Ink4a/Arf tumor suppressor does not modulate the degenerative conditions or tumor spectrum of the telomerase-deficient mouse

Abstract

The Rb/p16^Ink4a and p53/p19Arf tumor suppressor pathways have been linked to diverse cancer-relevant processes, including those governing the cellular responses to telomere dysfunction. In this study, we sought to provide direct genetic evidence of a role for the Ink4a/Arf tumor suppressor gene, encoding both p16^Ink4a and p19^Arf, in modulating the cellular and tissue phenotypes associated with telomere dysfunction by using the mTerc Ink4a/Arf mouse model. In contrast to the rescue associated with p53 deficiency, Ink4a/Arf deficiency did not attenuate the degenerative phenotypes elicited by telomere dysfunction in the late-generation mTerc−/− mice. Furthermore, in contrast to accelerated cancer onset and increased epithelial cancers of late-generation mTerc−/− p53 mutant mice, late-generation mTerc−/− Ink4a/Arf mutant mice experienced a delayed tumor onset and maintained the lymphoma and sarcoma spectrum. Consistent with the negligible role of Ink4a/Arf in the telomere checkpoint response in vivo, late-generation mTerc−/− Ink4a/Arf−/− tissues show activated p53, and derivative tumor cell lines sustain frequent loss of p53 function, whereas all early generation mTerc Ink4a/Arf−/− tumor cell lines remain intact for p53. In addition, the late-generation mTerc−/− Ink4a/Arf−/− tumors showed activation of the alternative lengthening of telomere mechanism, underscoring the need for adaptation to the presence of telomere dysfunction in the absence of p16^Ink4a and p19^Arf. These observations highlight the importance of genetic context in dictating whether telomere dysfunction promotes or suppresses age-related degenerative conditions as well as the rate of initiation and type of spontaneous cancers.

In cultured human cells, p53 and Rb are well established components of the machinery that monitors telomere status and executes checkpoints in the setting of telomere dysfunction (1–3). Correspondingly, mice doubly null for the telomerase RNA component (mTerc) and p53 genes have provided corroborating genetic evidence for a central role for p53 in the telomere checkpoint in vivo. Specifically, p53 deficiency results in near-complete rescue of adverse cellular consequences associated with telomere dysfunction, including cellular growth arrest and apoptosis, progressive organ system atrophy, and physiological compromise. Although enabling cellular growth and survival and masking degenerative phenotypes, these mTerc−/− p53 mutant mice experience a dramatic increase in cancer incidence, shift in tumor spectrum toward epithelial cancers, and accumulation of marked chromosomal aberrations targeting cancer-relevant loci for amplification and deletion (4–6).

Subsequent analyses of Atm, the upstream activator of p53, revealed that telomere dysfunction exacerbates the degenerative condition in late-generation mTerc−/− Atm−/− mice when compared with late-generation mTerc−/− Atm+/+ mice. The degenerative condition was characterized by depletion of tissue stem cell reserves, widespread organ system failure, accelerated aging and shortened lifespan, and near-complete suppression of lymphomagenesis (7, 8). The basis for this accelerated degeneration and cancer suppression may relate to several factors, including accelerated telomere erosion in ATM-deficient cells, preservation of an ATM-independent p53 response, and levels of chromosomal instability incompatible with cell viability (7–10). This result was not expected based on observations of a blunted checkpoint response to telomere deprotection in Atm or p53 mutant cultured human cells (3). Thus, despite the close linkage of ATM and p53 in DNA damage signaling, the opposite presentations of telomere dysfunction phenotypes in p53 versus Atm mutant settings underscores the complexity of the interactions between telomeres and specific components of the p53 signaling pathway and emphasizes the need to conduct systematic genetic analyses in vivo, in multiple cell types, and in different model systems.

The products of the Ink4a/Arf tumor suppressor locus are linked to the p53 pathway on a number of levels (11). This locus encodes p16^Ink4a and p19^Arf, which regulate Rb and p53, respectively (12). The p16^Ink4a protein functions to disrupt binding of cyclin-dependent kinase (Cdk) 4/6 to D-type cyclins, leading to decreased Cdk-directed Rb hyperphosphorylation and inactivation, resulting in G₁ phase cell cycle arrest. It has been reported that p16^Ink4a plays a modest role in response to DNA double-stranded breaks and telomere dysfunction in the absence of p53 (13, 14). This observation is in line with genetic evidence that Rb itself can contribute to DNA damage-induced growth arrest in part via p53-directed up-regulation of the Cdk inhibitor, p21^CIP1 (15, 16).

A number of studies indicate that p53-dependent telomere dysfunction and DNA damage responses are equivalent on molecular and cellular levels (3, 17, 18). Nevetheless, the linkage of p19^Arf to these p53-dependent processes has been less definitive. p19^Arf has been shown to inhibit Mdm2-mediated degradation of p53 (19–21) and is clearly required for activation of p53 in response to aberrant cell cycle entry (19) and oncogene activation (22–24). p19^Arf is not required for p53 induction by γ irradiation in vivo (25), although these observations appear to be at odds with its requirement in sustained cell cycle arrest after DNA damage in vitro (26). Consistent with the latter, p19^Arf deficiency has been reported to provide partial rescue of growth arrest after acute deprotection of telomeres by expression of a dominant-negative mutant of the telomere binding protein, Trf2, in mouse embryonic fibroblasts (MEFs) (27). However, these cell culture-based telomere deprotection findings appear to differ from those derived from in vivo analysis of late-generation mTerc−/− mice that experience telomere erosion (as opposed to deprotection) and are deficient for both p16^INK4a and p19^ARF (28). In these late-generation mTerc−/− Ink4a/Arf−/− mice, there is suppression of carcinogen-induced cancers, consistent with an intact telomere checkpoint in the development of tumorigenesis despite the absence of p16^Ink4a and p19^Arf. In line with these in vivo observations, there was a corresponding inhibition of Myc/Ras-induced transformation of primary late-generation mTerc−/− Ink4a/Arf−/− MEFs compared with early generation mTerc Ink4a/Arf−/− controls. The basis for the contrasting findings in these model systems is not clear and could relate to the use of carcinogens (as opposed to spontaneous cancers), differences in telomere erosion versus experimental deprotection, and/or in vitro versus in vivo experimental settings. For the same reasons, it is difficult to make clear comparisons between late-generation mTerc−/− Ink4a/Arf−/− versus mTerc−/− p53−/− studies, particularly given the use of carcinogens and lack of analysis of normal aging tissues conducted in the former study.

In this study, we sought to determine more rigorously the impact of p16^Ink4a and p19^Arf deficiency on the telomere checkpoint in both primary cells and tissues and in spontaneous tumor development, the latter with an emphasis on tumor kinetics and spectrum. The specific use of the Ink4a/Arf mutant allele here is motivated by the prominence of this allele in human cancers (29) and its relevance in pathways linked to cellular senescence and aging. The genetic studies of this report do not support a role for p16^Ink4a and p19^Arf in checkpoints activated by telomere dysfunction in the tissues and tumors examined in this mouse model.

Results

Ink4a/Arf Status Does Not Alter the Degree of Telomere Dysfunction or the Degenerative Cellular and Tissue Phenotypes Associated with Telomere Dysfunction.

As a first step toward an assessment of the impact of Ink4a/Arf deficiency on the telomere dysfunction phenotype, we measured the degree of telomere dysfunction in all three Ink4a/Arf genotypes. We monitored the anaphase bridge index (ABI), an in vivo biomarker of the degree of telomere dysfunction in dividing tissues that mirrors chromosomal fusion frequency, in successive mTerc−/− generations (7, 30) (see Materials and Methods for definition of ABI). The gastrointestinal (GI) tract is particularly well suited for in vivo ABI measurements given the high proliferative rates of the epithelia and the regional compartmentalization of proliferation in the crypts. There was the expected ABI increase in the late-generation G₄/G₅ mTerc−/− intestinal crypts compared with the early generation mTerc+/+, +/−, and G₁ mTerc−/− controls (referred to as early generation mTerc from this point onward) (Fig. 1A). Notably, the ABI was comparable in all three Ink4a/Arf genotypes in late-generation mTerc−/− mice, showing that Ink4a/Arf status does not influence the degree of telomere dysfunction.

Fig. 1.

Loss of p16^Ink4a and p19^Arf does not modulate the in vivo phenotype of telomere dysfunction. (A) Increase in the percentage of anaphases with anaphase bridging in the GI tract of late-generation mTerc−/− mice. Error bars represent mean ± SE (n = 4–7 for each group). No significant difference between late-generation mTerc−/− Ink4a/Arf+/+ and −/− mice (P = 0.1306). Twenty-five anaphases were counted per sample. (B) Representative examples of the degenerative phenotype seen in 12- to 13-week-old G₄ mTerc−/− Ink4a/Arf mutant mice compared with age-matched G₀ mTerc+/− Ink4a/Arf controls. (C) Increase in apoptotic bodies in the crypts of the GI tract in late-generation mTerc mutant mice with different Ink4a/Arf genotypes. No significant differences between the number of apoptotic bodies between the different Ink4a/Arf genotypes within the same generations (P = 0.9084). Error bars represent mean ± SE (n = 4–7 for each group). (D) p53 induction in late-generation mTerc−/− Ink4a/Arf+/+ and Ink4a/Arf−/− intestinal crypts. (E) Testes weights (Left) and histological sections (Right) of early and late-generation mTerc mutant mice with different Ink4a/Arf genotypes. Loss of Ink4a/Arf does not rescue germ cell depletion in late-generation mTerc−/− testes. Inset shows depletion of spermatogonia in late-generation mTerc−/− testes.

We next examined the impact of Ink4a/Arf status on the broad range of degenerative phenotypes reported previously in late-generation mTerc−/− mice. On gross inspection, late-generation mTerc−/− Ink4a/Arf+/+ and −/− mice presented with the same classical picture of generalized frailty, reduced body weight and size, hunched posture (kyphosis), hair graying, and alopecia, consistent with previous findings of mice with dysfunctional telomeres (31) (Fig. 1B; data not shown). At the tissue level, signs of telomere dysfunction were readily evident as multisystem degenerative pathologies, particularly in organ systems with high proliferative activity. The GI crypts of late-generation mTerc−/− Ink4a/Arf+/+ and −/− mice exhibited equally high levels of apoptosis (Fig. 1C), comparable to that in late-generation mTerc−/− tissues with intact p53 function (7, 30). Accordingly, immunohistochemical analysis revealed equally robust p53 expression in late-generation mTerc−/− Ink4a/Arf+/+ and −/− crypts (Fig. 1D). These findings also agree well with previous observations of a robust γ irradiation-induced p53 activation pattern in the small intestine and spleen of p19^ARF−/− mice (25). Similarly, in the testes, we observed the previously reported testicular atrophy and associated apoptotic depletion of germ cells in late-generation mTerc−/− mice (32), with comparable reduction in testes weights (P = 0.4131) and depletion of germ cells in all three Ink4a/Arf genotypes (Fig. 1E). When placed in the context of preserved p53 signaling in the late-generation mTerc−/− Ink4a/Arf−/− intestines, these observations indicate that p16^Ink4a and p19^Arf are dispensable for signaling from dysfunctional telomeres to p53 in these adult tissues in vivo.

Telomere Dysfunction Impedes Tumorigenesis in Ink4a/Arf Mutant Mice and Has No Impact on the Ink4a/Arf Mutant Tumor Spectrum.

Next, the incidence and spectrum of spontaneous tumors was monitored in various cohorts of mice wild type or mutant for mTerc and/or Ink4a/Arf. The early generation mTerc mice of the various Ink4a/Arf genotypes showed tumor incidence and median latencies similar to those reported previously for Ink4a/Arf wild-type and mutant mice (33) (Fig. 2A). Specifically, 84% of early generation mTerc Ink4a/Arf+/− mice and 98% of early generation mTerc Ink4a/Arf−/− mice succumbed to tumors with a median tumor-free latency of 71.6 and 38.6 wk, respectively. Relative to early generation mTerc Ink4a/Arf+/− and −/− controls, late-generation mTerc−/− Ink4a/Arf+/− and −/− mice showed a marked decrease in percent tumor death and an increase in tumor-free latency: from 84% to 20% and 71.6–104.9 wk for the late-generation mTerc−/− Ink4a/Arf+/− mice and from 98% to 43% and 38.6–65.3 wk for the late-generation mTerc−/− Ink4a/Arf−/− mice (Fig. 2 B and C).

Fig. 2.

Telomere dysfunction suppresses the incidence and delays the onset of tumors from the mTerc Ink4a/Arf mutant mice. (A) Kaplan–Meier analysis of tumor incidence of early generation mTerc mice with differing Ink4a/Arf genotypes (n = 34 for Ink4a/Arf+/+; n = 50 for Ink4a/Arf+/−; n = 62 for Ink4a/Arf−/−). (B) Kaplan–Meier analysis showing an increase of tumor latency in late-generation mTerc Ink4a/Arf+/− mice (n = 50 for early generation; n = 72 for late generation). (C) Kaplan–Meier analysis showing an increase of tumor latency in late-generation mTerc Ink4a/Arf−/− mice (n = 62 for early generation; n = 64 for late generation). Mice with multiple tumors were counted once.

With regard to tumor spectrum, early generation mTerc Ink4a/Arf mutant mice with intact telomeres succumbed largely to the previously reported Ink4a/Arf mutant profile of histiocytic sarcomas (34),** soft tissue sarcomas, and lymphomas (33, 35, 36). Comparison of early and late-generation mTerc−/− Ink4a/Arf+/− mice showed no significant changes in tumor spectrum. Specifically, the early generation mTerc Ink4a/Arf+/− mice sustained 35% histiocytic sarcomas, 30% soft tissue sarcomas, and 19% lymphomas, compared with the late-generation mTerc−/− Ink4a/Arf+/− spectrum of 39% histiocytic sarcomas, 23% soft tissue sarcomas, and 23% lymphomas (Fig. 3). Similarly, for the Ink4a/Arf−/− cohorts, early generation mTerc Ink4a/Arf−/− mice showed 57% histiocytic sarcomas, 24% soft tissue sarcomas, and 11% lymphomas, compared with the late-generation mTerc−/− Ink4a/Arf−/− profile of 59% histiocytic sarcomas, 28% soft tissue sarcomas, and 10% lymphomas (Fig. 3). In summary, telomere dysfunction serves only to decrease the incidence and delay the onset of spontaneous sarcomas and lymphomas and has no effect on tumor spectrum. These findings are in striking contrast to the impact of p53 mutation in the mTerc−/− model and suggest that p16^Ink4a and p19^Arf do not function as physiologically relevant sensors or mediators of the telomere dysfunction response during the development of these spontaneous cancers.

Fig. 3.

Telomere dysfunction did not change the tumor spectrums of the mTerc Ink4a/Arf mutant mice. Tumors were diagnosed histologically and graphed by percentages. Multiple tumors from the same mouse were counted as separate incidences (n = 38 for early generation mTerc Ink4a/Arf−/− mice; n = 27 for early generation mTerc Ink4a/Arf+/− mice; n = 13 for late-generation mTerc−/− Ink4a/Arf+/− mice; n = 29 for late-generation mTerc−/− Ink4a/Arf−/− mice).

Further analysis showed that there were no significant changes in the overall pathology, proliferative and apoptosis indices, degree of vascularization, and metastasis between the tumors from the early and late-generation mTerc Ink4a/Arf mutant mice [supporting information (SI) Fig. 6]. These observations indicate that, once tumors become established in late-generation mTerc−/− Ink4a/Arf mutant mice, the initial telomere dysfunction does not appear to exert a discernable impact on the biological behavior of these tumors.

Induction of Adaptive Responses to Telomere Dysfunction in the Ink4a/Arf Mutant Tumors.

The lack of a biological impact on advanced tumors, coupled with the importance of the p53 DNA damage checkpoint in modulating telomere dysfunction, and induction of p53 in the GI tract of late-generation mTerc−/−Ink4a/Arf−/− animals, prompted functional assessment of the p53-dependent checkpoint in the tumors from the late-generation mTerc−/− Ink4a/Arf−/− mice. Doxorubicin treatment of early passage tumor cell lines revealed that five of five early generation mTerc Ink4a/Arf−/− tumors maintained intact p53 response evidenced by p53 Ser-18 phosphorylation (37, 38) and associated induction of p21^CIP1. In contrast, 6 of 10 of the doxorubicin-treated late-generation mTerc−/− Ink4a/Arf−/− cultures failed to show p53 Ser-18 phosphorylation and p21^CIP1 induction (Fig. 4A). Sequencing of commonly mutated exons 5 through 9 (39) of the p53 gene by using DNA isolated from the frozen tumors did not reveal any mutations. Western blot analyses of frozen tumor extracts showed increased Mdm2 levels in tumors that had greatly attenuated p21 induction in response to DNA damage (data not shown), suggesting that the p53-dependent DNA damage checkpoint is compromised by overexpression of Mdm2, the major inhibitor of p53 (40, 41). The pressure to deactivate the p53-dependent DNA damage checkpoint in late-generation mTerc−/− Ink4a/Arf−/− tumors is consistent with the importance of p53 in the telomere checkpoint and further supports the view that p16^Ink4a and p19^Arf are not functionally equivalent to p53 and play negligible roles in telomere dynamics in murine tumorigenesis.

Fig. 4.

p53 checkpoint function and activation of ALT in the soft tissue sarcomas. (A) Compromised induction of p53 and p21, and Ser-18 phosphorylation of p53 in response to doxorubicin in a subset of the soft tissue sarcoma cell lines from late-generation mTerc−/− Ink4a/Arf−/− mice (L) compared with those from early generation mTerc Ink4a/Arf−/− mice (E). Cell lines were treated with 0.5 μg/ml doxorubicin and harvested at 0 and 16 h. Actin was used as loading control. (B) Telomere repeat fragment Southern of mTerc Ink4a/Arf−/− tumor cell lines showing activation of ALT in late-generation mTerc−/− Ink4a/Arf−/− tumors (L). Note increase in telomere length heterogeneity in late-generation mTerc−/− Ink4a/Arf−/− tumor cell lines (L). (C) Representative metaphase spreads of one early generation and two different late-generation sarcoma cell lines showing chromosomal fusions and heterogeneous telomere length in late-generation tumor cell lines. Late-generation mTerc−/− Ink4a/Arf−/− MEF with homogeneous short telomeres shown for comparison. Chromosomes were stained with DAPI (blue), and telomeric DNA was detected by FISH with Cy3-conjugated T₂AG₃ peptide nucleic acid probe (red). (D) Telomere length analysis by quantitative telomeric FISH of an early generation sarcoma cell line and two different late-generation sarcoma cell lines. Percent telomeres were plotted against TFU. Mean TFU and percent signal-free ends were 103.46, 181.94, and 141.96, and 0.9, 9.3, and 8.7%, respectively.

Another adaptive response to telomere dysfunction is the activation of the telomerase-independent alternative lengthening of telomere (ALT) pathway (42). Analysis of telomere lengths of the tumor cell lines by telomere repeat fragment Southern revealed that late-generation mTerc−/− Ink4a/Arf−/− cell lines have heterogeneous telomere lengths characteristic of ALT (Fig. 4B). Cytogenetic analysis showed heterogeneous telomere signals with chromosomal fusions lacking telomere signals at the sites of fusion along with very bright telomeres (Fig. 4C). Quantitative telomeric FISH showed that there was an increase in signal-free ends from 0.9% in early generation to 8.7–9.3% in late-generation mTerc−/− Ink4a/Arf−/− tumor cell lines (Fig. 4D). However, late-generation mTerc−/− Ink4a/Arf−/− tumor cell lines also have mean telomere lengths [181.84 and 141.96 telomere fluorescence units (TFUs)] that are comparable in length to the early generation tumor mTerc Ink4a/Arf−/− cells (103.46 TFUs), pointing to activation of ALT in these late-generation mTerc−/− Ink4a/Arf−/− tumors. Furthermore, telomere signals in late-generation mTerc−/− Ink4a/Arf−/− tumor cells were more heterogeneously distributed compared with the more clustered values in early generation mTerc Ink4a/Arf−/− tumor cells. This late-generation mTerc−/− Ink4a/Arf−/− telomere signature also contrasts with the homogenously dim signals in late-generation mTerc−/− Ink4a/Arf−/− MEFs that have not activated ALT (Fig. 4C; SI Fig. 7). These results, along with the increased tumor latency and decreased tumor incidence in mice with telomere dysfunction, show that late-generation mTerc−/− Ink4a/Arf mutant tumor cells experience biological pressures to incur p53 loss and activate ALT to overcome the suppressive effects of telomere dysfunction on tumorigenesis.

Telomere Dysfunction Attenuates the Immortalization and Growth Rate of Cultured Ink4a/Arf-Null Cells.

p19^Arf and, to a lesser extent, p16^Ink4a have been shown to play a role in the activation of cellular senescence and immortalization on passaging of primary mouse cells in culture (26, 43, 44). In MEFs, p19^Arf and p16^Ink4a contribute to the growth arrest incurred by telomere deprotection (14, 27). We passaged MEFs and astrocytes under the 3T3 protocol to assess their growth rates as a function of Ink4a/Arf and telomere status in vitro (Fig. 5). The G₀ and G₄ mTerc Ink4a/Arf+/+ MEFs and astrocytes senesced after 10 and 2 population doublings, respectively. The G₀ mTerc Ink4a/Arf−/− MEFs were immortal as reported previously (33). Consistent with previous reports (27), G₄ mTerc−/− Ink4a/Arf−/− MEFs showed an intermediate phenotype of decreased growth rates after 10 population doublings with steady slow proliferation thereafter, indicating that telomere dysfunction suppresses the growth rate, but not the immortalization potential, of Ink4a/Arf−/− MEFs. G₀ mTerc Ink4a/Arf−/− astrocytes showed continued proliferation as previously reported (45), whereas the G₄ mTerc−/− Ink4a/Arf−/− astrocytes senesced after 2.5 population doublings, indicating that telomere dysfunction constrains the immortalization potential of Ink4a/Arf−/− astrocytes. The capacity of telomere dysfunction to suppress the immortalization potential and/or growth rates of these cultured Ink4a/Arf−/− cell types reinforces the view of a modest or negligible role for p16^Ink4a and p19^Arf in the telomere checkpoint response. In addition, telomere dysfunction was able to activate p53 in late-generation mTerc−/− Ink4a/Arf+/+ and −/− ear fibroblasts (SI Fig. 8), confirming that signaling to p53 is independent of p16^Ink4a and p19^Arf.

Fig. 5.

Telomere dysfunction partially suppresses the growth and immortalization of Ink4a/Arf−/− MEFs and astrocytes. (A) 3T3 growth assay of G₀ and G₄ mTerc Ink4a/Arf MEFs. (B) 3T3 growth assay of G₀ and G₄ mTerc Ink4a/Arf mouse embryonic astrocytes (n = 2 each for G₀ mTerc+/− Ink4a/Arf+/+ and Ink4a/Arf−/−; n = 3 each for G₄ mTerc−/− Ink4a/Arf+/+ and Ink4a/Arf−/− for both MEFs and astrocytes).

Discussion

Studies on cell senescence have underscored the importance of Rb and p53 signaling in response to telomere dysfunction (1, 46). In this study, analysis of the mTerc Ink4a/Arf mouse model enabled further dissection of potential upstream signaling components in the telomere dysfunction checkpoint, specifically the contribution of p16^Ink4a and p19^Arf. We demonstrate that p16^Ink4a and p19^Arf deficiency does not attenuate p53 activation or the degenerative conditions that characterize mice with systemic telomere dysfunction. In addition, telomere dysfunction delayed tumor formation and had no impact on tumor spectrum in the Ink4a/Arf mutant mice. Finally, established tumors of late-generation mTerc−/− Ink4a/Arf−/− mice showed robust malignant phenotypes, yet acquired frequent loss of p53 function and activation of ALT. We conclude that, in contrast to p53, p16^Ink4a and p19^Arf do not impact on the telomere dysfunction checkpoint in vivo in both normal and neoplastic cells.

Telomeres are vital to the maintenance of normal cellular growth and viability (47), and loss of telomere function can influence tumorigenesis in a manner that is highly dependent on cell type and genotype. Telomere dysfunction appears to engage the p53-dependent double-stranded break DNA damage pathway. The role of p53 in telomere dysfunction signaling is well established in human cells and mice, as evidenced in the induction of p53 in late-generation mTerc−/− mouse tissues and alleviation of degenerative phenotypes in late-generation mTerc−/− p53−/− mice (1–4, 46). The lack of rescue of degenerative phenotypes in late-generation mTerc−/− Ink4a/Arf mutant mice is consistent with results in late-generation mTerc−/− Atm−/− mice, where telomere dysfunction activated p53 in an Atm-independent manner (7, 8).

Telomere dysfunction has been shown to suppress carcinogen-induced tumorigenesis in the Ink4a/Arf mutant mice (28), findings consistent with previous studies establishing that telomere dysfunction acts as a tumor suppressor when p53 function is intact (7, 8, 30, 48). When the p53 checkpoint is compromised, unchecked genomic instability caused by telomere dysfunction accelerates tumorigenesis as seen in late-generation mTerc−/− p53 mutant mice (4, 5). Despite the decrease in incidence and increase in latency of tumorigenesis in late-generation mTerc−/− Ink4a/Arf mutant mice, there were no significant differences in histology and malignant grade of these tumors when compared with those from Ink4a/Arf mutant mice with intact telomeres. Therefore, we propose that telomere dysfunction acts to suppress tumor initiation rather than progression in late-generation mTerc−/− Ink4a/Arf mutant mice. This finding differs from that of the mTerc Apc^Min model, wherein telomere dysfunction promotes the initiation of intestinal carcinomas but inhibits their progression (30). This difference could relate to the rate-limiting role of wild-type Apc loss in initiation of intestinal neoplasia (49), in which increased genomic instability in late-generation mTerc−/− Apc^Min intestinal epithelium would serve to promote this early and rate-limiting genetic event. Another formal, albeit remote, possibility may relate to attenuation of the telomere checkpoint response in cells haploinsufficient for Apc^Min. Finally, the suppression of tumor initiation in late-generation mTerc−/− Ink4a/Arf mutant mice may be because of apoptotic elimination or senescence of cellular compartments at risk for tumorigenesis in the setting of Ink4a/Arf mutation (i.e., progenitor cells of histiocytic and soft tissue sarcomas).

Telomere dysfunction has been shown to dramatically affect the tumor spectrum of late-generation mTerc−/− p53 mutant mice (5). The ongoing genomic instability permitted by p53 deficiency allowed for emergence of tumors with more complex karyotypes, reflected in a shift in tumor spectrum, from lymphomas and sarcomas in a setting of intact telomeres toward epithelial cancers in mice with telomere dysfunction (5). In contrast, the tumor spectrum of Ink4a/Arf mutant mice was not altered by telomere dysfunction. This may be because of an intact p53 checkpoint that eliminates genomically unstable cells and/or the emergence of ALT, which is expected to reduce the number of would-be cancer cells experiencing the breakage–fusion–bridge process and accumulation of regional amplifications and deletions (6). Both processes would function to impede the accumulation of a greater number of somatic mutations thought to be required for the development of epithelial cancers (50, 51).

p16^Ink4a and p19^Arf have prominent roles in the early stress-induced stage of cellular senescence in human cells, which is independent of telomere signaling. The lack of effect on in vivo phenotypes of p16^Ink4a and p19^Arf in this study shows that dysfunctional telomeres do not substantively signal to Rb and p53 via p16^Ink4a and p19^Arf in the tissues examined. This is again consistent with the minimal role of these genes in DNA double-stranded break signaling. The importance of p53 and its role in telomere signaling is also evident in the adaptive mutations in the late-generation mTerc−/− Ink4a/Arf−/− tumor cell lines. Although the p53 checkpoint is maintained in tumors with intact telomeres, a significant proportion of tumors with telomere dysfunction show defective p53 signaling in response to DNA damage. Furthermore, the activation of ALT in tumors with telomere dysfunction may serve as another avenue toward alleviating the antitumorigenic effects of telomere dysfunction, as we have observed previously in murine cell-based transformation systems (52).

Analysis of MEFs and astrocytes in vitro also showed that telomere dysfunction can inhibit cell growth and/or immortalization potential brought about by the loss of p16^Ink4a and p19^Arf. That is, consistent with the ability of telomere dysfunction in suppression of tumorigenesis in vivo, telomere dysfunction is able to induce an antiproliferative signal that is independent of p16^Ink4a and p19^Arf in vitro.

This study emphasizes the importance of p53 activation in the modulation of telomere dysfunction in vivo. The effect of telomere dysfunction on tumorigenesis and survival depends primarily on its ability to induce the p53 pathway. In the setting of compromised p53 signaling and defective telomeres, survival and expansion of cells experiencing genomic instability increases the probability of accumulation of protumorigenic mutations, which manifests as complex cancer karyotypes. In the presence of intact p53 signaling, the genomic instability caused by telomere dysfunction leads to the suppression of tumorigenesis, where premalignant cells have to accrue adaptive mutations to either bypass the p53 checkpoint or curb the rampant genomic instability by activating alternative pathways of telomere maintenance. Elimination of p16^Ink4a and p19^Arf does not mitigate the tumor suppressive phenotype of telomere dysfunction nor the activation of the p53 signaling pathway. These genetic studies may inform the application of antitelomerase therapy to specific human cancer genotypes, favoring the use of such therapies in tumors with intact p53 signaling on DNA damage.

Materials and Methods

Immunohistochemistry.

Standard immunohistochemical techniques were performed on formalin-fixed tissues by using mouse anti-proliferating cell nuclear antigen (Calbiochem, San Diego, CA) and rabbit anti-p53 (CM5; Vector Laboratories, Burlingame, CA). TUNEL assays were performed according to manufacturer's instructions on paraffin sections (ApopTag detection kit; Intergen, Purchase, NY).

Western Blot Analyses.

Samples were lysed in EBC buffer, and 40 μg of lysate per lane was resolved by SDS/PAGE and transferred onto PVDF membranes. The antibodies used were rabbit anti-p53 (CM5; Vector Laboratories), rabbit anti-phospho-p53 (Ser-15; Cell Signaling Technology, Danvers, MA), rabbit anti-p21 (M-19; Santa Cruz Biotechnology, Santa Cruz, CA), mouse anti-Mdm2 (2A10; Calbiochem), and goat anti-actin (I-19; Santa Cruz Biotechnology).

Quantitative Telomeric FISH.

Metaphase chromosomes and FISH of telomeric sequences with Cy-3 labeled T₂AG₃ peptide nucleic acid probe was performed as described (52). Ten metaphases of each sample were captured on the same day giving 1,538–2,308 telomeres per sample. TFU was analyzed by TFL-Telo (kind gift from P. Lansdorp, University of British Columbia, Vancouver, BC, Canada). Chromosomal ends at sites of fusion were included as signal-free ends. Because of the aneuploid nature of the tumor cell lines, percent telomeres were plotted.

Pulse-Field Gel Electrophoresis.

A total of 5 × 10⁴ tumor cells was embedded into 0.8% agarose plugs, digested overnight with 20 units of HinfI and RsaI, and analyzed as described (52).

Cell Culture and 3T3 Assay.

MEFs and astrocytes were prepared from day 13.5 embryos derived from G₀ and G₃ mTerc Ink4a/Arf mice and analyzed as described (52, 53).

p53 Mutational Analysis.

DNA was isolated from frozen tumor tissue, and the p53 gene was sequenced as described (54).

Statistical Analysis.

Statistical analyses were done by using GraphPad Prism 4 (GraphPadSoftware, San Diego, CA). Tumor incidence was plotted by using the Kaplan–Meier analysis. Only animals possessing histologically verified cancer were scored as an event. Statistical significance was measured by using the log-rank test.

Abbreviations

ABI

anaphase bridge index

ALT

alternative lengthening of telomere

GI

gastrointestinal

MEF

mouse embryonic fibroblast

TFU

telomere fluorescence unit.