p53-regulated Transcriptional Program Associated with Genotoxic Stress-induced Apoptosis.

Using a genome-wide approach, we sought the identification of p53-regulated genes involved in cellular apoptosis. To this end, we assessed the transcriptional response of HCT116 colorectal cancer cells during apoptosis induced by anticancer drug 5-Fluorouracil (5-FU) as the function of p53 status and identified 230 potential targets that are regulated by p53. Previously identified p53 targets known to be involved in growth arrest and apoptosis were observed to be induced thus validating the approach. Strikingly, we found that p53 regulates gene expression primarily through transcriptional repression (n=189) rather than activation (n=41) and selective blockade of p53-dependent gene repression resulted in the reduction in 5-FU induced apoptosis. Reporter and chromatin-immunoprecipitation (ChIP) assays demonstrated that p53 can suppress the promoter activities of three further studied candidate genes PLK, PTTG1 and CHEK1 but would only bind directly to PTTG1 and CHEK1 promoters, revealing that p53 can repress gene expression through both direct and indirect mechanisms. Moreover, RNA i-mediated knockdown of PLK and PTTG1 expression were sufficient to induce apoptosis, suggesting that repression of novel anti-apoptotic genes by p53 might contribute to a significant portion of the p53-dependent apoptosis. Our data supports the divergent functions of p53 in regulating gene expression that play both synergistic and pleiotropic roles in p53-associated apoptosis.


5
Apoptosis was measured using FACS analysis of cells in sub-G1 phase. Cells were harvested and fixed in 70% ethanol. The fixed cells were then stained with propidium iodide (50µg/ml) after treatment with RNase (100 µg/ml). The stained cells were analyzed for DNA content by fluorescence-activated cell sorting (FACS) in a FACScaliber (Becton Dickinson Instrument, San Jose, CA). Cell cycle fractions were quantified with CellQuest (Becton Dickinson).

Microarray Hybridization and Data Analysis
Total RNA was extracted with the use of Trizol reagent (Invitrogen, Carlsbad, CA) and the Qiagen RNAease Mini kit according to the manufacture's instructions (Valencia, CA). The methods for probe labeling reaction and microarray hybridization were described previously (26) except the total RNA was used directly for labeling. For all experiments, universal human reference RNA (UHR) (Stratagene, La Jolla, CA) was used to generate a reference probe for drug treated and untreated samples. 30 µg of total RNA from experimental samples or equal amount of UHR were labeled with Cy5 and Cy3, respectively, by using Superscript II Reverse Transcriptase (Invitrogen, Carlsbad, CA). The microarray hybridization, image process, and data normalization were as described previously (26). The log 2 ratios of each time point were then normalized for each gene to that of untreated cells (time 0) to obtain the relative expression pattern. The genes that showed substantial differences after drug treatment were selected based on a 2-fold change of expression value for at least two time points across all experiment conditions. A total of 1260 of ~19,000 genes met the criteria and were further analyzed using clustering and display programs (rana.stanford.edu/software) developed by Eisen et al (27).

Real-Time Quantitative RT-PCR
by guest on July 8, 2020 http://www.jbc.org/ Downloaded from Total RNA was extracted using RNeasy kit (Qiagen). 100 ng of total RNA from each sample was subjected to real time RT-PCR using ABI PRISM 7900 Sequence Detection System and SYBR Green master mix (Qiagen) according to manufacturer's protocol. Primers are available upon request. β-actin was used as an internal control for equal amount of RNA used.

Luciferase Reporter Gene Assay
For promoter reporter constructs, DNA fragments containing approximately 1-1.5 kb

Chromatin immunoprecipitation (ChIP) assay)
ChIP assays with HCT116 cells were carried out as described in Weinmann et al (28) and Wells and Farnham (29). Briefly, cells at different time point before and after 5-FU treatment were crosslinked with 1 % formaldehyde for 10 mins at room temperature.

Results:
Genome-wide effects in gene expression in response to 5-FU treatment in HCT116 and HCT116

p53 (-/-) cells
To study the regulation of gene expression by p53 during genotoxic drug-induced apoptosis, we treated the well-characterized colorectal cancer HCT116 cells and the p53-deficient subline with 5-FU, an antimetabolite anticancer drug known to induce p53-dependent apoptosis in these cells (24). Addition of 5-FU to the culture medium of HCT116 cells induced a strong p53 accumulation and apoptosis in a time-dependent manner. However, only a minimum apoptosis was induced in their p53 null counterparts (Fig.1, A  41 genes that were up-regulated whereas 239 genes that were down-regulated with induction occurring temporally earlier than repression (Fig. 1D). However, in HCT116 p53 (-/-) cells treated with 5-FU, these gene responses were either abolished or largely reduced suggesting that the major transcriptional effects by 5-FU are mediated by p53. A much smaller subset of genes were concomitantly increased following 5-FU treatment in both p53 wild-type and null HCT116 cells (Fig.1C, Cluster A). One gene of note in this cluster is Cyclin E2, which was significantly induced by 5-FU (~3 fold), and Cyclin E1, to a lesser degree (~ 1.5 fold). These genes may represent the general response to 5-FU and appeared to be independent of p53 status. The array results were confirmed in a select subset of these genes by real time RT-PCR (Fig. 2).

Genes affected by levels of p53 protein accumulation
Protein synthesis inhibitor Cycloheximide (CHX) has been shown to block p53 accumulation and to inhibit p53-dependent transcription (31). To identify genes whose by guest on July 8, 2020 http://www.jbc.org/ Downloaded from expression is directly influenced by p53 protein levels, we pretreated the culture with CHX for 30 min prior to the induction by addition of 5-FU and harvested the cells at different time points.
As predicted, pretreatment of HCT116 cells with 10 µg/ml CHX significantly abrogated 5-FUinduced p53 accumulation (Fig. 3A) and accordingly abolished 5-FU-induced apoptosis by 90 % (Fig. 3B). However, CHX treatment had negligible effect on low level of apoptosis observed in 5-FU-treated HCT116 p53 (-/-) cells (data not shown). This observation suggests the abrogation of p53 accumulation by CHX blocks 5-FU-induced apoptosis in HCT116 cells. Of 41 genes that were induced following p53 accumulation in the absence of CHX, 38 were either abolished or largely reduced in their induction in the presence of CHX (Fig. 3C). At all the time points analyzed, the expression of majority of p53 inducible genes are directly associated with p53 protein levels. In contrast to the p53-inducible genes, we found that approximately 40% of p53repressed genes (n=50) were insensitive to CHX treatment and therefore not p53 regulated (Fig.   3C). Thus, only those that are sensitive to blockade of p53 accumulation by CHX are considered truly p53 regulated genes. These genes are listed in Table 1. Note however, that since cycloheximide blocked p53 accumulation itself, we cannot discriminate between p53 primary and secondary target genes.
Induction of other known p53 targets such as MDM2, APAF, and KILLER/DR5 were also observed but to a lesser degree than those listed in Table 1. The successful detection of large number of previously identified p53 targets indicates that our system is robust and accurate in identifying p53 responsive genes. However, 3 out of 41 p53-inducible genes including known p53 targets GADD45A and PPMID (Fig. 3) were not sensitive to CHX treatment, indicating that these three genes must be regulated by p53 in a way that is distinct from other target genes. This observation seems to be in line with a previous report indicating that GADD45A is regulated by p53 through indirect transactivation (32). Of all p53-inducible genes, CDKN1A (encoding p21) was most strongly up-regulated (9-fold), which is consistent with a high binding affinity of p53 to p21 promoter (11). We also found that PUMA was most responsive among the detectable putative apoptotic targets induced by p53, supporting the previous reports that PUMA was a major apoptotic target mediating p53-induced apoptosis in colorectal cancer cells (33,34). We also identified 25 previously unidentified putative target genes up-regulated by p53. Of these 25, 19 (76%) were found to contain putative p53 binding sites in their promoter regions (data not shown). This compares to 29% in genes unresponsive to p53 induction. Interestingly, a candidate tumor suppressor gene SERPINB5 was also strongly induced by p53 at 24 h after 5-FU treatment (9-fold), pointing to a potential role of its gene product in mediating p53-mediated tumor suppressor function. The other two candidate genes of interest are DUSP5 &14, which encode dual specificity phosphatase 5 and 14, respectively. In line with these findings, a recent study shows that DUSP2, another member of this family, is a transcription target of p53 that functions in signaling apoptosis and growth arrest (15). Unexpectedly, we found that p53 activation also led to the induction of a number of genes associated with mitogenic responses. These genes include TGFα, SEK, TOP1, CNK and EPHA2 and all contain putative p53 binding sites in their promoters. In particular, TGFα and EPHA2 have been previously reported to be linked to the activation of MAPK signaling cascade that promotes cell growth (35,36). This implies that activation of p53 is accompanied by an induction of cellular mitogenic programme. This finding is consistent with recent studies showing that ability of p53 to induce apoptosis is counteracted by simultaneously augmenting opposing signals (37,38). As such, our identification of multiple by guest on July 8, 2020 http://www.jbc.org/ Downloaded from potential p53 target genes with growth-promoting functions is further evidence to support this notion.
p53 repressed genes p53-dependent activation of apoptotic targets has been extensively studied. However, the role of p53-mediated transcriptional repression in the induction of apoptosis is less investigated and in particular, has never been explored in a genomic scale. We observed that p53-mediated transactivation was maximally induced as early as 6-8 h after 5-FU treatment but induction of apoptosis did not become obvious until 12 h later when gene repression by p53 reached maximal (Fig. 1D). To examine whether p53-mediated transcriptional repression contributes to the apoptosis induced by 5-FU, CHX was added to HCT116 cells 6 h after 5-FU treatment when p53-mediated transactivation had been fully induced. As anticipated, this treatment caused a significance decrease in p53 accumulation at 12 h and abrogated all the p53 accumulation at 24 and 48 h in comparison with 5-FU treatment alone (Fig. 4A). In this manner, we were able to separate the immediate effects of p53 as compared to the late effects. As a consequence, no change in gene induction was noted, whereas gene repression was collectively affected ( Fig. 4 C & D). 5-FU plus CHX reduced apoptosis by 60% at 48 hours compared to 5-FU treatment alone. That no decrease in apoptosis was seen at 24 h implicates that p53-mediated transcriptional repression contributes to apoptosis later in the time course (Fig. 4B). Hence, the overall ability of p53 in inducing apoptosis appears to be tightly associated with its ability to repress gene expression. Among p53 repressed genes are those that function in mitosis (PLK, PTTG1, CHEK1, CDC20, CDC25B, CCNB1&2), and DNA replication and repair (MCMs, H2AX, NBS1, and RFC4)( Table 1). Previously known p53 repressed genes such as HSP70, CCNB2, TOP2A (39,40) were also included.

p53 represses the expression of PLK, PTTG1, and CHEK1 either directly or indirectly.
To gain further mechanistic insights of p53-dependent down-regulation of its target genes and to determine whether their down-regulation contributes to apoptosis, we selected three genes, PTTG1, PLK and CHEK1 for further evaluation because of their known importance in colorectal cancer biology and cell cycle checkpoint regulation (41-44). To further confirm the microarray data, we performed western blots to examine the effect on protein levels by p53. As anticipated, PLK, PTTG1 and Chk1 proteins were preferentially repressed in response to 5-FU in a time-dependent manner in HCT116 cells in comparison with HCT116 p53 (-/-) cells (Fig. 5A).
To determine whether their promoter sequences were responsive to p53, genomic fragments of ~ upper panel ). This fragment is located from nt-1429 to -1169. On the other hand, two out of 3 tested genomic fragments from PTTG1 promoter were found to be enriched following 5-FU treatment. The identified regions that show positive p53 binding are from nt -2624 to -2872 and from +102 to -177, respectively (Fig 6, lower panel). Importantly, the kinetics of p53 bindings to both PTTG1 and CHEK1 promoters are consistent with that of PTTG1 and CHEK1 RNA transcript levels following 5-FU treatment. Hence, the level of p53 binding inversely correlates with the accumulation of RNA transcripts. Together with the promoter assay, these results demonstrate that p53 interacts directly with p53-binding sites in CHEK1 and PTTG1 gene promoters and negatively regulate their transcription. However, we did not detect any p53 binding to the PLK1 promoter (data not shown). Therefore, PLK 1 promoter might be regulated by p53 through a mechanism that does not require direct p53 binding.

RNAi-mediated knockdown of PLK or PTTG1 directly causes apoptosis
p53 repressed gene targets BCL-2 and survivin have been shown to be anti-apoptotic (4,17,45). To test whether PLK and PTTG1 are required for maintaining cell survival or antiapoptotic, we used small interference RNA (siRNA) to specifically reduce their expression in HCT116 cells. Following transfection with siRNA for PLK, the level of PLK protein fell by more than 80% after 48 h (Fig. 7A). PTTG1 protein expression was similarly effectively reduced by PTTG1 siRNA. Consequently, HCT116 cells transfected with PLK or PTTG1 siRNAs showed a significant increase (~100%) in apoptosis in comparison to cells transfected with unrelated control siRNA (Fig. 7B). In addition, cells depleted of PLK or PTTG1, were more sensitive to 5-FU treatment compared to cells treated with unrelated control siRNA and this effect appeared to be more pronounced in HCT116 p53 (-/-) cells (data not shown) suggesting that inhibition of PLK or PTTG1 was sufficient to induce apoptosis in the absence of p53.
Together, these observations indicate that individual p53-repressed genes, such as PLK and PTTG1, appear to be important in maintaining the cell survival.

Discussion
Although a number of p53 targets have been identified and multiple cellular events involving p53 continue to increase in complexity, the entire picture of p53-regulated transcription in the regulation of apoptosis has not been fully explored. HCT116 cells exhibiting p53-dependent apoptosis in response to anticancer drug 5-FU provides a model system for identifying molecular events important in p53-mediated apoptosis of tumor cells. Using this system and a genome-wide approach, we have identified approximately 230 out of 19,000 genes responsive to 5-FU in a p53-dependent manner using a stringent selection algorithm. Validating this approach for discovery of novel targets regulated by p53, we have identified more than 20 previously known p53 target genes in our study.
There is now compelling evidence that the transcriptional activity of tumour suppressor p53 is required for its growth suppressing and tumour suppressing activities. Most studies have focused on the transactivation function of p53 because of the strong association between transactivation and tumour suppression. However, p53 is also able to repress gene expression but the functional consequence of this repression is largely unexplored. In this study, we show that p53 regulates the majority of gene transcription through transcriptional repression. We further demonstrate that transcriptional repression by p53 on a genomic scale is important for its ability to promote apoptosis since selective abrogation of gene repression function of p53 significantly diminishes the apoptotic response. As such, p53 may regulate apoptosis through a coordinated programme that includes both the activation of apoptotic genes (e.g., PUMA, NOXA, PIG3, FAS/CD95 and DUSP2) and the repression of potential survival genes (e.g., PLK, PTTG1, or CHEK1). Consistent with our observation is a recent report in which 80% of the p53-responsive genes have been found to be repressed rather than activated (17 Among genes induced by p53 activation, we found unexpectedly that p53 is able to induce multiple genes participating in cell proliferation or pro-survival pathways. Genes included in this category include early-growth-response genes SNK and CNK, also known as immediateearly genes that play important roles in regulating cell proliferation (51-53). We also found that      After 24 h, cells were lysed and luciferase activity was measured as a relative value of measured luminescence from firefly luciferase and Renilla luciferase from cotransfected pRL-null (internal control). Reporter activity induced by empty vector control was arbitrarily set as 1. Values represent the mean and variation of a typical experiment performed in triplicates.   Listed are genes that showed at least 2-fold changes in response to 5-FU treatment for at least two time points in HCT116 cells and these genes are sensitive to inhibition of p53 accumulation by CHX in the experiment shown in Fig. 3. Genes in bold are previously identified p53 targets.