Requirement for p53 and p21 to Sustain G2 Arrest After DNA Damage

See allHide authors and affiliations

Science  20 Nov 1998:
Vol. 282, Issue 5393, pp. 1497-1501
DOI: 10.1126/science.282.5393.1497


After DNA damage, many cells appear to enter a sustained arrest in the G2 phase of the cell cycle. It is shown here that this arrest could be sustained only when p53 was present in the cell and capable of transcriptionally activating the cyclin-dependent kinase inhibitor p21. After disruption of either the p53 or the p21 gene, γ radiated cells progressed into mitosis and exhibited a G2DNA content only because of a failure of cytokinesis. Thus, p53 and p21 appear to be essential for maintaining the G2 checkpoint in human cells.

DNA damaging agents in the form of γ radiation and chemotherapeutic drugs are the mainstays of most current cancer treatment regimens. This has stimulated much research to understand the cellular responses to DNA damage. After DNA damage, cells arrest at the transition from G1 to S phase (G1-S) or from G2 to M phase (G2-M) of the cell cycle, with DNA complements of 2n or 4n, respectively (1). Arrest at these checkpoints prevents DNA replication and mitosis in the presence of unrepaired chromosomal alterations. The proportion of cells that arrest at G1-S or G2-M depends on cell type, growth conditions, and the checkpoint controls operative in the cell (2). The G1-S arrest results, at least in part, from p53-regulated synthesis of the cell cycle inhibitor p21WAF1/CIP1 (3–5), which leads to inhibition of the cyclin-cdk complexes required for the transition from G1 to S phase. However, arrest in G2 after DNA damage occurs in both murine and human cells in the absence of p21 or p53 (3–6). This arrest is thought to result from activation of a protein kinase, Chk1, that phosphorylates and inhibits the function of the protein phosphatase Cdc25C (7). Inhibition of Cdc25C prevents the removal of inhibitory phosphates from Cdc2, a protein kinase that complexes with mitotic cyclins and is required for mitotic entry (8).

The integrity of the Chk1-Cdc25C-Cdc2 pathway in p21- or p53-mutant cells would appear to explain the prolonged G2-M arrest that occurs in such cells (3–7). However, there are some features of this arrest that have remained unexplained. Cells with disrupted p21 or p53 that are arrested in G2 can undergo DNA synthesis, in some cases resulting in cells with DNA contents of 8n or higher (9, 10). Such rereplication also occurs in cells blocked in mitosis (11), but it is unclear how this could occur in cells arrested before mitosis, as it is thought that negatively acting factors that prevent DNA synthesis must be degraded during the mitotic phase (12) and that positively-acting “licensing factors” required for the next S phase cannot traverse the nuclear membrane and therefore can enter the nucleus only during mitosis (13).

We therefore investigated G2-M arrest in p53-deficient cells in more detail. We initially used a panel of six human colorectal cancer cell lines, three with intact p53 genes and three with mutant genes. After irradiation, most of the cells in each culture were arrested with a DNA complement of 4n and >95% of the cells were in interphase (9, 14). To determine whether any of these cells entered M phase after irradiation, we treated them with nocodazole, a microtubule-disrupting agent that can trap cells in mitosis for several hours. In cells with wild-type p53, the mitotic index was very low after irradiation, as expected for cells truly blocked in G2 (Fig. 1A). Unexpectedly, a large fraction of the p53-mutant cells entered mitosis after irradiation (Fig. 1B).

Figure 1

Entry of colorectal cancer cell lines into mitosis after γ radiation. Cell lines with endogenous wild-type (A) or mutant (B) p53 genes were treated with nocodazole beginning 30 min after γ radiation (32). At the indicated times, cells were fixed, stained, and examined by fluorescence microscopy to determine the fraction of cells in mitosis (mitotic index). Cell lines were HCT116 (closed triangles), RKO (closed squares), SW48 (closed circles), DLD1 (open circles), HT29 (open diamonds), and Caco2 (open triangles).

These experiments suggested that p53 controls a G2checkpoint that prevents entry into mitosis after DNA damage. However, this conclusion was tempered by the fact that the six lines, all derived from colorectal cancers, might differ in ways other than p53 status. We therefore disrupted the p53 gene by homologous recombination in one of the wild-type p53-containing cell lines. The HCT116 cell line was chosen because it has apparently intact DNA damage–dependent and spindle-dependent checkpoints, and it is suitable for targeted homologous recombination (4, 9, 15). Two promoterless targeting vectors, each containing a geneticin- or hygromycin-resistance gene in place of genomic p53 sequences, were used to sequentially disrupt the two p53 alleles in HCT116 cells (16). Cells with the desired genotypes (Fig. 2A) were used to test the response to DNA damage. After treatment with γ radiation, parental cells (p53+/+) and those with one of the two p53 alleles disrupted (14) arrested in either G1 or G2, as expected for cells with intact checkpoints (Fig. 2B). The vast majority of cells with both p53 alleles disrupted (p53−/−) appeared to arrest at G2-M, with a DNA content of 4n, whereas a minor fraction had a DNA content of less than 2n, which is indicative of apoptosis (Fig. 2B). A substantial G2 arrest was observed in cells of all genotypes from 24 to 72 hours after 12-gray (Gy) of γ radiation (Fig. 2B). However, morphologic examination revealed mitotic cells in the p53−/− cell population within 24 hours after irradiation, whereas cells with one or two intact copies of p53 remained mitotically inactive for the duration of the experiment (Fig. 2C). All cell types exhibited mitotic arrest in response to nocodazole treatment in the absence of irradiation (Fig. 2D). A nocodazole-trapping experiment, however, confirmed that only p53−/−cells entered mitosis after irradiation (Fig. 2E). Thus, although G2 arrest was initiated after irradiation in all cells tested, this arrest was not sustained in the absence of functional p53 (Figs. 1 and 2E).

Figure 2

Targeted deletion of p53 in a colorectal cell line. (A) Southern blot after Hind III digestion of genomic DNA of selected clones (33). Fragments corresponding to the wild-type allele (wt) (2.5 kb), and neomycin (neo) (3.5 kb) and hygromycin (hyg) (3.7 kb) homologous integrants are shown. (B) Flow cytometric analysis of parental HCT116 cells (p53+/+) and HCT116 cells with targeted deletions of both (p53−/−) p53 alleles at the indicated time points after 12-Gy γ radiation. Flow cytometry was done as described (4). Ten thousand cells were analyzed in each experiment; n represents haploid DNA content and results are plotted on a logarithmic horizontal axis. Mitotic indices of parental cells (closed squares) and cells with targeted deletions of one (half-closed squares) or both (open squares) alleles of p53 are shown at the indicated times after 12-Gy γ radiation alone (C), nocodazole treatment alone (D), or 12-Gy γ radiation followed by nocodazole treatment (E).

Several potential mechanisms could account for these observations because p53 regulates the expression of many genes, including p21, that can affect the cell cycle (3–5, 17, 18). We examined the possible function of p21 in the maintenance of the G2arrest by using HCT116 cells in which the p21 genes were disrupted by homologous recombination (4). The p21−/− HCT116 cells arrested with a DNA content of 4n after γ radiation (4), but they continued to enter mitosis (Fig. 3, A and C). A control HCT116 line with targeted disruption of both alleles of the Smad4 gene (19) behaved identically to the parental HCT116 cells (Fig. 3, A to C). The G2checkpoint was activated after exposure of cells to 2-, 6-, and 12-Gy doses of γ radiation in wild-type, p53-, and p21-deficient cells (Fig. 3, D to F). However, the p53- and p21-deficient cells escaped the G2 arrest (Fig. 3, E and F), whereas the parental cells maintained it after exposure to 6 (Fig. 3D) or 12-Gy doses of γ radiation (Figs. 2E and 3C). The length of the G2 arrest in the p21- or p53-deficient cells depended on the dose of γ radiation, with the higher doses delaying the appearance of mitotic nuclei proportionately (Fig. 3, D to F). Escape from G2 arrest occurred earlier in p21- than in p53-deficient cells (Figs. 2E and 3C). This may be because some p53-independent p21 synthesis occurred after irradiation of p53−/−cells (Fig. 4A). These results are consistent with the fact that p53 is a major, but not sole, transcriptional regulator of p21 in mammalian cells (20).

Figure 3

Mitotic entry after γ radiation. Mitotic indices of parental HCT116 cells (closed circles), cells with targeted deletions of both p21 alleles (open squares), and cells with targeted deletions of both Smad4 alleles (closed triangles) after 12-Gy γ radiation (A), after nocodazole treatment alone (B), or after treatment with both 12-Gy γ radiation and nocodazole (C) (32). Response of wild-type HCT116 (D), p53-deficient (E), and p21-deficient derivatives (F) to lower doses of γ radiation in the presence of nocodazole. (G) Primary human fetal fibroblasts and their p21- and p53-deficient derivatives were treated with 12-Gy γ radiation plus nocodazole and were analyzed at the indicated times. Note that the scale on the y axis in (G) is different than in (D) to (F).

Figure 4

Protein expression and kinase activity after γ radiation. (A) Expression of p53 and p21 in parental HCT116 cells (+/+) and HCT116 cells with targeted deletion of one (+/−) or both (−/−) p53 alleles before and 36 hours after 12-Gy γ radiation assessed by immunoblot analysis (34). (B) Analysis of in vitro cyclin B1–associated histone H1 kinase activity in parental HCT116 cells (+/+) and in HCT116 cells with targeted deletions of both p21 alleles (−/−) at the indicated times after γ radiation. Whole cell extracts were immunoprecipitated with antibodies to cyclin B1 and assayed for kinase activity as described (35).

The most likely biochemical explanation for the entry into mitosis in the absence of p21 was lack of inhibition of the principal mitotic cyclin B1-cdc2 complex by p21 (21). The activity of this complex decreased within 12 hours after γ radiation in both cell types (Fig. 4B), which probably reflects activation of a checkpoint mechanism. This inhibition of cyclin B1-cdc2 kinase activity was not sustained in the absence of p21, as substantially increased activity was observed beginning 24 hours after DNA damage in p21-deficient cells (Fig. 4B).

To determine whether p21 and p53 are required to sustain the G2 arrest in cells other than colorectal cancer or epithelial cells, we disrupted the p53 gene by homologous recombination in normal human fibroblasts (22). Nocodazole trapping was then used to monitor the escape from G2 in parental fibroblasts and in a clone derived from the same fibroblasts in which the p21 genes had been disrupted by gene targeting (5). Again, the parental cells entered a sustained G2 arrest while a substantial fraction of both p21- and p53-deficient fibroblasts escaped G2 and entered mitosis (Fig. 3G).

Cells without p53 or p21 apparently proceed into mitosis after γ radiation but have a 4n DNA content (Fig. 2B) rather than the 2n DNA content expected for cells that had gone through mitosis. To investigate this contradictory result further, we stably transfected cells with a histone H2B–green fluorescent protein fusion vector to allow real-time visualization of the mitotic process (23). Time lapse experiments in the absence of nocodazole showed that 90 ± 6% of the p21−/− cells entered mitosis within 36 hours after 12-Gy γ radiation compared with less than 2% of the parental cells. The first stages of mitosis after irradiation of p21−/− (or p53−/− ) cells were indistinguishable from those in cells growing under normal conditions (Fig. 5A). After anaphase, however, the irradiated p53−/− and p21−/−cells never completed cytokinesis (Fig. 5B). These cells eventually flattened and the chromosomes decondensed, and >95% of the cells were found to contain abnormally shaped, multilobulated nuclei (Fig. 5, C, D, and E). A subset of these cells subsequently underwent programmed cell death. Staining with an antibody to the centrosome-specific γ-tubulin revealed that these cells always contained at least three centrosomes or pairs of centrosomes located in a cleft that likely was a remnant of the cleavage furrow associated with the failure of cytokinesis (Fig. 5, D and E). A large number of centrosomes, also observed in mouse cells that lack p53 (24), reflected the centrosome duplication that accompanies DNA synthesis (25) and was consistent with the fact that p53−/− and p21−/− cells often reenter S phase after irradiation, becoming tetraploid or octaploid (9,10).

Figure 5

Fluorescence microscopy of p21−/− cells. Clones stably expressing the histone H2B/GFP fusion protein were isolated and observed by time-lapse microscopy under bright-field (upper) and fluorescence (lower) illumination (23). (A) A p21−/− cell undergoing mitosis in the absence of radiation. The prometaphase-to-telophase transition time was 66 min. (B) An irradiated p21−/−cell. The prometaphase-to-telophase transition time was 114 min. Forty- eight hours after 12-Gy γ radiation, wild-type (C), p21−/− (D), and p53−/− (E) cells were fixed and stained with 4′,6-diamidino-2-phenylindole (left, blue) or immunostained with antibodies specific for the γ-tubulin component of centrosomes (36) (right, green). The nucleus in (D) was bilobed, with connections between the lobes visible in a different focal plane from the one shown.

These results demonstrate that induced expression of p21 and p53 is essential to sustain the G2 checkpoint after DNA damage in human cells. Although most research on p53- and p21-regulated checkpoints has focused on the G1-S transition, several previous observations are consistent with an important role for these genes in G2-M (26–29). The p21 protein is synthesized in G2 (27, 28), promotes a pause in late G2 under normal growth conditions (26, 28), and, when expressed exogenously, causes cells to arrest in G2 (29). In contrast, neither p53 nor p21 appears to play a major role in the spindle checkpoint because p53−/− and p21−/− cells respond normally to microtubule disruption (Figs. 2D and 3B). It is not yet clear whether cells with heavily damaged DNA fail to undergo cytokinesis because of a cytokinesis checkpoint (18, 30) or because of a simple mechanical problem.

Although p53 mutations provide cells with a selective growth advantage, such mutations burden them with a significant checkpoint deficit; they cannot respond normally to DNA-damaging agents and enter mitosis and subsequently replicate their genomes in the presence of DNA damage. Such checkpoint defects (31) may be exploited to treat the many cancers with abnormalities of p53 function.

  • * To whom correspondence should be addressed. E-mail: vogelbe{at}


View Abstract

Navigate This Article