Potent p53‐independent tumor suppressor activity of ARF in melanoma‐genesis

Abstract

It is generally accepted that human cancer is a resultof genetic mutations in molecular pathways controllingprocesses that normally harness the capacity for prolif-eration and survival. An emerging picture has begun toevolve from our understanding of these particular path-ways as they become disrupted in various cancers, in afashion which shows significant lineage-dependence.Cutaneous malignant melanoma, the most commonform of melanoma in man, appears to invariably harbormutations leading to activation of the canonical MAPK-pathway wherein the oncogene BRAF is found activatedin approximately 70% of all melanomas (Davies et al.,2002) as well as benign nevi (Pollock et al., 2003) andcorrespondingly NRAS in another 25% (Demunteret al., 2001) leaving only a small fraction unaccountedfor. In addition, disruption of the pRB-pathway throughloss of pRB, mutation or amplification of CDK4, amplifi-cation of CCND1, missense mutation or deletion of theINK4A-locus is similarly common. However, contrastingmany other human malignancies, melanomas exhibit astrikingly low frequency of canonical p53-mutations,amounting to approximately 20–25% of all cases(Ragnarsson-Olding et al., 2002). This is particularlynotable because of the discordance between the p53’swild-type status and the near-uniform relative refractori-ness of melanomas to numerous triggers of apoptosis,including cytotoxic chemotherapeutic agents. Addition-ally, disruption of the TP53-pathway through amplifica-tion of the HDM2 oncogene occasionally occurs, but itsupstream inhibitor ARF, the alternative product of theINK4A-locus (Kamijo et al., 1997), is a well-known mela-noma tumor-suppressor whose loss together withCDKN2A⁄p16 is relatively common during melanoma-genesis. It is broadly accepted that ARF functions toinhibit MDM2⁄HDM2, the canonical negative regulatorof p53, in mouse and man while there are additionalARF-interacting proteins that may execute importantfunctions various settings of tumorigenesis (Sherr,2006). The common occurrence of wild-type p53together with ARF (and CDKN2A⁄p16) deficiency in mel-anoma has appropriately raised the prospect of a thera-peutic opportunity based upon reactivating p53-function,to which end multiple lead drugs are in various stagesof development.In a recently published study, Merlino and colleagues(Ha et al., 2007) have utilized a powerful genetic modelof murine melanoma to assess the functional roles ofp53 and ARF in murine melanoma-genesis. Their dataclearly demonstrate p53-independent tumor suppressiveactivity of ARF- an observation which carries importantimplications for therapeutic targeting⁄activation of p53in melanoma patients.In this study, Ha et al. employ an established mousemodel that develops UV-induced cutaneous melanomaas well as rhabdomyosarcoma (RMS) via transgenicexpression of hepatocyte growth factor⁄scatter factor(Noonan et al., 2001), the ligand for the cMET receptortyrosine kinase. Merlino et al. previously showed thatmelanoma latency following UV irradiation was signifi-cantly reduced when the transgenic mice were crossedinto an INK4A⁄ARF-null background, suggesting thatloss of both the TP53- (through ARF) and pRB- (throughCDKN2A⁄p16) pathways likely contribute to develop-ment of UV-induced cutaneous melanoma. In this study,the authors have overlaid a more refined approach, ofseparating the functions of the two products from theINK4A-locus. They observe that ARF-nullizygosity exhib-its nearly the same frequency and timing of RMS devel-opment as genetic ablation of both products. Similarly,p53-nullizygousity exhibited similar frequency and timingas ARF-alone for induction of RMS, thus consistent withthe predicted role of the ARF-(MDM2)-p53-pathway fortumor suppression. Importantly, however, UV-inducedmelanoma incidence in these transgenic animalsshowed dependence on both CDKN2A and ARF statuswhereas p53-nullizygousity had no overt effect, suggest-ing that ARF tumor suppressor function is mechanisti-cally distinct from p53-function. The importance of thismelanoma-specific behavior is underscored by the dif-ferences seen in RMS tumors within the same geneticmodel. Interestingly, in a UV-independent RAS-driventransgenic model, CDKN2A, ARF and p53 all displayedimportant contributions as suppressors of melanomaincidence (Bardeesy et al., 2001; Sharpless and Chin,