After studying the genomics of multiple sclerosis (MS) for over a decade, Sergio Baranzini, a professor of neurology in the University of California’s Multiple Sclerosis Research Group, still could not trace the disease’s heritability. So Baranzini turned to epigenetics, hoping to identify heritable modifications other than DNA sequence variation that led to MS symptoms.

In 2010, Baranzini and colleagues performed genome-wide epigenetic analysis of three sets of identical twins where one of the twins had been diagnosed with MS (1). “The hypothesis was that the discordance among these twins would be in part due to either rare variation that arose after embryonic development, or epigenetic modifications that were different in the twins,” says Baranzini. But his team found nothing. The immune cells from the twins had identical epigenetic profiles, suggesting that the disease phenotype had little to do with epigenetic changes.

Nonetheless, several large-scale projects to further explore the epigenetic frontier—including the National Institutes of Health (NIH) $190 million Epigenetics Roadmap Initiative and the International Human Epigenome Consortium (IHEC)—have already saddled up. But some researchers are quietly warning those involved that these epigenetics projects may not live up to the hype.

Saddle up

Unlike the Human Genome Project, the proposed epigenome project would require multiple approaches to characterize different epigenetic marks, as well as the bioinformatics to integrate such data. Unable to be sequenced like a traditional genome, a complete epigenome requires the characterization of all known gene-regulating mechanisms not involving changes in DNA sequence: DNA methylation, histone modification, chromatin accessibility, and small non-coding RNA expression. To examine these different mechanisms, researchers use chromatin immunoprecipitation sequencing (ChIP-seq) of histone modifications, bisulfite sequencing of the cytosine methylation (MethylC-seq), mapping of active gene regulatory elements by sequencing DNase I–hypersensitive sites (DNase-seq), and sequencing of RNA expression profiles (RNA-seq).

In addition to the multiple techniques required, the project also faces other technical challenges. Commercially available ChIP-grade antibodies suffer from lot-to-lot variation, leading to nonspecific binding and questionable datasets. Also, epigenetic techniques still have certain limitations, such as the detection of 5-hydroxymethylcytosine (5-hmC), which reacts poorly with bisulfite. Furthermore, the management and integration of these different datasets presents another barrier.

Outside the mere technical hurdles, any large-scale project must accommodate for the fact that the epigenetic landscape is constantly changing: unlike the human genome, each tissue in an individual has a different epigenome. This means that instead of producing a single reference dataset, the project will have to characterize and compare different cell types from different individuals, both healthy and diseased, in order understand how these epigenetic mechanisms function on a systems level.

But Jones and his colleagues believed the time was right to start mapping out the project. A number of workshops followed, culminating in the National Institutes of Health (NIH)’s $190 million five-year Epigenomics Roadmap Initiative in January 2008. With the epigenomics initiative, the NIH sought to create a series of epigenome maps that would help to study epigenetic mechanisms, develop new epigenetic analysis techniques, and standardize practices and procedures for epigenomics.

As part of the initiative, the Roadmap Epigenomics Mapping Consortium was formed to provide a public database for human epigenomics data. Four Reference Epigenome Mapping Centers were established at the Massachusetts Institute of Technology, University of California, San Francisco, the Ludwig Institute for Cancer Research, and the University of Washington. A Data Analysis and Coordinating Center at Baylor College of Medicine coordinates experimental and analytical efforts to maximize consistency, data quality, and overall coverage of the epigenomic landscape.

In 2009, the consortium published its first human epigenome (2). Led by Joseph Ecker at the Salk Institute for Biological Studies, the team published maps of methylated cytosines from human embryonic stem cells and fetal fibroblasts, as well as small RNA expression and histone modification profiles. They identified several differences between the epigenomes of these two cell types.

But Jones still has his eye on a much loftier goal: 1000 human epigenomes in 10 years. This is the number that he believes will truly provide a framework to study the complete epigenetics landscape. One thousand epigenomes from different tissues and individuals would catalogue the different epigenetic marks in different cell types. Then researchers could really begin to make associations between these marks and phenotypes.

To pave the way, Jones organized a steering committee to outline the initiatives’ goals, standards, and collaborations. In January 2010, the steering committee officially launched the IHEC at a meeting in Paris, looking to raise the $130 million necessary for its first phase. The project would coordinate the mapping of epigenomes from not only the NIH’s Epigenomics Mapping Consortium but also from international efforts such as the European Epigenome Network of Excellence, the Danish National Research Foundation’s Center for Epigenetics, and the Australian Epigenetic Alliance. Rob Martienssen is a researcher at Cold Spring Harbor Laboratory and a member of the consortium’s interim executive committee. “For me, the importance of the project is understanding the chromosomes,” he says. “The immediate consequences of the genome in every cell type gives us an unparalleled idea of what these genes and signals are actually doing.”

Fool’s errand

“People work on [epigenetics] in many other ways, so I just don’t see the value of this,” says Kevin Struhl, an epigenetics researcher in the Department of Biological Chemistry and Molecular Pharmacology at Harvard Medical School who signed the letter. While certain that epigenetics is an important field, he says he doesn’t believe large-scale epigenome mapping projects will deliver any significant insights into cell biology and human disease, and that almost nothing will be gained from mapping 1000 epigenomes—let alone 100, 50, or even 10. “It’s a colossal waste of money.”

The IHEC has come under fire as well. In March 2010, following the official launch of the project, three researchers—Mark Ptashne from Sloan Kettering Institute, Oliver Hobert from Columbia University Medical Center, and Eric Davidson from Caltech—published a letter in Nature highlighting that epigenetic marks are the result of protein interactions with regulatory sequences in the genome (4). Therefore, they said, studying the control of gene expression by these regulatory sequences would probably provide more insights into cell diversity.

It doesn’t help either that the organizers of IHEC project continuously liken it to the Human Genome Project, which—despite its successes—has failed to live up to its initial claims of making a immediate and significant impact on human health. In the ten years since the first human genome was sequenced, researchers have been grappling with a much more complicated entity than initially anticipated, and the human reference genome is still incomplete: portions of each chromosome remain inaccessible despite advances in sequencing techniques and instruments.

On top of all this, there’s Baranzini’s 2010 negative MS twin study. In the study, the group analyzed the methylation patterns of CD4+ lymphocytes, whose function is disrupted in MS patients. Focusing on lymphocytes from three MS-discordant identical twins using reduced representation bisulfite sequencing—which targets sequences enriched with CpG dinucleotides for bisulfite conversion and sequencing—Baranzini’s group concluded that there was no epigenetic difference that could explain the disease. The identical twins’ epigenomes were identical despite one having the disease. Would a large-scale epigenome mapping project turn up evidence overlooked in a study like Baranzini’s, or rather produce another dead-end, providing no insight to the molecular origins of disease?

Martienssen is quick to point out that several other discordant twin studies have identified differences in methylation patterns related to schizophrenia, bipolar disorder, caudal duplication anomaly, Beckwith–Wiedemann syndrome, and autoimmune disease. For example, a 2010 study led by Spanish epigenetics researcher Esteban Ballestar found 49 differently methylated regions in identical twins discordant for the autoimmune disorder systemic lupus erythematosus (5). These methylation markers could be used not only for the clinical characterization of systemic lupus erythematosus patients, but also as new targets for drug development.

So why did Baranzini’s study fail to identify any epigenetic differences? Well, either epigenetics does not play a significant role in MS or Baranzini’s methods may have been insufficient. Taking a closer look at Baranzini’s methods, at least three UK researchers believe the jury is still out. Writing in Expert Reviews in November 2010, Wellcome Trust Center for Human Genetics researcher Lahiru Handunnetthi and colleagues pointed out that Baranzini’s study focused only on CD4+ lymphocytes, which are certainly not the only cell type that is involved with MS pathogenesis (6). They also wrote that reduced representation bisulfite sequencing provides poor coverage of the genome compared with other methylation analysis methods. Furthermore, they noted that methylation analysis alone does not cover the full spectrum of epigenetic marks; Baranzini did not analyze the complete epigenome of these twins.

Taming epigenetics

IHEC organizers believe that a growing pile of evidence is revealing that each epigenetic paradigm—each combination of DNA, sample history, and environmental factors—has a unique epigenome, making the mapping of a large group of epigenomes extremely valuable for research in disease, developmental biology, and basic biology. What will be most important is diversity, studying a broad range of cell types from a broad range of individuals at different stages of healthy and disease development, because—unlike the genome—the epigenome is constantly changing. “For example, comparing aging cells to newly differentiated cells will be extremely interesting,” says Martienssen.

Epigenetics is the Wild West of biomedical research; the standards vary widely from lab to lab, says Martienssen, and the vocabulary must be defined. An epigenome is not just a methylation map, but also includes histone maps, expression patterns of noncoding and small RNA, and chromatin accessibility maps. Each epigenetic mark requires a different detection approach, using techniques that are still being developed and improved. These methods need to be evaluated in terms of their resolution, comprehensiveness, accuracy, sensitivity, and cost. Data formats need to be developed to enable researchers to integrate their datasets with those from other labs.

One of the legacies of the Human Genome Project is its impact on sequencing technology, quality standards for sequencing, and data dissemination. Although these were the means to their end goal—a complete human genome reference assembly—it is these technologies that enable the possibility of clinical sequencing within the next decades. In December 2010, the IHEC steering committee released a draft of recommendations for epigenomics analysis. The draft provides a definition of the epigenome and the current best practices for high-resolution mapping of epigenetic marks. As sequencing technology continues to evolve, these epigenetic mapping methods will also improve.

“Five years ago, I would have told you that technology was the biggest challenge,” says Martienssen. “But right now mapping technology is so advanced that getting cells in a pure and biologically relevant state is the biggest challenge.” Because the epigenome changes from cell to cell, the consortium will reduce duplication and set purity standards for cell type profiling. Instead of just looking at one cell affected by a disease—CD4+ lymphocytes in MS patients, for example—the project will look at as many cell types as possible. Also, consortium members will need to work with a relatively small numbers of pure cells. One of the steering committee’s recommendations is that the homogeneity of these cell populations should be 95% or more.

But other technical issues—such as validation protocols for ChIP-grade antibodies and improved high-throughput 5-hmC detection techniques—remain on the consortium’s to-do list. In the end, the biomedical researchers may not judge the NIH’s Epigenetics Roadmap Initiative and IHEC on whether 1000 epigenomes will provide answers to unsolved mysteries about disease, but rather whether the project will improve epigenetic techniques as well as the accessibility of these tools, techniques, and data.

“Just like the early days of genetics, there are some aspects of epigenetics that are well-founded and accepted, others that are brand new and exciting, and others that may not hold water in the long run,” says Martienssen. “But just like genetics, the proof will be in the pudding.”

References

1. Baranzini, S.E., J. Mudge, J.C. van Velkinburgh, P. Khankhanian, I. Khrebtukova, N.A. Miller, L. Zhang, A.D. Farmer, C.J. Bell, et al. 2010. Genome, epigenome and RNA sequences of monozygotic twins discordant for multiple sclerosis. Nature 464:1351-6.
2. Lister, R., M. Pelizzola, R.H. Dowen, R.D. Hawkins, G. Hon, J. Tonti-Filippini, J.R. Nery, L. Lee, Z. Ye, et al. 2009. Human DNA methylomes at base resolution show widespread epigenomic differences. Nature 462:315-22.
3. Madhani, H.D., N.J. Francis, R.E. Kingston, R.D. Kornberg, D. Moazed, G.J. Narlikar, B. Panning, and K. Struhl. 2008. Epigenomics: a roadmap, but to where? Science 322:43-4.
4. Ptashne1, M., O. Hobert, and E. Davidson. 2010. Questions over the scientific basis of epigenome project. Nature 464: 487.
5. Javierre, B.M., A.F. Fernandez, J. Richter, F. Al-Shahrour, J.I. Martin-Subero, J. Rodriguez-Ubreva, M. Berdasco, M.F. Fraga, T.P. O'Hanlon, et al. 2010. Changes in the pattern of DNA methylation associate with twin discordance in systemic lupus erythematosus. Genome Res. 20:170-9.
6. Handunnetthi, L., A.E. Handel, and S.V. Ramagopalan. 2010. Contribution of genetic, epigenetic and transcriptomic differences to twin discordance in multiple sclerosis. Expert Rev Neurother. 10:1379-81.