Increased Susceptibility to Tumorigenesis of Ski-Deficient

Increased susceptibility to tumorigenesis of ski-de®cient heterozygous mice

Toshie Shinagawa1, Teruaki Nomura1, Clemencia Colmenares2, Miki Ohira3, Akira Nakagawara3 and Shunsuke Ishii*,1

1Laboratory of Molecular Genetics, RIKEN Tsukuba Institute, and CREST (Core Research for Evolutionary Science and Technology) Research Project of JST (Japan Science & Technology Corporation), 3-1-1 Koyadai, Tsukuba, Ibaraki 305-0074, Japan; 2Department of Cancer Biology, Lerner Research Institute, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio, OH 44195, USA; 3Division of Biochemistry, Chiba Cancer Center Research Institute, 666-2 Nitona, Chuo-ku, Chiba City, Chiba 260-8717, Japan

The c-ski proto-oncogene product (c-Ski) acts as a co- Introduction repressor and binds to other co-repressors N-CoR/ SMRT and mSin3A which form a complex with histone The v-ski gene was originally identi®ed as the deacetylase (HDAC). c-Ski mediates the transcriptional transforming gene of the avian Sloan-Kettering retro- repression by a number of repressors, including nuclear viruses, which transform chicken embryonic ®broblasts hormone receptors and Mad. c-Ski also directly binds to, (Li et al., 1986). The cellular homologue c-ski has been and recruits the HDAC complex to Smads, leading to identi®ed from several species, including human, inhibition of tumor growth factor-b (TGF-b) signaling. chicken, and Xenopus (Nomura et al., 1989; Stavnezer This is consistent with the function of ski as an et al., 1989; Sutrave and Hughes, 1989; Sleeman and oncogene. Here we show that loss of one copy of c-ski Laskey, 1993). The chicken c-ski proto-oncogene increases susceptibility to tumorigenesis in mice. When products (c-Ski) are nuclear proteins of 750 amino challenged with a chemical carcinogen, c-ski hetero- acids (Stavnezer et al., 1989; Sutrave and Hughes, zygous mice showed an increased level of tumor 1989; Sutrave et al., 1990a). The ski gene family formation relative to wild-type mice. In addition, c-ski- contains another member, sno (ski-related novel) gene de®cient mouse embryonic ®broblasts (MEFs) had (Nomura et al., 1989). The sno gene also has increased proliferative capacity, whereas overexpression transforming capacity, and overexpression of sno leads of c-Ski suppressed the proliferation. Furthermore, the to transformation of chicken embryo ®broblasts (Boyer introduction of activated Ki-ras into c-ski-de®cient et al., 1993). Ski is a unique oncogene in that, in MEFs resulted in neoplastic transformation. These addition to aecting cell growth, it is also involved in ®ndings demonstrate that c-ski acts as a tumor regulation of muscle dierentiation. Overexpression of suppressor in some types of cells. The level of cdc25A Ski induces muscle dierentiation in embryo ®broblasts mRNA, which is down regulated by two tumor (Colmenares and Stavnezer, 1989; Colmenares et al., suppressor gene products, Rb and Mad, was upregulated 1991) and causes postnatal hypertrophy of type II fast in c-ski-de®cient MEFs, whereas it decreased by over- muscle ®bers in transgenic mice (Sutrave et al., 1990b). expressing c-Ski in MEFs. This is consistent with the Furthermore, mice lacking c-ski display decreased fact that c-Ski acts as a co-repressor of Mad and Rb. myo®ber development in addition to abnormalities in These results support the view that the decreased pattern formation such as defects in neuronal and activities of Mad and Rb in ski-de®cient cells at least craniofacial patterning (Berk et al., 1997). Thus, c-Ski partly contribute to enhanced proliferation and suscept- has the multiple roles in the regulation of cellular ibility to tumorigenesis. Human c-ski gene was mapped proliferation and development. to a region close to the p73 tumor suppressor gene at the c-Ski is localized to the nucleus (Sutrave et al., 1990a) 1p36.3 locus, which is already known to contain multiple and binds to DNA in association with other cellular uncharacterized tumor suppressor genes. Oncogene factors (Nagase et al., 1990). Ski can function as a (2001) 20, 8100 ± 8108. transcriptional repressor via the speci®c DNA sequence (Nicol and Stavnezer, 1998). Recently we demonstrated Keywords: co-repressor; Ski/Sno; tumor suppressor; that c-Ski and Sno act as co-repressors (Nomura et al., Mad; Rb 1999). c-Ski directly binds to the two co-repressors N- CoR/SMRT (HoÈ rlein et al., 1995; Chen and Evans, 1995) and mSin3A (Ayer et al., 1995). mSin3A and N- CoR/SMRT also interact each other, and form a macromolecular complexes with histone deacetylase (HDAC), respectively (Heinzel et al., 1997; Alland et al., 1997; Hassing et al., 1997; Laherty et al., 1997; *Correspondence: S Ishii; E-mail: [email protected] Received 9 April 2001; revised 12 September 2001; accepted 18 Nagy et al., 1997). Ski and Sno are required for September 2001 transcriptional repression by multiple repressors in- ski acts as a tumor suppressor in mice T Shinagawa et al 8101 cluding Mad and thyroid hormone receptor b (Nomura repression by Mad and Rb (Nomura et al., 1999; et al., 1999), and modulate the transcriptional regula- Tokitou et al., 1999) suggest that ski may function as tion mediated by retinoid receptor a (Dahl et al., 1998). a tumor suppressor at least in some types of cells. The multiple roles of c-Ski described above in the To investigate the possibility that ski acts as a regulation of cellular proliferation and development tumor suppressor, we analysed the ski heterozygous may re¯ect the capacity of Ski to associate with a wide mutant mice. Here we demonstrate that loss of one range of dierent transcriptional regulators. In contrast copy of ski increases susceptibility to tumorigenesis to the role of c-Ski in the transcriptional repression in mice. elicited by multiple repressors, c-Ski also binds to transcriptional activator and inhibits its activity. c-Ski was recently demonstrated to directly bind to Smad Results proteins and recruits the HDAC complex to Smad proteins, leading to the inhibition of Smad proteins- Tumor susceptibility of ski+/7 mice dependent transcriptional activation (Sun et al., 1999; Luo et al., 1999; Akiyoshi et al., 1999; Xu et al., 2000). To examine the possibility that ski might act as a Smad proteins induce the transcription of a group of tumor suppressor, we compared the tumor suscept- target genes upon TGF-b stimulation (for a review, see ibility of ski+/7 mice and wild-type littermates during MassagueÂ and Wotton, 2000), indicating that c-Ski a course of intraperitoneal injections of the carcino- negatively regulates TGF-b signaling. Since TGF-b gen 9,10-dimethyl-1,2-benzanthracene (DMBA). Over inhibits cellular proliferation at least partly by inducing an observation period of 90 days, 96% of the wild- the expression of the Cdk4/6 inhibitor p15INK4B type male mice (n=23), 91% of the wild-type female (Hannon and Beach, 1994), this ®nding is consistent mice (n=22), 52% of the ski+/7 male mice (n=21), with the view that ski possesses the transforming and 30% of the ski+/7 female mice (n=20) survived capacity of an oncogene. this treatment and remained free of clinically- In contrast to its function as an oncogene, c-Ski is apparent tumors (P50.001) (Figure 1a). In total, 23 required to mediate transcriptional repression by two ski+/7 mice died with grossly-visible tumors, most of tumor suppressors, Mad and Rb (Nomura et al., which were malignant lymphomas as judged by 1999; Tokitou et al., 1999). Mad proteins, containing Giemsa staining and immunostaining. Twenty mice the bHLH domain, acts as transcriptional repressors developed lymphomas. Of these, ®fteen were T-cell, after heterodimerization with Max (Ayer et al., 1993). one was B cell, and four were of indeterminate origin The same target sequence of Mad/Max is also (negative for the T-cell and B-cell markers of CD3, recognized by a heterodimer of Myc/Max, which Thy 1.2, B220, and IgM). Two mice developed activates transcription. Myc/Max enhances cellular myelocytic leukemia and one mouse developed proliferation or transformation, whereas Mad/Max megakaryocytic leukemia. Histological analysis of leads to suppression of proliferation or induction of lymphomas in ski+/7 mice homogeneously expressing terminal dierentiation (Ayer and Eisenman, 1993; the T-cell marker Thy-1.2 (T lymphomas) is shown in Roussel et al., 1996; Chin et al., 1995). The role of Figure 1b. Mad as a tumor suppressor was demonstrated by the Most of the tumor suppressors examined to date observation that mutant mice lacking mxi1, a member ful®l Knudson's `two-mutation' criterion (Knudson, of the mad gene family, showed increased suscept- 1971, 1997), which dictates that loss of the wild-type ibility to tumorigenesis (Schreiber-Agus et al., 1998). allele or impaired expression of the tumor suppressor The product of the Rb tumor suppressor gene protein causes tumorigenesis in cells that normally negatively regulates the G1/S transition in the cell express the tumor suppressor gene. Studies of various cycle by silencing a group of target genes regulated by tumor suppressor genes have indicated that loss of E2F transcription factors (for a review, see Weinberg, expression from the wild-type allele is caused by 1995; Nevins, 1992). Rb recruits the HDAC complex multiple mechanisms including rearrangement or to E2F, and actively represses the transcription of deletion of the gene, point mutation, and methylation. S phase-speci®c genes (Brehm et al., 1998; Magnaghi- To determine whether expression of Ski protein from Jaulin et al., 1998; Luo et al., 1998). Thus, Mad and the wild-type allele in ski+/7 mice was lost in tumors, Rb play an important role in the negative regulation we carried out immunostaining of Ski protein in of the G1/S transition. Up to now, multiple target tumor cells. Among the nine DMBA-induced tumors genes, which are down-regulated by Mad or Rb, have examined, Ski protein could not be detected in three been identi®ed (for a review, see Grandori and of them (33%; Figure 1C). Thus, ski ful®ls Knudson's Eisenman, 1997; Muller and Helin, 2000). The most `two-mutation' criterion for a tumor suppressor gene famous among these is cdc25A (Galaktionov et al., (Knudson, 1997). Due to diculties of isolating 1996; Neufeld et al., 1998). Activation of cyclin- enough amounts of tumor cells, we could not analyse dependent kinases (Cdks) can be achieved through whether the wild-type allele in the remaining six their dephosphorylation by members of the Cdc25 tumors, all of which showed Ski expression, has a phosphatase family. By regulating Cdk activities, mutation or not. Therefore, we cannot exclude the Cdc25A plays an important role at the G1/S phase possibility that the mutated Ski protein is expressed in transition. Requirement of c-Ski for transcriptional these tumors (see Discussion).

Oncogene ski acts as a tumor suppressor in mice T Shinagawa et al 8102

Figure 2 Growth advantage conferred on MEFs by the loss of one copy of ski.(a) Growth curves of MEFs. The average and standard deviation of triplicate measurements are shown for each time point. (b) Cell cycle analysis. Exponentially growing wild- type (left) and ski7/7 (right) MEFs were stained with propidium iodide and analysed by FACS. The average population and standard deviation of triplicate measurements are shown Figure 1 Chemical carcinogen-induced tumor formation in ski +/ 7 mice. (a) Survival curves of DMBA-treated wild-type mice and ski+/7 littermates. For all informative cases, the cause of death was the development of a clinically-apparent and histologically-con®rmed tumor. (b) Histopathological analysis of tumors developed in DMBA-treated ski+/7 mice. The enlarged wild-type cells were 28.8% and 19.5%, respectively spleen of a tumor-bearing ski+/7 mouse (upper left) and a (Figure 2b), indicating an approximately 48% increase hematoxylin-eosin stained tumor of the spleen are shown (upper in the S phase population of ski7/7 cells relative to the right). The tumor is positive for the T-cell marker Thy-1.2 (lower 7/7 left), but not for the B-cell marker B220 (lower right). (c) wild type. These results suggest that ski MEFs have Immunostaining with anti-Ski antibody. Note that the para- a defect in cell cycle arrest. cortical area of the normal spleen of ski+/7 mice is stained with As a further measure of proliferative potential, we anti-Ski antibodies (right), but not the tumor cells (left) compared the abilities of ski7/7, ski+/7, and wild- type MEFs to form colonies in methylcellulose after infection with murine leukemia virus (MuLV) carrying the v-K-ras oncogene. ski7/7 MEFs and Growth properties of ski7/7 MEFs ski+/7 MEFs generated 74 ± 165 and 21 ± 47 colonies Since ski-de®cient mice were alive at E15.5 (Berk et al., per 103 cells, respectively, whereas wild-type MEFs 1997), mouse embryonic ®broblasts (MEFs) were generated only a few colonies (Table 1). Under the prepared from ski homozygous mutant embryos. We same conditions, p53-de®cient MEFs produced 230 next assayed the proliferation properties of exponen- colonies per 103 cells (data not shown). MEFs tially growing MEFs using growth curves and ¯ow expressing the v-K-ras oncogene were injected cytometry. Although the wild-type, ski+/7, and ski7/7 subcutaneously into nude mice, and tumor formation cells were morphologically indistinguishable at low was examined. Like p53-de®cient MEFs, ski7/7 density, ski+/7 and ski7/7 MEFs grew faster than MEFs expressing v-K-ras were signi®cantly tumori- wild-type MEFs derived from littermates (Figure 2a). genic in nude mice, although the tumor-forming In addition, ski+/7 and ski7/7 MEF monolayers capacity of wild-type and ski+/7 MEFs was achieved 30% higher cell densities than wild-type undetectable (Table 1, right column). These results MEFs (Figure 2a). Analysis of cell-cycle pro®les indicate that loss of the ski gene facilitated the indicated that the S phase populations of ski7/7 and transformation of cell cultures by activated ras.

Oncogene ski acts as a tumor suppressor in mice T Shinagawa et al 8103 Table 1 Loss of ski predisposes MEFs to tumorigenic transformation. The neoplastic transformation assay was performed using the MEFs. Three independent preparations of MEFs from wild-type, ski+/7,orski7/7 embryos at passage 3 were used. Wild-type embryos were littermates of the ski+/7 and ski7/7 embryos Number of colony/103 cells Tumors/number MEF No virus v-Ki-ras of injection

Wild type E15-1 0 1.7+0.6 ND E15-3 0 2.7+1.2 0/12 E15-6 0 4.3+1.5 ND ski+/7 E15-2 0 40.1+7.6 ND E15-4 0 21.7+1.1 0/12 E14-3 0 47.1+3.1 ND ski7/7 E15-5 0 74.1+24.5 12/12 E15-7 0 165.0+11.5 ND E14-2 0 92.2+11.1 ND

Left column, the number of colonies generated in a methylcellulose gel after infection with murine leukemia virus carrying the activated K-ras oncogene. Experiments were repeated three times, and the average number of colonies formed per 103 cells is indicated together with the strand deviation. Right column, number of tumors formed after injecting nude mice with cells expressing v-K-ras. ND, not determined

Suppression of growth of MEFs by c-ski Enhanced growth of the ski+/7 and ski7/7 MEFs compared to the wild-type MEFs suggested the role for c-ski in the negative regulation of cell proliferation. In addition to the loss-of- function study, we also investigated whether over expression of c-ski suppresses the growth of MEFs. For this purpose, we constructed the retrovirus vector encoding the human wild-type c- ski (pMSCV-ski-ir-GFP), and generated the recombinant viral stock of enough titer. Infection of wild-type MEFs with this recombinant virus led to high-level expression of c-Ski (Figure 3a). Although these cells were morphologically indistinguishable from the non- infected control MEFs, infection with the ski-expression virus almost completely blocked the growth of MEFs (Figure 3b). Analysis of cell-cycle pro®les indicated that the S phase populations of the mock- and ski virus-infected cells were 22.0% and 13.7%, respectively (Figure 3c). Thus, there was about a 38% decrease in the S phase population of ski virus-infected cells relative to the mock-infected cells. These results suggest that overexpression of c-Ski leads to the arrest of the cell cycle at G1 phase.

Figure 3 Eect of Ski on growth properties of MEFs. (a) Regulation of expression of the cdc25A gene by c-ski Construction of the retroviral expression plasmid pMSCV-ski-ir- GFP. Whole cell lysates were prepared from MEFs infected with We previously demonstrated that c-Ski and Sno are the c-Ski expression retrovirus or from control, non-infected required for the transcriptional repression mediated by MEFs, and used for Western blotting with anti-Ski monoclonal two tumor suppressors Mad and Rb (Nomura et al., antibodies. (b) Growth curves of MEFs. The average and standard deviation of triplicate measurements are shown for each 1999; Tokitou et al., 1999). Both Mad and Rb play time point. (c) Cell cycle analysis. Exponentially growing wild- an important role in the negative regulation of type (left) and ski7/7 (right) MEFs were stained with propidium cell cycle progression and down regulate the transcrip- iodide and analysed by FACS. The average population and tion of cdc25A (Galaktionov et al., 1996; Neufeld et standard deviation of triplicate measurements are shown al., 1998). We speculated that the increased proliferation of ski7/7 MEFs was at least partly due to the decreased activities of Mad and Rb. Therefore, we examined the cdc25A mRNA levels in wild-type and decreased to 34% of that of mock-infected cells ski7/7 MEFs (Figure 4a). The levels of cdc25A (Figure 4b). Theses results support the view that mRNA in ski+/7 and ski7/7 cells were 2.8- and 2.1- decreases in the activities of two tumor suppressors fold higher than that of the wild type, respectively. Mad and Rb in the ski-de®cient cells at least partly Consistent with this, the level of cdc25A mRNA in the contributed to enhanced proliferation and suscept- MEFs infected with the c-ski expression virus was ibility to tumorigenesis.

Oncogene ski acts as a tumor suppressor in mice T Shinagawa et al 8104

Figure 4 Expression of cdc25A.(a) cdc25A expression in ski- de®cient MEFs. Total RNA was isolated from con¯uent MEFs and expression of cdc25A and GAPDH were analysed by Northern blotting. The amount of cdc25A mRNA was normal- Figure 5 High resolution mapping of human c-ski gene. (a) ized with respect to the GAPDH mRNA level, and relative Mapping of the human ski gene using the RH panel. The position amounts are indicated as a bar graph. (b) Suppression of cdc25A of the human ski gene at the 1p36.3 locus was determined using expression by overexpression of c-Ski. MEFs were infected with the RH panel, and is shown schematically. The GeneBridge 4 the c-Ski expression retrovirus, and, 24 h after cultivating, RNA panel consisting of DNA samples from 93 hybrids was used. The was prepared from the ski virus-infected MEFs and from the PCR results were: 1000000002 1110000120 2222100000 control mock-infected MEFs. Northern blotting was performed 0000000000 0000000001 0020000000 1000110000 0000000000 similarly 0001010000 000 (Whitehead Institute/MIT Center for Genomic Research order). 0, 1, and 2 represent negative, positive, and ambiguous in the PCR assay, respectively. (b) High resolution FISH analysis. Two color FISH experiments demonstrate the relative distance between ski (green) and p73 (red) (left), between High resolution mapping of the human c-ski gene D1S80 (green) and p73 (red) (middle), and between ski (green) and D1S80 (red) (right). (c) Schematic representation of the To investigate whether the human ski gene is localized positions of the ski gene, D1S80 marker, and p73 gene. The at the locus where other suspected tumor suppressor distances between these three genes were determined by using the genes have been previously mapped, we determined the data from two color FISH experiments position of the human c-ski gene on human chromo- somes by using the RH mapping method. The human c-ski gene was mapped to the 1p36.3 locus. Among the markers contained in the GENEBRIDGE 4 RH Panel, 1995; Ichimiya et al., 1999; Imyanitov et al., 1999; the ski gene was located furthest from the centromere Mori et al., 1998). and 30.1 centi-ray away from the NIB1364 marker (Figure 5a). To precisely determine the position of the human c-ski gene, and especially its location with Discussion respect to that of the well-known tumor suppressor gene p73 also located in this region, we performed high Although ski is generally considered to be oncogene, resolution FISH mapping using three probes (D1S80 the susceptibility of ski+/7 mice to carcinogen-induced (Martinsson et al., 1995), p73 (Kaghad et al., 1997), tumors in this study indicates that ski acts as a tumor and ski) (Figure 5b,c). The data indicated that the suppressor at least in some types of cells. Recently, we D1S80 marker was located between the ski and p73 also observed that mice lacking one copy of the ski- genes, and separated by 285 kb (SEM: 70 kb) and related gene sno showed an increased level of tumor 610 kb (SEM: 150 kb) from the ski and p73 genes, formation relative to wild-type mice when challenged respectively. Thus, the human ski gene is localized at with a chemical carcinogen (Shinagawa et al., 2000). the 1p36.3 locus, a region already known to contain Thus, both Ski and Sno function as either oncoproteins uncharacterized tumor suppressor genes (Cheng et al., or tumor suppressors. Since mounting evidence

Oncogene ski acts as a tumor suppressor in mice T Shinagawa et al 8105 suggests that c-Ski and Sno can form a complex The level of cdc25A mRNA in ski7/7 cells was through their C-terminal coiled-coil region (Nagase et higher than that of the wild type (Figure 4). However, al., 1993; Heyman and Stavnezer, 1994), c-Ski and Sno upregulation of cdc25A gene may be one of the likely possess similar function. Spontaneous tumors multiple mechanisms for tumor formation by loss of have already been observed in sno-heterozygous mice c-Ski. Recently, we found that c-Ski is required for with low frequency (Shinagawa et al., 2000), but not in methyl CpG-dependent gene silencing via directly ski-heterozygous mice, suggesting that sno may be a interacting with methyl-CpG-binding protein, MeCP2 stronger tumor suppressor than ski. Ski protein was (Kokura et al., 2001). If expression of some proto- not detected in three out of nine DMBA-induced oncogenes is repressed by methylation of their tumors. There were no large solid tumors developed in promoter regions, loss of c-Ski would enhance ski+/7 mice, and all the tumor cells were entangled expression of those oncogenes, leading to transforma- with the normal cells. Due to these reasons, we could tion. Recently, it was reported that tumor suppressor not completely purify the enough amounts of tumor p53 represses transcription through interacting with co- cells. Therefore, we could not further analyse whether repressor mSin3A (Murphy et al., 1999). The p53- the wild-type allele in the remaining six tumors, which dependent transcriptional repression of some genes is showed Ski expression, has a mutation or not. As thought to be important for the p53-induced apoptosis. shown in Figures 2a and 4a, loss of either one or two Since c-Ski directly binds to mSin3A, it is not an alleles of c-ski led to higher cell density and increased unlikely possibility that both c-Ski and mSin3 are expression of Cdc25A gene. These results suggest that required for the p53-induced transcriptional repression. tumors were developed in c-ski+/7 mice both by loss of Thus, loss of c-Ski may increase the susceptibility to two c-ski alleles (Knudson type tumor suppressor) and tumorigenesis via multiple mechanisms. also by haploinsuciency. The human c-ski gene maps to 1p36.3 (Figure 5), Although they have been considered oncogenes to close to the reported location of multiple uncharacter- date, both c-ski and sno genes act as a tumor ized tumor suppressor genes for neuroblastoma (Cheng suppressor at least in some types of cells. Similar et al., 1995; Ichimiya et al., 1999), ovarian adenocarci- observations have been made concerning E2F-1 and nomas (Imyanitov et al., 1999), and chronic myelocytic PML. Enforced expression of E2F-1 in cultured cells leukemia (Mori et al., 1998). The sno gene is located induces transformation, but E2F-1-de®cient mice between 3q26.31 and 3q26.32 (Shinagawa et al., develop various types of tumors (Yamasaki et al., unpublished results), where one of the highest 1996; Field et al., 1996). The PML gene was originally frequencies of loss of constitutional heterozygosity in identi®ed as a PML-RAR fusion oncogene that caused human osteosarcoma has also been mapped (Kruzelock acute promyelocytic leukemias (APL); however, later et al., 1997). Elderly people sustain a high rate of studies showed it to be a tumor suppressor gene epithelial cancers by loss of tumor suppressor gene, (Doucas and Evans, 1996; Wang et al., 1998). Since whereas mice carrying common tumor suppressor gene both c-Ski and Sno inhibit TGF-b-induced transcrip- mutations typically develop lymphomas and soft tissue tional activation (Sun et al., 1999; Luo et al., 1999; sarcomas. Due to this species variance, the loss of c-ski Akiyoshi et al., 1999; Xu et al., 2000), Ski-mediated or sno may increase the susceptibility of mice to inhibition of TGF-b signaling likely contributes at least lymphomas, but yield other type(s) of cancer in human. partly to Ski-induced transformation of TGF-b- The results in this study raise the possibility that loss of sensitive cells. On the other hand, c-Ski and Sno are human ski gene family function may increase the needed for the transcriptional repression mediated by susceptibility of human to certain types of tumors. Mad and Rb (Nomura et al., 1999; Tokitou et al., 1999). Since both Mad and Rb appear to induce cell cycle arrest (Zhang et al., 1999; Goodrich and Lee, 1992; Leone et al., 1997), the defect in cell cycle arrest Materials and methods observed in ski7/7 MEFs could be at least partly due to decreased Mad and/or Rb activity. In fact, the Histological analysis and immunohistochemistry cdc25A gene, which is one of the most well-known The generation of ski-de®cient mutant mice was described target genes of Rb and Mad (Galaktionov et al., 1996; previously (Berk et al., 1997). Mice were backcrossed onto Neufeld et al., 1998), is up regulated in ski7/7 MEFs the C57BL/6J background for at least four generations. and down-regulated by over expression of c-Ski (Figure Tissues were ®xed in 4% paraformaldehyde, dehydrated and 4). Similar cell cycle arrest and upregulation of cdc25A embedded in paran. Sections (5 mm) were stained with gene was also observed in sno+/7 MEFs (Shinagawa et haematoxylin and eosin according to standard procedures. al., 2000). Expression of c-ski mRNA peaks in mid G1 Paran sections of 4 mm were used for immunohistochem- phase of the cell cycle in hematopoietic cells, while istry. Phycoerythrin-conjugated anti-Thy-1.2, anti-CD3-e, and expression of sno mRNA is maximal in early to mid anti-B220 antibodies (30-H12, 500A2, and RA3-6B2, respec- G1, although its levels oscillate throughout the rest of tively; Pharmingen) were used for immunostaining. For Ski the cell cycle (Pearson-White et al., 1995). Together protein staining, deparanized sections were incubated in 3% with our results, these observations suggest that c-Ski H2O2 for 10 min to block endogenous peroxidase. Specimens and Sno play an important role in the negative were placed in microwave for 10 min in the presence of regulation of the G1/S transition. 10 mM sodium citrate (pH 6.0) and then treated with 0.3%

Oncogene ski acts as a tumor suppressor in mice T Shinagawa et al 8106 Triton X-100 in PBS for 10 min at room temperature. Northern analysis Samples were then incubated overnight at 378C with a mixture of anti-Ski monoclonal antibodies 1 ± 1, 9 ± 1, 11 ± 1, For the comparison of wild-type and ski-de®cient MEFs, cells and 16 ± 1 (1 : 50). Horseradish peroxidase-conjugated anti- were seeded at 36105 cells per 10 cm dish and exponentially mouse IgG1 antibody was used as the secondary antibody growing cells were harvested 2 days later to prepare RNA. and DAB as the chromogen. Isotype-matched mouse anti- To examine the eect of over expression of c-Ski, wild-type body (IgG1) was used as a negative control. MEFs were infected with the retrovirus expressing c-Ski as described above (MOI=1) for 24 h. Then, the medium was replaced and cells were cultivated for an additional 24 h DMBA tumorigenic treatment before RNA preparation. Total RNA was isolated by Isogen A lipid emulsion of 9,10-dimethyl-1,2-benzanthracene (Nippon Gene, Japan) and 20 mg of total RNA was (DMBA) (0.5% w/w, Sigma) was administered to 2 ± 3 electrophoresed in a 1% agarose formaldehyde gel, and then month-old mice three times at 2-week intervals via intraper- transferred to a Hybond-N+ membrane (Amersham). The itoneal injection (17.5 mg/kg body weight). Mice were blot was hybridized with cdc25A or GAPDH 32P-labeled monitored regularly and killed when their overall health probes. was compromised. A full histopathological survey was conducted on the ®rst 8 ± 10 mice in each cohort; thereafter, Chromosomal localization detailed pathological analysis was restricted to the site of tumor involvement and organs of potential metastatic Radiation hybrid (RH) panel (Genebridge 4 RH Panel, involvement. Research Genetics, Inc.) screening was generally done according to the manufacturer's protocol. The following two primers were used for PCR: hski-1, 5'- Analysis of embryonic fibroblasts CGTGGTGGGAGGCGAGAAGC-3' (human c-ski mRNA Mouse primary embryonic ®broblasts (MEFs) were isolated position 411-430), and hski-2, 5'-CCAGCAGCCCCTTA- from embryos at 14.5 and 15.5 d.p.c. For growth curves, cells CACTTG-3' (human c-ski mRNA position 769 ± 750). The at passage 3 were grown from a starting density of 36104 human ski DNA fragment was ampli®ed by 30 cycle of 948C cells per well in 24-well plates in DMEM supplemented with for 45 s, 638C for 30 s, and 728C for 1 min. PCR products 10% fetal bovine serum. Nine independent experiments were were electrophoresed on 2% agarose gel. Data were analysed carried out from three dierent MEF preparations. For cell by accessing the server at http://www-genome.wi.mit.edu/cgi- cycle analysis, exponentially growing cells were suspended in bin/contig/rhmapper.pl. High resolution mapping was per- 50 mg/ml propidium iodide (Sigma), 4 mM sodium citrate, formed essentially as described by Lawrence et al. (1990) 150 mM NaCl, 0.5 mg/ml RNase A, and 0.05% Nonidet P- using ¯uorescence in situ hybridization (FISH) signal 40, and were analysed by FACScan (Becton Dickinson). Cell mapping and the PAC clones encoding human c-ski gene, cycle distributions were analysed using the CellQuest and p73 gene, or D1S80 marker. Puri®ed PAC clone DNAs were ModFit LT software packages. For colony-formation assays, labeled with either biotin dATP or digoxigenin dUTP by nick MEFs were infected with murine leukemia virus (MuLV) translation. Labeled probes were combined with sheared carrying the activated K-ras oncogene (MOI=1). After 48 h, human DNA and hybridized as dierentially labeled pairs to 36104 cells were suspended in 1.3% methylcellulose gel interphase nuclei derived from PHA stimulated peripheral dissolved in culture medium and overlaid on an agarose bed blood lymphocytes in a solution containing 50% formamide, composed of 0.53% agarose in culture medium. Colonies 10% dextran sulfate, and 26SSC. Speci®c hybridization were scored 3 weeks after plating. To test the tumorigenicity signals were detected by incubating the hybridized slides in of the cell clones, subcutaneous injection of nude mice ¯uorescein conjugated anti-digoxigenin antibodies as well as (BALB/c nu/nu; Clea Japan Inc.) was performed with 106 Texas red avidin. Estimation of the distance between two cells in 200 ml of DMEM without FCS. Cells were scored as locations was performed as described by van den Engh et al. tumorigenic if a visible nodule more than 2 mm in diameter (1992). appeared at the site of injection at 10 days after injections.

Generation of retroviral expression plasmid for human c-ski Abbreviations The human full-length c-ski cDNA was cloned into the 5' side c-ski, cellular ski proto-oncogene; c-Ski, c-ski gene pro- of the ribosome entry site of the MSCV (murine stem cell duct; DMBA, 9,10-dimethyl-1,2-benzanthracene; HDAC, virus)-based retroviral vector (Persons et al., 1997) to histone deacetylase; MEFs, mouse embryonic ®broblasts; generate the plasmid pMSCV-ski-ir-GFP. The ecotropic virus TGF-b, tumor growth factor-b packaging cells BOSC23 (Pear et al., 1993) were transfected with the pMSCV-ski-ir-GFP DNA, and conditioned media were prepared. The viral titer in conditioned media was Acknowledgments approximately 2.66105 particles/ml as assessed by transfer of We are grateful to N Nomura for advice on mapping, M the GFP marker to wild-type MEFs. MEFs were infected Noda for the kind gift of MuLV, and E Stavnezer for with this virus (MOI=1) for 24 h, and after replacement of helpful discussions. This work was supported in part by medium, growth curve and cell cycle analysis were made as grant HD30728 from the National Institutes of Health to C described above. Colmenares.

Oncogene ski acts as a tumor suppressor in mice T Shinagawa et al 8107 References

Akiyoshi S, Inoue H, Hanai J, Kusanagi K, Nemoto N, Kokura K, Kaul SC, Wadhwa R, Nomura T, Khan MM, Miyazono K and Kawabata M. (1999). J. Biol. Chem., Shinagawa T, Yasukawa T, Colmanares C and Ishii S. 274, 35269 ± 35277. (2001). J. Biol. Chem., 276, 34115 ± 34121. Alland L, Muhle R, Hou Jr H, Potes J, Chin L, Screiber- Knudson Jr AG. (1971). Proc. Natl. Acad. Sci. USA, 68, Agus N and DePinho RA. (1997). Nature, 387, 49 ± 55. 820 ± 823. Ayer DE and Eisenman RN. (1993). Genes Dev., 7, 2110 ± Knudson Jr AG. (1997). Ann. NY Acad. Sci. USA, 833, 58 ± 2119. 67. Ayer DE, Kretzner L and Eisenman RN. (1993). Cell, 72, Kruzelock RP, Murphy EC, Strong LC, Naylor SL and 211 ± 222. Hansen MF. (1997). Cancer Res., 57, 106 ± 109. Ayer DE, Lawrence QA and Eisenman RN. (1995). Cell, 80, Laherty CD, Yang W-M, Sun J-M, Davie JR, Seto E and 767 ± 776. Eisenman RN. (1997). Cell, 89, 349 ± 356. Berk M, Desai SY, Heyman HC and Colmenares C. (1997). Lawrence JB, Singer RH and McNeil JA. (1990). Science, Genes Dev., 11, 2029 ± 2039. 249, 928 ± 932. Boyer PL, Colmenares C, Stavnezer E and Hughes SH. Leone G, DeGregori J, Sears R, Jakoi L and Nevins JR. (1993). Oncogene, 8, 457 ± 466. (1997). Nature, 387, 422 ± 426. Brehm A, Miska EA, McCance DJ, Reid JL, Bannister AJ Li Y, Turck CM, Teumer JK and Stavnezer E. (1986). J. and Kouzarides T. (1998). Nature, 391, 597 ± 601. Virol., 57, 1065 ± 1072. Chen JD and Evans RM. (1995). Nature, 377, 454 ± 457. LuoK,StroscheinSL,WangW,ChenD,MartensE,ZhouS Cheng NC, Van Roy N, Chan A, Beitsma M, Westerveld A, and Zhou Q. (1999). Genes Dev., 13, 2196 ± 2206. Speleman F and Versteeg R. (1995). Oncogene, 10, 291 ± Luo RX, Postigo AA and Dean DC. (1998). Cell, 92, 463 ± 297. 473. Chin L, Schreiber-Agus N, Pellicer I, Chen K, Lee H-W, Magnaghi-Jaulin L, Groisman R, Naguibneva I, Robin P, Dudast M, Cordon-Cardo C and Depinho RA. (1995). Lorain S, Le Villain JP, Troalen F, Trouche D and Harel- Proc. Natl. Acad. Sci. USA, 92, 8488 ± 8492. Bellan A. (1998). Nature, 391, 601 ± 605. Colmenares C and Stavnezer E. (1989). Cell, 59, 293 ± 303. Martinsson T, Sjoberg RM, Hedborg F and Kogner P. Colmenares C, Teumer JK and Stavnezer E. (1991). Mol. (1995). Cancer Res., 55, 5681 ± 5686. Cell. Biol., 11, 1167 ± 1170. MassagueÂ J and Wotton D. (2000). EMBO J., 19, 1745 ± Dahl R, Kieslinger M, Beug H and Hayman MJ. (1998). 1754. Proc. Natl. Acad. Sci. USA, 95, 11187 ± 11192. Mori N, Morosetti R, Spira S, Lee S, Ben-Yehuda D, Schiller Doucas V and Evans RM. (1996). Biochim. Biophys. Acta., G, Landol® R, Mizoguchi H and Koeer HP. (1998). 1288, 25 ± 29. Blood, 92, 3405 ± 3409. Field SJ, Tsai F-Y, Kuo F, Zubiaga AM, Kaelin Jr WG, Muller H and Helin K. (2000). Biochem. Biophys. Acta., Livingston DM, Orkin SH and Greenberg ME. (1996). 1470, M1 ± M12. Cell, 85, 549 ± 561. Murphy M, Ahn J, Walker KK, Homan WH, Evans RM, Galaktionov K, Chen X and Beach D. (1996). Nature, 382, Levine AJ and George DL. (1999). Genes Dev., 13, 2490 ± 511 ± 517. 2501. Goodrich DW and Lee WH. (1992). Nature, 360, 177 ± 179. Nagase T, Mizuguchi G, Nomura N, Ishizaki R, Ueno Y and Grandori C and Eisenman RN. (1997). Trends Biochem. Sci., Ishii S. (1990). Nucleic Acids Res., 18, 337 ± 343. 22, 177 ± 181. Nagase T, Nomura N and Ishii S. (1993). J. Biol. Chem., 268, Hannon GJ and Beach D. (1994). Nature, 371, 257 ± 261. 13710 ± 13716. Hassing CA, Fleischer TC, Billin AN, Schreiber SL and Ayer Nagy L, Kao H-Y, Chakravarti D, Lin RJ, Hassig CA, Ayer DE. (1997). Cell, 89, 341 ± 347. DE, Screiber SL and Evans RM. (1997). Cell, 89, 373 ± HeinzelT,LavinskyRM,MullenT-M,SoÈ derstroÈ mM, 380. LahertyCD,TorchiaJ,YangW-M,BrardG,NgoSD, Neufeld TP, de la Cruz AF, Johnston LA and Edgar BA. Davie JR, Seto E, Eisenman RN, Rose DW, Glass CK and (1998). Cell, 93, 1183 ± 1193. Rosenfeld MG. (1997). Nature, 387, 43 ± 48. Nevins JR. (1992). Science, 258, 424 ± 429. Heyman HC and Stavnezer E. (1994). J. Biol. Chem., 269, Nicol R and Stavnezer E. (1998). J. Biol. Chem., 273, 3588 ± 26996 ± 27003. 3597. HoÈ rlein AJ, NaÈ aÈ r AM, Heinzel T, Torchia J, Gloss B, Nomura T, Khan MM, Kaul SC, Dong H-D, Wadhwa R, Kurokawa R, Ryan A, Kamei Y, SoÈ derstroÈ mM,Glass Colmenares C, Kohno I and Ishii S. (1999). Genes Dev., CK and Rosenfeld MG. (1995). Nature, 377, 397 ± 404. 13, 412 ± 423. Ichimiya S, Nimura Y, Kageyama H, Takada N, Sunahara Nomura N, Sasamoto S, Ishii S, Date T, Matsui M and M, Shishikura T, Nakamura Y, Sakiyama S, Seki N, Ohira Ishizaki R. (1989). Nucleic Acids Res., 17, 5489 ± 5500. M, Kaneko Y, McKeon F, Caput D and Nakagawara A. Pear WS, Nolan GP, Scott ML and Baltimore D. (1993). (1999). Oncogene, 18, 1061 ± 1066. Proc. Natl. Acad. Sci. USA, 90, 8392 ± 8396. Imyanitov EN, Birrell GW, Filippovich I, Sorokina N, Pearson-WhiteS,DeaconD,CrittendenR,BradyG,Iscove Arnold J, Mould MA, Wright K, Walsh M, Mok SC, N and Quesenberry PJ. (1995). Blood, 86, 2146 ± 2155. Lavin MF, Chenevix-Trench G and Khanna KK. (1999). Persons DA, Allay JA, Allay ER, Smeyne RJ, Ashmun RA, Oncogene, 18, 4640 ± 4642. Sorrentino BP and Nienhuis AW. (1997). Blood, 90, Kaghad M, Bonnet H, Yang A, Creancier L, Biscan JC, 1777 ± 1786. Valent A, Minty A, Chalon P, Lelias JM, Dumont X, Roussel MF, Ashmun RA, Sher CJ, Eisenman RN and Ayer Ferrara P, McKeon F and Caput D. (1997). Cell, 90, 809 ± DE. (1996). Mol. Cell. Biol., 16, 2796 ± 2801. 819.

Oncogene ski acts as a tumor suppressor in mice T Shinagawa et al 8108 Schreiber-Agus N, Meng Y, Hoang T, Hou Jr H, Chen K, Tokitou F, Nomura T, Khan MM, Kaul SC, Wadhwa R, Greenberg R, Cordon-Cardo C, Lee H-W and DePinho Yasukawa T, Kohno I and Ishii S. (1999). J. Biol. Chem., RA. (1998). Nature, 393, 483 ± 487. 274, 4485 ± 4488. Shinagawa T, Dong H-D, Xu M, Maekawa T and Ishii S. van den Engh G, Sachs R and Trask BJ. (1992). Science, 257, (2000). EMBO J., 19, 2280 ± 2291. 1410 ± 1412. Sleeman JP and Laskey RA. (1993). Oncogene, 8, 67 ± 77. Wang ZG, Delva L, Gaboli M, Rivi R, Giorgio M, Cordon- Stavnezer E, Brodeur D and Brennan L. (1989). Mol. Cell. Cardo C, Grosveld F and Pandol® PP. (1998). Science, Biol., 9, 4038 ± 4045. 279, 1547 ± 1551. Sun Y, Liu X, Eaton EN, Lane WS, Lodish HF and Weinberg RA. (1995). Cell, 81, 323 ± 330. Weinberg RA. (1999). Mol. Cell, 4, 499 ± 509. Xu W, Angelis K, Danielpour D, Haddad MM, Bischof O, Sutrave P, Copeland TD, Showalter SD and Hughes SH. Campisi J, Stavnezer E and Medrano EE. (2000). Proc. (1990a). Mol. Cell. Biol., 10, 3137 ± 3144. Natl. Acad. Sci. USA, 97, 5924 ± 5929. Sutrave P and Hughes SH. (1989). Mol. Cell. Biol., 9, 4046 ± Yamasaki L, Jacks T, Bronson R, Goillot E, Harlow E and 4051. Dyson NJ. (1996). Cell, 85, 537 ± 548. Sutrave P, Kelly AM and Hughes SH. (1990b). Genes Dev., 4, Zhang HS, Postigo AA and Dean DC. (1999). Cell, 97, 53 ± 1462 ± 1472. 61.

Oncogene