as p21) (Prince et al., 2004). Ectopic resulting Tbx2-expressing cells are larger, withbinucleate expression of Tbx3 together with oncogenic Ras or and lobular nuclei containing double the number of in embryonic mouse fibroblasts leads to cellular . Moreover, these cells had an increase in transformation and suppression of apoptosis (Carlson frequency of several features of genomic instability such et al., 2002). Furthermore, both Tbx2 and Tbx3 are as missegregation, chromosomal rearrange- expressed in the developing breast (Jerome-Majewska ments and polyploidy. While grossly abnormal, these cells et al., 2005) and their are amplified and/or still divide and give rise to cells that are resistant to the overexpressed in some breast tumors (Sinclair et al., chemotherapeutic drug cisplatin. Furthermore, this is 2002; Packham and Brook, 2003) and in certain breast shown to be neither species nor cell type dependent, as cancer cell lines (Wu et al., 2002; Fan et al., 2004). Tbx2 ectopically expressing Tbx2 in a murine melanoma cell is also overexpressed in 50% of pancreatic cancer cell line also induce mitotic defects and polyploidy. These lines (Mahlamaki et al., 2002) and in melanomas where results have important implications for our understanding it was shown to function as an anti-senescence factor of the role of Tbx2 in tumorigenesis because polyploidy (Vance et al., 2005). These results suggest that T-box frequently precedes aneuploidy, which is associated with factors may contribute to oncogenesis through suppres- high malignancy and poor prognosis. sing senescence but whether additional mechanisms are Oncogene (2008) 27, 976–984; doi:10.1038/sj.onc.1210701; involved remains to be elucidated. published online 13 August 2007 Genomic instability, a hallmark of transformed cells, often culminates in aneuploidy as a result of failure to Keywords: Tbx2; T-box factor; polyploidy; chromosomal execute proper cell division. Signaling checkpoints that instability elicit temporary delays in cell cycle progression serve as an important defense against genomic instabi- lity (Elledge, 1996; Paulovich et al., 1997). These checkpoints, which involve a combination of cyclins, Introduction cyclin-dependent kinases and cyclin-dependent kinase inhibitors, can be activated during G1,G2 or the mitotic The T-box family is characterized spindle assembly checkpoints (Rieder and Maiato, 2004; by a highly conserved DNA-binding domain, the T box, Huang et al., 2005). In particular failure to detect which binds specific promoter sequences in target genes. specific mitotic defects has been associated with poly- Members of this family play a crucial role in embryonic ploidy and cancer. These defects include the missegrega- tion of defective chromosomes (for example, unrepaired DNA), misattached kinetochore or a defective mitotic Correspondence: Dr S Prince, Department of Human Biology, Faculty spindle. Ordinarily, when these defects are detected, a of Health Sciences, University of Cape Town, Anzio Road, cell may undergo mitotic catastrophe, a process invol- Observatory, Western Province 7925, South Africa. E-mail: [email protected] ving the generation of binucleate, tetraploid cells which Received 10 April 2007; accepted 20 June 2007; published online 13 are eliminated and hence prevented from reentering the August 2007 cell cycle in a p21- and -dependent manner (Bunz Ectopic expression of Tbx2 results in polyploidy E Davis et al 977 et al., 1998; Andreassen et al., 2001). In transformed BrdU-positive cells expressed as a percentage of the cells, however, these tetraploid cells are able to reenter total number of cells from 10 fields of view for each cell the cell cycle through bypassing the relevant check- line. Whereas, similar numbers of CT-E cells had points. Genomic instability is thought to drive tumor- incorporated BrdU as the CT-Tbx2 cell lines at 1 h, igenesis and the resulting increase in DNA content after 8 h 97% of the CT-E cells had incorporated BrdU promotes the ability to adapt to changing physiological compared to 76 and 48% for the CT-Tbx2(2) and CT- conditions and the emergence of chemotherapeutic Tbx2(3), respectively. These results thus confirm that drug-resistant cells (Nigg, 2002). Tbx2 does indeed have a negative effect on cell Although Tbx2 has been reported to be upregulated proliferation. Furthermore, we observed that the CT- in a variety of cancers, this study shows that while Tbx2 Tbx2 cells had a morphology that was significantly is expressed in normal lung fibroblasts it is down- different to that of the CT-E and untransfected cells. regulated in several of its transformed counterparts. To The CT-Tbx2 cells were larger and their cell borders less investigate the implication of this downregulation of defined (Figure 1e). Tbx2 upon cellular transformation, Tbx2 was reex- pressed in a transformed lung fibroblast cell line. These Tbx2-expressing cells experience an initial block in Ectopic Tbx2 expression induces a G2/M arrest cytokinesis. With increasing passage, however, the cells Several of the CT-Tbx2 cells had large nuclei or were acquire the ability to bypass this block, resulting in the binucleate suggesting that they fail to undergo cytokin- formation of genetically unstable polyploid cells with esis. This was confirmed by flow cytometry, which increased resistance to the chemotherapeutic drug revealed that Tbx2-expressing cells had a cell cycle cisplatin. profile very different to normal cycling cells as seen for the CT-E cells (Figure 1f). Whereas 73% of the CT- Tbx2 cells had a DNA content of 4n or greater, only 15% of the CT-E cells were 4n. These results suggested Results that the Tbx2-expressing cells were arrested at either a G2 or M checkpoint. We therefore explored the Reexpression of Tbx2 in CT-1 cells results in a decrease in possibility that Tbx2 may be exerting an effect on key cell proliferation rate and an altered morphology cell cycle proteins involved in regulating these two We have previously shown that Tbx2 mRNA and checkpoints by comparing the levels of p53, p21, cyclin protein is expressed in normal WI-38 lung fibroblast cell B1 and p14ARF (human homolog of p19ARF) in CT-E and lines, but not in the SV40-transformed WI-38 (SV40WI- CT-Tbx2 cells. No detectable differences in the levels of 38), the Co-60 gamma irradiated WI-38 (CT-1) and the the p53, p21 and cyclin B1 proteins were seen when they human fibrosarcoma (HT1080) cell lines (Teng et al., were standardized to tubulin levels in western blot 2007). analyses (Figure 1g). Interestingly, p14ARF levels were To understand the significance of Tbx2 downregula- higher in the CT-Tbx2 cells, which could be as a result of tion in the transformation of human lung fibroblasts, an the cells being blocked in G2. Furthermore, as shown in expression construct in which the cytomegalovirus Figure 1f (see arrow), Tbx2 induced apoptosis in a small promoter drives expression of human Tbx2 was number of CT-Tbx2 cells as indicated by a sub-G1 peak. transfected into CT-1 cells. A number of G418-resistant We next tested whether ectopically expressing Tbx2 in clones were tested for Tbx2 expression. Figure 1a shows another cell type would result in similar defects seen in the presence of Tbx2 protein in two representative the CT-Tbx2 cells. For these experiments we chose the clones (CT-Tbx2(2) and CT-Tbx2(3)), which is absent in K1735 murine melanoma cell line that does not express cells transfected with the empty vector (CT-E) and the endogenous Tbx2 and infected them with a pBabePuro parental CT-1 cells. Figure 1b shows that WI-38 cells retrovirus expressing SV5-tagged Tbx2. Following have much higher levels of Tbx2 protein compared to selection with puromycin individual clones were isolated the CT-Tbx2 clones and therefore any subsequent and examined for Tbx2 expression by reverse transcrip- changes observed in the CT-Tbx2 cells were not due to tion (RT)–PCR (Figure 2a). We were unable to detect abnormally high levels of Tbx2. Tbx2 protein levels using an anti-Tbx2 antibody, While expanding the CT-E and CT-Tbx2 cell lines, it indicating that Tbx2 protein is expressed at low levels was observed that the Tbx2-expressing cells grew much compared to cell lines that naturally express endogenous slower than the CT-E cells. To explore this further, the Tbx2. Although several clones were isolated all exhib- growth rate of these cell lines was compared using a ited a similar phenotype and the results obtained for hemocytometer to count cells over a 9-day culture two, clones 4 and 10, are presented. As seen in the CT- period. In keeping with our observations, the CT-Tbx2 Tbx2 cell lines, both of the K1735 cell lines expressing cells had a decreased cell proliferation rate (Figure 1c), Tbx2 analysed proliferated at a slower rate than the growing approximately 2.5 times more slowly than the parental cells (Figure 2b) and exhibited an altered CT-E cells on day 9. To confirm the effect seen for Tbx2 morphology (Figure 2c). Furthermore, flow cytometry on cell proliferation we performed 5-bromo-2-deoxyur- (Figure 2d) revealed that the K1735-Tbx2 cells display idine (BrdU) incorporation assays in which CT-E and an abnormal cell cycle profile with a substantial increase the two CT-Tbx2 cell lines were pulsed with BrdU in the G2 fraction and in some instances the emergence for 1 and 8 h. Figure 1d shows the total number of of an 8n (G2/M tetraploid) peak was also seen

Oncogene Ectopic expression of Tbx2 results in polyploidy E Davis et al 978 abeCT-E CT-Tbx2 CT-1 CT-E CT-Tbx2(2) CT-Tbx2(3) CT-E CT-Tbx2(2) WI-38

Tbx2 Tbx2

Tubulin Tubulin Western blot Western blot

c 175000 d CT-E 120 150000 CT-E CT-E(2) 100 125000 CT-Tbx2(2) CT-Tbx2(2) 80 CT-Tbx2(3) CT-Tbx2(3) 100000 60 (%) 75000 40

Cell Number 50000 20

25000 BrdU Incorporation 0 0 1 8 0123456789 BrdU Incubation (Hours) Time (Days) f g CT-E CT-Tbx2(2) CT-E(2) CT-E CT-E(2) CT-Tbx2(2) CT-Tbx2(3) CT-E CT-Tbx2(2) CT-Tbx2(3)

p53 p21

Number of events ARF

Number of events cyclinB1 p14 2N 4N 2N 4N DNA content DNA content Tubulin Tubulin

Figure 1 Ectopic Tbx2 expression affects cell proliferation and morphology. (a) Establishment of CT-E and CT-Tbx2 cell lines. Western blot analyses were used to confirm that the established CT-Tbx2 cell lines express Tbx2 protein. (b) Tbx2 protein levels in the CT-Tbx2 clones are much lower than the endogenous Tbx2 in WI-38 normal lung fibroblasts. Western blot analysis was used to compare levels of Tbx2 protein in CT-Tbx2(2) and WI-38 cell lines. (c) CT-Tbx2 cells have reduced proliferative ability. Growth curve assays performed over a 9-day period, with cells seeded in triplicate and counted on a hemocytometer. Solid lines represent two CT-Tbx2 clones and dashed lines two CT-E clones. Results represent the means7s.d. of two individual experiments. (d) BrdU incorporation assays in which cells were pulsed with BrdU for 1 and 8 h and BrdU-positive nuclei visualized by fluorescence microscopy. Results show an average of 10 fields of view with BrdU-positive cells being expressed as a percentage of total cells counted. White bars represent CT-E cells, gray bars CT-Tbx2(2) cells and black bars CT-Tbx2(3) cells. (e) CT-Tbx2 cells are larger and flatter than CT-E cells. Phase contrast images of CT-E (left panel) and CT-Tbx2 (right panel) cells photographed at indicated magnifications. (f) CT-Tbx2 cells are initially blocked in G2/M. The cell cycle status in CT-E and CT-Tbx2 cell lines was determined by measuring their DNA content using fluorescence-activated cell sorting analyses. Arrow points to CT-Tbx2 cells that show a sub-G1 peak characteristic of cells undergoing apoptosis. (g) Tbx2 expression does not affect the levels of key cell cycle regulators, but results in increased p14ARF levels. Western blot analyses with cell extracts from CT-E and CT-Tbx2 cell lines were used to compare the indicated proteins using antibodies listed in ‘Materials and methods’.

(Figure 2d, right hand histogram), further indicating a do indeed have double the number of chromosomes defect in mitosis. These results confirmed that the effects compared to that contained in CT-E cells. Metaphase of Tbx2 in CT-Tbx2 cells were not species or cell type spreads (Figure 3b) confirmed that all CT-Tbx2 cells dependent. analysed had at least 160 chromosomes compared to a maximum of 85 chromosomes seen in CT-E cells. As expected, the normal WI-38 cells had 46 chromosomes. Ectopic Tbx2 expression induces mitotic defects and The observation that CT-Tbx2 cells have an increased polyploidy number of chromosomes suggested that overexpression With increasing passage of the CT-Tbx2 cell lines, of Tbx2 may be enhancing chromosomal instability fluorescence-activated cell sorting (FACS) analyses (CIN). Indeed, results from experiments in which revealed the gradual appearance of an 8n (G2/M Giemsa-banding analysis was performed on metaphase tetraploid) peak (Figure 3a). These results are in chromosomes reveal that compared to CT-E cells CT- agreement with previous reports that checkpoint activa- Tbx2 cells have an increase in the frequency of several tion is often only transient, with some cells ‘slipping’ chromosomal abnormalities indicative of CIN past the arrest and producing a tetraploid population (Figure 3c). For example whereas dicentric and giant due to defective cell division (Storchova and Pellman, marker chromosomes were never seen in CT-E control 2004). We next investigated whether the CT-Tbx2 cells cells, 78% of the CT-Tbx2 metaphase spreads had

Oncogene Ectopic expression of Tbx2 results in polyploidy E Davis et al 979 K1735 80 K1735-Tbx2(4) 5 70 60 K1735-Tbx2(10) 50 K1735-Tbx2(10) K1735 K1735-Tbx2(4) 40 Tbx2 30 20

cell number x 10 cell number 10 0 G3PDH 012345678 RT-PCR Time (Days)

K1735 K1735-Tbx2(4) K1735-Tbx2(10)

K1735 K1735-Tbx2(4) K1735-Tbx2(10) Number of events Number of events Number of events

2N 4N 2N 4N 2N 4N 8N DNA content DNA content DNA content

Figure 2 Ectopic expression of Tbx2 in murine K1735 cells result in similar defects seen in the CT-Tbx2 cells. (a) Establishment of K1735-Tbx2(4) and of K1735-Tbx2(10) cell lines. Semi-quantitative RT–PCR was used to show expression of Tbx2 mRNA in the K1735-Tbx2(4) and K1735-Tbx2(10) cell lines compared to undetectable levels in the K1735 parental cells. (b) Growth curve assays performed over an 8-day period with cells counted on a hemocytometer show that the Tbx2-expressing cells proliferate at a slower rate than the parental cells. Dashed lines represent two K1735-Tbx2 clones and solid lines the parental cells. (c) Tbx2-expressing K1735 cells are larger than the parental cells, as seen in phase contrast images of K1735 (left panel), K1735-Tbx2(4) (middle panel) and K1735-Tbx2(10) (right panel) cells. (d) K1735-Tbx2 cells are initially blocked in G2/M, with the emergence of a G2/M tetraploid (8n) population. The cell cycle status in K1735 and K1735-Tbx2 cell lines was determined by measuring their DNA content using fluorescence-activated cell sorting analyses.

between one and three dicentric chromosomes (arrows) Polyploid cells with CIN can arise from a wide variety and 44% displayed giant marker chromosomes (D). In of cell cycle defects including chromosome missegrega- addition, while metaphase spreads of both the CT-E tion, defective mitotic spindle function and defective control and CT-Tbx2 cells contained very small marker cytokinesis (reviewed by Storchova and Pellman, 2004). chromosomes they occurred with twice the frequency in Therefore, to identify defects in CT-Tbx2 cells that the CT-Tbx2 cells (arrowheads). Furthermore, by could have triggered mitotic catastrophe and subsequent comparison to the CT-E control cells, CT-Tbx2 cells polyploidy, we examined anaphase and telophase had almost no apparently normal chromosomes. Nu- chromosomes of CT-Tbx2 cells by immunofluorescence merous derivative and unidentifiable chromosomes (n), microscopy. Figure 3e shows examples of trailing/ which must have resulted from translocations or lagging chromosomes and chromatin bridges in between chromosomal fusions, predominated. CT-Tbx2 cells separating anaphase chromosome masses (anaphase also displayed nuclear abnormalities characteristic of bridges), which was seen in, on average, 46% of cells CIN. For example, the presence of micronuclei was undergoing mitosis. These mitotic defects may be the detected in 23% of cells analysed and many cells were reason for the initial cell cycle block seen in CT-Tbx2 binucleate or had large lobular nuclei (Figure 3d), with cells because anaphase bridges present a physical barrier the average size of CT-Tbx2 nuclei being 0.17 mm2 to furrow formation that explain why the cells are compared to an average of 0.11 mm2 obtained for unable to undergo successful cytokinesis. We also the CT-E cells. examined the effect of transiently expressing Tbx2 in

Oncogene Ectopic expression of Tbx2 results in polyploidy E Davis et al 980 a b WI-38 CT-E CT-Tbx2 CT-E CT-Tbx2 ~ 46 ~ 80 ~ 160 chromosomes chromosomes chromosomes Number of events Number of events 2N4N 2N 4N 8N DNA content DNA content c d CT-Tbx2 CT-E Der(10)

e

Figure 3 Tbx2 expression results in cells exhibiting several features of polyploidy and chromosomal instability (CIN). (a) With increasing passage CT-Tbx2 cells undergo mitotic slippage resulting in the emergence of a G2/M tetraploid (8n) population. The cell cycle status of the indicated cell lines was determined by DNA staining and fluorescence-activated cell sorting analyses. (b) CT-Tbx2 cells have double the DNA content of CT-E cells. Metaphase spread assays were performed to confirm that CT-Tbx2 cells have a DNA content of 8n. (c) CT-Tbx2 cells have enhanced CIN. Giemsa-banding analysis for CT-Tbx2 cells reveal dicentric chromosomes (arrows), giant marker chromosomes (D), very small marker chromosomes (arrowheads), unidentifiable or marker chromosomes (n) and a derivative of chromosome 10 (Der(10)). (d) CT-Tbx2 cells display nuclear abnormalities indicative of polyploidy and CIN. Phase-contrast images (upper panel) of CT-Tbx2 cells at  40 magnification with corresponding immunofluorescence images of DAPI- labeled nuclei (lower panel). Arrowheads indicate examples of a lobular nucleus, a binucleate cell and a multinucleate cell with micronuclei. (e) Anaphase CT-Tbx2 cells display trailing chromosomes and anaphase bridges. To determine the effect of Tbx2 expression on chromosomal segregation the DNA of CT-E and CT-Tbx2 cells was stained with propidium iodide and the cells viewed by immunofluorescence. Lagging chromosomes and anaphase bridges seen in CT-Tbx2 cells are shown with arrows.

other cell lines with no endogenous Tbx2 and the concentrations of 17, 16 and 12 mM cisplatin, respec- resulting cells also exhibit lagging chromosomes and tively. To exclude the possibility that the apparent anaphase bridges as described in this study (our increased resistance of CT-Tbx2 to cisplatin was due to unpublished observations). a percentage of the cells not dividing BrdU, incorpora- tion assays were carried out. The cells were grown with or without 50 mM cisplatin for 48 h in the presence of Effect of Tbx2 overexpression on cisplatin resistance BrdU and processed for BrdU incorporation by micro- Reports have shown a relationship between genetic scopy. Figure 4c shows that in the absence of the drug all alterations, such as gene amplification and polyploidy, of the CT-E and CT-Tbx2 cells have incorporated BrdU in tumors and the development of resistance to and have therefore divided at least once. Furthermore, in chemotherapeutic agents (Rothenberg and Ling, 1989; the presence of 50 mM cisplatin there are considerably Baroja et al., 1998). We therefore compared the effect of more viable BrdU-positive CT-Tbx2 cells compared to the anticancer drug, cisplatin, on the growth rate of the the CT-E cells (Figure 4d). These results confirmed that CT-Tbx2 and CT-E cells. The cells were cultured for CT-Tbx2 cells that survive cisplatin treatment are indeed 72 h, either in the presence or absence of cisplatin, and dividing. Taken together, the results show that Tbx2- cell viability measured using the 3-(4,5-dimethylthiazol- expressing cells are more resistant to the anticancer drug 2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. As cisplatin than the control cells. shown in Figure 4a, while 50 mM cisplatin killed all the CT-E cells, approximately 15% of CT-Tbx2 cells (as compared to untreated cells) continued to survive. Importantly, even at 166.7 mM of cisplatin approxi- Discussion mately 8% of CT-Tbx2 cells were still viable (results not shown). Fifty percent growth inhibition (IC50)intwo Several studies have implicated Tbx2 in tumorigenesis CT-Tbx2 clones and CT-E cells (Figure 4b) was noted at but whether it involves its anti-senescence function or

Oncogene Ectopic expression of Tbx2 results in polyploidy E Davis et al 981 0.35 CT-E 0.30 IC50 = 11.82 0.25 CT-Tbx2(2) IC50 = 17.36 0.20 20 CT-Tbx2(3) 0.15 15 IC50 = 16.31 10 0.10 5 0.05 CELLS 0 0.00 Corrected OD [595nm] –5 –0.05 % OF UNTREATED CT-Tbx2(2) CT-Tbx2(3) CT-E –4 –3 –2 –10123 CELL LINES Log [concentration]

0 µM cisplatin 50 µM cisplatin

CT-E CT-Tbx2(2) CT-Tbx2(3) CT-E CT-Tbx2(2) CT-Tbx2(3)

Figure 4 CT-Tbx2 cells have increased resistance to the chemotherapeutic drug cisplatin. (a and b) Two CT-Tbx2 clones and a CT-E cell line were incubated with increasing concentrations of cisplatin for 72 h and cell viability determined using the 3-(4,5- dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay. (a) Cell survival, expressed as a percentage of untreated cells, when incubated with 50 mM cisplatin. (b)IC50 curves showing 50% growth inhibition in the indicated cell lines. (c and d) Cells were grown in the presence of BrdU and 0 or 50 mM cisplatin for 48 h to determine whether the CT-E and CT-Tbx2 cell lines were dividing during the cisplatin treatment. Fluorescence microscopy shows that (c) in the absence of the drug all CT-E and CT-Tbx2 cells incorporate BrdU and (d) after treatment with 50 mM cisplatin substantially more CT-Tbx2 cells are viable compared to control cells. Corresponding phase-contrast images are shown in the top panels. Photographs taken at  20 magnification. additional mechanisms has not been elucidated. CIN is cells have much lower levels of the Tbx2 protein thought to be a key mechanism driving the genomic compared to the normal lung fibroblasts, WI-38. The alterations associated with tumorigenesis. Defects in changes in phenotype observed in the CT-Tbx2 cells are chromosome segregation during mitosis are one of the therefore not due to gross overexpression of Tbx2. major causes of CIN in cells. This study demonstrates Whether Tbx2 is able to induce the same genetic that reexpression of Tbx2 in a transformed lung instability in normal cells has not been reported possibly fibroblast cell line results in cells that exhibit an increase because they have more robust cell cycle checkpoints in the frequency of several features of genetically which would lead to the elimination of tetraploid cells. unstable cells. These included chromosomal rearrange- This may be the reason why we (unpublished data), and ments resulting in derivative and unidentifiable chromo- others (Butz et al., 2004), have found ectopic expression somes and small marker chromosomes. Importantly, of Tbx2 in several cell lines to be toxic, compromising giant marker and dicentric chromosomes were present in attempts at establishing Tbx2-expressing cell lines. only the Tbx2-expressing cells. Interestingly, small Indeed, this is the first published report characterizing marker chromosomes have been shown to be useful to the properties of genetically engineered Tbx2-expressing cancer cells because they can harbor extra copies of cell lines. genes that make a cell resistant to drugs used in Understanding the mechanism by which Tbx2 induces chemotherapy (Danesi et al., 2003). The increased increased genomic instability will be an important topic number of these chromosomes in CT-Tbx2 cells may of future work. More specifically how Tbx2 causes the explain why, although grossly abnormal, these cells are doubling of chromosomes, resulting in polyploidy, in still capable of dividing and give rise to cells that are already transformed lines may provide a clue to this resistant to the chemotherapeutic drug cisplatin. The mechanism. Polyploidy can be generated from a wide results of this study suggest that one mechanism by variety of errors in cell division (Nigg, 2002; Storchova which Tbx2 may contribute to tumorigenesis is through and Pellman, 2004). For example, anaphase chromo- inducing genomic instability. Importantly, the CT-Tbx2 some segregation and cytokinesis are tightly coordinated

Oncogene Ectopic expression of Tbx2 results in polyploidy E Davis et al 982 processes that have to be completed with high fidelity Namba et al., 1980), and the K1735 murine melanoma cell and are therefore regulated by complex surveillance lines were maintained in Dulbecco’s modified Eagles medium mechanisms. When defects in either of these processes supplemented with 10% fetal calf serum (FCS), 100 U/ml are detected a cell cycle arrest is imposed. In the event penicillin and 100 mg/ml streptomycin. Cells were maintained 1 that these surveillance mechanisms are compromised at 37 C in an atmosphere of 5% CO2. cells can ‘slip past’ this arrest (Storchova and Pellman, 2004) resulting in polyploid cells. Initially, FACS Generation of stable cell lines expressing Tbx2 The CT-1 cells were derived by cobalt 60 g-irradiation of the analysis of CT-Tbx2 cells indicated that they were normal WI-38 lung fibroblast cell line. To generate stably experiencing a cell cycle block in G2 or M, which transfected cell lines, CT-1 cells were transfected with either prevented cells from undergoing cytokinesis. The pre- the empty expression vector pcDNA3.1 ( þ ) or with this vector sence of lagging chromosomes and anaphase bridges in containing the full-length human Tbx2 cDNA (Lingbeek et al., CT-Tbx2 cells supports the possibility that the block is 2002) using the standard calcium phosphate precipitation in M phase and is due to a mitotic spindle checkpoint method (Teng et al., 2007). Among subcloned cell lines, two arrest. This suggests that the CT-1 cells do indeed have a cell lines were chosen for subsequent analysis: CT-Tbx2 functional mitotic spindle checkpoint. However, with contained the pcDNA3.1-Tbx2 construct while CT-E con- increasing passage of CT-Tbx2 cells, a subpopulation of tained pcDNA3.1 empty vector. The K1735 murine melanoma cells was able to proceed through the cell cycle cell line that does not express endogenous Tbx2 was infected with a pBabePuro retrovirus expressing SV5-tagged Tbx2. undetected with a DNA content of 8n. It is therefore Individual clones were selected with 1 mg/ml puromycin to possible that altered Tbx2 protein levels affect chromo- establish the K1735-Tbx2 cell lines. some segregation as well as negatively regulating normal surveillance mechanisms that would otherwise eliminate RT–PCR polyploid cells. We speculate that Tbx2 may be doing For the RT–PCR isolation of the Tbx2 cDNA total RNA was this, through regulating key cell cycle regulators subjected to RT with avian myeloblastosis virus reverse especially since Tbx2 has been shown to interact directly transcriptase (Boehringer, UK) followed by first strand cDNA with and repress the cell cycle regulators p19ARF (Jacobs synthesis (Amersham First Strand cDNA synthesis kit) using et al., 2000; Brummelkamp et al., 2002) and p21 (Prince appropriate primers. et al., 2004). We show that the changes observed in our Tbx2-expressing cells are not associated with alterations Microscopy in p53, p21 or cyclin B protein levels but with increased Cells grown on glass coverslips were fixed in 4% paraformal- levels of p14ARF. Whether the increase in p14ARF is due to dehyde at room temperature for 20 min and permeabilized in a direct regulation of the gene by Tbx2, or a secondary 0.2% Triton X-100 in phosphate-buffered saline (PBS) for 10 min. Cells were incubated in the dark with 1 mg/ml 406- effect induced by Tbx2 is not known. diamidino-2-phenyl indole in PBS for 10 min and visualized by Our results are consistent with, and provide func- fluorescence microscopy. tional support for, reports identifying a role for Tbx2 in the cell cycle. In a recent article by Bilican and Goding Western blot analysis (2006) Tbx2 protein levels were shown to be tightly Cells were harvested and solubilized at 41C in radioimmuno regulated during the various phases of the cell cycle with precipitation assay buffer (150 mM NaCl, 1% Triton X-100, the levels peaking at G2 and being greatly diminished in 0.1% sodium dodecyl sulfate (SDS), 20 mM Tris (pH 7.5), 1% mitosis. These results suggest that Tbx2 may play an deoxycholate and a cocktail of protease inhibitors), centri- 1 important role in G2. Interestingly, in CT-Tbx2 cells fuged at 12 000 g for 20 min at 4 C and the supernatants Tbx2 protein levels follow the same general pattern as recovered. Protein concentrations were determined using the described by Bilican and Goding (2006), except that we bicinchoninic acid assay (Pierce, Rockford, IL, USA), accord- find (data not shown) that Tbx2 protein levels remain ing to the manufacturer’s instructions, with bovine serum high in M phase. It is therefore possible that when Tbx2 albumin as the standard. Equal amounts of protein were loaded and separated on 10–15% SDS-polyacrylamide gels is either ectopically expressed or overexpressed as is the and transferred to Hybond C (Amersham, Amersham, UK). case in some cancers, the mechanism by which the Membranes were blocked for 1 h at room temperature with protein is degraded after G2 is compromised and hence PBS containing 5% nonfat dry milk, probed with appropriate the protein interferes with mitotic events, which may primary antibodies followed by peroxidase-conjugated anti- account for the lagging chromosomes and anaphase mouse or anti-rabbit antibody (1:5000) and visualized by bridges seen in CT-Tbx2 cells. The results presented here enhanced chemiluminescence (Pierce). The primary antibodies therefore provide additional evidence for a role for Tbx2 used were mouse monoclonal anti-Tbx2 62-2 (Prince et al., 2004), mouse monoclonal anti-tubulin, rabbit polyclonal anti- in G2 and may have important implications for our understanding of the exact role that Tbx2 plays in the p53, anti-p21 and anti-p19 (Santa Cruz Biotechnology, CA, cell cycle. USA) and mouse monoclonal anti-cyclin B1 (Transduction Laboratories, Franklin Lakes, NJ, USA).

Growth curves Materials and methods Short-term growth of the CT-Tbx2 cell lines was compared to that of the CT-E control cell lines as described previously Cell culture (Prince et al., 2003). Cells were seeded in triplicate at 104 cells/ WI-38 human embryonic lung fibroblasts, their in vitro well in 12-well plates, collected by trypsinization and counted transformed counterparts WI-38 CT-1 (referred to as CT-1; on a hemocytometer at 2–3 days intervals. As an alternative

Oncogene Ectopic expression of Tbx2 results in polyploidy E Davis et al 983 assay for proliferation, cells were plated at 4 Â 104 on sterile 3 s to 1 min in a trypsin (DIFCO, Maryland, MD, USA, 1:250) glass coverslips in 35 mm dishes and allowed to attach. The solution (0.1 g trypsin in 100 ml isotonic buffer). Slides were cells were then grown in medium containing 10 mM BrdU for 1 rinsed for a few seconds in a jar with FCS (2–3 ml FCS in 50 ml and 8 h followed by fixing with Carnoy’s Fixative (1:3 acetic isotonic buffer) and then rinsed in isotonic buffer and acid:methanol) at À201C for 20 min. For immunostaining, the incubated for 1.5–5 min in a Coplin jar with Giemsa stain cultures were incubated in 2 N hydrochloric acid at 371C for (EM Science, Gibbstown, NJ, USA). 1 h, neutralized in 0.1 M borate buffer (pH 8.5), rinsed with PBS containing 0.05% Tween-20 (PBS/T) and incubated in PBS/T with 5% swine serum for 30 min at 371C. BrdU was Resistance to cisplatin detected with the anti-BrdU mouse monoclonal antibody Cisplatin, an alkylating chemotherapeutic drug, was kindly (6 mg/ml, Roche, Indianapolis, IN, USA) for 30 min at 371C, provided by Pharmachemie (Pty) Ltd (Haarlem, Holland). followed by a secondary immunoglobulin G coupled to Alexa Cells were seeded in quadruplicate in a 96-well plate (3000 488 (1:1000, Molecular Probes, Eugene, OA, USA) for 30 min cells/well) and incubated for 72 h with cisplatin at concentra- at 371C. Cells were rinsed with PBS/T, incubated in propidium tions ranging from 0 to 166.7 mM in culture medium. Cell iodide (PI, 20 mg/ml) for 20 min, mounted onto slides and viability was determined using the MTT assay according to the visualized by fluorescence microscopy. manufacturer’s instructions (Roche). As an assay for cell viability, cells were plated at 3 Â 104 on sterile glass coverslips in 35 mm dishes and allowed to attach. The cells were then Flow cytometry grown in medium containing cisplatin at concentrations Cells were collected by trypsinization, washed twice with PBS, ranging from 0 to 50 mM in the presence of 10 mM BrdU for suspended in 2 ml of cold PBS and fixed in 8 ml of 70% cold 48 h, followed by fixing with Carnoy’s Fixative (1:3 acetic acid/ ethanol for at least 30 min at À201C. Cells were pelleted by methanol) at À201C for 20 min. Immunostaining for BrdU was centrifugation, washed twice with PBS and treated for 15 min as described above for growth curves. at 371C with 50 mg/ml RNase A. Cells were stained at room temperature with PI solution (2 mM MgCl2,10mM Pipes buffer, 0.1 M NaCl, 0.1% Triton X-100, 0.01 mg/ml PI) and subjected to analysis in a Beckman Coulter Cytomics FC500 Acknowledgements flow cytometer. We thank the following cytogeneticists from the National Metaphase spreads Health Laboratory Services for assistance with the metaphase Cells grown to 70% confluency on glass coverslips were spreads and G-banding analysis: Ronnie Smart and Glynnis arrested in metaphase by treatment with 0.1 mg/ml of colcemid Shutte (Groote Schuur Hospital), Jeanette Brusnicky (Tyger- for 1 h at 371C. Following hypotonic treatment in 0.56% KCl berg Hospital) and Dr George Rebello (Department of Human solution at 371C for 1 h, the cells were fixed in 1:3 acetic Genetics, University of Cape Town). We also thank Professors acid:methanol fixative at 371C for 1 h. Evaluation of the Sue Kidson, and Dorothy Bennett, Drs Rafiq Omar and chromosome sets was carried out after conventional Giemsa Shaheen Mowla, Deeya Ballim and Amaal Abrahams for their staining. Metaphase chromosome spreads were prepared on critical feedback on this manuscript. This work was supported acid-cleaned microscope slides using the standard method by grants from the SA Medical Research Council, the National (Priest, 1977). For G-banding staining, slides were incubated Research Foundation and the University of Cape Town.

References

Andreassen PR, Lacroix FB, Lohez OD, Margolis RL. Carlson H, Ota S, Campbell CE, Hurlin PJ. (2001). A WAF1 (2001). Neither p21 nor 14-3-3s prevents G2 pro- dominant repression domain in Tbx3 mediates transcrip- gression to mitotic catastrophe in human colon carci- tional repression and cell immortalization: relevance to noma cells after DNA damage, but p21WAF1 induces stable mutations in Tbx3 that cause ulnar-mammary syndrome. G1 arrest in resulting tetraploid cells. Cancer Res 61: Hum Mol Genet 10: 2403–2413. 7660–7668. Carlson H, Ota S, Song Y, Chen Y, Hurlin PJ. (2002). Tbx3 Baroja A, de la Hoz C, Alvarez A, Vielba R, Sarrat R, impinges on the p53 pathway to suppress apoptosis, Arechaga J et al. (1998). Polyploidization and exit from cell facilitate cell transformation and block myogenic differen- cycle of mechanisms of cultured melanoma cell resistance to tiation. Oncogene 21: 3827–3835. methotrexate. Life Sci 62: 2275–2282. Danesi R, De Braud F, Fogli S, De Pas T, Di Paolo A, Bilican B, Goding CR. (2006). Cell cycle regulation of the Curigliano G et al. (2003). Pharmacogenetics of anticancer T-Box factor tbx2. Exp Cell Res 312: 2358–2366. drug sensitivity in non-small cell lung cancer. Pharmacol Rev Brummelkamp TR, Kortlever RM, Lingbeek M, Trettel F, 55: 57–103. MacDonald ME, van Lohuizen M et al. (2002). TBX-3, the Elledge S. (1996). Cell cycle checkpoints: preventing an gene mutated in ulnar-mammary syndrome, is a negative identity crisis. Science 274: 1664–1671. regulator of p19ARF and inhibits senescence. J Biol Chem 227: Fan W, Huang X, Chen C, Gray J, Huang T. (2004). 6567–6572. TBX3 and its isoform TBX3+2a are functionally dis- Bunz F, Dutriaux A, Lengauer C, Waldman T, Zhou S, Brown tinctive in inhibition of senescence and are overexpressed JP et al. (1998). Requirement for p53 and p21 to sustain G2 in a subset of breast cancer cell lines. Cancer Res 64: arrest after DNA damage. Science 282: 1497–1501. 5132–5139. Butz NV, Campbell CE, Gronostajski RM. (2004). Differ- Huang X, Tran T, Zhang L, Hatcher R, Zhang P. (2005). entiation target gene activation by Tbx2 and Tbx2VP16: DNA damage-induced mitotic catastrophe is mediated by evidence for activation domain-dependent modulation of the Chk1-dependent mitotic exit DNA damage checkpoint. gene target specificity. Gene 342: 67–76. Proc Natl Acad Sci USA 102: 1065–1070.

Oncogene Ectopic expression of Tbx2 results in polyploidy E Davis et al 984 Jacobs JJL, Keblusek P, Robanus-Maandag E, Kristel P, Prince S, Carreira S, Vance K.W, Abrahams A, Goding CR. Lingbeek M, Nederlof PM et al. (2000). Senescence bypass (2004). Tbx2 directly represses the expression of the p21 screen identifies TBX2, which represses Cdkn2a (p19ARF) and (WAF1) cyclin-dependent kinase inhibitor. Cancer Res 64: is amplified in a subset of human breast cancers. Nat Genet 1669–1674. 26: 291–299. Prince S, Wiggins T, Hulley PA, Kidson SH. (2003). Jerome-Majewska LA, Jenkins GP, Ernstoff E, Zindy F, Sherr Stimulation of melanogenesis by tetradecanoylphorbol-13 CJ, Papaioannou VE. (2005). Tbx3, the ulnar-mammary acetate (TPA) in mouse melanocytes and neural crest cells. syndrome gene, and Tbx2 interact in mammary gland Pigment Cell Res 16: 26–34. development through a p19Arf/p53-independent pathway. Rieder CL, Maiato H. (2004). Stuck in division or passing Dev Dyn 234: 922–933. through: what happens when cells cannot satisfy the spindle Lingbeek ME, Jacobs JJL, van Lohuizen M. (2002). The T- assembly checkpoint. Dev Cell 7: 637–651. box repressors TBX2 and TBX3 specifically regulate the Rothenberg M, Ling V. (1989). Multidrug resistance: mole- tumor suppressor gene p14ARF via a variant T-site in the cular biology and clinical relevance. J Natl Cancer Inst 81: initiator. J Biol Chem 277: 26120–26127. 907–910. Mahlamaki EH, Barlund M, Tanner M, Gorunova L, Sinclair CS, Adem C, Naderi A, Soderberg CL, Johnson M, Hoglund M, Karhu R et al. (2002). Frequent amplification Wu K et al. (2002). TBX2 is preferentially amplified in of 8q24, 11q, 17q, and 20q-specific genes in pancreatic BRCA1- and BRCA2-related breast tumors. Cancer Res 62: cancer. Genes Chromosomes Cancer 35: 353–358. 3587–3591. Naiche LA, Harrelson Z, Kelly RG, Papaioannou VE. (2005). Storchova Z, Pellman D. (2004). Polyploidy and aneuploidy: T-box genes in vertebrate development. Ann Rev Genet 39: implications for genome stability and cancer. Nature Rev 219–239. Mol Cell Biol 5: 45–54. Namba M, Nishitami K, Kimoto T. (1980). Characteristics of Teng H, Davis E, Abrahams A, Parker I, Mowla S, Prince S. WI-38 cells (WI-38 CT-1) transformed by treatment with (2007). A role for Tbx2 in the regulation of the a2(1) Co-60 gamma rays. Gann 71: 300–307. collagen gene in human fibroblasts. J Cell Biochem [E-pub Nigg EA. (2002). Centrosome aberrations: cause or conse- ahead of print, 3 April 2007]. quence of cancer progression? Nat Rev Cancer 2: 815–825. Vance K, Carreira S, Brosch G, Goding CR. (2005). Tbx2 is Packham E, Brook JD. (2003). T-box genes in human overexpressed and plays an important role in maintaining disorders. Hum Mol Genet 12: 37–44. proliferation and suppression of senescence in melanomas. Paulovich AG, Toczyski DP, Hartwell LH. (1997). When Cancer Res 65: 2260–2268. checkpoints fail. Cell 88: 315–321. Wu Z, Earle J, Saito S, Anderson CW, Appella E, Xu Y. Priest JH. (1977) In: Medical Cytogenetics and Cell Culture, (2002). Mutation of mouse p53 Ser23 and the response to 2nd edn. Lea and Febiger: Philadelphia, p 118. DNA damage. Mol Cell Biol 22: 2441–2449.

Oncogene