MCB
Home Help [Feedback] [For Subscribers] [Archive] [Search] [Contents]
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplemental material
Right arrow Other Versions of this Article:
MCB.01316-06v1
27/15/5336    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowReprints and Permissions
Right arrow Copyright Information
Right arrow Books from ASM Press
Right arrow MicrobeWorld
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Marusyk, A.
Right arrow Articles by DeGregori, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Marusyk, A.
Right arrow Articles by DeGregori, J.

 Previous Article  |  Next Article 

Molecular and Cellular Biology, August 2007, p. 5336-5351, Vol. 27, No. 15
0270-7306/07/$08.00+0     doi:10.1128/MCB.01316-06
Copyright © 2007, American Society for Microbiology. All Rights Reserved.

p53 Mediates Senescence-Like Arrest Induced by Chronic Replicational Stress{triangledown} ,{dagger}

Andriy Marusyk,1 Linda J. Wheeler,2 Christopher K. Mathews,2 and James DeGregori1*

Department of Biochemistry and Molecular Genetics, Program in Molecular Biology, Integrated Department of Immunology, University of Colorado at Denver Health Sciences Center, Aurora, Colorado 80045,1 Department of Biochemistry and Biophysics, Oregon State University, Corvallis, Oregon 97331-73052

Received 18 July 2006/ Returned for modification 1 September 2006/ Accepted 13 May 2007


   ABSTRACT
Top
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Previous studies have shown that exposure of cells to high levels of replicational stress leads to permanent proliferation arrest that does not require p53. We have examined cellular responses to therapeutically relevant low levels of replicational stress that allow limited proliferation. Chronic exposure to low concentrations of hydroxyurea, aphidicolin, or etoposide induced irreversible cell cycle arrest after several population doublings. Inhibition of p53 activity antagonized this arrest and enhanced the long-term proliferation of p53 mutant cells. p21CIP1 was found to be a critical p53 target for arrest induced by hydroxyurea or aphidicolin, but not etoposide, as judged by the ability of p21CIP1 suppression to mimic the effects of p53 disruption. Suppression of Rad51 expression, required for homologous recombination repair, blocked the ability of mutant p53 to antagonize arrest induced by etoposide, but not aphidicolin. Thus, the ability of mutant p53 to prevent arrest induced by replicational stress per se is primarily dependent on preventing p21CIP1 up-regulation. However, when replication stress is associated with DNA strand breaks (such as with etoposide), up-regulation of homologous recombination repair in response to p53 disruption becomes important. Since replicational stress leads to clonal selection of cells with p53 mutations, our results highlight the potential importance of chronic replicational stress in promoting cancer development.


   INTRODUCTION
Top
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
TP53 (p53), the "guardian of the genome," is a central player in cellular responses to a variety of stresses, including DNA damage, inappropriate growth signaling, and hypoxia (90). Depending on the cellular context, intensity of stimulation, and other factors, activation of p53 can result in temporary arrest of proliferation, allowing cells to repair the damage and preserve clonal survival, or in contrast, activation of p53 might lead to clonal elimination via either apoptosis or permanent proliferation arrest. Given that p53 is mutated in roughly half of human malignancies and that most malignancies retaining wild-type p53 have other defects in the p53 pathway, loss of normal p53 function is an important step in the evolution of cancers.

Senescence, a permanent proliferation arrest, is one possible outcome of p53 activation. Senescence can be induced by shortened telomeres, hyperactive oncogenic signaling, or DNA damage (78). Induction of senescence is thought to involve the p53-p21CIP1 and p16INK4A-retinoblastoma protein (Rb) pathways, although the relative contributions of these pathways may vary depending on the cell type (16). As for apoptosis, senescence is a potent mechanism to inhibit the proliferation of cells with activated oncogenes or excessive DNA damage. Recently, the induction of senescence has been revealed as an important physiological block to tumorigenesis induced by oncogene activation (8, 20, 22, 61). Induction of senescence is also associated with tumor responses to genotoxic stress (55, 73, 86). Still, although disruption of p53 in tumor cells treated with genotoxic agents was reported to antagonize phenotypic aspects of senescence, it did not improve long-term survival, as cells that avoided senescence were eliminated by apoptosis or mitotic catastrophe (19, 86).

In the context of inappropriate oncogenic signaling, p53 disruption provides mutated cells with a selective advantage, allowing abnormal proliferation while preventing apoptotic death or growth arrest (54). However, in contexts of genotoxic stress, the advantage of losing p53 is not as clear. Although the loss of p53-dependent apoptosis might be expected to increase tumor cell survival following chemotherapy or radiation therapy, the loss of p53-dependent cell cycle arrest may sensitize cancer cells to genotoxic agents, as cell cycle arrest facilitates the repair of DNA damage and entering mitosis with unrepaired DNA damage can be lethal (12, 75). The inability of p53 mutation to promote the sustained proliferation of cells with DNA damage raises the question of what insults can select for p53 mutations during carcinogenesis.

One of the insults that activates p53 is replicational stress, resulting from the inhibition of DNA synthesis (90). Replicational stress can be caused by several mechanisms, including decreased deoxynucleoside triphosphate (dNTP) synthesis, inhibition of replicative polymerases, or impeding of replication fork progression by obstacles, such as DNA strand breaks or DNA adducts. Inhibition of DNA replication leads to the generation of extended regions of single-stranded DNA, which is coated by replication protein A and sensed by ATRIP, leading to activation of ATR-CHK1 kinase signaling (97). The activated checkpoint maintains replication fork stability (allowing resumed DNA replication when dNTPs become available again) and prevents mitosis with incompletely replicated chromosomes (68). Also, the collapse of stalled replication forks might lead to the formation of DNA double-strand (DS) breaks, resulting in the activation of the ATM-CHK2 axis. Because different types of insults lead to different types of DNA abnormalities and different ratios of DNA single-strand to DS breaks, the checkpoint responses and DNA repair pathways needed to deal with the stress might vary, complicating the understanding of replicational-stress responses.

The most commonly used experimental agent to induce replicational stress is the inhibitor of ribonucleotide reductase, hydroxyurea (HU). Cells progress through G1 and into S phase at a normal rate in the presence of HU but accumulate in early S phase with hyperphosphorylated Rb, high levels of a subset of E2F target gene products, and high cyclin E-dependent kinase activity (49). Despite activating p53, S-phase arrest induced by HU treatment has been shown not to require p53 (50). Experiments at the Prives laboratory demonstrated that up to 48 h of HU treatment induces minimal or delayed accumulation of the p53 target gene products p21CIP1 (henceforth called p21), Mdm2, cyclin G, and Gadd45, while PIG3 was effectively induced (31). Notably, another group has shown that HU, like {gamma}-irradiation, induces p53-dependent transcriptional increases in the expression of target genes like p21 and Mdm2, perhaps reflecting cell line-specific differences (64). Regardless, prolonged exposure of rat and human fibroblasts to high concentrations of HU results in irreversible cell cycle arrest irrespective of p53 status (7).

Inhibition of replication has been associated with up-regulation of homologous recombination (HR) and the formation of foci containing Rad51, a key protein in HR repair (HRR) (56, 70). The increase in HR following replication inhibition correlates with increased DNA DS breaks. At least in some models, HRR appears to be the major mechanism to repair DNA DS breaks at stalled replication forks (1). Not surprisingly, HRR-deficient cell lines are hypersensitive to HU, as measured by clonogenic survival (56, 70), and Rad51 overexpression promotes clonogenic survival in response to HU or etoposide (34), but not {gamma}-irradiation (47). Interestingly, survival following HU treatment is more than 1,000-fold better than following a {gamma}-irradiation dose that produces similar amounts of DS breaks (56), suggesting that replication-associated DS breaks are more efficiently repaired or that DS breaks produced by HU are different from the DS breaks produced by {gamma}-irradiation.

p53 potently inhibits HR through repression of Rad51 expression and inhibitory interactions with several components of the HR machinery (Rad51, Rad54, and replication protein A) at sites of DNA damage or stalled replication forks (46, 51, 69, 83). Disruption of p53 leads to substantial up-regulation of HR independent of p53's ability to induce G1 cell cycle arrest (71). Null p53 mutation also leads to increased replication-associated DNA DS breaks, although p53 status did not affect clonogenic survival following continuous exposure to replication inhibitors (46).

The majority of reports on the effects of replicational stress on cell proliferation and the role of p53 in this response have used acute replicational stress at levels far exceeding replication inhibition typically experienced by cells in vivo. In addition, these studies have mostly employed immortalized cell lines, which might possess defects in damage response pathways. In this study, we examined the role of p53 in cellular responses to chronic exposure to low, clinically relevant levels of replication inhibitors. In contrast to previous studies, which concluded that replicational-stress-induced arrest does not require p53, we demonstrate that p53 and its target, p21, mediate senescence-like arrest induced by chronic replicational stress.


   MATERIALS AND METHODS
Top
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Cell lines and cell cultures. Human foreskin fibroblasts (HFF) (from the ATCC), REF52 cells, and Phoenix-Ampho cells (from G. Nolan, Stanford University) were cultured in high-glucose Dulbecco's modified Eagle's medium in the presence of 10% fetal bovine serum. MCF10A cells (from Cambrex) were cultured in Ham's 12-Dulbecco's modified Eagle's medium containing 5% equine serum, 10 µg/ml insulin, 500 ng/ml hydrocortisone, 20 ng/ml epithelial growth factor, and 100 ng/ml cholera toxin. For proliferation assays, 2 x 105 cells (HFF and MCF10A) or 1 x 105 cells (REF52) were plated per 6-cm dish in medium with or without drugs. After 3 to 4 days of culture (when the cells were still subconfluent), the cells were trypsinized, counted, and replated at the original density. For colony formation assays, 100 to 2,000 cells were seeded per 6-cm or 10-cm plate and cultured in drug-free medium for 7 to 10 days, fixed in a solution of 12.5% acetic acid and 30% methanol for 15 min, and stained with 0.1% crystal violet solution in water for 4 h. Colonies with approximately 50 or more cells were counted, and the percentage of plated cells that formed colonies was determined.

Retroviral constructs and infections. Mouse stem cell virus (MSCV)-internal ribosome entry site (ires)-green fluorescent protein (GFP) constructs expressing dominant-negative (DN) p53 (with the Pro polymorphism at position 72) and DDp53 (the "dimerization domain" of p53) have been previously described (6). Short hairpin RNA (shRNA) constructs targeting p16Ink4a (in pSIN-puro vector) and p53 (in MSCV-miR30-puro vector) were the gift of S. Lowe (Cold Spring Harbor Laboratory). The shRNA construct targeting p21CIP1 (in pSuper-Retro vector) was the gift of R. Bernards (The Netherlands Cancer Institute). Additional shRNA-expressing pSuper-Retro constructs were generated using algorithms by Oligoengine. The shRNA constructs used in our experiments targeted the following human sequences: p21 (GACCATGTGGACCTGTCAC), p16 (GAGGAGGTGCGGGCGCTGC), Rad51 (AGAGCAGTGTGGCATAAAT [construct A], TGAAGCCAAAGCTGATAAA [construct B], and AGACCCAGCTCCTTTATCA [construct C]), p53 (TGGAGGATTTCATCTCTTGTAT), and luciferase (CGTACGCGGAATACTTCGATT). MSCV-ires-GFP and pSuper-Retro retroviruses were prepared by transient transfection of Phoenix-Ampho packaging cells, together with pCL-AMPHO helper plasmid, using ExGen 500 transfection reagent (Fermentas) following previously described protocols. Transduction of cells with retroviruses was performed using three rounds of infection at 3-h intervals with 50%-diluted virus-containing medium supplemented with 6 µg/ml Polybrene (Sigma). Over 95% infection efficiencies with MSCV-ires-GFP viruses were achieved in HFF and REF52. Transduced MCF10A cells were sorted using a MoFlo (Cytomation, Fort Collins, CO) cell sorter for GFP-expressing cells to achieve >98% homogeneous populations. Cells infected with pSuper-Retro viruses were selected and kept in the presence of 2.5 µg/ml puromycin.

Western blotting. Cell extracts were prepared in gel shift lysis buffer (50 mM HEPES, pH 7.9, 250 mM KCl, 0.1% NP-40, 0.1 mM EGTA, 0.1 mM EDTA, 10% glycerol). Western blotting procedures were performed following Millipore protocols. The antibodies used were anti-p21 (Santa Cruz sc-397), anti-p16 (Neo Markers Ab-6), anti-total p53 (Novo Castra CM5 and Santa Cruz sc-6243), anti-Rad51 (UBI 3C10), anti-phospho-p53 (Ser15), anti-phospho-p53 (Ser20), anti-phospho-CHK1 (Ser345), anti-phospho-CHK2 (Thr68) (Cell Signaling), anti-tubulin (NeoMarkers Ab-2), anti-CHK1 (Santa Cruz G4), total Rb (a 50/50 mixture of PharMingen 14001A and Santa Cruz sc-102 and IF-8), anti-PIG-3 (Exalpha Biologicals), anti-BAX (Santa Cruz P-19), anti-ß-actin (Santa Cruz sc-1616), and anti-lamin A/C (Santa Cruz sc7292). The secondary antibodies used were anti-mouse antibody-horseradish peroxidase (HRP), anti-rabbit antibody-HRP, and anti-goat antibody-HRP (Bio-Rad). The blots were developed using Immobilon substrate (Millipore).

Immunocytochemistry and senescence-associated ß-galactosidase analysis. Immunofluorescence detection of {gamma}H2A.X foci was as described previously (36), using primary mouse monoclonal antibody (Upstate; JBW301) and secondary antibody conjugated to Alexa Fluor-488 (Molecular Probes). Senescence-associated ß-galactosidase activity was detected as described previously (27); DAPI (4',6'-diamidino-2-phenylindole) staining was used to distinguish individual cells. Images were obtained with a BX51 microscope (Olympus, Melville, NY) and a Penguin 600CL camera (Pixera, Los Gatos, CA). Photoshop 7.0 (Adobe Systems, Mountain View, CA) was used to process the images.

dNTP pool extraction and analysis. Subconfluent cells grown in 15-cm plates were drained of medium and then incubated in 8 ml of ice-cold 80% methanol at –20°C for 1 h to extract nucleotide pools. The extracts were heated for 3 min in a boiling-water bath, followed by centrifugation for 20 min at 17,000 x g. The supernatant was transferred to a fresh tube and dried under a vacuum. The residue was dissolved in sterile water and stored at –20°C for later analysis. Analysis of the dNTP pools in each extract was carried out by the polymerase-based method as described previously (79), with some modifications. Reaction mixtures (25 µl) contained 100 mM HEPES buffer (pH 7.5), 10 mM MgCl2, 0.1 unit of Escherichia coli Pol I Klenow fragment (United States Biochemical), 0.25 µM oligonucleotide template, 5 µg of bovine serum albumin (New England Biolabs), and 1.25 µCi (1 Ci = 37 GBq) of [3H]dATP (Amersham Pharmacia Biosciences) or [3H]dTTP (Perkin-Elmer Life Sciences). Incubations were carried out for 45 min at 37°C.

BrdU incorporation assays and cell cycle analysis. Cells were cultured in the presence of 10 µM bromodeoxyuridine (BrdU) (Sigma) for 24 h and then harvested and fixed in 70% ethanol at 4°C for at least 1 h. BrdU incorporation was detected by using fluorescein isothiocyanate-linked anti-BrdU (PharMingen or Dako) according to the manufacturer's protocols, except that 0.2 mg/ml pepsin was added at the HCl denaturation step to improve antibody accessibility. After being washed, the cells were resuspended in 300 µl of 10 µg/ml propidium iodide (Roche) in 1x phosphate-buffered saline, incubated overnight, and analyzed by flow cytometry. For cell cycle distribution assays, cells were fixed in 70% ethanol, incubated at 4°C for at least 1 h, and then subjected to propidium iodide staining as described above. Fluorescence was detected with a Coulter Epics XL (Beckman Coulter) or FACSCalibur (Becton Dickinson) cytometer.

Statistical analysis. Statistical analysis was performed with Prizm 4 software (Graphpad). For determining P values, two-tailed Student tests were performed. Pearson correlation was used to determine correlation coefficients.


   RESULTS
Top
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Disruption of p53 activity prevents proliferative arrest by low levels of DNA synthesis inhibitors. For our initial experiments, we used the rat fibroblast cell line REF52. Although REF52 cells are immortalized, they possess intact Rb and p53 pathways, similar to primary cells (77). We cultured REF52 cells in the continuous presence of the ribonucleotide reductase inhibitor HU at 100 to 150 µM and the replicative polymerase inhibitor aphidicolin (APH) at 200 nM. These concentrations were about 10-fold lower than those typically used to fully arrest proliferation, and the concentration of HU was similar to that present in the sera of individuals treated with HU in the clinic (see Discussion). To inhibit p53, we stably expressed either DNp53 or DDp53 using MSCV. The highly prevalent human DNp53 tumor mutant has the R175H mutation, leading to inhibition of DNA binding and potent DN activity toward p53 and, to some extent, p73 (5, 28, 39, 80). DDp53 encodes only the multimerization domain of p53 (amino acids 302 to 390) and is thought to inhibit only p53 (not p73) (32), as DNp53 and p73 interact via their DNA binding domains (38). For controls, we transduced cells with empty vector. Both DN forms of the protein led to accumulation of similarly high total p53 protein levels (see Fig. S1A in the supplemental material; for cells expressing DDp53, stabilization of endogenous p53 is evident).

Given that HU-induced cell cycle arrest has been shown not to require p53 (7, 31, 50) and that p53 is required for preventing catastrophic mitosis and for survival following HU- and APH-mediated inhibition of DNA synthesis (85), one would not expect p53 inhibition to improve long-term cellular proliferation in the presence of HU or APH. We found that continuous exposure of REF52 cells to HU or APH resulted in the inhibition of proliferation; after 8 to 10 population doublings, the cells completely ceased proliferating (Fig. 1B and C). On the other hand, and in contrast to the prevailing notion that p53 does not mediate HU-induced arrest, cells with p53 disrupted by expression of DDp53 maintained constant proliferation rates in the presence of HU or APH, although proliferation was reduced relative to untreated cells (Fig. 1B and C). Of note, p53 status did not significantly affect the proliferation of cells in the absence of replicational stress (Fig. 1A). Disruption of p53 via expression of DNp53-R175H overcame drug-induced proliferation arrest in a manner indistinguishable from DDp53 expression (see Fig. S1B in the supplemental material). Following incubation with APH or HU, most of the vector-transduced cells displayed the flattened, enlarged morphology characteristic of senescent cells, while most DDp53- or DNp53-expressing cells retained normal morphology (see Fig. S2 in the supplemental material). Steady proliferation rates were maintained by DDp53-expressing cells over 60 days of monitoring, and during this time, the culture expanded by more than 50 population doublings without obvious signs of decline (data not shown). Finally, we did not observe any changes in cell death with drug treatment in either control or p53-disrupted cultures.


Figure 1
View larger version (17K):
[in this window]
[in a new window]

  		FIG. 1. Inhibition of p53 in rat fibroblasts improves proliferation under chronic replicational stress. REF52 fibroblasts expressing either DDp53 or vector (V) were cultured in drug-free medium (A) or with 150 µM HU (B), 200 nM APH (C), or 200 nM VP16 (D). The cells were counted every 3 or 4 days upon passage. The cumulative numbers of total population doublings are plotted (each culture in triplicate), with standard deviations indicated (obscured when the standard deviation is very small).

 
Both HU and APH treatments stall replication forks, increasing the presence of single-stranded DNA that triggers activation of the ATR-CHK1 checkpoint response. Although a fraction of the stalled replication forks may collapse, creating DNA DS breaks (56, 70), the majority of replication forks are likely to be resolved without collapse, and DNA DS breaks should be rare in primary cells exposed to replicational stress. In contrast, topoisomerase inhibitors, such as the topoisomerase II inhibitor etoposide (VP16), which is a commonly used antineoplastic drug, induce single-stranded and DS DNA breaks in the vicinity of replication forks (57), thus creating replicational stress and DNA damage. We asked whether p53 disruption could result in a long-term proliferative advantage when replicational stress was coupled with substantial DNA damage. We cultured REF52 cells in the presence of 200 nm VP16, a concentration that immediately impaired but did not completely prevent expansion of control cells. As observed for HU and APH, expression of DDp53 allowed cells to avoid proliferative arrest and to continue to expand in the presence of VP16 (Fig. 1D; see Fig. S2 in the supplemental material). Thus, p53 is required for the efficient arrest of proliferation induced by continuous exposure to drugs that impair DNA synthesis, either with or without high-level DNA strand breaks.

p53 is required for the accumulation of senescent cells during chronic replicational stress. Because vector-expressing cells completely halted proliferation under continuous exposure to low levels of replicational stress, exhibiting changes in cell morphology characteristic of senescence, we asked whether the arrest was permanent. Indeed, when released into drug-free medium after 2 weeks of being cultured with HU or APH, vector cells lost the ability to form colonies and to incorporate BrdU despite maintaining viability (Fig. 2A; not shown for HU), confirming the permanent nature of the arrest. In contrast, a substantial fraction of cells expressing DDp53 were capable of incorporating BrdU and forming colonies, consistent with persistent expansion of cultures in the presence of replicational stress. In fact, the kinetics of senescent-cell accumulation mirrored the kinetics of inhibition of cell proliferation (correlation coefficients, r = –0.9529 with P < 0.01 for vector cells and r = –0.8048 with P < 0.05 for DDp53-expressing cells), suggesting that accumulation of senescent cells was the cause of the observed reduction in proliferation rates and the eventual proliferation halt (Fig. 2B). Notably, after the initial decline, we did not observe changes in the percentage of cells capable of forming colonies in drug-treated DDp53-expressing cultures, suggesting a lack of selection for additional mutations. Although p53 disruption does not prevent arrest in all cells, a sufficient fraction (about half) of cells are capable of maintaining proliferation to provide a substantial net expansion of the population.


Figure 2
View larger version (33K):
[in this window]
[in a new window]

  		FIG. 2. Disruption of p53 antagonizes the induction of permanent proliferation arrest under mild replicational stress. (A) REF52 cells expressing either DDp53 (DD) or vector (V) were cultured in the presence of 200 nM APH or no drug (n/tr) for 2 weeks and then released into APH-free medium containing 10 µM BrdU for 24 h. Aliquots were plated for colony formation assays prior to adding BrdU. Following incubation with BrdU, the cells were harvested, stained with propidium iodide and fluorescein isothiocyanate-conjugated anti-BrdU antibody, and subjected to flow cytometric analysis. Average values for clonogenic survival and percentages of cells incorporating BrdU are plotted for independent duplicate cultures. Representative flow cytometry profiles are shown with gates for BrdU-positive cells. (B) REF52 cells expressing either DDp53 or vector were cultured as for Fig. 1 in the presence of 200 nM APH or no drug for 7 days (in triplicate). At each passage, aliquots of cells were plated for colony formation assays. The number of population doublings and percent clonogenic survival are plotted for each passage. The difference between vector- and DDp53-expressing cells became statistically significant (P < 0.05) by day 7. (C) After being cultured in the presence of 200 nM APH, 100 µM HU, or no drugs, vector and DDp53 cells were released into drug-free medium for 48 h, at which point they were lysed and expression of cyclin A2 and tubulin was analyzed by immunoblotting analysis. (D) REF52 cells expressing either DDp53 or vector were cultured in the presence of 150 µM HU for 10 days or in the presence of 2 mM HU for 2 or 3 days and then plated for colony formation assays. Average values of independent duplicate cultures are shown. Standard deviation is indicated (obscured when very small). "*" indicates a P value from 0.01 to 0.05; "**" indicates a P value from 0.001 to 0.01.

 
Another hallmark of senescence is the irreversible inhibition of E2F-dependent transcription (63). Indeed, vector control cells failed to up-regulate cyclin A2 expression 48 h post-drug release after being cultured in the presence of APH or VP16 for 2 weeks (Fig. 2C). In contrast, DDp53-expressing cells displayed significant cyclin A2 levels, albeit reduced compared to those of cells that had not been subjected to chronic replicational stress. Similarly to DDp53, expression of DNp53 was capable of maintaining clonogenic survival following prolonged exposure to APH or VP16 (see Fig. S1C in the supplemental material).

Since our results were in contrast with a report demonstrating p53 independence of permanent proliferative arrest resulting from exposure to high levels of replicational stress (using 2 mM HU) in REF52 cells (7), we tested the effect of p53 disruption on cell responses to high concentrations of HU (2 mM) or APH (10 µM). Consistent with the observation of Borel et al., disruption of p53 failed to prevent permanent proliferation arrest by high levels of replicational stress (Fig. 2D and data not shown for APH). Therefore, the p53 dependence of the arrest is specific to low levels of replicational stress.

Potentially, p53 disruption could provide resistance to low levels of replicational stress by reversing the effects of HU or APH on dNTP levels or DNA synthesis. To address these possibilities, we measured dNTP levels in control and DDp53-expressing cultures grown in the presence of low concentrations of APH or HU, or no drug, at a point when vector cells were still proliferating (after 5 days of treatment). We found that treatment with HU or APH induced similar dNTP perturbations irrespective of p53 status (Fig. 3A). One hundred micromolar HU reduced dNTP levels up to twofold, while treatment with 200 nM APH increased dNTP levels (likely due to decreased dNTP consumption) both in control cells and in cells expressing DDp53. However, control cells ceased proliferating with continued passages, while p53 mutant cells were capable of continuous proliferation. In comparison, brief (5-h) treatment with high concentrations of HU (2 mM) resulted in more severe depletion of dNTPs (more than 10-fold for dATP) in p53-proficient and -inhibited cells (Fig. 3B). We also noticed that the expression of DDp53 in untreated cells resulted in modest, but consistent, reductions in the levels of all four dNTPs (Fig. 3A and B), despite increases in the fraction of cells in S phase (data not shown). These reductions in dNTP pools could reflect decreased p53-dependent expression and activity of ribonucleotide reductase subunits (84, 94). Interestingly, a recent paper has suggested another potential mechanism for such a decrease: the p53 target TIGAR up-regulates the pentose phosphate pathway, for which intermediate metabolites are important precursors of DNA synthesis (84), and thus, disrupting p53 function might lead to down-regulation of dNTP synthesis. Finally, S-phase progression was similarly slowed following treatment with both high and low concentrations of HU or APH for 24 h irrespective of p53 status (data not shown). Therefore, the effect of p53 disruption cannot be attributed to reduced effectiveness of these replication inhibitors on their direct targets, and the inhibition of p53 does not appear to counteract the direct consequences of replicational stress on the cell cycle.


Figure 3
View larger version (35K):
[in this window]
[in a new window]

  		FIG. 3. p53 disruption does not restore dNTP levels. (A) REF52 cells were grown for 5 days in the presence of 100 µM HU or 200 nM APH (or no drug [n/tr], as indicated) and then harvested for dNTP extraction. dNTP levels were determined as described in Materials and Methods. All plates were subconfluent at harvest. (B) REF52 cells were grown for 5 h in the presence of 2 mM HU (or no drug, as indicated) and then harvested for dNTP measurements as in panel A. Standard deviations of duplicate samples are indicated for both panels A and B (obscured when very small).

 
p53 is also required for chronic replicational-stress-induced permanent arrest of human cells. Although REF52 cells provide a convenient experimental system, some p53-dependent responses may differ in humans, since human cells possess more elaborate tumor suppression mechanisms. We therefore sought to determine how p53 disruption affects responses to low levels of replicational stress in human cells. As for REF52 cells, primary HFF transduced with vector control ceased proliferation after prolonged exposure to low levels of replicational stress induced by APH, HU (Fig. 4A), or VP16 (data not shown), while stable expression of DDp53 allowed cells to maintain proliferation for at least 2 months (Fig. 4A and data not shown). Similar results were obtained using DNp53 (data not shown).


Figure 4
View larger version (36K):
[in this window]
[in a new window]

  		FIG. 4. Proliferation arrest induced by mild replicational stress is p53 dependent in human fibroblasts. HFF (starting at passage 10) expressing either DDp53 (DD) or vector (V) cultured with 150 µM HU or 150 nM APH were counted upon passage every 3 or 4 days. Cumulative population doublings are plotted (each culture in duplicate), with standard deviations indicated (obscured when very small). n/tr, not treated. (B) HFF expressing V or DD were cultured as for panel A in the presence of either 150 µM HU or no drug for 12 days and then stained for senescence-associated ß-galactosidase (SA-ß-Gal) activity. At least 200 cells were counted from duplicate cultures for each condition. "**" indicates a P value from 0.001 to 0.01. (C) HFF expressing V, DD, or an shRNA targeting p21 were cultured as for panel A in the presence of either 200 nM VP16, 150 nM APH, or no drug for 21 days and then analyzed for cyclin A2 and tubulin expression by immunoblotting. (D) HFF expressing MSCV-ires-GFP vector or shRNA against p53 were harvested for Western blot analysis for p53, p21, and lamin A/C proteins. (E) HFF expressing V, DDp53, or p53 shRNA were cultured in drug-free medium or medium containing 150 µM HU or 150 nM APH (each culture in duplicate). The cells were counted every 3 or 4 days upon passage. The cumulative numbers of total population doublings are plotted.

 
Similarly to REF52 cells, most of the vector-transduced cells developed flattened senescence-like morphology upon continuous exposure to the replication inhibitors, while a substantial fraction of cells expressing DDp53 maintained normal morphology (Fig. 4B). While we could not analyze clonogenic survival of HFF due to poor proliferation at clonal densities, we found that following cessation of cell proliferation, vector-expressing cells lost the ability to incorporate BrdU and failed to resume proliferation after release into drug-free medium (data not shown). Also, at the time of proliferation arrest, almost all of the vector-expressing cells were positive for senescence-associated ß-galactosidase activity at pH 6, a commonly used marker for distinguishing senescent cells (27), while less than half of DDp53-expressing cells were positive for the marker (Fig. 4B). Thus, inhibition of p53 substantially reduces, but does not completely abrogate, senescence induced by chronic replicational stress in human fibroblasts. As expected, HFF vector control cells were unable to upregulate cyclin A2 levels following release into drug-free medium after 2 weeks of incubation with 200 nM APH or 150 nM VP16 (Fig. 4C). In contrast, disruption of p53 via DDp53 expression allowed robust activation of cyclin A2 expression despite drug treatment. Finally, the effect of the expression of DN forms of p53 on proliferative arrest induced by low levels of replicational stress could be recapitulated by stable expression of an shRNA against p53 (Fig. 4D and E), arguing that the effects of DN forms of the protein are specific to p53 inhibition.

Epithelial cancers constitute the majority of human malignancies, and p53 disruption is frequent in carcinomas. We therefore tested whether similar p53 dependence in induction of senescence-like arrest by low levels of replicational stress could be observed in human epithelial cells. For this purpose, we used MCF10A cells, a spontaneously immortalized diploid breast epithelial cell line that was originally derived from a subcutaneous mastectomy (82). MCF10A cells possess wild-type p53 genes and normal p53-dependent cell cycle arrest in response to DNA damage (33, 82). Retroviral expression of DNp53 in the MCF10A cells led to levels of p53 protein similar to those in T47D breast carcinoma cells expressing an endogenous DNp53 mutant (2) (Fig. 5C; the MCF7 breast carcinoma cell line that had intact p53 [17] was used as a control). As for human and rat fibroblasts, continuous exposure of MCF10A cells to low levels of replicational stress halted proliferation and reduced clonogenic survival, while disruption of p53 via expression of DNp53 prevented the complete arrest of proliferation and improved clonogenic survival (Fig. 5A and B). Notably, although HU substantially inhibited clonogenic survival in DNp53-expressing cells, DNp53-expressing cells still displayed about 10-fold-higher clonogenic survival than vector controls. Importantly, the ability of DNp53 to antagonize permanent proliferation appeared to be induced specific for replicational stress and does not reflect a general p53 dependence for cellular responses to low levels of DNA damage, as DNp53 expression failed to provide a survival advantage when cells were treated with the radio-mimicking drug bleomycin at concentrations that allowed a fraction of cells to maintain clonogenic outgrowth (Fig. 5B).


Figure 5
View larger version (22K):
[in this window]
[in a new window]

  		FIG. 5. Proliferation arrest induced by mild replicational stress is p53 dependent in human epithelial cells. (A) MCF10A cells expressing vector (V) or DNp53 were cultured in the presence of either 150 µM HU or with no drug (n/tr). Population doublings are plotted, and standard deviations are indicated. (B) MCF10A cells grown as for panel A in the presence of 150 µM HU for 3 weeks or in the presence of 0.2 µg/ml bleomycin (Bleo) for 11 days were plated for colony formation assays. Without drug treatment, DNp53 MCF cells exhibited about 50% more colonies than vector-expressing MCF cells (the average plating efficiencies were 190% and 130% for untreated DNp53 and vector MCF cells, respectively, indicating that some colonies might be derived from cells emigrating from other colonies), and thus the results are shown as percent colony formation relative to untreated cells transduced with the same virus. "**" indicates a P value from 0.001 to 0.01. (C) p53 expression in V MCF10A, DNp53 MCF10A, MCF7, and T47D breast cancer cells (with endogenous mutated p53) was determined by Western blotting.

 
Checkpoint induction by low levels of replicational stress. Given the p53 dependence of cellular responses to low levels of replicational stress, we asked whether this treatment regimen induces changes in p53 phosphorylation and protein levels. Consistent with observations made by Gottifredi et al. using 2 mM HU (31), we found that 150 µM HU induced activating p53 phosphorylation at serine 15 as early as 2 h of treatment (Fig. 6A), suggesting activation by ATR (87), and phosphorylation levels increased with prolonged incubation. Similar p53 phosphorylation was observed with 200 nM APH (data not shown). However, the extent of p53 activation was substantially reduced compared to phosphorylation resulting from treatment with 2 mM HU, a concentration that arrests S-phase progression and induces p53-independent cell cycle arrest with prolonged exposure (7). We could also detect low levels of activating phosphorylation of p53 at serine 20 (Fig. 6B), but only after 6 days of culture with HU. Total p53 protein levels changed little, even after prolonged exposure to low concentrations of HU (Fig. 6B) or APH (data not shown).


Figure 6
View larger version (66K):
[in this window]
[in a new window]

  		FIG. 6. p53 and CHK1 are activated in response to low levels of replicational stress. (A and B) Subconfluent HFF (passage 12) were grown in the presence of 150 µM or 2 mM HU for the indicated periods of time and then harvested and analyzed by Western blotting with antibodies against p53 phospho-Ser15, p53 phospho-Ser20, total p53, or tubulin. (C) HFF expressing either DDp53 or vector control were cultured in the presence of 150 µM HU or 150 nM APH for 0, 24, or 72 h (0, 1, or 3 days) or 72 h followed by 24-h release into drug-free medium (3' days) and then harvested and analyzed with antibodies against total CHK1, CHK1 phospho-Ser345, CHK2 phospho-Thr68, or actin. Vector-expressing cells treated with 2 mM HU overnight or irradiated with 600 rad served as positive controls. The arrow indicates the CHK2 phospho-Thr68-specific band; a nonspecific band migrated just above the CHK2 phospho-Thr68 band. (D) HFF were cultured in the presence of 200 nM APH for 3 weeks or in the presence of 10 µg/ml bleomycin for 3 days (when net proliferation of cells ceased) and then plated on microscope slides in drug-free medium for 36 h, fixed, and stained with {gamma}H2AX antibody and counterstained with DAPI. Representative images are shown. The percentage of cells exhibiting {gamma}H2AX foci, scored for at least 150 cells per condition, is indicated in the lower left corner of each image. The arrows point to examples of nuclei with {gamma}H2AX foci.

 
Given that p53 is phosphorylated at serines 15 and 20 by activated checkpoint pathways and that blocking replication primarily induces the ATR-CHK1 pathway (21, 95), we analyzed levels of activated CHK1 phosphorylated at Ser345. We observed CHK1 Ser345 phosphorylation following 24 h of HU and, consistently to a lesser extent, APH treatment in control cells (Fig. 6C). Cells expressing DDp53 also exhibited clear CHK1 Ser345 phosphorylation following HU and APH treatments, indicating that the failure of these treatments to induce permanent proliferation arrest in p53-inhibited cells does not result from reduced checkpoint activation. Notably, exposure to HU and APH treatments led to down-regulation of total CHK1 levels independently of p53 status, and this down-regulation correlated with the decrease in Ser345 CHK1 from 24 to 72 h.

It has been proposed that irrespective of the initial stress, induction of senescence requires acquisition of unrepaired DNA DS breaks, and therefore, cellular senescence can be regarded as a permanent DNA damage response state (89). Since replication stress stalls replication forks and some of the stalled forks might collapse, producing DNA strand breaks, we tested whether the vector-arrested cells that had been driven to the state of senescence-like arrest by low levels of replicational stress displayed evidence of unrepaired DNA DS breaks. We could not detect increased activating phosphorylation of CHK2 on Thr68 in control cells subjected to low concentrations of HU and APH, suggesting that no substantial DNA DS breaks were generated (Fig. 6C). Interestingly, we observed some basal CHK2 phosphorylation in DDp53-expressing cells, which did not increase upon treatment with low levels of HU or APH.

We analyzed cells for the presence of {gamma}H2A.X foci, a widely used marker of DNA DS breaks, although {gamma}H2A.X foci are also detected during replicational stress in the absence of detectable DNA strand breaks (91). We observed increased levels of nuclear {gamma}H2A.X in response to culture in low-dose HU or APH for 2 days (see Fig. S3 in the supplemental material), but drug release for 1 day returned {gamma}H2A.X to near basal levels. Disruption of p53 led to increased basal and induced levels of {gamma}H2A.X (see Fig. S3 in the supplemental material). Since DDp53-expressing cells uniformly exhibit {gamma}H2A.X staining and yet are proliferating, and since DNA strand breaks should be incompatible with long-term proliferation, we believe that {gamma}H2A.X detects replicational stress in these cells, not DNA DS breaks. Although in cells driven into senescence by 3 weeks of culture in low levels of HU or APH we could detect an increase in the fraction of cells displaying {gamma}H2A.X foci (similar to foci induced by the radio-mimicking drug bleomycin), upon release from the drugs the majority of these permanently arrested cells were free of detectable foci (Fig. 6D). We therefore conclude that accumulation of unrepaired DNA DS breaks is not a major mechanism of senescence induction by replicational stress. Nonetheless, the absence of detectable unrepaired DNA damage in senescent cells could still be consistent with DNA damage driving cells into permanent arrest but being subsequently repaired.

Recent reports have demonstrated that oncogene-induced senescence is dependent on DNA damage signaling induced by abnormalities in replication and that inhibition of DNA damage checkpoint signaling can overcome senescence (4, 26, 58). We therefore decided to test whether inhibition of checkpoint signaling could overcome the permanent proliferation arrest caused by low levels of replicational stress. First, we tested the effect of caffeine, a commonly used inhibitor of ATR and ATM protein kinases (72). Although caffeine treatment was reported to rescue cells from oncogene-induced senescence (4), we found that concentrations of caffeine that had minimal effect on cellular replication in the absence of replicational stress completely blocked replication and markedly compromised clonogenic survival of REF52 cells in the presence of 100 µM HU (see Fig. S4A in the supplemental material). Knockdown of Chk2 also did not prevent HU-mediated replicational senescence (see Fig. S4B in the supplemental material) Thus, in marked contrast to oncogene-induced senescence, replicational-stress-induced senescence is potentiated by inhibition of checkpoint signaling.

The cyclin-dependent kinase (CDK) inhibitor p21 is the critical p53 target mediating permanent arrest in response to replicational stress. p21 is an important p53 target regulating the G1-S checkpoint (14, 15, 25, 35, 93). Considering that p21 is thought to be the essential p53 target in senescence induced by telomere erosion (13), we decided to examine p21 levels after chronic exposure to replicational stress. We found that treatment with both APH and VP16 induced p21 accumulation by 3 days of treatment, and expression of DDp53 prevented p21 accumulation (Fig. 7A). Similar results were obtained for HU (data not shown). G1-to-S-phase progression requires hyperphosphorylation of Rb by CDKs, relieving repression of E2F-dependent transcription and promoting entry into S phase. Consistent with the role of p21 as a CDK inhibitor, up-regulation of p21 in cells treated with APH and VP16 correlated with decreased hyperphosphorylated, inactive forms of Rb, while phosphorylation of Rb was maintained in the presence of DDp53 (Fig. 7A).


Figure 7
View larger version (42K):
[in this window]
[in a new window]

  		FIG. 7. p21 is an essential p53 target mediating arrest in response to APH. (A) HFF expressing DDp53 or vector control were cultured in the presence of 150 nM APH or 200 nM VP16 for the indicated number of days, harvested, and analyzed by Western blotting with antibodies against Rb, p21, or actin. (B) HFF expressing pSuper-Retro (pSR) vector or pSR encoding anti-p21 shRNA (p21shRNA) were treated with 2 mM HU for 24 h to induce the expression of p21 and then harvested and analyzed by Western blotting using antibodies against p21 or actin. Untreated pSR-expressing cells were used as a control for basal p21 expression levels. (C) HFF expressing pSR, pSR-sh-p21, or pSR plus DDp53 cultured in the presence of 150 nM APH, 200 nM VP16, or no drug were counted every 3 or 4 days upon passage. The numbers of population doublings are plotted (each culture in duplicate), with standard deviations indicated (obscured when very small). (D) Cells cultured as for panel A in the presence of 150 nM APH were harvested and analyzed with antibodies against p53, PIG3, BAX, cyclin A2 (cyclin A2 was used as a marker for proliferation arrest), and tubulin.

 
To test the functional significance of preventing p21 accumulation, we knocked down p21 expression by stably expressed shRNAs. p21 knockdown efficiently prevented up-regulation of p21 by HU (Fig. 7B) but did not affect proliferation under normal culturing conditions (Fig. 7C). However, when cells were cultured with APH, p21 knockdown prevented the induction of proliferation arrest, and the effect of p21 disruption was indistinguishable from the effect of p53 disruption (Fig. 7C). As expected, knockdown of p21 did not prevent activating phosphorylation of p53 at serine 15 and upregulation of the p53 target PIG3 (see Fig. S5 in the supplemental material). Interestingly, p21 disruption only partially and temporarily prevented the inhibition of proliferation induced by VP16 treatment, suggesting that other p53 targets besides p21 are involved in the ability of DDp53 to antagonize arrest induced by replicational stress associated with DNA strand breaks. Importantly, cells expressing the shRNA against p21 maintained cyclin A2 expression following treatment with APH, but not VP16 (Fig. 4C). The ability of p21 knockdown to prevent APH, but not VP16, arrest was confirmed using clonogenic assays with two independent shRNA sequences targeting rat p21 in REF52 cells (data not shown). Thus, p21 is the key target of p53 required for proliferation arrest induced by replicational stress but only partially accounts for the ability of p53 to arrest cells in response to VP16.

It has been previously reported that during replicational senescence (induced by critically short telomeres) in human fibroblasts, p53 is preferentially recruited to promoters of genes that regulate cell cycle arrest but not apoptosis (40). We therefore analyzed the expression of p53 target genes implicated in regulation of apoptotic response. We found that treatment with 150 nM APH led to modest p53-dependent increases in protein levels of the proapoptotic p53 targets BAX and PIG3 (62, 66) (Fig. 7D). The modest induction of these targets appears insufficient to promote apoptosis, as we did not observe cell death in response to replicational stress (data not shown).

Senescence induced by low levels of replicational stress is p16INK4A independent. Induction of senescence by shortened telomeres appears to occur through independent p53-p21 and p16-Rb pathways (36, 41), with the relative contributions of these pathways varying among different cell types (16). Because disruption of p53 did not prevent induction of senescence by low levels of replicational stress in all cells, we hypothesized that induction of p16 might be inducing permanent proliferation arrest in a subset of DDp53-expressing cells. Indeed, exposing cells to chronic mild levels of APH or HU increased p16 protein levels, and this induction was substantially increased in cells expressing DDp53 (Fig. 8A).


Figure 8
View larger version (27K):
[in this window]
[in a new window]

  		FIG. 8. Permanent arrest induced by chronic replicational stress does not require p16. (A) HFF expressing DDp53 (DD) or vector (V) were cultured in the presence of 150 µM HU or 200 nM APH for the indicated periods of time and then harvested and analyzed by Western blotting with antibodies against p16 or tubulin. (B) DDp53-expressing cells stably expressing pSR vector or pSR encoding shRNA against p16 (p16shRNA) were cultured with no drug or with 150 µM HU to induce p16 expression for 3 days and then harvested and analyzed for p16 and tubulin protein levels. (C) HFF expressing DDp53 and either SR or p16shRNA were cultured in the presence of 150 µM HU or no drug and were counted upon passage. The numbers of population doublings are plotted (each culture in duplicate), with standard deviations indicated.

 
To test the functional significance of p16 induction, we knocked down the expression of p16 using stably expressed shRNA (Fig. 8B). Surprisingly, preventing p16 up-regulation in vector- or DDp53-expressing cells had no significant effect on cellular proliferation during HU (Fig. 8C) or APH (data not shown) treatment. The results were confirmed with an independent shRNA construct (data not shown). We therefore conclude that senescence-like arrest of HFF induced by chronic replicational stress does not require p16 and that p16 induction is not the cause of permanent proliferation arrest in a subpopulation of DDp53-expressing cells.

It should be noted, however, that given various degrees of contribution of p53-p21 and p16-Rb pathways to the induction of senescence in different cells, we cannot exclude contributions of p16 to senescence induced by replicational stress in cells other than HFF. Nonetheless, consistent with the lack of a requirement for the p16-Rb pathway for chronic-stress-induced arrest, downregulation of Rb expression using retrovirus- expressed shRNAs in HFF also failed to prevent induction of senescence-like arrest induced by HU, APH, or VP16 (see Fig. S6 in the supplemental material; data not shown).

Upregulation of HRR is critical for the ability of p53 mutation to overcome VP16- but not APH-induced arrest. The ability of p53-deficient cells to maintain long-term proliferation suggests that if DNA damage is indeed the cause of the permanent arrest by chronic low levels of replicational stress, p53 mutant cells need to repair DNA DS breaks before the cells enter mitosis. Given the reported role of p53 in inhibiting HR and substantial increases in Rad51 protein levels in DDp53-expressing cells that we observed (Fig. 9A), we speculated that increases in HRR might be one of the mechanisms allowing cells with mutated p53 to avoid permanent arrest under replicational stress.


Figure 9
View larger version (27K):
[in this window]
[in a new window]

  		FIG. 9. Rad51 is essential to prevent proliferation arrest in response to VP16, but not APH. (A) HFF expressing DDp53 or vector (V) were cultured in the presence of 150 µM HU for the indicated periods of time and then harvested and analyzed by Western blotting with antibodies against Rad51 or tubulin. (B) HFF stably expressing shRNA against luciferase (pSR-GL2) or Rad51 (three different constructs, A, B, and C) were analyzed by Western blotting for expression of Rad51 and tubulin. The most efficient construct, C, was used for the experiment shown. (C) HFF expressing DDp53 and either shRNA against luciferase or Rad51 cultured in the presence of 150 nM APH or 200 nM VP16 were counted upon passage. The numbers of population doublings are plotted (each culture in duplicate), with standard deviations indicated (obscured when very small). (D) Ratios of population doublings of Rad51 shRNA-expressing cells relative to population doublings of luciferase shRNA controls are plotted. "*" indicates a P value from 0.01 to 0.05.

 
To test this hypothesis, we disrupted the expression of Rad51 in DDp53-expressing cells using stably expressed shRNA (Fig. 9B) and asked how Rad51 down-regulation affects cellular responses to replicational stress caused by APH and VP16. We found that expressing shRNAs against Rad51 (shown for construct C; not shown for construct A) reduced the proliferation of DDp53-expressing cells under normal culture conditions, consistent with essential roles of Rad51 (81, 88). The presence of 150 nM APH led to additional inhibition of proliferation rates, but surprisingly, the extents of the inhibition relative to proliferation with no drugs were similar in Rad51 shRNA- and control luciferase shRNA-expressing cells (Fig. 9C and D). In other words, Rad51 knockdown did not sensitize cells to APH treatment (similar results were obtained with HU [data not shown]). In marked contrast, culturing cells in the presence of 200 nM VP16 led to a much more dramatic inhibition of proliferation of Rad51 shRNA-expressing cells compared to control shRNA (Fig. 9C and D), revealing major differences in the requirements for HRR in dealing with replicational stress caused by simple inhibition of DNA replication and inhibition of DNA replication caused by generation of DNA strand breaks. Notably, the results with Rad51 inhibition are consistent with the p21 shRNA data and together suggest that although preventing p21 up-regulation is sufficient to prevent p53-dependent arrest in response to APH, the ability of p53 inhibition to overcome VP16-mediated arrest also involves a Rad51-dependent pathway.

Chronic replicational stress selects for p53 mutations. The ability of DDp53 to antagonize the induction of proliferation arrest under low levels of replicational stress suggests that subjecting a cell population to such stress would select for p53 mutants if the effect of p53 disruption is cell autonomous. To test this prediction, we mixed DDp53-expressing HFF that also coexpressed GFP with nontransduced HFF at a 1:50 ratio and cultured this mixture in the presence of HU, APH, or no drug. Culturing these mixed cell populations under normal conditions led to only modest selection for DDp53-expressing cells after many population doublings. In contrast, when cells were subjected to replicational stress induced by HU or APH, DDp53-expressing cells acquired a much stronger selective advantage (Fig. 10A). Thus, mild replicational stress can substantially increase the selective advantage conferred by p53 mutation, leading to clonal expansion of p53 mutant cells. The increased target size of p53 mutant cells should lead to increased possibilities for secondary mutations, which together with abrogated "guardian of the genome" function, could promote tumorigenesis.


Figure 10
View larger version (13K):
[in this window]
[in a new window]

  		FIG. 10. Mild replicational stress selects for cells with p53 disruption. (A) Early-passage HFF expressing DDp53 and GFP were mixed with nontransduced cells at a 1:50 ratio. The mixtures were cultured in the presence of either 150 µM HU, 200 nM APH, or no drugs for 13 days, and percentages of GFP-positive cells and cumulative population doublings were determined at each passage. Percentages of DDp53 (GFP)-expressing cells are plotted against population doublings, with standard deviations indicated. Note that more robust inhibition of proliferation induced by 200 nM APH in this experiment is associated with stronger selection for DNp53-expressing cells. (B) Model. p53 mutation (indicated by X) that is neutral under normal conditions provides cells with a selective advantage when the cell population is subjected to replicational stress. Chronic replicational stress leads to senescence (denoted by dotted cell outlines), removing these cells from the proliferative pool and reducing competition for niches. As chronic replicative senescence is largely (but not entirely) abrogated by p53 mutation, cells with disrupted p53 pathways will gain a significant competitive advantage.

 

   DISCUSSION
Top
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
We have shown that human and rodent cells chronically exposed to mild levels of replicational stress enter a senescence-like arrest that is dependent on p53 and its target, p21. Importantly, disruption of p53 not only allows cells to avoid arrest, but also does so without compromising long-term survival as assayed by clonogenic assays. In fact, low-level replicational stress robustly selects for cells with mutant p53. The improved long-term survival of p53 mutant cells under mild replicational stress is in marked contrast with the effects of p53 disruption on cellular responses to acute genotoxic stress. It is important to distinguish short-term abrogation of cell cycle arrest and/or apoptosis mediated by p53 mutation from long-term clonogenic survival (11, 12). Analyses of short-term survival of drug-treated cells often do not correlate with long-term clonogenic proliferation, and the latter better correlates with treatment responses in vivo (11, 12). Although p53–/– mouse embryo fibroblasts are highly resistant to VP16-induced apoptosis, clonogenic survival following 1 or 24 h of VP16 treatment over a range of doses is indistinguishable between p53+/+ and p53–/– cells (12, 46). Clonogenic cell survival following {gamma}-irradiation also does not correlate well with p53 mutation status in human tumor cells (12) or mouse embryo fibroblasts (24). Furthermore, although short-term (2-day) survival assays using the National Cancer Institute's panel of tumor cells indicated a strong positive correlation between wild-type p53 status and cell death induced by 86 clinically used anticancer agents (92), clonogenic assays using the same cell lines with cisplatin or mitomycin C failed to find any correlation (10). Finally, although senescence induction by doxorubicin has been shown to be partially p21 and p53 dependent, p53–/– and p21–/– tumor cells that avoid senescence in response to DNA damage still die by apoptosis or mitotic catastrophe a few days later, and the clonogenic survival of isogenic tumor cells mutated for p21 or p53 is actually reduced (19).

Hematopoietic cells appear to represent an exception to the lack of correlation between p53 status and long-term survival in response to DNA damage, as p53 loss promotes chemoresistance in mouse models of Myc-induced lymphomas (74), primary p53–/– and p53+/– mouse hematopoietic progenitors exhibit greater resistance to {gamma}-irradiation-induced inhibition of colony formation than p53+/+ progenitors (48, 52), and for lymphoid malignancies, p53 mutation is clearly associated with poor prognosis following chemotherapeutic treatment of patients (12, 75). As another exception, fibroblasts engineered to express dominant oncoproteins (such as adenoviral E1A) that sensitize the cells to DNA damage-induced apoptosis also exhibit p53-dependent responses to chemotherapy and radiation therapy, which translate into differences in tumor growth in xenograph models (9, 53).

It has been proposed that irrespective of the initial stimulus, senescence is caused by DNA damage that has not been repaired and that remains unrepaired throughout the senescent state (89). Multiple reports have demonstrated that replicational stress might induce formation of DNA DS breaks (1, 56, 70, 76), although these reports used higher levels of replicational stress and immortalized or tumor cell lines. While we show that a subset of replicational-stress-induced senescent cells indeed displayed stable {gamma}H2A.X foci, perhaps indicating DS breaks, the majority of cells were focus free. Moreover, although high levels of replicational stress induced another marker of DNA DS breaks, phosphorylation of CHK2 at Thr68, we did not observe CHK2 phosphorylation under treatment with low concentrations of replication inhibitors that induced senescence-like arrest. Moreover, inhibition of checkpoint signaling failed to antagonize proliferation arrest induced by low levels of replicational stress. Thus, chronic replicational-stress-induced senescence does not appear to depend on unrepairable DNA DS breaks. Because even a single unrepaired DNA DS break is sufficient to kill a cell (29), the long-term proliferative advantage conferred by p53 disruption during replicational stress suggests either that the arrest is not caused by DNA damage or that the damage is efficiently repaired before cells enter mitosis.

HR is considered to be the major mechanism for repair of replication-associated DNA DS breaks (1, 56, 70, 76). Because p53 disruption dramatically up-regulates HRR, we tested whether upregulation of HRR is important for the ability of DDp53 to overcome permanent arrest induced by replicational stress. We found that the ability of p53 inhibition to overcome permanent arrest induced by VP16, but not APH, was highly dependent on Rad51 (and thus HRR). This result is consistent with the ability of p21 knockdown to mimic the effect of p53 disruption in avoiding arrest induced by APH, but not VP16. While our results conflict with a reported role of HR in repairing replicational-stress-associated DNA damage (57, 70), those studies used Chinese hamster ovary-derived cells, which display much higher sensitivity to HU than human and rat fibroblasts. The most likely interpretation of our results is that stalled replication forks induced by low levels of APH can be resolved by nonrecombinogenic pathways (or that residual Rad51 levels are sufficient for HRR-mediated resolution) and that most stalled forks do not collapse. In contrast, VP16 presumably induces a substantially larger fraction of replication forks to collapse, and efficient resolution of collapsed replication forks requires HRR. In summary, our results suggest that replication-associated DNA DS breaks likely do not cause the permanent senescence-like arrest induced by chronic inhibition of DNA replication (by HU or APH) but do contribute to arrest induced by drugs that cause DNA strand breaks (like VP16).

Although most cellular proliferation presumably occurs under low-stress conditions, proliferation under replicational stress in vivo is not uncommon. The most obvious examples include the use of chemotherapeutic agents. HU has been used for over 40 years as an antitumor/cytostatic drug and is a frequently used medication for cytoreduction (18). For example, patients treated for chronic myelogenous leukemia can be treated with an oral HU regimen resulting in average plasma concentrations in the 100 to 200 µM range (45), which is similar to concentrations employed in our experiments (100 to 150 µM). Although when used as a single agent HU treatment has not been associated with increased cancer risk (96), patients treated with chemotherapeutic regimens that include VP16 are at increased risk of treatment-related acute myeloid leukemia (67). Furthermore, combining antimetabolites (such as methotrexate, a folate antagonist that inhibits dNTP synthesis) with VP16 increases the frequencies of treatment-induced leukemias relative to VP16 alone, suggesting that drugs that impair dNTP synthesis might function as tumor promoters by selecting for oncogenic events that avoid drug-induced senescence. Finally, patients undergoing bone marrow transplantation have an increased risk of new solid cancers (mostly carcinomas) (23), and these patients are frequently treated with VP16 as part of their conditioning regimen. Notably, the low concentrations of VP16 used in our experiments are even lower than those often experienced by patients in the clinic (37). However, concentrations experienced by target cells in the patient may be different from concentrations in plasma. Although secondary malignancies resulting from VP16 treatment are thought to result from mutagenic effects of the drug, our data indicate that replicational stress induced by this and other drugs might promote the selection of p53 mutant cells, which could synergize with increased mutation rates.

Previous evidence suggests that replicational stress can select for oncogenic mutations in vivo. In a recent report, we demonstrated that replication-impaired environments select for hematopoietic progenitors bearing particular oncogenic mutations, promoting leukemogenesis (6). Furthermore, clones of p53 mutant keratinocytes have been reported to occur in normal human skin, and higher sun exposure leads to increased size and frequency of the p53 mutant clones (42). Sun exposure is expected to create replicational stress resulting from UV light-induced pyrimidine dimers that stall the progression of replication forks. Based on the results presented here, we propose that chronic replicational stress should lead to selection of clones with p53 pathway disruption in vivo, therefore increasing the target size for additional mutations and driving tumorigenesis (Fig. 10B). In addition, continued replication in the presence of inappropriate or unbalanced levels of dNTPs can be mutagenic (43, 59), which could compound the increased genomic instability already associated with p53 mutation.

Finally, senescence induced by telomere shortening, DNA damage, or excessive oncogenic signaling appears to represent an important tumor-suppressive mechanism (60). On the other hand, the induction of senescence might be a double-edged sword. First, senescent fibroblasts have been shown to promote carcinoma development (44, 65). Second, although the activation of DNA damage-signaling pathways during oncogene-induced hyperreplication leads to tumor-suppressive senescence (3, 30), these events should also select for mutations (such as in p53) that avoid senescence (4, 26, 58). Finally, our results suggest that replicational stress, and perhaps other conditions that cause senescence, can select for mutations in the p53 pathway, thus driving tumorigenesis (Fig. 10B). In other words, by removing nonmutated cells from the proliferative pool, conditions promoting senescence can provide a competitive advantage for cells with p53 mutations. Thus, although senescence almost certainly limits tumorigenesis at a cell-autonomous level, conditions promoting senescence might also stimulate oncogenic progression via non-cell-autonomous mechanisms.


   ACKNOWLEDGMENTS
 
J.D. is supported by the National Institutes of Health (RO1 CA109657) and by a Scholar Award from the Leukemia and Lymphoma Society. During this study, C.K.M. was supported by grants from the National Science Foundation (MCB 9906576 and 0130760) and the Army Research Office (LS45039 and LS49557). Assistance from the Cancer Center Flow Cytometry Core was supported by grant 2 P30 CA 46934.

We thank Scot Lowe and Rene Bernards for providing shRNA constructs. We also thank Crystal Van Peursem and Vadym Zaberezhnyy for technical assistance; Siddharth Singh for preliminary experiments with p16 knockdown; Bob Sclafani and Heide Ford for their critical review of the manuscript; and K. Helm, C. Childs, and M. Ashton of the Cancer Center Flow Cytometry Core.


   FOOTNOTES
 
* Corresponding author. Mailing address: Department of Biochemistry and Molecular Genetics, Program in Molecular Biology, Integrated Department of Immunology, University of Colorado at Denver Health Sciences Center, Aurora, CO 80045. Phone: (303) 724-3230. Fax: (303) 724-3215. E-mail: james.degregori{at}uchsc.edu Back

{triangledown} Published ahead of print on 21 May 2007. Back

{dagger} Supplemental material for this article may be found at http://mcb.asm.org/. Back


   REFERENCES
Top
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
1. Arnaudeau, C., C. Lundin, and T. Helleday. 2001. DNA double-strand breaks associated with replication forks are predominantly repaired by homologous recombination involving an exchange mechanism in mammalian cells. J. Mol. Biol. 307:1235-1245.[CrossRef][Medline]

2. Bartek, J., R. Iggo, J. Gannon, and D. P. Lane. 1990. Genetic and immunochemical analysis of mutant p53 in human breast cancer cell lines. Oncogene 5:893-899.[Medline]

3. Bartkova, J., Z. Horejsi, K. Koed, A. Kramer, F. Tort, K. Zieger, P. Guldberg, M. Sehested, J. M. Nesland, C. Lukas, T. Orntoft, J. Lukas, and J. Bartek. 2005. DNA damage response as a candidate anti-cancer barrier in early human tumorigenesis. Nature 434:864-870.[CrossRef][Medline]

4. Bartkova, J., N. Rezaei, M. Liontos, P. Karakaidos, D. Kletsas, N. Issaeva, L. V. Vassiliou, E. Kolettas, K. Niforou, V. C. Zoumpourlis, M. Takaoka, H. Nakagawa, F. Tort, K. Fugger, F. Johansson, M. Sehested, C. L. Andersen, L. Dyrskjot, T. Orntoft, J. Lukas, C. Kittas, T. Helleday, T. D. Halazonetis, J. Bartek, and V. G. Gorgoulis. 2006. Oncogene-induced senescence is part of the tumorigenesis barrier imposed by DNA damage checkpoints. Nature 444:633-637.[CrossRef][Medline]

5. Bergamaschi, D., M. Gasco, L. Hiller, A. Sullivan, N. Syed, G. Trigiante, I. Yulug, M. Merlano, G. Numico, A. Comino, M. Attard, O. Reelfs, B. Gusterson, A. K. Bell, V. Heath, M. Tavassoli, P. J. Farrell, P. Smith, X. Lu, and T. Crook. 2003. p53 polymorphism influences response in cancer chemotherapy via modulation of p73-dependent apoptosis. Cancer Cell 3:387-402.[CrossRef][Medline]

6. Bilousova, G., A. Marusyk, C. C. Porter, R. D. Cardiff, and J. DeGregori. 2005. Impaired DNA replication within progenitor cell pools promotes leukemogenesis. PLoS Biol. 3:e401.[CrossRef][Medline]

7. Borel, F., F. B. Lacroix, and R. L. Margolis. 2002. Prolonged arrest of mammalian cells at the G1/S boundary results in permanent S phase stasis. J. Cell Sci. 115:2829-2838.[Abstract/Free Full Text]

8. Braig, M., S. Lee, C. Loddenkemper, C. Rudolph, A. H. Peters, B. Schlegelberger, H. Stein, B. Dorken, T. Jenuwein, and C. A. Schmitt. 2005. Oncogene-induced senescence as an initial barrier in lymphoma development. Nature 436:660-665.[CrossRef][Medline]

9. Bristow, R. G., A. Jang, J. Peacock, S. Chung, S. Benchimol, and R. P. Hill. 1994. Mutant p53 increases radioresistance in rat embryo fibroblasts simultaneously transfected with HPV16-E7 and/or activated H-ras. Oncogene 9:1527-1536.[Medline]

10. Brown, J. M. 1997. NCI's anticancer drug screening program may not be selecting for clinically active compounds. Oncol. Res. 9:213-215.[Medline]

11. Brown, J. M., and L. D. Attardi. 2005. The role of apoptosis in cancer development and treatment response. Nat. Rev. Cancer 5:231-237.[CrossRef][Medline]

12. Brown, J. M., and B. G. Wouters. 1999. Apoptosis, p53, and tumor cell sensitivity to anticancer agents. Cancer Res 59:1391-1399.[Abstract/Free Full Text]

13. Brown, J. P., W. Wei, and J. M. Sedivy. 1997. Bypass of senescence after disruption of p21CIP1/WAF1 gene in normal diploid human fibroblasts. Science 277:831-834.[Abstract/Free Full Text]

14. Brugarolas, J., C. Chandrasekaran, J. I. Gordon, D. Beach, T. Jacks, and G. J. Hannon. 1995. Radiation-induced cell cycle arrest compromised by p21 deficiency. Nature 377:552-557.[CrossRef][Medline]

15. Brugarolas, J., K. Moberg, S. D. Boyd, Y. Taya, T. Jacks, and J. A. Lees. 1999. Inhibition of cyclin-dependent kinase 2 by p21 is necessary for retinoblastoma protein-mediated G1 arrest after gamma-irradiation. Proc. Natl. Acad. Sci. USA 96:1002-1007.[Abstract/Free Full Text]

16. Campisi, J. 2005. Senescent cells, tumor suppression, and organismal aging: good citizens, bad neighbors. Cell 120:513-522.[CrossRef][Medline]

17. Casey, G., M. Lo-Hsueh, M. E. Lopez, B. Vogelstein, and E. J. Stanbridge. 1991. Growth suppression of human breast cancer cells by the introduction of a wild-type p53 gene. Oncogene 6:1791-1797.[Medline]

18. Chabner, B. 2001. Cancer chemotherapy and biotherapy: principles and practice, 3rd ed. Lippincott Williams and Wilkins, Philadelphia, PA.

19. Chang, B. D., Y. Xuan, E. V. Broude, H. Zhu, B. Schott, J. Fang, and I. B. Roninson. 1999. Role of p53 and p21waf1/cip1 in senescence-like terminal proliferation arrest induced in human tumor cells by chemotherapeutic drugs. Oncogene 18:4808-4818.[CrossRef][Medline]

20. Chen, Z., L. C. Trotman, D. Shaffer, H. K. Lin, Z. A. Dotan, M. Niki, J. A. Koutcher, H. I. Scher, T. Ludwig, W. Gerald, C. Cordon-Cardo, and P. P. Pandolfi. 2005. Crucial role of p53-dependent cellular senescence in suppression of Pten-deficient tumorigenesis. Nature 436:725-730.