Oncogene (2002) 21, 6779 – 6790 ª 2002 Nature Publishing Group All rights reserved 0950 – 9232/02 $25.00 www.nature.com/onc Human p14Arf: an exquisite sensor of morphological changes and of short-lived perturbations in cell cycle and in nucleolar function The´ re` se David-Pfeuty*,1 and Yolande Nouvian-Dooghe1 1UMR 146 du CNRS, Institut Curie-Recherche, Baˆtiment 110, Centre Universitaire, 91405 Orsay Ce´dex, France The human Ink4a/Arf tumor suppressor locus encodes integrity of cell cycle checkpoint controls (Hartwell and two distinct products: p16Ink4a which prevents phosphor- Weinert, 1989). Defects in checkpoint controls can ylation and inactivation of the retinoblastoma protein trigger either cell death or survival of cells carrying and, p14Arf, a nucleolar protein which activates the genetic anomalies and prone to develop genomic function of the tumor suppressor p53 protein in the instability which is the hallmark of cancer cells nucleoplasm in response to oncogenic stimulation through (Hartwell, 1992). It has been realized for some time an as yet ill-defined mechanism. Here we show that the that the p53 tumor suppressor gene is a component of level of endogenous p14Arf and its balance between the DNA damage checkpoints in mammalian cells since its nucleolus and the nucleoplasm in HeLa cells are function is required to prevent traverse across the G1/S exquisitely sensitive to changes in cell morphology and and G2/M boundaries in the presence of many types of to short-lived perturbations in cell cycle and in nucleolar genotoxic stress (O’Connor, 1997; Giaccia and Kastan, function such as those induced by the cyclin-dependent 1998) via promoting cell cycle arrest or apoptosis. It kinase inhibitor, roscovitine, and the casein kinase II and has been realized for even longer that p53 is mutated or RNA synthesis inhibitor, DRB. Most remarkably, lost in more than 50% of human cancers (Hollstein et whereas p14Arf predominantly concentrates in the al., 1994) which provides a hint that it would also serve nucleolus of interphase cells and transiently disappears in vivo to bar cell proliferation in inappropriate between metaphase and early G1 under normal growth environmental settings, a conclusion which has been conditions, it massively and reversibly accumulates in the consolidated by all kinds of in vivo and in vitro nucleoplasm of postmitotic and S-phase cells upon short- investigations (Ko and Prives, 1996; Levine, 1997; term treatment with roscovitine and, at a lesser extent, Agarwal et al., 1998). The possibility that the function DRB. In line with the fact that the nuclear level of p53 of p53 in tumor suppression and in DNA damage reaches a peak between mid-G1 and the G1/S border in response might represent two distinct facets of its p53-expressor cells which lack Arf expression, these activity has emerged recently from the discovery that results provide a clue that, in p53+/Arf+ cells, Arf oncogenic stimulation can induce its activation via a proteins might serve both to speed and to amplify p53- mechanism quite different from that generally used by mediated responses in conditions and cell cycle periods in DNA damages. Whereas the latter utilizes protein which the mechanisms involved in p53 stabilization and kinases to phosphorylate p53 at various sites mostly activation are not fully operational. They further suggest within its N-terminal transactivation domain and to that human endogenous p14Arf might activate p53 disrupt its interaction with its target and negative pathways in physiologic situations by acting inside the regulator, Mdm2 (Prives, 1998), the first relies on the nucleoplasm, especially when normal cell cycle progres- expression of the Arf proteins (Quelle et al., 1995; sion and nucleolar function are compromised. Della Valle et al., 1997), encoded by the late-identified Oncogene (2002) 21, 6779 – 6790. doi:10.1038/sj.onc. Arf (alternate reading frame) transcript of the Ink4a 1205871 gene (Quelle et al., 1995; Duro et al., 1995; Stone et al., 1995; Mao et al., 1995). Human p14Arf, like mouse Keywords: p14Arf; cell cycle regulation; tumor suppres- p19Arf, stabilizes p53 owing to its capability to interact sion; cell morphology; cyclin-dependent kinase directly with Mdm2 (Pomerantz et al., 1998; Zhang et al., 1998; Kamijo et al., 1998; Stott et al., 1998) and, thereby, to inhibit the Mdm2-mediated nuclear export Introduction (Tao and Levine, 1999; Zhang and Xiong, 1999), ubiquitination and degradation of p53 (Honda and Faithful cell reproduction depends on the orderly Yasuda, 1999). Arf upregulation in primary cells completion of S and M phase which relies on the exposed to hyperproliferative signals such as myc (Zindy et al., 1998), E1A (de Stanchina et al., 1998), E2F1 (Bates et al., 1998; Dimri et al., 2000), oncogenic ras (Palmero et al., 1998; Lin and Lowe, 2001) and v- *Correspondence: T David-Pfeuty; E-mail: The´ re` [email protected] abl (Radfar et al., 1998; Cong et al., 1999) seems to be Received 20 February 2002; revised 8 July 2002; accepted 15 July directly involved in stabilizing p53, the physiologic 2002 outcome being cell cycle arrest or apoptosis and, p14Arf subnuclear localization and function T David-Pfeuty and Y Nouvian-Dooghe 6780 possibly, after diverting the p53 pathway, cell trans- response because it would permit to accumulate at the formation (Sherr, 1998; Sharpless and DePinho, 1999). same time high levels of Mdm2 ready to eliminate p53 A rationale for these observations is provided by the as soon as the triggering signal is switched off. This fact that Arf is a direct target gene of some hypothesis that Arf might work at extending the transcription factors of the E2F family (Degregori et duration of reversible p53-mediated cell cycle arrest is al., 1997; Bates et al., 1998) whose function is required already consistent with the report that NARF2 cells to promote S phase entry (Harbour and Dean, 2000; arrested by inducible expression of p14Arf resume Nevins, 2000). Altogether, these data have led to the normal cycling upon removal of the inducer after 2 concept that, via Arf, p53 would also participate in days (Stott et al., 1998). Whether this hypothesis is also oncogene checkpoints that guard cells against hyper- relevant to physiologic situations clearly deserves proliferative signals (Sherr, 1998). further investigation. The fact that the co-expression Yet, the mechanism how Arf proteins activate p53 is of p14Arf and p53 does not occur in tumor cells still unclear because the subcellular compartment in suggests, not only, that it is incompatible with in vitro which p53 and Mdm2 localize preferentially is the tumor cell growth, but also, that some important nucleoplasm and for Arf, it is the nucleolus. Based on aspects of the in vivo Arf regulation are preserved in the results of enforced co-expression experiments, one p53-negative tumor cell lines. Identifying parameters model proposes that Arf proteins function owing to involved in controlling the level of expression and their ability to sequester Mdm2 in the nucleolus and, subnuclear targeting of endogenous p14Arf in this type thence, to trap Mdm2-free p53 inside the nucleus (Tao of tumor cells might therefore be a source of and Levine, 1999; Weber et al., 1999, 2000; Sherr and information not to ignore more than those issued from Weber, 2000) or in PML-like nuclear bodies (Zhang the experiments of ectopic overexpression. and Xiong, 1999). It can not be excluded however that Here we report that both the level of expression and Arf proteins could utilize other tricks to activate p53 subnuclear targeting of endogenous p14Arf are highly given that the nucleolar localization of p14Arf is not variable among clonal variants of the HeLa cell line requested for its function in human cell lines (Llanos et which differ in their morphological appearance. Most al., 2001) and that Mdm2 does not relocalize to the remarkably also, we uncovered that endogenous, but nucleolus following p19Arf-dependent, ras-induced cell not ectopically-expressed, p14Arf disappears in meta- cycle arrest in primary mouse keratinocytes (Lin and phase when the nucleoli disintegrate and re- Lowe, 2001). The major caveat of in vitro studies accumulates in the newly-reformed nucleoli of the aiming at deciphering the mechanism of p53 activation daughter cells at the beginning of G1. Intriguingly by Arf proteins in physiologic situations comes from however, it massively and reversibly accumulates in the the fact that they all rely on the artifice of ectopic nucleoplasm of early G1 and S-phase cells upon short- overexpression because Arf is undetectable in untrans- term treatments with roscovitine and, at a lesser extent, formed cells and its co-expression with p53 does not DRB, that disrupt nucleolar organization. The implica- occur in tumor-derived ones (Pomerantz et al., 1998; tions of these results in relation to the possible in vivo Stott et al., 1998; Bates et al., 1998; Lindstro¨ m et al., functions of the Arf proteins are discussed. 2000). That is why a more conceptual approach to the problem should also be considered. In a recent report Results (David-Pfeuty et al., 2001), we have shown that a short-term treatment of human breast carcinoma Clonal variability in levels and subnuclear localization of MCF-7 cells and non-immortalized IMR-90 fibroblasts endogenous p14Arf with the Cdk inhibitor, roscovitine (Meijer et al., 1997), and the CKII and RNA synthesis inhibitor, DRB Human p14Arf has been reported to localize predomi- (Granick, 1975; Haaf and Ward, 1996), induces a nantly to the nucleolus (Pomerantz et al., 1998; Kamijo reversible activation of p53 that is coupled with a et al., 1998; Stott et al., 1998; Zhang and Xiong, 1999), reversible nucleolar fragmentation and blockade of 45S more precisely to its granular component (Lindstro¨ m et pre-rRNA processing and a reversible brake in G1/S al., 2000; Karayan et al., 2001), in a pattern very transit.