|Home | About | Journals | Submit | Contact Us | Français|
The p53 tumor suppressor protein acts as a transcription factor to modulate cellular responses to a wide variety of stresses. In this study we show that p53 is required for the downregulation of FoxM1, an essential transcription factor that regulates many G2/M-specific genes and is overexpressed in a multitude of solid tumors. After DNA damage, p53 facilitates the repression of FoxM1 mRNA, which is accompanied by a decrease in FoxM1 protein levels. In cells with reduced p53 expression, FoxM1 is upregulated after DNA damage. Nutlin, a small-molecule activator of p53, suppresses FoxM1 levels in two cell lines in which DNA damage facilitates only mild repression. Mechanistically, p53-mediated inhibition of FoxM1 is partially p21 and retinoblastoma (Rb) family dependent, although in some cases p21-independent repression of FoxM1 was also observed. The importance of FoxM1 to cell fate was indicated by the observation that G2/M arrest follows FoxM1 ablation. Finally, our results indicate a potential contribution of p53-mediated repression of FoxM1 for maintenance of a stable G2 arrest.
p53, a tumor suppressor protein that is frequently mutated in human primary tumors (Petitjean et al., 2007), directs such processes as apoptosis, cell cycle arrest and senescence through transcriptional regulation of genes that control cell growth and survival (Das et al., 2008; Riley et al., 2008). Although a majority of p53 studies focus on transactivation, multiple gene array studies have shown that p53 also represses the transcription of a substantial number of genes (Wang et al., 2001; Mirza et al., 2003; Robinson et al., 2003; Sax et al., 2003; Kho et al., 2004; Sun, 2006). As binding of p53 to its canonical response element is not always necessary for repression (Gridasova and Henry, 2005; Imbriano et al., 2005), p53 makes use of several mechanisms to impair mRNA expression from specific promoters (reviewed in Ho and Benchimol (2003) and Laptenko and Prives (2006)). Mechanisms that do not involve direct interactions of p53 with the repressed promoter often rely on activation of the cyclin-dependent kinase inhibitor p21/WAF1 (Spitkovsky et al., 1997; Gottifredi et al., 2001; Lohr et al., 2003).
Although several reports have linked p53-mediated repression to p53-induced apoptosis (reviewed in Ho and Benchimol, 2003), many key cell cycle regulators are also repressed in a p53-dependent manner (Spurgers et al., 2006). In particular, p53-mediated repression of c-myc was shown to be necessary for G1 arrest (Ho et al., 2005). Many reports also document the ability of p53 to repress important G2/M regulators such as cyclin B (Krause et al., 2000; Innocente and Lee, 2005), Plk1 (Jackson et al., 2005; Incassati et al., 2006), Cdc2 (Yun et al., 1999; Taylor et al., 2001), Cdc25C (Krause et al., 2001; St Clair et al., 2004; St Clair and Manfredi, 2006) and survivin (Hoffman et al., 2002; Mirza et al., 2002; Raj et al., 2008). Although p53-independent mechanisms can block entry into mitosis after DNA damage, a role for p53 has been established at the G2/M transition (reviewed in Taylor and Stark, 2001; Giono and Manfredi, 2006). Specifically, p53 and p21 are crucial for the maintenance of a stable G2 arrest and for prevention of aberrant mitotic entry after DNA damage (Bunz et al., 1998).
To further analyse the ability of p53 to regulate the cell cycle, and in particular the G2/M transition, we sought to identify targets of p53 that have critical roles in cell cycle progression. Although studies of global gene regulation by p53 have identified an abundance of p53-repression targets, relatively few repression targets have been the subject of careful analysis. Tremendous insight into the full p53 tumor suppressor arsenal can be gained by studying the mechanism and function of p53-mediated repression of important target genes.
Using microarray analysis, two studies independently identified FoxM1 (Trident, MPP-2) as a gene the mRNA of which is significantly downregulated upon p53 expression/activation (Bhonde et al., 2006; Spurgers et al., 2006). FoxM1, a forkhead-family transcriptional activator, is a major regulator of both the G2/M transition and mitotic progression, and is required for proper cell proliferation (Laoukili et al., 2007). Its importance is underscored by the fact that FoxM1−/− mice die in utero at day E18.5 (Krupczak-Hollis et al., 2004). Not only is FoxM1 expression regulated by the cell cycle (it initiates during late G1/S phase, peaks in G2/early mitosis and declines in late mitosis/early G1; reviewed in Laoukili et al. (2007) and Park et al. (2008)), FoxM1 also drives cell cycle progression by transactivating key G2/M regulatory genes, including polo-like kinase-1, survivin, centromere proteins A/B/F, aurora kinase B, S-phase kinase-associated protein-2, Csk1, cell division cycle 25 homolog B (S. pombe), cyclin B and NIMA (never in mitosis gene a)-related kinase 2 (Laoukili et al., 2005; Wang et al., 2005; Fu et al., 2008). FoxM1 null mouse embryonic fibroblasts or human cells treated with small interfering RNA (siRNA) show reduced expression of such genes and increased levels of the cell cycle inhibitors p27 and p21 (Laoukili et al., 2005; Wang et al., 2005). Consequently, FoxM1-ablated cells experience a prolonged G2 phase and delays in mitotic entry (Laoukili et al., 2005; Wonsey and Follettie, 2005).
FoxM1 is commonly upregulated in human carcinomas that originate from different tissues (Pilarsky et al., 2004) and drives tumor development by stimulating proliferation (as reviewed in Laoukili et al., 2007). FoxM1 expression is upregulated, and in many cases required, in several cancer types: basal cell carcinoma (Teh et al., 2002), hepatocellular carcinoma (Kalinichenko et al., 2004), glioblastoma (Liu et al., 2006), primary breast cancer (Wonsey and Follettie, 2005), lung cancer (Kim et al., 2006), prostate cancer (Kalin et al., 2006) and gastric cancer (Zeng et al., 2009). Furthermore, FoxM1 has been implicated in tumor formation (Gusarova et al., 2007), angiogenesis (Wang et al., 2007), metastasis (Chandran et al., 2007) and inhibition of oxidative-stress-induced senescence through down-regulation of the p19(Arf)–p53 pathway (Li et al., 2008).
On the basis of the gene expression array data that FoxM1 levels inversely correlated with p53, our goal was to confirm that FoxM1 is a bonafide p53 repression target and, if so, to elucidate the mechanism and implications of such repression. In this study we show that pro-proliferative FoxM1 is indeed downregulated by the tumor suppressor p53 in multiple cell types after p53 stabilization or activation in a manner that involves the p21/Rb/E2F axis. The data provided in this paper also suggest that FoxM1 repression by p53 may contribute to and be critical for the long-term maintenance of the G2 arrest.
To analyse the role of p53 in the regulation of FoxM1 mRNA levels, two cell lines that express tetracycline-regulated wild-type p53 were used. The H24 (Chen et al., 1996; Baptiste et al., 2002) and MCF7-24 cells (Zhu et al., 2000), which are derivatives of H1299 and MCF7 cells, respectively, were grown in the presence or absence of tetracycline. The p53 induction after tetracycline withdrawal correlated with the downregulation of FoxM1 mRNA levels (Figure 1a). This suggests that p53 expression, in the absence of any stimuli, facilitates FoxM1 transcriptional repression. p53 levels were also stabilized in MCF7 cells treated with daunorubicin, a DNA-damaging agent (Figure 1b), and in this study FoxM1 mRNA levels were repressed to a much greater extent (that is, by a factor of 5) after daunorubicin treatment, with an accompanying decrease in FoxM1 protein levels. MCF7 cells that constitutively express short hairpin RNA to p53 did not repress FoxM1 at either the mRNA or the protein levels, and in fact showed upregulation of FoxM1, confirming a p53-dependent mechanism. Furthermore, it is unlikely that the observed downregulation of FoxM1 is a secondary consequence of cell cycle arrest. Daunorubicin treatment caused arrest of both MCF7 wild-type and MCF7 shp53 cells in the G2/M phase of the cell cycle (Figure 1c). As FoxM1 levels are known to peak in these cell cycle phases, the downregulation of FoxM1 levels, specifically observed in the wild-type p53-expressing cells, is not due to p53-mediated cell cycle arrest at a point in the cell cycle when FoxM1 levels are inherently low.
As p21 is an important mediator of p53-dependent repression, we compared FoxM1 protein and mRNA levels in p53-null H1299 derivative cell lines engineered to express inducible (‘tet-off’) wild-type p53, p53Q22/S23 and p21 (Figure 2a). As before, FoxM1 was significantly repressed upon induction of wild-type p53. The two substituted amino acids in p53Q22/S23 are located within the first transactivation domain of p53 and render it largely impaired in its ability to induce p21 as well as many other p53 target genes (Baptiste et al., 2002). The representative immunoblot shows that p53Q22/S23 accumulated to higher levels than wild-type p53, most likely because of its inability to transactivate murine double minute 2 (MDM2), the major negative regulator of p53 (Iwakuma and Lozano, 2003), and, as expected, only very weakly induced p21. Furthermore, p53Q22/S23 was unable to facilitate FoxM1 mRNA downregulation. Conversely, induction of p21 in the absence of p53 was sufficient for repression of FoxM1 mRNA. The magnitude of FoxM1 repression after p21 expression was comparable to, though slightly less than, that observed after expression of wild-type p53. Thus, induction of p21 is both necessary and sufficient for FoxM1 mRNA repression in H1299 cells.
To assess the importance of p21 levels induced in cells with endogenously expressed p53, MCF7 cells were treated with siRNA to p21 in combination with daunorubicin. As previously shown, MCF7 cells repressed FoxM1 mRNA and protein, whereas p53-knockdown cells induced FoxM1 levels in response to daunorubicin treatment (Figure 2b). Furthermore, MCF7 cells treated with siRNA versus p53 behaved identically to their shp53-expressing counterparts, ruling out clonal variation as an explanation for the lack of repression observed in cells with constitutively low p53 levels. Markedly, cells treated with siRNA versus p21 were unable to repress FoxM1 mRNA. In fact, under this condition, an increase in FoxM1 protein levels was observed, similar to what occurred in cells with reduced p53 expression. We conclude that p21 expression is necessary for FoxM1 repression in daunorubicin-treated MCF7 cells.
FoxM1 activity has been shown to be crucial to hepatocyte DNA replication and mitosis (Wang et al., 2002). We therefore extended our studies to examine FoxM1 regulation in HepG2 cells (Figure 2c), a hepatocellular cancer cell line that harbors endogenous wild-type p53. Furthermore, HepG2 cells in which FoxM1 activity has been partially inhibited undergo significant apoptosis (Gusarova et al., 2007). As shown in Figure 2c, accumulation of p53 protein after daunorubicin treatment led to downregulation of FoxM1 mRNA levels in HepG2 cells (right panel). Cells treated with siRNA versus p53 or siRNA versus p21 were unable to repress FoxM1 message and stabilize FoxM1 protein levels as observed in MCF7 cells, confirming both p53 and p21-dependent repression of FoxM1 (right and left panels). Intriguingly, FoxM1 protein levels and mRNA levels were also somewhat increased after knockdown of basal p53 levels (left and middle panels, compare −dauno control siRNA with −dauno sip53), suggesting a role for basal levels of p53 in the maintenance of FoxM1 levels.
p21 exerts regulation of the cell cycle largely through inhibition of cyclin-dependent kinases and maintenance of Rb in a hypophosphorylated state, allowing Rb to complex with and inhibit E2F family members. The FoxM1 promoter contains two putative E2F-binding sites that are in close proximity to its transcription start site (Laoukili et al., 2007). To determine whether the p21-dependent repression of FoxM1 observed in daunorubicin- treated MCF7 cells works through Rb family activation, MCF7 cells were treated with a combination of siRNA to three Rb family members (Rb, p130 and p107) (Figure 3a). As before, MCF7 cells treated with control siRNA repressed FoxM1 mRNA. MCF7 cells treated with siRNA to the Rb family members showed partial abrogation of FoxM1 downregulation at the mRNA level (approximately twofold less repression was observed; Figure 3a). The complete rescue of repression may be unachievable because of incomplete knockdown of the three Rb family members, although it is possible that p21 acts through distinct pathways to inhibit activation of FoxM1.
To further analyse the involvement of the Rb family members in FoxM1 repression, we asked whether E2F1 can activate FoxM1. Depletion of E2F1 levels using two concentrations of siRNA caused decreasing FoxM1 mRNA and protein levels that were especially apparent at the highest amount of siRNA (Figure 3b). This suggests that E2F1 can exert control over FoxM1 expression, and that the canonical p53/p21/Rb/E2F1 pathway is likely to contribute to FoxM1 repression in daunorubicin-treated MCF7 cells.
We extended our studies of FoxM1 to two other cancer cell lines, U2OS (osteosarcoma) and HCT116 (colon carcinoma) cells, and found that, in contrast to HepG2 and MCF7 cells, FoxM1 mRNA repression after daunorubicin treatment was very mild (Figures 4a and b). Such repression, although modest, was completely reversed in HCT116 cells lacking either p53 or p21 (Figure 4b). In fact, a marked increase in FoxM1 protein levels was observed after daunorubicin in both HCT116 and U2OS cells, suggesting that DNA damage acts as a positive regulator of FoxM1 protein levels in the absence of substantial FoxM1 mRNA repression.
To determine whether these two cell lines were unable to effectively repress FoxM1 in any context, we chose to activate p53 by treating cells with nutlin-3, a small molecule that binds to and disrupts the interaction between MDM2 and p53 (Vassilev et al., 2004). Nutlin inhibits both MDM2-dependent p53 ubiquitination and MDM2-dependent inhibition of p53-dependent transcription.
Stabilization of p53 by nutlin-3 led to a dramatic repression of FoxM1 at both the mRNA and protein levels in U2OS and HCT116 cells (Figures 4a and b), whereas HCT116 p53−/− cells were unable to repress FoxM1 at the mRNA or protein level. On the other hand, nutlin-treated HCT116 p21−/− cells showed mild FoxM1 mRNA repression. Yet, considering the strong repression observed in wild-type cells, nutlin-mediated repression can be considered to be largely p21 dependent.
As FoxM1 mRNA levels were indeed reduced in HCT116 p21−/− cells after nutlin treatment, this suggests that p53 must be able to work in a p21-independent manner to repress FoxM1. Intriguingly, p21-independent repression of FoxM1 was also observed in the MCF7-24 tetracycline-regulated p53 cells (Figure 4c). These cells were pretreated with p21 siRNA with subsequent induction of p53. Although FoxM1 mRNA repression was not particularly strong after ectopic p53 expression in this cell line, the repression was largely, if not completely, p21-independent, a result in sharp contrast with the p21-dependent repression observed in MCF7 cells after DNA damage.
To deduce the biological consequence of p53-dependent repression of FoxM1, we determined the outcome of FoxM1 ablation in MCF7 cells. The cells were treated with two different siRNAs that reduce FoxM1 levels to different extents, with FoxM1 siRNA no. 1 giving the most efficient knockdown (especially notable at the mRNA levels, data not shown). After treatment with either of the FoxM1 siRNAs, the MCF7 cells arrested with 4N content, assessed using fluorescence-activated cell sorting analysis, indicative of cells arrested at G2/M phase of the cell cycle (Figure 5a). Furthermore, MCF7 cells expressing shp53 also arrested in G2/M phase of the cell cycle, ruling out the possibility that the arrest observed after siFoxM1 was due to p53 activation. This suggests that FoxM1 is necessary for proper passage of MCF7 cells through G2.
The extent of G2/M arrest observed after FoxM1 siRNA no. 1 was similar to that observed after daunorubicin treatment of untransfected MCF7 cells. As such, the contribution of p53-mediated repression of FoxM1 to G2/M arrest was assessed. However, even cells expressing shp53 arrested in G2/M in response to daunorubicin (Figure 5a). In fact, these cells arrested to a greater extent in G2 than did p53-expressing cells because of an inability to retain cells in G1, a function ascribed primarily to the action of p53. The observation that shp53-expressing cells arrest in G2 is not surprising as p53 is not thought to be necessary for initial G2/M arrest (Kastan et al., 1991; Bunz et al., 1998; Taylor and Stark, 2001).
To evaluate the contribution of FoxM1 repression to arrest, we assayed for the ability of MCF7 cells and MCF7 shp53-expressing cells to maintain a stable G2 arrest after acute daunorubicin treatment by modifying a previously described approach (Bunz et al., 1998). Both cell lines were treated with daunorubicin for 2 h and allowed to recover for 72 h after removal of the drug (Figure 5b). In a subsequent experiment, cells were treated with daunorubicin as above and also trapped in mitosis by the addition of taxol (Figure 5c), a mitotic poison that prevents the depolymerization of microtublues. In the latter case, mitotic indexes were calculated based on the percentage of cells expressing histone H3 that has been phosphorylated at Serine 10, a mark closely associated with the condensation of chromatin observed during mitosis (Hendzel et al., 1997).
Both MCF7 wild-type and shp53-expressing cells were strongly arrested to similar levels in G2/M, 72 h after acute daunorubicin treatment (in the absence of taxol, Figure 5b). Despite similar extents of G2 arrest, shp53-expressing cells entered mitosis at a significantly higher rate than cells that retained wild-type p53 (a twofold greater mitotic index, Figure 5c). These data show that p53 has a role in the long-term maintenance of G2 arrest after DNA damage.
To determine the possible contribution of p53-mediated repression of FoxM1 in the stability of G2 arrest, shp53-expressing cells (which fail to repress FoxM1 after daunorubicin treatment) were treated with FoxM1 siRNA no. 1 before daunorubicin treatment. Strikingly, ablation of FoxM1 completely restored the mitotic index of shp53-expressing cells to the same level as that observed for daunorubicin-treated wild-type p53 cells (Figure 5c, compare bars indicated by arrows). Daunorubicin caused FoxM1 repression in MCF7 cells but not in shp53 cells (Figures 1b and and2b),2b), suggesting that p53-mediated repression of FoxM1 can contribute to and is likely to be important for maintenance of a stable G2 arrest.
In this particular assay, the acute daunorubicin treatment yielded greater than twofold FoxM1 mRNA repression (data not shown), but did not give rise to the strong downregulation of FoxM1 protein observed in Figures 1b and and2b,2b, most likely because of the stabilizing effect of DNA damage on FoxM1 protein. Nonetheless, the negative regulation of FoxM1 protein levels can be clearly shown after taxol treatment (Figure 5d). In general, FoxM1 levels increased after taxol treatment (data not shown). Despite the high levels of FoxM1, the activation of p53/p21 after acute daunorubicin treatment resulted in significantly lower levels of FoxM1 (lane 2) compared with both untreated wild-type cells (lane 1) and shp53-expressing cells with and without daunorubicin treatment (lanes 3 and 4).
This study demonstrates that FoxM1, an essential transcription factor that controls the expression of many G2/M target genes, is downregulated by p53. Although ectopic p53 expression results in a reduction of FoxM1 mRNA levels, DNA damage cooperates with p53 (perhaps through modification of p53) to more potently repress FoxM1 mRNA. Nonetheless, DNA damage in the relative absence of p53 (shp53) gives rise to an increase in both FoxM1 mRNA and protein over basal levels, supporting previous work that identifies DNA damage as a positive regulator of FoxM1 protein stability (Tan et al., 2007). Thus, DNA damage seems to cause the activation of multiple signaling pathways that converge to fine-tune FoxM1 levels.
As has been found for multiple targets of p53-mediated repression (Kannan et al., 2001; Lohr et al., 2003; Shats et al., 2004), our data in MCF7, HepG2 and H1299 cells establish FoxM1 as an indirect repression target in which downregulation depends on p21. The Rb family (Rb, p130, p107) operates downstream of p21 and has been implicated in p53-mediated repression (Gottifredi et al., 2001; Taylor et al., 2001; Shats et al., 2004; Jackson et al., 2005). Similar to some FoxM1 target genes including Plk1 (Jackson et al., 2005), our results indicate that the Rb family has a role in FoxM1 repression. Although it is unknown whether E2F1 directly activates FoxM1, two putative E2F1 sites have previously been identified in the FoxM1 promoter (Laoukili et al., 2007). The finding that E2F1 contributes to FoxM1 expression further implicates Rb’s involvement in FoxM1 repression.
An intriguing possibility is that p21 may also function to inhibit other transcription factors that are responsible for FoxM1 activation. The FoxM1 promoter contains multiple putative transcription factor binding sites, and has been shown to be downstream of both Gli1 (Teh et al., 2002) and c-Myc (Fernandez et al., 2003; Blanco-Bose et al., 2008) transcription factors. Interestingly, p21 is known to inhibit c-Myc-dependent transcription through direct interaction that disrupts the c-Myc–Max complex (Kitaura et al., 2000). In addition to putative E2F1-binding sites, the FoxM1 promoter contains a B-Myb binding site and the cis-regulatory module CHR-NF-Y (Linhart et al., 2005). p21 could again participate here by impairing NF-Y function through inhibition of cyclin-dependent kinase 2 (Yun et al., 1999, 2003). Furthermore, the presence of CDE/CHR elements in a gene’s promoter often correlates with an indirect repression by p53 (Badie et al., 2000; St Clair et al., 2004). In addition, our data reveal that a portion of FoxM1 repression can be considered p21 independent in both nutlin-treated HCT116 cells and after ectopic p53 expression in MCF7-24 cells. Although the mechanism is unexplored, p53-dependent inhibition of the factors listed above could effectuate this repression. In fact, a well-characterized interaction between p53 and NF-Y is known to directly inhibit NF-Y-dependent transcription (Imbriano et al., 2005). Alternatively, the induction of microRNAs by p53 (reviewed in He et al. (2007) and Vousden and Prives (2009)) could lead to the observed p21-independent FoxM1 repression.
We observe that the basal levels of p53 contribute to FoxM1 regulation in HepG2 cells. This reflects the findings of a study in which p53 represses expression of the cell-surface molecule CD44 under basal conditions (Godar et al., 2008), allowing cells to respond to apoptotic signals that would otherwise be blocked by CD44. Similarly, survivin (a FoxM1 target gene), has been shown to be regulated by the basal levels of p53 and Rb (Raj et al., 2008). As FoxM1 is implicated in negative regulation of the cell cycle inhibitors, p21 and p27 (Wang et al., 2005, 2007; Chan et al., 2008; Xia et al., 2008; Penzo et al., 2009), repression of FoxM1 might be necessary for full p53-dependent responses to stress/treatments.
Furthermore, as loss of p53 and Rb are frequent occurrences in tumors, their absence may contribute to FoxM1 misregulation and thereby adversely affect the ability to inhibit cellular proliferation. In fact, while this manuscript was being prepared, one report showed that p21 (which we have shown to be an important mediator of p53-dependent FoxM1 repression) is required for proper FoxM1 suppression during dextrose-mediated inhibition of liver regeneration after partial hepatectomy (Weymann et al., 2009). This finding confirms the importance of p21 in FoxM1 regulation and also highlights a biological context in which precise regulation of FoxM1 levels is crucial to inhibition of proliferation. As such, it is likely that the inability to downregulate FoxM1 after loss of the tumor suppressor p53 will contribute greatly to biological processes such as carcinogenesis.
Although DNA damage facilitated FoxM1 repression in MCF7 and HepG2 cells, two other cell lines (U2OS and HCT116) were surprisingly unable to effectively repress FoxM1 mRNA and stabilize FoxM1 protein levels after daunorubicin treatment. This finding is reminiscent of a study in which doxorubicin caused p53/Rb-dependent downregulation of human telomerase reverse transcriptase in MCF7 but not in HCT116 cells (Shats et al., 2004). However, we show that nutlin-3 does cause notable downregulation of FoxM1 mRNA and protein in both HCT116 and U2OS cells. Nutlin-3 activates p53 in the absence of DNA-damage signaling by disrupting the interaction between p53 and MDM2 (its major negative regulator). As disruption of the p53– MDM2 complex is thought to be complete after nutlin treatment, it is possible that residual MDM2–p53 complexes remaining after ectopic p53 expression or after DNA damage curb the ability of p53 to down-regulate FoxM1. Alternatively, these particular cell lines may require high p21 levels observed only after nutlin treatment (Figures 4a and b) to repress FoxM1 (and possibly other indirect targets such as human telomerase reverse transcriptase).
As FoxM1 is often overexpressed in cancers, the biological outcome of p53-mediated repression of FoxM1 was an important goal of this study. In MCF7 cells, siRNA to FoxM1 causes a p53-independent G2 arrest. This underscores the importance of FoxM1 and supports a rich literature that depicts FoxM1 as a pro-proliferative transcription factor. As p53 is a major regulator of the cell cycle, it was of great interest to determine the contribution of FoxM1 repression to decisions about cell fate—namely, how does FoxM1 repression affect the cell cycle?
Although only cells that retain wild-type p53 repress FoxM1, this repression cannot be a requirement for initial G2 arrest, as both MCF7 wild-type and shp53-expressing cells arrest in G2 after DNA damage (Figure 5a). However, p53 is thought to be more important for maintenance of G2 arrest rather than its initiation (Taylor and Stark, 2001), as cells without p53 aberrantly enter mitosis with damaged DNA (an event that could lead to chromosomal aberrations, anueploidy and potentially give rise to pro-tumorigenic cells). The Rb family members have also been shown to be important for a stable G2 arrest and cell cycle exit from G2 (Jackson et al., 2005). On the contrary, FoxM1 has been shown to be important for entry into mitosis in both mouse cells and human osteosarcoma cells (U2OS cells) (Wang et al., 2005).
The finding that FoxM1 siRNA rescues the aberrant mitotic entry ofMCF7 shp53-expressing cells reveals the role of p53-mediated repression of FoxM1. That is, for conditions in which p53-mediated repression is robust, the downregulation of a single target, FoxM1, could facilitate stable G2 arrest. Although it is likely that p53 uses several pathways to maintain a stable G2 arrest, the repression of FoxM1 provides an elegant way for p53 to achieve global cellular changes. Notably, many reported targets of p53-mediated repression (Jackson et al., 2005; Spurgers et al., 2006) overlap with the set of genes induced by FoxM1 (Laoukili et al., 2005, 2007; Wang et al., 2005; Fu et al., 2008). As FoxM1 is a master regulator of factors that facilitate the G2/M transition and regulate mitotic events, p53-mediated repression of FoxM1 may cause indirect negative regulation of such factors, ultimately leading to stable arrest and maintenance of genomic integrity.
The HepG2 cells were maintained in RPMI/10% fetal bovine serum. All other cells were maintained in Dulbecco’s modified Eagle’s medium/10% fetal bovine serum. H24 (Baptiste et al., 2002) and MCF724 cells (Zhu et al., 2000) are derivatives of the H1299 and MCF7 cell lines, respectively, that are engineered to inducibly express either wild-type p53, p53Q22/S23 or p21 under control of tetracycline-regulated promoters (‘tet-off’). MCF7 cells that stably express a short hairpin RNA to p53 (MCF7 shp53-expressing cells) were grown in Dulbecco’s modified Eagle’s medium/10% fetal bovine serum with 2.5 μg/ml puromycin. Both MCF7-derivative cell lines were kindly provided by Dr Xinbin Chen. HCT116 wild-type cells and their derivatives HCT116 p53−/− and p21−/− were generously provided by B Vogelstein. MCF7 (mammary carcinoma), HepG2 (hepatocellular carcinoma), U2OS (osteosarcoma) and HCT116 (colorectal carcinoma) cells were treated with daunorubicin (0.22 μm; Oncogene Research Products, La Jolla, CA, USA) or nutlin-3 (10 μM; Sigma-Aldrich, St Louis, MO, USA) for 24 h unless otherwise noted.
The siRNA duplexes from Qiagen (Valencia, CA, USA) were used individually at 50 nmol unless otherwise noted, and transfected into cells using Dharmafect 1 reagent (Dharmacon, Lafayette, CO, USA) according to the manufacturer’s instructions. Control siRNA (either Qiagen All-Stars or siLuciferase) represented a negative control siRNA. The RNA oligo sequences were as follows: siFoxM1 no. 1 (published as siFoxM1 no. 2 in Wang et al., 2005) GGACCACUUUCCCUACUUUdTdT, siFoxM1 no. 2 (Wonsey and Follettie, 2005) CCUUUCCCU GCACGACAUGdTdT, sip21 CUUCGACUUUGUCACCGA GdTdT, sip53 CUACUUCCUGAAAACAACGdTdT, sip130 (Jackson and Pereira-Smith, 2006) GAGCAGAGCUUAAUC GAAUUU, sip107 (Jackson and Pereira-Smith, 2006) CAAG AGAAGUUGUGGCAUAUU, siE2F1 GUCACGCUAUG AGACCUCAdTdT, siRB AAGATACCAGATCATGTCAGA siLuciferaseGL3 CTTACGCTGAGTACTTCGAdTdT.
Cell extracts were analysed according to standard western blotting procedures using enhanced chemiluminescence (Amersham, Piscataway, NJ, USA) or fluorescence through use of the Odyssey system (LI-COR, Lincoln, NE, USA). The monoclonal antibodies DO1 or 1801 were used to detect p53. The monoclonal antibody XZ131 was used to detect Rb. The following polyclonal antibodies were purchased from Santa Cruz (Santa Cruz, CA, USA): anti-FoxM1 (C-20) sc-502, anti-p21 (C-19) sc-397, anti-p130 (C-20) sc-317, anti-p107 (C-18) sc-318, and anti-E2F1 (C-20) sc-193. Anti-Actin (A2066) was purchased from Sigma (St Louis, MO, USA).
Quantitative reverse transcription–polymerase chain reaction was used to quantify relative changes in mRNA expression. RNA was isolated from cell cultures using the Qiagen RNeasy Mini Kit. Complementary DNA was amplified using Qiagen Quantitect reverse transcription kit (Qiagen, Valencia, CA, USA). PCR was carried out on an Applied Biosystems Prism 7300 using the SYBR green dye (Applied Biosystems, Foster City, CA, USA). The mRNA expression was assayed in triplicate and normalized to RPL32 mRNA expression. The relative levels were calculated using the Comparative-Ct Method (ΔΔCT method). Primers were designed with Primer Express (Applied Biosystems): RPL32-F 5′-TTCCTGGCTCA CAACGTCAAG-3′, RPL32-R 5′-TGTGAGGGATCTCGG CAC-3′, FoxM1-F 5′-TGCCCAGATGTGCGCTATTA-3′, FoxM1-R 5′-TCAATGCCAGTCTCCCTGGTA-3′, E2F1-F 5′-AGATGGTTATGGTGATCAAAGCC-3′, E2F1-R 5′-AT CTGAAAGTTCTCCGAAGAGTCC-3′.
The error bars were derived from the s.d. of multiple experiments. Levels of significance were calculated using the Student’s t-test. Statistical significance (P<0.05) is denoted with an asterisk.
Cells pellets were washed with phosphate buffered saline and fixed/permeabilized with 50% ice-cold ethanol. Pellets were washed and resuspended in 50 μg/ml ribonuclease A and 62.5 μg/ml propidium iodide. Samples were analysed on the Becton Dickinson FACSCalibur (Becton Dickinson, Franklin Lakes, NJ, USA). The percentages of cells in various phases of the cell cycle were quantified using the ModFit LT Version 3.0 program (Verity Software House, Topsham, ME, USA). The error bars were derived from the s.d. of multiple experiments.
To determine the mitotic index, the Phospho-Histone H3 (Ser10) Antibody (Alexa Fluor 488 Conjugate) (no. 9708) from Cell Signaling Technology (Danvers, MA, USA) was used. The manufacturer’s recommended flow cytometry protocol for intracellular staining using conjugated primary antibodies (Cell Signaling) was followed with slight modification (cells were permeabilized in 50% ethanol). The samples were analysed for positive incorporation of the Alexa Fluor 488 conjugated phospho-histone H3 (Ser10) antibody on the FACSCalibur (Becton Dickinson). Mitotic index was calculated by dividing the number of Alexa Fluor 488 positive-staining cells with the total cells counted (as assessed by propidium iodide staining). Error bars were derived from the s.d. of multiple experiments.
Ella Freulich provided expert technical assistance. Members of the Prives laboratory are thanked for their support and encouragement. This work was supported by the National Institutes of Health Grant CA77742.
Conflict of interest
The authors declare no conflict of interest.