Cancer Cell Previews

with regard to the precise localization of chemotherapy (Ma et al., 2013, 2014). T., Mignot, G., Ullrich, E., et al. (2009). Nat. Med. 15 DC subsets within the tumor bed. Broz Moreover, direct purification of intratu- , 1170–1178. + et al. (2014) found that both CD11b moral DC subsets yields functional tumor Hildner, K., Edelson, B.T., Purtha, W.E., Diamond, + DC1 and CD103 DC2 were preferentially antigen-presenting cells that are able to M., Matsushita, H., Kohyama, M., Calderon, B., located in collagen-rich zones distal to the prime naive T cells in vitro (Broz et al., Schraml, B.U., Unanue, E.R., Diamond, M.S., et al. (2008). Science 322, 1097–1100. tumor nodules where TAM1 and TAM2 2014) and elicit anticancer immune cells were found. In contrast, Ruffell response upon adoptive transfer in vivo Kroemer, G., Galluzzi, L., Kepp, O., and Zitvogel, L. et al. (2014) report that CD103+ cells (Ma et al., 2013). These results reinforce (2013). Annu. Rev. Immunol. 31, 51–72. were dispersed throughout the tumor the idea that the tumor may be considered Ma, Y., Adjemian, S., Mattarollo, S.R., Yamazaki, stroma in the proximity of macrophages. as a full-blown lymphoid organ, in which T., Aymeric, L., Yang, H., Portela Catani, J.P., Although Ruffell et al. (2014) found all steps of cellular immune responses Hannani, D., Duret, H., Steegh, K., et al. (2013). no change in the localization of CD103+ starting with appropriate presentation of Immunity 38, 729–741. cells after treatment with paclitaxel and tumor antigens by dendritic cells occur Ma, Y., Mattarollo, S.R., Adjemian, S., Yang, H., aCSF-1, Ma et al. (2013) found that in situ. Aymeric, L., Hannani, D., Portela Catani, J.P., Du- CD11b+ cells exhibited a selective ret, H., Teng, M.W., Kepp, O., et al. (2014). Cancer 74 tropism for dying tumor cells after doxoru- Res. , 436–445. REFERENCES bicin treatment. Ruffell, B., Chang-Strachan, D., Chan, V., Rose- Irrespective of these discrepancies, nbusch, A., Ho, C.M.T., Pryer, N., Daniel, D., Broz, M., Binnewies, M., Boldajipour, B., Nelson, Hwang, S., Rugo, H.S., and Coussens, L.M. however, the accumulating evidence sug- A., Pollock, J., Erle, D., Barczak, A., Rosenblum, (2014). Cancer Cell 26, this issue, 623–637. gests that some DC subpopulations can M., Daud, A., Barber, D., et al. (2014). Cancer Cell 26, this issue, 638–652. cross-present tumor antigens within the Stoll, G., Enot, D., Mlecnik, B., Galon, J., Zitvogel, 3 cancer without needing to migrate to Fridman, W.H., Page` s, F., Saute` s-Fridman, C., and L., and Kroemer, G. (2014). OncoImmunology , lymph nodes. Thus, lymphadenectomy Galon, J. (2012). Nat. Rev. Cancer 12, 298–306. e27884. fails to affect the anticancer immune Ghiringhelli, F., Apetoh, L., Tesniere, A., Aymeric, Zitvogel, L., Galluzzi, L., Smyth, M.J., and response elicited by anthracycline-based L., Ma, Y., Ortiz, C., Vermaelen, K., Panaretakis, Kroemer, G. (2013). Immunity 39, 74–88.

Building through Breaking: The Development of Cancer Neochromosomes

Joshua J. Waterfall1 and Paul S. Meltzer1,* 1Genetics Branch, Center for Cancer Research, National Cancer Institute, 37 Convent Drive, Bethesda, MD 20892, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.ccell.2014.10.013

In this issue of Cancer Cell, Garsed and colleagues combine chromosome flow sorting and deep sequencing to characterize the structure of oncogene-containing neochromosomes in liposarcoma and provide evidence that they are generated by a combination of multiple dynamic and destructive processes.

Loss of genomic integrity in cancer has normal chromosome. By combining The heterogeneous collection of malig- many different manifestations. In this issue chromosome flow sorting and deep nant tumors of adipose tissue known as of Cancer Cell, Garsed et al. (2014) investi- sequencing, Garsed et al. (2014) charac- liposarcomas constitutes roughly 20% of gate one of the most convoluted products terize the structure of these remarkably all sarcomas (Dei Tos, 2014). The most of this genomic instability—the neochro- large and highly rearranged structures common subtype of liposarcoma, ac- mosome, characteristic of well-differenti- and propose a model for their genesis counting for nearly half of all cases, is the ated/dedifferentiated liposarcoma (WD/ and growth. Undergoing multiple rounds WD/DDLPS, also referred to as atypical DDLPS). The term ‘‘neochromosome’’ de- of such catastrophic events as chromo- lipomatous tumor in some circumstances. scribes a marker chromosome whose thripsis, breakage-fusion-bridge cycles, Although the higher grade DDLPS is origin cannot be determined by conven- and centromere erosion, their survival is a metastatic in roughly 20% of cases, the tional chromosome banding techniques, testament to the power of selection and primary difficulty of this malignancy is which emphasizes their extreme diver- the ability of tumors to leverage destructive associated with local aggressiveness gence in size and structure from any processes for their own benefit. and recurrence. Early cytogenetic studies

Cancer Cell 26, November 10, 2014 ª2014 Elsevier Inc. 593 Cancer Cell Previews

of WD/DDLPS tumors revealed the strik- tural rearrangements (Greenman et al., opment can be studied in living cells, and ing, consistent presence of ring or giant 2012; Sanborn et al., 2013) in the WD/ the scope of neochromosome sequencing marker chromosomes (Dei Tos, 2014), a DDLPS neochromosomes. The life history is limited to relatively few cell culture sour- feature now recognized as characteristic that emerges from these WD/DDLPS ces. Therefore, of necessity, the exact na- of the disease. These neochromosomes neochromosomes is apparently a tale ture of the earliest events in this process were subsequently shown to contain of disaster upon disaster; it is quite sur- remains inferential. However, the general- high-level amplifications of chromosome prising that anything functional, let alone ity of the structural model presented by 12 region q13-q15 (Dei Tos, 2014), and beneficial for the cell, can be so created. Garsed et al. (2014) could be tested by further mapping pointed to the proto-on- The initial episomal structure appears whole genome sequencing of tumor cogenes MDM2 and CDK4 (Berner et al., to have been generated by stitching DNA samples if it becomes practical to 1996; Nilbert et al., 1994). Mdm2 inhibits together some of the donor loci, including assemble chromosomes de novo with p53-mediated apoptosis and arrest while the chromosome 12 regions, in a nonam- the development of improved long read Cdk4 phosphorylation of Rb1 blocks its plifying process highly reminiscent of sequencing technologies. Other ques- interaction with E2F transcription factors chromothripsis (Stephens et al., 2011). tions also remain. How is chromothripsis governing the G1-S checkpoint. Thus, The inferred double minutes apparently initiated, and why is neochromosome for- this co-amplification provides a selective continue to receive donor sequences mation so strongly associated with certain benefit by coordinating increased survival through additional chromothripsis, but types of cancer? MDM2 and CDK4 ampli- and proliferation. the second stage of their development fication as double minutes or recognizable While the selective benefit of amplifying is dominated by progressive amplifica- insertions in other chromosomes occurs in 12q13-15 may appear straightforward, tion and deletion of material through many different types of cancer, but 12q- the exact structure of this neochromo- breakage-fusion-bridge cycles. While containing neochromosomes are much some and the mechanism by which it generation of double minutes by chromo- more restricted. Intriguingly, they are develops has remained a mystery. The thripsis has been demonstrated previ- particularly enriched in malignant, but WD/DDLPS neochromosomes contain ously (Sanborn et al., 2013; Stephens typically less metastatic, sarcomas such so much genetic material that they are, et al., 2011), Garsed et al.’s model for as- as WD/DDLPS and parosteal osteosar- in fact, the largest chromosomes in the sembly and amplification of the initial coma (O¨ rndal et al., 1992). What favors as- tumor cells. Garsed et al. (2014) begin episomal structures to the much larger sembly of chromosome fragments into by leveraging this to separate the neo- neochromosome is novel. While typical large structures specifically in these histo- chromosomes from the rest of the chro- double minutes do not contain centro- types, and precisely how does this occur? mosome complement by flow-sorting meres and sort randomly to daughter cells Is there an underlying susceptibility to this multiple WD/DDLPS cell lines. Consistent at division, the centromere biology of type of genomic instability in the mesen- with the neochromosome being the most these evolving WD/DDLPS neochromo- chymal lineage that gives rise to these tu- remarkable cytogenetic finding in such tu- somes is apparently quite dynamic. Stan- mors that could be defined biochemically mors, copy number analysis shows that dard alphoid centromeres are acquired (Crasta et al., 2012)? Conversely, do these nearly all of the amplified genetic material but then degraded and lost, while neo- cells experience an elevated selective in the cell occurs on the neochromosome. centromeres are established at other benefit for this type of structure? The Garsed et al. (2014) then use high-depth loci, including across rearrangement inactivation of p53 prior to catastrophic paired-end sequencing of the enriched junctions on the neochromosomes. genomic processes such as chromothrip- neochromosomes to identify how they The final, stabilizing event for the neo- sis has been demonstrated before are stitched together. While each neo- chromosomes appears to be linearization (Rausch et al., 2012). In these cases, it is chromosome is unique, they share and telomere acquisition. While both cir- thought that the initial TP53 mutation several features. First, they each contain cular and linear forms of the neochromo- either facilitates the initiation of or survival a ‘‘core’’ of highly, but unequally, ampli- somes are often found in WD/DDLPS after chromothripsis. In WD/DDLPS, how- fied material from hundreds of genomic tumors, these cell lines contain only linear ever, it appears that chromothripsis pre- loci. Across cell lines, Garsed et al. forms. Linearization appears to be a late cedes and, in fact, leads to p53 inhibition. (2014) found that the median length of event and is accomplished with the acqui- Finally, if the process of neochromosome the donor sequence is 23 Mb, which is sition, but not amplification, of large telo- formation is so turbulent and dynamic, amplified on average 10-fold, with some meric fragments from diverse donor chro- why are the other chromosomes relatively loci amplified considerably more. While mosomes. These additional sequences spared? Is the specificity for chromosome each neochromosome incorporates and extend the length of the neochromo- 12 completely explained by selection for amplifies many different genomic regions, somes by over 100 Mb. 12q genes, or are there structural factors the only shared sequence across all the Garsed et al. (2014) provide a compel- that confer a predilection to this process? neochromosomes is 1.4 Mb from chro- ling description of the architecture of these The thought-provoking study by Garsed mosome 12, which includes MDM2, neochromosomes, and the proposed et al. (2014) has made significant CDK4, and several other genes. model of their genesis and development inroads into the long-standing problem of Garsed et al. (2014) proceed to utilize is also highly consistent with the sequence the chromosome mechanics underlying the information in paired-end sequencing data. Unfortunately, there is no experi- 12q-amplification in WD/DDLPS and will both to reveal the composition as well as mental model in which the process of certainly stimulate additional investigation to model the temporal ordering of struc- 12q-containing neochromosome devel- of this remarkable phenomenon.

594 Cancer Cell 26, November 10, 2014 ª2014 Elsevier Inc. Cancer Cell Previews

REFERENCES Kim, B.W., Kowarsky, M., et al. (2014). Cancer Rausch, T., Jones, D.T.W., Zapatka, M., Stu¨ tz, Cell 26, this issue, 653–667. A.M., Zichner, T., Weischenfeldt, J., Ja¨ ger, N., Remke, M., Shih, D., Northcott, P.A., et al. (2012). Berner, J.M., Forus, A., Elkahloun, A., Meltzer, Greenman, C.D., Pleasance, E.D., Newman, S., Cell 148, 59–71. P.S., Fodstad, O., and Myklebost, O. (1996). Genes Yang, F., Fu, B., Nik-Zainal, S., Jones, D., Lau, Chromosomes Cancer 17, 254–259. K.W., Carter, N., Edwards, P.A.W., et al. (2012). Sanborn, J.Z., Salama, S.R., Grifford, M., Brennan, Genome Res. 22, 346–361. Crasta, K., Ganem, N.J., Dagher, R., Lantermann, C.W., Mikkelsen, T., Jhanwar, S., Katzman, S., A.B., Ivanova, E.V., Pan, Y., Nezi, L., Protopopov, Chin, L., and Haussler, D. (2013). Cancer Res. 73, Nilbert, M., Rydholm, A., Wille´ n, H., Mitelman, F., A., Chowdhury, D., and Pellman, D. (2012). Nature 6036–6045. 482, 53–58. and Mandahl, N. (1994). Genes Chromosomes Cancer 9, 261–265. Dei Tos, A.P. (2014). Histopathology 64, 38–52. Stephens, P.J., Greenman, C.D., Fu, B., Yang, F., O¨ rndal, C., Mandahl, N., Rydholm, A., Wille´ n, H., Bignell, G.R., Mudie, L.J., Pleasance, E.D., Lau, Garsed, D.W., Marshall, O.J., Corbin, V.D.A., Hsu, Brosjo¨ , O., Heim, S., and Mitelman, F. (1992). K.W., Beare, D., Stebbings, L.A., et al. (2011). A., Di Stefano, L., Schro¨ der, J., Li, J., Feng, Z.-P., Cancer Genet. Cytogenet. 60, 170–175. Cell 144, 27–40.

ChIP-ping Away at EWS/ETS Transcription Networks

Christopher T. Denny1,2,3,* 1Division of Hematology/Oncology, Department of Pediatrics, Gwynne Hazen Cherry Memorial Laboratories 2Jonsson Comprehensive Cancer Center 3Molecular Biology Institute University of California, Los Angeles, Los Angeles, CA 90095, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.ccell.2014.10.016

In this issue of Cancer Cell, Riggi and colleagues use a genomic approach to define two distinct molecular mechanisms through which the chimeric EWS/FLI1 oncoprotein regulates target genes in Ewing sarcoma, expanding a framework upon which to model the target gene network and test strategies for antagonizing growth of this tumor.

Twenty two years ago, Delattre and tion factor. However, the portion of EWS incomplete. In general, forced expression Thomas pried open the door on Ewing that was fused to FLI1 was found to be un- of any single EWS/FLI1 target gene did sarcoma biology by reporting that the structured, making it poor bait for protein not recapitulate the EWS/FLI1 phenotype t(11,22) found in the large majority of interaction screens and a difficult target in cells. Conversely, target gene inhibition these tumors created a fusion gene be- to develop small molecule antagonists frequently did not completely shut down tween EWS and FLI1 (Delattre et al., against (Ng et al., 2007). EWS/FLI1, like the cell transformation effects of EWS/ 1992). Finally, there was an unambiguous many of the oncogenic chimeric fusions FLI1. biomarker for a malignancy that, up until found in other sarcomas, earned the ‘‘un- This suggested that EWS/FLI1 trans- then, had been diagnosed primarily by druggable’’ label. The fact that in the last formed cells through the cumulative effect exclusion. Of even greater importance two decades there has been only one of transcriptionally modulating a network was the prevalent belief that a somatic candidate small molecule that specifically of genes. Inherent in such systems is a mutation that was present in 85% of Ew- targets the EWS/FLI1 fusion, suggests central robustness that can tolerate loss ing tumors must be playing crucial onco- that, at least so far, this reputation is of one or more nodes while still maintaining genic roles. This contention was quickly deserved (Barber-Rotenberg et al., 2012). the overall effect of the network (Friedman born out. Not only could ectopic expres- So, if directly antagonizing EWS/FLI1 and Perrimon, 2007). As success in the sion of EWS/FLI1 avidly transform cells, seemed unfeasible, perhaps targeting search for EWS/FLI1’s Achilles heel was but inhibition of the fusion consistently genes that were transcriptionally modu- proving progressively more unlikely, the induced growth arrest of Ewing sarcoma lated by the fusion would be a therapeuti- need to generate a comprehensive map tumor derived cell lines. Who could ask cally more tractable strategy. As molecu- of the target gene network became more for a better therapeutic target for this lar methods advanced, the number of urgent. Although earlier work provided deadly disease? EWS/FLI1 target genes that were identi- snapshots of this network, it was difficult And then things got difficult. Structural fied geometrically increased. The prob- to paste the pictures together into a function analyses of EWS/FLI1 indicated lem was that demonstrating biologic coherent whole. What was needed was a that the fusion was mediating its biologic relevance of these candidates proved to wide-angle lens through which to view effects by acting as an aberrant transcrip- be labor intensive, unpredictable, and the broad EWS/FLI1 target gene network.

Cancer Cell 26, November 10, 2014 ª2014 Elsevier Inc. 595