Reexpression of hSNF5 in Malignant Rhabdoid Tumor Cell Lines Causes Cell Cycle Arrest through a p21^CIP1/WAF1-Dependent Mechanism

Malignant rhabdoid tumor (MRT) is a rare and extremely aggressive childhood cancer. MRT was initially described as an unfavorable histologic type of pediatric renal tumor, a variant of Wilms' tumor (1). Whereas the most common locations occur in the kidney and central nervous system, MRT also arise in almost any site (2, 3). Despite significant advances in the treatment and outcome of other pediatric tumors, for MRTs diagnosed before the age of 6 months, patient survival at 4 years drops to ∼8.8% (4). Therefore, improved patient outcome requires a better understanding of malignant rhabdoid tumorigenesis and the development of novel therapeutic strategies.

In the past several years, the discovery of deletions and mutations at 22q11.2 involving hSNF5/INI1 has contributed to the clarification of pathogenesis of MRT (5). The finding that genetic alterations in MRTs are usually limited to hSNF5 mutations and deletions implicates the loss of hSNF5 function as the primary cause of these tumors. Now, hSNF5 function is recognized as being lost in almost 100% of MRTs (6, 7). Therefore, the elucidation of hSNF5 function should lead to the identification of the key molecular steps necessary for MRT tumorgenesis.

hSNF5 is one of the core subunits of the SWI/SNF chromatin remodeling complex that also includes an ATPase subunit (either BRG1 or BRM), BAF155, and BAF170. SWI/SNF complexes are ATP-dependent chromatin remodeling complexes that regulate gene transcription by causing conformational changes in chromatin structure, as well as by cooperation with histone acetylation complexes (8). In human cells, studies have shown a role for transcriptional regulation by SWI/SNF complexes in the control of cell growth, tissue differentiation, and embryo development in multiple tissues (9). Furthermore, loss of BRG1 function has been observed in malignant tumors, including lung, pancreatic, breast, and prostate cancer (10–13). Several new SWI/SNF members, such as BAF180, have been found to form different subsets of SWI/SNF complexes with distinct functions (14–16). To understand how the SWI/SNF complex regulates gene expression in a complex and precise manner has become increasingly important.

Recently, several reports have shown that hSNF5 plays key roles in cell cycle control, differentiation, and oncogenic transformation. Reexpression of hSNF5 induces G₁ cell cycle arrest in MRT cell lines, accompanied by upregulation of p16^INK4A and downregulation of cyclin D₁, cyclin A, and phosphorylated retinoblastoma protein (pRb), suggesting a key role for these genes in MRT cell cycle control (17–20). Kia and colleagues reported that reexpression of hSNF5 mediates eviction of polycomb complex proteins, such as BMI-1, from epigenetically silenced promoters of the INK4b-ARF-INK4a locus followed by their activation (21). Furthermore, some reports showed that hSNF5 controls the differentiation of MRT cells (22, 23) and hSNF5 loss changes gene transcription epigenetically and contributes to oncogenesis without genomic instability (24).

Our previous study showed that reexpression of hSNF5 induced cell cycle arrest even in the absence of p16 ^INK4A expression (25). This finding suggested that other genes besides p16^INK4A play a critical role at early time points of G₁ cell cycle arrest induced by hSNF5. Therefore, in this study, we determined the mechanism of G₁ cell cycle arrest induced by hSNF5 in MRT cells within 24 hours after reexpression using adenoviral vectors. We show that induction of p21^WAF1/CIP1 appears at the onset of hSNF5-induced growth arrest and precedes p16^INK4A expression. Furthermore, we show that p21^WAF1/CIP1 knockdown inhibits hSNF5-induced G₁ cell cycle arrest. We also show differences in the histone methylation changes at these two promoters after hSNF5 reexpression. Finally, we demonstrate that p21^WAF1/CIP1 shows both p53-dependent and p53-independent mechanisms of induction after hSNF5 reexpression. Our results suggest that p21^WAF1/CIP1 plays a key role in hSNF5 control of cell growth, and hSNF5 loss may alter p21^WAF1/CIP1 transcription by a different mechanism than that reported for the p16 ^INK4A promoter in MRT cells.

A204.1 (American Type Culture Collection, ATCC), G401.6 (ATCC), TTC642 (Dr. Timothy Triche, Children's Hospital of Los Angeles), and NIH3T3 (Dr. Stuart Aaronson, National Cancer Institute) cells were cultured in RPMI 1640, and UNC N3T cells were cultured in bronchial epithelial growth medium (26). 293FT cells were cultured in DMEM containing 10% fetal bovine serum. The Ad/pAdEasyGFPINI-SV⁺ adenoviral vectors expressing hSNF5 and coexpressing green fluorescent protein (GFP; designated Ad-hSNF5) and the Ad/pAdEasyGFP expressing GFP (designated Ad-GFP) were previously published (20). To achieve infection of over 90% of cells, we infected at a multiplicity of infection (MOI) of 20 for the A204.1 cell line and 200 for the TTC642 cell line.

Western blotting was carried out as described previously (25). Western analyses of proteins were carried out by using anti-p21^CIP1/WAF1 (AB1; Calbiochem), anti-p16^INK4a (G175-1239; BD Pharmingen), anti-pRb (G3-245; BD Pharmingen), anti-actin (A2066; Sigma), anti-p53 (DO-1; Santa Cruz), anti-cyclin A (H-432; Santa Cruz Biotechnology), anti-hSNF5 (BD Transduction Laboratories), BMI-1 (upstate cloneF6; Millipore), and horseradish peroxidase–conjugated antirabbit or antimouse IgG (GE Healthcare).

RNA was extracted using the RNeasy mini kit (Qiagen), and 1 μg was used for cDNA synthesis primed with random primers (Invitrogen). cDNA was analyzed using TaqMan (Applied Biosystems) quantitative real-time reverse transcription–PCR (QT-PCR) analysis, with β-actin as the reference gene in each reaction. Reactions were performed on an ABI 7900 HT sequence detection system (Applied Biosystems), and relative quantification was determined using the 2^−ΔΔC_t method (27). The primers used for p16^INK4a QT-PCR were 5′-CTGCCCAACGCACCGAATA-3′ and 5′-GCGCTGCCCATCATCATGA-3′. The probe used for p16^INK4a QT-PCR was 5′-CTGGATCGGCCTCCGACCGTA-3′. The TaqMan gene expression assay primer/probe set Hs00355782_m1 (Applied Biosystems) was used for p21^CIP1/WAF1, the primer/probe set Hs01034249_m1 (Applied Biosystems) was used for p53, and the primer/probe set Hs99999903_m1 (Applied Biosystems) was used for β-actin.

Chromatin immunoprecipitation (ChIP) was carried out as described by Donner and colleagues (28). Immunoprecipitation was performed with an antibody specific to hSNF (Dr. Tony Imbalzano), histone H3 trimethylation of lysine 4 (H3K4me3; ab8580; Abcam), BRG-1 (J1; Dr. Weidong Wang), BMI-1 (upstate cloneF6; Millipore), normal rabbit IgG (sc-2027; Santa Cruz Biotechnology), normal mouse IgG (sc-2025; Santa Cruz Biotechnology), or p53 (DO-1; Calbiochem). DNA present in each immunoprecipitation was quantified by QT-PCR using gene-specific primers on an ABI 7000 sequence detection system. All expression values were normalized against input DNA. Antibody specificity was also determined for each cell line (Supplementary Fig. S1). The primer sequences are shown in Supplementary Table S1.

Lentivirus was generated using 293FT cells following the protocol of Kafri and colleagues (29). Either pLKO.1, a nontarget small hairpin RNA (shRNA) control vector (SHC002; Sigma), an equal mixture of five types of NM_00039 p21 MISSION shRNA lentiviral transduction particles (TRCN0000040123, TRCN0000040124, TRCN0000040125, TRCN0000040126, and TRCN0000040127), or NM_000546 p53 MISSION shRNA lentiviral transduction particles (TRCN0000003756), obtained from Sigma, were cotransfected with the packing construct ΔNRF (from Dr. Tal Kafri, University of North Carolina; ref. 29) and the VSV-G envelope expression plasmid (pMDK64; from Dr. Matthias Kaeser, Salk Institute) into 293FT cells with FuGene (Roche). pLKO.1 is a negative control containing an insert sequence that does not target any human or mouse gene but will activate the RNAi pathway. For infection, cells were incubated with lentiviral particles and polybrene and then selected with puromycin. At least three puromycin-resistant colonies of A204.1 and TTC642 cells were isolated and expanded for further characterization.

Cell cycle analyses were performed according to the procedure of Huang and colleagues (30). Percentages of cells within each of the cell cycle compartments were determined by flow cytometry (CyAn; Dako) and analyzed with ModFit software (Verity).

We have previously shown that both hSNF5 and p16^INK4A can induce G₁ cell cycle arrest in MRT cell lines at 72 hours after transfection (19). However, hSNF5 also induced G₁ cell cycle arrest without p16^INK4A induction in the same time frame (25). Because of the extended period between hSNF5 transfection and the characterization of cell cycle arrest in these previous studies, we characterized the effects of hSNF5 expression on the growth of two MRT cell lines within 24 of infection with Ad-hSNF5 and Ad-GFP (negative control) adenoviruses.

The induction of hSNF5 protein expression in the A204.1 and TTC642 cells by adenoviral infection is shown in Fig. 1A. No hSNF5 expression was detected in either MRT cell line in the absence of infection or after Ad-GFP infection. However, infection with Ad-hSNF5 led to hSNF5 expression as early as 12 hours postinfection, followed by a time-dependent increase in hSNF5 expression levels in both MRT cells lines.

We next tested the effects of hSNF5 reexpression on cell cycle regulation by flow cytometry. Both cell lines infected with Ad-hSNF5 showed cell cycle arrest 24 hours after infection, characterized by the presence of nearly 80% of cells in the G₁ phase of the cell cycle and the presence of <10% of cells in the S phase (Fig. 1B). The percentage of Ad-hSNF5– and Ad-GFP–infected cells in the S phase was significantly different at 24 hours after infection. Similar results were found at 48 hours postinfection. These results showed that the G₁-S cell cycle progression was inhibited by 24 hours after hSNF5 reexpression in these MRT cell lines.

We previously showed that hSNF5 reexpression induced the downregulation of cyclin A, the dephosphorylation of pRb, and the upregulation of p16^INK4A and p21^CIP1/WAF1 expression at 3 days after transfection (25). To determine whether these changes also occurred simultaneously with hSNF5-induced cell cycle arrest at 24 hours postinfection, we examined the expression of cyclin-dependent kinase (CDK) inhibitors, especially p16^INK4A and p21^CIP1/WAF1, as well as their downstream targets by Western blotting (Fig. 1C). We observed increased p21^CIP1/WAF1 and decreased phosphorylated pRb and cyclin A levels at 24 hours after Ad-hSNF5 infection compared with Ad-GFP control and uninfected control in both A204.1 and TTC642 cells. In A204.1, p16^INK4A protein levels were increased slightly at 12 hours after Ad-hSNF5 infection and showed a further increase at 24 hours after infection compared with control cells. In contrast, p16^INK4A protein expression was absent at baseline in TTC642, with a slight increase at 24 hours, followed by a marked increase at 48 hours after infection. On the other hand, p53 was not significantly changed in both MRT cell lines (Fig. 1C).

We next examined whether the increase in p21^CIP1/WAF1 protein levels resulted from an increase in its mRNA levels by QT-PCR. We found the level of p21^CIP1/WAF1 mRNA increased within 12 hours after Ad-hSNF5 infection in comparison with Ad-GFP infection in both MRT cell lines. In TTC642, p21^CIP1/WAF1 also increased more with Ad-GFP infection than uninfected control, especially at 48 hours (Fig. 2A).

In our previous study, we showed hSNF5-regulated p21^CIP1/WAF1 and p16^INK4A transcription in A204.1 at 4 days after hSNF5 plasmid transfection by ChIPs (25). Therefore, we analyzed the chromatin status at p21^CIP1/WAF1 promoter, −2,283 kb (p53 high-affinity binding site) and −1,391 kb (p53 low-affinity binding site) in both A204.1 and TTC642 cells (28), at 24 hours after Ad-hSNF5 infection to clarify the mechanism of p21^CIP1/WAF1 activation by hSNF5. ChIP data confirmed that hSNF5 bound to both −2,283 kb site and −1,391 kb site in either cell line. Furthermore BRG1 is also recruited by hSNF5 induction to both sites in A204.1, but only at the −2,283 site in TTC642 (Fig. 2B).

Moreover, our previous reports also suggested hSNF5 recruits p53 to the p21 promoter (25). Therefore, we determined whether hSNF5 recruitment to the p21 promoter affected p53 binding. In A204.1, p53 binding is increased after hSNF5 reexpression, with a higher amount detected at the high-affinity −2,283 kb site (Fig. 2B). In contrast, we did not observe a difference in p53 recruitment between Ad-hSNF5 infection and Ad-GFP infection in TTC642, although the difference in binding between the two affinity sites remained (Fig. 2B). We next determined the effect of hSNF5 reexpression on H3K4me3, a chromatin mark associated with gene activation (31). H3K4me3 decreased after hSNF5 reexpression at both −2,283 kb and −1,391 kb sites in both A204.1 and TTC642 cells (Fig. 2B).

Because our results indicated that hSNF5 reexpression activated p21^CIP1/WAF1 transcription with p53 recruitment in A204.1 cells but without p53 recruitment in TTC642 cells, we next assessed the role of p53 in p21^CIP1/WAF1 transcription in the MRT cell lines. We established two independently derived p53 stable knockdown MRT cell lines from both A204.1 and TTC642 cells using lentiviral vectors encoding a shRNA targeting p53 mRNAs. We also developed a negative control cell line using a lentiviral vector encoding a shRNA targeting a nonmammalian sequence (pLKO.1). By QT-PCR and Western blotting, all p53 knockdown cells (A204.1 p21KD and TTC642 p21 KD) showed significant decreases in the p53 mRNA levels along with the protein levels of p53 and p21^CIP1/WAF1 compared with the parental cells or the control cells (A204.1 pLKO.1 and TTC642 pLKO.1; Fig. 3A).

We next determined whether the reduction in p53 expression affected p21^CIP1/WAF1 transcription induced by hSNF5. Infection of the pLKO.1 and p53KD cells with Ad-hSNF5 or Ad-GFP resulted in increased levels of p21^CIP1/WAF1 mRNA at 24 hours after Ad-hSNF5 infection in pLKO.1 cells as in the parental cell lines (Fig. 3B). However, whereas the increase of p21^CIP1/WAF1 mRNA by hSNF5 reexpression was significantly inhibited in all A204.1 p53KD cells, the increase of p21^CIP1/WAF1 mRNA by hSNF5 reexpression was not significantly different among TTC642, TTC642 pLKO.1, and all TTC642 p53KD cells (Fig. 3B). These results suggested that the upregulation of p21^CIP1/WAF1 transcription by hSNF5 reexpression was operated through a p53-dependent mechanism in A204.1 cells and through a p53-independent mechanism in TTC642 cells.

Although A204.1 expresses p16^INK4A mRNA and protein, TTC642 does not show detectable expression of p16^INK4A protein. However, reexpression of hSNF5 caused upregulation of p16^INK4A protein in both MRT cell lines (Fig. 1C). We, therefore, examined whether the increase of p16^INK4A protein resulted from an increase in its mRNA levels by QT-PCR. We found that p16^INK4A mRNA levels increased within 24 and 48 hours after Ad-hSNF5 infection in A204.1 and TTC642 cells, respectively (Fig. 4A). These results showed that the increase in p21^CIP1/WAF1 mRNA occurs earlier than the increase in p16^INK4A mRNA.

Because BMI-1 represses transcription at p16^INK4A locus (32) and an earlier report indicated that p16^INK4A transcription is activated by induction of hSNF5 via BMI-1 eviction (21), we first confirmed that hSNF5 binding at the p16^INK4A promoter increased at 24 hours after Ad-hSNF5 infection in both cell lines (Fig. 4B). We next determined the binding of BMI-1 to the p16^INK4A promoter as an indication of polycomb complex silencing. In TTC642, we observed BMI-1 binding was also significantly less after infection of Ad-hSNF 24 and 48 hours after infection compared with control infected cells (Fig. 4B). We also found a modest increase in BRG-1 binding on the p16^INK4A promoter after hSNF5 reexpression at 24 hours, followed by a dramatic increase in H3K4me3 binding at 48 hours in TTC642 cells (Fig. 3B and C). These results seem consistent with hSNF5 reexpression increasing the binding of the SWI/SNF complex to the p16^INK4A promoter accompanied by polycomb eviction at 24 hours followed by H3K4 methylation and activation of p16^INK4A transcription at 48 hours in TTC642 cells.

In contrast, we detected little BMI-1 binding on the p16^INK4A promoter in A204.1 cells, even in the absence of hSNF5 expression (Fig. 4B). We also observed that hSNF5 reexpression had little effect on the binding of BRG-1 and the level of H3K4 methyation at this promoter (Fig. 4B and C). These results suggested an absence of polycomb complex silencing at the p16^INK4A promoter in the A204.1 cells. Therefore, we examined the expression of BMI-1 in our MRT cell lines by Western blotting. The results in Fig. 4F show that the A204.1 cell line fails to express detectable BMI-1 protein compared with the TTC642 cell line (Fig. 4D). The absence of BMI-1 may explain the basal level of p16^INK4A mRNA observed in A204.1 cells.

Because our results indicated that hSNF5 reexpression activated p21^CIP1/WAF1 transcription earlier than p16^INK4A transcription, we next assessed the role of p21^CIP1/WAF1 in hSNF5-induced cell cycle arrest in the MRT cell lines. We established three independently derived p21^CIP1/WAF1 stable knockdown MRT cell lines from both A204.1 and TTC642 cells using lentiviral vectors encoding a shRNA targeting p21^CIP1/WAF1 mRNAs and a negative control cell line (pLKO.1). By QT-PCR and Western blotting, all p21^CIP1/WAF1 knockdown cells (A204.1 p21KD and TTC642 p21 KD) showed significant decreases in the mRNA levels along with the protein levels of p21^CIP1/WAF1 compared with the parental cells or the control cells (A204.1 pLKO.1 and TTC642 pLKO.1; Fig. 5A).

We next determined whether the reduction in p21^CIP1/WAF1 expression affected hSNF5-induced cell cycle arrest. Therefore, we infected the pLKO.1 and p21KD cells with Ad-hSNF5 or Ad-GFP and assayed the effects on cell cycle by flow cytometry. We observed a similar G₁ cell cycle arrest induced by Ad-hSNF5 infection in pLKO.1 cells after 24 hours as in the parental cell lines (Fig. 5B). However, the inhibition of the G₁-S cell cycle progression by hSNF5 reexpression was significantly inhibited in all A204.1 p21KD cells and TTC642 p21KD cells (Fig. 5B). These results indicated that p21^CIP1/WAF1 upregulation contributes to the inhibition of the G₁-S cell cycle progression by hSNF5 at 24 hours after Ad-hSNF5 infection.

Because p21^CIP1/WAF1 knockdown caused inhibition of G₁ cell cycle arrest induced by hSNF5 reexpression, we examined whether the effect occurred at the level of pRb phosphorylation. Although p21^CIP1/WAF1 expression increased after hSNF5 reexpression in A204.1 pLKO.1 and TTC642 pLKO.1 cell lines, we observed limited increase of p21^CIP1/WAF1 and limited reduction of phosphorylated pRb and cyclin A at 24 hours after Ad-hSNF5 infection in A204 p21KD cells and TTC642 p21KD cells (Fig. 6A and B). The expression of p16^INK4A was not significantly increased at 24 hours after Ad-hSNF5 infection in A204.1 p21KD cells. In TTC642 p21KD cells, the p16^INK4A protein increased slightly at 24 hours after Ad-hSNF5 infection (data not shown) and then increased significantly at 48 hours, similar to the TTC642 parent cell line (data not shown). The change of p21^CIP1/WAF1 mRNA was similar to the results seen with p21^CIP1/WAF1 protein levels (data not shown). Taken together, these results indicated that the G₁ arrest induced by hSNF5 reexpression strongly correlated with dephosphorylation of pRb through p21^CIP1/WAF1 activation.

By studying the mechanism of G₁ cell cycle arrest induced by hSNF5 at a time point concomitant with the induction of growth arrest, our study shows three important observations. First, hSNF5 directly regulates the p21^CIP1/WAF1 and p16^INK4A loci through different mechanisms among MRT cell lines. Second, hSNF5 regulates p21^CIP1/WAF1 through either a p53-dependent or a p53-independent mechanism. Finally, p21^CIP1/WAF1 has a key role in hSNF5-induced cell growth arrest in MRT cell lines.

Reexpression of hSNF5 increased p21^CIP1/WAF1 transcriptional activity immediately through the recruitment of BRG1 to the p21^CIP1/WAF1 promoter. Similarly, some reports have suggested that BRG1 associates with the p21^CIP1/WAF1 promoter and activates it along with hSNF5 (33, 34). Does this recruitment of the SWI/SNF complex lead to an interaction with another transcription factor? Lee and colleagues (35) suggested that the BRG-1 interacts with p53 and activates the p21^CIP1/WAF1 promoter in a p53-dependent manner. In contrast, Liu and colleagues (36) and Hendricks and colleagues (33) suggested that BRG-1 activates the p21^CIP1/WAF1 promoter in p53-independent manner. Indeed, previous reports have shown that multiple mechanisms can activate the p21^CIP1/WAF1 promoter, including p53-dependent, Sp1- or Sp3-dependent, or CDK8-dependent mechanisms (37). Our results in the A204.1 cell line seem consistent with the former model. We observed increased p53 levels at the p21^CIP1/WAF1 promoter after hSNF5 reexpression and reduction in p53 expression inhibited upregulation of p21^CIP1/WAF1 induced by hSNF5. This finding, at a minimum, supports the notion that reexpression of hSNF5 in these cells facilitates the recruitment of p53 to the p21^CIP1/WAF1 promoter. However, our studies did not determine whether this occurs through a direct interaction between p53 and hSNF5. We also need to identify the upstream signal that initiates p53 binding to the p21^CIP1/WAF1 promoter.

In contrast, p53 levels on p21^CIP1/WAF1 did not change in the TTC642 cells after hSNF5 reexpression nor did decreased p53 levels affect the ability of hSNF5 to increase p21^CIP1/WAF1 transcription. Whereas we cannot exclude the possibility that a low level of p53 protein remains at the p21^CIP1/WAF1 promoter, sufficient for transcriptional activation, it seems that hSNF5 reexpression may operate through a different mechanism in these cells. Whereas activation of p21^CIP1/WAF1 transcription by hSNF5 in both cell lines seems associated with recruitment of BRG-1, the transcription factors recruited to the p21^CIP1/WAF1 promoter may differ. Additional ChIP analyses of the p21^CIP1/WAF1 promoter in the TTC642 cell line will clarify this matter. In addition, our result showed a modest increase in p21^CIP1/WAF1 transcription after Ad-GFP infection compared with the uninfected control. We believe that the high MOI of adenovirus required for infection of this cell line caused upregulation of p21^CIP1/WAF1 through apoptotic induction (38).

We also showed that reexpression of hSNF5 increases p16^INK4A transcriptional activity after p21^CIP1/WAF1 upregulation. In the TTC642 cell line, hSNF5 expression caused decrease of BMI-1 and increase of BRG-1 at the p16^INK4A promoter. The level of BMI-1 at 48 hours was less than that at 24 hours, and the decrease of BMI-1 was correlated with an increase of H3K4me3. These results concur with the report by Kia and colleagues showing that hSNF5 induced a decrease of BMI-1 and an increase of H3K4me3 on the p16^INK4A promoter and that BRG1 was necessary for activation of the p16^INK4A promoter by hSNF5 (21). On the other hand, the A204.1 cell line expresses low basal levels of p16^INK4A consistent with a lack of expression of BMI-1. Our results seem in accord with the report that p16^INK4A expression is directly regulated by polycomb proteins, such as BMI-1 and EZH2 (39). Moreover, while hSNF5 appeared at p16^INK4A promoter after reexpression followed by an increase in p16^INK4A mRNA in the A204.1 cell line, BRG-1 and H3K4me3 levels at the promoter did not change significantly. Therefore, in the absence of polycomb silencing, p16^INK4A mRNA could increase more rapidly in the A204.1 cell line than in the TTC642 cell line. Regardless, the data still show that the hSNF5-induced increase in p16^INK4A levels in the A204.1 cell line follows the p21^CIP1/WAF1 upregulation regardless of BMI-1 expression. The mechanism for the increase in p16^INK4A mRNA in the A204.1 cell line requires further investigation.

The difference in H3K4me3 patterns between the p21^CIP1/WAF1 and p16^INK4A promoters was unexpected. The significant increase in H3K4me3 at the p16^INK4A promoter (−450 kb) in the TTC642 cell line after hSNF5 reexpression seems consistent with the activation of transcription from this promoter. Although the levels of this modification did not change in the A204.1 cell line, this may reflect the active p16^INK4A transcription present in these cells before hSNF5 reexpression. However, in both cell lines, H3K4me3 decreased at the p53 binding sites in the p21^CIP1/WAF1 promoter (−2,283 kb and −1,391 kb) after hSNF5 reexpression, although p21^CIP1/WAF1 transcription increased. One possible explanation for this observation might come from hSNF5 reexpression activating SWI/SNF complex activity resulting in nucleosome repositioning by chromatin remodeling. Therefore, the subsequent change in H3K4me3 positioning on the p21^CIP1/WAF1 promoter would be reflected by a decreased signal in our ChIP assay.

In our study, p21^CIP1/WAF1 knockdown experiments showed that inhibition of p21^CIP1/WAF1 expression partially inhibited the efficiency of hSNF5-induced G₁ cell cycle arrest in MRT cell lines. The failure to completely abrogate the growth arrest may result from the residual activation of p21^CIP1/WAF1 protein in the RNAi-expressing MRT cell lines. In addition, the increasing levels of p16^INK4A protein may also begin to affect the cells because we observed a complete cell cycle arrest at 48 hours in the p21 knockdown cells.³

Many MRTs arise under in infants under the age of 6 months and in neonates (43). Therefore, it seems plausible that a significant number of MRTs arise from the loss of hSNF5 in stem cells or progenitor cells during development. The expression of p16^INK4A in stem cells is restricted by the expression of BMI-1 (44). Indeed, mouse embryo cells do not display expression of p16^INK4A (45) and stem cells in young mice (8–12 weeks old) do not express detectable p16^INK4A mRNA (46). Thus, a strong possibility exists that p16^INK4A was already silenced in the cell that gave rise to the TTC642 cell line before hSNF5 loss occurred.

Our results may indicate that a decrease in p21^CIP1/WAF1 expression after hSNF5 loss may signify one key event during MRT development. For example, the absence of p21^CIP1/WAF1 expression can increase hematopoietic stem cell proliferation (47) or maintain neural stem cells by playing a significant role in regulating their proliferation (48). Our recent studies showing cooperation between SNF5 loss and pRb family inactivation in the acceleration of formation of spinal cord MRTs in mice support this notion (49). These results suggest that p21^CIP1/WAF1 and its downstream targets may regulate the boundary between quiescence and proliferation in stem cells and progenitor cells.

In conclusion, our results showed that, whereas hSNF5 reexpression in MRT cells increases both p21^CIP1/WAF1 and p16^INK4A expression during the induction of G₁ cell cycle arrest, p21^CIP1/WAF1 upregulation precedes p16^INK4A. Although our studies firmly substantiate p21^CIP1/WAF1 as a key target for hSNF in cell cycle regulation, the role of hSNF5 within the activities of SWI/SNF complex and gene regulation seem complex. Studies from other laboratories also implicate a role for hSNF5 in the regulation of cellular differentiation, cell migration, and DNA repair (40, 41, 50). However, the establishment that SNF5 loss alters p21^CIP1/WAF1 expression during MRT tumorigenesis provides an important new target for therapy in a tumor with limited options for treatment.

No potential conflicts of interest were disclosed.

We thank Dr. Matthias Kaeser for his excellent technical input and thoughtful discussions, Dr. Cindy Wright (Medical University of South Carolina) for the Ad-SNF5 and Ad-GFP reagents, Dr. Scott Randell (University of North Carolina) for the UNC N3T cell line, and Dr. Tony Imbalzano (University of Massachusetts School of Medicine) for the SNF5 antiserum.

Grant Support: National Cancer Institute Public Health Service grants CA91048 (B.E. Weissman) and R01-CA104213 (E.S. Knudsen) and American Brain Tumor Association Chad Dunbar Postdoctoral Fellowship (A. Charboneau).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.