View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector p r e v i e w s

However, there are several key points cell-autonomous tumor-promoting activ- Selected reading that emanate from this interesting report. ity, in apposition to cell-nonautonomous Kovacic, B., Stoiber, D., Moriggl, R., Weisz, E., These findings serve as an important host immune response—in this case each Ott, R.G., Kreibich, R., Levy, D.E., Beug, H., extension of the concept of immunoed- attributable to deficiency of the same Freissmuth, M., and Sexl, V. (2006). Cell, iting in tumor surveillance (Shankaran allele, STAT1. Finally, this report dem- this issue. et al., 2001), and the importance of NK onstrates an expanding paradigm that Lee, C.K., Gimeno, R., and Levy, D.E. (1999). J. cells in shaping the phenotypic charac- some products may serve both as Exp. Med. 190, 1451–1464. teristics of tumors. These data corrobo- tumor suppressors and as tumor promot- Lee, C.K., Rao, D.T., Gertner, R., Gimeno, R., rate and support previous findings on the ers. Ultimately, clear insights into these Frey, A.B., and Levy, D.E. (2000a). J. Immunol. role of STAT1 in NK cell function, and in pathogenetic processes should inform 165, 3571–3577. regulation of MHC class I expression. It more effective therapeutic approaches to Lee, C.K., Smith, E., Gimeno, R., Gertner, R., seems likely, based on these data, that cancer. and Levy, D.E. (2000b). J. Immunol. 164, 1286– there will be functions of STAT1 that are 1292. 1 both dependent and independent of type Jennifer L. Rocnik and Lesinski, G.B., Anghelina, M., Zimmerer, J., I IFN signaling, an arena that should be D. Gary Gilliland1,* Bakalakos, T., Badgwell, B., Parihar, R., Hu, Y., of considerable interest for further investi- Becknell, B., Abood, G., Chaudhury, A.R., et al. gation. These experiments also elegantly (2003). J. Clin. Invest. 112, 170–180. 1Howard Hughes Medical Institute, demonstrate the value of modeling cancer Harvard Medical School, Karp Family Shankaran, V., Ikeda, H., Bruce, A.T., White, J.M., Swanson, P.E., Old, L.J., and Schreiber, R.D. in vivo in murine models of disease, and Research Laboratories, 1 Blackfan Circle, (2001). Nature 410, 1107–1111. the importance of in vivo experimentation Room 5210, Boston, Massachusetts 02115 to dissect out the evolutionary “winner” in *E-mail: [email protected] DOI 10.1016/j.ccr.2006.06.015

Of mice and men: Cancer gene discovery using comparative

With the proliferation of high-throughput technologies to profile the cancer genome, methods to distinguish causal from bystander genetic events are needed. Two recent reports by Zender et al. and Kim et al. in Cell use genetically defined mouse models to serve as biological filters to mine the human cancer genome. Integration of high-resolution copy number profiles of mouse tumor models and human tumors identified cIAP1 and Yap as in human hepatocellular carcinoma, while NEDD9 was identified as a gene in human melanoma. Together, these reports demonstrate that a comparative oncogenomics approach can identify causally involved in oncogenesis and metastasis.

High-throughput techniques are identify- in oncogenesis and metastasis in human By ROMA, four of seven Myc-driven HCCs ing remarkable heterogeneity in the can- tumors (Figure 1). contained a focal (containing 12 cer genome, implying that dysregulation Comparative oncogenomics identi- genes) on 9qA1. of numerous genes and pathways can fies cIAP1 and Yap as driving onco- Using the mouse model as a biological lead to oncogenesis and cancer progres- genes in hepatocellular carcinoma filter, the authors interrogated 48 human sion. Furthermore, as progress, In the study by Zender et al., the authors HCCs profiled in parallel. Intriguingly, two they demonstrate genomic instability, isolated hepatoblasts from the liver of of these human HCCs harbored focal leading not only to the acquisition of muta- −/− mice and overexpressed Myc, acti- amplifications on chromosome 11q22 tions conferring selective advantages, but vated Akt, or oncogenic Ras before rein- (syntenic to mouse 9qA1). To identify the also to noncontributing bystander lesions. troducing these cells into recipient mice driving gene, the authors determined the With plans to characterize all livers. They focused on tumors develop- expression of all overlapping genes from in human cancers, such as The Cancer ing from p53−/−;myc-induced hepatoblasts, the syntenic . Not one, but two Genome Atlas (Bonetta, 2005), novel which were histologically consistent with genes, cIAP1 and Yap, showed increased techniques to identify the lesions driving human hepatocellular carcinomas (HCCs). expression at the mRNA and protein level oncogenesis/metastasis from secondary In an effort to identify spontaneously in all mouse and human tumors contain- mutations are urgently needed. Two stud- acquired lesions contributing to tumori- ing the amplicon. ies by Zender et al. (2006) and Kim et al. genesis, they used a form of array com- To confirm the oncogenic effects of (2006) in the June 30 issue of Cell dem- parative genomic hybridization (aCGH) cIAP1 and Yap in vivo, the authors overex- onstrate the usefulness of comparative termed representational pressed these two genes in the p53−/−;myc oncogenomic approaches using defined microarray analysis (ROMA) to search hepatoblasts that were used to generate mouse models to identify driving genes for focal regions of copy number change. the tumors. Expression of cIAP1 or Yap

 cancer cell July 2006 p r e v i e w s

accelerated tumor formation and Finally, the authors showed that increased tumor burden. Similarly, NEDD9 expression induced the knockdown of cIAP1 or Yap in activation of FAK, a key regulator of tumors containing the 9qA1 ampli- focal adhesion contacts, and dem- con inhibited tumor growth. Finally, onstrated that NEDD9 modulates the authors demonstrated that the formation of focal contacts in cIAP1 and Yap act synergistically melanocytes, suggesting a mecha- in the 9qA1 amplicon, as coexpres- nism for its effects on invasion and sion of cIAP1 and Yap accelerated metastasis. tumor growth more than expres- Advantages and limitations sion of either gene alone. of mouse models to identify Comparative oncogenomics ­driving cancer genes identifies NEDD9 as a mediator An important aspect of both studies of metastasis in melanoma was the use of mouse models with Kim et al. employed a similar defined genetic backgrounds. This approach, using an inducible Ras provides the proper genetic back- model of nonmetastatic melanoma ground to confirm the oncogenic/ to generate “escapers,” or clones metastatic effects of identified can- that underwent secondary genomic didate genes, as these effects may events conferring metastatic capa- only occur in cooperation with the bility. The authors characterized two lesions used to generate the tumor such escapers with increased pro- models. For example, Zender et liferation and invasion compared al. demonstrated that cIAP1 and to the parental tumors and profiled Yap overexpression did not affect these samples using genome-wide tumor formation in p53−/−;Akt or aCGH. Both escapers showed Ras hepatoblasts. Likewise, Kim a focal amplification on mouse et al. showed that NEDD9 was chromosome 13, and the overlap- unable to increase invasion or ping region contained eight genes. metastasis in melanocytes in the Examining the expression of these absence of activated Ras or Raf. eight genes in Ras- and Met-driven Unfortunately, for many tumor melanomas, NEDD9 was the only types, the initiating genetic events gene with increased expression in driving tumorigenesis and the pre- samples with and without genomic cursor cells harboring such events amplification. are unclear, precluding the devel- The authors used this result to opment of such defined models. In guide the examination of a data- addition, the experimental strate- base of human aCGH profiles gies utilized by Zender et al. and from melanocytic lesions, where Kim et al., which were designed to chromosome 6p25-24 (syntenic to identify focal amplifications driv- mouse chromosome 13) was found ing oncogenesis and metastasis, to be amplified in 36% of metastat- would not be able to identify driv- ic melanomas, 8% of melanomas, ing mutations or chromosomal and none of the benign nevi. They rearrangements. confirmed the overexpression of Integrative approaches to NEDD9 at the transcript and protein Figure 1. Biological filtering using defined mouse models to ­identify genes causally levels in 35%–52% of metastatic identify driving mutations controlling oncogenesis/metastasis in involved in cancer melanomas using quantitative RT- human tumors Another important result observed PCR and immunohistochemistry on Schematic of the experimental approach used by Zender et al. by both Zender et al. and Kim et al. and Kim et al. to identify genes involved in tumor initiation or tissue microarrays. To confirm the metastasis. Defined mouse models of tumorigenesis/metastasis was that the oncogenic amplifica- role of NEDD9 in the development were chosen to mimic the initiation or progression of hepato- tions that they observed in human of metastasis in their escapers, cellular carcinoma or melanoma, respectively. Isolated tumors tumors were relatively rare. For the authors knocked down NEDD9 from these models were profiled by array comparative genomic example, Zender et al. observed hybridization (aCGH)-based techniques to identify focal regions expression, significantly inhibiting of copy number change. These profiles were then used to filter amplification of the cIAP1 and Yap the proliferation and invasion of aCGH data from human tumors to identify minimally conserved loci in only 2 of 48 (4%) human the two escapers. Additionally, the amplified regions between the mouse and human tumors. The hepatocellular carcinomas, while expression of all genes in the amplified regions in both the mouse authors demonstrated that overex- models and human tumor samples was determined. Candidate Kim et al. observed amplification pression of NEDD9 increased the gene(s) were chosen by increased expression in all mouse and of the region harboring NEDD9 in in vitro and in vivo metastatic capa- human samples with the amplification. Importantly, both groups 36% of metastatic melanomas. were then able to confirm the effect of their candidate gene(s) bility of primary melanocytes and in vitro or in vivo in the same defined genetic model by retroviral- These results likely reflect the nonmetastatic melanoma cells. mediated overexpression or shRNA knockdown. substantial genetic heterogene- cancer cell July 2006  p r e v i e w s

ity that underlies tumorigenesis and the subset of melanomas (Garraway et al., Scott A. Tomlins1 and development of metastasis. Recently, our 2005). Adler et al. also integrated expres- Arul M. Chinnaiyan1,2,3,4,* group developed a strate- sion and copy number profiles from breast gy termed Cancer Outlier Profile Analysis cancers to identify coordinated amplifica- 1Department of Pathology (COPA) in an effort to identify “ tion of Myc and CSN5 as a regulator of the 2Program in Bioinformatics outliers,” or genes with marked overex- wound response profile that is predictive 3Department of Urology pression in a fraction of cases, which of poor outcome (Adler et al., 2006). 4The Comprehensive Cancer Center characterize genes involved in high-level Amassing genomic scale data for University of Michigan Medical School, copy number changes or translocations, human tumors is becoming relatively 1301 Catherine Street, Ann Arbor, such as those described above. COPA routine. The rapid advancement of Michigan 48109 identified the Ets transcription factors high-throughput techniques suggests *E-mail: [email protected] ERG and ETV1 as outliers across mul- that complete genomic profiles of tiple prostate cancer profiling studies. tumor samples may soon be realized. Selected reading Characterizing cases with overexpres- Integrative approaches will be needed to sion of ERG or ETV1, we demonstrated sift through this mass of data to identify Adler, A.S., Lin, M., Horlings, H., Nuyten, D.S., that these samples harbored recurrent the driving genetic lesions underlying van de Vijver, M.J., and Chang, H.Y. (2006). Nat. Genet. 38, 421–430. gene fusions with the androgen-regulat- cancer development and progression. ed gene TMPRSS2, and these fusions The results of Zender et al. and Kim et al. Bonetta, L. (2005). Cell 123, 735–737. occur in the majority of prostate cancers demonstrate that integrative approaches Garraway, L.A., Widlund, H.R., Rubin, M.A., Getz, (Tomlins et al., 2005). This approach was based on defined mouse models can G., Berger, A.J., Ramaswamy, S., Beroukhim, R., useful in the analysis of tumors without be used as one such method to identify Milner, D.A., Granter, S.R., Du, J., et al. (2005). Nature 436, 117–122. biologically relevant mouse models and driving oncogenic or metastatic events may prove useful in mining mouse mod- in human tumors. Kim, M., Gans, J.D., Nogueira, C., Wang, A., Paik, els of cancer as well. J.-H., Feng, B., Brennan, C., Hahn, W.C., Cordon- Cardo, C., Wagner, S.N., et al. (2006). Cell 125, In addition to using mouse models Acknowledgments 1269–1281. or bioinformatic strategies, other groups Tomlins, S.A., Rhodes, D.R., Perner, S., have used alternative integrative analy- This work was supported in part by the Department of Defense (W81XWH-06-1-0224), the National Dhanasekaran, S.M., Mehra, R., Sun, X.W., ses to identify driving genetic events in Institutes of Health (U54 DA021519-01A1 to Varambally, S., Cao, X., Tchinda, J., Kuefer, R., et tumorigenesis. For example, Garraway et A.M.C.), and the Cancer Center Bioinformatics al. (2005). Science 310, 644–648. al. performed an integrative analysis com- Core (Support Grant 5P30 CA46592). S.A.T. is sup- Zender, L., Spector, M.S., Xue, W., Flemming, P., bining gene expression and copy number ported by a Rackham Predoctoral Fellowship and is Cordon-Cardo, C., Silke, J., Fan, S.-T., Luk, J.M., profiles of the NCI60 panel of cancer cell a Fellow of the Medical Scientist Training Program. Wigler, M., Hannon, G.J., et al. (2006). Cell 125, A.M.C. is supported by a Clinical Translational 1253–1267. lines to identify and validate MITF as a Research Award from the Burroughs Welcome lineage-specific oncogene amplified in a Foundation. DOI 10.1016/j.ccr.2006.06.013

Differential utilization of two ATP-generating pathways is regulated by p53

A fundamental property of cancer cells is the preferential utilization of glycolysis over aerobic respiration to produce ATP. Renewed interest in understanding the mechanism underlying this metabolic shift in energy production is broadening our understanding of the relationship between cancer and cellular metabolism. In a recent article, Matoba et al. report that the p53 tumor suppressor regulates the expression of SCO2, a protein that is required for the assembly of cytochrome c oxidase (COX), a multimeric protein complex required for oxidative phosphorylation. The implication of these findings is that aerobic respira- tion is compromised in cells that lack functional p53.

In contrast to normal cells, cancer cells respiration, tumor cells have a much high- metabolism. A commonly held view is that have a high glycolytic rate and produce er rate of glucose uptake than normal cells constitutive upregulation of glycolysis is high levels of lactate even in the pres- (Figure 1). The physiological significance likely to be an adaptation to hypoxia that ence of oxygen (Figure 1). This metabolic of the Warburg effect has been controver- develops as tumor cells grow progres- shift to a higher rate of aerobic glycolysis sial since its discovery over 80 years ago, sively further away from their blood supply is commonly referred to as the Warburg and now there is renewed and vigorous (Gatenby and Gillies, 2004). Interestingly, effect. Because glycolysis produces ener- interest in understanding the relation- cells derived from tumors continue to uti- gy (ATP) far less efficiently than aerobic ship between cancer and altered energy lize glycolysis in culture under normoxic

 cancer cell July 2006