p53 Prevents Entry into Mitosis with Uncapped Telomeres

DNA Damage Response Factors at Human Telomeres during Mitosis

Telomere uncapping leads to checkpoint activation and recruitment of DNA damage response factors to the telomere (e.g., 53BP1, MDC1, γH2AX, and ATM), visualized as microscopically defined telomere-dysfunction-induced foci (TIFs, reviewed in [5]). Telomeres become transiently uncapped in every cell cycle following their replication in S phase, and the ensuing ATM-dependent response is thought to promote the reassembly of protective telomeric structures in G2 [4]. To monitor telomere recapping at the G2/M transition under physiological conditions, we synchronized human HeLa 1.2.11 cells by double thymidine block and release. γH2AX foci were detectable in G2, and a subset of these localized to telomeres, indicative of TIFs (see Figure S1A, available online). Upon entry into mitosis, several sites of γH2AX labeling persisted, and these preferentially colocalized with telomeres labeled with an anti-TRF2 antibody (Figure 1A). The frequency of these mitotic TIFs was comparable to that of interphase TIFs (Figures S1A), suggestive of persistence of G2-uncapped telomeres into mitosis. To confirm γH2AX localization at telomeres, we visualized TIFs on mitotic chromosomes by colocalization of γH2AX and telomeric fluorescent in situ hybridization (FISH) signals (Figures 1B and 1C). In addition to γH2AX, we also detected MDC1 at mitotic telomeres (Figure S1B), a damage response mediator at the G2/M transition [6]. In contrast, 53BP1, another damage response protein [7], was not detected at mitotic telomeres. HeLa 1.2.11 cells carry relatively long telomeres (∼ 17 kb [8]). We therefore also analyzed HeLa OHIO cells with shorter telomeres (∼ 3.4 kb [8]) and again observed γH2AX labeling of mitotic telomeres (Figures 1B and 1C). The number of γH2AX-labeled telomeres in mitotic HeLa OHIO cells was even higher than in HeLa 1.2.11 cells (Figure 1D).

γH2AX staining at deprotected telomeres can extend over approximately 570 kb into the subtelomeric region in human cells [9]. Consistently, the chromatin domain covered by γH2AX often extended beyond the telomeric FISH signal or the region of TRF2 staining in our experiments (Figures 1A–1C). To address whether the γH2AX signal indeed originated from telomeric DNA, we analyzed γH2AX binding to telomere DNA by using chromatin immunoprecipitation (ChIP; Figure 2A). We used HeLa 1.2.11 cells either untreated or arrested in mitosis by addition of colcemid. An antibody against γH2AX efficiently pulled down telomeric DNA from mitotic cells, whereas less enrichment was observed in unsynchronized cells. An antibody against the telomere-associated protein TRF2 was used as a positive control that precipitated telomeric DNA with similar yield in both cases. Preimmune serum served as a control for nonspecific binding, and no precipitation was observed when a nontelomeric rDNA probe was used instead of a telomere probe for detection. These results confirm specific association of γH2AX with telomeres, which can clearly be seen in cells arrested in mitosis.

Mitotic TIFs Mark Short, Uncapped Telomeres

We next addressed whether a correlation existed between telomere length and the occurrence of mitotic TIFs. Our initial results suggested that HeLa OHIO cells with shorter telomeres show a somewhat higher incidence of telomeric γH2AX staining during mitosis compared to HeLa 1.2.11 cells (Figure 1D). This observation was confirmed in mitotic chromosome preparations of HeLa 1.2.11 cells, where we observed that nuclei with stronger telomeric FISH signals, indicative of long telomeres (Figure 2B, long), typically lack TIFs. In contrast, TIFs were abundant in metaphases with shorter telomeres, on the basis of the FISH signal intensity (Figure 2B, short). To quantitatively assess whether γH2AX labeling was linked to short telomere length, we measured the intensities of individual telomeric FISH signals in mitotic HeLa 1.2.11 cells by using the Q-FISH technique and recorded their γH2AX status in the same preparation. This revealed that shorter telomeres were indeed more likely to be labeled by γH2AX. Twenty percent of γH2AX-positive telomeres, but only 7% of all telomeres, were shorter than 10 kb in length (Figure 2B).

To determine whether mitotic γH2AX labeling is indicative of uncapped telomeres, we artificially created telomere uncapping by overexpression of dominant-negative TRF2 lacking both Myb and N-terminal basic domain (TRF2^ΔMΔB [10]) in human T4 cells under control of a doxycyclin-repressible promoter. TRF2^ΔMΔB levels increased over 7 days after removal of doxycyclin from the growth medium (Figure 2C). Before induction, only a few TIFs were detected on spread metaphase chromosomes, similar to what we observed in HeLa cells. In contrast, 7 days after induction of TRF2^ΔMΔB, the majority of mitotic telomeres were labeled by γH2AX (Figure 2C). We conclude that uncapped telomeres display mitotic γH2AX staining. TIFs persisting into mitosis in cells with intact telomere structure may therefore reflect telomeres that have remained uncapped after their replication in S phase.

The p53/p21 Pathway Prevents Mitotic Entry with Uncapped Telomeres and Telomere Fusion

Telomere capping after DNA replication is thought to be completed during the G2 phase of the cell cycle [4]; therefore, the widespread presence of mitotic TIFs in a range of human cell lines came as a surprise. However, the cells so far used in our studies lacked functional p53. To investigate whether p53 plays a role in preventing mitotic uncapped telomeres, we first analyzed the human U2OS cell line that is proficient for p53 function [11]. γH2AX labeling on mitotic chromosomes was rarely seen but increased to frequencies similar to HeLa cells after siRNA-mediated depletion of p53 or p21 from these cells (Figure S2). To confirm the dependence of mitotic TIFs on the p53 status, we analyzed isogenic mouse embryonic fibroblasts (MEFs) carrying wild-type p53 or a targeted deletion in the p53 gene [12]. γH2AX labeling of mitotic chromosomes was again rarely seen in wild-type MEFs but occurred frequently in p53^−/− MEFs (Figure 3A). As was the case in human HeLa cells, Q-FISH analysis revealed that γH2AX staining was preferentially observed at short telomeres of p53^−/− MEFs (data not shown). These results suggest that p53 is required either for efficient capping of telomeres in G2 before the time of mitotic entry or to prevent progression into mitosis while telomere capping reactions are ongoing. p53 has been shown to delay mitotic entry in response to DNA damage by upregulation of its downstream target p21 [13]. We observed mitotic TIFs in MEFs carrying a p21 deletion [14] as frequently as in MEFs lacking p53 (Figure 3A). This suggests that one role of p53 is to delay cell-cycle progression into mitosis by p21 induction when uncapped telomeres persist. Association of γH2AX with sites close to the telomeres of mitotic chromosomes has been reported in late generation Terc^−/− T cells [15]. When we quantified mitotic TIFs in Terc^−/− shp53 MEFs [16] with short telomeres (G4; ~20.88 kb average), we found a substantial increase compared to wild-type MEFs (∼36.4 kb average) (Figures 3A and 3B). This confirms our observation that shorter telomeres are more prone to persisting in an uncapped state.

To obtain insight into the fate of uncapped telomeres in the absence of p53, we examined the frequency of chromosome abnormalities in p53^−/− MEFs (Figure 3C). As expected, we observed an increased incidence of DNA breaks and complex structural aberrations in p53^−/− MEFs, as compared to p53^+/+ MEFs established from three embryos derived for each genotype (data not shown). In addition, we observed an increase in chromosome and chromatid fusions in p53^−/− MEFs. Importantly, we observed a significant 4.4-fold increase in the frequency of chromosome and chromatid fusions that displayed telomeric DNA at the fusion site (Figure 3C). End-to-end fusions containing telomeric sequences are a characteristic signature of telomeric dysfunction [10]. These fusions may break in subsequent mitoses, thus starting a bridge-fusion-breakage cycle [17] that may contribute to the overall genomic instability of p53-deficient cells. Our results demonstrate that p53 is required to protect telomeres from end-to-end fusions.

To address whether p53 affects the telomere capping reaction, we synchronized cells from a pair of isogenic human colorectal cancer cell lines, proficient or deleted for p53 (HCT116 p53^+/+ or p53^−/−) [13], by double thymidine block and release. We monitored the occurrence of TIFs at the G2/M transition, 7–11 hr after the second release. At 7 hr, both p53^+/+ and p53^−/− cells had completed DNA replication and stained positive for the G2 specific marker CENP-F [18] (Figure S3). Most cells in both populations exhibited more than two TIFs at this time. By the time cells entered mitosis, as seen by the increase in phosphohistone H3 staining, TIFs were no longer observed in p53^+/+ cells. In contrast, TIFs remained detectable in most p53^−/− cells throughout mitosis and as cells re-entered G1, at 10 hr after release. This suggests that p53 is required for efficient completion of telomere capping and that the p53 response pathway may have an impact on the capping reaction. Alternatively, homologous recombination reactions that are part of telomere capping may become disfavored as p53^−/− cells enter mitosis prematurely [19, 20]. As reported previously [13] and as shown here by staining with phosphohistone H3 (Figure S3), a marker for the G2/M transition, mitotic entry of p53^−/− cells was somewhat advanced compared to p53^+/+ cells.

Uncapped Telomeres Trigger ATM- and p53/p21-Dependent G2/M Arrest

To directly address the ability of p53 to delay the G2/M transition in the presence of uncapped telomeres, we disrupted telomere protective structures by depleting components of the telomeric complex in mouse and human cells and compared mitotic indexes between p53^+/+ and p53^−/− cells. p53^+/+ and p53^−/− primary MEFs carrying homozygous floxed TRF1 alleles (TRF1^F/F) [21] were infected with retroviruses encoding Cre recombinase or vector control. Deletion of TRF1, as monitored by western blotting, resulted in marked telomere uncapping visualized by γH2AX-labeled telomeres (Figure 4A). TIF quantification in interphase cells revealed no impact of p53 status on TIF frequency (data not shown). To analyze mitotic entry of these cells after telomere uncapping, we determined the fraction of cells in mitosis at the indicated times after infection (Figure 4B). p53^+/+ cells displayed a very low mitotic index after uncapping, compared to TRF1-positive control cells. p53^−/− cells showed a higher mitotic index in the TRF1-proficient background compared to p53^+/+ cells, and this was not affected after loss of telomere protection by TRF1 deletion. FACS analysis of DNA content (Figure 4B) confirmed that the low mitotic index in p53^−/− cells with uncapped telomeres was due to an arrest of cells in G2/M rather than G1/S.

Similarly, p53 was required to prevent mitotic entry of human cells with uncapped telomeres (Figure S4A). When TRF2 was depleted with siRNA in p53^+/+ or p53^−/− HCT116 cells, the ability of cells to enter mitosis was significantly impaired only in the presence of functional p53. Additionally, human cells depleted of p21 and TRF2 entered mitosis at a higher rate compared to p21-proficient cells (Figure S4B). This suggests that a p53-dependent pathway recognizes uncapped telomeres and delays mitotic entry via activation of p21. Consistent with this idea, we observed p21 induction after siRNAmediated depletion of TRF2 (Figure S4B).

TIF formation at uncapped telomeres during S phase is ATM dependent [2]. Similarly, we found that mitotic TIFs were strongly reduced when we treated HeLa 1.2.11 cells with the ATM inhibitor Ku55933 (Figure S4C). In comparison, MO59J cells deficient in DNA-PK activity [22] showed similar levels of mitotic TIFs to those of HeLa and MO59K control cells. Instead, Ku55933 addition prevented mitotic TIF formation in MO59J and MO59K cells to the same degree as in HeLa cells. This demonstrates that γH2AX labeling of mitotic telomeres depends on ATM, but not DNA-PK, characteristic for the response to uncapped telomeres.

To gain mechanistic insight into the control of p53 activation at the G2/M transition, we monitored in synchronized human cells ATM-dependent p53 phosphorylation at Ser15, an event known to occur in response to DNA damage [23, 24]. Ser15 phosphorylation occurred in S phase and also at the G2/M transition monitored by FACS analyses of DNA content and phospho-H3 staining (Figure 4C). Concomitant activation of p21 was observed only in G2/M, emphasizing its importance at this stage of the cell cycle. Another target of p53, 14-3-3σ [25], showed a similar induction pattern to that of p21, raising the possibility that 14-3-3σ is also involved in the response to telomere damage. To test this hypothesis, we synchronized U2OS cells that have been depleted of TRF2 by using siRNA (Figure 4D). Although increased ATM-dependent Ser15 p53 phosphorylation, as well as p21 induction, occurred specifically in response to TRF2 depletion, 14-3-3σ expression was not significantly augmented. This suggests that p21, and to a lesser degree 14-3-3σ, act downstream of p53 in response to telomere dysfunction at the G2/M transition.

Traditionally, much research on p53 has focused on its role in regulating cell-cycle arrest at the G1/S transition in response to DNA damage and other types of cellular stress. An ATM-and p53-dependent checkpoint that detects altered telomeric state at the G1/S transition was proposed to act as a regulator of the cellular response to telomere dysfunction [26]. However, several reports also support a role for the p53 pathway at the G2/M transition (reviewed in [25, 27]). Here we provide evidence that telomere uncapping during normal cell-cycle progression triggers a p53-dependent checkpoint at the G2/M transition. The nature of the signal could be the linear telomere structure transiently generated during DNA replication in S phase, which resembles an unrepaired DSB. It has been proposed that telomeres in a linear form trigger a DNA damage response to initiate the reactions required for elongation, processing, and capping [4], and reactions related to those of DNA repair by homologous recombination are required to form a protective t loop structure [20]. Our results suggest that ATM, as a sensor of linear telomeres, at least in part uses p53 as an effector to delay mitotic entry and halt the cell cycle in a state conducive for the capping processes to be completed. The physiologic p53 response to uncapped telomeres, which naturally occur in every round of cell division after DNA replication, extends our understanding of this important tumor suppressor.

Cell Culture and Treatments

Human T4 [10], HeLa 1.2.11, HeLa OHIO, U2OS, HCT116 wild-type and p53^−/− cells [13], and MO59J and MO59K [22] (all obtained from Cancer Research UK Cell Services) were cultivated in monolayers in Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen) supplemented with antibiotics (penicillin and streptomycin, Sigma Aldrich) and 10% fetal bovine serum (FBS, Invitrogen). Primary MEFs were isolated from day 13.5 embryos embryos as previously described [28] and cultivated in DMEM medium (Invitrogen) supplemented with antibiotics (penicillin and streptomycin, Sigma Aldrich) and 10% FBS (Invitrogen). p21^−/− MEFs [14] were a gift from M. Serrano (CNIO). Isogenic wild-type and p53^−/− MEFs, isogenic TRF1^F/F p53^−/− and TRF1^F/F p53^+/+, and MEFs established from C57BL6 embryos of 4^th generation (G4) Terc^−/− immortalized by p53 depletion (shp53 [29]) have been previously described [12, 16, 21]. ATM inhibitor Ku-55933 ([30]; a gift from G. Smith) was added at 10 μM final concentration for 1 hr prior to collection. Mitotic cell arrest was carried out by addition of 0.1 μg/ml colcemid (GIBCO) to human cells and 0.2–0.3 μg/ml to MEF cultures, followed by 4–12 hr incubation at 37°C. The synchronization of HeLa 1.2.11, HCT116, and U2OS cells was performed with a double thymidine block. In brief, exponentially growing cells were arrested by addition of thymidine (2 mM) to the growth media for 16 hr, followed by two washes in HBSS and release into fresh media for 10 hr. Cells were then arrested a second time by addition of thymidine (2 mM) and 16 hr incubation, then washed as above and released into fresh media.

Antibodies

The following antibodies were used for immunoblotting: rabbit polyclonal antisera raised against human TRF2 [31], mouse TRF2 (2390 [32]), human histone H3 (a gift from A. Verreault), Ser15-p53 (Cell Signaling), and human p21 (C-19, Santa Cruz); and mouse monoclonal antibodies raised against phosphorylated ATM-Ser1981 (Cell Signaling), mouse TRF1 (TRF-78, Abcam), p53 (DO-1, Santa Cruz Biotechnology), 14-3-3σ (Upstate), and α-tubulin [31] (Cancer Research UK Monoclonal Antibody Service). Additional antibodies used for imunoflourescence detection were mouse monoclonal antibodies raised against MDC1 (Abcam) and phosphorylated histone H2AX-Ser 139 (Upstate) and a rabbit polyclonal antiserum raised against 53BP1 (Novus) and CENP-F (Abcam). Antibodies used in ChIP analyses were rabbit polyclonal antibodies anti-phosphorylated histone H2AX-Ser 139 (Upstate) and anti-human TRF2.

Preparation of Metaphase Spreads

Mitotic cells were collected by mitotic shake off and swollen in hypotonic buffer (10 mM Tris-HCl, pH 7.5, 10 mM NaCl, 5 mM MgCl₂) at 37°C for 5 min. For IF-FISH, cells were either spun onto coverslips via a cytospin centrifuge (Cytospin 4, Fisher) or spread across a slide presoaked in 2% paraformaldehyde as described [33] and dried. For Q-FISH, cells were fixed in a freshly prepared 3:1 mix of methanol:glacial acetic acid. Nuclear preparations were dropped onto slides presoaked in 45% acetic acid and left to dry overnight.

Immunofluorescence and FISH

Metaphase spreads prepared as above were fixed in 4% paraformaldehyde with 0.1%–0.5% Triton X-100 in PBS, permeabilized with 0.5% Triton X-100 in PBS, and subjected to immunofluorescence staining as described [31]. In brief, metaphase spreads were incubated in antibody dilution buffer (1% goat serum, 0.3% BSA, 0.005% Triton X-100 in PBS) for 10 min, then in primary antibody overnight. After three washes in antibody dilution buffer supplemented with 0.005% Triton X-100, secondary antibody was added for 1 hr. All antibody incubations were at room temperature. Samples were then washed and fixed again in 4% paraformaldehyde in PBS. After three washes in PBS, slides were briefly dried at room temperature. FISH was performed as described [31] with 15 μg/ml Cy3-conjugated [CCCTAA]₆-PNA telomeric probe (Applied Biosystems). Chromosomes were visualized with DAPI.

Q-FISH, Mitotic Index, and Chromosome Analyses

Primary MEFs were isolated from day 13.5 embryos from p53^−/− [12] and p21^−/− mice [14] as described [34], passaged once and treated with 0.1 μg/ml colcemid for 4 hr, trypsinized, and fixed as above for metaphase spreads. Q-FISH analyses and statistical analyses of chromosomal aberrations and telomere length were carried out as described [28, 31]. Fifty anaphases per cell line from two independent cell samples of each genotype were scored. One fusion event was defined as the result of the joining of two or more chromosomes. The end-to-end fusion index was calculated as the number of events per total number of metaphases. For determining the mitotic index, dried metaphase spreads were mounted in DAPI and visualized by fluorescence microscopy. Nuclei with condensed, evenly stained chromosomes were counted as mitotic. Between 350 and 500 nuclei were scored for each time point.