Loss of the hSNF 5 Gene Concomitantly Inactivates p 21 CIP / WAF 1 and p 16 INK 4 a Activity Associated with Replicative Senescence in A 204 Rhabdoid Tumor Cells

hSNF5, the smallest member of the SWI/SNF chromatin remodeling complex, is lost in most malignant rhabdoid tumors (MRT). In MRTcell lines, reexpression of hSNF5 induces G1 cell cycle arrest, elevated p16 , and activated replicative senescence markers, such as B-galactosidase (B-Gal) and plasminogen activator inhibitor-1. To compare the replicative senescence caused by hSNF5 in A204 cells to normal cellular senescence, we examined the activation of both p16 and p21. Analogous to normal cellular senescence, both p16 and p21 were up-regulated following hSNF5 restoration. Furthermore, we found that hSNF5 bound the p16 and p21 promoters, suggesting that it directly regulates transcription of these genes. Using p16 RNA interference, we showed its requirement for the replicative senescence caused by hSNF5 but not the growth arrest. Instead, p21 remained activated by hSNF5 in the absence of high p16 expression, apparently causing the growth arrest in A204. Interestingly, we also found that, in the absence of p16, reexpression of hSNF5 also increased protein levels of a second cyclin-dependent kinase (CDK) inhibitor, p18. However, our data show that lack of hSNF5 does not abrogate cellular responsiveness to DNA damage or growth-inhibitory factors. In summary, our studies suggest that hSNF5 loss may influence the regulation of multiple CDK inhibitors involved in replicative senescence. (Cancer Res 2005; 65(22): 10192-8)


Introduction
SWI/SNF complexes are ATP-dependent chromatin remodeling complexes, which regulate gene transcription by causing conformational changes in chromatin structure (1).SWI/SNF complexes regulate expression of up to 6% of yeast genes (2).In humans, an increasing number of SWI/SNF targets have been identified, including genes involved in cell growth, tissue differentiation, embryo development, and diseases, such as cancer (3)(4)(5)(6)(7)(8).Therefore, to understand how the SWI/SNF complex regulates gene expression has become increasingly important.Recently, several groups have reported a promoter-specific functional interaction between SWI/SNF and histone acetylation complexes in regulating gene expression (9)(10)(11).SWI/SNF complexes have 10 to 12 conserved members ranging in size from 250 kDa (BAF250) to 47 kDa (hSNF5, also known as INI1/BAF47/SmarcB1).Several new SWI/SNF members have also been found that form different subsets of SWI/SNF complexes with distinct functions (12,13).Taken together, these results indicate that SWI/SNF complexes cooperate with gene regulation machinery to control gene transcription in a complex and precise manner.
Clinical studies have shown that loss of the hSNF5 gene occurs in virtually 100% of rhabdoid tumors (a kind of malignant pediatric tumors), and hSNF5 germ line mutations predispose individuals to rhabdoid tumors (14)(15)(16)(17)(18).Further studies have shown that reexpression of hSNF5 in human rhabdoid tumor cell lines causes a G 1 cell cycle arrest, flattened cell morphology, activation of senescence-associated proteins, down-regulation of a subset of E2F target genes, and up-regulation of p16 INK4a (19)(20)(21)(22).In addition, animal studies show that SNF5-null mice are embryonic lethal at E6.5, whereas the SNF5 heterozygous mice are predisposed to rhabdoid tumor development (23).This accumulated evidence firmly establishes SNF5 as a rhabdoid tumor suppressor gene in both humans and mice.
Hayflick and Moorhead (24) showed that normal human fibroblasts possess a limited life span in culture and eventually undergo senescence.Senescence is a physiologic irreversible process of terminal G 1 cell cycle arrest caused by short telomeres, DNA damage, disrupted chromatin organization, and some oncogenes, such as activated RAS, etc. Senescence of normal cells follows activation of the cyclin-dependent kinase (CDK)-cyclin inhibitors, p21 CIP1/WAF1 and p16 INK4a , either in a consecutive manner or in an independent fashion (25)(26)(27)(28).Multiple reports have shown that many tumor cell lines undergo a terminal growth arrest after reexpression of tumor suppressor genes.This process shares some specific features with senescent normal cells, such as a characteristic change in cell shape, a constant G 1 DNA content, absence of S phase in response to mitogens, retention of metabolic activity, resistance to apoptotic death, etc. (29).However, whether the same mechanisms that drive terminal growth arrest in both normal cells and tumor cells remains an open question.Therefore, the term ''cellular senescence'' refers to the changes that normal human cells show at the end of their life span, whereas ''replicative senescence'' generally refers to abnormal cells in culture undergoing a terminal growth arrest (29).
We and others have shown previously that rhabdoid tumor cells apparently undergo replicative senescence after hSNF5 reexpression, including G 1 cell cycle arrest, a flattened cell morphology, and p16 INK4a induction (19)(20)(21)(22).Other groups also confirmed this observation and showed induction of h-Gal expression (20).However, the mechanism by which hSNF5 reexpression induces replicative senescence remains unclear.In this report, we show that hSNF5 induces replicative senescence by directly regulating p21 CIP1/WAF1 and p16 INK4a transcription in the A204 rhabdoid cell line.We also show that both retinoblastoma (Rb) and p53 pathways are involved in this process.Additionally, loss of hSNF5 does not abrogate cellular responsiveness to g-radiation, transforming growth factor-h (TGF-h), and staurosporine treatments.These results implicate a role for hSNF5 in control of cell proliferation and senescence and a potential mechanism to account for the embryonic lethality of SNF5-null mice and the development of rhabdoid tumors in infant patients.

Materials and Methods
Cell lines, plasmids, transfection, and transforming growth factor-B treatment.A204, MvLu1, normal human fibroblasts, and FaDu cells were maintained in RPMI 1640 supplemented with 10% fetal bovine serum at 37jC in 5% CO 2 .A 1.2-kb Flag-tagged PCR product encoding the larger fulllength splicing variant of hSNF5 plus a NH 2 -terminal Flag sequence was cloned into the BamHI and XhoI sites of pCDNA3 (Invitrogen, Carlsbad, CA) or the BamHI and SalI sites of pBabe-puro ( from Dr. Channing Der, University of North Carolina School of Medicine, Chapel Hill, NC) to generate pCDNA3-Flag-hSNF5 or pBabe-Flag-hSNF5.The expression constructs, pDON-p16 RNA interference (RNAi) and pH2B-green fluorescent protein (GFP), were kind gifts of Dr. Scott Lowe (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY) and Dr. G. Wahl (Salk Institute, San Diego, CA).The pCMV-p16 and pCDNA3-HA-p21 were gifts of Dr. Yue Xiong (University of North Carolina, Chapel Hill, NC).pBabe-p21 was generated by cloning a full-length p21 CIP1/WAF1 fragment into the EcoRI/SalI sites of pBabe-puro ( from Dr. Channing Der).Plasmids were transfected using LipofectAMINE 2000 (Invitrogen) according to the manufacturer's instructions.A204, MvLu1, and FaDu cells were treated with 4 ng/mL TGF-h1 (R&D Systems, Minneapolis, MN) for 5 days.
Real-time reverse transcription-PCR.Total mRNA was extracted by RNeasy mini kit (Qiagen, Valencia, CA) and analyzed by the TaqMan (Applied Biosystems, Foster City, CA) quantitative real-time reverse transcription-PCR (RT-PCR) using b-actin as reference gene in each reaction.The change of cycle number (DCt) was obtained by subtraction of the cycle number (Ct) at which the fluorescence exceeded the threshold of detection for h-actin from that for the target genes.Fold change was calculated with 2 DCt .The primers used for p16 INK4a real-time RT-PCR were 5V -CTGCCCAACGCACCGAATA-3V and 5V -GCGCTGCCCATCATCATGA-3V .
Flow cytometry and bromodeoxyuridine staining.Forty-eight hours after transfection with the appropriate expression vectors, 5 Â 10 6 cells were trypsinized, washed in PBS, fixed in 70% ethanol overnight, and then stained with propidium iodide solution [20 Ag/mL propidium iodide (Molecular Probes, Eugene, OR), 200 Ag/mL RNase A (Sigma), and 0.1% Triton X-100 (Sigma) in PBS] for at least 0.5 hour.More than 5,000 gated GFP-positive cells were analyzed for DNA content on a FACScan flow cytometer (Becton Dickinson, San Diego, CA).Cell cycle distribution was determined by using ModFit LT software (Verity Software House, Topsham, ME).For bromodeoxyuridine (BrdUrd) staining, cells were treated with 1:1,000 BrdUrd (Amersham Biosciences) 36 hours after transfection and stained with 1:50 anti-BrdUrd antibody (Accurate Chem.& Sci.Corp.) 12 hours later.BrdUrd-positive cells were counted in at least 100 GFP-positive cells.Error bars were calculated from at least three independent experiments.
Apoptosis assay.Cells (3 Â 10 6 ) were seeded on 100-mm plates and grown overnight.Staurosporine, diluted in DMSO, was added to the medium at a final concentration of 100 nmol/L.Fresh medium with 100 nmol/L staurosporine was placed onto the cells every other day.Cells were harvested for Western blotting at 0, 1, 2, 4, 5, and 6 days after staurosporine treatment.

Results
hSNF5 directly regulates p16 INK4a and p21 CIP1/WAF1 transcription.Previous studies have shown that both p21 CIP1/WAF1 and p16 INK4a protein levels increase when normal human fibroblasts undergo cellular senescence (25,26).In our earlier report, we showed that hSNF5 could induce growth arrest, flattened cell morphology, down-regulation of cyclin A, and up-regulation of p16 INK4a , which all indicated the rhabdoid cells might undergo replicative senescence (19).However, we did not investigate the potential role of p21 CIP1/WAF1 in this process.Therefore, we checked p16 INK4a and p21 CIP1/WAF1 expression by Western blotting 3 days after transfection of hSNF5 into the A204 cells (Fig. 1A).Similar to normal human fibroblasts, both p16 INK4a and p21 CIP1/WAF1 expression increased (Fig. 1A).Furthermore, both cyclin A and hyperphosphorylated Rb expression decreased consistent with a G 1 -S growth arrest.To show that the increase of p16 INK4a and p21 CIP1/WAF1 expression resulted from gene transcription, we used real-time RT-PCR to quantify p16 INK4a and p21 CIP1/WAF1 mRNA levels.From days 3 to 4, p16 INK4a mRNA levels increased from 4-to 7-fold versus the vector control, whereas the p21 CIP1/WAF1 mRNA levels stayed at 2-fold compared with the vector control (Fig. 1B).Chromatin immunoprecipitation data further confirmed that hSNF5 regulated p16 INK4a and p21 CIP1/WAF1 transcription by binding to the appropriate promoter regions of each gene and not to more remote promoter regions (Fig. 1C).These results establish that hSNF5 directly regulates p16 INK4a and p21 CIP1/WAF1 transcription.
Overexpression of p16 INK4a alone is sufficient to cause growth arrest and senescence, whereas p21 CIP1/WAF1 can only induce growth arrest in A204 cells.Because p16 INK4a and p21 CIP1/WAF1 are up-regulated by hSNF5, we asked whether increased expression of either gene alone was sufficient to cause growth arrest and/or senescence in A204 cells.To answer this question, we expressed p16 INK4a or p21 CIP1/WAF1 at high levels in A204 cells and tested their ability to cause growth arrest and senescence.Because p21 CIP1/WAF1 only increased modestly after hSNF5 overexpression, we expressed it at two different levels using two expression vectors (pCDNA3 and pBabe) and they both caused a significant growth arrest (Supplementary Fig. S1A and S1B).Both BrdUrd and colony formation data showed that overexpression of p16 INK4a , p21 CIP1/WAF1 , or hSNF5 could cause a growth arrest in A204 cells (Fig. 2A and B).However, only p16 INK4a and hSNF5 but not p21 CIP1/WAF1 could induce h-Gal, a senescence marker, in A204 cells (Fig. 2C).Surprisingly, p16 INK4a overexpression did not activate other senescence markers, such as PML and PAI-1, whereas reexpression of hSNF5 did induce their expression (Fig. 2D).In conclusion, the induction of p21 CIP1/WAF1 alone resulted in growth arrest but not replicative senescence.Overexpression of p16 INK4a caused growth arrest but only activated a subset of the senescence markers.Only reexpression of hSNF5 activated both growth arrest and induction of replicative senescence.These results show that hSNF5 plays a critical role in the onset of replicative senescence and implicates its requirement for p16 INK4a induction of cellular senescence.
p16 INK4a is required for replicative senescence caused by hSNF5 but not for the growth arrest.A previous report has suggested that p16 INK4a induction was solely responsible for hSNF5's ability to induce a G 1 arrest (33).However, when we inhibited p16 expression by RNAi, we still observed a G 1 arrest (Fig. 3A).In contrast, it seemed that the ability of hSNF5 to induce Figure 1.hSNF5 directly binds to p16 INK4a and p21 CIP1/WAF1 promoters and activates transcription.A204 cells were transfected with plasmids encoding either pBabe (vector) or pBabe-Flag-hSNF5.Cells were selected in 0.6 Ag/mL puromycin for either 3 or 4 days and characterized as follows: A, 3 days after selection, total cellular protein was isolated, separated by SDS-PAGE, and immunoblotted for the indicated proteins.B, total mRNA was extracted and measured by real-time PCR.For each experiment, the mRNA level relative to vector control was calculated.Columns, mean of three individual experiments; bars, SD.C, hSNF5 binds to both p16 INK4a and p21 CIP1/WAF1 promoters, whereas p53 only binds to p21 CIP1/WAF1 promoter when hSNF5 is present.Chromatin immunoprecipitation assays were done on p16 INK4a and p21 CIP1/WAF1 promoters after 4 days of puromycin selection as described in Materials and Methods.p16 INK4a -0.6k and p16 INK4a -2.2k primers were f600 and 2,200 bp upstream of p16 INK4a transcription start site.Similarly, p21 CIP1/WAF1 -2.3k and p21 CIP1/WAF1 -3.5k primers were near 2,300 and 3,500 bp upstream of p21 CIP1/WAF1 transcription start site.IgG was the negative control.AcH3 was used as a positive control.Flag antibody was used to pull down Flag-hSNF5.
replicative senescence, as measured by PML activation, was substantially compromised (Fig. 3A).Because this experiment as well as those of Oruetxebarria et al. relied on transient transfections of p16 INK4a RNAi, we generated two A204 clones with stable expression of p16 INK4a RNAi to better gauge its role in hSNF5 effects (ref.34; Fig. 3B and C).Again, the ability of hSNF5 to induce replicative senescence, as measured by h-Gal induction, was substantially compromised, whereas the G 1 growth arrest remained intact in both p16 INK4a knockdown clones (Fig. 3D and E).These data suggest that the induction of p16 INK4a expression was required for replicative senescence, but not the immediate G 1 arrest caused by hSNF5, consistent with the experiments above.
p21 CIP1/WAF1 remains up-regulated by hSNF5 in the absence of p16 INK4a induction.To determine the basis of the growth arrest caused by hSNF5 in the absence of p16 INK4a induction, we examined the expression of other CDK inhibitor (CDKI) genes after cotransfection of p16 INK4a RNAi and hSNF5 into A204 cells.We observed that p21 CIP1/WAF1 protein levels remained high in the presence of p16 INK4a RNAi by hSNF5 gene compared with the vector (Fig. 3A).These results suggest that p21 CIP1/WAF1 compensation might play the primary role in the sustained growth arrest in the absence of p16 INK4a induction.However, we also investigated whether other CDKI family members might also contribute.Therefore, we screened for expression of p15 INK4b , p18 INK4c , p19 INK4d , p27 KIP1 , and p57 KIP2 proteins after reexpression of hSNF5 in the vector controls and p16 INK4a knockdown clones by Western blotting.We found that p18 INK4c was also activated by hSNF5 when p16 INK4a was down-regulated by transient transfection of p16 INK4a  RNAi and in stable p16 INK4a knockdown clones (Fig. 3A; data not shown).In contrast, none of the other CDKI proteins were altered (data not shown).These results implicate p18 INK4c as another target for hSNF5 regulation and provide an additional mechanism for the ability of hSNF5 to regulate the cell cycle.
hSNF5 loss does not inactivate growth arrest or apoptosis caused by other external signals.To determine whether loss of hSNF5 caused a general disruption of cell cycle growth arrest, we examined cell proliferation and protein expression after g-radiation, exposure to TGF-h, and staurosporine treatments.Surprisingly, the A204 cell line responded in a similar fashion as normal human fibroblasts (35,36).After g-radiation, p53 was activated, p21 CIP1/WAF1 expression increased, and a G 1 cell cycle arrest occurred (Fig. 4A and B).Staurosporine treatment induced an immediate growth arrest and a delayed apoptosis in A204 cells (Fig. 4C and D).Furthermore, A204 cells underwent growth arrest after exposure to exogenous TGF-h (Fig. 4E).Therefore, loss of hSNF5 does not abrogate other growth arrest or apoptotic signals caused by DNA damage and growth inhibitors implicating a block in replicative senescence as the primary effect.

Discussion
Rhabdoid tumor is a rare, highly malignant, and invasive pediatric renal or extrarenal neoplasm.Recent studies have shown that loss of hSNF5 expression plays a critical role in rhabdoid tumor development (18).However, it is still unknown how hSNF5 contributes to A204 cells were transfected with plasmids encoding pCDNA3, pCDNA3-Flag-hSNF5, pCMV-p16, pCDNA3-HA-p21, pBabe, or pBabe-hSNF5.A, A204 cells were cotransfected with the above plasmids along with pH2B-GFP at 5:1 ratio.Forty-eight hours after transfection, the percentage of S-phase cells was determined by BrdUrd (BrdU ) staining as described in Materials and Methods.For each experiment, the S-phase fraction relative to vector control was calculated.Columns, mean of three individual experiments; bars, SD.B, 24 hours after transfection, cells were selected in 0.6 mg/mL neomycin.The plates were fixed and stained with Coomassie blue after 12-day selection.The data were edited by NIH Image 1.62.For each experiment, the colony number relative to vector control was calculated.Columns, mean of three individual experiments; bars, SD.C, 24 hours after transfection, cells were fed with 0.6 mg/mL neomycin and selected for 6 days before the h-Gal assay.Columns, mean of three individual experiments; bars, SD.D, cells were selected in 0.6 mg/mL neomycin or 0.6 Ag/mL puromycin 24 hours after transfection for 6 or 4 days.Protein expression was examined by immunoblotting for the indicated proteins.NS, nonspecific bands recognized by the polyclonal hSNF5 antisera and used as a control for equal loading.
this rare tumor's growth.Abrogation of Rb/p16 INK4a or p53 tumorsuppressive pathways is a common mechanism for tumor cells to escape cell death and maintain proliferation.In rhabdoid tumors, both Rb and p53 were found to be wild-type (37).Therefore, whether abrogation of these pathways contributes to the etiology of rhabdoids and how hSNF5 loss might influence their activities remain important questions.Our data indicate that loss of hSNF5 may affect both pathways to promote rhabdoid tumor growth.
hSNF5 and retinoblastoma pathway.Our previous article concluded that activation of p16 INK4a expression might represent one mechanism by which hSNF5-mediated cell cycle arrest occurs following reexpression in hSNF5-deficient rhabdoid tumor cell lines (19).Later, another group reported that hSNF5 could bind to p16 INK4a promoter and enhance its mRNA expression in the G401 malignant rhabdoid tumor (MRT) cell line (33).In this article, we showed that hSNF5 regulated p16 INK4a transcription and directly bound to its promoter in another rhabdoid tumor cell line, A204, further establishing p16 INK4a as a direct downstream target of hSNF5.Loss of hSNF5 might disable the up-regulation of p16 INK4a during critical junctures in embryonic development and lead to abnormal cell proliferation.This vital role for hSNF5 could explain the embryonic lethality of hSNF5À/À mice and a predisposition to rhabdoid tumors in hSNF5+/À mice.Therefore, loss of hSNF5 may be one of the multiple mechanisms, such as overexpression of cyclin D/CDK4, loss of p16 INK4a , or loss of Rb, to suppress the p16 INK4a -Rb-E2F pathway during tumorigenesis.
hSNF5 and p53 pathway.Several other groups found that p21 CIP1/WAF1 was regulated by BRG1 and was required for BRG1-induced growth arrest (38,39).We showed that p21 CIP1/WAF1 was also regulated by hSNF5 and might be responsible for the growth arrest caused by hSNF5 in the absence of elevated p16 INK4a .Our chromatin immunoprecipitations and quantitative real-time RT-PCR data showed that hSNF5 bound to its promoter and activated p21 CIP1/WAF1 transcription.This regulation might still require p53's function based on our observation that p53 was recruited to p21 CIP1/WAF1 promoter by hSNF5, consistent with a previous study (8).Furthermore, our study found that hSNF5 was not required for p53's ability to cause growth arrest or apoptosis in rhabdoid tumor cell lines.This result apparently conflicts with a previous finding that hSNF5 was required for p53 induction of p21 CIP1/WAF1 in osteosarcoma cell lines (8).However, these differences could reflect the requirement of dissimilar factors in different cell types for p53 activity or a different spectrum of mutations in other genes in these cell lines.In addition, we also cannot rule out the possibility that the increase in p21 CIP1/WAF1 expression observed after DNA damage occurs independently of p53 expression.
Our results also differ from a previous report that showed a requirement of p16 INK4a in hSNF5 caused growth arrest that could not be bypassed through an alternate pathway (33).This apparent difference might arise from the use of different rhabdoid tumor cell lines with different origins and/or gene expression patterns.For example, the G401 cell line used by the other group does not express the tumor progression marker CD44, whereas the A204 cell line does (33).Alternatively, it could reflect the level of p16 INK4a reduction in each cell line or the level of expression of the hSNF5 transgene.Because we used stable p16 INK4a knockdown cell lines Twenty-four hours after transfection, cells were selected in 0.6 Ag/mL puromycin and 0.6 mg/mL neomycin (for vector clones and p16KD#1) or 0.4 mg/mL neomycin (for p16KD#2).After 6 days of selection, cells were examined for induction of replicative senescence by the h-Gal assay.Columns, mean of three individual experiments; bars, SD.E, A204, A204 vector clones, and A204 p16 INK4a knockdown clones were transfected with plasmids encoding either pCDNA3 or pCDNA3-Flag-hSNF5 along with a GFP expression plasmid.Forty-eight hours after transfection, cells were analyzed by flow cytometry.For each experiment, the S-phase fraction relative to vector control was calculated.Columns, mean of three individual experiments; bars, SD.
for our studies, they might provide a more accurate assessment of physiologic functions than transient transfection experiments.We are currently testing these possibilities by doing parallel studies in other rhabdoid cell lines, including G401.
Recently, hSNF5 was found to play a general role in the mitotic checkpoint through the p16 INK4a -cyclinD/CDK4-pRb-E2F pathway (22).The hSNF5 mutants, which lose their ability to control chromosome segregation, can still cause a growth arrest (22).It will be interesting to investigate the role of p21 CIP1/WAF1 in the mitotic checkpoint control and the growth arrest caused by those hSNF5 mutants.
Role of hSNF5 in cellular senescence.Reexpression of hSNF5 in rhabdoid tumor cell lines consistently induced replicative senescence in our studies.How closely does this process resemble cellular senescence that occurs in normal human fibroblasts?In A204 cells, we observed an increase of p21 CIP1/WAF1 and p16 INK4a after overexpression of hSNF5, similar to what happens in normal human fibroblasts during cellular senescence (26).Furthermore, some reports have suggested that p16 INK4a was required for the cellular senescence in normal human fibroblasts (26).Similarly, we found that loss of p16 INK4a abrogated the function of hSNF5 to cause replicative senescence in the MRT cell line.Interestingly, p16 INK4a was sufficient to induce h-Gal but not other senescence markers, such as PML and PAI-1, in the absence of hSNF5 in the MRT cell line.This indicates that p16 INK4a requires hSNF5 to induce the full spectrum of replicative senescence markers in the MRT cell line.The role of p21 CIP1/WAF1 in normal cellular senescence remains less defined.Our studies suggest that p21 CIP1/WAF1 induction could only cause a G 1 cell cycle arrest in the absence of hSNF5 without apparent senescence.Therefore, p21 CIP1/WAF1 induction may not be sufficient for initiation of a senescence program but might still be required for it.A recent report by Herbig et al. implicates p16 INK4a and p21 CIP1/WAF1 acting in independent pathways to cellular senescence (27).Our results are also consistent with this model.Interestingly, their data also show a strong correlation between p16 INK4a induction and h-Gal expression.
hSNF5 and cyclin-dependent kinase inhibitor genes.We found that loss of p16 INK4a induction seems to result in compensatory induction of other CDKI family members.Whereas most experiments showed a decrease in the levels of p21 CIP1/WAF1 protein 3 to 6 days after hSNF5 expression, we consistently found higher levels after hSNF5 expression in the absence of p16 INK4a induction.Thus, our data imply that cross-talk may occur among CDKI proteins, such as p16 INK4a , p18 INK4c , and p21 CIP1/WAF1 , which involve hSNF5.Further investigation is required to define the role of CDKI proteins in hSNF5-mediated cellular senescence.In addition, previous studies suggested that some of the CDKI family members, such as p18 INK4c and p19 INK4d , might play roles in embryonic development (40,41).It will be interesting to study the For each experiment, the S-phase fraction relative to untreated control was calculated.Columns, mean of three individual experiments; bars, SD.D, A204 cells were treated with 100 nmol/L staurosporine and harvested at indicated time points.Total proteins were isolated, separated, and immunoblotted for the indicated proteins.E, TGF-h treatment induces growth arrest in the absence of hSNF5.Cells (2.5 Â 10 4 ) were seeded on glass chamber slides and grown overnight.Twenty-four hours after plating, the cells were treated with TGF-h 1 (4 ng/mL) for 5 days.The percentage of S-phase cells was determined by BrdUrd staining as described in Materials and Methods.For each experiment, the S-phase fraction relative to untreated control was calculated.Columns, mean of three individual experiments; bars, SD.
function of hSNF5 in this process.More importantly, because hSNF5 is a core member of SWI/SNF complex, the relevance of SWI/SNF complex to hSNF5's regulation of CDKI proteins needs to be further explored.
How do these studies provide insight into the mechanism of rhabdoid tumor development?Considering the unusual nature of these tumors (i.e., low incidence, high aggressiveness, undifferentiated phenotype, normal karyotypes, or no or infrequent mutations in Rb and p53 pathway members), one might envision a unique mechanism.If hSNF5 expression is required for procession of normal development at several distinct points, then its loss may only initiate tumor development at those times.Therefore, loss of hSNF5 expression at other junctures in development or in postembryonic tissues may prove irrelevant to normal cellular functions.Further studies into the role of hSNF5 and the SWI/SNF complex in normal embryonic development should address these issues.The availability of several mouse models with conditional loss of hSNF5 will accelerate this process (42).

Figure 2 .
Figure 2. Overexpression of p16INK4a alone is sufficient to cause growth arrest and senescence in A204 cells, whereas p21 CIP1/WAF1 can only introduce growth arrest.A204 cells were transfected with plasmids encoding pCDNA3, pCDNA3-Flag-hSNF5, pCMV-p16, pCDNA3-HA-p21, pBabe, or pBabe-hSNF5.A, A204 cells were cotransfected with the above plasmids along with pH2B-GFP at 5:1 ratio.Forty-eight hours after transfection, the percentage of S-phase cells was determined by BrdUrd (BrdU ) staining as described in Materials and Methods.For each experiment, the S-phase fraction relative to vector control was calculated.Columns, mean of three individual experiments; bars, SD.B, 24 hours after transfection, cells were selected in 0.6 mg/mL neomycin.The plates were fixed and stained with Coomassie blue after 12-day selection.The data were edited by NIH Image 1.62.For each experiment, the colony number relative to vector control was calculated.Columns, mean of three individual experiments; bars, SD.C, 24 hours after transfection, cells were fed with 0.6 mg/mL neomycin and selected for 6 days before the h-Gal assay.Columns, mean of three individual experiments; bars, SD.D, cells were selected in 0.6 mg/mL neomycin or 0.6 Ag/mL puromycin 24 hours after transfection for 6 or 4 days.Protein expression was examined by immunoblotting for the indicated proteins.NS, nonspecific bands recognized by the polyclonal hSNF5 antisera and used as a control for equal loading.

Figure 3 .
Figure 3. Loss of p16 INK4a high expression does not influence the growth arrest but reduces the senescence caused by hSNF5.A, p21 CIP1/WAF1 and p18INK4c are up-regulated by hSNF5 transiently in the absence of p16 INK4a .A204 cells were cotransfected with plasmids encoding either pDON (vector) or pDON-p16 RNAi and either pCDNA3 (vector) or pCDNA3-hSNF5 at a 5:1 ratio.Twenty-four hours after transfection, cells were selected in 0.6 Ag/mL puromycin.Total proteins were extracted, separated by SDS-PAGE, and immunoblotted for the indicated proteins after 6 days of selection.B, p16 INK4a protein levels were greatly reduced in the two p16 INK4a knockdown clones.Total protein was extracted, separated, and immunoblotted for p16INK4a .C, p16 INK4a mRNA levels were also reduced in the two p16 INK4a knockdown clones.Total mRNA was extracted and quantified by real-time RT-PCR.The mRNA level relative to vector control was shown.Columns, mean of three individual experiments; bars, SD.D, A204, A204 vector clones, and A204 p16 INK4a knockdown clones were transfected with plasmids encoding either pCDNA3 or pCDNA3-Flag-hSNF5.Twenty-four hours after transfection, cells were selected in 0.6 Ag/mL puromycin and 0.6 mg/mL neomycin (for vector clones and p16KD#1) or 0.4 mg/mL neomycin (for p16KD#2).After 6 days of selection, cells were examined for induction of replicative senescence by the h-Gal assay.Columns, mean of three individual experiments; bars, SD.E, A204, A204 vector clones, and A204 p16 INK4a knockdown clones were transfected with plasmids encoding either pCDNA3 or pCDNA3-Flag-hSNF5 along with a GFP expression plasmid.Forty-eight hours after transfection, cells were analyzed by flow cytometry.For each experiment, the S-phase fraction relative to vector control was calculated.Columns, mean of three individual experiments; bars, SD.

Figure 4 .
Figure 4. hSNF5 loss does not influence growth arrest or apoptosis caused by other external signals.A, treatment with g-radiation elevates p53 protein levels and causes a G 1 growth arrest in the absence of hSNF5.A204 cells were either treated with 5-Gy 137 Cs g-rays (Gammacell 40) or untreated as shame control.After 48 hours, cells were fixed, stained with propidium iodide, and analyzed for DNA content by flow cytometry.The S-phase ratio was determined using ModFit LT.Columns, mean of three individual experiments; bars, SD.B, A204 cells were treated or untreated (shame) with 5-Gy 137 Cs g-rays (Gammacell 40) and harvested at indicated time points.Total proteins were isolated, separated, and immunoblotted for the indicated proteins.C, staurosporine treatment causes growth arrest and delayed apoptosis in the absence of hSNF5.Normal human fibroblasts and A204 cells were treated with 100 nmol/L staurosporine and analyzed by flow cytometry 48 hours later.For each experiment, the S-phase fraction relative to untreated control was calculated.Columns, mean of three individual experiments; bars, SD.D, A204 cells were treated with 100 nmol/L staurosporine and harvested at indicated time points.Total proteins were isolated, separated, and immunoblotted for the indicated proteins.E, TGF-h treatment induces growth arrest in the absence of hSNF5.Cells (2.5 Â 10 4 ) were seeded on glass chamber slides and grown overnight.Twenty-four hours after plating, the cells were treated with TGF-h 1 (4 ng/mL) for 5 days.The percentage of S-phase cells was determined by BrdUrd staining as described in Materials and Methods.For each experiment, the S-phase fraction relative to untreated control was calculated.Columns, mean of three individual experiments; bars, SD.