Main

In recent years, haplo-insufficiency has been demonstrated for several key tumour suppressor genes, including ATM, APC, BRCA1, BRCA2, p27, PTEN, SMAD4 and Nkx3.1 (reviewed in Cook and McCaw, 2000; Santarosa and Ashworth, 2004; Payne and Kemp, 2005). p53 is a crucial tumour suppressor gene often referred to as the guardian of the genome, and is found mutated or deleted in over 50% of human cancers (Hollstein et al., 1991). p53 orchestrates tumour suppression by acting as a central decision-making node via its functional capacity as a transcription factor. Thus, p53 couples detection of diverse stresses to regulation of the cell cycle, programmed cell death and DNA repair (reviewed in Vogelstein et al., 2000; Vousden, 2002; Vousden and Lu, 2002). However, despite 25 years of intensive research, a complete description of p53's role in the cell and the fundamental mechanisms by which it operates remains to be elucidated (Fojo, 2002; Hofseth et al., 2004). For example, although loss of one wild-type p53 allele is associated with increased tumorigenesis in mice and in humans, the underlying molecular mechanism remains unclear (Varley et al., 1997; Venkatachalam et al., 1998, 2001).

Li–Fraumeni Syndrome (LFS) is an inherited human condition characterized by a germline mutation in one p53 allele and greatly increased cancer incidence (Varley et al., 1997). p53 mutant alleles are unusual in that their mutant p53 protein products may retain some residual activity (Soussi and Lozano, 2005). Thus, explanations for the increased tumorigenesis observed in LFS include: dominant-negative p53 mutants, oncogenic gain of function p53 mutants or an increase in p53 inhibitors (Blagosklonny, 2000; Parant and Lozano, 2003). However, generation of p53 knockout cell and mouse models has shown that such events are exceptions rather than the rule. Moreover, simply reducing p53 gene dosage to a p53+/− genotype promotes cancer formation (Venkatachalam et al., 1998, 2001). Thus, p53+/− cells, mice and humans can exhibit an attenuated tumour-suppressive phenotype despite expressing transactivation-competent wild-type p53 protein (Camplejohn et al., 1995; Williams et al., 1997; Venkatachalam et al., 1998). To explore this paradox, we have used a human cell model of reduced wild-type p53 gene dosage. HCT116 are human epithelial cells of colorectal adenocarcinoma origin in which p53 responses are intact (Waldman et al., 1995; Bunz et al., 1998). Established isogenic clones of defined p53 ploidy have been generated by targeted homologous recombination (Bunz et al., 1998). Hence, HCT116 p53+/+, p53+/− and p53−/− cells provide an informative model for investigating the molecular mechanism linking reduced p53 gene dosage with attenuated tumour suppressor function in humans. Since loss of p53 is the most common genetic mutation in cancer (Hollstein et al., 1991), understanding the impact of a haploid p53 genotype is fundamental to the pathology of LFS and also imperative in understanding whether loss of one or both wild-type p53 genes is the critical event in early tumorigenesis.

Basal levels of p53 according to p53 gene dosage

Basal levels of p53 messenger RNA (mRNA) and protein were analysed in HCT116 p53+/+, p53+/− and p53−/− cells. Unexpectedly, both RT–PCR and quantitative real-time PCR indicated that p53 mRNA levels were approximately four-fold lower in p53+/− cells than p53+/+ cells under basal, nonstressed conditions (Figure 1a). p53 protein is constitutively maintained at tightly regulated and extremely low nuclear levels in the absence of stress. Remarkably, Western blotting indicated that total p53 protein levels were also approximately fourfold lower in p53+/− cells than p53+/+ cells under nonstressed conditions (Figure 1b). Thus, a two-fold reduction in p53 gene dosage reduced basal expression of p53 mRNA and p53 protein by ∼4-fold.

Figure 1
figure 1

Wild-type p53 gene dosage affects basal p53 mRNA and protein levels under nonstressed conditions. HCT116 isogenic p53+/+, +/− and −/− cells were incubated under nonstressed conditions. p53 genotype is indicated below each bar. All results are representative of at least three independent cell experiments. (a) Analysis of p53 mRNA levels. Total RNA was isolated (RNeasy Kit, Qiagen). Panels show 1.5% agarose gel electrophoresis (200 ng/mL ethidium bromide) of p53 RT–PCR products after 27 cycles, and LaminA/C after 25 cycles from the same aliquot as an internal loading control. Bar chart data represents the mean±s.d. of three determinations by quantitative real-time PCR (SYBRgreen Quantitect kit, Qiagen; Opticon2 instrument, MJ Research) of p53 mRNA levels according to p53 gene dosage, corrected by the corresponding LaminA/C mRNA levels and expressed as a percentage of p53+/+ sample. p53+/− data were tested for significance relative to p53+/+ using Student's T-test: *P<0.01. Note: Both LaminA/C and GAPDH were routinely used as duplicate internal controls for qRT–PCR data throughout these studies and both gave equivalent results (data not shown). Primers for p53 – forward: 5′-ATG GAG GAG CCG CAG TCA GAT-3′; reverse: 5′-GCA GCG CCT CAC AAC CTC CGT C-3′; for LaminA/C – forward: 5′-ATG GAG ACC CCG TCC CAG CG-3′; reverse: 5′-TAT ACT GCT CCA CCT GGT CCT CAT GC-3′. (b) Analysis of p53 protein levels. Whole-cell protein lysates were prepared and ∼80 μg protein loaded per lane as in Rubbi and Milner (2003). Total p53 was probed by immunoblot (mouse anti-p53 DO1 monoclonal antibody). Actin levels were also probed as an internal loading control (mouse anti-actin, mAB1501, Chemicon). p53 protein band intensity was corrected by actin using densitometry to generate the bar chart.

Full size image

Analysis of p53 stress response according to p53 gene dosage

3H-methyl-thymidine is a radioactive DNA-labelling reagent, incorporated into DNA during replication, which can cause cell stress in the form of DNA damage (Yanokura et al., 2000; Hu et al., 2001; Mirzayans et al., 2003). UV irradiation also induces stress by generating DNA damage such as cyclobutane pyrimidine dimers and 6-4-photoproducts (Latonen and Laiho, 2005). Levels of p53 mRNA remained unchanged following incubation with 3H-methyl-thymidine or UV irradiation (Figure 2a). However, the fourfold difference in p53 mRNA levels between p53+/+ and p53+/− cells persisted following both forms of cell stress (Figure 2a).

Figure 2
figure 2

p53 mRNA and protein levels according to wild-type p53 gene dosage following cell stress. (a) Quantitative real-time PCR analysis of p53 mRNA levels before/after two modes of cell stress. HCT116 isogenic p53+/+, +/− and −/− cells were incubated in nonstressed conditions or treated with either 8 kBq/mL of 3H-methyl-thymidine (Amersham, Stock=88 μCi/mmol) for 48 h or exposed to 10 J/m2 UV irradiation (at a fluency of 2 J/m2/s; Rubbi and Milner, 2003) and harvested after 48 h. Total RNA isolation, qRT–PCR, and primers for p53 and LaminA/C exactly as in Figure 1. p53 and LaminA/C mRNA levels were analysed from the same RNA aliquot as an internal control. Bar chart data represent mean±s.d. of three determinations of p53 mRNA levels corrected by corresponding LaminA/C levels and expressed as a percentage of p53+/+ control sample. p53 genotype and cell treatments are indicated below each bar. Neg. Control refers to dH2O-only PCR control. (b) p53 protein stabilization following UV irradiation. HCT116 isogenic p53+/+, +/−, and −/− cells were incubated in nonstressed conditions or UV-irradiated as in (a). Immunoblotting for total p53 and actin as an internal control exactly as in Figure 1b. p53 protein band intensity was corrected by actin using densitometry to generate the bar chart data. p53 genotype and cell treatments are indicated. (c) p53 protein stabilization across a gradient of cell stress. HCT116 isogenic p53+/+, +/− and −/− cells were incubated with a spectrum (0–12 kBq/ml) of 3H-methyl-thymidine for 72 h. Immunoblotting for total p53 and actin as an internal control in whole-cell protein lysates exactly as in Figure 1b. p53 protein band intensity was corrected by actin using densitometry to generate the bar chart data (see also Figure 3a for p53 and actin band images). Results are representative of three independent cell experiments. (d) Relationship between p53 mRNA and p53 protein levels with increasing cell stress according to p53 gene dosage. Relative p53 protein levels and p53 mRNA levels in HCT116 p53+/+ and +/− cells are depicted to summarize the results of Figures 1 and 2. The X-axis represents the cell stress gradient examined. A vertical increase on the Y-axis represents increased levels. ‘∼ × 4’ indicates an approximately × 4-fold difference in both p53 mRNA and protein between p53+/+ and +/− cells. ‘∼ × 5’ indicates an approximately × 5-fold increase in p53 protein levels with increasing cell stress.

Full size image

Following cell stress, the rate of p53 translation is increased, while the rate of p53 protein degradation is decreased, leading to elevated p53 levels and transactivation of p53-target genes (Vousden, 2002). Cell exposure to UV irradiation increased p53 protein levels approximately fivefold in both p53+/+ and p53+/− cells (Figure 2b). Remarkably therefore, a fourfold difference in p53 protein levels was maintained between p53+/+ and p53+/− cells despite p53 protein stabilization. Cell incubation with increasing 3H-methyl-thymidine concentrations demonstrated a similar differential in p53 protein levels between p53+/+ and p53+/− cells (Figures 2c and 3a). Approximately fourfold less p53 was detected in p53+/− than p53+/+ cells across the stress spectrum examined (Figure 2c). Therefore, despite a large difference in absolute p53 protein level (∼4-fold), the degree of p53 stabilization was similar (∼5-fold) for both p53+/+ and p53+/− cells (Figure 2d).

Figure 3
figure 3

Effect of p53 stress response on p53 target-gene regulation according to p53 gene dosage. (a) Protein product levels of p53 and p53-target genes according to wild-type p53 gene dosage across a gradient of cell stress. HCT116 isogenic p53+/+, +/− and −/− cells were incubated with a gradient of 3H-methyl-thymidine activities. Immunoblotting was performed exactly as in Figure 1b or: total p53 (mouse anti-DO-1); p53 phosphorylated at serine15 (p53Ser15P, mouse anti-p53 Serine15P, Cell Signalling Tech.); p21waf1 (rabbit anti-p21); Mdm2 (mouse anti-2A9, in-house hybridoma), SIRT1 (rabbit anti-H300) and actin (mouse anti-actin, mAb1501, Chemicon) as a loading control. All antibodies from Santa Cruz unless otherwise stated. 3H-methyl-thymidine activity, p53 genotype and molecular weight are indicated. Results are representative of three independent cell experiments. (b) p21waf1 (p21) and Mdm2 mRNA levels following two modes of cell stress. Cells treated and quantitative real-time PCR performed as in Figure 2a, but here for p21 and Mdm2. Bar chart data represent mean±s.d. of three determinations of p21waf1 mRNA (upper panel) or Mdm2 mRNA (lower panel) corrected by corresponding LaminA/C levels. p53+/− and p53−/− data were tested for significance relative to p53+/+ for each treatment using Student's t-test: *P<0.1, **P<0.05. Primers for p21waf1 were forward: 5′-AGG CAC CGA GGC ACT CAG AG-3′; reverse: 5′-AGT GGT AGA AAT CTG TCA TGC TG -3′. Primers for Mdm2 were forward: 5′-AAGAGACCCTGGTTAGACCAAAGC-3′; reverse: 5′-TTTCTTCTGTCTCACTAATTGCTCT-3′. LaminA/C primers same as for Figure 2a. p53 genotype and cell treatments are indicated below each bar. (c) Survivin and SIRT1 mRNA levels with increasing UV-irradiation doses over a time course. HCT116 p53+/+ and +/− cells were exposed to 0, 5, 10 or 20 J/m2 UV irradiation and harvested after the times indicated. Quantitative real-time PCR was performed as in Figure 2a, but here for survivin and SIRT1. Bar chart data represent mean±s.d. of three determinations of survivin mRNA (upper panel) or SIRT1 mRNA (lower panel) corrected by corresponding GAPDH levels as an internal control. p53+/− data were tested for significance relative to p53+/+ using Student's t-test: *P<0.05. Primers for survivin were: forward: 5′-GCATGGGTGCCCCGACGTTG-3′; reverse: 5′-TCAATCCATGGCAGCCAGCTG-3′; for SIRT1, forward: 5′-TCAGTGTCATGGTTCCTTTGC-3′; reverse: 5′-AATCTGCTCCTTTGCCACTCT-3′ (Ford et al., 2005); for GAPDH, forward: 5′-CGGAGTCAACGGATTTGGTCGTAT-3′; reverse: 5′-AGCCTTCTCCATGGTGGTGAAGAC-3′. p53 genotype and cell treatments are indicated below each bar.

Full size image

p53 phosphorylation at serine 15 is associated with p53 stabilization and activation in response to DNA damage (Bean and Stark, 2001; Bode and Dong, 2004). Western blot probes indicated that serine 15 phosphorylation increased sharply with increasing stress; however, levels were clearly lower in p53+/− cells than p53+/+ cells, both before and after stress (Figure 3a). Hence, the results for p53 serine 15 phosphorylation paralleled the results for total p53 levels. This indicates that p53+/− cells failed to upregulate an important post-translational mechanism for p53 stabilization to compensate for the difference in total p53 levels.

The above results demonstrate that loss of one wild-type p53 allele is associated with four-fold reduction of p53 protein and mRNA, a difference that persists under both stressed and nonstressed conditions. These observations were unexpected and of great potential interest given the importance of p53 protein levels for its tumour-suppressive functions (Vousden, 2002). Three intriguing aspects emerge from a simplified summary of our observations of the relationship between p53 protein and mRNA levels in p53+/+ and +/− cells (Figure 2d). First, a twofold reduction in p53 gene dosage results in an × 4-fold reduction in p53 mRNA. This implies that a constitutive mechanism must operate at the level of p53 transcription or mRNA stability, whose effect is dependent on both p53 alleles being present. The p53 mRNA transcript is known to contain 3′-regulatory elements for translational repression (Fu et al., 1999), and protein-binding to mRNA 3′-UTRs is a common mechanism of regulating mRNA stability (de Moor et al., 2005). Conversely, an auto-regulatory mechanism may be involved at the transcriptional level, since p53 can also bind to its own promoter region (Deffie et al., 1993), which also contains a negative regulatory element (Bienz-Tadmor et al., 1985).

Secondly, on comparison of p53+/+ and +/− cells under nonstressed conditions, we observed a close correlation between p53 mRNA and basal p53 protein levels: both being ∼4-fold lower in p53+/− cells (Figure 2d). Thirdly, despite the fourfold reduction in p53 protein levels, p53 stabilization and post-translational modification were parallel in p53 +/+ and +/− cells following stress (Figures 2c, d and 3a). It has been reasonably assumed that damage-transducing circuitry upstream of p53 is identical in HCT116 p53 isogenic clones under conditions of equivalent stress (Waldman et al., 1995; Bunz et al., 1998). Thus, the results here indicate that p53+/− cells are either unable to detect or compensate for subnormal p53 protein levels by translational or post-translational mechanisms. This is particularly surprising, given the multiple mechanisms that have evolved to maintain p53 protein levels for tumour suppression. Similarly, the p53 mRNA results also imply that adult human cells are either unable to detect or compensate for variations in p53 mRNA. We hypothesize that levels of p53 mRNA are dominant over translational and post-translational regulatory mechanisms for p53 protein levels in the cell, such that p53 mRNA levels are a critical determinant of p53 function. Indeed, inducible overexpression of wild-type p53 from exogenous constructs depends on this phenomenon and can induce arrest/apoptosis in the absence of cell stress. p53 mRNA levels have been linked with tissue radiosensitivity in mice (Komarova and Gudkov, 1998). It is noteworthy that p53 mRNA accumulates to high levels during embryogenesis, and that embryonic cells also display enhanced sensitivity to p53-dependent apoptosis (Rogel et al., 1985; Gottlieb et al., 1997). Conversely, RNA interference knockdown of p53 mRNA to subnormal levels was recently shown to also constitutively reduce p53 protein levels and attenuate p53-dependent regulation of arrest/apoptosis in human cells (Hemann et al., 2003).

p53 target-gene regulation

p53 target-gene regulation was investigated to determine if the observed difference in p53 protein levels affected its efficiency as a transcription factor (Figure 3). A critical downstream mediator of p53-dependent cell cycle arrest is the p53-target gene p21waf1/cip1 (here-in referred to as p21; El-Deiry et al., 1994; Waldman et al., 1995; Bunz et al., 1998). Following cell stress, p53 transactivates p21 expression, which in turn enforces cyclin-dependent kinase inhibition and cell cycle arrest. Here we found that p21 mRNA was induced approximately three- and 4.5-fold in p53+/+ cells after 3H-methyl-thymidine incubation and UV irradiation, respectively (Figure 3b). In contrast, p21 mRNA levels were very low p53−/− cells under nonstressed conditions and failed to increase following cell stress (Figure 3b). This indicated that p21 mRNA induction was wholly p53-dependent in this assay. Interestingly, basal p21 mRNA levels in p53+/− cells were lower than in p53+/+ cells under nonstressed conditions, and p21 induction was severely attenuated following stress (Figure 3b). These results prompted investigation of p21 protein levels. p21 protein levels in p53−/− cells remained unchanged across the stress spectrum examined (Figure 3a). Notably, in nonstressed conditions, p21 protein was lower in p53+/− cells than in p53+/+ cells. p53-dependent p21 induction was observed strongly, but differentially in both p53+/+ and p53+/− cells in that p53+/− cells contained less p21 protein than p53+/+ cells at corresponding levels of stress (Figures 3a). Thus, p21 protein levels paralleled p53 protein levels, being lower in p53+/− cells than p53+/+ cells. These results demonstrate that loss of one p53 allele is associated with reduction of p21 mRNA and protein levels. Therefore, p53 is haplo-insufficient for maintaining a normal p21 response. These observations also suggest a constitutive role for p53 in sustaining basal levels of p21 mRNA and protein under nonstressed conditions, in agreement with previous reports of constitutive p53 function in the absence of stress (Espinosa and Emerson, 2001; Kaeser and Iggo, 2002; Espinosa et al, 2003) and in the developing mammalian embryo (Gottlieb et al., 1997). The results are also consistent with a previously suggested hypothesis that under nonstressed conditions basal p53 protein levels are limiting, so that p53-dependent functions might be especially sensitive to p53 expression levels (Gottlieb et al., 1997; Venkatachalam et al., 1998).

Mdm2 is the principal negative regulator of p53 stabilization, inducing p53 degradation and hindering access of transcriptional cofactors to the N-terminal p53 transactivation domain. Mdm2 is also a p53-target gene and therefore p53 stabilization is tightly linked to p53-dependent Mdm2 induction, forming a negative feedback loop that acts to limit p53 protein levels and transactivity (reviewed in Freedman et al., 1999; Iwakuma and Lozano, 2003). Hence, Mdm2 mRNA and protein levels were investigated in order to explore the observed differences in p53 protein levels. Mdm2 protein levels in HCT116 p53+/+, +/− and −/− cells exposed to cell stress indicated p53-dependent Mdm2 induction (Figure 3a). In particular, Mdm2 protein levels closely correlated with p53 levels, being lower in p53+/− cells than p53+/+ cells across the cell stress spectrum examined. Mdm2 mRNA levels were also p53-dependent under both stressed and nonstressed conditions (Figure 3b). Notably, loss of one p53 allele was associated with attenuated Mdm2 mRNA levels. Thus, differential Mdm2 expression correlated with p53 protein levels according to p53 gene dosage. Importantly, these findings indicate that the p53-Mdm2 negative feedback loop is intact, and therefore that aberrant Mdm2 activity is not responsible for the reduced p53 protein levels in p53+/− cells.

Survivin is an inhibitor of apoptosis that is repressed by p53 at the transcriptional level (Hoffman et al., 2002; Mirza et al., 2002). Survivin mRNA levels were assessed to measure the efficiency of p53-dependent transrepression according to p53 gene dosage. Survivin mRNA levels fell dramatically with increasing stress insult in p53+/+ cells (Figure 3c). However, repression of survivin mRNA was absent in p53+/− cells (Figure 3c). The results indicate p53 is haplo-insufficient for transrepression of survivin, since loss of one p53 allele abrogated this function. The results also imply that a threshold p53 protein level is required in order to enable repression at the survivin promoter.

SIRT1 is a NAD-dependent protein deacetylase known to be involved in metabolic homeostasis, development, differentiation, longevity, stress tolerance and suppression of apoptotic pathways (Giannakou and Partridge, 2004). Under nonstress conditions, SIRT1 and FOXO4 regulate apoptosis in HCT116 cells (Ford et al., 2005). Following cell stress, SIRT1 has an antiapoptotic influence through modulation of the FOXO transcription factors, promoting FOXO-dependent arrest over apoptosis (Brunet et al., 2004). SIRT1 can also deacetylate and thereby downregulate p53 directly (Luo et al., 2001). Reciprocal p53-dependent regulation of SIRT1 transcription was postulated via two p53 binding sites identified in the SIRT1 promoter (Nemoto et al., 2004). However, no evidence of this was observed in HCT116 cells (Ford et al., 2005). Hence, the nature of the physiological relationship between SIRT1 and p53 following cell stress remains unresolved. HCT116 p53 isogenic human cells were ideally suited to resolve the stress-induced endogenous p53–SIRT1 relationship. SIRT1 mRNA levels were transiently repressed following higher stress in p53+/+ cells (Figure 3c). Regulation of SIRT1 mRNA levels after stress was sensitive to loss of one p53 allele (Figure 3c). Under nonstressed conditions, a similar level of SIRT1 protein was apparent in HCT116 p53+/+ and −/− cells, in agreement with previous findings (Ford et al., 2005). In response to increasing stress, SIRT1 protein levels were constant at low stress, but decreased in p53+/+ cells at higher stress (Figure 3a). In contrast, at equivalent levels of stress SIRT1 protein levels remained constant in p53+/− cells and were moderately increased in p53−/− cells (Figure 3a). Thus, levels of SIRT1 mRNA and protein were significantly decreased under the higher levels of stress examined in p53+/+ cells. The observations in p53+/− cells indicate that p53-dependent regulation of SIRT1 is sensitive to the lower levels of p53 protein associated with loss of one p53 allele. Taken together with previous reports, the current findings support an antagonistic relationship between p53 and SIRT1 following cell stress, for which p53 is haplo-insufficient.

Cell cycle distribution according to p53 gene dosage

p53-dependent p21 transactivation is one of the best characterized biochemical circuits capable of negatively regulating cell cycle progression (El-Deiry et al., 1994; Bunz et al., 1998). p21 induction and p21-dependent cell cycle arrest are central to tumour suppression in vivo (Williams et al., 1997; Geng et al., 2004). In light of the attenuated p53-p21 antiproliferative circuit observed at the molecular level (see Figure 3a and b), we next investigated cell cycle distribution according to p53 gene dosage. Overall, cell cycle distribution varied in a p53-dependent manner both under stressed and nonstressed conditions (Figure 4a and b). Surprisingly, p53+/− cells displayed an intermediate cell cycle distribution phenotype. The p53–p21 antiproliferative circuit is critical for G1 checkpoint integrity (El-Deiry et al., 1994; Bunz et al., 1998). Loss of one p53 allele resulted in an intermediate proportion of cells in G1-phase before and after stress (Figure 4a and b), correlating with the observed attenuation of p53-dependent p21 transactivation in these cells. The proportion of sub-G1 cells following exposure to stress is an indicator of apoptosis. Loss of one p53 allele resulted in an intermediate proportion of sub-G1 cells after stress (Figure 4c), correlating with the observed attenuation of p53-dependent survivin transrepression in these p53+/− cells (see: Figure 3c). Taken together, these results indicate a dose-dependent relationship between cell cycle distribution and p53 gene dosage even under nonstressed conditions. In addition, the results suggest that p53 is haplo-insufficient for G1-checkpoint integrity in human cells under both stressed and nonstressed conditions.

Figure 4
figure 4

Cell cycle distribution is dependent on p53 gene dosage. (a, b) Examination of cell cycle distribution according to wild-type p53 gene dosage before and after cell stress. HCT116 isogenic p53+/+, +/− and −/− cells were incubated for 48 h before either (a) mock treatment or (b) UV irradiation with 10 J/m2 as in Figure 2b. At 24 h after treatment, cells were harvested, fixed with 70% ethanol and stained with 30 μg/ml propidium iodide (PI) overnight before FACS analysis (Becton Dickinson FACSCalibur, 10 000 events/sample). Histograms display cell cycle distribution with two peaks of 2n and 4n cells. Cell cycle gating and analysis were carried out using Cell Quest software as in Allison and Milner (2003). Percentage of cell population in G1-phase was derived by histogram deconvolution using Cyclchred software, and is displayed in each panel. p53 genotype, cell counts and PI intensity are indicated. (c) Percent of total cell population in subG1-phase versus gradient of cell stress according to p53 gene dosage. Cell treatments were exactly as for (a), except that here UV-irradiation doses were 0, +5, +10 or +20 J/m2. Cell cycle distribution was analysed and percentage of cell population in subG1 was derived as for (a).

Full size image

In this work, we show that the p53 gene is haplo-insufficient: (i) for transactivation of p21 and Mdm2, (ii) for transrepression of survivin, and (iii) for a novel role in regulating SIRT1 expression after stress. Attenuation of the p53–p21 antiproliferative circuit is most likely central to concomitant variations in cell cycle distribution also observed here with loss of one p53 allele. Furthermore, under nonstress conditions, evidence for constitutive maintenance of p21 expression by basal levels of p53 was indicated. These observations are consistent with a requirement for a p53 protein threshold for regulation of p53-target genes. The observations with p21 and survivin demonstrate that the p53 threshold phenomenon applies to both transactivation and transrepression of p53-target genes.

Mutation or deletion of one wild-type p53 allele occurs in over 50% of human cancers (Hollstein et al., 1991), while loss of one p53 allele has been postulated to reduce the threshold for transformation (Gottlieb et al., 1997; Venkatachalam et al., 2001). In addition, aberrant cell cycle regulation provides an environment more conducive to accumulation of oncogenic lesions and initiation of tumorigenesis (Bunz et al., 1998; Geng et al., 2004). This work establishes that loss of one wild-type p53 allele in human cells is associated with an ∼4-fold reduction in p53 mRNA and p53 protein levels, reduced p53 stress responses, and attenuated p53-dependent cell cycle regulation. We conclude that: (i) cells that have lost one wild-type p53 allele are unable to compensate for the reduced levels of p53 mRNA and p53 protein which they exhibit; (ii) p53 mRNA levels are a critical determinant of p53 function; (iii) p53 displays haplo-insufficiency for transactivation and transrepression of specific target genes. Thus, reduction of wild-type p53 mRNA levels may provide a molecular basis to explain p53 haplo-insufficiency, and therefore also the increased tumorigenesis observed in mice and LFS humans who have suffered loss of one wild-type p53 allele.