NATURE|Vol 454|7 August 2008 FEATURE Moving AHEAD with an international human epigenome project A plan to ‘genomicize’ research and pave the way for breakthroughs in the prevention, diagnosis and treatment of human disease. Me The American Association for Cancer Figure 1 | Epigenetic Research Human Epigenome Task Force mechanisms. The coding c and structural information and the European Union, Network of in the base sequence of DNA Excellence, Scientific Advisory Board g g is organized in chromatin to form multiple epigenomes. It is now possible to define whole epigenomes, c c DNA cytosine methylation and representing the totality of epigenetic marks g covalent modifications of the in a given cell type. Epigenetic processes are tails of histones and histone essential for packaging and interpreting the Me variants contribute information , are fundamental to normal develop- to nucleosomal remodelling ment and are increasingly recognized as being Me machines that render genes and involved in human disease. Epigenetic mech- non-coding RNAs susceptible anisms include, among other things, histone to transcription. Transcription modification, positioning of histone variants, factors (not shown) also play a nucleosome remodelling, DNA methylation, major part in the competence Me and organization of the small and non-coding RNAs (Fig. 1). These epigenome. The AHEAD project mechanisms interact with transcription fac- will map epigenetic marks in a tors and other DNA-binding proteins to regu- defined set of epigenomes. late gene-expression patterns inherited from cell to cell. The patterns underlie embryonic development, differentiation and cell identity, transitions from a stem cell to a committed cell and responses to environmental signals such as hormones, nutrients, stress and damage. Ac Although epigenomic changes are herit- Chromatin remodeller able in somatic cells, drug treatments could Me potentially reverse them. This has significant implications for the prevention, diagnosis and treatment of major human diseases and Histones Transcription for ageing. Diseases to be targeted could include diabetes, cardio pulmonary diseases, Me Histone tails Rett syndrome, other neurological disorders, Ac imprinting disorders, autoimmune diseases Me and cancer, in which missteps in epigenetic Non-coding RNAs programming have been directly impli- Chromosome cated. Indeed, several inhibitors of chroma- tin- modifying enzymes including histone deacetylase (HDAC) inhibitors and DNA methyltransferase (DNMT) inhibitors have now been approved by the US Food and Drug Administration or are in clinical trials with good prognosis for tumour regression. Epige- netic therapy is now a reality, but to maximize the potential of such therapeutic approaches, the human epigenome, and we urge the AHEAD project is to provide high-resolution it is crucial that there be a more comprehen- community to join in a coordinated effort in reference epigenome maps. These maps would sive characterization of the epigenetic changes support of the Alliance for the Human Epig- be of great use in basic and applied research, that occur during normal development, adult enome and Disease (AHEAD) to help solve would have an immediate impact on under- cell renewal and disease, and of the relation- the problems of cancer and other intractable standing many diseases, and would lead to ships between genetic and epigenetic variation diseases. Just as the Human the discovery of new means to control the and their impact on health. provided a reference ‘normal’ sequence for diseases. Although the project should have a The time is right for a major effort to decode studying human disease, the goal of the human focus, it will be essential to use model

711 FEATURE NATURE|Vol 454|7 August 2008

Box 1 | A European success story European researchers in the as well as 350 laboratories late 1990s were quick to take worldwide that participate in up the fast-emerging field of ongoing activities through the epigenetics, building on early NoE website. Expansion of the discoveries in X-chromosome NoE resulted in the creation of inactivation, Polycomb Associate Members and, most regulation, dosage compensation, importantly, the NETs. Through the imprinting, heterochromatin, NET programme, talented young DNA methylation, nucleosome European investigators in their first modification and RNA interference four years of independent research (RNAi). The Epigenome Network are selected to participate and of Excellence (NoE) was created provided with moderate financial by the European Commission support. The NET programme has in 2004 with the long-term aim been a resounding success and of tackling major epigenetic has given a powerful impetus to a questions in the post-genomic era. cohort of young research talent. Under the direction of Thomas The NoE programme of activities Jenuwein of the Research Institute has promoted a common identity of Molecular Pathology in Vienna, and generated numerous in lay terms are provided in seven momentum achieved should be assisted by Phil Avner of the collaborations through training and languages. Several NoE members nurtured, supported and exploited Pasteur Institute in Paris and mobility programmes, workshops, are involved in biotechnology to broaden connections with Geneviève Almouzni of the Curie meetings and dissemination of companies and technology transfer biotechnology, agriculture and Institute in Paris, the intention was technologies. In 2007 alone, has been the subject of NoE particularly with the biochemical to create a durable structure for the NoE collectively published workshops. and pharmaceutical industries epigenetics research in Europe, some 300 research articles. These achievements have that are such key components promoting collaborations, sharing A major effort has also been resulted from a conjunction of of the European technological technologies and resources, made to reach out to the broader circumstance, hard work and the landscape. With the worldwide and training and fostering the community with a scientific talent assembled by the NoE. interest in epigenetics, this seems development of leaders in the field. website (www.epigenome-noe. Europe is in many respects pre- an inopportune moment for the Now, after four of the five years net), and to the general public, eminent in the field of epigenetics European Commission to reduce planned, the NoE includes an through conferences, newspaper and the NoE could serve as a model support for this research and risk original core of 25 members articles, and a public website for similar programmes worldwide. losing momentum and brainpower. plus 35 associate members and (www.epigenome.eu), where But, as its five-year mandate On the contrary, the opportunity 23 Newly Established Teams news, information, explanations approaches its end, the future of should be seized to forge close (NETs): a total of 83 laboratories of the concepts, significance and the NoE remains uncertain. The links with international efforts such from 12 European countries, applications of epigenetic research organization, coordination and as AHEAD. organisms to obtain mechanistic insights as to Project), chromatin profiling (HEROIC, a computational infrastructure for data shar- the functionality of epigenomic parameters or High-Throughput Epigenetic Regulatory ing and display that puts epigenetic data in ‘codes’. Studying, genome-wide, the increasing Organization In Chromatin), and treatment the hands of non-epigenetic investigators1. number of interacting epigenetic mechanisms of neoplastic disease (EPITRON, EPIgenetic The American Association for Cancer in a spectrum of cell types will involve data Treatment Of Neoplastic Disease). A spe- Research (AACR) organized a Human Epi- generation on a massive scale. An international cial function is provided by the Epigenome genome Workshop in June 2005 and also a project would provide the infra- Network of Excellence (NoE), created by the follow-up workshop in July 2006 that focused structure needed to ensure that epigenom- European Commission in 2004 (see Box 1). on planning an international project to map a ics research can be integrated with genomic In the past few years, there have been defined number of human epigenomes (http:// data, and be utilized efficiently to advance the several efforts to organize the epigenetics www.aacr.org/page9673.aspx)2. The consen- knowledge of human health and disease. Here, research community in the United States sus of workshop participants was that there are we discuss the benefits of the AHEAD frame- and develop the support and structure for a compelling reasons from both scientific and work to coordinate and plan an international human epigenome project. A 2004 National public-health perspectives to initiate a human Human Epigenome Project. Cancer Institute (NCI)-sponsored Epige- epigenome project that could take full advan- netic Mechanisms in Cancer Think Tank tage of advances in several existing US and Early steps concluded that the development of a US European initiatives. Individual investigators have been study- human epigenome project of analogous On the heels of these workshops, the AACR ing epigenetics for several decades; however, scope to the should Human Epigenome Task Force, a cross-disci- concerted efforts to organize the epigenom- be of the highest priority (http://www.cancer. plinary group of international investigators, ics research community are quite recent, and gov/think-tanks-cancer-biology/page7). This was formed to design a strategy and develop none has sought to engage the community on was followed by an NCI workshop in 2005 a timetable for the implementation of an a broad-based, international level. attended by programme staff from many international Human Epigenome Project. As Europe has a strong tradition for epige- National Institutes of Health (NIH) institutes. described below, the task force recommends netics research that has been recognized Key recommendations were: 1) comprehen- the formation of AHEAD to coordinate a by European Union funding programmes sive analysis of reference epigenomes, focused transdisciplinary, international project to map and by individual national initiatives. More on stem cells and key differentiated lineages a defined subset of robust epigenetic markers than €50 million (US$79 million) has been that can be modelled experimentally; 2) in a limited number of human tissues at dif- allocated to networks and consortia that development of a standardized set of reagents, ferent stages of development and to develop focus on central epigenetic questions such as including monoclonal antibodies and com- a bioinformatics infrastructure to support DNA methylation (HEP, Human Epigenome mon tissue substrates; and 3) development of the collection of epigenomic data. The efforts

712 NATURE|Vol 454|7 August 2008 FEATURE

and reflect the current and past environments Table 1 | Potential reference cell systems for epigenomic analysis of individual cells. Epigenetic programming Organ system Tissue progenitor Differentiated cell Altered disease state may even precede lineage choice and amplify Embryo Embryonic stem cells signals from the environment. A major goal of Haematopoietic CD34+ Polymorphonuclear Acute myelogenous leukemia, AHEAD would be to map the epigenome in leukocytes, lymphoid, etc. Acute lymphoblastic leukemia, Chronic myelogenous leukemia, normal tissues and differentiating cells so that Myelodysplastic syndrome they could be compared with disease states, Breast CD24+and/or CD44+ Myoepithelial, luminal, Breast cancer including but not restricted to cancer, that fibroblasts arise in these tissues. Therefore, the selection Liver Oval cell Hepatocytes, stromal cells Cirrhosis, viral disease, of which cell types to use and which epigenetic Hepatocellular carcinoma modifications to examine deserves careful consideration. Many environmental factors, of this group will surely be bolstered by the as a stem-cell phenotyope, proliferation, such as ageing, hormonal milieu, diet and recent and exciting decision of the NIH to fund differentiation, senescence and stress, using infection must be taken into account. Empha- epigenomics research as an interagency Road- common core specimens from different cell sis should be placed on comparing embryonic map initiative. types, and to develop a bioinformatics infra- stem cells, adult stem and precursor cells, and Asia, too, has been active in fostering epi- structure to support the collection and integra- related differentiated cells of primary human research, with a major emphasis tion of epigenomic data. The precise objectives tissues, with the diseased state. placed on disease epigenomes, especially those of such a project still need to be delineated. The reference systems selected should meet in liver and gastric cancers. An international However, we would almost certainly expect several criteria: 1) cells should be easy to sam- meeting, Genome-wide Epigenetics 2005, was AHEAD to 1) provide complete epigenome ple in a reproducible fashion (renewable if pos- held in Tokyo3. Scientists from Yonsei Uni- maps at very high resolution for important sible); 2) cell numbers should be sufficient for versity (South Korea), the National Cancer histone modifications across diverse cel- analyses of DNA methylation and chromatin Center (Japan), the Shanghai Cancer Insti- lular states in both human and mouse; 2) to modifications; 3) cell progenitors should be tute (China) and the Genome Institute (Sin- complete and catalogue epigenome maps of identified that can be suitably manipulated gapore) also organized meetings to facilitate model yeasts (Saccharomyces cerevisiae and and harvested in a pure state; 4) where possible, the exchange of information in epigenomics S. pombe), plants (Arabidopsis and rice), and cells should be amenable to tissue reconstruc- and held their first meeting in Seoul in 2006, animals (Caenorhabditis elegans and Dro- tion and three-dimensional model systems; followed by another conference in Osaka in sophila); 3) to deliver a high resolution DNA and, of course, 5) systems must provide insight 2007. In December 2006, a Japanese Society methylation map of the entire human genome into key differentiation and related disease for Epigenetics was formed. Clearly, Asia is in defined cell types and a landmark map for states. Some examples of suitable systems are now poised to contribute strongly to global transcription start sites of all protein coding given in Table 1. epigenomics research. genes and a representative number of other Specific modifications to histones often National support for the Human Epigenome features throughout the genome; 4) define correlate with gene activation or repression Project is also mounting in Australia with non-coding and small RNAs; and 5) to estab- — for example lysine acetylation and trimeth- the formation of the Australian Alliance for lish a bioinformatics platform including a ylation of lysine 4 of histone H3 (H3K4me3) Epigenetics to begin in late 2008. Australian relational database, website and suite of ana- are permissive for gene activation whereas meetings devoted to epigenetics were initi- lytical tools to organize, integrate and display H3K9me2 and H3K27me3 correlate with ated in 1996 in Heron Island with Susan Clark whole epigenomic data on model organisms transcriptional silencing. Often, activating of the Garvan Institute of Medical Research, and humans. AHEAD will develop standard- and repressive histone marks co-exist at gene Sydney, holding the first workshop on bisul- ized procedures that assure data quality in the start sites, reflecting perhaps epigenetic het- phite sequencing and this has been followed choice of reagents (antibodies), the experi- erogeneity among otherwise similar cells, by biannual meetings, the most recent being mental procedures (a minimal set of biologi- and it is the balance between these marks that held in Perth in November 2007 to showcase cal replicates) and data analysis (appropriate determines gene expression states4–6. Because Australia’s strengths in the global epigenetic statistical procedures). there are so many chromatin modifications, research arena. AHEAD would differ from and comple- it may prove challenging to examine all of ment other ongoing projects such as ENCODE them in every cell type. We recommend that The scope of AHEAD (ENCyclopedia Of DNA Elements). The the initial set of reference epigenomes focus AHEAD would aim to provide reference ENCODE project is focused on defining the on several of the better understood covalent epigenomes for key cellular states, such functional sequences in the genome, whereas markers, such as those listed in Table 2, that AHEAD would define the patterns of epige- could be mapped in a ‘first pass’ project defin- Box 2 | Heritability netic regulation occurring at those sequences ing reference epigenomes. Epigenetics is often defined as somatically in different cell states. The potential synergies Several high-profile studies, including the heritable changes in gene expression that between these two projects in terms of devel- human ENCODE maps7 and genome-wide do not involve changes in base sequence. opment of innovative molecular and compu- DNA methylation patterns are known to tational techniques would enhance each of Table 2 | Recommended histone markers with be heritable through S phase of the cell them. It would be valuable to coordinate such known function to define reference epigenomes cycle, however it has been more difficult to individual projects in the interests of efficiency Histone modification Function demonstrate mechanisms for the heritability and potential scientific insights. Some of these H3K9ac, H4K16ac Transcriptional ‘ON’ states of other chromatin modifications that will be objectives are exemplified by the current and H3K4me mapped. Also, the abilities of transcription efforts on model organisms carried out by the H3K9me and Transcriptional ‘OFF’ states factors to reprogram the epigenome in modENCODE (Model Organism ENCyclope- H3K27me normal and experimental situations need dia Of DNA Elements) programme. H3K36me Elongation; primary to be considered. The AHEAD project will transcripts help illuminate processes of heritability and H3K9me and HP1 Chromosome organization reprogramming by providing genome-wide Reference epigenomes H4K20me and Stress/damage response reference maps. In a multicellular organism, there are many potential epigenomes that define each cell type H2BS14ph

713 FEATURE NATURE|Vol 454|7 August 2008 profiling for histone modifications in mouse robustness of modification-specific antibodies. allowing cell-type specific epigenomic and human cells8,9, have been published Rapid progress has also been made towards profiles to be determined through cell sorting recently. Although they present valuable characterizing the global distribution of cyto- or immunoprecipitation. It is anticipated that first-generation epigenetic maps, these efforts sine methylation. Genome-scale assays for new technologies such as massively parallel largely focused on transcriptional regula- detecting this epigenetic modification fall sequencing will make the generation of these tion and functional sequences. We still need into two general categories. The gold-stand- profiles so straightforward that the limitations to know more about epigenetic control and ard technique for reading out the methyla- will be in the data analysis and the ability to plasticity in intergenic regions, mechanisms tion state of individual cytosines is bisulphite take advantage of such genetic and cell bio- of heritability (Box 2), the role of repetitive sequencing developed by Clark and Marianne logical tools. Reference epigenomes for sev- elements, non-coding and small RNAs, and Frommer at the University of Sydney10, in eral model organisms including S. cerevisiae, finally how epigenetic marking contributes to which unmethylated cytosines are converted Arabidopsis, Drosophila, C. elegans, mouse and chromosome segregation, stress response and to uracils and read as thymine, while meth- others will form an integral part of AHEAD. transducing signals from the environment. ylated ones are protected from conversion. Clearly, AHEAD would enable coordination Although this method yields precise nucle- and expansion of these projects and rapid Advances in technology otide-resolution data, bisulphite sequencing translation into human studies. Our understanding of the epigenome has been is challenging to scale — regions of interest transformed in recent years by a succession of must be sequenced several times in order Computational challenges technological innovations. Approaches involv- for methylation patterns in a given cell type As a comprehensive human epigenome ing microarrays and, most recently, ultra-high- to be appreciated. A recent study showed the project will require a strong bioinformatics throughput sequencing technology have been potential of combining bisulphite methodol- platform, an early priority of AHEAD would applied to map chromatin modifications, cyto- ogy with ultra-high-throughput sequencing be to establish a central relational database sine methylation and non-coding RNAs across by deciphering the entire Arabidopsis ‘DNA and a web interface to present data to the chromosomes and even entire . methylome’ at nucleotide resolution11. Still, scientific community. This website would Genome-scale studies of histone modifica- bisulphite studies of the much larger human include analytical and statistical tools that tions and other aspects of chromatin structure methylome have thus far been limited to rela- could be used dynamically to process data have typically relied on an immunological tively small subsets of genome12. For more for visualization. It would have tremendous procedure, chromatin immunoprecipitation, comprehensive coverage, many investiga- utility for investigators, allowing immediate in which specific antibodies are used to enrich tors have turned to alternative approaches access to detailed maps for a locus of inter- chromatin. For example, an antibody against that involve the isolation of methylated est in the same way that the Human Genome histone H3 acetylated at lysine 9 can be used (or unmethylated) fractions of genome by Project provides sequence data. By initially to isolate genomic regions associated with methylation-sensitive restriction or immuno- focusing on establishing data interoperability, this activating modification. Isolated DNA precipitation with a methylcytosine-specific it is hoped that the database, or later versions is then interrogated by microarray hybridi- antibody. The isolated fractions are then inter- of it, would allow for the simultaneous display zation or by deep sequencing. Microarrays rogated on microarrays. Several successes have of integrated epigenomic parameters on the are a well-established readout, particularly been described recently using this genome entire human genome. In addition, not yet suited to studies that require high coverage of fractionation/microarray approach, including addressed is how the complex system of epige- a small subset of a given genome (for example, a profile of promoter methylation in human nomic regulation should be treated as a whole: annotated gene promoters). Resolution on fibroblasts13 and drafts of the complete Arabi- the data could be a powerful resource for the the order of the nucleosome (a few hundred dopsis methylome14–16. emerging field of , allowing bases) can be achieved with sufficiently dense insights into the stability of the epigenomic tiling arrays. Recent studies — several dozen Model organisms networks and how they can be perturbed in whole mammalian genome data sets have Model organisms have led the way in under- disease states. already been reported8,9 — have leveraged standing epigenetic mechanisms of gene regu- The development of these resources would emerging ultra-high-throughput sequencing lation. Position effect variegation, imprinting, be coordinated with parallel efforts already technology to ‘deep-sequence’ chroma- transposon silencing, DNA methylation and under way as part of the caBIG (the NIH Can- tin-immunoprecipitated DNA to generate histone modification were all discovered in cer Biomedical Informatics Grid) and Cancer genome-wide chromatin state maps. This model organisms and later found in humans. Genome Atlas initiatives, among others. These sequencing approach offers potential advan- RNA interference and its role in epigenetics resources could be distributed through the tages over arrays in precision, throughput were also first described in model organisms MGED (Microarray Gene Expression Data and genome coverage. Notably, sequencing (plants and worms). Society) and the BioConductor (open-source requires orders-of-magnitude less DNA than The development of infrastructure to study software) organizations. In addition, one could microarrays, and this enhanced sensitivity model-organism epigenomes is under way and build on the sophisticated approaches of exist- should enable studies of primary tissues, dis- is having a major impact on human epigenome ing genome browsers used for data access and ease samples, and other limited cell popula- research, with hundreds of whole-genome or visualization, such as those at University of tions of high biological importance. whole-chromosome profiles of histone modi- California, Santa Cruz, and ENSEMBL, to The quality of the data derived from stud- fications and DNA methylation already pub- develop innovative means of data representa- ies relying on chromatin immunoprecipi- lished in yeast, Arabidopsis, Drosophila and tion and web-based data analysis that would be tation depends mainly on the quality of the the mouse. Importantly, mutants in epigenetic of immense value to and foster collaborations antibodies used, and there is an essential need modification, gene silencing, development among the scientific community. for high-titre polyclonal antibodies with high and metabolism, as well as disease models, specificity. Standardization of monoclonal can be readily obtained and manipulated, and Summary and perspectives antibodies has proven difficult because of will allow the epigenetic pathways that are Dramatic changes in technologies have made low avidity and only limited epitope recogni- responsible for controlling genomic output to AHEAD eminently feasible. Successful coor- tion. Reference maps defined by AHEAD, as be determined first in model organisms. Other dination of this multifaceted project through well as those already developed from studies tools such as lineage-specific green fluorescent an AHEAD International Steering Commit- in human cells and model organisms, would protein reporters and tagged histones have tee would be essential if rapid progress is to be provide a platform to assess the quality and been engineered in mouse and Arabidopsis, made, and it will require leadership from many

714 NATURE|Vol 454|7 August 2008 FEATURE

11. Cokus, S. J. et al. Nature 452, 215–219 (2008). stakeholders at all levels. The major challenges are important in helping establish the infra- 12. Eckhardt, F. et al. Nature Genet. 38, 1378–1385 (2006). of AHEAD include the initial selection of the structure needed to advance the field. We must 13. Weber, M. et al. Nature Genet. 39, 457–466 (2007). most important epigenomic markers, the iden- now move AHEAD in earnest to achieve the 14. Zhang, X. et al. Cell 126, 1189–1201 (2006). 15. Zilberman, D., Gehring, M., Tran, R. K., Ballinger, T. & tification of appropriate, pure cell populations, goals of an international Human Epigenome Henikoff, S. Nature Genet. 39, 61–69 (2007). and the handling and presentation of the data. Project. ■ 16. Vaughn, M. W. et al. PLoS Biol. 5, e174 (2007). None of these challenges is insurmountable, and there are enormous potential scientific 1. Feinberg, A. P., Jones, P. A. & Ault, G. National Cancer Acknowledgements We gratefully acknowledge Institute (NCI), Division of Cancer Biology (DCB) Kimberly Sabelko from the AACR for her assistance and public health benefits. Workshop on Defining the Epigenome: Addressing the AHEAD would effectively integrate epi- Value and Scope of a Human Epigenome Project. http:// in editing this manuscript. We also thank the AACR research that is currently being con- dcb.nci.nih.gov/Workshoprpt.cfm. and Margaret Foti for supporting and funding the 2. Jones, P. A. & Martienssen, R. Cancer Res. 65, 11241–11246 establishment of the Human Epigenome Task Force. ducted on a piecemeal basis around the world (2005). and would pave the way for breakthroughs in 3. Lieb, J. D. et al. Cytogenet. Genome Res. 114, 1–15 (2006). Author contributions P.A.J., S.B.B., B.E.B., A.P.F., understanding normal and pathological proc- 4. Azuara, V. et al. Nature Cell Biol. 8, 532–538 (2006). J.M.G., T.J., R.M., T.U., and V.P., C.D.A., S.C.E., J.R. and esses. This research would offer significant sci- 5. Bernstein, B. E. et al. Cell 125, 315–326 (2006). C.W. all contributed to this manuscript. 6. Guenther, M. G., Levine, S. S., Boyer, L. A., Jaenisch, R. & entific opportunities to maximize translational Young, R. A. Cell 130, 77–88 (2007). Author Information Correspondence and requests research to prevent and cure human diseases. 7. Birney, E. et al. Nature 447, 799–816 (2007). for materials should be addressed to P.A.J. (e-mail: The selection of epigenetics as a new NIH 8. Barski, A. et al. Cell 129, 823–837 (2007). [email protected]). 9. Mikkelsen, T. S. et al. Nature 448, 553–560 (2007). Roadmap initiative in the United States and 10. Clark, S. J., Harrison, J., Paul, C. L. & Frommer, M. F. Nucleic the European Union’s Network of Excellence Acids Res. 22, 2990–2997 (1994). See Tech Feature, page 795.

The American Association for Cancer Research Human Epigenome Task Force Peter A. Jones1, Trevor K. Archer2, Stephen B. Baylin3, Stephan Beck4, Shelley Berger5, Bradley E. Bernstein6, John D. Carpten7, Susan J. Clark8, Joseph F. Costello9, Rebecca W. Doerge10, Manel Esteller11, Andrew P. Feinberg12, Thomas R. Gingeras13, John M. Greally14, Steven Henikoff15, James G. Herman3, Laurie Jackson-Grusby16, Thomas Jenuwein17, Randy L. Jirtle18, Young-Joon Kim19, Peter W. Laird1, Bing Lim20, Robert Martienssen21, Kornelia Polyak22, Henk Stunnenberg23, Thea Dorothy Tlsty24, Benjamin Tycko25, Toshikazu Ushijima26 and Jingde Zhu27

The European Union, Network of Excellence, Scientific Advisory Board Vincenzo Pirrotta28, C. David Allis29, Sarah C. Elgin30, Peter A. Jones1, Robert Martienssen21, Jasper Rine31 and Carl Wu32

1USC/Norris Comprehensive Cancer Center, Los Angeles, California 90089, USA. 2NIH–NIEHS, Research Triangle Park, North Carolina 27709, USA. 3Johns Hopkins University, Baltimore, Maryland 21231, USA. 4Wellcome Trust Sanger Institute, Hinxton, Cambridge CB10 1SA, UK. 5The Wistar Institute, Philadelphia, Pennsylvania 19104, USA. 6Massachusetts General Hospital, Charlestown, Massachusetts 02129, USA. 7TGen, Phoenix, Arizona 85004, USA. 8Garvan Institute of Medical Research, Sydney, NSW 2010, Australia. 9Department of Neurological Surgery, University of California, San Francisco, San Francisco, California 94143, USA. 10Purdue University, West Lafayette, Indiana 47907, USA. 11Spanish National Cancer Center, Madrid, Spain. 12Departments of Medicine, Oncology and Molecular Biology & Genetics, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, USA. 13Affymetrix, Santa Clara, California, 95051, USA. 14Albert Einstein College of Medicine, Bronx, New York 10461, USA. 15Fred Hutchinson Cancer Research Center, Seattle, Washington 98104, USA. 16Harvard Medical School, Boston, Massachusetts 02115, USA. 17Research Institute of Molecular Pathology, Vienna 1030, Austria. 18Duke University Medical Center, Durham, North Carolina 27710, USA. 19Yonsei University, Seoul 120-719, Korea. 20Genome Institute of Singapore, Singapore 138672. 21Cold Spring Harbor Laboratory, Cold Spring Harbor, New York 11724, USA. 22Dana-Farber Cancer Institute, Boston, Massachusetts 02115, USA. 23Radboud University, Nijmegen 6525, the Netherlands. 24Department of Pathology, University of California, San Francisco, San Francisco, California, 94143, USA. 25Columbia University, New York, New York 10032, USA. 26National Cancer Center Research Institute, Tokyo 104-0045, Japan. 27Shanghai Cancer Institute, Shanghai 200032, China. 28Rutgers University, Piscataway, New Jersey 08854, USA. 29Rockefeller University, New York, New York 10021, USA. 30Washington University, Saint Louis, Missouri 63130, USA. 31University of California Berkeley, Berkeley, California 94720-3202, USA. 32National Cancer Institute, Bethesda, Maryland 20892, USA.

715