Research ArticleCancer

Differential p53-Independent Outcomes of p19Arf Loss in Oncogenesis

+ See all authors and affiliations

Science Signaling  18 Aug 2009:
Vol. 2, Issue 84, pp. ra44
DOI: 10.1126/scisignal.2000053


One reported function of the tumor suppressor p19Arf is to stabilize p53, providing a critical checkpoint in the response to oncogenic insults. Acute loss of Pten leads to an increase in the abundance of p19Arf, p53, and p21 proteins as part of a fail-safe senescence response. Here, we report that loss of p19Arf in prostate epithelium does not accelerate—but rather partially inhibits—the prostate cancer phenotype of Pten-deficient mice. Moreover, cellular senescence and a further decrease in the number of pre-neoplastic glands were observed in prostates of the Pten-p19Arf double-mutant mice. In both prostate epithelium and primary mouse embryo fibroblasts (MEFs), the increase in p53 protein abundance found upon loss of Pten was unaffected by the simultaneous loss of p19Arf. However, in contrast to that in the prostate epithelium, p19Arf deficiency in MEFs lacking Pten abolished cell senescence and promoted hyperproliferation and transformation despite the unabated increase in p53 abundance. Consistent with the effect of p19Arf loss in Pten-deficient mouse prostate, we found that in human prostate cancers, loss of PTEN was not associated with loss of p14ARF (the human equivalent of mouse p19Arf). Collectively, these data reveal differential consequences of p19Arf inactivation in prostate cancer and MEFs upon Pten loss that are independent of the p53 pathway.


ARF (alternative reading frame), one of two products of the INK4a/ARF locus, plays a critical role in suppressing tumor initiation and progression, and accumulates in response to such oncogenic events as loss of Pten or activation of Ras or Myc (13). p19Arf-deficient mice are predisposed to development of sarcomas, lymphomas, and pulmonary and mammary adenocarcinomas (4), but not prostate cancer. ARF (p14ARF in human and p19Arf in mouse) is functionally coupled to p53 through its inhibition of Mdm2-mediated p53 degradation, and overexpression of p19Arf leads to p53-dependent growth arrest and cellular senescence in vitro and in vivo in some tissues (2, 5, 6). The possibility that ARF might act independently of p53 has been suggested on the basis of binding of ARF to proteins other than MDM2 (murine double minute protein 2), although its p53-independent functions remain controversial and poorly understood (6).

PTEN (phosphatase and tensin homolog deleted from chromosome 10) was identified as a tumor suppressor gene that is frequently mutated or deleted in human cancers, including prostate cancer, in which up to 70% of cases show loss or mutation of one allele of PTEN at presentation (79). Complete germline deletion of Pten in mice results in embryonic lethality at embryonic day 6.5 (E6.5) to E9.5 (10), highlighting the importance of Pten during mouse embryonic development. Mice lacking a single Pten allele survive and develop various tumors in adulthood (10). Reduction of PTEN protein leads to activation of phosphatidylinositol-3-OH kinase (PI3K) and thereby to hyperactivation of the AKT-mTOR (mammalian target of rapamycin) signaling pathway (1113). Surprisingly, however, short-term inactivation of both alleles of Pten triggers a cellular senescence response that suppresses cancer progression through activation of the ARF-p53 pathway (1, 14, 15). Bypassing senescence through concomitant inactivation of Pten and p53 leads to a lethal prostate cancer phenotype in mice (1). In agreement with the mouse models, PTEN loss–induced senescence also occurs in human cells (16). Overall, these findings suggest that p19Arf, through its inhibition of Mdm2-mediated p53 degradation, plays a critical role in opposing prostate tumorigenesis in response to loss of Pten.

Here, we identify unexpected roles for ARF in cellular senescence, transformation, and cancer initiation that are uncoupled from the p53 pathway.


Loss of p19Arf does not accelerate prostate tumorigenesis

We previously showed that concomitant inactivation of Pten and Trp53 accelerates tumor progression in Pten-deficient mice by overcoming Pten loss–induced cellular senescence and, as a result, double-mutant mice with specific inactivation of Pten and Trp53 in prostate epithelium invariably die of prostate cancer at 7 months of age (1). p19Arf accumulates after short-term loss of Pten both in mouse embryo fibroblasts (MEFs) and in early prostatic intraepithelial neoplasia (PIN) lesions, in which the cellular senescence program is active (1). This, together with the notion that p19Arf promotes the stabilization of p53, prompted us to examine the role of p19Arf in prostate tumorigenesis driven by loss of Pten. We generated prostate-specific [probasin (PB)-Cre4–driven] Pten-p19Arf double-mutant mice following the same strategy described previously (1). Briefly, we crossed PtenloxP/+;PB-Cre4 transgenic mice with p19Arf−/− mice to obtain the p19Arf−/−;PtenloxP/loxP;Pb-Cre4 mice, hereafter referred to as p19Arf−/−;Ptenpc−/−. To minimize the effects of variations in genetic background among individual genotypes, a cohort of wild-type (Wt), p19Arf−/−, Ptenpc−/− single-mutant, and p19Arf−/−;Ptenpc−/− double-null mice were generated from F2 offspring and intercrossed for more than three generations (fig. S1A).

Because p19Arf and p53 are functionally coupled (6, 17), we reasoned that p19Arf−/−;Ptenpc−/− double-mutant mice would, at least in part, phenocopy the cancer acceleration observed in Pten-Trp53 double-mutant mice. We followed a cohort of 77 animals over a period of 17 months (Fig. 1A), subjecting them to biweekly magnetic resonance imaging (MRI) analysis as described previously (1). Unlike the lethal prostate cancer phenotype of Pten-Trp53 double-null mice, concomitant loss of Pten and p19Arf did not affect disease survival relative to that in p19Arf−/− mice (Fig. 1A; nonsignificant differences between p19Arf−/− and p19Arf−/−;Ptenpc−/− compound mutant mice; P > 0.05, Kaplan-Meier survival test). Indeed, all p19Arf−/−;Ptenpc−/− double-mutant mice died by 13 months of age of tumors typically found in p19Arf mutant mice, rather than of prostate cancer. Overt tumor masses in the prostate region were detected by 6 months of age in both Ptenpc−/− mutant mice and p19Arf−/−;Ptenpc−/− double-mutant mice, but not in p19Arf−/− or in Wt cohort (Fig. 1, B and C). The prostate lobes of Ptenpc−/− mutant mice and p19Arf−/−;Ptenpc−/− double-mutant mice were visibly enlarged to a similar extent. Prostate tumors from p19Arf−/−;Ptenpc−/− double-mutant mice were comparable in size to those from Ptenpc−/− mutant mice by MRI visualization (fig. S1B) and after biopsy (anterior prostates; Fig. 1B). The average tumor weight of p19Arf−/−;Ptenpc−/− double-mutant mice was indistinguishable from that in Ptenpc−/− mutant mice (Fig. 1C; 339 ± 52 and 336 ± 87 mg, respectively; n = 10 mice, P = 0.91 > 0.05, t test). These results show that p19Arf does not play a major tumor suppressive role in Pten loss–driven prostate tumorigenesis and highlight a phenotypic discordance between Trp53 loss and p19Arf loss in progression of prostate cancer in Pten-null mice.

Fig. 1

Loss of p19Arf does not accelerate prostate tumorigenesis. (A) Cumulative survival analysis (Kaplan-Meier plot) for p19Arf−/− (blue line), Ptenpc−/− (black line), and p19Arf−/−;Ptenpc−/− (red line). (B) Actual sizes of representative tumors from anterior prostates (AP) of Wt, p19Arf−/−, Ptenpc−/−, and p19Arf−/−;Ptenpc−/− double-mutant mice at 6 months of age. (C) Comparison of AP tumor masses from Ptenpc−/− and p19Arf−/−;Ptenpc−/− mice at 6 months of age indicates that p19Arf deficiency does not accelerate the tumorigenesis in Pten-deficient mice (n = 10 mice for each group, P = 0.91 > 0.05, t test).

Loss of p19Arf constrains prostate cancer driven by Pten deficiency

To investigate the functional consequences of p19Arf loss in Pten loss–driven prostate tumorigenesis, prostates were collected from 11- to 16-week-old double-mutant mice and histopathological analysis [hematoxylin and eosin (H&E) staining] was performed on all three prostatic lobes [anterior prostate (AP), ventral prostate (VP), and dorsolateral prostate (DLP)]. Prostates from p19Arf−/− mice, like those from Trp53 mutant mice, failed to show pathological alterations such as hyperplasia or PIN; glandular architectures of p19Arf−/− prostates were indistinguishable from those of age-matched Wt mice (n = 10 mice; Fig. 2A). Indeed, no visible alterations were observed in the prostate morphology of p19Arf−/− mice over a period of 1 year.

Fig. 2

Loss of p19Arf constrains prostate cancer progression. (A) Histopathological analysis (H&E staining) of ventral prostates in Wt, p19Arf−/−, Ptenpc−/−, and p19Arf−/−;Ptenpc−/− double-mutant mice at 11 to 16 weeks of age. (B) Quantification of HG-PIN in Wt, p19Arf−/−, Ptenpc−/− and p19Arf−/−;Ptenpc−/− double-mutant mice at 11 to 16 weeks of age (n = 4 mice for each group). (C) IHC staining of pAkt (indicated by arrows) in adjacent sections from ventral prostates of (A). Insets indicate the lower magnification. Error bars in (B) represent SD for a representative experiment performed in triplicate. (D) Ki-67 in adjacent sections from ventral prostates of (A) and an inset showing the Ki-67 positivity in PIN lesions in p19Arf−/−;Ptenpc−/− double-mutant mice prostate.

Analysis of the p19Arf−/−;Ptenpc−/− double-null mutant mice revealed a diminution of the Ptenpc−/− phenotype (Fig. 2, A and B). The VP of double-null mutant mice (but not the AP and DLP) showed a decreased incidence of high-grade PIN (HG-PIN) in many individual glands compared with those of Ptenpc−/− mice (Fig. 2A, right panels). Many glands still showed Wt-like glandular features with a single layer of epithelial cells (Fig. 2A and insets) compared to multicellular layered or cribiform architecture of epithelial cells in glands with HG-PIN. Quantification revealed a significant decrease in the percentage of glands with HG-PIN in the VP of p19Arf−/−;Ptenpc−/− double-null mice compared with those from Ptenpc−/− mice (Fig. 2B; 93 and 49% HG-PIN, respectively; P = 0.0024 < 0.005), whereas no marked difference was found in the AP and DLP between these two genotypes. Similarly, p19Arf−/−;Ptenpc−/− double-null mice showed a reduced epithelial component in tumors compared to Pten-null mutant mice at 6 months of age (fig. S1C). Immunohistochemical (IHC) analysis indicated that p19Arf loss did not alter Pten-PI3K pathway signaling, as determined by measuring Akt activation (phospho-Akt) and its membrane recruitment in p19Arf−/−;Ptenpc−/− double-null mutant mice (Fig. 2C). Indeed, epithelial cells positive for phospho-Akt were detected as part of monolayers from p19Arf−/−;Ptenpc−/− double-null prostates, indicating that progression to PIN was impaired in p19Arf−/−;Ptenpc−/− compound mutant mice (Fig. 2B) despite Akt activation consequent to the complete loss of Pten. Additionally, cell proliferation as assayed by Ki-67 immunostaining was markedly decreased in glands in which PIN diminution was observed (Fig. 2D). Together, these results support the notion that loss of p19Arf fails to promote prostate cancer driven by loss of Pten.

Loss of p19Arf does not affect accumulation of p53 protein and cellular senescence in mouse Pten-null prostates

Next, we investigated the effect of p19Arf loss on increased p53 abundance and the senescence response in Pten-null prostates. To do so, prostates were collected from p19Arf−/−;Ptenpc−/− double-mutant mice at 11 weeks of age to evaluate the activity of senescence-associated β-galactosidase (SA-β-gal). Ptenpc−/− mutants showed a strong senescence response as determined by SA-β-gal activity (Fig. 3A). p19Arf−/−;Ptenpc−/− double-null mutant prostates retained a senescence response comparable to that of Ptenpc−/− mutants (Fig. 3A and right panels for a representative example from anterior prostate lobes; arrows denote senescent cells), whereas prostates from age-matched Wt and p19Arf−/− mice showed low to undetectable SA-β-gal activity (Fig. 3A). Indeed, the percentage of senescent cells in p19Arf−/−;Ptenpc−/− double-null mutant mice prostates was indistinguishable from that in Ptenpc−/− (Fig. 3B; 19.6 ± 5.3 and 17.8 ± 4.3%, respectively; P = 0.33 > 0.05, t test), both of which had nearly 15 times the number of senescent cells as did Wt prostates (1.3 ± 0.87%). The increase in p53 protein and percentage of p53-positive cells was unaffected in the prostatic epithelium of p19Arf−/−;Ptenpc−/− double-null mutant mice compared to that in Ptenpc−/− mice (Fig. 3, C and D). Moreover, p53 function was intact in compound mutant prostates, as determined by an increase in the abundance of p21, a transcriptional target of p53 (fig. S2).

Fig. 3

In vivo p19Arf-p53 uncoupling and cellular senescence contribute to cancer suppression in compound mutant mice. (A) Histopathology and senescence analysis of 11-week-old prostates, stained as indicated for H&E, and senescence-associated β-galactosidase (β-gal) in anterior prostates (AP). (B) Quantification (percentage of cells) of the β-gal staining seen on (A) sections from 11-week-old mice (19.6 ± 5.3% for p19Arf−/−;Ptenpc−/− double-mutant mice compared with 17.8 ± 4.3% for Ptenpc−/− mice; P = 0.33 > 0.05, t test). (C) p53 staining on AP sections from 11-week-old mice; arrows denote the positive staining of p53 in nucleus of epithelial cells. (D) Quantification of the p53 staining seen on (C) sections from 11-week-old mice. Senescence and histopathology analysis of 11-week-old prostates. Representative sections from three mice were counted for each genotype. Error bars in (B) and (D) represent SD for a representative experiment performed in triplicate.

Our results therefore show that after loss of Pten, p53 accumulation and execution of senescence are independent of p19Arf, suggesting that in prostate cancer, the functions of p19Arf are unrelated to the p53 pathway.

Loss of p19Arf bypasses senescence and promotes transformation independently of p53 in Pten-null MEFs

We used primary MEFs to investigate the independence of p19Arf and p53 signaling and the biological consequences of this uncoupling. First, we determined whether the increased abundance of p53 protein and senescence response that occur with Pten loss were affected by p19Arf loss in MEFs. Following the same strategy reported previously (1), we prepared a series of conditional p19Arf-Pten (p19Arf−/−;PtenΔ/Δ) MEFs and acutely deleted Pten through retroviral delivery of Cre-PURO-IRES-GFP (Cre–puromycin-resistant–internal ribosome entry site–green fluorescent protein). In the p19Arf-deficient context, virus-mediated expression of Cre recombinase led to efficient excision of LoxP alleles of conditional Pten and, subsequently, an increase in Akt phosphorylation after Pten inactivation in MEFs (Fig. 4A). Consistent with the in vivo prostate data, increased abundance in p53 and p21 proteins after loss of Pten was unaffected in p19Arf−/−;PtenΔ/Δ MEFs (Fig. 4, A and C), strengthening the notion that p19Arf is not required for the increase in p53 abundance that follows loss of Pten. Quantification revealed that p53 protein abundance in p19Arf−/−;PtenΔ/Δ MEFs was similar to that in PtenΔ/Δ MEFs, around 2.5 times that of Wt (Fig. 4B). Moreover, the half-life of p53 in p19Arf−/−;PtenΔ/Δ cells was similar to that in Wt cells (Fig. 4D), further supporting the notion that p53 increase induced by Pten deletion is independent of the inhibition of Mdm2-mediated p53 degradation by p19Arf. The half-life of p21, a downstream target of p53, was also comparable to that in Wt cells (fig. S3A), suggesting that p53 signaling is unaffected by loss of p19Arf.

Fig. 4

p19Arf-p53 uncoupling in primary MEFs. (A) Western blot of lysates of primary MEFs with indicated antibodies in Wt, PtenΔ/Δ, and p19Arf−/−-PtenΔ/Δ double-null cells. β-Actin is used as a loading control. (B) Quantification of protein abundance in primary MEFs for p53, p16, and pRB from (A). (C) Western blotting of cellular lysates of primary MEFs with indicated antibodies in p19Arf−/−, p19Arf−/−-PtenΔ/+, and p19Arf−/−-PtenΔ/Δ double-null cells. (D) Inhibition of protein synthesis by cycloheximide (CHX) combined with Western blot at the indicated times (minutes) shows that the half-life of p53 in p19Arf−/−-PtenΔ/Δ double-null MEFs is comparable to that in Wt MEFs. Right panel: quantification of p53 half-life from the Western blots in the left panel normalized to β-actin. Blue circles, Wt-Cre; red squares, p19Arf−/−-PtenΔ/Δ-Cre. (E) Cellular senescence assay of Wt, p19Arf−/−, PtenΔ/Δ, p19Arf−/−-PtenΔ/Δ double-null cells. (F) Growth curves of primary MEFs, infected with retroviral Cre (under puromycin selection) followed over a 6-day period: p19Arf−/−-PtenΔ/Δ double-null (red squares), PtenΔ/Δ (black squares), p19Arf−/− (black circles), and Wt cells (blue triangles). (G) Transformation assay as determined by colony formation in soft agar from (E); PtenΔ/Δ;p53Δ/Δ double-null cells served as positive control. Error bars for (B), (E), (F) and (G) indicate SD for a representative experiment performed in triplicate.

Despite a sharp increase in p53 abundance, loss of p19Arf resulted in inhibition of the senescence response (Fig. 4E); SA-β-gal activity was undetectable in p19Arf−/−;PtenΔ/Δ cells (Fig. 4E and fig. S3B). In addition, combined loss of Pten and p19Arf led to a significant increase in cell proliferation (Fig. 4F, P = 0.042 < 0.05, t test) and cell transformation as determined by soft agar transformation assay (Fig. 4G), as compared to p19Arf null cells. Because the p16Ink4a-RB–E2F pathway plays critical roles in senescence and cell proliferation, the abundance of their proteins was evaluated in MEFs. Increased phospho-RB (pRB, Ser780) and p16Ink4a protein abundances were observed in both p19Arf−/−;PtenΔ/Δ and PtenΔ/Δ MEFs (Fig. 4, A and B, lower panel). The increase in RB phosphorylation at Ser780 was also observed in p19Arf−/−;PtenΔ/+ cells but not in p19Arf−/− cells, indicating that phosphorylation of RB is triggered even in cells lacking a single copy of Pten (Fig. 4C). E2F-1 and its transcriptional target, proliferating cell nuclear antigen (PCNA), were increased in p19Arf−/−;PtenΔ/Δ MEFs compared to p19Arf−/− and PtenΔ/Δ single-mutant cells (Fig. 4A, lower right panel). The increase in E2F-1 abundance was restricted to the compound mutant cells, p19Arf−/−;PtenΔ/Δ and p19Arf−/−;PtenΔ/+, suggesting that loss of p19Arf cooperates with loss of Pten to promote cell proliferation in association with the increase in E2F-1 abundance (Fig. 4C).

Next, we analyzed the status of E2F-1 and PCNA in mouse prostate epithelium. Immunostaining of the ventral prostate of 11-week-old mice revealed a marked increase in E2F-1 abundance in the Ptenpc−/− mutant mice, which was attenuated in prostates from p19Arf−/−;Ptenpc−/− double-mutant mice (Fig. 5A). These results correlated with the decrease in PIN observed in these glands. Additionally, PCNA staining, commonly used as a marker for proliferation, showed a pattern similar to that of Ki-67 (Fig. 2D). PCNA abundance was decreased in non-PIN areas of prostate epithelium from p19Arf−/−;Ptenpc−/− double-mutant mice relative to PIN lesions in Ptenpc−/− mice and PIN regions of p19Arf−/−;Ptenpc−/− double-mutant mice (Fig. 5B, lower right panel and inset). These data suggest that the opposite effects of p19Arf loss on cell proliferation in MEFs and in the prostate epithelium correlate with modulation of E2F-1 abundance.

Fig. 5

Loss of p19Arf in Pten-null mutant mice results in decreased E2F-1 and PCNA up-regulation in prostates. IHC staining for E2F-1 (A) and PCNA (B) of ventral prostates in Wt, p19Arf−/−, Ptenpc−/−, and p19Arf−/−;Ptenpc−/− double-mutant mice at 11 weeks of age. PIN lesions in p19Arf−/−;Ptenpc−/− double-mutant mice prostate in (B) are represented in an inset. Arrows denote E2F-1 positive cells (A) or PCNA positive cells (B).

Human prostate cancer biopsies with loss of PTEN retain p14ARF expression

Based on these data, we hypothesized that p14ARF loss would not be selected in human prostate cancer in the context of PTEN loss. To test this hypothesis, we analyzed the status of PTEN and p14ARF in tumor tissue microarrays (TMAs) from human prostate specimens. We found that complete loss of p14ARF was extremely rare in human prostate cancer (Fig. 6A). In contrast to other types of cancer (18), increased p14ARF abundance correlated with disease aggressiveness [Fig. 6A and (14)]. Furthermore, we found a correlation between PTEN loss and p14ARF overexpression in prostate cancers (Fig. 6B; n = 129, P < 0.000001, χ2 test). Loss of p53 function has been linked to accumulation of ARF as a feedback mechanism (19), which is thought to depend on the negative transcriptional regulation of ARF by p53 (20). Analysis of p53 abundance by IHC staining did not show a significant correlation between p53 immunoreactivity and p14ARF accumulation (χ2 test, P = 0.16; fig. S4A).

Fig. 6

Overexpression of p14ARF in human prostate cancer specimens correlate with PTEN loss. (A) p14ARF protein abundance in prostate cancer biopsies. Right: representative images for p14ARF staining in specimens at low- and high-severity grades. Left: graph showing the correlation of disease aggressiveness and p14ARF abundance; 0, tumor cells negative for p14ARF; 1, tumor cells focally positive for p14ARF; 2, tumor cells diffusely positive for p14ARF. (B) Correlation of PTEN and p14ARF protein abundance in human prostate specimens by IHC. Representative images showing p14ARF staining in specimens with normal PTEN abundance (1) and PTEN loss (0). Yellow indicates abundance 2, orange abundance 1, and red abundance 0. (C) Schematic representation of the findings highlighted in this study.

However, determining the abundance of p53 by IHC analysis does not provide information about its genomic or mutational status. Therefore, we performed correlation analysis between the genomic status of TP53 and the abundance of messenger RNA (mRNA) p14ARF (CDKN2A). In the subset of prostate cancer biopsies analyzed (n = 103), 81% of the biopsies showed Wt TP53, 16% hemizygous TP53 loss, and 2% homozygous TP53 loss. Only 1 patient out of 103 (1%) presented a mutation in TP53 (together with hemizygous TP53 loss). These data suggest that inactivation of TP53 by mutation and genomic loss is an infrequent event in prostate cancer. We then analyzed the correlation between TP53 status and p14ARF mRNA expression. Genomic loss of TP53 did not correlate with p14ARF mRNA expression (P = 0.6; fig. S4B). Taken together, our results suggest that, in prostate cancer, the loss of p14ARF is selected against and lead us to speculate that prostate tumors lacking PTEN might in fact select for the retention or even the enhancement of p14ARF function.


The pathogenesis of prostate cancer, the second leading cause of cancer-related deaths among men in Western societies (21, 22), is a complex process involving alterations in the activity of multiple oncogenes and tumor suppressor genes (2325). Of these, loss of PTEN function and translocation of TMPRSS22-ERG (transmembrane serine protease 22–Ets-related gene) represent the most frequent events (8, 12, 26). Following complete loss of Pten, p53-dependent cellular senescence provides a fail-safe mechanism to oppose cancer progression; concomitant loss of p53 and Pten abolishes this senescence response and results in lethal prostate cancer (1). In general, inhibition of Mdm2 by the tumor suppressor ARF represents one of the most important molecular mechanisms for the up-regulation of p53. (2730). In line with a tumor-suppressive role for ARF, p19Arf-deficient mice develop lymphomas, sarcomas, and adenocarcinomas (4), dying by 1 year of age. The observation that p19Arf loss did not phenocopy the loss of Trp53 in the Pten-null prostate, with no abrogation of senescence, but rather a decrease in the initiation of prostate cancer, is consistent with the notion that ARF could affect in vivo tumor growth in a tissue-specific manner (31). These results provide a plausible explanation for the observation that increased p14ARF abundance correlates with loss of PTEN and prostate cancer aggressiveness, in turn suggesting that prostate cancer development may select against loss of ARF function.

Both in prostate epithelium and in MEFs, loss of p19Arf did not affect the marked increase in p53 abundance observed in the context of Pten deficiency. Several groups have identified p53-independent functions of ARF through its interaction with various proteins (6). Our results suggest that ARF may exert distinct functions in a tissue-specific and p53-independent manner, possibly resulting from interactions with different targets in different cell types. In line with this hypothesis, we studied the localization and abundance of the ARF-interacting protein NPM (nucleophosmin, also called B23) and did not detect any changes upon compound Pten-p19Arf loss (fig. S5), suggesting NPM is not involved in the p53 pathway in vitro and in vivo upon Pten loss.

In MEFs, however, the concomitant loss of Pten and p19Arf resulted in the abrogation of senescence and in hyperproliferation and transformation. The differential requirement of ARF with Pten loss could involve different wiring of Mdm2-p53 in MEFs and prostate epithelium. However, Mdm2 inhibitors (Nutlin-3) promote p53 up-regulation in both experimental settings (32), confirming that in both MEFs and prostate epithelial cells, p53 is under the control of Mdm2. Further, in both settings, p53 is markedly increased even in the absence of ARF. This suggests that, with Pten loss, ARF exerts different functions in MEFs and prostate epithelium.

In conclusion, these findings establish cell context-specific functions for ARF in transformation, senescence, and cancer initiation that are independent of the Mdm2-p53 pathway. They also suggest that the role of ARF in cancer may vary, depending on tumor genetic milieu and tissue-cell type. Our findings also underscore the need for a careful reevaluation of the status and role of p14ARF in human cancer. A deeper understanding of the p53-independent functions regulated by ARF may provide previously unknown and critical therapeutic targets for cancer treatment.

Materials and Methods

Generation of p19Arf and Pten double-mutant mice

PtenloxP/loxP, PB-Cre4, and p19Arf−/− mice were maintained as described (1, 4, 33). Male PtenloxP/+;PB-Cre4 mutant mice were crossed with female p19Arf−/−mice to produce mouse colonies. Then F1 male p19Arf-/+;PtenloxP/+;PB-Cre4 mice were mated with F1 female p19Arf-/+;PtenloxP/+ mice to obtain F2 offspring, which were used to generate the following genotypes: Wt, p19Arf deficient mice (p19Arf−/−), prostate-specific Pten mutant (PtenloxP/loxP;PB-Cre4, referred to as Ptenpc−/−), and prostate-specific p19Arf-Pten double-mutant (p19Arf−/−;PtenloxP/loxP;PB-Cre4, referred to as p19Arf−/−;Ptenpc−/−) mice. For genotyping, DNA extracted from mouse tails was subjected to polymerase chain reaction (PCR) analysis with the same PCR primers and conditions as described (1). All experimental animals were kept in a mixed genetic background of C57BL/6J X129/Sv. Animal experiments were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee.

MRI and histopathology

Individual mice were subjected to MRI assessment for the detection of prostate tumors as described (1). Tissues were fixed in 10% neutral-buffered formalin (Sigma) overnight, rinsed with phosphate-buffered saline (PBS), and stored in 70% ethanol at 4°C. Tissues were processed for ethanol dehydration and embedded in paraffin (Histoserv Inc., Gaithersburg, MD) according to standard protocols. Tissue sections (5 μm) were prepared for H&E staining and antibody detection.

Cell proliferation, transformation, and senescence assays

MEFs were prepared from individual embryos of various genotypes and early-passage MEF cultures (P1-P2) as described (33) and were then infected with retroviruses expressing Cre-PURO-IRES-GFP or empty vector (without Cre) (1). MEFs were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum, 2 mM glutamine, and penicillin and streptomycin (100 U/ml; GIBCO). To achieve a high titer of retroviral particles, 2 × 106 Phoenix cells were seeded per 10-cm poly-d-lysine–coated culture dish (BD) followed by overnight culture in an incubator at 37°C with 5% CO2. pMSCV-Cre-PURO-IRES-GFP or empty vector was transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. Forty-eight hours after transfection, retroviral supernatants were collected and filtered through a 0.45-μm filter. For retroviral infection, MEFs at passage 1 or 2 were plated at a density of 3 × 105 to 4 × 105 cells per 10-cm culture dish and incubated with 10 ml of freshly made viral supernatant containing polybrene (5 μg/ml; Calbiochem). To increase infection efficiency, MEFs were exposed to viral supernatants a second time, followed by replacement with fresh medium after 8 hours. After the second infection, MEFs were cultured for an additional 48 hours and selected with puromycin (2 μg/ml) in 10 ml of fresh medium (Sigma) for 2 days. Selected MEFs at passage 5 were used for growth curves and Western blot and cellular senescence analysis. To evaluate cell proliferation, MEFs were plated at 2 × 104 cells per well in 12-well plates in triplicate, and cells were counted with a hematocytometer at 2, 4, 6, and 8 days. To determine senescence, MEFs were plated at 1 × 104 cells per well of a six-well plate in triplicate, and after 4 days, SA-β-gal was detected with the senescence detection kit (Calbiochem) and quantified (more than 200 cells per sample). For prostate tissue, frozen sections at 6-μm thickness were stained for β-gal as above. For transformation assay, selected MEFs (3 × 104) at passage 5 were suspended in medium containing 0.3% agar onto solidified 0.6% agar per well of a six-well plate. The ability to grow in soft agar was assessed by size and number of colonies counted after 21 days.

Western blot and immunohistochemistry

MEF lysates were prepared with radioimmunoprecipitation assay buffer [1× PBS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, and protease inhibitor cocktail (Roche)]. The following antibodies were used for Western blot: mouse monoclonal antibody directed against Pten [anti-Pten (6H2.1, Cascade BioScience)], rabbit polyclonal anti-p53 (CM5, Novocastra), rabbit polyclonal anti-Akt and phospho-Ser473 of Akt (Cell Signaling), rabbit polyclonal phospho-RB (Ser780, Cell Signaling), rabbit polyclonal anti-p19Arf (C-19, NeoMarkers), rabbit polyclonal anti-p21 (C-19, Santa Cruz), rabbit polyclonal anti-p16 (M-156, Santa Cruz), rabbit polyclonal anti–E2F-1 (C-20, Santa Cruz), mouse monoclonal anti-PCNA (SC-56, Santa Cruz), rabbit polyclonal anti-NPM (3542, Cell signaling), and mouse monoclonal anti–β-actin (AC-74, Sigma). To determine the half-life of p53 and p21 proteins, MEFs (Pten+/+-Cre and p19Arf−/−-PtenΔ/Δ-Cre, at 80% confluence and equal passage number) were treated with cycloheximide (30 μg/ml; Sigma) and harvested at the indicated times for Western analysis. For IHC analysis, sections were stained for phospho-Akt (Ser473) (mouse monoclonal antibody, Cell Signaling), p53 (FL-393, rabbit polyclonal antibody, Santa Cruz), p21 (F-5, mouse monoclonal antibody, Santa Cruz), rabbit polyclonal anti–E2F-1 (C-20, Santa Cruz), Ki-67 (Novacastra), rabbit polyclonal anti-NPM (3542, Cell signaling), and mouse monoclonal anti-PCNA (SC-56, Santa Cruz).

Tumor tissue microarrays

Prostate TMAs were constructed with a fully automated Beecher Instrument (ATA-27). The study cohort comprised prostate tumor biopsies from Memorial Sloan-Kettering Cancer Center (MSKCC). All biopsies were evaluated at MSKCC, and the histological diagnosis and Gleason score were based on established standard criteria. Use of tissues was approved by institutional review board waivers and by the Human Biospecimen Utilization Committee. TMAs were stained with antibody directed against p14ARF (clone 4C6/4, #2407, mouse monoclonal, Cell Signaling, 1:250 dilution) and scored by the following criteria: 0, tumor cells negative; 1, tumor cells focally positive; 2, tumor cells diffusely positive. PTEN staining (clone 6H2.1, mouse monoclonal, Cascade Bioscience, 1:75 dilution) and scoring were performed as previously published (34). Secondary antibodies used were biotinylated horse anti-mouse (1:500) and avidin-biotin (1:25) (Vector Lab).


We thank the members of the Pandolfi laboratory for their insightful comments and helpful discussions. Special thanks also extend to C. Sherr for reagents; M. Stuart for critical reading and editing of the manuscript; K. Manova, A. Barlas, and V. Gueorguiev from the Molecular Cytology Core Facility at Memorial Sloan-Kettering Cancer Center (MSKCC) for assistance with IHC analysis; and C. Le, C. Matei, and M. Lupu from the Small Animal Imaging Core at MSKCC for MRI analysis. This work was supported by National Cancer Institute (NCI) grants U01 CA-84292 and R01 CA-82328 to P.P.P. and partially by NCI grant U54 CA-91408 to Z.C. A.C. was supported by a European Molecular Biology Laboratory long-term fellowship. Z.C., A.C., and P.P.P. conceived and designed the experiments and wrote the manuscript. Z.C., A.C., H.-K.L., A.E., N.B., and A.A. performed the experiments.

Supplementary Materials

Fig. S1. Generation of p19Arf−/−;Ptenpc−/− mice and visualization of prostate tumors.

Fig. S2. Immunohistochemical analysis of p21 in mouse prostate.

Fig. S3. In vitro characterization of p19Arf−/−-PtenΔ/Δ double-null MEFs.

Fig. S4. Lack of correlation between p53 and p14ARF expression in human prostate cancer biopsies.

Fig. S5. Nucleophosmin abundance and localization in MEFs.

References and Notes

View Abstract

Related Content

Navigate This Article