Increased DNA Damage Sensitivity and Apoptosis in Cells Lacking the ...

MOLECULAR AND CELLULAR BIOLOGY, Apr. 2006, p. 2661–2674 0270-7306/06/$08.00⫹0 doi:10.1128/MCB.26.7.2661–2674.2006 Copyright © 2006, American Society for Microbiology. All Rights Reserved.

Vol. 26, No. 7

Increased DNA Damage Sensitivity and Apoptosis in Cells Lacking the Snf5/Ini1 Subunit of the SWI/SNF Chromatin Remodeling Complex Agnes Klochendler-Yeivin,1* Eli Picarsky,2 and Moshe Yaniv 3 Department of Animal and Cell Biology, The Institute for Life Sciences, The Hebrew University of Jerusalem, Givat-Ram, Jerusalem 91904, Israel1; Department of Pathology, Hadassah-Hebrew University Medical Center, Jerusalem 91120, Israel 2; and Gene Expression and Disease Unit, Department of Developmental Biology, Pasteur Institute, Paris, France3 Received 6 October 2005/Returned for modification 4 November 2005/Accepted 5 January 2006

The gene encoding the SNF5/Ini1 core subunit of the SWI/SNF chromatin remodeling complex is a tumor suppressor in humans and mice, with an essential role in early embryonic development. To investigate further the function of this gene, we have generated a Cre/lox-conditional mouse line. We demonstrate that Snf5 deletion in primary fibroblasts impairs cell proliferation and survival without the expected derepression of most retinoblastoma protein-controlled, E2F-responsive genes. Furthermore, Snf5-deficient cells are hypersensitive to genotoxic stress, display increased aberrant mitotic features, and accumulate phosphorylated p53, leading to elevated expression of a specific subset of p53 target genes, suggesting a role for Snf5 in the DNA damage response. p53 inactivation does not rescue the proliferation defect caused by Snf5 deficiency but reduces apoptosis and strongly accelerates tumor formation in Snf5-heterozygous mice. tumor cell lines (74). Mice heterozygous for a Brg1-null mutation are cancer prone and develop tumors of epithelial origin without loss of heterozygosity (11). It is still unclear whether the tumor suppressor function of Snf5 depends on SWI/SNF activity. Although it has been shown that the expression of several Brg1 target genes is not affected by the absence of Snf5 and that the SWI/SNF complex remains intact in Snf5-deficient cells (17), the similar phenotypes of Snf5 and Brg1 knockout mice suggest that both genes function in partially redundant pathways. The molecular mechanisms dictating the tumor-suppressing functions of Snf5 and Brg1 are not fully understood. The antiproliferative activity of Brg1 has been shown to depend on the repression of specific E2F target genes (those for cyclin E and cyclin A and cdc2) via its association with the retinoblastoma gene product (pRb) (65, 80). Reexpression of Brg1 in mutant cell lines can restore pRb function via induction of the cyclin-dependent kinase inhibitors p15INK4b and p21cip1 (27). Similarly, reexpression of SNF5 in human rhabdoid cell lines causes G0/G1 arrest. This may occur via downregulation of specific cyclin-encoding genes or via direct activation of p16INK4a, implicating the cyclin D/CDK4-pRb-E2F pathway in tumor formation (6, 51, 68, 81). Studies of mouse models for Snf5 inactivation indicate that Snf5 has a dual role: it prevents tumorigenesis, and paradoxically, it is required for cell survival, as both the Snf5-null trophectoderm and inner-cell-mass-derived cell lineages die of apoptosis (38). Furthermore, mice with a conditionally inactivated Snf5 locus die and display bone marrow failure, indicating that Snf5 is required for the survival of hematopoietic cells (58). These results suggest that malignancy due to loss of Snf5 depends on the specific cellular context and/or additional mutations, whereas most cells will not survive the biallelic inactivation event. To understand better the molecular basis underlying the survival or lethal phenotype triggered by Snf5 inactivation, we used a Cre/lox-conditional targeting approach to disrupt Snf5 in cultured primary murine embryonic fibroblasts (MEFs) (61). MEFs represent a cell type that has been

Chromatin-remodeling complexes modulate nucleosome structure in an ATP-dependent manner (40, 49). The 2-MDa multisubunit SWI/SNF complex was the first such complex identified in yeast and is highly conserved among eukaryotes (14, 54). Mammalian SWI/SNF complexes are heterogenous but have several common subunits, including SNF5 (Ini1), BAF155/BAF170, and one of the highly homologous ATPase subunits (Brm or Brg1) (46, 72). SWI/SNF complexes play important roles in transcriptional regulation (both activation and repression) and contribute to the control of diverse cellular processes, such as proliferation and differentiation (21, 39, 44, 59, 63). Gene-targeting experiments with mice have shown that the mammalian SWI/SNF complex is essential for early embryonic development, since homozygous inactivation of genes encoding different subunits (Brg1, Snf5, or Swi3/Srg3) results in peri-implantation lethality (11, 22, 37, 38, 57). Accumulating genetic evidence has defined SNF5/Ini1 as a tumor suppressor gene in humans and mice. Homozygous inactivating mutations or deletions in the SNF5 gene in humans are associated with malignant rhabdoid and atypical teratoid/ rhabdoid tumors (7, 62, 69). These are rare, but very aggressive, pediatric tumors that arise primarily in the brain and kidney. Furthermore, we and others have shown that Snf5 functions as a tumor suppressor gene in mice. Heterozygous Snf5⫹/⫺ animals develop tumors at a high incidence via loss of heterozygosity (LOH) at the Snf5 locus. These tumors share histological features with their human counterparts and develop at very specific sites, mainly in the nervous system (22, 38, 57). Recently, it was shown that mice with a reversible conditional mutation of Snf5 developed early, and fully penetrant, T-cell lymphomas (58). Several studies have also revealed that Brg1 is mutated or deleted in a variety of human

* Corresponding author. Mailing address: Department of Animal and Cell Biology, The Institute for Life Sciences, The Hebrew University of Jerusalem, Givat-Ram, Jerusalem 91904, Israel. Phone: 00972-2-6585945. Fax: 00972-2-6585417. E-mail: [email protected]. 2661

2662

KLOCHENDLER-YEIVIN ET AL.

widely used to identify the consequences of gene ablation in cell cycle control. We report that inactivation of Snf5 in MEFs impairs cell growth and survival. This phenotype includes hypersensitivity to genotoxic stress and signs of defective mitosis and occurs concomitantly with p53 induction and altered expression of several key players involved in cell cycle regulation. Although p53 inactivation cannot rescue the growth arrest phenotype in cultured Snf5-deficient MEFs, we show that it significantly reduces the apoptotic response and considerably accelerates the onset of rhabdoid tumor formation in Snf5heterozygous mice.

MATERIALS AND METHODS Mice. The targeting vector and the conditional-deletion procedure have been described in detail elsewhere (21). The floxed allele contains two loxP sites, the first one located 1 kb upstream of exon 1 and the other one 0.6 kb downstream of exon 2. Genotyping of mice and embryos was performed by PCR using primers flanking the second loxP site (5⬘-CTTGCCAGGTGAGCAGTCTG and 5⬘-GC CACCAGCCAGATGTCATAC). Mice carrying a null allele of Snf5 (designated Snf5⫺) have been described previously (38). Snf5-heterozygous mice were bred with mice carrying a mutated null p53 allele (33). All the animals were on a mixed 129/SV ⫻ C57BL/6 genetic background. Survival curves were compiled from animals that died or were sacrificed when seriously ill or displaying an obvious tumor. MEF generation, culture, and infection. Primary MEFs were isolated from day 13.5 postcoitum embryos by standard methods. Briefly, the brains and livers were used for genotyping, and the remainder of the embryos were treated with trypsin, washed once in phosphate-buffered saline (PBS), and cultured in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal calf serum (FCS). MEFs at passage 2 or 3 were infected with adenovirus type 5-cytomegalovirus Cre (AdCre) (University of Iowa Gene Transfer Vector Core) in 2% FCS-DMEM at a multiplicity of infection of 100 per cell. At 18 to 24 h after the start of infection, the virus-containing medium was removed and replaced with fresh 10% FCS-DMEM. Southern blot analysis. Genomic DNA was extracted from mock- or AdCreinfected MEFs and from tumor samples to monitor Cre-mediated deletion and LOH, respectively. Tumor DNA was digested with PstI, separated by electrophoresis on a 1% agarose gel, and transferred to a Hybond N⫹ membrane (Amersham) using standard Southern blotting techniques. Blots were probed with a radiolabeled genomic 5⬘ external probe as described previously (38). Cell proliferation assays. For growth curves, 300,000 mock- or AdCre-infected MEFs were seeded in duplicate 2 days after infection. At the indicated time periods, cell cultures were trypsinized and washed, and live-cell numbers were determined by trypan blue exclusion and plated at the same initial density. To monitor cells in S phase, we performed BrdU incorporation assays. Cells plated on coverslips were labeled with 10 ␮M BrdU for 30 min. Cells were then fixed in ice-cold methanol for 10 min, incubated in 2 N HCl for 10 min, and washed with 0.1 M sodium borate (pH 8.5) and then with PBS. Incorporation of BrdU was monitored by immunostaining with anti-BrdU antibodies (Ab-3; Neomarkers) and with fluorescein isothiocyanate-conjugated goat anti-mouse immunoglobulin Gs (IgGs). The percentage of BrdU-labeled nuclei was quantified by counting at least 300 DAPI (4⬘,6⬘-diamidino-2-phenylindole)-counterstained nuclei for each MEF preparation. Genotoxin sensitivity. Three days after infection, mock- and AdCre-infected MEFs plated at 5 ⫻ 104 to 10 ⫻ 104 cells/well in six-well dishes were exposed to UV light (254 nm) or doxorubicin at the doses indicated in Fig. 5. UV exposure was achieved using StrataLinker source (Stratagene Corp.) after gentle aspiration of the culture medium. Cell survival was determined 24 h (doxorubicin) or 3 days (UV) later by trypan blue exclusion and expressed as a percentage compared to that of untreated, nonconfluent cells. All experiments were conducted in triplicate, and the cell survival rate for each replicate was determined twice at each experimental point. Thus, each data point represents the mean of six observations. Flow cytometry. Cells (1 ⫻ 106 to 2 ⫻ 106) were harvested at various time points after infection, pelleted, and washed in PBS. Cells were fixed in ice-cold 70% ethanol. After fixation, cells were washed in PBS and stained in PBS containing 50 ␮g/ml propidium iodide (Sigma) and 100 ␮g/ml RNase A. Flow cytometry analysis was carried out (BD Biosciences) and analyzed using CellQuest software.

MOL. CELL. BIOL. Protein analysis. Whole-cell extracts were prepared in lysis buffer (20 mM phosphate buffer [pH 7], 250 mM NaCl, 30 mM Na4P2O7, 0.1% NP-40, 5 mM EDTA) containing the complete cocktail of protease inhibitors (Roche). Protein levels were determined by Western blot analyses by following established protocols. The rabbit polyclonal anti-SNF5 was described previously (48). Antibodies against p21 (F-5), cyclin E (M-20), and cyclin A (C-19) were obtained from Santa Cruz. Antibodies against ␣-tubulin were from Sigma. The phospho-p53 (Ser15) antibodies were obtained from Cell Signaling Technology, and p19ARF antibodies were from Novus Biologicals. The mouse monoclonal anti-p53 (248 and 421) and anti-mdm2 (2A10 and 4B2) were a kind gift from Y. Haupt (Hebrew University, Jerusalem, Israel). Secondary antibodies coupled to horseradish peroxidase were purchased from Jackson Immunoresearch Laboratories. Detection was performed by chemiluminescence. Semiquantitative RT-PCR. We carried out reverse transcription (RT)-PCR on total RNA prepared with the RNeasy mini kit (QIAGEN). RNA (2 ␮g) was reverse transcribed into cDNA using random primers and Moloney murine leukemia virus reverse transcriptase (Promega). cDNA aliquots (0.8 ng and 2.4 ng) were amplified by PCR and labeled with [␣-32P]dCTP, using the minimum number of cycles to obtain a clear signal in the linear range. The following are the number of cycles and the sequences of 5⬘ and 3⬘ primers used for each of the tested genes encoding the indicated proteins, as follows: GAPDH (glyceraldehyde-3-phosphate dehydrogenase), 22 cycles, 5⬘-GCCTGGAGAAACCTGC CAAG and 5⬘-CTCCTTGGAGGCCATGTAGG; SNF5, 27 cycles, 5⬘-TCCGG GATCAAGATAGGAACAC and 5⬘-TGGAATGTGTGCTGAAGGGAG; p16, 26 cycles, 5⬘-CAGACAGACTGGCCAGGGC and 5⬘-GAGAAGGTAGGGTC CTC; p19ARF, 29 cycles, 5⬘-GAGGGTTTTCTTGGTGGAGTTC and 5⬘-GAG AAGGTAGGGTCCTC; mdm2, 24 cycles, 5⬘-TCACAGTCTATCAGACA GGAG and 5⬘-TCCCCTTATCGTGAAGC; p21, 24 cycles, 5⬘-CACGTGGCCT TGTCGCTGTC and 5⬘-CACACAGAGTGAGGGCTAAGG; Puma, 29 cycles, 5⬘-AGCACTTAGAGTCGCTG and 5⬘-AGGGTGAGGGTCGGTGTCG; Bax, 28 cycles, 5⬘-CCCTGTGCAACTAAAGTGCCC and 5⬘-ACCCCTCCCAATAA TTACAAAAG; thymidine kinase (TK), 26 cycles, 5⬘-GCAGCATCTTGAACCT GGTG and 5⬘-CTCAGTTGGCAGAGTTGTATTG; dihydrofolate reductase (DHFR), 29 cycles, 5⬘-TTGTGACAAGGATCATGCAGG and 5⬘-ACTAGGGTT GGGGTGGCTC; cyclin E1, 26 cycles, 5⬘-CTGGACAAAGCCCAAGCAAAG and 5⬘-AGGCCAGCAACCGCCATGG; E2F1, 26 cycles, 5⬘-GGAGAAGTCACGCT ATGAAACC and 5⬘-CTATGACCATCTGTTCTGAGG; p73 with an N-terminal transactivation domain (TA-p73), 38 cycles, 5⬘-CAGACAGCACCTACTTTGACC and 5⬘-GGTATTGGAAGGGATGACAGG; p73 without the N-terminal transactivation domain (⌬N-p73), 38 cycles, 5⬘-CACGAGCCTACCATGCTTTAC and 5⬘-GGTATTGGAAGGGATGACAGG. Each RT-PCR was performed at least twice with RNAs extracted from two infection experiments using different sets of MEF preparations. Immunofluorescence and histology. Cells plated on coverslips were fixed with 4% paraformaldehyde for 15 min and permeabilized with 0.5% Triton X-100 in PBS for 10 min. Coverslips were washed with PBS, incubated for 1 h with blocking buffer (1% bovine serum albumin, 0.05% Tween 20 in PBS), and incubated with anti-␣-tubulin (dilution, 1:1,000) or anti-␥-tubulin (dilution, 1:300) primary antibodies (Sigma) for 2 h. Coverslips were rinsed three times with PBS and incubated for 1 h with fluorescein isothiocyanate-conjugated secondary anti-mouse antibodies (Calbiochem). After counterstaining with DAPI and mounting using Vectashield (Vector Laboratories), immunofluorescence was monitored and images were digitally recorded. Tumors were fixed in phosphate-buffered 4% formaldehyde (pH 7.4), embedded into paraffin, and cut in 5-␮m sections. Sections were stained with hematoxylin and eosin.

RESULTS Conditional inactivation of Snf5 in MEFs causes cell cycle arrest and cell death. To circumvent the embryonic lethality inherent in homozygous-Snf5 deletion, we used the Cre/loxP recombination system of bacteriophage P1 to generate conditional-Snf5-knockout mice. The Snf5lox allele carries two loxP sites abutting the 5⬘ regulatory sequences and the first two exons of the murine Snf5 gene (Fig. 1A). We obtained Snf5lox/lox mice at the normal frequency, confirming that the floxed Snf5 allele did not have an adverse effect on the development of animals. By contrast, homozygosity for the deleted allele was lethal at peri-implantation, as observed with the Snf5

VOL. 26, 2006

Snf5 DISRUPTION IN MURINE PRIMARY FIBROBLASTS

2663

FIG. 1. Proliferative defects and cell death in Snf5-depleted MEFs. (A) Schematic representation of the wild-type (Snf5⫹), floxed (Snf5L), and deleted (Snf5D) alleles of Snf5. Exons 1 to 3 are indicated by boxes and loxP sites by filled triangles. Relevant restriction sites (EcoRI [E], EcoRV [EV]) and the position of the probe are indicated. Cre-mediated recombination eliminates exons 1 and 2. Southern blot (B) and Western blot (C) analyses of DNA and lysates prepared from mock- and AdCre-infected MEFs with the indicated genotypes, 4 and 6 dpi. For Southern blot analysis, the DNA was digested with EcoRI and EcoRV and the Snf5 locus was identified by hybridization with the probe indicated in panel A. (D) Exit of Snf5-deficient MEFs from the cell cycle. Cells of indicated genotypes are shown 6 days after infection (magnification, ⫻100). (E) Representative growth curves of mock- and AdCre-infected MEFs with the indicated genotypes. The cells were placed in culture at equal densities, infected, counted, and passaged every 2 days at the same density. Experiments were performed on two different MEF cultures of each genotype, and duplicate plates were counted. The curves represent the means of these values.

null allele, confirming that the Cre-mediated deletion inactivates Snf5 (data not shown). To study the effect of Snf5 loss on primary cell proliferation, day 13.5 postcoitum embryos were isolated from parental crosses of Snf5lox/⫹ mice or

crosses between Snf5lox/lox and Snf5⫹/⫺ mice (38) to generate Snf5lox/⫹, Snf5lox/lox, and Snf5lox/⫺ MEFs. The cells were infected with an AdCre, and we confirmed by Southern blotting that the infection resulted in a virtually com-

2664


MOL. CELL. BIOL.

FIG. 2. Snf5 depletion impairs cell cycle progression and increases the rate of apoptosis under growth-inhibiting conditions. (A) Cell cycle analysis of mock- and AdCre-infected Snf5lox/⫹ and Snf5lox/⫺ MEFs, 6 days after infection. The cells were cultured in 10% serum-containing medium or serum starved for 72 h before the analysis. DNA content was determined by propidium iodide staining and flow cytometric analysis. (B) BrdU incorporation assays of Snf5lox/⫹, Snf5lox/⫺, and Snf5lox/lox MEFs that were mock or AdCre infected. The cells were incubated with BrdU for 30 min and processed for immunostaining. The graph is representative of three independent experiments performed on different preparations of MEFs.

plete deletion of the floxed sequence of the gene (Fig. 1B). Western blot analysis revealed that 4 days after infection, the level of Snf5 protein in infected Snf5lox/lox MEFs dropped significantly (Fig. 1C). The protein level remained unchanged in mock- or Cre-infected Snf5lox/⫹ cells, indicating the existence of a compensation mechanism that adjusts the SNF5 protein level in heterozygotes. Transcriptional compensation of Snf5 in heterozygous mouse has been described previously (23). We measured the proliferation rate of wild-type, heterozygous, and Snf5-deficient cells. The difference in the growth rates between control- and Cre-treated Snf5lox/⫹ MEFs was modest, suggesting that ectopic Cre expression from our adenoviral vector did not affect cell growth to any appreciable extent. In contrast, Cre-mediated excision of Snf5 from Snf5lox/ lox/⫺ lox and Snf5 MEFs severely impaired their growth (Fig. 1E). Moreover, 6 days postinfection (dpi), morphological changes were observed in AdCre-infected Snf5lox/lox MEFs, which did not occur in infected Snf5lox/⫹ MEFs. These changes included a significant number of detached cells, with the remaining adherent cells displaying elongated and flattened or rounded morphology (Fig. 1D). In situ detection of ␤-galactosidase activity at pH 6.0 was performed on AdCre-infected Snf5lox/lox and Snf5lox/⫹ MEFs to assess for senescence (16).

Loss of Snf5 did not affect the percentage of senescence-associated, ␤-galactosidase-positive cells (data not shown). To investigate whether Snf5-depleted cells were arrested at a specific stage of the cell cycle, we analyzed the cells for their DNA content by flow cytometry (Fig. 2A). At 4 days postinfection, unsynchronized knockout cells had a profile of DNA content similar to that of control cells. However, 6 days after infection, we observed a modest but consistent decrease in the percentage of cells in the G1 phase and a corresponding increase in the fraction of cells with sub-G1 DNA content. When Snf5-depleted cells were cultured in low serum for 3 days, the apoptotic sub-G1 population increased significantly, reaching up to 30% of the total population. To examine whether Snf5deficient cells proceed through DNA replication, we monitored BrdU incorporation. By 6 days postinfection, Snf5-null cells exhibited a 50 to 70% decrease in the percentage of cells replicating their DNA (Fig. 2B). Since Snf5 depletion did not affect the percentage of cells in S phase (Fig. 2A), this result may reflect a slower progression of the replication forks or unscheduled DNA replication with subsequent triggering of an S-phase checkpoint. Further, the fluorescence-activated cell sorter (FACS) profile of serum-starved, Snf5-depleted MEFs revealed an increased subpopulation with S-phase DNA con-

VOL. 26, 2006


tent yet no BrdU incorporation (Fig. 2A and data not shown), a profile consistent with a defective G1 arrest coupled with activation of an S-phase checkpoint. The similar cell cycle profiles and the unaltered BrdU incorporation of control- and AdCre-infected Snf5lox/⫹ cells indicated that expression of Cre itself had no significant effect on the ability of cells to proliferate or progress through the cell cycle. Therefore, the data presented above indicate that loss of Snf5 in primary fibroblasts impairs cell viability and cell cycle progression at multiple stages of the cell cycle. Snf5 deletion does not result in increased expression of G1/S E2F target genes. Previous studies suggested that the SWI/SNF complex participates in the regulation of cell cycle progression via its role in pRb-mediated corepression of E2F1 target genes (39, 59). We reasoned that the phenotype observed might result from inappropriate activation of E2F1. We therefore examined the effect of Snf5 deletion on the expression of several E2F1-responsive genes. Interestingly, there were no detectable changes in the levels of cyclin E, E2F1 (Fig. 3A), and cyclin A mRNA (not shown). Furthermore, upon serum starvation, cyclin E and cyclin A levels were reduced in both Snf5-positive and -negative cells (data not shown), indicating that in primary fibroblasts, Snf5 is not required for Rb-mediated repression of G1/S-specific E2F target genes. These results are consistent with a previous study showing that human SNF5, unlike human BRG1, is not required for Rb-mediated cell cycle arrest (6). DHFR and TK, both E2F1 target genes involved in the synthesis of deoxynucleotides essential for DNA synthesis, were, on the contrary, downregulated in Snf5null cells (Fig. 3A). Real-time PCR confirmed that the transcriptional expression of both genes was decreased by 2-fold and 1.5-fold, respectively (data not shown). This finding is consistent with Snf5-depleted cells being impaired in S-phase progression, as suggested by the decrease in BrdU incorporation. p53 accumulates in Snf5-depleted cells. We next investigated whether the increased apoptosis that we observed in the absence of Snf5 could be related to an upregulation of proapoptotic genes. We therefore examined the expression of p53 and several of its target genes in mock- and AdCre-infected Snf5lox/⫹, Snf5lox/lox, and Snf5lox/⫺ MEFs (Fig. 3B and C). By 4 days after infection, Snf5 deletion led to an increase in p53 protein levels compared to those of control cells. Serum withdrawal, between days 4 and 6, caused a further increase in the level of p53 protein in Snf5-depleted MEFs compared to that of their wild-type or heterozygous counterparts (Fig. 3B). p53 activation was accompanied by a two- to threefold transcriptional induction of the p53-responsive genes Mdm2 and Puma (which encodes a BH3-only proapoptotic Bcl-2 family member). In contrast, we did not observe effects of Snf5 depletion on the protein levels of p21 and Bax, two other p53 target genes that mediate the p53-induced G0/G1 arrest and apoptosis, respectively (Fig. 3C). The absence of p21 induction may explain the reduction in the G1 population of Snf5-null cells, as detected by FACS analysis. Additional p53-responsive genes (Gadd45, Noxa, Apaf1) involved in growth arrest or apoptosis were unaffected by Snf5 loss despite the p53 accumulation (data not shown). We also verified the expression pattern of the p53 family member p73. The p73 gene contains two distinct promoters, P1 and P2, that are used to generate transcripts

2665

FIG. 3. Expression analysis of E2F target genes and induction of p53 in Snf5-deficient cells. Semiquantitative RT-PCR analysis of E2Fresponsive genes (A) and p53 target genes (C). Total RNA was isolated from mock (⫺)- and AdCre (⫹)-infected Snf5lox/lox, Snf5lox/⫺, and Snf5lox/⫹ MEFs, 6 days postinfection. PCR was carried out under linear amplification conditions with 0.01 ␮l and 0.03 ␮l of the 25-␮l cDNA sample. One representative experiment of four is shown. Similar results were obtained with AdCre-infected Snf5lox/lox and Snf5lox/⫺ MEFs (C). GAPDH expression is shown as a control. (B) Western blot analysis of lysates prepared from mock- or AdCre-infected Snf5lox/⫺ and Snf5lox/⫹ MEFs, 4 and 6 days after infection. The cells were grown in 10% serum containing medium or in 0.1% serum (low serum) for 72 h as indicated.

encoding TAp73 and ⌬Np73, respectively (77). TAp73 expression is directly induced by E2F1 and contributes to E2F1mediated apoptosis via activation of p53 target genes (19, 32, 45, 79). The ⌬Np73 promoter is activated by p53 and has been

2666


MOL. CELL. BIOL.

FIG. 4. p19ARF induction and genotoxic stress in Snf5-depleted cells. (A, C) Semiquantitative RT-PCR analysis of Snf5, p19ARF (A), and p16-INK4A (C). Total RNA was isolated from mock (⫺)- and AdCre (⫹)-infected Snf5lox/lox and Snf5lox/⫹ MEFs, 4 and 6 days after infection as indicated (A) and from Snf5lox/lox, Snf5lox/⫺, and Snf5lox/⫹ MEFs 6 days after infection (C) and assayed as described in the legend to Fig. 3. GAPDH expression is shown as a control (A). The GAPDH control for panel C is included in Fig. 3A. (B) Protein levels of p19ARF in mock- and AdCre-infected Snf5lox/⫺ and Snf5lox/⫹ MEFs, 4 and 6 days after infection, were determined by Western blot analysis. The cells were grown in 10% serum-containing medium or in 0.1% serum (low serum) for 72 h as indicated. (D) Phosphorylation of p53 at serine 18 is detected in Snf5-depleted MEFs. Lysates from mock- or AdCre-infected Snf5lox/lox and Snf5lox/⫹ MEFs were analyzed by Western blot analysis. Extracts from cells harvested 16 h after UV irradiation (10 J/m2) were included as a positive control.

shown to act as a dominant inhibitor of p53 and TAp73 in several experimental systems (36, 70). ⌬Np73 was strongly induced in SnfF5-null MEFs (Fig. 3C), whereas TAp73 transcripts were undetectable and were not affected by Snf5 ablation (not shown). We have confirmed by real-time PCR that ⌬Np73 transcripts are undetectable in Snf5-positive MEFs and induced in Snf5-deficient MEFs (data not shown). We conclude that the p53 protein that accumulates in Snf5deficient MEFs selectively upregulates a specific subset of genes, among them Puma, which may contribute to apoptosis, and ⌬Np73, which could attenuate the p53-dependent response to Snf5 loss. Mechanisms of p53 activation. Several types of cellular stress, including DNA damage, activate the p53 pathway and promote cell cycle arrest or apoptosis (50, 56). In addition,

aberrant activation of oncogenes, such as Ras, E1A, or Myc, stimulates the transcription of the ARF tumor suppressor gene (5, 15, 52, 82). When induced, the p19ARF protein sequesters Mdm2 to the nucleolus and blocks its shuttling to the cytoplasm (66, 73). This prevents p53 degradation by Mdm2 and leads to increased p53 stability. We therefore examined a possible effect of Snf5 loss on ARF gene expression at both the RNA and protein levels. When Snf5lox/lox and Snf5lox/⫺ MEFs were infected with AdCre, we observed induction of ARF transcription by 4 days postinfection (Fig. 4A). This correlated with increased accumulation of p19ARF protein at 6 dpi (Fig. 4B). ARF is encoded by the alternative reading frame of the INK4a locus, which also encodes the p16 cyclin-dependent kinase (Cdk4/Cdk6) inhibitor. In contrast to ARF, p16INK4a was downregulated in MEFs lacking Snf5, in agreement with pre-

VOL. 26, 2006


vious studies that showed that SNF5 activates p16INK4a transcription (Fig. 4C) (6, 51). We have confirmed p16INK4a downregulation (2.5-fold reduction) in Snf5-deficient MEFs by realtime PCR (data not shown). The fact that the delayed kinetics of p19ARF protein increase versus p53 accumulation in Snf5depleted cells suggested that the initial activation of p53 might involve an alternative pathway. Genotoxic stress induces p53 via posttranslational modifications of the protein. A well-characterized modification triggered by DNA damage is the phosphorylation of serine 15 (or the corresponding serine, serine 18, in mice), which increases p53 protein stability and/or activity (4, 64). To determine whether Snf5 loss-induced p53 activation occurs through DNA damage signaling, we analyzed the phosphorylation status of Ser18 in Snf5-depleted MEFs. As shown in Fig. 4D, phosphorylated p53 accumulated in AdCreinfected Snf5lox/lox MEFs but not in Cre-treated Snf5lox/⫹ cells. p53 phosphorylation was induced by UV irradiation in both normal and Snf5-depleted cells. The level of phosphorylation caused by Cre-mediated deletion of Snf5 roughly corresponded to that induced by 10 J/m2 of UV irradiation. Furthermore, the kinetics of phosphorylation coincided with p53 accumulation. Thus, our results suggest that the cellular response to DNA damage followed by p19ARF upregulation activates p53 in Snf5-null MEFs. Defective DNA repair in Snf5-null cells. The DNA damage response in Snf5-null MEFs could result from failure to repair spontaneous DNA damage caused by oxidative stress occurring under standard culture conditions (53), from defects in DNA replication, such as collapsed replication forks, or from impaired chromosome condensation or segregation. Recent studies indicate that the yeast SWI/SNF complex can stimulate nucleotide excision repair (NER) of damaged chromatin in vitro (20, 25, 26). This repair pathway is used for the removal of UV-induced cyclobutane-pyrimidine dimers and pyrimidine (6–4) photoproducts. To test whether Snf5 was indeed required for a proper cellular response to such damage, we examined the viability of AdCre-infected Snf5lox/lox, Snf5lox/⫺, and Snf5lox/⫹ MEFs after irradiation with increasing doses of UV irradiation. Snf5-null MEFs were 3.5- to 6-fold more sensitive to UV radiation than Snf5-heterozygous MEFs (Fig. 5A). This finding indicates that Snf5 plays an essential role in the ability to survive UV-induced DNA damage, suggesting that it could function in DNA repair in vivo. In normal MEFs, UV radiation triggers a p53-independent G1 checkpoint and a G2/M checkpoint activated by the ATR/ Chk1 or p38 kinase pathway. Although our findings raise the possibility that DNA excision pathways are impaired in the absence of Snf5, they do not exclude a role for Snf5 in checkpoint control. Snf5-null cells also displayed increased sensitivity, although less marked, to DNA double-strand breaks caused by the genotoxic agent doxorubicin (Fig. 5B). Thus, Snf5 deletion causes increased sensitivity to distinct genotoxic agents, establishing that the gene plays an essential role in the ability to survive DNA damage. Aberrant mitotic features in Snf5-null cells. Snf5-null MEFs also displayed aberrant nuclear morphology, which is characteristic of impaired mitosis and/or cytokinesis failure. We detected a two- to threefold increase in micronucleus formation in these cells in comparison with that of heterozygous MEFs (Fig. 5C). Micronuclei are indicative of lagging chromosomes

2667

and can appear as a result of a failure to activate the G2/M checkpoint in the presence of incomplete replication or repair or from abnormal spindle organization. Furthermore, we could detect interphase nuclei with bridging chromatin, a feature that results from unbalanced chromosomal segregation or defective cytokinesis. As centrosomes play a central role in spindle assembly, we determined the centrosome number in Snf5depleted cells versus that of heterozygotes. The frequency of Snf5-null cells containing more than two centrosomes was more than double that seen in heterozygous cells. Most Snf5null cells with increased numbers of centrosomes had large, partially cleaved, bilobed nuclei or micronuclei, features highly suggestive of aberrant cytokinesis (Fig. 5D). Consistently, a recent study reports that loss of SNF5 function promotes polyploidization and genomic instability in human cells (71). Growth properties of p53-null, Snf5-deficient MEFs. To examine further the contribution of p53 induction to the growth and survival defects caused by Snf5 ablation, we prepared MEFs from p53⫺/⫺; Snf5lox/⫹ mouse offspring and determined the growth rate of AdCre-infected p53⫺/⫺; Snf5lox/lox and p53⫺/⫺; Snf5lox/⫹ cells over 10 days. The absence of p53 did not alleviate the decreased proliferation capacity of Snf5-depleted cells (Fig. 6A). By 12 days postinfection, AdCre-infected p53⫺/⫺; SNF5lox/lox cultures resumed a proliferation rate similar to that of Snf5lox/⫹ cultures. However, Western blot analysis revealed that in the AdCreinfected p53⫺/⫺; Snf5lox/lox cells, the level of Snf5 increased progressively (Fig. 6B). Moreover, the cells remaining in culture 15 days postinfection were heterozygous, harboring one floxed and one deleted Snf5 allele (Fig. 6C). Thus, prolonged culture of AdCre-infected p53⫺/⫺; Snf5lox/lox MEFs led to the selective loss of Snf5-depleted cells and enrichment in the small cell population that had escaped biallelic deletion. However, inactivation of p53 protected Snf5-deficient MEFs from cell death. Cre-infected p53⫺/⫺; Snf5 lox/lox MEFs showed a ⬎2-fold decrease in the percentage of apoptotic cells (Fig. 6D). This observation correlated with the absence of Puma induction in these cells, in contrast with Cre-infected p53 wildtype; Snf5lox/lox MEFs (Fig. 6E). Further, the enhanced apoptosis observed in serum-deprived Snf5⫺/⫺ MEF cells was also strikingly reduced in p53-null, Snf5-depleted MEFs (Fig. 6D). From these experiments, we conclude that in this cellular context, Snf5 loss triggers a p53-dependent apoptotic response, in which Puma might be involved, and a p53-independent proliferation arrest. p53 mutation accelerates rhabdoid tumor development in Snf5ⴙ/ⴚ mice. It was also interesting to monitor the consequence of the absence of p53 in our tumor-prone Snf5-heterozygous mice. Would the absence of an apoptotic response accelerate tumor formation? A comparison of the tumor incidence of Snf5⫹/⫺ mice with that of Snf5⫹/⫺; p53⫺/⫺ compound-mutation mice revealed a strong genetic interaction between the inactivation of these two tumor suppressor genes (Fig. 7A). Whereas 28% of Snf5⫹/⫺ mice developed tumors with a mean latency of 42 weeks, 100% of the Snf5⫹/⫺; p53⫺/⫺ mice (n ⫽ 49) developed tumors within 19 weeks (2 months earlier than animals homozygous for the p53 mutation alone). Interestingly, while p53 deficiency accelerated tumorigenesis, the distribution of the tumors in the different anatomic locations, as well as their type (sarcomas, some of them with

2668


MOL. CELL. BIOL.

VOL. 26, 2006


2669

FIG. 6. p53-null background does not rescue the slowed proliferation phenotype in Snf5-depleted cells but reduces apoptosis. (A) Growth curves of mock- and AdCre-infected p53-null MEFs with the indicated genotype for the Snf5 locus. The experiment was performed as described for Fig. 1E, and the graph shows a representative of two independent experiments. Western blot (B) and Southern blot (C) analyses of DNA and lysates prepared from mock- and AdCre-infected p53-null MEFs with the indicated genotypes, 5, 12, and 15 days postinfection. (D) p53 wild-type and p53-null MEFs with the indicated genotypes at the Snf5 locus were infected with AdCre and grown in 10% or 0.1% serum-containing medium for 72 h. The percentage of cell death was determined by trypan blue exclusion in three independent experiments. The percentages of cells with a sub-G1 content of DNA as determined by FACS analysis were similar. (E) Semiquantitative RT-PCR analysis of Puma. Total RNA was isolated from mock (⫺)- and AdCre (⫹)-infected MEFs with the indicated genotype 6 days after infection and assayed as described in the legend to Fig. 3. GAPDH expression is shown as a control.

rhabdoid features), was not altered. Fifty-five percent of the tumors developing in Snf5⫹/⫺; p53⫺/⫺ mice arose in the nervous system (brain, cerebellum [shown in Fig. 7C], or spinal cord). Subcutaneous tumors developed in other sites, including

cheek, limb, eye, and axillary sites, as observed in Snf5-heterozygous animals (Table 1). Twenty-five percent of the animals developed two concurrent tumor foci affecting different tissues, such as brain or cerebellum and cheek or limb; these

FIG. 5. Snf5-null cells display increased sensitivity to genotoxic agents, abnormal mitotic features, and amplification of centrosomes. Mock- and AdCre-infected MEFs with the indicated genotypes were plated in six-well culture dishes and exposed to increasing doses of UV light (A) or doxorubicin (B) at day 3 after infection. The percentage of cell death in each well was assessed by trypan blue exclusion 72 h (A) or 24 h (B) after treatment. Each data point represents the mean of six observations. Similar results were obtained in three independent experiments. The difference in UV hypersensitivity between AdCre-infected Snf5lox/lox and Snf5lox/⫺ MEFs is apparently due to experimental variability. All the Snf5-deficient clones (Cre-infected Snf5lox/lox or Snf5lox/⫺) tested were three- to sixfold more sensitive to UV radiation than their heterozygous counterparts, independently of whether the infected cells carried two floxed alleles or one floxed and one null allele. (C, D) Snf5lox/⫺ and SNF5lox/⫹ MEFs were infected with AdCre and immunostained 4 days after infection. (C) AdCre-infected Snf5lox/⫺ MEFs were stained with anti-␣-tubulin (green) and counterstained with DAPI. White arrows indicate micronuclei, and yellow arrows point to interphase bilobed nuclei with bridging chromatin. (D) AdCre-infected Snf5lox/⫺ MEFs were stained for centrosomes with anti-␥-tubulin (green) and counterstained with DAPI. Micronuclei (C) and centrosomes (D) of infected Snf5lox/⫺ MEFs and Snf5lox/⫹ MEFs were counted for 100 cells per sample in three independent experiments.

2670

MOL. CELL. BIOL.


FIG. 7. Snf5 and p53 deficiency cooperate in tumorigenesis in vivo. (A) Kaplan-Meier tumor-free survival curves of Snf5⫹/⫺ (n ⫽ 32), p53⫺/⫺ (n ⫽ 25), and Snf5⫹/⫺; p53⫺/⫺ (n ⫽ 49) mice (confidence interval [C.I.], 95%). The curves were compared pairwise by the log rank test (P ⬍ 0.0001). The mean age of survival was 12.5 weeks for Snf5⫹/⫺; p53⫺/⫺ mice and 26 weeks for p53⫺/⫺ mice. At 60 weeks, 72% of Snf5⫹/⫺ mice were still alive and tumor free. (B to D) Photomicrographs of hematoxylin- and eosin-stained slides from Snf5⫹/⫺; p53⫺/⫺ mice (magnification, ⫻400). Section of normal cerebellum (B) versus section showing infiltration of malignant cells into the molecular layer (C). Black and red arrows point to mitotic and apoptotic figures, respectively. (D) Representative section of a brain rhabdoid tumor. The arrows point to rhabdoid cells with typical prominent nucleoli and eosinophilic hyaline cytoplasmic inclusions. (E) LOH at the Snf5 locus is observed in tumors (T) from Snf5⫹/⫺; p53⫺/⫺ mice. Southern blot analysis of DNA extracted from tumors shows that the wild-type Snf5 allele is greatly reduced, while the ratio of mutant to wild-type alleles is roughly 1 in adjacent normal tissues (N).

tumors displayed histological characteristics of Snf5-associated lesions (undifferentiated sarcoma, some of them with rhabdoid features). Histological examination of 22 tumors has been performed. Among these 22 tumor samples, 12 (60%) displayed rhabdoid features (presence of typical rhabdoid cells) (Fig. 7D). The other lesions were defined as high-grade sarcomas (Table 1). Importantly, histological analysis revealed that Snf5⫹/⫺; p53⫺/⫺ mice did not develop thymic lymphoma, a tumor type associated with mutation of p53. We believe that the early death of p53-null, Snf5-heterozygous mice due to the development of Snf5-associated tumors has precluded the further development of lymphomas. Southern blot analysis of DNA from six tumors demonstrated that loss of the wild-type Snf5 allele accompanied tumorigenesis in all tested tumors (Fig. 7E). Thus, loss of p53 accelerates Snf5-associated tumorigenesis without altering the tumor spectrum in Snf5-hetero-

zygous mice, implying that inactivation of p53 and Snf5 cooperate in tumorigenesis in specific tissues in mice. DISCUSSION The SNF5/INI1 protein was initially identified as a core subunit of the mammalian SWI/SNF complex (48, 55). Gene inactivation in mice revealed that Snf5 and several other core subunits of the SWI/SNF complex are essential for early mouse development (11, 22, 37, 38, 57). In culture, Snf5⫺/⫺ cells from the inner cell mass and trophoblasts fail to proliferate and undergo apoptosis (38). Paradoxically, biallelic inactivation of the SNF5 gene is associated with aggressive rhabdoid tumors in humans or mice (39, 59). These data suggest that Snf5 may be essential in some cell types, whereas other cells become highly malignant in its absence. To unravel essential and growth-control-


VOL. 26, 2006 TABLE 1. Incidence and location of tumors in mice with mutations in Snf5 and p53 % found in mice of indicated genotype Tumor parameter

⫹/⫺

Snf5 (n ⫽ 37)

Snf5⫹/⫺; p53⫺/⫺ (n ⫽ 22)

p53⫺/⫺ (n ⫽ 15)

Type Rhabdoida Lymphoma Sarcoma

27 0 73

60 0 40

0 75 25

Location Nervous system Limb Cheek Thymus Other

57 5 8 0 30

55 8 10 0 27

0 0 0 68 32

a Rhabdoid tumors are diagnosed by the presence of typical rhabdoid cells in the histological section as described in the legend to Fig. 7. The higher proportion of rhabdoid tumors in the Snf5⫹/⫺; p53⫺/⫺ mice compared with that of Snf5⫹/⫺ mice might be biased as the histological data for the tumors developing in Snf5⫹/⫺ mice have been taken from our previous study (38). It is worth mentioning that also in humans, the histological and immunohistochemical characteristics of rhabdoid tumors are imprecise and have led to diagnostic controversies.

ling functions of this gene, we studied the consequence of Snf5/Ini1 deletion in primary mouse fibroblasts by using the Cre/loxP system. The results presented here demonstrate that Snf5 depletion affects several distinct pathways, including the DNA damage response pathway and cell proliferation. Snf5 deletion causes an impaired response to genotoxic stress and growth arrest. By deleting Snf5 in primary MEFs, we have established that this gene is essential for cell growth and survival under normal conditions in culture. The growth defect is attributable, at least in part, to an increase in the incidence of spontaneous apoptotic cell death, which was even higher under serum-starved conditions. This phenotype is consistent with the early embryonic death of Snf5-null mice and bone marrow failure in mice in which Snf5 was conditionally targeted (58). However, increased cell death alone cannot account for the proliferative defect of Snf5-deficient MEFs. A large number of cells arrested in the G1, S, and G2/M phases still remained viable 6 days after AdCre infection. Moreover, p53-null; Snf5-null cells did not exhibit significant apoptosis but still failed to grow. Growth inhibition in primary cells can be the result of defects in receiving growth- and survival-promoting signals or of failure to perform other essential functions of the cell. The data that we have obtained in this study suggest that Snf5deficient cells accumulate DNA damage due to a defect in their DNA repair process. Consistently, Snf5 loss was accompanied by increased expression of phosphorylated p53, as discussed below. This was corroborated by the higher sensitivity of Snf5-null MEFs to DNA-damaging UV irradiation. Increased sensitivity to the double-strand-break-inducing agent doxorubicin was also observed. These findings agree with previous studies that showed that immobile nucleosomes inhibit in vitro repair of UV-induced DNA lesions (24). Moreover, yeast SWI/SNF has been directly implicated in stimulating NER on nucleosomal substrates in vitro (20, 25, 26). It is plausible that SWI/SNF enhances repair by facilitating access to UV-dam-

2671

aged DNA sites. Loss of Snf5 might impair this function. We cannot formally exclude the possibility that the increased UV radiation sensitivity of Snf5⫺/⫺ cells reflects downregulation of genes involved in DNA repair. However, transcriptome analysis following Snf5 deletion did not reveal any decrease in the expression of known genes encoding NER components (data not shown). Another link between the SWI/SNF complex and chromosome integrity was suggested by the isolation of a SWI/ SNF-related complex containing the BRCA1 protein (9). In this context, it is important to recall recent studies that revealed an essential role for the INO80 chromatin-remodeling complex in double-strand break repair in yeast (18, 47, 67). However, unlike the INO80 complex, the SWI/SNF complex does not contain ATPases of the AAA⫹ family, proteins related to the bacterial RuvB protein, a double-hexameric DNA helicase involved in Holliday junction resolution. Increased sensitivity to genotoxic agents may be caused not only by defects in DNA repair but also by disruption of cell cycle checkpoint responses to DNA damage. Pertinent examples come from cells deficient in ATR or Chk1 genes, which code for DNA damage checkpoint proteins. These cells exhibit progressive proliferative defects and are highly sensitive to genotoxic injury (10, 78). In addition, cell cycle arrest could be triggered by a p53-independent checkpoint response to spontaneous DNA damage. Snf5-deficient cells showed signs of severe misregulation of mitosis, with an increase in micronucleus formation indicative of lagging chromosomes, and the presence of lobulated nuclei connected by bridging chromatin, which reflects cytokinesis failure. These features are characteristic of DNA damage-induced mitotic catastrophe or could reflect defects in the spindle attachment checkpoint and chromosome segregation (3, 12). Mitotic catastrophe can also be triggered by the DNA damage-induced, Chk1-dependent checkpoint (30). Interestingly, the yeast SWI/SNF-related RSC complex has been implicated in chromatin structures at centromeres and kinetochore function, and mutation of sfh1, an Snf5 homolog, causes chromosome missegregation (13, 28). Moreover, RSC has a role in cohesin loading onto chromosome arms and establishment of arm cohesion (29). In human cells, the RSC-related SWI/SNF-B complex localizes at kinetochores and spindle poles in prometaphase cells, suggesting that it could play a role there in mitosis (76). Finally, a study published recently shows that SNF5 mutation impairs chromosome segregation in human cells and that loss of SNF5 results in elevated poly- and aneuploid cells (71). Thus, it is possible that SNF5 and the SWI/SNF complex play multiple roles in controlling DNA damage and mitosis. Snf5 inactivation and the pRb-E2F pathway. Our understanding of the molecular parameters involved in Brg1- or Snf5-mediated regulation of cellular proliferation has relied mostly on overexpression experiments, namely, reintroducing the missing gene into cell lines and analyzing the expression of key cell cycle regulators. These experiments led to the conclusion that both Brg1 and Snf5 induce G0/G1 arrest via downregulation of a specific subset of E2F-regulated genes in collaboration with pRb (39, 59). Additional studies showed that Snf5 reexpression at physiological levels reinforces pRb-mediated transcriptional repression via induction of the cyclin-dependent kinase inhibitor p16INK4A (6, 51). Indeed, we found that Snf5 inactivation in MEFs caused p16INK4A downregula-

2672


tion. However, the reduction in p16INK4A expression level was not associated with upregulation of E2F target genes and accelerated cell cycle progression. In contrast, inactivation of Snf5 in primary MEFs impaired cell proliferation and survival while the expression levels of the E2F-responsive genes, such as cyclin A, cyclin E, cdc2, and E2F1, were not affected. Moreover, we observed a decrease in the transcription of the TK and DHFR genes, which may contribute to the reduction in the fraction of surviving cells that replicate their DNA. It is possible, however, that p16INK4A downregulation loosens the regulation of cell cycle checkpoints through pathways that do not involve pRb-E2F. The discrepancy with previous studies could also result from cell-type-specific modes of action of Snf5. Alternatively, Snf5 could be required in our system, not only for pRb-mediated repression but also for E2F-mediated transcriptional activation of these specific genes. The only E2F target gene that we found induced by Snf5 depletion was p19ARF. However, additional regulators of p19ARF have been identified, i.e., both activators (DMP-1, AP1) and transcriptional repressors (Bmi-1, Tbx2, Tbx3, and Twist) (2, 31, 34, 42, 43). Thus, it remains unclear whether Snf5 disruption activates p19ARF via E2F. Snf5 and the p53 pathway. When ARF is induced, it activates p53 indirectly, by binding to Mdm2 and inhibiting p53-directed E3 ubiquitin ligase activity and by sequestering Mdm2 into the nucleolus (73). Activated p53 induces transcription of genes involved in cell cycle arrest and apoptosis. The growth and survival defects that we observed in Snf5depleted cells were indeed accompanied by p53 accumulation. The p53 protein that accumulates in Snf5-depleted MEFs is phosphorylated on serine 18. This modification plays an essential role in both the stabilization and activation of the p53 protein. Although p53 phosphorylation is typically induced by genotoxic stress via ATM/ATR kinase activation, a recent study contradicts previous data and indicates that p19ARF can also induce ATR- and Chk1-dependent phosphorylation of p53 (60). However, since p53 phosphorylation and stabilization preceded the accumulation of p19ARF, we favor the first hypothesis. Interestingly, only a subset of p53-responsive genes, including Mdm2, Puma, and ⌬N-p73, were activated following p53 induction in Snf5-null cells, whereas p21, Bax, and Gadd45a transcription levels remained unaltered. Chromatin structures at the promoters of p53 target genes might dictate the differential requirement for Snf5. Promoters with an open configuration might be activated by p53 in the absence of Snf5, whereas other regulatory sequences might require chromatin-remodeling activity to potentiate p53-mediated transcriptional activation. Consistently, research from several laboratories has implicated BRG1 and SNF5 in the activation of p21 gene transcription through both p53-dependent and -independent mechanisms (27, 35, 41). The increased expression of proapoptotic genes like Puma in the absence of p21 activation, as well as the activation of p19ARF coupled with the decrease in p16INK4A expression, could favor apoptosis over DNA repair and survival in Snf5-null cells. In contrast with the enhanced apoptosis, the decreased proliferation of Snf5-null MEFs is p53 independent, as p53 inactivation did not abolish this phenotype. One could postulate that the accumulation of spontaneous DNA damage and chromosome segregation defects are responsible for the proliferation defect that is observed in the absence of p53. In vivo, p53

MOL. CELL. BIOL.

and Snf5 inactivation synergize to promote tumorigenesis such that all Snf5⫹/⫺; p53⫺/⫺ mice succumb to Snf5-associated tumors before 4 months of age. The early development of these tumors can account for the absence of p53-associated lesions in the compound-mutation mice. It is worth noting that the tumor spectrum of our Snf5-heterozygous mutants (on both p53 wildtype and -null backgrounds) is different from that of the previously published Snf5-conditional mice (i.e., mature T-cell lymphomas) (58). Tumor development in our system depends on loss of heterozygosity at the Snf5 locus, whereas Snf5 inactivation in the conditional system is mediated by Cre recombinase, which has been shown to be highly expressed in the thymus and spleen. It is therefore plausible that the different knockout systems generate distinct tumor spectrums. The acceleration of tumor onset by p53 deletion can be explained by different mechanisms. First, p53 deficiency may increase the proportion of cells that lose the remaining wild-type Snf5 allele or accelerate the timing of the loss of heterozygosity at the Snf5 locus. Second, p53 inactivation could impair the differentiation of cells that become malignant upon Snf5 inactivation. Finally, p53 deficiency might enhance the survival of tumor cells after the LOH event at the Snf5 locus. Considering the fact that tumors in Snf5⫹/⫺ mice arise primarily in the nervous system, it is interesting to note that p53 is involved in key survival/ death checkpoints in both peripheral and central neurons (1, 75). Moreover, p53 is involved in the differentiation of oligodendrocyte precursor cells, and overexpression of ⌬N-p73, the dominant negative p53 family member that is strongly induced upon Snf5 loss in MEFs, inhibits oligodendrocyte precursor cell differentiation (8). Thus, the cooperation between p53 and Snf5 inactivation in tumorigenesis may extend beyond the genomic instability which is characteristic of p53-deficient cells. In conclusion, our present study clearly demonstrates that Snf5, a member of the core SWI/SNF chromatin remodeling complex, is essential for normal cell growth. Snf5 deletion results in multiple dysfunctions, ranging from a failure to survive DNA damage to an unbalanced transcription program driven by activated p53. The failure to respond correctly to DNA damage and defects in chromosome segregation may generate additional mutations and explain why certain cell types escape apoptosis and become malignant in the absence of Snf5.

ACKNOWLEDGMENTS We are grateful to Moshe Oren for helpful discussions and Jonathan Weitzman for critical reading of the manuscript. We thank Aharon Razin, Ruthy Shemer, and the friends from the Department of Cell Biochemistry and Human Genetics (Hadassah Medical School, Hebrew University) for their help and contribution to this work. This work was supported by the Israel Cancer Research Fund (ICRF project grant) and Association for International Cancer Research grants 00-221 and 03-109 (A.K.-Y.).

ADDENDUM IN PROOF After submission of this paper, Isakoff et al. (Proc. Natl. Acad. Sci. USA 102:17745–17750, 2005) published a related study demonstrating apoptosis, polyploidy, and growth arrest in primary fibroblasts lacking Snf5/Ini1, as well as synergy with p53 inactivation in tumor formation.


VOL. 26, 2006 REFERENCES

1. Aloyz, R. S., S. X. Bamji, C. D. Pozniak, J. G. Toma, J. Atwal, D. R. Kaplan, and F. D. Miller. 1998. p53 is essential for developmental neuron death as regulated by the TrkA and p75 neurotrophin receptors. J. Cell Biol. 143: 1691–1703. 2. Ameyar-Zazoua, M., M. B. Wisniewska, L. Bakiri, E. F. Wagner, M. Yaniv, and J. B. Weitzman. 2005. AP-1 dimers regulate transcription of the p14/ p19ARF tumor suppressor gene. Oncogene 24:2298–2306. 3. Andreassen, P. R., F. B. Lacroix, O. D. Lohez, and R. L. Margolis. 2001. Neither p21WAF1 nor 14-3-3sigma prevents G2 progression to mitotic catastrophe in human colon carcinoma cells after DNA damage, but p21WAF1 induces stable G1 arrest in resulting tetraploid cells. Cancer Res. 61:7660–7668. 4. Banin, S., L. Moyal, S. Shieh, Y. Taya, C. W. Anderson, L. Chessa, N. I. Smorodinsky, C. Prives, Y. Reiss, Y. Shiloh, and Y. Ziv. 1998. Enhanced phosphorylation of p53 by ATM in response to DNA damage. Science 281:1674–1677. 5. Bates, S., A. C. Phillips, P. A. Clark, F. Stott, G. Peters, R. L. Ludwig, and K. H. Vousden. 1998. p14ARF links the tumour suppressors RB and p53. Nature 395:124–125. 6. Betz, B. L., M. W. Strobeck, D. N. Reisman, E. S. Knudsen, and B. E. Weissman. 2002. Re-expression of hSNF5/INI1/BAF47 in pediatric tumor cells leads to G1 arrest associated with induction of p16ink4a and activation of RB. Oncogene 21:5193–5203. 7. Biegel, J. A., L. Tan, F. Zhang, L. Wainwright, P. Russo, and L. B. Rorke. 2002. Alterations of the hSNF5/INI1 gene in central nervous system atypical teratoid/rhabdoid tumors and renal and extrarenal rhabdoid tumors. Clin. Cancer Res. 8:3461–3467. 8. Billon, N., A. Terrinoni, C. Jolicoeur, A. McCarthy, W. D. Richardson, G. Melino, and M. Raff. 2004. Roles for p53 and p73 during oligodendrocyte development. Development 131:1211–1220. 9. Bochar, D. A., L. Wang, H. Beniya, A. Kinev, Y. Xue, W. S. Lane, W. Wang, F. Kashanchi, and R. Shiekhattar. 2000. BRCA1 is associated with a human SWI/SNF-related complex: linking chromatin remodeling to breast cancer. Cell 102:257–265. 10. Brown, E. J., and D. Baltimore. 2003. Essential and dispensable roles of ATR in cell cycle arrest and genome maintenance. Genes Dev. 17:615–628. 11. Bultman, S., T. Gebuhr, D. Yee, C. La Mantia, J. Nicholson, A. Gilliam, F. Randazzo, D. Metzger, P. Chambon, G. Crabtree, and T. Magnuson. 2000. A Brg1 null mutation in the mouse reveals functional differences among mammalian SWI/SNF complexes. Mol. Cell 6:1287–1295. 12. Bunz, F., A. Dutriaux, C. Lengauer, T. Waldman, S. Zhou, J. P. Brown, J. M. Sedivy, K. W. Kinzler, and B. Vogelstein. 1998. Requirement for p53 and p21 to sustain G2 arrest after DNA damage. Science 282:1497–1501. 13. Cao, Y., B. R. Cairns, R. D. Kornberg, and B. C. Laurent. 1997. Sfh1p, a component of a novel chromatin-remodeling complex, is required for cell cycle progression. Mol. Cell. Biol. 17:3323–3334. 14. Carlson, M., and B. C. Laurent. 1994. The SNF/SWI family of global transcriptional activators. Curr. Opin. Cell Biol. 6:396–402. 15. de Stanchina, E., M. E. McCurrach, F. Zindy, S. Y. Shieh, G. Ferbeyre, A. V. Samuelson, C. Prives, M. F. Roussel, C. J. Sherr, and S. W. Lowe. 1998. E1A signaling to p53 involves the p19(ARF) tumor suppressor. Genes Dev. 12: 2434–2442. 16. Dimri, G. P., X. Lee, G. Basile, M. Acosta, G. Scott, C. Roskelley, E. E. Medrano, M. Linskens, I. Rubelj, O. Pereira-Smith, M. Peacocke, and J. Campisi. 1995. A biomarker that identifies senescent human cells in culture and in aging skin in vivo. Proc. Natl. Acad. Sci. USA 92:9363–9367. 17. Doan, D. N., T. M. Veal, Z. Yan, W. Wang, S. N. Jones, and A. N. Imbalzano. 2004. Loss of the INI1 tumor suppressor does not impair the expression of multiple BRG1-dependent genes or the assembly of SWI/SNF enzymes. Oncogene 23:3462–3473. 18. Downs, J. A., S. Allard, O. Jobin-Robitaille, A. Javaheri, A. Auger, N. Bouchard, S. J. Kron, S. P. Jackson, and J. Cote. 2004. Binding of chromatinmodifying activities to phosphorylated histone H2A at DNA damage sites. Mol. Cell 16:979–990. 19. Fontemaggi, G., I. Kela, N. Amariglio, G. Rechavi, J. Krishnamurthy, S. Strano, A. Sacchi, D. Givol, and G. Blandino. 2002. Identification of direct p73 target genes combining DNA microarray and chromatin immunoprecipitation analyses. J. Biol. Chem. 277:43359–43368. 20. Gaillard, H., D. J. Fitzgerald, C. L. Smith, C. L. Peterson, T. J. Richmond, and F. Thoma. 2003. Chromatin remodeling activities act on UV-damaged nucleosomes and modulate DNA damage accessibility to photolyase. J. Biol. Chem. 278:17655–17663. 21. Gresh, L., B. Bourachot, A. Reimann, B. Guigas, L. Fiette, S. Garbay, C. Muchardt, L. Hue, M. Pontoglio, M. Yaniv, and A. Klochendler-Yeivin. 2005. The SWI/SNF chromatin-remodeling complex subunit SNF5 is essential for hepatocyte differentiation. EMBO J. 24:3313–3324. 22. Guidi, C. J., A. T. Sands, B. P. Zambrowicz, T. K. Turner, D. A. Demers, W. Webster, T. W. Smith, A. N. Imbalzano, and S. N. Jones. 2001. Disruption of Ini1 leads to peri-implantation lethality and tumorigenesis in mice. Mol. Cell. Biol. 21:3598–3603. 23. Guidi, C. J., T. M. Veal, S. N. Jones, and A. N. Imbalzano. 2004. Transcrip-

24. 25. 26. 27. 28. 29. 30.

31. 32. 33. 34. 35. 36. 37.

38. 39. 40. 41.

42. 43. 44. 45.

46. 47.

48. 49. 50.

2673

tional compensation for loss of an allele of the Ini1 tumor suppressor. J. Biol. Chem. 279:4180–4185. Hara, R., J. Mo, and A. Sancar. 2000. DNA damage in the nucleosome core is refractory to repair by human excision nuclease. Mol. Cell. Biol. 20:9173– 9181. Hara, R., and A. Sancar. 2003. Effect of damage type on stimulation of human excision nuclease by SWI/SNF chromatin remodeling factor. Mol. Cell. Biol. 23:4121–4125. Hara, R., and A. Sancar. 2002. The SWI/SNF chromatin-remodeling factor stimulates repair by human excision nuclease in the mononucleosome core particle. Mol. Cell. Biol. 22:6779–6787. Hendricks, K. B., F. Shanahan, and E. Lees. 2004. Role for BRG1 in cell cycle control and tumor suppression. Mol. Cell. Biol. 24:362–376. Hsu, J.-M., J. Huang, P. B. Meluh, and B. C. Laurent. 2003. The yeast RSC chromatin-remodeling complex is required for kinetochore function in chromosome segregation. Mol. Cell. Biol. 23:3202–3215. Huang, J., J. M. Hsu, and B. C. Laurent. 2004. The RSC nucleosomeremodeling complex is required for Cohesin’s association with chromosome arms. Mol. Cell 13:739–750. Huang, X., T. Tran, L. Zhang, R. Hatcher, and P. Zhang. 2005. DNA damage-induced mitotic catastrophe is mediated by the Chk1-dependent mitotic exit DNA damage checkpoint. Proc. Natl. Acad. Sci. USA 102:1065– 1070. Inoue, K., M. F. Roussel, and C. J. Sherr. 1999. Induction of ARF tumor suppressor gene expression and cell cycle arrest by transcription factor DMP1. Proc. Natl. Acad. Sci. USA 96:3993–3998. Irwin, M., M. C. Marin, A. C. Phillips, R. S. Seelan, D. I. Smith, W. Liu, E. R. Flores, K. Y. Tsai, T. Jacks, K. H. Vousden, and W. G. Kaelin, Jr. 2000. Role for the p53 homologue p73 in E2F-1-induced apoptosis. Nature 407:645–648. Jacks, T., L. Remington, B. O. Williams, E. M. Schmitt, S. Halachmi, R. T. Bronson, and R. A. Weinberg. 1994. Tumor spectrum analysis in p53-mutant mice. Curr. Biol. 4:1–7. Jacobs, J. J., K. Kieboom, S. Marino, R. A. DePinho, and M. van Lohuizen. 1999. The oncogene and Polycomb-group gene bmi-1 regulates cell proliferation and senescence through the ink4a locus. Nature 397:164–168. Kang, H., K. Cui, and K. Zhao. 2004. BRG1 controls the activity of the retinoblastoma protein via regulation of p21CIP1/WAF1/SDI. Mol. Cell. Biol. 24:1188–1199. Kartasheva, N. N., A. Contente, C. Lenz-Stoppler, J. Roth, and M. Dobbelstein. 2002. p53 induces the expression of its antagonist p73 Delta N, establishing an autoregulatory feedback loop. Oncogene 21:4715–4727. Kim, J. K., S.-O. Huh, H. Choi, K.-S. Lee, D. Shin, C. Lee, J.-S. Nam, H. Kim, H. Chung, H. W. Lee, S. D. Park, and R. H. Seong. 2001. Srg3, a mouse homolog of yeast SWI3, is essential for early embryogenesis and involved in brain development. Mol. Cell. Biol. 21:7787–7795. Klochendler-Yeivin, A., L. Fiette, J. Barra, C. Muchardt, C. Babinet, and M. Yaniv. 2000. The murine SNF5/INI1 chromatin remodeling factor is essential for embryonic development and tumor suppression. EMBO Rep. 1:500–506. Klochendler-Yeivin, A., C. Muchardt, and M. Yaniv. 2002. SWI/SNF chromatin remodeling and cancer. Curr. Opin. Genet. Dev. 12:73–79. Langst, G., and P. B. Becker. 2004. Nucleosome remodeling: one mechanism, many phenomena? Biochim. Biophys. Acta 1677:58–63. Lee, D., J. W. Kim, T. Seo, S. G. Hwang, E. J. Choi, and J. Choe. 2002. SWI/SNF complex interacts with tumor suppressor p53 and is necessary for the activation of p53-mediated transcription. J. Biol. Chem. 277:22330– 22337. Lingbeek, M. E., J. J. Jacobs, and M. van Lohuizen. 2002. The T-box repressors TBX2 and TBX3 specifically regulate the tumor suppressor gene p14ARF via a variant T-site in the initiator. J. Biol. Chem. 277:26120–26127. Maestro, R., A. P. Dei Tos, Y. Hamamori, S. Krasnokutsky, V. Sartorelli, L. Kedes, C. Doglioni, D. H. Beach, and G. J. Hannon. 1999. Twist is a potential oncogene that inhibits apoptosis. Genes Dev. 13:2207–2217. Martens, J. A., and F. Winston. 2002. Evidence that Swi/Snf directly represses transcription in S. cerevisiae. Genes Dev. 16:2231–2236. Melino, G., F. Bernassola, M. Ranalli, K. Yee, W. X. Zong, M. Corazzari, R. A. Knight, D. R. Green, C. Thompson, and K. H. Vousden. 2004. p73 induces apoptosis via PUMA transactivation and Bax mitochondrial translocation. J. Biol. Chem. 279:8076–8083. Mohrmann, L., and C. P. Verrijzer. 2005. Composition and functional specificity of SWI2/SNF2 class chromatin remodeling complexes. Biochim. Biophys. Acta 1681:59–73. Morrison, A. J., J. Highland, N. J. Krogan, A. Arbel-Eden, J. F. Greenblatt, J. E. Haber, and X. Shen. 2004. INO80 and gamma-H2AX interaction links ATP-dependent chromatin remodeling to DNA damage repair. Cell 119: 767–775. Muchardt, C., C. Sardet, B. Bourachot, C. Onufryk, and M. Yaniv. 1995. A human protein with homology to Saccharomyces cerevisiae SNF5 interacts with the potential helicase hbrm. Nucleic Acids Res. 23:1127–1132. Muchardt, C., and M. Yaniv. 1999. ATP-dependent chromatin remodelling: SWI/SNF and co. are on the job. J. Mol. Biol. 293:187–198. Oren, M. 1999. Regulation of the p53 tumor suppressor protein. J. Biol. Chem. 274:36031–36034.

2674


51. Oruetxebarria, I., F. Venturini, T. Kekarainen, A. Houweling, L. M. Zuijderduijn, A. Mohd-Sarip, R. G. Vries, R. C. Hoeben, and C. P. Verrijzer. 2004. P16INK4a is required for hSNF5 chromatin remodeler-induced cellular senescence in malignant rhabdoid tumor cells. J. Biol. Chem. 279:3807– 3816. 52. Palmero, I., C. Pantoja, and M. Serrano. 1998. p19ARF links the tumour suppressor p53 to Ras. Nature 395:125–126. 53. Parrinello, S., E. Samper, A. Krtolica, J. Goldstein, S. Melov, and J. Campisi. 2003. Oxygen sensitivity severely limits the replicative lifespan of murine fibroblasts. Nat. Cell Biol. 5:741–747. 54. Peterson, C. L. 1996. Multiple SWItches to turn on chromatin? Curr. Opin. Genet. Dev. 6:171–175. 55. Phelan, M. L., S. Sif, G. J. Narlikar, and R. E. Kingston. 1999. Reconstitution of a core chromatin remodeling complex from SWI/SNF subunits. Mol. Cell 3:247–253. 56. Prives, C., and P. A. Hall. 1999. The p53 pathway. J. Pathol. 187:112–126. 57. Roberts, C. W., S. A. Galusha, M. E. McMenamin, C. D. Fletcher, and S. H. Orkin. 2000. Haploinsufficiency of Snf5 (integrase interactor 1) predisposes to malignant rhabdoid tumors in mice. Proc. Natl. Acad. Sci. USA 97:13796– 13800. 58. Roberts, C. W., M. M. Leroux, M. D. Fleming, and S. H. Orkin. 2002. Highly penetrant, rapid tumorigenesis through conditional inversion of the tumor suppressor gene Snf5. Cancer Cell 2:415–425. 59. Roberts, C. W., and S. H. Orkin. 2004. The SWI/SNF complex—chromatin and cancer. Nat. Rev. Cancer 4:133–142. 60. Rocha, S., M. D. Garrett, K. J. Campbell, K. Schumm, and N. D. Perkins. 2005. Regulation of NF-kappaB and p53 through activation of ATR and Chk1 by the ARF tumour suppressor. EMBO J. 24:1157–1169. 61. Sauer, B., and N. Henderson. 1989. Cre-stimulated recombination at loxPcontaining DNA sequences placed into the mammalian genome. Nucleic Acids Res. 17:147–161. 62. Sevenet, N., A. Lellouch-Tubiana, D. Schofield, K. Hoang-Xuan, M. Gessler, D. Birnbaum, C. Jeanpierre, A. Jouvet, and O. Delattre. 1999. Spectrum of hSNF5/INI1 somatic mutations in human cancer and genotype-phenotype correlations. Hum. Mol. Genet. 8:2359–2368. 63. Sif, S., A. J. Saurin, A. N. Imbalzano, and R. E. Kingston. 2001. Purification and characterization of mSin3A-containing Brg1 and hBrm chromatin remodeling complexes. Genes Dev. 15:603–618. 64. Siliciano, J. D., C. E. Canman, Y. Taya, K. Sakaguchi, E. Appella, and M. B. Kastan. 1997. DNA damage induces phosphorylation of the amino terminus of p53. Genes Dev. 11:3471–3481. 65. Strobeck, M. W., K. E. Knudsen, A. F. Fribourg, M. F. DeCristofaro, B. E. Weissman, A. N. Imbalzano, and E. S. Knudsen. 2000. BRG-1 is required for RB-mediated cell cycle arrest. Proc. Natl. Acad. Sci. USA 97:7748–7753. 66. Tao, W., and A. J. Levine. 1999. P19(ARF) stabilizes p53 by blocking nucleocytoplasmic shuttling of Mdm2. Proc. Natl. Acad. Sci. USA 96:6937–6941. 67. van Attikum, H., O. Fritsch, B. Hohn, and S. M. Gasser. 2004. Recruitment of the INO80 complex by H2A phosphorylation links ATP-dependent chromatin remodeling with DNA double-strand break repair. Cell 119:777–788.

MOL. CELL. BIOL. 68. Versteege, I., S. Medjkane, D. Rouillard, and O. Delattre. 2002. A key role of the hSNF5/INI1 tumour suppressor in the control of the G1-S transition of the cell cycle. Oncogene 21:6403–6412. 69. Versteege, I., N. Sevenet, J. Lange, M. F. Rousseau-Merck, P. Ambros, R. Handgretinger, A. Aurias, and O. Delattre. 1998. Truncating mutations of hSNF5/INI1 in aggressive paediatric cancer. Nature 394:203–206. 70. Vossio, S., E. Palescandolo, N. Pediconi, F. Moretti, C. Balsano, M. Levrero, and A. Costanzo. 2002. DN-p73 is activated after DNA damage in a p53dependent manner to regulate p53-induced cell cycle arrest. Oncogene 21: 3796–3803. 71. Vries, R. G., V. Bezrookove, L. M. Zuijderduijn, S. K. Kia, A. Houweling, I. Oruetxebarria, A. K. Raap, and C. P. Verrijzer. 2005. Cancer-associated mutations in chromatin remodeler hSNF5 promote chromosomal instability by compromising the mitotic checkpoint. Genes Dev. 19:665–670. 72. Wang, W., Y. Xue, S. Zhou, A. Kuo, B. R. Cairns, and G. R. Crabtree. 1996. Diversity and specialization of mammalian SWI/SNF complexes. Genes Dev. 10:2117–2130. 73. Weber, J. D., L. J. Taylor, M. F. Roussel, C. J. Sherr, and D. Bar-Sagi. 1999. Nucleolar Arf sequesters Mdm2 and activates p53. Nat. Cell Biol. 1:20–26. 74. Wong, A. K., F. Shanahan, Y. Chen, L. Lian, P. Ha, K. Hendricks, S. Ghaffari, D. Iliev, B. Penn, A. M. Woodland, R. Smith, G. Salada, A. Carillo, K. Laity, J. Gupte, B. Swedlund, S. V. Tavtigian, D. H. Teng, and E. Lees. 2000. BRG1, a component of the SWI-SNF complex, is mutated in multiple human tumor cell lines. Cancer Res. 60:6171–6177. 75. Xiang, H., D. W. Hochman, H. Saya, T. Fujiwara, P. A. Schwartzkroin, and R. S. Morrison. 1996. Evidence for p53-mediated modulation of neuronal viability. J. Neurosci. 16:6753–6765. 76. Xue, Y., J. C. Canman, C. S. Lee, Z. Nie, D. Yang, G. T. Moreno, M. K. Young, E. D. Salmon, and W. Wang. 2000. The human SWI/SNF-B chromatin-remodeling complex is related to yeast rsc and localizes at kinetochores of mitotic chromosomes. Proc. Natl. Acad. Sci. USA 97:13015–13020. 77. Yang, A., and F. McKeon. 2000. P63 and P73: P53 mimics, menaces and more. Nat. Rev. Mol. Cell Biol. 1:199–207. 78. Zachos, G., M. D. Rainey, and D. A. Gillespie. 2003. Chk1-deficient tumour cells are viable but exhibit multiple checkpoint and survival defects. EMBO J. 22:713–723. 79. Zaika, A., M. Irwin, C. Sansome, and U. M. Moll. 2001. Oncogenes induce and activate endogenous p73 protein. J. Biol. Chem. 276:11310–11316. 80. Zhang, H. S., M. Gavin, A. Dahiya, A. A. Postigo, D. Ma, R. X. Luo, J. W. Harbour, and D. C. Dean. 2000. Exit from G1 and S phase of the cell cycle is regulated by repressor complexes containing HDAC-Rb-hSWI/SNF and Rb-hSWI/SNF. Cell 101:79–89. 81. Zhang, Z.-K., K. P. Davies, J. Allen, L. Zhu, R. G. Pestell, D. Zagzag, and G. V. Kalpana. 2002. Cell cycle arrest and repression of cyclin D1 transcription by INI1/hSNF5. Mol. Cell. Biol. 22:5975–5988. 82. Zindy, F., C. M. Eischen, D. H. Randle, T. Kamijo, J. L. Cleveland, C. J. Sherr, and M. F. Roussel. 1998. Myc signaling via the ARF tumor suppressor regulates p53-dependent apoptosis and immortalization. Genes Dev. 12: 2424–2433.