A New Twist on Epigenetics How Rediscovered Chemical Tags on DNA and RNA Are Shaking up the Study of Gene Expression

NEWS FEATURE

A new twist on epigenetics How rediscovered chemical tags on DNA and RNA are shaking up the study of gene expression.

BY CASSANDRA WILLYARD

ome big ideas seem to appear out chemical marks that decorate DNA and its of nowhere, but in 2008 Chuan He protein scaffold. These marks act like a chem- NATURE deliberately went looking for one. The ical notation, telling the cell which genes to SUS National Institutes of Health had just express and which to keep silent. As such, the launched grants to support high-risk, high- epigenome helps to explain how cells with impact projects, and He, a chemist at the Uni- identical DNA can develop into the multitude versity of Chicago in Illinois, wanted to apply. of specialized types that make up different tis- But he needed a good pitch. sues. The marks help cells in the heart, for He had been studying a family of proteins example, maintain their identity and not turn that repair damaged DNA, and he began to into neurons or fat cells. Misplaced epigenetic NIK SPENCER/ BY ILLUSTRATION suspect that these enzymes might also act on marks are often found in cancerous cells. RNA. By a stroke of luck, he ran into molecu- When He and Pan began working together, lar biologist Tao Pan, who had been investi- most epigenetic research focused on the tags gating specific chemical marks, called methyl associated with DNA and the histone proteins groups, that are present on RNAs. The pair that it wraps around. But more than 100 dif- worked in the same building at the Univer- ferent types of chemical mark had been iden- sity of Chicago, and began meeting regularly. tified on RNA, and nobody knew what they From those conversations, their big idea did. Some of the enzymes He was studying took shape. could strip off methyl groups, and He and Pan At the time, biologists were getting excited wondered whether one of them might work about the epigenome — the broad array of on RNA. If the marks could be reversed, they

406 | NATURE | VOL 542 | 23 FEBRUARY©2017 M 2017ac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

FEATURE NEWS

might constitute an entirely new histones. To He, it seemed like proof

NATURE way of controlling gene expression. Reading, writing and regulation of an RNA-based system of gene In 2009, they got funding to hunt A lot of the research on epigenetic modi cations has been concentrated regulation. on methyl marks on cytosine bases in DNA. Recent studies have for reversible marks on RNA and brought methylated adenine bases in both DNA and RNA into focus. the proteins that erase them. The identi cation of proteins that write, read or erase these marks EVIDENCE MOUNTS NIK SPENCER/ Nine years later, such research suggests their importance in the regulation of gene expression. He’s group wasn’t the only one 6 has given birth to an ’ome of its DNA thinking about m A. In 2012, two own, the epitranscriptome. He and teams of researchers independently others have shown that a methyl published the first maps of where group attached to adenine, one of m6A appears6,7. The studies revealed the four bases in RNA, has crucial more than 12,000 methylated sites roles in cell differentiation, and may on mRNAs originating from about contribute to cancer, obesity and 7,000 genes. “After years in the dark, more1,2. In 2015, He’s lab and two we were instantly facing a wide other teams uncovered the same 6mA appears on the DNA of vista,” wrote Dan Dominissini, an chemical mark on adenine bases in some organisms. It is found author of one of the studies, in an mRNA clustered near the beginnings 8 DNA (methyl marks had previously of genes and may either essay in Science . been found only on cytosine), sug- boost or silence production The maps showed that the of messenger RNA. 6 gesting that the epigenome may be Writer distribution of m A is not random. even richer than previously imag- protein Its location suggested that the mark ined3. Research has taken off. “I might have a role in alternative splic- think we’re approaching a golden ing of RNA transcripts, a mechanism

age of epigenomics and epitran- m6A is found on that allows cells to produce multiple scriptomics,” says Christopher messenger RNA. The Eraser versions of a protein from a single proteins that read it Mason, a geneticist at Weill Cor- protein Reader gene. carry out various protein nell Medical College in New York functions that regulate Over the past few years, research- City. “We can actually start to see all protein production. ers have identified some of the these modifications that we knew machinery involved in regulating have been there for decades.” these marks. Each requires a writer Nucleus to place it, an eraser to remove it and MARKING THE MESSENGER a reader to interpret it (see “Read- The governing rule of molecular ing, writing and regulation’). As the biology — the central dogma — Cytoplasm identities of these proteins emerged, holds that information flows from scientists have come to understand DNA to messenger RNA to protein. that m6A affects not only RNA splic- Many scientists therefore viewed ing, but also translation and RNA mRNA as little more than a courier, stability. carrying the genetic information One m6A reader, for example, encoded in a cell’s nucleus to the makes mRNA degrade faster by protein factories in the cytoplasm. Shuttle to degradation Enhance translation Inuence alternative shuttling it to decay sites in the That’s one reason why few research- sites in the cell into protein splicing cell. Another m6A reader promotes ers paid much attention to the protein production by shepherding modifications made to mRNA. methylated RNA to the ribosome. They weren’t a secret, though. The mark actually study these modifications,” Pan says. It Whether m6A directs a cell to produce a pro- that pushed He to the forefront of epitran- requires powerful mass spectrometry and high- tein or destroy a transcript depends on the loca- scriptomics was first discovered on mRNA throughput sequencing techniques. tion of the mark and on the reader that binds in 1974 (ref. 4). Fritz Rottman, an organic Two members of He’s lab at the time, to it. But understanding how this selection chemist at Michigan State University in East Ye Fu and Guifang Jia, pushed forward any- works has been a major challenge, says Gideon Lansing, was trying to understand the role of way, focusing on a protein called FTO, part Rechavi, a geneticist at Tel Aviv University in RNA in regulating gene expression when he of the family of methyl-stripping enzymes Israel who was involved in the mapping of m6A. stumbled across a methyl group on adenine. that He’s group had been studying. Fu and Jia What is clear is that m6A has fundamental The modified base is called N6-methyl thought that it might remove methyl groups roles in cell differentiation. Cells that lack the adenosine, a mouthful that’s commonly from RNA, but they struggled to identify its mark get stuck in a stem- or progenitor-like shortened to m6A. target. Fu and his colleagues began to syn- state. That can be lethal: when He and his col- Rottman and his colleagues wrote that RNA thesize snippets of RNA that contained dif- leagues disabled the m6A writer in mice, many methylation could be a way to select certain ferent modifications, to determine whether embryos died in utero. transcripts for translation into protein. “But FTO could remove them. It was slow going. He has a possible explanation for the role of that was all speculation,” says Karen Friderici, Over the course of three years, the team faced m6A. Each time a cell changes from one state an author of the 1974 paper and a geneticist a string of failures, “I almost thought I would to another — such as during differentiation — at Michigan State University. The team didn’t never find the function,” Fu says. the mRNAs in it must change too. This change have a good way to investigate the mark’s true Finally, in 2010 the team decided to in mRNA content, which He calls a transcrip- function. “It was the beginning of molecular test FTO’s activity on m6A — the methy tome switch, requires precision and careful biology. We didn’t have many of the tools that lated adenine. The mark disappeared. timing. He thinks that the methyl marks might are available now,” she says. The team had shown for the first time be a way for cells to synchronize the activity of More than three decades later, He and Pan that RNA methylation was reversible5, thousands of transcripts. found the tools still lacking. “It’s very difficult to just like the marks found on DNA and Self-described ‘RNA geek’ Wendy Gilbert, a

©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. Al23l ri g FEBRUARYhts reserved. 2017 | VOL 542 | NATURE | 407

NEWS FEATURE biologist at the Massachusetts Institute of Tech- but they did detect 6mA. What’s more, the levels points out that researchers haven’t yet shown nology in Cambridge, says that He’s explana- seemed to be higher in the less-fertile genera- that the mark passes from one generation of tion is plausible. “One of the things that I really tions, “raising the possibility that it could indeed cells to its progeny, a hallmark of epigenetic like about Chuan’s presentations over the last be a carrier of this non-genetic information”, he modifications. couple of years is his effort to try to speak to says. The result came as a surprise. Researchers what is the most important aspect of the mark,” had looked for 6mA in multicellular organisms MINING FOR MORE MARKS she says. But she points out that there are other before, but they weren’t able to find it because it As some researchers dive deep to try to under- ways to coordinate the expression of large is present at such low levels. stand the function of m6A and 6mA, others are groups of genes, such as microRNAs, small bits Greer’s lab head, Yang Shi, knew that He had looking for new modifications. Last year, He, of RNA that do not code for proteins and that uncovered 6mA in algae, and asked him for Rechavi and their colleagues reported15 the help to silence genes. “I don’t know that m6A is help. When He heard what Shi had found, he discovery of another methyl mark on adenine the only way that you could do that,” she says. was excited. “We decided we’re going to do this in RNA called N1-methyladenosine (m1A). together,” He says. A couple of months later, This mark also seems to promote translation, THE A’S HAVE IT He met a researcher in China who had found although the underlying mechanism is differ- Although scientists have long known that RNA 6mA in the fruit fly Drosophila. “I almost fell ent from that of 6mA. He says it might also carries a host of modifications that decorate all to the floor,” He says. In April 2015, the three have a role in synchronizing transcripts for the four of its bases, mammalian DNA seemed to papers came out simultaneously in Cell10–12. transcriptome switch. have only a few marks, all on cytosine. The most Andrew Xiao, who studies epigenetics at Then, in January, Jaffrey and his colleagues common modification in mammals, 5-methyl- Yale University in New Haven, Connecticut, reported on yet another kind of modification that occurs near the caps of mRNAs. The researchers found that mRNAs with this mark — called m6Am — are more stable “It could indeed be a carrier of because their caps are harder to remove16. “It’s exciting to people that the landscape of poten- this non-genetic information.” tially regulated messenger RNA modifications that might influence gene expression could be an order of magnitude more complex than we cytosine or 5mC, is so important that it’s often read the articles with interest. Xiao and his thought before,” Gilbert says. referred to as the ‘fifth base’, after A, C, T and G. colleagues had identified 6mA in mammalian Along with these new discoveries also But He wondered whether there might be other cells, but they hadn’t published their results. come scientific squabbles. Jaffrey’s work16 marks hiding in the genome. Bacteria carry the “Literally we thought nobody will take interest suggests that FTO, which He identified as DNA equivalent of m6A — called N6-methyl- in this field,” Xiao says. The Cell articles proved an m6A eraser, actually targets m6Am. And adenine or 6mA. “They use the methylation to him wrong. “We realized we should hurry up.” in October, He’s group reported17 that the distinguish between their own DNA or foreign Within the year, Xiao’s group and another enzyme Xiao flagged as a 6mA eraser on DNA DNA,” says Eric Greer, a biochemist at the Bos- led by John Gurdon of the Gurdon Institute actually does a better job of stripping m1A off ton Children’s Hospital in Massachusetts. But in Cambridge, UK, showed that 6mA can be a particular type of RNA. But such ambigui- researchers struggled to confirm its presence in found at exceedingly low levels in a variety ties are to be expected in a field that’s experi- more complex organisms. of vertebrate species including mouse and encing a scientific gold rush. In 2013, He’s postdoc Fu had found an human13,14. When Xiao’s group looked at the “We are only in the beginning of the story,” intriguing paper from the 1970s, which distribution of the mark in mouse embryonic Rechavi says. And as the techniques improve, showed that algal DNA contains methylated stem cells, the team found the strongest peaks scientists will be able to see these marks more adenine9. “Nobody ever knew the function, on the X chromosome. Here, the mark seemed clearly. The wealth of research possibilities and nobody ever followed up,” Fu says. to be involved in silencing gene expression. makes Mason feel “euphoric”, he says. “It’s Fu and another postdoc, Guan-Zheng Luo, The researchers also identified an enzyme that like the most exciting time to be working in decided to take the investigation further and seems to be a 6mA eraser14. the field.” ■ SEE TECHNOLOGY FEATURE P.503. map the distribution of 6mA in the DNA of the Xiao is still unravelling the function of 6mA. alga Chlamydomonas. They found it in more He says that it seems to be crucial at certain Cassandra Willyard is a science journalist in than 14,000 genes. And the distribution wasn’t developmental stages, acting like a molecular Madison, Wisconsin. random: 6mA clustered around the places switch — barely present one moment, then 1. Ben-Haim, M. S., Moshitch-Moshkovitz, S. & where transcription begins. “We saw some there’s a surge, and then it disappears. Rechavi, G. Cell Res. 25, 3–4 (2015). 2. Lin, S., Choe, J., Du, P., Triboulet, R. & Gregory, R. I. periodic pattern of the peaks. It’s like one peak “His paper was absolutely a bombshell,” Mol. Cell 62, 335–345 (2016). after another,” Fu says. It might be promoting says Samie Jaffrey, a researcher at Weill 3. Luo, G.-Z., Blanco, M. A., Greer, E. L., He, C. & Shi, Y. gene activation, they reasoned. Cornell Medical College. “It really showed Nature Rev. Mol. Cell Biol. 16, 705–710 (2015). Nearly 2,000 kilometres away in Boston, functional roles for 6mA.” Both He and Shi say 4. Desrosiers, R., Friderici, K. & Rottman, F. Proc. Natl 6 6 Acad. Sci. USA 71, 3971–3975 (1974). Greer and his colleagues had found mA in they have also found mA in mammalian cells, 5. Jia, G. et al. Nature Chem. Biol. 7, 885–887 (2011). the genome of a worm, Caenorhabditis elegans. but haven’t yet published their results. 6. Meyer, K. D. et al. Cell 149, 1635–1646 (2012). Greer, a postdoc at the time, had been study- However, the significance of 6mA isn’t yet 7. Dominissini, D. et al. Nature 485, 201–206 (2012). 8. Dominissini, D. Science 346, 1192 (2014). ing epigenetic inheritance using a C. elegans clear, Shi says. He points out that even with 9. Hattman, S., Kenny, C., Berger, L. & Pratt, K. mutant that becomes less fertile with each suc- the latest technology, the modification is only J. Bacteriol. 135, 1156–1157 (1978). cessive generation. He wanted to understand borderline detectable and its precise location 10. Greer, E. L. et al. Cell 161, 868–878 (2015). 6 11. Fu, Y. et al. Cell 161, 879–892 (2015). how this infertility is transmitted from one cannot be mapped. And the pattern of mA will 12. Zhang, G. et al. Cell 161, 893–906 (2015). generation to the next. Caenorhabditis elegans probably vary from tissue to tissue. 13. Koziol, M. J. et al. Nature Struct. Mol. Biol. 23, 24–30 had long been thought to lack methyl marks, There are still big questions to untangle. (2016). 14. Wu, T. P. et al. Nature 532, 329–333 (2016). but Greer decided to double-check using Mamta Tahiliani, a geneticist at New York Uni- 15. Dominissini, D. et al. Nature 530, 441–446 (2016). antibodies that can bind specific methylated versity School of Medicine in New York City 16. Mauer, J. et al. Nature 541, 371–375 (2017). bases. He and his colleagues didn’t find any 5mC, calls the 6mA work “incredibly exciting”, but 17. Liu, F. et al. Cell 167, 816–828.e16 (2016).

408 | NATURE | VOL 542 | 23 FEBRUARY©2017 M 2017ac milla n| PCORRECTEDublishers Li mited, p17art oMARCHf Spri nger N2017ature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

FEATURE NEWS

CORRECTION The News Feature ‘A new twist on epigenetics’ (Nature 542, 406–408; 2017) omitted a reference (M. J. Koziol et al. Nature Struct. Mol. Biol. 23, 24–30; 2016) that demonstrated 6mA in mammalian cells.