[Journal Nature] Nature -.. August.2.2012

THIS WEEK WORLD VIEW The time is right RECHARGE New lithium- ADAPTATION Early peoples EDITORIALS to crack down on research air batteries come with bounced back from misconduct p.7 extended life p.8 climate change p.9 Unfortunate oversight Scientists must remember that however irrelevant their involvement in industry might seem to them, others will see it differently — only full disclosure will avert the taint of scandal. ydraulic fracturing, or ‘fracking’, a technology that revolution- to fracture geological formations and release natural gas or oil. The ized the natural-gas industry, has been surrounded by contro- technology has been in use for decades, and practised properly, the Hversy in recent years. So, when environmental experts at the report suggested, it is safe and poses little risk to the environment. University of Texas at Austin produced a report in February that gave This over-arching conclusion seems reasonable in view of what we the technique a fairly clean bill of health, they received widespread know today, although scientists continue to sift through contradictory news coverage, including in the pages of Nature (see Nature 482, 445; evidence. And Groat’s explanation of his role also sounds plausible 2012). The study was billed as an independent analysis. Yet last week it — but that is all the more reason for him to have openly disclosed his emerged that its lead author is a well-paid board member of an energy ties to the industry. company that is actively involved in fracking. After the link was revealed by the Public Accountability Initiative, The failure to declare this involvement was an unfortunate mistake a non-profit watchdog in Buffalo, New York, university officials to make, not least because the man who made it is a respected senior announced plans to review the study. But even if the review exonerates scientist who headed the US Geological Survey under US presidents the panel and endorses its findings, it is unlikely to remove the taint of Bill Clinton and George W. Bush — and is therefore experienced scandal. Rather than cutting through the confusion on fracking, the enough to understand the role that politics and perception have in report is likely to contribute to it. sensitive issues such as energy development. Yet Charles ‘Chip’ Groat, Experts in many fields bounce between academia, government and associate director of the University of Texas at Austin Energy Institute, industry during their careers. Universities could not exclude people failed to disclose that he holds a significant number of shares in the who have industry connections from their ranks, nor would they want Houston-based Plains Exploration & Production Company, and that to. The same goes for government. There is also nothing inherently he earned more than US$400,000 from the company last year. In a wrong with universities accepting donations from industry to conduct 23 July statement to Bloomberg news, he said that disclosing his posi- studies, as long as the proper protections are put in place. The key is tion on the board “would not have served any meaningful purpose transparency, because that is the basis for trust between institutions relevant to this study”. and the wider public, which is especially important when people are Groat says that his position on the board did not affect the outcome buffeted by confusing, contradictory and inflammatory information. of the study and that he did not interfere with the findings of his col- What the public needs, and what scientists must deliver, is reliable leagues. The study found no evidence of groundwater contamination information that is honest about both its methods and its inevitable from fracking, which pumps fluid into the ground at high pressure biases. What it needs is full disclosure. ■ main public funder of UK physical-sciences research, the Engineering Marching orders and Physical Sciences Research Council (EPSRC), to cut the num- ber of proposals it receives and to prioritize research that addresses national priorities or comes with economic spin-offs (see page 20). Scientists unhappy with policy are right to take Echoing their Canadian counterparts, the scientists argued that the to the streets. changes would endanger blue-skies research in chemistry, physics and mathematics. But unlike Canada’s protests, the UK campaign has yet to win support from the wider scientific community. he mock funeral — an idea so good that scientists had it twice. In part, that is because the campaign targets a single, specific Last month, about 2,000 researchers marched on Parliament funder and so is not seen as relevant to UK science as a whole. Some THill in Ottawa, carrying a coffin that signified, they said, the researchers have dismissed the coffin parade as an overreaction to a “death of evidence”. The scientists were protesting against a series spat between a few disenfranchised scientists and the EPSRC. Others of cuts by Canadian Prime Minister Stephen Harper’s conservative worry that a public protest that exposes disunity in the ranks of science government that they believed threatened basic research and under- at a time of economic chaos could result in cuts to the science budget. mined expert advice in areas such as environmental policy. And in Perhaps, but if it is an isolated spat, then why did people with little May, physical scientists drove a horse-drawn Victorian hearse to the personal stake in the EPSRC’s policies join in the protests? And the British Prime Minister’s residence in Downing Street, London, this calls by dissenters to close ranks — to keep calm and to carry on — time to mark the demise of UK science. ignore the fact that science funding is a political question. To make a The Downing Street stunt was to protest against moves made by the point in a political arena, scientists must stand up and be counted. ■ 2 A U GUST 2012 | V O L 488 | N A TURE | 5 © 2012 Macmillan Publishers Limited. All rights reserved

WORLD VIEW A personal take on events The time is right to confront misconduct After a generation of denial, research leaders are finally treating scientific fraud with the seriousness it deserves, says Colin Macilwain. ne problem with having worked as a journalist for a long time scientists now believe that. They know that misconduct exists and that, is that every story comes with a feeling of déjà vu. You keep unchecked, it can undermine public regard for science and scientists. Othinking: I’ve been here before. So it is refreshing to report Two major studies to be released in the next year reflect this shift in one issue where something has actually changed: the vexed and attitude. Significantly, they have been instigated by leading scientists. perennial problem of research misconduct, which scientific leaders One study, by the InterAcademy Council, is looking at international are finally taking seriously. Talking to several leaders in recent weeks, aspects of misconduct. Sharp disparities in investigative procedures I have found that their mood has hardened — and not before time. — and the lack of any such procedures, or responsible officials, at For too long, scientists’ instinctive defensiveness has produced many institutions outside the United States — are problematic, given general denial that misconduct constitutes a serious problem. that an increasing proportion of research involves collaborators from I arrived in Washington DC to work for Nature in 1993, in the more than one country. aftermath of congressional hearings into allegations of miscon- Robbert Dijkgraaf, co-chairman of the Inter Academy Coun- duct involving a paper by biologists David Baltimore and Thereza cil, is one of the people leading the study. He hopes that, when its Imanishi-Kari at the Massachusetts Institute of Technology in Cam- findings are released this year, governments and research agencies bridge. The researchers were correctly found around the world will use them as a template innocent. But the case led an independent to improve training and enforcement of good commission chaired by reproductive biologist research conduct. Kenneth Ryan to call for a much more rigorous CURRENT SCIENTIFIC The second study, by the US National Acad- approach to the investigation of misconduct. emy of Sciences, will report in 2013. It is likely to Ryan was shot down in flames by scientific LEADERS HAVE THE call for far-reaching changes in how US agencies officials and his recommendations were ignored. define and police misconduct. Since the 2000 They were delivered to the US Department of OPPORTUNITY decree, agencies have regarded only ‘falsifica- Health and Human Services, which kicked them TO TAKE THE tion, fabrication and plagiarism’ as misconduct: upstairs to the White House. The administration the academy may call for this definition to be of then-president Bill Clinton sat on the findings INITIATIVE AND widened in line with an emerging global con- until 2000, when it issued a bland federal miscon- sensus to include most other sorts of unethical duct decree. And that was in the United States — STAMP DOWN behaviour, such as falsely attributed authorship. the world’s dominant scientific power and the one Last December, for example, Canada estab- that had done the most to address misconduct. ON FRAUD. lished a Tri-Agency Framework for the Respon- Countermeasures elsewhere have been even sible Conduct of Research at its main funding feebler. In Germany, for example, no university agencies. The framework oversees publicly and had an integrity officer until 2011, and it is still privately funded research and has a secretariat to difficult for institutions there to sanction proven fraudsters. Some support university misconduct investigations. judges consider academic freedom of expression to be paramount — Britain is also finally taking some faltering steps to address the and say that it would be violated if a university were to request scien- issue. In July, universities adopted a voluntary concordat that obliges tists to retract a paper. them to investigate misconduct allegations. Some research leaders Worldwide, however, research integrity is now very much in the want to leave it at that but others, led by Michael Rawlins, chairman of spotlight. Prominent cases in the United Kingdom, South Korea, the the UK National Institute for Health and Clinical Excellence, demand Netherlands and Canada in recent years have each had a disturbing further action to ensure that cases are properly investigated. and powerful impact in their respective locales. Current scientific leaders have the opportunity to take the initiative Considerable hard data have emerged on the scale of misconduct. and stamp down hard on fraud. Next year’s National Academy study A metastudy (D. Fanelli PLoS ONE 4, e5738; 2009) and a detailed won’t use language as divisive as Ryan’s, but it could usher in a more screening of all images in papers accepted by The Journal of Cell consistent US system to handle misconduct, which could percolate Biology (M. Rossner The Scientist 20 (3), 24; 2006) each suggest that around the globe. The international report will help governments and roughly 1% of published papers are fraudulent. That would be about agencies to pursue miscreants across borders. Together, the studies 20,000 papers worldwide each year. represent a historic opportunity to deal with what is, perhaps, the At the time of the Baltimore case, it was widely NATURE.COM single most potent threat to science’s prestige. ■ argued that research misconduct was insignifi- Discuss this article cantly rare — and irrelevant to the progress of online at: Colin Macilwain writes about science policy from Edinburgh, UK. science, which would self-correct. Few senior go.nature.com/6knitx e-mail: [email protected] 2 A U GUST 2012 | V O L 488 | N A TURE | 7 © 2012 Macmillan Publishers Limited. All rights reserved

Selections from the RESEARCH HIGHLIGHTS scientific literature NEUROSCIENCE responses. Populating the skin with a skin bacterium restored Blind mice can immunity to the germ-free sense light animals. N. PAVITT/JAI/CORBIS Science http://dx.doi. A small molecule restores light org/10.1126/science.1225152 sensitivity to blind mice when (2012) it is injected into their eyes. For a longer story on this research, Richard Kramer at the see go.nature.com/8ahyc3 University of California, Berkeley, and his colleagues ELECTROCHEMISTRY studied a mouse model of retinitis pigmentosa — a Rechargeable form of blindness in which Li–air battery light-sensing rod and cone cells in the retina degenerate. Lithium–air batteries The researchers applied the promise to greatly exceed molecule, AAQ, to retinas the energy-storage capacity isolated from the mice and of conventional lithium-ion found that it triggered retinal batteries and a study shows ganglion cells — most of which that they can retain 95% of are normally light-insensitive their capacity even after — to increase their firing rate 100 recharges. in response to light. Other Current is generated in work has suggested that the GENOMICS lithium–air batteries when molecule functions by blocking lithium ions from the anode potassium ion channels in Hunter-gatherer genes react with oxygen from the the membranes of neurons, air — rather than with a boosting their excitability. In limited volume of oxidizing behavioural tests, blind mice Three African populations that rely mainly on hunting and agent, as in conventional treated with AAQ showed signs gathering possess a trove of previously unrecorded genetic batteries. Peter Bruce of light sensitivity. diversity. and his colleagues at the The use of this and related Sarah Tishkoff at the University of Pennsylvania in University of St Andrews, molecules could restore Philadelphia and her team sequenced the full genomes of five UK, created a lithium–air vision less invasively than individuals from each of three populations: Cameroonian battery using an electrolyte of other proposed methods, the Pygmies, and the Hadza (pictured) and Sandawe people from dimethylsulphoxide, through researchers say. Tanzania. The researchers’ trawl uncovered 13.4 million which the lithium ions flow, Neuron 75, 271–282 (2012) variants — more than 3 million of which have never been and a porous gold cathode seen before. where oxygen is reduced MICROBIOLOGY Genes involved in immunity, metabolism, taste, before it reacts with the smell and reproduction seem to have evolved since the lithium ions. These materials Skin bacteria different populations split — a sign of adaptation to local seem to prevent the side- boost immunity environments. In the Pygmies, recent changes in genes reactions that have quickly involved in the function of the pituitary gland, which secretes degraded the performance of Microbes living in mammalian growth and other hormones, could explain their short stature. previous lithium–air batteries. guts have an important role All the hunter-gatherers sampled showed signatures of gene Science http://dx.doi. in intestinal immunity and it flow from now-extinct human species. This has been seen org/10.1126/science.1223985 seems that those living on the before mainly in non-African populations, supporting the (2012) skin are similarly crucial for idea that breeding between various human species occurred tuning immune responses to regularly. ANIMAL BEHAVIOUR skin pathogens. Cell http://dx.doi.org/10.1016/j.cell.2012.07.009 (2012) Yasmine Belkaid at the US For a longer story on this research, see go.nature.com/ss7rzr Sex is costly National Institute of Allergy for squid and Infectious Diseases in Bethesda, Maryland, and her a subset of immune cells, with a skin parasite, the germ- For a squid, mating can take up colleagues compared mice produced fewer immune- free mice developed a greater to three hours, and the resulting with microbes on their skin stimulating molecules in number of parasites per skin energy losses could put the with germ-free mice raised germ-free animals than in lesion than the controls, and animal at a disadvantage in aseptic conditions. T cells, control mice. When infected also showed impaired T-cell around predators and reduce 8 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

RESEARCH HIGHLIGHTS THIS WEEK M. NORMAN/MUSEUM VICTORIA COMMUNITY papers in science The most viewed CHOICE CLIMATE MODELLING Aerosols keep down monsoon rain Tiny airborne particles called ✪ HIGHLY READ on www.agu.org atmospheric aerosols tend to reduce the week of 23 July summer monsoon rainfall over most of South Asia. Dilip Ganguly and his colleagues at the Pacific Northwest National Laboratory in Richland, Washington, used a simplified atmosphere–ocean model to simulate the effects of foraging opportunities. of two rhesus monkeys, so changes in the levels and composition of atmospheric aerosols Amanda Franklin and her that the neurons fired when — from local and distant sources — on South Asia’s mean colleagues at the University blue light was delivered into monsoon rainfall. Increased local emissions of aerosols such of Melbourne in Victoria, the brain by an optical cable. as black carbon — which absorbs sunlight and produces a Australia, collected wild Stimulating these neurons as warming effect that tends to reduce cloud cover — weakened dumpling squid (Euprymna the monkeys performed an the monsoon rains in most of South Asia. Aerosols from tasmanica; mating male and eye-movement task changed outside Asia also contributed to the overall reduction in female pictured). They tested the latencies of eye movements rainfall. the creatures’ swimming in both animals. Functional Only over northwest India, where aerosol emissions from endurance in a tank with a magnetic resonance imaging local forest and grass fires are thought to be decreasing, did constant current, before and revealed that the stimulation the mean summer monsoon rainfall increase. after the squid mated. Mating induced distinct patterns of J. Geophys. Res. http://dx.doi.org/10.1029/2012JD017508 (2012) halved the time taken for brain activity during the task. males and females to become Curr. Biol. http://dx.doi. exhausted, but both regained org/10.1016/j.cub.2012.07.023 winter-like conditions. Because GENE THERAPY their energy within 30 minutes (2012) the researchers found the ash of copulation. at several archaeological sites Gene fix repairs Knowing this cost could PALAEOANTHROPOLOGY in Europe and North Africa, hearing contribute to a better they were able to link events understanding of the evolution Resilient to in Neanderthal and human Gene therapy has restored of reproductive behaviours, natural disasters evolution with the timing of hearing for up to 18 months in such as promiscuity, in squid, climatic changes. Early modern mice that were born deaf. the researchers say. A study of ancient humans started to displace The animals are missing the Biol. Lett. http://dx.doi. volcanic ash found at Neanderthals from parts of gene that encodes the protein org/10.1098/rsbl.2012.0556 key archaeological Europe before the eruption VGLUT3. Lack of VGLUT3 (2012) sites suggests that and subsequent cooling, renders inner hair cells of Neanderthals (pictured) and their activities appear the ear’s cochlea incapable of NEUROSCIENCE and early modern to have been unaffected sending electrical signals to the BRIDGEMAN ART LIBRARY/GETTY humans were more by these events. brain. Lawrence Lustig at the Light control in resilient to climate Indeed, in parts of University of California, San monkey brains change and natural central and Francisco, and his team used a disasters than is eastern Europe, virus to deliver the Vglut3 gene Using a technique that often assumed. Neanderthals into the cochleas of these mice. makes it possible to control John Lowe seem to have After one week, the researchers the activity of specific at the Royal become extinct detected auditory responses engineered neurons with light, Holloway well before in the creatures’ brains, and neuroscientists have modified University of the eruption within two weeks, the animals behaviour in primates. London in occurred. showed an increased startle Optogenetics can help Egham, UK, and Early response to sound. researchers to figure out the his colleagues modern humans The results could bode well role of individual neurons, but analysed probably placed for humans, the researchers it has previously been used microscopic shards greater pressure suggest, because VGLUT3 to control behaviour only in of volcanic ash from on Neanderthals is also associated with a rare rodents and invertebrates. a major eruption that than did volcanic form of human deafness. Wim Vanduffel at Harvard occurred in Europe eruptions or Neuron 75, 283–293 (2012) Medical School in Boston, some 40,000 years ago. climate change, the Massachusetts, and his The volcano spewed so researchers suggest. NATURE.COM colleagues genetically modified much climate-cooling Proc. Natl Acad. Sci. USA For the latest research published by neurons in the premotor and ash that the event http://dx.doi.org/10.1073/ Nature visit: prefrontal cortex brain regions probably created pnas.1204579109 (2012) www.nature.com/latestresearch 2 A U GUST 2012 | V O L 488 | N A TURE | 9 © 2012 Macmillan Publishers Limited. All rights reserved

SEVEN DAYS The news in brief POLICY Stem-cell ruling The US Food and Drug Administration (FDA) has now been given legal backing T. ALLOFS/MINDEN PICTURES/FLPA for its attempts to regulate a US clinic that offers therapies involving a patient’s processed stem cells. Such treatments can now be classified as drugs, a US District Court in Washington DC ruled on 23 July, in relation to the FDA’s injunction against Regenerative Sciences, a stem-cell clinic in Broomfield, Colorado. The ruling could pave the way for the agency to regulate other stem-cell clinics. See page 14 for more. Data exemption A cross-party group of India curbs tiger tourism politicians has recommended that England’s laws on freedom of information be India’s Supreme Court has placed an interim ban in some national reserves, but that other parks modified to protect universities on tourists visiting central parts of the country’s would hardly be affected, because they already from having to release data 40 or so tiger reserves, to protect the dwindling keep core areas off-limits. The order came in prematurely. The nation’s population of the endangered big cats. The response to a petition from conservationist Ajay universities have complained 24 July ruling — which the court will re-examine Dubey at the non-governmental organization that the Freedom of on 22 August — still allows tourists into fringe Prayatna in Bhopal. According to a 2010 census, Information Act might be used areas of reserves (‘buffer zones’). Park managers India is home to about 1,700 wild tigers — more to force the release of research said that the ruling would devastate tourism than half of the world’s total. findings and data before they are ready for publication. On 26 July, Parliament’s Justice Select and to set up a US$500,000 of Texas at Austin, did not Yoweri Museveni told people Committee agreed, saying that environmental-law scholarship disclose his industry ties when to avoid physical contact. the existing ‘pre-publication in Sangji’s name. But charges the report (see go.nature. According to the Uganda Virus exemption’ section of the remain against Sangji’s com/sopiwm) was released in Research Institute in Entebbe, act should be amended. See supervisor, the organic February. The university says the outbreak involves the go.nature.com/cf1fia for more. chemist Patrick Harran; his that it is reviewing the study. Sudan subtype of the virus, case has been postponed until See page 5 for more. which in a 2000–01 Ugandan Lab-death charges 5 September. See go.nature. outbreak killed 224 people — A landmark criminal com/hmoden for more. EVENTS 53% of identified cases. prosecution over an accident in a US academic laboratory Industry ties Ebola outbreak Greenland melt reached a partial conclusion Doubt has been cast on a The first widespread outbreak Satellite observations revealed on 27 July. In a deal that saw supposedly independent study of Ebola haemorrhagic fever massive surface melting across criminal charges dropped, into the risks of fracking (the since 2009 has killed 14 people the Greenland ice sheet last the regents of the University pumping of high-pressure in the Kibaale district of month as a dome of unusually of California accepted fluids into shale to force out western Uganda, the World hot air settled over the region, responsibility for laboratory natural gas) after its lead author Health Organization said on NASA scientists announced on conditions three-and-a-half confirmed last week that he is 29 July. Twenty cases have been 24 July. Between 8 and 12 July, years ago, when 23-year-old on the board of directors for reported since the beginning the area subject to melting Sheharbano Sangji died in a an energy company actively of July, but the presence of increased from 40% to 97% lab fire at the University of involved in the practice, a ebolavirus was not officially of the ice sheet — an extent California, Los Angeles. The position that earned him more confirmed until last week. After unprecedented in three decades regents also agreed to put in than US$400,000 last year. the virus spread to the capital, of space observations. The place stringent safety measures Charles Groat, of the University Kampala, Ugandan President previous record was 55%. But 1 0 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

SEVEN DAYS THIS WEEK POLARIS/EYEVINE the event falls within the realm Atomic Energy Commission, COMING UP accusing him of being too of natural variability: ice-core close to the nuclear industry records suggest that extreme melting occurs roughly once and playing down the health every 150 years, with the most the Fukushima Daiichi NASA tries to land its recent event in 1889. risks from last year’s disaster at 6 AUGUST nuclear plant. Curiosity rover on Mars. RESEARCH See page 16 for more. Museum head mars.jpl.nasa.gov/msl Warming redux The Smithsonian Institution’s The Berkeley Earth Surface National Museum of Natural 510 AUGUST Temperature (BEST) study the authors recommend that History in Washington DC has The Ecological Society released the second part the roughly US$300-million announced its next director: of America meets in of its independent analysis budget for the US Antarctic Kirk Johnson, a geologist Portland, Oregon, to of the global land-surface- Program (USAP) be increased who specializes in plant discuss preserving, temperature record on by 6%, and that the programme fossils from the Cretaceous utilizing and sustaining 29 July. The findings — that divert 6% of its planned science period. Johnson currently Earth’s ecosystems. the planet has warmed over spending to infrastructure. The serves as vice-president of www.esa.org/portland the past 250 years owing to USAP devotes nine times more research and collections and human influence — are not person days in Antarctica to as chief curator at the Denver news to climate scientists. But logistics efforts than it does to Museum of Nature & Science in August — had been team leader Richard Muller, a actual research, and after the in Colorado. He will take on requested by the University physicist at the University of upgrade, the balance should the Smithsonian museum’s of Copenhagen, Penkowa’s California, Berkeley, is being tilt more towards research, the US$68-million budget and 126 former employer, in February criticized by researchers for authors add. See go.nature. million artefacts and specimens 2011. Penkowa had already publicizing the BEST results com/dvb9y9 for more. on 29 October. Johnson resigned and been sentenced before they have been peer replaces Cristián Samper, who for embezzling money reviewed. The BEST team has PEOPLE stepped down on 23 January to from the Danish Society for not yet published any of its head the Wildlife Conservation Neuroscience. Two of her findings in journals, despite Nuclear safety Society in New York. papers have been officially posting its first results online Japan’s new nuclear regulatory retracted, and a report from last October. See go.nature. commission will probably be Misconduct charge the Danish Committee on com/euvydr for more. headed by radiation physicist A once high-flying Danish Scientific Dishonesty is also Shunichi Tanaka. On 26 July, neuroscientist, Milena expected later in summer. See FUNDING a parliamentary committee Penkowa, is suspected of go.nature.com/eakrbd for (covering both lower and upper “potentially intentional more. Antarctica upgrade houses) proposed Tanaka as misconduct” involving US research facilities in head of the commission, which 15 research papers, according Physics millions Antarctica, such as the Polar will be launched in September to a leaked report from an A lucrative prize for Star icebreaker (pictured), and will be affiliated with the international committee fundamental physics was need an overhaul, says a report environment ministry. But investigating her case. The launched on 31 July, with nine to the US National Science some observers objected to report — published on 25 July researchers each receiving Foundation, released on Tanaka, a former deputy by Danish newspaper BT but US$3 million. The prize is 23 July. To pay for the upgrade, chair of the cabinet’s Japan due to be released officially sponsored by Yuri Milner, a Russian billionaire who once studied for a physics PhD. Milner chose the first SOURCE: THOMSON REUTERS/ WWW.PWCMONEYTREE.COM US venture capitalists seem to US biotech venture-capital funding ($US billion) 1.4 40 Biotech share of total US venture capital (%) Juan Maldacena, Nathan TREND WATCH BIOTECH VENTURE-CAPITAL DIPS winners: Alan Guth, Andrei Linde, Nima Arkani-Hamed, Investment in biotechnology frms by US venture capitalists dropped sharply in the frst half of 2012. Seiberg, Edward Witten, be avoiding the biotechnology 1.6 sector in favour of information Alexei Kitaev, Maxim 35 Kontsevich and Ashoke Sen, technology, according to who will, in turn, select future numbers released on 20 July in 1.2 30 the US National Venture Capital winners. The prize will be 25 1.0 announced annually and is Association’s ‘MoneyTree’ report. Investments in biotech accompanied by an ad hoc 20 0.8 firms dropped to a 9-year low 0.6 and a $100,000 prize for of around US$700 million in promising junior researchers. the second quarter of 2012. And 10 See go.nature.com/mwaays for they accounted for fewer than 0.4 15 award for ‘exceptional cases’ 10% of all venture-capital deals 0.2 5 more. in that quarter — a much lower 0 0 proportion than usual (see 2003 2005 2007 2009 2011 NATURE.COM chart). For daily news updates see: www.nature.com/news 2 A U GUST 2012 | V O L 488 | N A TURE | 1 1 © 2012 Macmillan Publishers Limited. All rights reserved

NEWS IN FOCUS AGRICULTURE Keeping a SPACE The Mars rover’s PUBLISHING Biologists flirt SOCIAL SCIENCE Can maths cattle plague in its nail-biting descent to with physics-style make sense of the cycles grave p.15 the red planet p.16 prepublication p.19 of history? p.24 G. DRIESSENS For the first time, researchers can trace cell lineage within a growing tumour. In this skin tumour, the cells labelled red all arose from a single stem cell. ONCOLOGY Cancer stem cells tracked The master builders that underlie tumour growth may inform treatment strategies. BY MONYA BAKER going to be a paradigm shift in the way that removes cells from their natural environment chemotherapy efficacy is evaluated and how and may change their behaviour. “You can see ancer researchers can sequence tumour therapeutics are developed”. Instead of test- what a cell can do, but not what cells actually cells’ genomes, scan them for strange ing whether a therapy shrinks a tumour, for do,” says Cédric Blanpain of the Free University Cgene activity, profile their contents for instance, researchers would assess whether it of Brussels, who co-led the skin study . 1 telltale proteins and study their growth in labo- kills the right sorts of cell. All three research groups tried to address this ratory dishes. What they have not been able Underlying this scenario is the compelling knowledge gap by using genetic techniques to to do is track errant cells doing what is more but controversial hypothesis that many tumours track cells. Parada and his co-workers began relevant to patients: forming tumours. Now are fuelled by ‘cancer stem cells’ that produce the by testing whether a genetic marker that three groups studying tumours in mice have other types of cancer cell, just as ordinary stem labels healthy adult neural stem cells but not 1–3 done exactly that . Their results support the cells produce normal tissues. Previous studies their more specialized descendents might ideas that a small subset of cells drives tumour have tested this idea by sorting cells from a can- also label cancer stem cells in glioblastoma, a growth and that curing cancer may require cer biopsy into subsets on the basis of factors type of brain cancer. When they did so, they those cells to be eliminated. such as cell-surface markers, and injecting them found that all tumours contained at least a It is too soon to know whether these results into laboratory mice. In few labelled cells — presumably stem cells. — obtained for tumours of the brain, the gut principle, those cells that NATURE.COM Tumours also contained many unlabelled 2 and the skin — will apply to other cancers, says generate new tumours For a web focus on cells . The unlabelled cells could be killed Luis Parada at the University of Texas South- are the cancer stem cancer metabolism, with standard chemotherapy, but the tumours western Medical Center in Dallas, who led the cells. But sceptics point see: quickly returned. Further experiments showed brain study . But if they do, he says, “there is out that transplantation go.nature.com/hlahwj that the unlabelled cells originated from 2 2 A U GUST 2012 | V O L 488 | N A TURE | 1 3 © 2012 Macmillan Publishers Limited. All rights reserved

NEWS IN FOCUS labelled predecessors. When chemotherapy THERAPEUTICS was paired with a genetic trick to suppress the labelled cells, Parada says, the tumours shrank back into “residual vestiges” that did not resem- FDA’s claims over ble glioblastoma. Meanwhile, Hans Clevers, a stem-cell biolo- gist at the Hubrecht Institute in Utrecht, the Netherlands, and his colleagues focused on the stem cells upheld gut. They had previously shown that a genetic marker that labels healthy gut stem cells also labels stem cells in benign intestinal tumours, Drug watchdog wins right to regulate controversial 4 which are precursors of cancer . In their lat- 3 est study , he and his team engineered mice to therapies. carry a gene for a drug-inducible marker that, when activated, causes labelled cells to make molecules that fluoresce one of four colours. BY DAVID CYRANOSKI state law, rather than to regulation by the FDA. This experiment yielded single-colour tumours The court disagreed on both counts, consisting of several cell types, suggesting that court decision on 23 July could help noting that “the biological characteristics of each tumour arose from a single stem cell. To to tame the largely unregulated field the cells change during the process”, and that check that stem cells continued to fuel the A of adult stem-cell treatments. The US this, together with other factors, means the tumours, Clevers added a second, low dose of District Court in Washington DC affirmed cells are more than “minimally manipulated”. the drug, triggering a few of the stem cells to the right of the Food and Drug Administra- Leigh Turner, a bioethicist at the Univer- change colour. This produced streams of cells tion (FDA) to regulate therapies made from sity of Minnesota in Minneapolis, agrees. “It in the new colour, showing that stem cells were a patient’s own processed stem cells. The case is much too simplistic to think that stem cells consistently producing the other cell types. hinged on whether the court agreed with the are removed from the body and then returned For the skin study, Blanpain and his group FDA that such stem cells are drugs. to the body without a ‘manufacturing process’ labelled individual tumour cells, without target- The judge concurred, upholding an injunc- that includes risk of transmission of com- 1 ing stem cells specifically . They found that cells tion brought by the FDA against Regenera- municable diseases,” he says. “Maintaining showed two distinct patterns of division: they tive Sciences, based in Broomfield, Colorado. the FDA’s role as watchdog and regulatory either produced a handful of cells before peter- Under the treatment sold by the firm, stem authority is imperative.” ing out, or went on to produce many cells. Once cells are isolated from patients’ bone marrow, Centeno says that the FDA injunction again, the results pointed to a distinct subset of processed, and the resulting cells injected applies to only one of his company’s four cells as the engine of tumour growth. What’s back into the patients stem-cell products — one that requires more, as tumours became more aggressive, they to treat joint pain. The “Maintaining 4–6 weeks of processing. The procedure will were more likely to produce new stem cells — FDA calls this pro- the FDA’s role still be available: after the 2010 injunction, the which can divide indefinitely — and less likely cedure the “manu- as watchdog company moved its treatment location to an to produce differentiated cells, which can divide facturing, holding for and regulatory affiliated Cayman Island clinic. only a limited number of times. That could be a sale, and distribution authority is Centeno plans to continue providing the key to halting tumour development early, says of an unapproved bio- other three procedures, also used for joint Blanpain. Rather than eradicating cancer stem logical drug product”, imperative.” pain, in the United States. In those treatments, cells, for example, therapies could try to coax and in August 2010, the cells are reinjected within two days. Cen- them to differentiate into non-dividing cells. ordered Regenerative Sciences to stop offer- teno claims that those cells are “minimally The papers provide clear experimental ing the treatment (see Nature 466, 909; 2010). manipulated”, and that the FDA sees them as evidence that cancer stem cells exist, says During investigations leading up to the the “practice of medicine” and “has no issues” Robert Weinberg, a cancer researcher at the injunction, the FDA also found that, because with them. Indeed, until 25 July, a graphic on Whitehead Institute in Cambridge, Massachu- of flaws in its cell processing, the company the Regenerative Sciences website claimed setts. “They have made a major contribution to was violating regulations on “adulteration” that these three procedures were “FDA validating the concept of cancer stem cells,” he that are meant to ensure patients’ safety. approved”. says. But cancer cells probably also act in more Jeanne Loring, a regenerative-medicine In fact, the FDA has not approved these complex ways than those observed, he warns. scientist at the Scripps Research Institute in procedures, and Centeno did not provide For example, non-stem cells within the tumour La Jolla, California, says that the decision documentation to support his claims that the might de-differentiate into stem cells. will send a warning to other entrepreneurs agency views the three treatments as outside The next step, the three groups say, is figur- offering unapproved stem-cell treatments. its purview. The graphic was removed after ing out how the cells tracked in these experi- “So many people want to start these compa- Nature’s enquiries. ments relate to putative cancer stem cells nies. They say, ‘FDA? What FDA?’.” Doug Sipp, a stem-cell ethics and regula- identified by years of transplantation studies. Chris Centeno, the medical director of tion expert at the RIKEN Centre for Devel- Researchers are already busy hunting for ways Regenerative Sciences and one of two major- opmental Biology in Kobe, Japan, worries to kill these cells; now they have more tools to ity shareholders, told Nature that he plans to that more stem-cell companies might now tell whether such a strategy will work. ■ appeal against the ruling. During the case, set up shop outside the United States to the company claimed that the cells in its avoid regulation, as Regenerative Sciences 1. Driessens, G., Beck, B., Caauwe, A., Simons, B. D. ‘Regenexx’ procedure are not significantly has done. “Other US stem-cell outfits have & Blanpain, C. Nature http://dx.doi.org/10.1038/ nature11344 (2012). modified before they are reinjected, so the close ties with partner clinics in Mexico and 2. Chen, J. et al. Nature http://dx.doi.org/10.1038/ procedure should be considered routine other neighbouring countries, which are tra- nature11287 (2012). medical practice. The company also argued ditionally regulatory havens for other forms 3. Schepers, A. G. Science http://dx.doi.org/10.1126/ science.1224676 (2012). that because all the processing work is done in of fringe medicine as well. I suppose it will be 4. Barker, N. et al. Nature 457, 608–611 (2009). Colorado, the procedure should be subject to business as usual in such places,” Sipp says. ■ 1 4 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

IN FOCUS NEWS FAO/F. PALADINI with ministries of agriculture and veterinary services worldwide, and wrote “to virtually everyone they could think of”, says Ulaeto. By last week, the FAO and OIE had identified some 40 labs. “They were a bit surprised at how many laboratories did have virus,” he says. The list remains confidential, but it includes labs from some 20 countries, thought to be mainly in Africa, the Middle East and Asia, where rinderpest outbreaks were common until recently, and a handful of established rin- derpest research centres, such as the Institute for Animal Health in Pirbright, UK, and the Plum Island Animal Disease Center in New York state. One worrying aspect was that some virus samples were found to be held in facilities that had inadequate biosafety levels. Fears of an accidental release are grounded in experience. After smallpox was eradicated, a lab accident in Birmingham, UK, resulted in two infections and one death. And an acci- dental release of foot-and-mouth virus from the Pirbright facility, which houses a high- biosecurity, world-reference laboratory for Mouth lesions are a sign of rinderpest, which has long decimated cattle throughout the world. both foot-and-mouth and rinderpest, caused an outbreak in the United Kingdom in 2007. INFECTIOUS DISEASE Active research on rinderpest has waned as the disease has been brought under control Officials act to secure over the past few decades, says Michael Baron, a rinderpest researcher at the Pirbright cen- tre. He and others say that the biggest threat is cattle-plague virus from long-forgotten samples of virus from past research programmes, and serum and other samples collected for diagnostic or other pur- poses, that may be lurking in lab freezers. Rin- Risk of accidental reintroduction shadows rinderpest derpest vaccine strains, which are stocked in many countries and consist of live attenuated eradication effort. virus, are also a concern. In theory, they could revert to wild type and cause disease outbreaks. Until the world is certain that rinderpest is BY DECLAN BUTLER World Organisation for Animal Health (OIE). gone for good, vaccine strains will probably This October, the JAC will probably issue need to be maintained in high-security labs inderpest, a devastating cattle disease, the first of a series of guidelines for an inter- in several regions so that they can be shipped has not been seen in the wild for a national oversight system. With the help of swiftly to any outbreak, says Baron. But he says R decade, but it lives on in scores of labs. ad hoc expert groups, the JAC would approve that just a couple of pure-research labs would Twelve months after the world celebrated the official repositories of the virus and ensure that be enough to pursue the valuable scientific success of a years-long vaccination campaign they meet tough biosafety standards. The com- opportunities that rinderpest still offers. that made rinderpest only the second disease mittee would also approve all future research Although the virus is closely similar to the after smallpox to be eradicated, animal-health on live rinderpest virus to ensure that its human measles virus, for example, cattle don’t authorities are turning to the next task: making benefits outweigh the risks. catch measles and humans don’t catch rin- sure that a lab release — accidental or inten- The FAO and OIE don’t have the authority derpest. Understanding why this is so could tional — doesn’t lead to a resurgence. to impose such measures on member states, provide insight into the pathology and basic Rinderpest is as deadly to cattle as highly but last year, countries gave the organiza- biology of viruses, Baron says. Of more imme- pathogenic H5N1 avian flu is to chickens. tions a mandate by endorsing a moratorium diate interest, investigators would also like to In past decades, outbreaks ripped through on research and declaring that the remaining know whether vaccines can be developed herds and wiped out up to 90% of animals, virus samples should be destroyed or shipped against another related virus, the sheep and often leaving famine, and sometimes war, in to approved high-security labs. The approach is goat disease called peste des petits ruminants, their wake. “Its eradication is a huge, huge, modelled on the post-eradication phase of the that might also protect against rinderpest. That achievement that has happened largely under smallpox campaign, which saw the number of would eliminate the need to keep any stocks of the radar of most of the virology and scientific labs holding the virus reduced from 76 in 1976 live attenuated rinderpest virus at all. community,” says David Ulaeto, a member of a to just 2 in 1984. Baron’s home lab contains more than seven-person multidisciplinary Joint Advisory To identify labs that NATURE.COM 100 different rinderpest virus isolates, which Committee (JAC) on rinderpest that was set up might still hold rinder- To follow the last he says represent “basically the history of the to consolidate the eradication by the Rome- pest virus, the FAO days of rinderpest, disease”. He intends to sequence them all in the based Food and Agriculture Organization of carried out extensive lit- visit: next few years — so that they can be recreated the United Nations (FAO) and the Paris-based erature searches, liaised go.nature.com/1kvnni if ever needed — and then destroy them. ■ 2 A U GUST 2012 | V O L 488 | N A TURE | 1 5 © 2012 Macmillan Publishers Limited. All rights reserved

NEWS London IN FOCUS Pathfnder (Sojourner) 1997 Thanks to the guided-entry system, the elliptical area in which Curiosity is projected to set down is orders of magnitude smaller than those for previous Mars landers. 20 m 0.75 m s –1 Previous landings have used airbags to cushion the fnal part of the fall. But Curiosity is too heavy, so engineers arranged to use the rover’s wheel-suspension system as landing gear. A bridle made of three nylon cords and an ‘umbilical cord’ for data unspools, dropping the rover 7.5 metres beneath the descent stage, which is still descending under the power of four retrorockets. Once the rover stops moving, the bridle cords are severed and the descent stage flies away to land at least 150 metres from the rover. The fnal touchdown is the second-riskiest part of the landing, after the parachute deployment. The 0.7% risk is divided between different terrain hazards. Rocks and slopes could in rare cases flip the rover over, or Curiosity could land in a crater too deep to escape or on a mesa too MORE ONLINE NATURE.COM/CURIOSITY Spirit/Opportunity 2004 Phoenix 2008 Curiosity 20 km A PRECISE AIM steep to descend. 2 A U GUST 2012 | V O L 488 | N A TURE | 1 7 Viking 1976 T–13 s SKY CRANE T–0:00 TOUCHDOWN 1.6 km 80 m s –1 The parachute is jettisoned, along with the back shell, and eight retrorockets — similar to those used by Viking — begin to fre. They frst jerk the spacecraft sideways by 300 metres to get it out of the way of the parachute. “If we didn’t do that we’d get hit in the back of the head,” says Steve Sell, deputy operations lead for entry, descent and landing at the JPL. Flyaway T–53 s BACK-SHELL SEPARATION T–2 min 4 s HEAT-SHIELD SEPARATION 8 km 125 m s –1 After the heat shield is expelled, six radar antennas turn on. They have a feld of view of 3 degrees each, and provide the frst independent check on the internal estimate of the spacecraft’s altitude. “We fnally get eyes, and it’s absolutely vital,” says Steven Lee, Curiosity’s guidance, navigation and control-systems manager. To test the system, engineers strapped the antennas to an FA-18 fghter jet and flew it at the ground. SOURCE: NASA/JPL-CALTECH; ILLUSTRATION: J. KRZYSZTOFIAK 125 km 5,900 m s –1 Just before entering the atmosphere, the spacecraft sheds two 75-kilogram tungsten weights, shifting its centre of mass and creating a crude ability to generate lift. “We’re flying a brick,” says Allen Chen, the JPL’s operations lead for entry, descent and landing. The ability to change the lift vector lets the spacecraft compensate for unexpected fluctuations in atmospheric density, reducing the size of the potential landing area and making it possible to land in Gale Crater. All the manoeuvres have to be done by dead reckoning, using the spacecraft’s internal gyroscopes. 11 km 405 m s –1 The parachute, nearly 16 metres in diameter, is expelled by a mortar blast. Of the 1.7% overall risk of failure, about 1% is due to potential problems with the parachute, such as tangled lines or turbulent oscillations. The parachute design strays little from that used in the successful Viking landings, but there have been only limited tests at the speeds and densities that the parachute will experience in the Martian atmosphere — just a few drops T–6 min 43 s GUIDED ENTRY s s T–2 min 28 s PARACHUTE DEPLOYMENT Sojourner (1997) T–5 min 28 s 8 PEAK HEATING Lightweight carbon tiles, similar to those on NASA’s 1999 Stardust comet-sample- return mission, protect the spacecraft from steel-melting temperatures of up to 2,100 ˚C. T–5 min 18 s PEAK DECELERATION All parts of the spacecraft had to be designed to withstand accelerations 15 times as strong as Earth’s gravity. Time to touchdown Altitude Velocity ROVER EVOLUTION The size of a small car at almost one tonne, Curiosity is 5 times heavier than the Spirit and Opportunity rovers and 56 times heavier than Sojourner. Unlike its solar-panelled predecessors, it is powered by a radioisotope thermoelectric generator. Spirit/ Curiosity Opportunity (2012) (2004) 7 MINUTES OF TERROR The Curiosity rover prepares to plunge down to Mars. fter an eight-month journey to Mars, success for NASA’s Curiosity rover will hinge on a few crucial moments. The larg- est and most complicated piece of machinery ever sent to the red planet, Curi- osity will begin its seven-minute fall through the wispy atmosphere at 05:24 utc on 6 August. On Earth, mission scientists will be un able to do anything but wait and hope for the signal that the six-wheeled remote laboratory is rest- ing safely in the feeble Martian sunlight. If Curiosity lands successfully in Gale Crater, it will eventually trundle over to a 5.5-kilo- metre-tall stack of layered deposits ringed by water-altered minerals. Ascending the mound, the rover will chart hundreds of millions of years of geology and help researchers to deduce whether life could ever have existed on Mars. But first it has to arrive. On its way down, the spacecraft will fire 76 charges, adopt 6 configu- rations and slow from 6 kilometres per second to a standstill. It will be the first craft since the PLANETARY SCIENCE BY ERIC HAND A

NEWS IN FOCUS NUTRITIONAL SUPPLEMENTS Lawsuit challenges anti-ageing claims Former executive sues manufacturer of pill meant to rejuvenate cells. BY BRENDAN BORRELL protects chromosomes. But, she adds, “we Patton alleges in the affidavit, Egan threw his would need to test it rigorously”. keys at his boss and demanded to settle things ou’re going to be hearing from my The active ingredient of TA-65 was isolated “man to man”. An employee broke up the con- attorney,” Brian Egan told his boss from the herb Astragalus membranaceus and frontation, and Egan stormed out, says Patton.   “Yon his last day of work, nearly a patented by Geron, a biopharmaceutical firm Egan denies Patton’s version of events. year ago. Last week, Egan filed a class-action in Menlo Park, California. Research spon- On 19 September, Egan told a potential lawsuit that accuses Telomerase Activation sored by TA Sciences and other companies TA Sciences partner in Spain that he had Sciences (TA Sciences) in New York of engag- has shown that the compound can lengthen developed cancer while taking TA-65. Patton 3 2 ing in deceptive business practices in promot- telomeres in mice and humans , but Greider and TA Sciences sued Egan for defamation in ing a proprietary herbal extract intended to and others are sceptical of the assay used. March, saying that he had lost the company reverse the effects of ageing. Calvin Harley, president of Telome Health in $2 million in sales. Patton says that TA Sci- The lawsuit threatens to put the science of Menlo Park, spearheaded the studies as chief ences believes that telomeres — repetitive nucleotide sequences scientific officer at Geron. He stands by the “A compound if Egan had cancer, that protect the ends of chromosomes during conclusion that TA-65 is a “weak telomerase that can he had it before he DNA replication — on trial. But Noel Patton, activator”. However, TA Sciences sells the pill lengthen started taking TA-65. president of TA Sciences, denies all of the alle- as a nutritional supplement, or ‘nutraceutical’, telomeres would Egan stands by gations. “We stand by what we say,” he adds. rather than a drug, so the firm’s health claims be excellent. his allegations, and The connection between cellular ageing and have not been evaluated by the US Food and But we would has now launched a telomere length is rooted in solid research. Drug Administration (FDA). need to test it broader attack on the Telomeres become shorter every time a cell rigorously.” company’s science divides, and when they are lost cells can no DIFFICULT RELATIONSHIP with his class-action longer reproduce. The enzyme telomerase In May 2011, Patton hired Egan to help to suit, which he filed on can lengthen telomeres, possibly slowing or expand TA Sciences’ reach in foreign mar- 23 July, in the New York State supreme court, 1 reversing degenerative diseases. In one study , kets. Egan was required to take TA-65 twice along with another man who took TA-65. The mice genetically engineered to lack functional a day, he later wrote in a discrimination com- suit challenges statements on TA Sciences’ telo merase showed brain degeneration and plaint, so “that I could tell customers that I website, including the assertion that TA-65 shrunken testes, but those effects were reversed was also taking the product, and that it was can lengthen short telo meres. when the enzyme was reactivated. safe and effective”. Patton denies that it was Greider doubts that TA-65 caused Egan’s Such findings have sparked a lot of hype obligatory. cancer, but agrees that the science behind it and encouraged a cottage industry of compa- On 14 September, Egan says, he told Pat- is murky. A telomere-lengthening compound nies that assess a person’s ‘biological age’ on ton that he had been diagnosed with prostate would be a boon to patients dying of bone- the basis of their telomere length. But TA Sci- cancer. The next day, according to Egan, Pat- marrow failure and pulmonary fibrosis, she ences has taken the buzz further: it sells a pill ton fired him and said that his prostate cancer says, and firms could be expected to explore called TA-65, which it says can lengthen short could ruin the company. Egan says that when its pharmaceutical potential. “I don’t think a telomeres. The pill brings in an annual revenue he was fired, he was offered a cash settlement company would be selling it on the side as a of US$6 million in the United States alone. to keep quiet about his cancer, but turned it nutraceutical,” she says. ■ “A compound that can lengthen telo meres down. would be excellent,” says Carol Greider, a Patton denies Egan’s version of events. 1. Jaskelioff, M. et al. Nature 469, 102–106 (2011). 2. Bernardes de Jesus, B. et al. Aging Cell 10, molecular biologist at Johns Hopkins Uni- According to an affidavit that Patton filed in 604–621 (2011). versity in Baltimore, Maryland, who shared a the discrimination suit, Egan was fired for 3. Harley, C. B. et al. Rejuvenation Res. 14, 45–56 Nobel prize for her work on how telomerase meagre sales. On being told of his dismissal, (2011). TOP STORY MORE NEWS VIDEO MORE Electronic ● Dormant HIV gets rude awakening Termites R. HANUS go.nature.com/yvssoy sensor explode ONLINE rivals ● Storms may speed ozone loss to defend above the United States go.nature.com/ their sensitivity of human qwmcf4 colony skin go.nature. ● The skin’s secret surveillance go.nature.com/ com/isxuqc system go.nature.com/slhnr2 i2vxrp 1 8 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

IN FOCUS NEWS SOURCE:ARXIV.ORG BIOLOGY OPENS UP 30 Growth in Projected David Reich of Harvard Medical School in Boston, Massachusetts, publishes routinely in Quantitative-biology papers are a small but rising fraction of Nature and the Public Library of Science jour- submissions to the ArXiv preprint database. nals, and co-author Carlos Bustamante, of Number of submissions (thousands) 10 8 6 4 particle physics Annual growth in submissions (%) 20 other felds author Joseph Pickrell, also at Harvard Medical 14 Stanford University School of Medicine in Cali- Astrophysics submissions fornia, is a leader in the field. Reich says that first outstrips 25 12 Theoretical high-energy Quantitative biology School, suggested using arXiv. Reich and the other co-authors saw no good reason not to post 15 the manuscript there. “It could be an example of 10 the younger generation coming in and finding 5 this sort of thing natural,” says Ginsparg. For most life scientists, however, pre-publi- 2 0 0 says statistical geneticist Graham Coop of 1991 1995 1999 2003 2007 2011 –5 2004 2006 2008 2010 2012 cation is still “more of a trickle” than a trend, the University of California, Davis. He and his postdoc, statistician Peter Ralph, posted a PUBLISHING paper on 16 July analysing genetic relatedness among neighbouring European populations, Geneticists eye the and Coop remains bullish about arXiv’s poten- tial. “Biology will soon have to embrace this trend fully: the speed of discussion, comment potential of arXiv and pre-publication review allowed is needed in biology more than most fields,” he says. Yet some population geneticists still feel that those posting to arXiv are sticking their necks Population biologists turn to pre-publication server to gain out. Some biology journals, such as those pub- lished by the Ecological Society of America in wider readership and rapid review of results. Washington DC, for example, expressly pro- hibit pre-publication in citable public archives. And there are the concerns over establishing BY EWEN CALLAWAY papers posted to the section in the past tended who was first to a discovery. to report esoteric models and methods, often But Ginsparg says that pre-publication he preprint server arXiv.org is perhaps from physical scientists dabbling in biology. is more likely to stop scientists from being best known as the preserve of theoretical Rich in mathematics and steeped in the open- scooped. In many physics fields, publication on Tphysicists and astrophysicists. But 2008 data traditions of genomics, population genet- arXiv is what counts for claiming priority, and saw an influx of submissions of unpublished ics would seem the ideal candidate to dip its toes journal reviewers can use the server to check manuscripts, or preprints, by condensed-matter into pre-publication, which brings the advan- that discoveries are correctly attributed. An physicists who wanted to stake claims to the tages of speed and open discussion. “I grew authoring history that accompanies all arXiv fast-moving subject of iron-based superconduc- up in the physics community,” says Richard papers also allows scientists to arbitrate dis- tors called pnictides. Now the life sciences may Neher, a population geneticist at the Max putes over priority. In the 21 years since arXiv be on the cusp of their own ‘pnictide moment’, Planck Institute for Developmental Biology in began, Ginsparg has seen astrophysicists, com- with population geneticists leading the charge. Tübingen, Germany, “and putting things up on puter scientists and others go from sceptics to In the past month, leading research groups arXiv is a natural thing for me to do.” Neher devotees. “Once a community adopts arXiv, it have posted to arXiv high-profile papers on has co-authored more than ten submissions to never seems to relinquish it,” he says. ■ 1 the genetic history of southern Africans and arXiv since 2004. 1. Pickrell, J. K. et al. Preprint at http://arxiv.org/ 2 Europeans . Other prominent population Despite such examples, one of the best- abs/1207.5552 (2012). geneticists have submitted methods-based known life-sciences preprints on arXiv comes 2. Ralph, P. & Coop, G. Preprint at http://arxiv.org/ papers to the server, which is hosted by Cornell from microbiology, not population genetics. abs/1207.3815 (2012). 3. Ginsparg, P. Nature 476, 145–147 (2011). University in Ithaca, New York. The number Posted as a rebuttal to a 2011 Science paper 4. Wolfe-Simon, F. et al. Science 332, 1163–1166 (2011). of biology papers on the server is still small in reporting that a strain of Halomonas bacteria 5. Reaves, M. L. et al. Preprint at http://arxiv.org/ comparison with physical-sciences preprints from a Californian lake could incorporate arse- abs/1201.6643 (2012). 4 (see ‘Biology opens up’), but Paul Ginsparg, a nic into its DNA , the preprint appeared on the 6. Reaves, M. L. et al. Science 337, 470–473 (2012). 5 theoretical physicist at Cornell who founded server in January before its publication in Sci- 6 arXiv in 1991 (ref. 3), welcomes what he hopes ence this month . Yet a population geneticist still CORRECTIONS could be a sea change. played a part — the paper’s co-author, Leonid The graph in the News story ‘Gene data to “It’s wonderful if biologists are belatedly Kruglyak, of Princeton University in New Jersey hit milestone’ (Nature 487, 282–283; 2012) joining the late twentieth century,” he quips. and the Howard Hughes Medical Institute, miscounted data sets in ArrayExpress for “Welcome to the party; better late than never.” works in the field. Kruglyak says he will con- the years 2003–11. The corrected graph Life-sciences papers have existed on arXiv sider arXiv for future studies from his lab. can be seen online at go.nature.com/2wrlpx. almost since its inception, but biologists have Another attention-grabbing submission typically shied away from this approach amid by prominent geneticists, posted on 23 July, The News Feature ‘Beta test’ (Nature 487, fears of getting scooped, or of offending jour- compares genomic variation in 22 African 160–162; 2012) gave the wrong affiliation nals. It was only in 2003 that the site inau- populations to suggest an ancient genetic link for Stefan Schönert. He is at the Technical 1 gurated a section specifically for papers on between people in southern and eastern Africa . University Munich. quantitative biology, or ‘q-bio’ for short. Yet One of the paper’s senior authors, geneticist 2 A U GUST 2012 | V O L 488 | N A TURE | 1 9 © 2012 Macmillan Publishers Limited. All rights reserved

Physical scientists protested against funding reforms with a mock funeral for British science. Duel to the death Physicists, chemists and mathematicians in the United Kingdom are furious about funding reforms that they say threaten blue-skies research. he horse-drawn Victorian BY ANANYO BHATTACHARYA a series of controversial reforms. hearse canters past Lon- Some were intended to cut the SANG TAN/AP Tdon’s Houses of Parliament and round Parlia- overwhelming number of grant proposals that the ment Square before trotting smartly down Whitehall. EPSRC receives, by limiting resubmissions and tem- Straggling behind it comes an eclectic mix of scientists, porarily blocking people who submitted too many ranging from students with body piercings to tweed- unsuccessful applications from sending in more. A jacketed professors. The spectacle makes an odd addi- second set of reforms included requirements that tion to the spring afternoon traffic, bringing to mind grant applicants explain how their research might an elegant, if unusual, state funeral. generate economic or other benefits, as well as a vig- The truth, however, is much stranger. “We’re protest- orous overhaul of the EPSRC’s research portfolio. ing,” explains one PhD student to a puzzled tourist, The changes incensed many physical scientists, “against our research funder.” who protested that the policy to blacklist grant That funder is the Engineering and Physical Sci- applicants was draconian. They complained that ences Research Council (EPSRC), the government the EPSRC’s decision to exert more control over body that holds the biggest public purse for phys- the fields it funds risked sidelining peer review and ics, mathematics and engineering research in the would favour short-term, applied research over curi- United Kingdom. Facing a growing cash squeeze osity-driven, blue-skies work in a way that would be and pressure from the government to demonstrate detrimental to British science. The souring relation- the economic benefits of research, in 2009 the coun- ship between the EPSRC and parts of its constituency cil’s chief executive, David Delpy, embarked on reached a conspicuously public nadir in May, when 2 0 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

FEATURE NEWS disaffected researchers launched the ‘Sci- in any further funding applications for bang for their buck. In 2009, the research ence for the Future’ campaign with the hearse 12 months — a policy that initially hit more councils started to require that funding appli- stunt, which ended by delivering the coffin, than 200 people. cants submit a two-page Pathways to Impact signifying the death of British science, and a The changes outraged physical scientists. statement, summarizing how they intended petition demanding the “immediate reform Within 2 weeks of the blacklisting policy to maximize the societal or economic ben- of the EPSRC’s policies” to the prime PHYSICS OF FUNDING efits of a project — through commer- cialization of results, for example, or minister in Downing Street. In a letter SOURCE: TOP: EPSRC/C. HAYES/UNIV. NOTTINGHAM; BOTTOM: EPSRC United States accused the EPSRC of Grant applications in physical sciences 2,000 34% 27% 30% 32% 22% 27% Submitted researchers describe in a separate to The Daily Telegraph newspaper in through public outreach. In Novem- The UK’s main physical-science funder introduced policies ber 2011, the EPSRC added a ‘national support of the protestors, nine Nobel to cut grant submissions in 2009 (top), partly in response to importance’ criterion, requiring that laureates in the United Kingdom and a declining budget (bottom). statement “the extent to which the “manipulating the process of peer Funded research proposed has the potential, review” and “establishing favouritism 1,500 over 10–50 years, to meet national schemes”. Success rate The battle could be played out else- strategic needs”. 1,000 where. Many government funding Similar requirements are already bodies are facing diminishing budg- routine at some other funding agen- 500 30% cies, such as the NSF, which has asked ets in the wake of the global financial crisis, and increasing pressure from project’s ‘broader impacts’ on science politicians to show that the research 2010–11 and society since 1997. That require- they are funding will contribute to ment has come in for some criticism economic growth. “This is a challenge (see Nature 475, 141; 2011) — but not for research funding agencies across 1,000 0 2004–05 2006–07 2008–09 Predicted applicants to explain their proposed 900 budget the globe,” says Julia Lane, an expert 800 the level of anger that the new request in science policy who formerly worked EPSRC budget (£ millions) adjusted for inflation seemed to inspire in Britain’s physi- at the National Science Foundation 700 Budget review cal scientists. The impact statements, (NSF) in Arlington, Virginia, and is 600 trims funding for they said, showed that the EPSRC is next four years now a senior managing economist at 500 inappropriately favouring short-term the American Institutes for Research 2004 2006 2008 2010 2012 2014 projects that will have economic ben- in Washington DC. “And they’re all efits. “It is changing the fundamen- struggling to provide enough informa- tal ethos of research to make it more tion for policy-makers so that they can keep being announced, more than 1,200 people responsive to the market,” says physicist Philip funding going for basic research.” had signed an online petition demanding that Moriarty of the University of Nottingham, an In Britain, it looks likely that researchers will it be rescinded (see Nature 458, 391; 2009). outspoken critic of the EPSRC reforms. And have to live with this new reality: Delpy and his In May 2009, the EPSRC was forced to water the national-importance criterion, say some, supporters say that the policies are an unavoid- down the policy by delaying its introduc- is simply asking for the impossible. “It’s very able reaction to a slumping budget and that tion to April 2010, and allowing researchers hard to justify the economic importance of there will be no U-turns. According to Delpy’s to apply for one grant during the 12-month work that might not become applicable to real- detractors, however, the EPSRC provides a ‘cooling-off’ period. The EPSRC says that world problems for decades,” says postdoctoral lesson in how not to implement such reforms. only ten researchers are currently blacklisted. mathematician Will Merry at the University of “They shot themselves in the foot by alienat- Clarke is one of those ten. Six years ago he Cambridge, UK. (Merry’s research problem, ing the community that they’re here to serve,” concurrently held three EPSRC grants — more how to solve the motions of three or more says Paul Clarke, a synthetic organic chemist at than any other researcher at the same career interacting bodies, was first set out by Isaac the University of York, and one of the EPSRC’s stage. But in January, after submitting a third Newton in 1687 — suggesting that science most vociferous critics. unsuccessful proposal, he was told that he sometimes has to take the long road.) had been automatically barred from sending Martin Rees, former president of the Royal OPENING ROUND in more. Clarke, who works on the chemi- Society in London and a cosmologist at the The tensions started rising in 2007, shortly cal origins of life and on synthesizing natu- University of Cambridge, says he worries that after Delpy, a physicist by training, left his ral compounds, says that before the EPSRC the new requirements could affect how young post managing the research portfolio at Uni- banned resubmissions, he was able to address researchers apply for their first grant. “It’s versity College London to take command of critiques of rejected proposals and send them going to make them slant their application in the EPSRC and its research budget of some back for re-review — after which they would a way that might not be optimal from the point £800 million (US$1.3 billion). Delpy faced a often be successful. Rather than being a prob- of view of the research,” he says. The council problem: an overwhelming number of grant lem, he views the large number of proposals should focus on making sure that the “bright- applications and a flat budget were starting to as a sign of scientists with many good ideas, est people don’t get discouraged”, adds Rees, push up rejections. In 2008, the success rate and advocates widening the pool of reviewers who says he finds the idea of asking research- for applications dropped from a typical 30% by mandating that those who submit EPSRC ers to write about the potential future national to 26% overall and even lower in some fields grants also review a certain number of propos- importance of their work “absurd”. (see ‘Physics of funding’). In March 2009, the als each year. But the EPSRC had more reforms up its EPSRC announced that from the following The EPSRC was also under growing pres- sleeve. In July 2011, it published the first of month researchers could no longer resub- sure to demonstrate the impact of its invest- three phases in its Shaping Capability strategy, mit grant proposals that had been rejected, ment in research. This grew out of the 2006 which divided the organization’s research port- a policy that has since been implemented at Warry report, Increasing the Economic Impact folio into more than 100 fields, with the aim several other UK research councils. At around of Research Councils, as well as subsequent of maintaining or expanding areas of national the same time, it also barred researchers reports, demanding that all the UK research importance and excellence, such as catalysis with a record of rejections from sending councils show that they are getting the most and energy storage, and shrinking others such 2 A U GUST 2012 | V O L 488 | N A TURE | 2 1 © 2012 Macmillan Publishers Limited. All rights reserved

NEWS FEATURE as mathematical physics and mobile comput- ing. The council revealed that future postdocs SANG TAN/AP and other fellowships would be funded only in areas that were in line with the strategy; at the outset, for example, mathematics postdocs would be funded only in statistics and applied probability. To the critics, that decision was further evidence that the council ranked short-term pay-off above blue-skies research, in this case favouring the needs of the City, London’s large financial sector, which is hungry for statisti- cal expertise. The policy had an immediate impact for Merry, who finished a 12-month EPSRC doctoral-prize fellowship and strug- gled to find a postdoc at home. He is now set to start one at ETH Zurich in September. Merry says that all the maths PhD students he knows at the University of Cambridge are following a similar path. “The most visible outcome of the change in funding is that to the best of my knowledge they’re all going abroad next year,” he says. In May, physical scientists called for research-council reform at the House of Commons in London. Some researchers also have a broader con- cern: that EPSRC administrators are taking the EPSRC states that contributing to the ways of working you are going to create some over the role of working scientists in decid- country’s economic competitiveness is one of degree of upset”. ing how money should be spent and, as a the body’s three main aims. The prominence Britt Holbrook, a philosopher at the Univer- consequence, are funding mediocre research placed on it in recent years is largely the result sity of North Texas in Denton, who special- in arbitrarily selected areas. Instead, they say, of government pressure, Delpy says. “You izes in science policy, says that the EPSRC’s the EPSRC should focus on supporting the must realize that the term ‘economic impact’ biggest mistake lay in banning resubmis- best-quality science in any area of its remit, was something that was imposed on us by sions and blacklisting researchers. “It’s totally as judged by peer review. “Non-scientists are the Treasury.” What’s more, the EPSRC does understandable that did not go down well,” making decisions that impact on the future not expect a precise forecast of what impacts he says. “If you stop people from resubmit- spend of science money — and that is wrong,” a research proposal could have decades down ting, that cuts out a large part of the value of says Tony Barrett, a synthetic organic chem- the line, he says; rather, researchers are encour- peer review.” But Holbrook, who co-authored ist at Imperial College London and one of the aged to describe how any impacts that they can an influential review of the NSF’s broader- organizers of the Science for the Future cam- foresee might be speedily achieved. impacts criterion, says that the critics are paign. “They are not qualified to make those Delpy maintains that peer review is para- fighting a losing battle against the impact decisions.” mount at the agency. The decision to limit agenda. “People are working against the politi- The uproar over impact and national impor- the EPSRC’s maths postdocs to statistics cal and economic realities,” he warns, and tance grew so loud that, in November 2011, was based on an independent international would be better off trying to shape the drive the House of Lords Science and Technol- review of maths commissioned by the coun- for impact, rather than block it. ogy Committee held a session to discuss the cil and published in 2010, which singled out The hearse parked outside Downing Street, policies with Delpy and civil engineer John the field as an area of serious concern. “We however, suggests that this advice is falling Armitt, then the council’s chair. The EPSRC decided we had to do something to get new on deaf ears; the organizers of Science for also agreed to further discussions with the blood into statistics,” he says. “This is not a the Future say that they are planning further scientific community before rolling out the response to government or the call of the City.” protests over the summer. And Delpy’s critics final phase of the Shaping Capability strategy, Mathematicians can turn to the Royal Society have little time for his explanations, saying, for which it did in March 2012. The announce- or the Leverhulme Trust in London for fund- example, that they are sceptical of EPSRC fig- ment, which filled in funding details for just ing, he points out. And although the research ures showing no decline in blue-skies research, over 50 fields, met a frosty, but somewhat less council’s budget is set to decline — by 6% in given that the council defines the term itself. angry, reception. cash terms between 2010–11 and 2014–15 — Clarke prefers to point to figures showing that he says that the proportion of funding going in 2010–11, the EPSRC funded 151 proposals MAKING IMPACT to discovery-led or blue-skies research has in the physical sciences, excluding engineer- Delpy stoutly defends his organization and stayed roughly flat at around 50–60% of the ing, down from more than 500 in 2004–05. its reforms. The policies to cut grant applica- total. “Where’s the evidence the system is bro- Yet there are more than 3,000 scientists in the tions have worked, he says: the success rates ken?” he asks. United Kingdom who are eligible to apply for for applications are back up to around 30–35% Delpy does regret the rift that the reforms EPSRC funding. — “a healthy level of competition”. have opened up between the council and “People would rather not submit than He rejects the criticisms of the impact and researchers. “Of course we could have done submit, get blacklisted and hence be seen as national-importance strategies. The focus on things better,” he concedes, particularly in a failing academic by their department,” says economic impacts is nothing new, he says, communicating what the council was doing Clarke. “There is now a culture of fear in aca- nor does it come from his own experience in and why. “I would have liked to have been able demic departments.” ■ applied bioengineering. (He developed tech- to carry the community with me and have niques for monitoring premature babies.) The every one feel that they were engaged.” At the Ananyo Bhattacharya is Nature’s chief online 1994 royal charter that officially established same time, he says, “if you change any of the editor. 2 2 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

NEWS IN FOCUS CYCLES OF VIOLENCE The motivating issues vary, but episodes of violent political upheaval in the United States are surprisingly regular. ~1920 NUMBER OF 80 VIOLENT EVENTS PER 5 YEARS Racial tensions, unrest among ~1870 workers and anticommunist feeling surged in the aftermath of the First World War. 60 Racial, class and political tensions peaked during and after the Civil War. ~1970 40 The civil-rights movement, social change and opposition to the Vietnam War marked the tumultuous 1960s. 20 RACIALLY MOTIVATED LABOUR OR ECONOMIC POLITICAL 0 VILIGANTE 1780 1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000 HISTORY AS SCIENCE Advocates of ‘cliodynamics’ say that they can use scientific methods to illuminate the past. But historians are not so sure. BY LAURA SPINNEY trends and outbursts of violence in the United States, and has come to the conclusion that ometimes, history really does seem to a new wave of internal strife is already on its SOURCE: REF. 1 1 repeat itself. After the US Civil War, for way . The peak should occur in about 2020, Sexample, a wave of urban violence fuelled he says, and will probably be at least as high as by ethnic and class resentment swept across the one in around 1970. “I hope it won’t be as the country, peaking in about 1870. Internal bad as 1870,” he adds. strife spiked again in around 1920, when race Turchin’s approach — which he calls clio- riots, workers’ strikes and a surge of anti- dynamics after Clio, the ancient Greek muse of Communist feeling led many people to think history — is part of a groundswell of efforts to that revolution was imminent. And in around apply scientific methods to history by identify- 1970, unrest crested once more, with violent stu- ing and modelling the broad social forces that dent demonstrations, political assassinations, Turchin and his colleagues say shape all human riots and terrorism (see ‘Cycles of violence’). societies. It is an attempt to show that “history To Peter Turchin, who studies population is not ‘just one damn thing after another’”, says dynamics at the University of Connecticut in Turchin, paraphrasing a saying often attributed Storrs, the appearance of three peaks of politi- to the late British historian Arnold Toynbee. cal instability at roughly 50-year intervals is not Cliodynamics is viewed with deep scepti- a coincidence. For the past 15 years, Turchin cism by most academic historians, who tend has been taking the mathematical techniques to see history as a complex stew of chance, that once allowed him to track predator–prey individual foibles and one-of-a-kind situa- cycles in forest ecosystems, and applying them tions that no broad-brush ‘science of history’ to human history. He has analysed historical will ever capture. “After a century of grand records on economic activity, demographic theory, from Marxism and social Darwinism to 2 4 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

FEATURE NEWS BETTMANN/CORBIS structuralism and postmodernism, Periods of rioting and upheaval most historians have abandoned have recurred roughly every 50 the belief in general laws,” said Rob- years in US history. ert Darnton, a cultural historian at Harvard University in Cambridge, quality, measured by the ratio of Massachusetts, in a column written expectancies — and wealth ine- such as political instability should 1871 wage. Choosing appropriate prox- in 1999. the largest fortune to the median Most think that phenomena ies can be a challenge, because rel- be understood by constructing evant data are often hard to find. detailed narratives of what actu- No proxy is perfect, the researchers ally happened — always looking concede. But they try to minimize for patterns and regularities, but the problem by choosing at least never forgetting that each out- two proxies for each variable. break emerged from a particular Then, drawing on all the sources time and place. “We’re doing what they can find — historical data- bases, newspaper archives, ethno- can be done, as opposed to aspiring TOPICAL PRESS AGENCY/GETTY who studies early-modern history colleagues plot these proxies over after what can’t,” says Daniel Szechi, graphic studies — Turchin and his time and look for trends, hop- at the University of Manchester, ing to identify historical patterns UK. “We’re just too ignorant” to identify meaningful cycles, he adds. and markers of future events. For general laws, thanks to tools such 1926 example, it seems that indicators But Turchin and his allies con- tend that the time is ripe to revisit cal cooperation unravels when a period of instability or violence as nonlinear mathematics, simu- of corruption increase and politi- lations that can model the inter- is imminent. Such analysis also actions of thousands or millions of allows the researchers to track the individuals at once, and informat- order in which the changes occur, ics technologies for gathering and so that they can tease out useful analysing huge databases of his- correlations that might lead to torical information. And for some cause–effect explanations. N. BOENZI/NEW YORK TIMES CO./GETTY “Historians need to abandon the 1967 When Turchin refined the con- academics, at least, cliodynamics ENDLESS CYCLES can’t come a moment too soon. cept of cliodynamics with two col- habit of thinking that it’s enough to informally point to a sample of leagues — Sergey Nefedov of the Institute of History and Archaeol- cases and to claim that observations ogy in Yekaterinburg, Russia, and generalize,” says Joseph Bulbulia, who studies the evolution of reli- gion at Victoria University of Wel- State University for the Humani- ties in Moscow — the researchers lington in New Zealand. Andrey Korotayev of the Russian found that two trends dominate FROM ECOLOGY TO HISTORY the data on political instability. Turchin conceived cliodynamics The first, which they call the secu- during what he jokingly calls a lar cycle, extends over two to three midlife crisis: it was 1997, he was centuries. It starts with a relatively 40 years old, and he had come to egalitarian society, in which supply feel that all the major ecological questions about population dynamics and demand for labour roughly balance out. In time, the population had been answered. History seemed to be the next frontier — perhaps grows, labour supply outstrips demand, elites form and the living stand- because his father, the Russian computer scientist Valentin Turchin, had ards of the poorest fall. At a certain point, the society becomes top-heavy also wondered about the existence of general laws governing societies. with elites, who start fighting for power. Political instability ensues and (The elder Turchin’s dissident writings about the origins of totalitarian- leads to collapse, and the cycle begins again. ism were among the reasons that the Soviet Union exiled him in 1977, Superimposed on that secular trend, the researchers observe a shorter after which he moved his family to the United States.) cycle that spans 50 years — roughly two generations. Turchin calls this What is new about cliodynamics isn’t the search for patterns, Turchin the fathers-and-sons cycle: the father responds violently to a perceived explains. Historians have done valuable work correlating phenomena social injustice; the son lives with the miserable legacy of the resulting such as political instability with political, economic and demographic conflict and abstains; the third generation begins again. Turchin likens variables. What is different is the scale — Turchin and his colleagues this cycle to a forest fire that ignites and burns out, until a sufficient are systematically collecting historical data that span centuries or even amount of underbrush accumulates and the cycle recommences. millennia — and the mathematical analysis of how the variables interact. These two interacting cycles, he says, fit patterns of instability across In their analysis of long-term social trends, advocates of cliodynam- Europe and Asia from the fifth century bc onwards. Together, they ics focus on four main variables: population numbers, social structure, describe the bumpy transition of the Roman Republic to the Roman state strength and political instability. Each variable is measured in sev- Empire in the first century bc. He sees the same patterns in ancient eral ways. Social structure, for example, relies on factors such as health Egypt, China and Russia, and says that they explain the timing of last inequality — measured using proxies including quantitative data on life year’s Egyptian uprising, which took the regime of then-president 2 A U GUST 2012 | V O L 488 | N A TURE | 2 5 © 2012 Macmillan Publishers Limited. All rights reserved

NEWS FEATURE Hosni Mubarak by surprise. At the time, the Egyptian economy was scepticism among historians in general. The essential weakness of any growing and poverty levels were among the lowest in the developing attempt to make predictions based on trends, says Szechi, is the appall- world, so the regime could reasonably have expected stability. In the ing patchiness of historical information. Records can be preserved or decade leading up to the revolution, however, the country saw a quadru- destroyed by chance: in 1922, for example, fighting in the Four Courts pling of graduates with no prospects — a marker of elite overproduction area of Dublin during the Irish Civil War led to a fire that destroyed the and hence, Turchin argues, trouble. country’s entire medieval archive. More generally, says Szechi, knowl- Turchin has also applied this approach to other historical puzzles, edge tends to pool around narrow subject areas. “We can tell you in great such as how religions grow. Several models have been proposed. One is detail what the grain prices were in a few towns in southern England that they grow in a linear fashion as nonbelievers spontaneously ‘see the in the Middle Ages,” he says. “But we can’t tell you how most ordinary light’. Another model holds that the number of converts increases expo- people lived their lives.” nentially, like infections with a contagious disease, as outsiders come Concerted efforts are now under way to fill those holes. Harvey into contact with growing numbers of converts. Using several independ- Whitehouse, an anthropologist at the University of Oxford, UK, is over- ent proxies, Turchin has mapped conversions to Islam in medieval Iran seeing the construction of a database of information about rituals, social 2 and Spain, and found that the data fit the contagion model most closely . structure and conflict around the globe since records began. It is a huge Using the same techniques, he has also shown that the model describes undertaking, involving historians, archaeologists, religious scholars, the expansion of Christianity in the first century ad, and of Mormonism social scientists and even neuroscientists, and it will take decades to since the Second World War. complete — assuming that funding can be found beyond the Claudio Cioffi-Revilla, a computer social scientist at George Mason UK government’s current 5-year commitment. But White- University in Fairfax, Virginia, welcomes cliodynamics as a natural house believes that complement to his own field: doing simulations using ‘agent-based’ “MOST HISTORIANS HAVE the research that is computer models. Cioffi-Revilla and his team are developing one such feeding the database Valley region in East Africa, a populous area that is in the grip of a ABANDONED THE BELIEF IN will complement model to capture the effects of modern-day climate change on the Rift Turchin’s approach by households and allows them to interact, following rules such as sea- GENERAL LAWS.” throwing light on the drought. The model starts with a series of digital agents representing immediate triggers of sonal migration patterns and ethnic alliances. The researchers have political violence. He 3 already seen labour specialization and vulnerability to drought emerge argues , for example, spontaneously, and they hope eventually to be able to predict flows of that for such violence to happen, individuals must begin to identify refugees and identify potential conflict hotspots. Cioffi-Revilla says that strongly with a political group. One powerful way for groups to cement clio dynamics could strengthen the model by providing the agents with that identification is through rituals, especially frightening, painful or rules extracted from historical data. otherwise emotional ones that create a body of vivid, shared memories. “People form the impression that the most profound insights they GLOBAL TRENDS have into their own personal history are shared by other people,” says Cliodynamics has another ally in Jack Goldstone, director of the Center Whitehouse, who explored this fusion of identities in an as-yet unpub- for Global Policy at George Mason University and a member of the lished survey of revolutionary brigades in Misrata, Libya, last December, Political Instability Task Force, which is funded by the US Central Intel- along with his colleague Brian McQuinn, an anthropologist at Oxford ligence Agency to forecast events outside the United States. Goldstone who studies civil wars. Only once such fusion has occurred do people has searched for clio dynamic patterns in past revolutions, and predicts become willing to fight and die for the group, he says. Therefore, if that Egypt will face a few more years of struggle between radicals and Turchin’s prediction of unrest in the United States around 2020 is cor- moderates and 5–10 years of institution-building before it can regain rect, Whitehouse would expect the next few years to see an increase stability. “It is possible but rare for revolutions to resolve rapidly,” he in tightly knit US groups whose rituals have a threatening quality but says. “Average time to build a new state is around a dozen years, and promise great rewards. many take longer.” Turchin can’t say who those groups might be, what cause they will But Goldstone cautions that cliodynamics is useful only for looking be fighting for or what form the violence will take. Previous bouts of at broad trends. “For some aspects of history, a scientific or clio dynamic turbulence were not dominated by any one issue, he says. But he already approach is suitable, natural and fruitful,” he says. For example, “when sees the warning signs of social strife, including a surplus of graduates we map the frequency versus magnitude of an event — deaths in vari- and increasing inequality. “Inequality is almost always a bad thing for ous battles in a war, casualties in natural disasters, years to rebuild a societies,” he says. state — we find that there is a consistent pattern of higher frequencies That said, Turchin insists that the violence is no more inevitable than at low magnitudes, and lower frequencies at high magnitudes, that fol- an outbreak of measles. Just as an epidemic can be averted by an effective lows a precise mathematical formula.” But when it comes to predicting vaccine, violence can be prevented if society is prepared to learn from unique events such as the Industrial Revolution, or the biography of history — if the US government creates more jobs for graduates, say, or a specific individual such as Benjamin Franklin, he says, the conven- acts decisively to reduce inequality. tional historian’s approach of assembling a narrative based on evidence But perhaps revolution is the best, if not the only, remedy for severe is still best. social stresses. Gintis points out that he is old enough to have taken Herbert Gintis, a retired economist who is still actively researching part in the most recent period of turbulence in the United States, which the evolution of social complexity at the University of Massachusetts helped to secure civil rights for women and black people. Elites have Amherst, also doubts that cliodynamics can predict specific historical been known to give power back to the majority, he says, but only under events. But he thinks that the patterns and causal connections that it duress, to help restore order after a period of turmoil. “I’m not afraid of reveals can teach policy-makers valuable lessons about pitfalls to avoid, uprisings,” he says. “That’s why we are where we are.” ■ and actions that might forestall trouble. He offers the analogy of avia- tion: “You certainly can’t predict when a plane is going to crash, but Laura Spinney is a freelance writer in Lausanne, Switzerland. engineers recover the black box. They study it carefully, they find out 1. Turchin, P. J. Peace Res. 49, 577–591 (2012). why the plane crashed, and that’s why so many fewer planes crash today 2. Turchin, P. Historical Dynamics (Princeton Univ. Press, 2003). than used to.” 3. Swann, W. B. Jr, Jetten, J., Gómez, Á., Whitehouse, H. & Bastian, B. Psychol. Rev. None of these arguments, however, has done much to soften 119, 441–456 (2012). 2 6 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

COMMENT BRAIN Extraordinary tale of C CULTURE Exhibition EVOLUTION A brief look at ECOLOGY Post mortems show Francis Crick teaching Galileo c celebrates African curious human behaviour, bats killed by wind-turbine views of the cosmos p.30 0 consciousness theory p.29 vi from tickling to burping p.31 blades, not air pressure p.32 he Mars Curiosity rover, which all NASA/SPL Ttouch down on the red planet safely space scientists fervently hope will this week, is a prime example of an expen- sive and complicated NASA mission. With a landing scheme involving 76 pyrotechnic devices firing on time and a US$2.5-billion price tag, it is a high-risk endeavour. By contrast, the Mars Atmosphere and Volatile Evolution Mission (MAVEN) is a project being run out of our laboratory in Colorado to explore Mars’s upper atmosphere and ionosphere. It is set to launch in 2013 for about $500 million. It is on budget, on schedule and promises compelling science. Yet the Scout programme, under which such small Mars missions were funded, has recently been axed. The planetary exploration flagship pro- grammes and the vastly over-budget James Webb Space Telescope are symptomatic of a core problem in space research. Increas- ingly, NASA’s focus is on big projects that promise to return tremendous science benefits. But these programmes absorb most of the available funding for space research. They shift resources away from efficient and effective principal investigators (PIs) at universities, an approach in which a single person is responsible to NASA for the success of a mission, and towards bureau- cratic NASA centres. This is the wrong direction for space research, especially in a time of scarce funding. In my opinion, we need to turn civilian space-policy thinking on its head. Missions managed by PIs should be the highest priority for NASA, not the lowest. I am not talking about the ‘faster, better, cheaper’ approach of the 1990s, with skeleton crews of engi- neers at NASA centres. I am talking about SAMPEX, the first of NASA’s inexpensive Small Explorer satellites in orbit, is due back to Earth this year. missions led by university scientists with a real passion for research. This strategy would reduce budgetary overruns, increase the Let academia lead frequency of launches and enhance excite- ment like few other things could. THREE-WAY PARTNERSHIP space science At its beginning, the US civilian space programme was crafted as a three- way partnership between government (NASA), industry and academia. From NASA must put more of its money into thrifty missions the famous 1945 report of engineer and policy adviser Vannevar Bush, Science — led by principal investigators, says Daniel N. Baker. The Endless Frontier, through to the 2 A U GUST 2012 | V O L 488 | N A TURE | 2 7 © 2012 Macmillan Publishers Limited. All rights reserved

COMMENT unflinching commitment of NASA’s second administrator, James Webb, the founding fathers of space research put universities at the NASA/JPL-CALTECH centre of almost all NASA science activities. Since then, university researchers have brought innovation and nimbleness to hard- ware development, have exercised tender loving care of space instruments and have provided a necessary antidote to govern- ment stagnation. Universities have been a fertile training ground for thousands of space engineers and researchers, who have learned to be creative while sticking to budgets and schedules. This has been shown statistically in an analysis of historical data by David Bearden and his colleagues at the Aero- space Corporation in El Segundo, Califor- nia, due to be published in September. The central and indispensable role of universities in space work is now under immense stress as budgets tighten and NASA withdraws to its centres and core NASA’s costly Curiosity rover will hopefully touch down on Mars to begin work this week. industry contractors. Most people who knew and understood the essential nature and to the ever-increasing pace of changes just to give them accounts to charge to. Many of the three-way space partnership are gone. that are occurring in the atmosphere, oceans, centre-led missions are costly because they Those who have replaced them in policy and land surface, cryosphere and ecosystems. focus more on maintaining jobs than on get- leadership roles may not have realized or ting the biggest scientific bang for the buck. absorbed the lessons of the early days. THE RIGHT BALANCE There is ample support in the space- University labs are being driven out of busi- Some things cannot be done in modest research community for a more balanced ness. In the recent ‘Earth Venture’ mission PI mode. A dedicated flagship mission is space programme. Many US National selection, four and a half of the five concepts needed to develop, for instance, the multiple- Research Council (NRC) reports and decadal selected were for missions led by NASA instrument spacecraft necessary for going to surveys have made clear calls for more PI-led centres. (The half comes from a collaboration the challenging environments of the outer missions. And NASA’s Explorer programme between a NASA centre and a university.) planets. The same is true for large-aperture has used PIs to study focused space physics Many space hardware- facility-class astrophysics programmes and astrophysics. Similarly, the Discovery and development groups “University (including the James Webb Space Telescope). New Frontiers programmes in its Solar System that were thriving as labs are being I am not arguing that NASA should gut its exploration division are PI missions to study recently as five years driven out of centres. Before my present university posi- planetary-science issues of moderate scale. But ago are now defunct. business.” tion, I was a laboratory director at NASA’s the general trend is away from such missions. Those that remain Goddard Space Flight Center in Greenbelt, Planning groups at NASA should work are struggling. Students are finding fewer Maryland, from 1987 to 1994. I saw first- with the NRC, the other National Acad- opportunities for experimentation and are hand what immense strengths could be emies and the Office of Management and not being trained to do things cheaply in a mustered with a critical mass of engineering Budget to shift towards the kind of balanced ‘hands-on’ fashion. Space research has fallen and science talent. In the current frenzy to programme that I advocate here. Allocating into a vicious negative-feedback loop. cut federal budgets, there is a real danger of a few hundred million more dollars from For example, all 15 of the high-priority losing vital and unique capabilities at centres NASA’s $5-billion space-research budget missions recommended for initiation by that have taken years to build and hone. to the PI end of the spectrum could work NASA in the 2007 US National Academies’ There have already been staff reductions, wonders, in my view. decadal survey, Earth Science and Applica- such as at the Jet Propulsion Laboratory With government budgets tightening, tions from Space, are being implemented in Pasadena, California, and the planned space research should be revived in universi- by collaborations between centres and budget cuts mean that more losses may be in ties because they are the best places to foster industry. Costs for the first set of missions the offing. We must not allow the navigation, innovative thinking and to get science done have ballooned by factors of two to three. propulsion and communication skills that in an affordable way. They are also where we According to a mid-term assessment of the enable space to be explored to slip through must train the scientists and engineers who survey, in-sourcing work to centres and the our fingers. If we do, we may never again be will bring an aggressive, nimble mindset to use of ‘directed’ missions rather than com- able to traverse the rings of Saturn, nor land a brighter, future NASA. ■ petitive PI-class missions were among the on an enticing asteroid — nor one day plumb reasons for rising costs. the depths of Europa’s oceans. Daniel N. Baker is professor of astrophysical With insufficient funds, missions are But we must not allow centres to be and planetary sciences and director of the being cancelled or delayed. As ageing maintained at exorbitant staffing levels Laboratory for Atmospheric and Space spacecraft begin to fail, the United States is irrespective of cost. Too many institutions Physics at the University of Colorado, Boulder, in grave danger of losing its ability to view employ workers who are performing routine USA. He was chief of the Laboratory for Earth from space. Soon, it will be unable to and often-unnecessary functions. I have Extraterrestrial Physics at NASA’s Goddard provide decision-makers with the informa- recently seen dozens of extra managers and Space Flight Center from 1987 to 1994. tion they need to respond to natural hazards engineers assigned to NASA programmes e-mail: [email protected] 2 8 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

BOOKS & ARTS COMMENT UNIVERSAL HISTORY ARCHIVE/GETTY Galileo Galilei, pictured centre, after a painting by H. J. Detouche, is imagined in Giulio Tononi’s Phi as an explorer of consciousness. NEUROSCIENCE A quest for consciousness Christof Koch marvels at a journey that explains mind–body theory through a fantastical lens. n the end, consciousness is all that mat- Galileo negotiates information theory in the twentieth century, ters. So writes Giulio Tononi, whose some tricky concepts scholars averred a link between information Istunningly original scientific fantasy, on a road long trodden and conscious experience without work- Phi, is a distant echo of that great deduction by neuro scientists and ing out what that might be or could imply. by René Descartes. Tononi, a neuroscien- neurologists seeking Tononi’s integrated-information theory tist, psychiatrist and expert on sleep and to track consciousness does both. Proceeding from two axioms consciousness, is also that rarest of modern down to its lair in the that are rooted in everyday phenomenal scholars — an idealist. In this category-defy- brain. Even if we could experience, the theory defines a measure ing book, he presents his quantitative theory point to this biophysi- (the eponymous Φ) that is associated with of how brain produces mind as a voyage of cal mechanism, and Phi: A Voyage from every system that consists of causally inter- discovery imagined for Galileo Galilei. those nerve cells, as the Brain to the acting parts. This measure is high if a system In Tononi’s literary telling of this story, mediators of the phe- Soul constitutes a single entity above and beyond GIULIO TONONI Francis Crick teaches Galileo basic neuro- nomenal experience Pantheon: 2012. 384 its parts (integration) and if it is endowed science. Galileo learns that the brain is the of red, we would still pp. £19.99, $30 with a large repertoire of discriminable states seat of the mind, and that consciousness flees need to ask — why (information). The more integrated infor- when neurons turn on and off together dur- these particular mechanisms and neurons? mation any system has, the more conscious ing deep sleep or seizures, as the pair meet Why not others? Historically, the great chal- it is. This framework, couched in a proba- scholars, scientists, doctors and artists from lenge has been to explain how conscious- bilistic language, also captures the unique the Enlightenment to the modern era. The ness emerges from highly organized matter intrinsic quality of experience — why blue, vast cast includes Descartes, Nicolaus Coper- without invoking magic, soul-stuff or exotic for example, is more similar to red than to nicus, Charles Darwin, Sigmund Freud, Mar- physics. pain or smell. cel Proust and, eventually, Alan Turing. With the advent of Claude Shannon’s In Phi, this is conveyed through a 2 A U GUST 2012 | V O L 488 | N A TURE | 2 9 © 2012 Macmillan Publishers Limited. All rights reserved

COMMENT BOOKS & ARTS series of dazzling thought experi- ments aided by cameos from Shannon and philosophers Spinoza, Leibniz and Thomas Nagel (the only living person to figure in the book). Through them, Galileo understands how the algebra of integrated information is turned into the geometry of conscious experiences, and how this links to the physiology and the anatomy of the brain. In the book’s final third, Tononi lays out the implications of his theory. He discusses a number of points about con- sciousness: that it ceases in death and dementia, does not require language or knowledge of self, exists in animals in graded forms and can be present, to some degree, in the fetus. Hell, Tononi emphasizes, is all in the mind. One of the most chilling charac- ters in Phi is the Master, an amalgam of the captain in Franz Kafka’s 1914 short story In the Penal Colony and the “This is a story for grown Grand Inquisitor Gavin Jantjes’ untitled painting depicts a Khoisan girl creating the Milky Way. from Fyodor Dos- men, not a toyevsky’s novel consoling tale The Brothers Kara- ASTRONOMY for children.” mazov (1880). The Master’s obsession Under African skies is creating perfect never-ending pain by manipulating the brain’s informational content. In the final chapter, the Man- Ivan Semeniuk follows the gaze of artists from cultures nequin, a stand-in for Mephistopheles, throws up some logical paradoxes before that have interpreted the heavens for millennia. leaving the dying Galileo reunited with his beloved daughter. Phi is extraordinary. In its appeal to the azing up at a sky full of stars is one of The country was selected this year, along imagination, it bears some resemblance the most universal of human expe- with Australia, to host the Square Kilome- to Edwin Abbott’s Flatland novella (1884) Griences, cutting across cultures and, tre Array, which will be the world’s largest or Douglas Hofstadter’s Gödel, Escher, one imagines, stretching back to the dawn of radio telescope; that association adds to the PHOTOGRAPH: F. KHOURY Bach (Basic Books, 1979). Yet its language humanity. Yet artistic depictions of the heav- sense of interplay between the scientific and is more poetic, and full of cultural refer- ens in popular culture are predominantly the spiritual that weaves its way through the ences and images — film stills and often European — from Johann Bayer’s engrav- exhibition. The show seamlessly bridges modified coloured photos of artworks. ings of the constellations in his the centuries, uniting pieces as Endnotes to each chapter link the alle- 1603 star atlas Uranometria diverse as traditional moon gories and metaphors in the text to the to the swirling brilliance of masks from Côte d’Ivoire science. Van Gogh’s 1889 painting and Trembling Field, an I believe that in the fullness of time, the The Starry Night. interactive sculpture by quantitative framework outlined in Phi An exhibition at the South African Karel Nel. Nel will prove to be correct. Consciousness is US National Museum of is resident artist with the tightly linked to complexity and to infor- African Art, part of the Cosmic Evolution Survey, mation, with profound consequences for Smithsonian Institution a project that focuses on understanding our place in the evolving in Washington DC, may a two-square-degree field Universe. As Crick says to Galileo, this is help to change that. It of the sky to see how the a “story for grown men, not a consoling showcases a range of con- Universe has changed over tale for children”. ■ temporary and historical time. pieces by African artists. “Africa has a long and Christof Koch is chief scientific officer All are connected in one rich history of keen observa- at the Allen Institute for Brain Science way or another to the Sun, tion of the heavens,” says the in Seattle, Washington, and professor of Moon or stars. exhibition’s curator, Christine biology and engineering at the California African Cosmos: Stel- Mullen Kreamer. “Works METROPOLITAN MUSEUM OF ART Institute of Technology in Pasadena, lar Arts was sponsored in California. large part by the govern- Figures from Central Africa e-mail: [email protected] ment of South Africa. bear lunar patterns. 3 0 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

BOOKS & ARTS COMMENT Books in brief Curious Behavior: Yawning, Laughing, Hiccupping, and Beyond Robert R. Provine HARVARD UNIVERSITY PRESS 246 pp. £18.95 (2012) How can farting, sneezing and other marginal biological realities illuminate humanness? Neuroscientist Robert Provine turns an evolutionary lens on everything from the gross to the faintly improper. The ‘contagiousness’ of yawning, for instance, hints at the roots of empathy and herd behaviour. Burping and farting were involved in the development of speech, says Provine. And tickling may play a part in our early understanding that we are distinct beings (you can’t tickle yourself). An exercise in ‘small science’ — some of it speculative, all of it fascinating. The Guardian of All Things: The Epic Story of Human Memory Michael S. Malone ST MARTIN’S PRESS 304 pp. £18.99 (2012) Memory is a kind of relay, with each generation passing its torch on to the next — creating a conduit for thought and civilization through the eons. In his evocative book, technology writer Michael Malone traces that history from the brain’s evolution and the development of speech and writing to advances in recording, the rise of technology and the shifts in ownership of memory from the tribal elect to the masses. The book is packed with gems, including a passage on the twelfth century, when Greek and Arabic science infused Europe, filling its libraries and helping to seed its universities. of art can allows us African Cosmos: access to that history, Stellar Arts and that knowledge.” National Museum of Dreamland: Adventures in the Strange Science of Sleep African Art, Washington The journey begins DC. David K. Randall NORTON 304 pp. £17.99 (2012) on territory that is Sleep occupies us for one-third of our lives. So insomnia, nightmares, Until 9 December both ancient and 2012. deprivation and other aspects of bad sleeping are an obsession familiar, with a series for thousands. The tipping point for journalist David Randall was of pieces from pharaonic Egypt. Representa- sleepwalking into a wall. Astonished by a specialist’s admission of tions of cosmic deities and celestial objects ignorance about the condition, Randall set out to uncover research such as the bright star Sirius are reminders and shine a light into some dark corners. The entertaining result of the night sky’s prominent role in the ritu- covers plenty of territory, from the medieval habit of dividing nightly als and beliefs of civilizations along the Nile. sleeps to the link between vacuum cleaners and sleep apnea. Along The exhibition goes on to leap across the the way, Randall picks up the basics for crafting a healthy snooze. Sahara and forward in time. Far more exotic to non-African eyes are items that date back only a century or so: a bowl and lid from Air: The Restless Shaper of the World Nigeria representing the domains of Earth William Bryant Logan NORTON 416 pp. £13.99 (2012) and sky, or a Dogon stool from Mali, which Arboriculturist William Bryant Logan follows Oak (Norton, 2005) depicts human ancestral figures descend- and Dirt (Norton, 2007) with this splendid exploration of the ing from the heavens to populate the land “floating world” of air — our planet’s invisible skin. Starting with below. the tornadoes that hit New York in 2010, he both warns of and One of the more striking of the con- celebrates the often turbulent and dangerous action of atmosphere. temporary works, an untitled painting by Logan delivers vast amounts of science with brevity and elegance, South African artist Gavin Jantjes, playfully and is as breezy describing the billion tonnes of dust that blow reverses the theme of genesis from above. from African deserts to fertilize the Amazon as he is discussing the Based on a Khoisan myth from southern echolocation skills of some people with sight impairment. Africa, it depicts the story of a girl dancing around a fire. She throws glowing embers high into the night, thereby creating the The Big Muddy: An Environmental History of the Mississippi and Milky Way, the dominant feature of the Its Peoples, from Hernando de Soto to Hurricane Katrina southern sky. Christopher Morris OXFORD UNIVERSITY PRESS 336 pp. £22.50 (2012) So might the creative sparks tossed sky- Seven years ago this month, Hurricane Katrina triggered massive wards from such an exhibition serve to flooding in the valley of the Mississippi. The environmental backstory illuminate a continent’s worth of artistic of the catastrophe is as rich as river sediment, and historian achievement and potential. ■ Christopher Morris takes us through 500 years of it. The valley’s metamorphosis from vast wetland staked out by France and Spain to Ivan Semeniuk is Nature’s chief of a patchwork of development — drained swamp, levees, deforestation, correspondents in Washington. industry and poor urban planning — is powerfully recounted. 2 A U GUST 2012 | V O L 488 | N A TURE | 3 1 © 2012 Macmillan Publishers Limited. All rights reserved

Correspondence Budget cuts leave in compact-disc players to sanitation improvements. Bat deaths from US science lagging magnetic resonance imaging in Hong Yang, Jim A. Wright wind turbine blades hospitals and much more. University of Southampton, UK. As a student of science, I’m If it doesn’t want to fall [email protected] You suggest that wind turbines thrilled that scientists at CERN, behind, the United States Stephen W. Gundry University kill bats as a result of air-pressure Europe’s particle-physics lab, should be following the lead of of Bristol, UK. changes when they fly through have proved the existence of other nations that are investing the wake of a spinning blade the Higgs boson and advanced in science and technology to (the barotrauma hypothesis). our understanding of the benefit their economies. Better lives, not just However, this is likely to be Universe. But as an American, William J. Richards Hall contraceptives only a minor cause of bat deaths I’m somewhat saddened. Had Institute of Public Policy, New (Nature 486, 310–311; 2012). congressional budget-cutters Jersey, USA. Last month’s London Summit on The barotrauma hypothesis been less short-sighted two [email protected] Family Planning, hosted by the has been criticized as based decades ago, the Higgs boson Bill & Melinda Gates Foundation on erroneous interpretations might have been discovered and the UK government’s of bat injuries (K. E. Rollins by a US-led team instead of by Improve access to Department for International et al. Vet. Pathol. 49, 362–371; a European consortium. On sanitation in China Development, has been hailed 2012). Evidence from bat 4 July, no less. as a resounding success. A total carcasses shows that blunt- In 1993, Congress cancelled We hope that last month’s of US$2.6 billion was pledged to force trauma from the spinning funding for the Superconducting raising of drinking-water provide 120 million women and blades is a much more common Super Collider near Waxahachie, standards in China will help girls in developing countries with killing mechanism (see also Texas, after sinking US$2 billion to speed up improvements in access to family-planning services S. M. Grodsky et al. J. Mammal. into an 87-kilometre particle the country’s sanitation. As by 2020. In measuring the success 92, 917–925; 2011). accelerator that promised to in India, sanitation remains of this welcome campaign, the We hope that this finding establish the United States as inadequate for a rapidly delivery of social change should will be useful in mitigating the the leader in physics research. developing country (Nature also be taken into account. effects of wind turbines on bat Two years later, funding was 486, 185; 2012). The hosts emphasize that mortality. approved by CERN to build the In 2010, 477 million people in results will be rapid and Angelo Capparella, Sabine Large Hadron Collider near China (36% of the population) quantifiable, for example Loew Illinois State University, Geneva, Switzerland. did not have access to improved in terms of the number of Normal, Illinois, USA. US science is facing a growing sanitation such as a ventilated contraceptives supplied. But [email protected] threat from a well-funded pit latrine or a flush toilet piped reducing unwanted pregnancies David K. Meyerholz University anti-science movement, abetted to a sewer system (WHO/ requires other improvements of Iowa, Iowa City, USA. by those corporations and UNICEF Progress on Drinking in women’s lives, such as better politicians opposed to any Water and Sanitation, 2012). education for girls and reduced research that conflicts with their There are national disparities as child mortality (J. Drèze and Giants all around — own vested interests. well, with 74% of people having M. Murthi Popul. Dev. Rev. apart from the squid Apathy towards basic research improved access to sanitation 27, 33–63; 2001), outreach by in the United States is coupled in urban areas in 2010, but only community-health workers and A smile would have crossed the with an increasing reluctance 56% in rural areas. Provision of women’s empowerment (see, late Andrew Huxley’s face at your to invest in science projects sanitation facilities for disabled for example, go.nature.com/ description of his “experiments that do not have a foreseeable people is sparse. bpjgma), and quality family- on the axon of the giant squid” pay-off. But let’s not forget China’s growing population planning programmes. Such (Nature 486, 10–11; 2012). that the pioneers of quantum and urbanization make factors are harder to quantify. Huxley was giant, the axon was mechanics in the 1900s — Niels sewage treatment a particular Focusing simply on what can giant, but the squid were quite Bohr, Albert Einstein and Erwin challenge. Although about be measured encourages short- average in size. Schrödinger — were unable 73% of urban sewage is treated term, narrow interventions Jonathan C. Horton University to offer any practical ideas (China Statistical Yearbook rather than broader, longer- of California, San Francisco, about commercial uses for the on Environment; 2010), more term strategies. For instance, California, USA. subatomic particles, quarks and than 95% of waste water in value-for-money criteria make [email protected] leptons they were bringing to rural areas drains untreated it tempting to sidestep national light at the time. However, if you into rivers and lakes (X. Sun health-care systems, when are reading this on a computer, et al. Chinese Agr. Sci. Bull. 26, supporting these is crucial to CONTRIBUTIONS tablet or smart phone, you have 384–388; 2010). the delivery of appropriate Correspondence may be quantum mechanics to thank. The country has now technologies in developing sent to correspondence@ An estimated 30–35% of increased its surveillance of countries. nature.com after consulting today’s US gross domestic freshwater pollution so that the Devi Sridhar University of the guidelines at go.nature. product is based on inventions new drinking-water standards Oxford, UK. com/cmchno. Alternatively, derived from quantum theory, can be met. This should [email protected] readers may comment from semiconductors in catalyse the government into Karen Grépin New York online: nature.com/nature. computer chips and lasers investing more in nationwide University, USA. 3 2 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

CAREERS GENDER IMBALANCE European universities COLLABORATIONS Agreement will send NATUREJOBS For the latest career aim to eliminate inequalities p.122 US researchers to Europe p.122 listings and advice www.naturejobs.com STOCK4B CREATIVE/GETTY COLUMN The roots of research misconduct Mentors should understand what causes misconduct among trainees — and keep in mind some possible remedies, argues William Neaves. hat is the most effective way for men- Stowers Institute), I dealt with three cases of sci- which the mentor incorporated in a manu- tors to prevent misconduct among entific misconduct. Each led to an admission of script that was then submitted for publication. Wtrainees? First, they should make misconduct, sanctions against the perpetrator After belatedly examining the postdoc’s lab sure that those trainees understand the impor- by the US Office of Research Integrity (ORI) notebook, the mentor discovered the discrep- tance of research integrity. Consistently model- and public disclosure of the person’s identity. ancy between data collected and data included ling good practice beats lecturing hands down, One case also led to the retraction of several in figures, and withdrew the manuscript — but and discussing ethical guidelines at laboratory publications. In none of the cases was there any not before it had been accepted by a journal. meetings helps the team to appreciate honesty wrongdoing on the part of the mentors. When confronted, the postdoc confessed, and — and the grim consequences of misconduct. In the first case, a postdoctoral fellow run- was fired by the host institution. The case took But mentors should also understand the moti- ning 90-minute experiments found that data a twist when the postdoc formally accused the vation behind some acts of misconduct, and the points tended to plateau after the first 30 min- mentor of encouraging misconduct by pres- steps they can take to make sure that misguided utes. Concluding that nothing of interest hap- suring trainees to generate data. The host trainees don’t commit scientific fraud. pened in the final hour, the postdoc started institution conducted an inquiry according While dean of a US biomedical institution fabricating those data points. By taking this to ORI standards and found no evidence that more than a decade ago (before my time at the shortcut, the postdoc quickly generated data, other trainees in the lab perceived unusual 2 A U GUST 2012 | V O L 488 | N A TURE | 1 2 1 © 2012 Macmillan Publishers Limited. All rights reserved

CAREERS GENDER IMBALANCE pressure to produce results. no radio activity. When assayed by the rehired Action plans for equality Fortunately for the scientific reputation of postdoc, these specimens yielded radio- the mentor, he required lab members to main- activity that only he could have added to the The 21 members of the League of tain bound notebooks that included details scintillation vials. When first confronted, he European Research Universities of all experiments and data. The case taught denied every thing. The ORI reviewed results (LERU) are set to implement strategies him always to scrutinize the relevant notebook of the investigation and concluded that the to eliminate gender bias in scientific entries before submitting a manuscript. rehired postdoc had engaged in misconduct. research. Women, Research and In the second misconduct case, a mentor Only then did he acknowledge his guilt. The Universities: Excellence Without Gender had asked a postdoc to purify a protein sam- mentor and co-authors from multiple institu- Bias, a LERU report published on ple so that only a single band remained in a tions retracted four high-profile publications 10 July, outlines recommendations for western-blot assay. Instead, the postdoc used that had been based on the fabricated data. universities, funders, policy-makers and a physical mask so that only one band was On reflection, what could a mentor have publishers to improve gender balance recorded. A technician found the discarded done to prevent this debacle? Simply keep- across the European Union. Institutions mask and took it to the mentor, who con- ing experimental and control specimens have agreed to create and launch action fronted the postdoc; he admitted falsifying blinded during analysis would have suf- plans to mitigate bias, and set up gender- the results. The postdoc was fired by the host ficed. More over, I believe that a mentor in equality offices. A LERU target group will institution and sanctioned by the ORI. such circumstances should hear alarm bells meet this November and annually from What if the technician had not discov- if only one per- next spring to monitor progress, says ered the mask? The mentor could still have “Mentors son in a lab can get Kurt Deketelaere, secretary-general of taken steps to safeguard the integrity of the should not avoid the assay to work. LERU in Leuven, Belgium. The European work and his own reputation. Having urged a discussion Whenever results Commission is using the report in its the postdoc to purify the sample until a blot on research depend on human construction of the European Research showed only one band, he could have sought integrity just manipulation or Area (see ‘Research area first steps’). evidence of the purification steps in the post- because of measurement, team doc’s lab notebook. And he could have asked their own members should a second member of the research team to discomfort.” verify each other’s COLLABORATIONS verify that the results were reproducible. work. When a new US–European deal The final case wreaked havoc on a men- person joins the lab, the mentor can make it tor’s research programme. It started when a clear that verification practices do not reflect US early-career researchers have a new graduate student falsified a cell-killing assay mistrust. For consistency, co-workers should route for in-person collaborations with and fabricated data to support the mentor’s also repeat the mentor’s measurements. European colleagues. US National Science favoured hypothesis. The fraud continued Do such practices prevent fraud? They cer- Foundation (NSF) postdoctoral fellows when the mentor retained the culprit as a tainly make it more difficult. Just as impor- or recipients of Faculty Early Career postdoc, on the basis that no one else could tantly, they protect against inadvertent error Development awards will be able to “make the assay work properly”. Over the and subconscious bias. Many of us wish for spend 6–12 months in teams funded by course of several years, the postdoc manu- data that support our theories, and trainees European Research Council grants under factured data for multiple publications. may anticipate outcomes that would please the an agreement announced on 13 July. After the postdoc left the lab to join a bio- mentor. In general, evidence would suggest David Stonner, director of the NSF office technology company, the mentor assigned a that very few trainees curry favour by fabricat- of international science and engineering new postdoc to perform the assay. When he ing data, but mentors should be careful not to in Arlington, Virginia, says that the could not get the expected results, the new encourage misconduct by signalling their dis- partnerships could lead to positions for postdoc personally paid the former postdoc appointment when a trainee’s data confound postdocs and help researchers to build to perform the assay for him. Sure enough, expectations. The chances of falsification or and expand networks. “Getting young the results supported the mentor’s theory. But fabrication of results are greatly reduced when researchers into European research the former postdoc would not show the new a lab uses only blinded specimens and when networks early in their career will pay postdoc how he performed the assay. other lab members are always responsible for dividends for years,” says Stonner. The former postdoc then left his biotech- independently verifying reproducibility. nology job, and the mentor rehired him, In my experience, mentors often avoid assigning him responsibility for the assay. discussing scientific misconduct with lab EUROPE But the rehired postdoc would only perform members, perhaps out of a misguided con- Research area first steps the assay late at night, after everyone had left. cern that doing so might imply mistrust. Frustrated, the new postdoc hid in the lab one There are ways, however, to circumvent this. The European Commission on 17 July night and saw the culprit pipette a radioactive For example, mentors could broach the topic signed agreements with the first five label directly into scintillation vials, without by first discussing the increasing incidence stakeholders of the European Research any attempt to recover it from experimental of retractions (up tenfold in the past decade; Area. The scheme aims to boost research in samples of labelled cells that had been exposed see Nature 478, 26–28; 2011). In that way, the European Union (EU) by: coordinating to the (hypothetical) killing agent. The new they can engage trainees without calling into member states, funders and research postdoc reported his observations to the men- question anyone’s integrity. organizations; making pensions mobile; tor, who immediately informed the dean. Mentors should not avoid a discussion on improving gender equality (see ‘Action The dean learned that the mentor had research integrity just because of their own plans for equality’); and opening hiring never given blinded specimens to the culprit. discomfort. The potential consequences for practices across borders. EU research is To avoid having to rely entirely on the new careers and reputations are too severe. ■ currently fragmented, with little exchange postdoc’s testimony, the university’s chief aca- of information and restricted mobility, demic officer advised the mentor to set up a William Neaves is the president emeritus of said Máire Geoghegan-Quinn, European sting operation. The mentor prepared speci- the Stowers Institute for Medical Research in commissioner for research, in a statement. mens labelled as experimental that contained Kansas City, Missouri. 122 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

FUTURES SCIENCE FICTION DEAD MEAT Interactive art. BY POLENTH BLAKE I pitied them for their squeamishness. I’d refused the plastic coverings, pre- JACEY rue art is seamless. That’s why I ferring to wander the halls in a mini crafted my new gallery out of meat. skirt and tank top. I wanted to feel TPatrons squelched around the the blood on my skin. entrance hall in plastic coats and boots. One Most of my middling work was man touched the walls gingerly. His hand inconsequential. The same old came away red with blood. designs, as I struggled to find “Welcome to Dead Meat II,” a voice my way. I ignored the pieces greeted him from the wall. The voice holes in favour of the patrons. A lined the entrance room, each with its own child watched the floor, set of vocal cords and lungs. They’d been pressing his foot against modelled after mine. Although the tone the flesh to see it ooze. He was right, the organic quality was missing. realized that the true art was Higher functions needed the computer’s the building. It made it all control and it never quite got it. worthwhile. The entrance had information on the A man fainted and nicked original Dead Meat. It had been a stroke the wall with his watch as he of genius, if I do say so myself. Meat bricks slumped into it. Pores opened around a metal frame. I’d fused the meat in the floor to reabsorb the blood from together with transglutaminase, so it looked the cut, while the edges of the tear pulled like it’d grown as one solid lump. together. Designing the repair nanobots Genius, but temporary. All meat goes had been the hardest part. A true immune bad in the end. Still, Clarissa Brookes was a system would react with or without control, worldwide name before the last piece rotted but the nanobots needed the computer. They “Get the back-up systems.” away. I had no trouble getting sponsors for were the final seam in the construction. For They left and I turned my attention to the this gallery. the next Dead Meat, I’d do away with the last wounds. Even without control, the nanobots The door to the first section opened as I of the machinery. would run a basic routine of sealing meat approached. Far superior to any mechanized The doors jerked open. Two security to meat. That was something I had. The door. The muscles could open it slowly for guards hurried through, in the manner of bomber, and several guards, were splat- the casual wanderer or pull back quickly people who don’t want to panic the patrons. tered over the walls. I picked up a mangled when a child ran through. The door under- The door muscles trembled in response. arm and placed it against the bleeding door stood. Cold metal and plastic couldn’t mimic I followed the guards, resting a reassuring muscle, until the arm fused over the damage. that. hand on the doorframe as I passed. I started the same process on the large My sponsors had insisted on displays Muffled shouts came from farther in hole, but before long the blood was up to of my older work. I’d resisted at first, but I the gallery. No doubt one of the protestors, my ankles. Too much for the pores to take came around to seeing the merit. My patrons spouting about abominations of nature and back in. learnt with me. Paintings of daisies growing other nonsense. As though a living gallery “It’ll be an hour, ma’am,” a guard reported out of women. An elephant’s head flowed was worse than cars and aeroplanes. At least from the door, with a glance at the arm into a whale’s body. The works were crude, my work would merge back into the earth lodged there. but the concept of seamless design was lurk- in the end. When I was younger, in the days when ing in the brush strokes. One of the guard’s radios hissed out a I thought that painting was the ultimate “They’re a bit creepy, aren’t they?” a message. Too quiet to hear, but the guards expression, I’d have been angry. Now, anger woman noted to her companion. were running now. was just another seam, keeping me from I resisted commenting. Some patrons A dull boom reverberated through the what I had to do. I climbed into the wound were slower learners than others. meat. Every door muscle spasmed. and rested my head against the damaged For my first sculpture, I’d sewn the front I sprinted past the guards towards the flesh. The guard’s shouts faded as my wound ends of two sheep together. The fleece was sound. A stream of blood flowed from a torn healed around me. arranged to cover the seams, but I knew that door muscle, but it was minor compared My eyes opened in every room. I mar- they were there. Stitches are still stitches, no with the damage in the room. It had been a velled at the seamless perfection of the matter how well you hide them. precise attack. The bomber set it off by the artwork wandering my corridors, in their Blood oozed from the ceiling more freely wall nearest the control centre, taking out shades of wonder and fear. in the next room. A few splashes landed on the meat and a chunk of the computer. The In my entrance room, I greeted my art. the hood of one patron. muscles would survive on their own for a “Welcome to Dead Meat III.” ■ After a whispered con- NATURE.COM while, but without a computer to coordinate versation with his part- Follow Futures on repair, Dead Meat II would bleed to death. Polenth Blake (www.polenthblake.com) ner, they headed back Facebook at: The guards arrived behind me. “Ma’am?” lives in England with her pet cockroaches, towards the entrance. go.nature.com/mtoodm one asked. and prefers her meat cooked. 124 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

BRIEF COMMUNICATIONS ARISING Models of grid cells and theta oscillations ARISING FROM M. M.Yartsev, M. P. Witter & N. Ulanovsky Nature 479, 103–107 (2011) Grid cells recorded in the medial entorhinal cortex (MEC) of freely and in putative velocity-controlled oscillatory inputs identified as inter- moving rodents show a markedly regular spatial firing pattern whose neurons within the septohippocampal circuit . 7 1 underlying mechanism has been the subject of intense interest. Yartsev et al. recorded the firing of grid cells from bats trained to 1 Yartsev et al. report that the firing of grid cells in crawling bats does crawl within the recording environment, a behaviour that they per- not show theta rhythmicity ‘‘causally disproving a major class of form very slowly (a mean speed of 3.7 cm s 21 versus 17.6 cm s 21 in computational models’’ of grid cell firing that rely on oscillatory our rat data), often stopping entirely (supplementary figure 11 in 2–7 interference . However, their data may be consistent with these ref. 1). The authors found grid cells with very low firing rates (a mean models, with the apparent lack of theta rhythmicity reflecting slow peak rate of 0.56 Hz versus 5.14 Hz in our data) and little significant movement speeds and low firing rates. Thus, the conclusion of theta modulation. However, matching movement speed is important Yartsev et al. is not supported by their data. for comparisons involving theta. At low speeds movement-related In oscillatory interference models, path integration is performed by theta rhythmicity is strongly attenuated 12 and the need for path velocity-dependent variation in the frequencies of theta-band oscilla- integration is reduced. Equally importantly, low firing rates impede tions, which combine to generate the grid-cell pattern 2–4,6,7 . In addi- detection of theta rhythmicity (5–10 Hz), which requires periods con- tion, learned associations to environmental sensory inputs (possibly taining plenty of spikes fired within tens to hundreds of milliseconds mediated by place cells) ensure that grids are spatially stable over time of each other (something that is absent in bat interspike interval and are sufficient to maintain firing in familiar environments 2,3,8 .In histograms; supplementary figure 2b in ref. 1). rats, the majority of grid cells show theta-modulated firing 9,10 , and the We examined whether differences in movement speeds and firing model predicts specific relationships between modulation frequency, rates between the rat data and the bat data could explain the apparent 4 running velocity and grid scale , which have been verified in grid cells 11 lack of theta rhythmicity in bat grid cells. We took random samples of 25 cells from a representative data set of 85 grid cells recorded in rat MEC (Fig. 1a, bottom row), extracted periods of slow running to Model, reduced Down-sampled c Model a Rat data bd rat data fring rate match bat movement speeds, and duplicated this data until it exceeded 5.73 Hz 0.71 0.54 Hz 0.70 4.83 Hz 0.93 0.44 Hz 0.65 the duration of the longest bat trial (60min). We then randomly discarded spikes to match the mean firing rates of each of the 25 published bat grid cells. From the 25 down-sampled rat cells matching each bat grid cell, we selected the one with the median theta index as representative. This process was repeated 10 times. Subjecting the 10 sets of 25 down-sampled cells to the analyses of Yartsev et al. produced ×10 –3 ×10 –5 ×10 –3 ×10 –5 a relative absence of theta rhythmicity (Fig. 1b, fourth row). So, if rats 2 4 6 4 Figure 1 | Down-sampled rat grid cells and oscillatory interference 1 2 3 reproduce bat grid-cell firing. a, b, The firing of grid cells in rats (a) resembles 2 1 grid-cell firing in bats if the rat data are down-sampled to match the low firing 0 0 0 0 rates and slow movements of the bat data (b). c, d, The oscillatory interference 0 250 500 0 250 500 0 250 500 0 250 500 model simulates theta-modulated grid cell firing in rats (c), and also apparently Time lag (ms) un-modulated grid-cell firing in bats when firing rates are reduced (d). 15 80 a–d, Toprow, examplefiring-rate maps (peak rate andgridness, above). Second row, example spike-train autocorrelograms. Third row, distributions of Cell count 10 5 40 10 5 gridness scores. Fourth row, distributions of theta modulation (theta index). Grid cells have gridness . 0.33 (red line). ‘Theta-modulated cells’ have a theta 0 0 0 index of $ 5 (red line). The theta index exceeded the 95th percentile for that cell’s temporally shuffled spike times for 58% of rat cells (a) but only for 2% of 0 1.0 0 1.0 0 1.0 0 1.0 Gridness cells down-sampled to match the bat data (b; averaged over 10 samples of 25 cells). This rises to 14% if speed is not down-sampled, 20% if only the 25 most 15 100 strongly theta-modulated rat cells are used and 72% for the 25 most strongly 40 Cell count 10 5 50 20 20 theta-modulated cells, if speed is unmatched. However, we do not consider this last cell population to be comparable to the bat grid cells because of the pre- 10 selection of only the most strongly theta-modulated cells and the difference in 0 0 0 0 movement speed between running rats andcrawling bats. Theta index, gridness 0 5 10 15 0510 15 0510 15 0510 15 and shuffling follow ref. 1 (in which theta index is theta power divided by mean Theta index power 0–50 Hz), except for a, bottom row, which shows theta index calculated 18 e following ref. 13 (that is, theta power divided by mean power 0–125 Hz), giving higher values that match the proportions of theta-modulated cells in ref. 13 Cell count 12 6 (which range from 62% in layer V, where most bat cells were recorded, to 90% in layer III). e, Schematic showing how theta-modulated inhibitory spike trains producing spikes when exceeding a threshold (middle, red dashed line). Spatial 0 (top, black ticks) drive the grid cell’s membrane potential (middle, black trace), 0 20 40 firing fields (bottom) are defined by constructive interference (top, grey lines Theta index show theta modulation; middle, grey line shows the resulting interference 12 pattern), but the underlying oscillations are undetectable at low firing rates (see Time (s) Methods for details). 2A U G U S T 2 0 1 2|V O L4 8 8 |N A T U R E|E 1 ©2012 Macmillan Publishers Limited. All rights reserved

BRIEF COMMUNICATIONS ARISING moved as slowly as bats and their grid cells fired as infrequently, rat email: [email protected] grid cells would show bat levels of gridness (below the higher levels 2 University College LondonInstitute of Neurology,Queen Square, London seen in rats), and theta modulation would be very hard to detect. WC1N 3BG, UK. Most importantly, to disprove the model requires knowing how 3 University College London Institute of Behavioural Neuroscience, 26 much theta rhythmicity it predicts in low-firing-rate cells. Bedford Way, London WC1H 0AP, UK. Simulations (using code adapted from ref. 7) with strong theta modu- 4 University College London Department of Cell and Developmental lation and typical firing rates for rats (Fig. 1c) also lack significant Biology, Gower Street, London WC1E 6BT, UK. theta modulation when firing rates are reduced to bat levels (Fig. 1d, 5 Sainsbury Wellcome Centre for Neural Circuits and Behaviour, fourth row). Although spatially modulated firing is driven by inter- University College London, Gower Street, London WC1E 6BT, UK. ference between theta-modulated inputs, the theta rhythmicity is undetectable in low-rate spike trains (Fig. 1e). Received 18 November 2011; accepted 26 April 2012. Local field potentials and multi-unit activity were also reported in 1 bats , but these reflect the physical arrangement and coherence of 1. Yartsev, M. M., Witter, M. P. & Ulanovsky, N. Grid cells without theta oscillations in the entorhinal cortex of bats. Nature 479, 103–107 (2011). populations of cells, which may vary between species and are not 2. Burgess, N., Barry, C., Jeffery, K. J. & O’Keefe, J. A grid and place cell model of path addressed by the model (although spatially offset grids require integration utilizing phase precession versus theta. http://cdn.f1000.com/ 7 phase-offset oscillators , suggesting no overall phase preference in posters/docs/225 (Computational Cognitive Neuroscience Conference Poster, 2005). the model). Finally, consistent with the model, grids might be set 3. Burgess, N., Barry, C. & O’Keefe, J. An oscillatory interference model of grid cell up through oscillatory interference during the initial training of the firing. Hippocampus 17, 801–812 (2007). bats to not fly out of the box (by physically blocking from above), and 4. Burgess, N. Grid cells and theta as oscillatory interference: theory and predictions. maintained (at lower firing rates) by learned sensory associations Hippocampus 18, 1157–1174 (2008). 5. Giocomo, L. M., Zilli, E. A., Fransen, E. & Hasselmo, M. E. Temporal frequency of during subsequent slow crawling in the now highly familiar box. subthreshold oscillations scales with entorhinal grid cell field spacing. Science 315, 1719–1722 (2007). Methods 6. Hasselmo, M. E. Grid cell mechanisms and function: contributions of entorhinal The activity of 85 grid cells was recorded from superficial and deep layers of rat persistent spiking and phase resetting. Hippocampus 18, 1213–1229 (2008). 2 MEC during 20 min foraging in 1-m arenas using standard procedures . 8 7. Welday, A. C., Shlifer, I. G., Bloom, M. L., Zhang, K. & Blair, H. T. Cosine directional tuning of theta cell burst frequencies: evidence for spatial coding by oscillatory Random samples of 25 cells were speed matched by removing periods of fast interference. J. Neurosci. 31, 16157–16176 (2011). 21 running, retaining periods of $0.5 s, until the median speed was 3.7 cm s . 8. Barry, C., Hayman, R., Burgess, N. & Jeffery, K. J. Experience-dependent rescaling Speed-matched data were duplicated and concatenated to exceed the duration of entorhinal grids. Nature Neurosci. 10, 682–684 (2007). of the longest bat trial (60 min). Cell firing rates were down-sampled by randomly 9. Hafting, T., Fyhn, M., Molden, S., Moser, M. B. & Moser, E. I. Microstructure of a removing spikes, in turn, to match the mean firing rate of each of the 25 bat spatial map in the entorhinal cortex. Nature 436, 801–806 (2005). 10. Hafting, T., Fyhn, M., Bonnevie, T., Moser, M. B. & Moser, E. I. Hippocampus- grid cells (mean rate taken as 25% of the peak rates found by Yartsev et al. independent phase precession in entorhinal grid cells. Nature 453, 1248–1252 (range of mean rates, 0.03–0.40 Hz)). Spike-train autocorrelograms combined (2008). 11 the individual autocorrelograms from each period of slow running and were 11. Jeewajee, A., Barry, C., O’Keefe, J. & Burgess, N. Grid cells and theta as oscillatory mean-normalized to avoid low-frequency power reducing the theta index (com- interference: electrophysiological data from freely moving rats. Hippocampus 18, pare with figure 4g in ref. 1). Grid cells were simulated as leaky integrate-and-fire 1175–1185 (2008). neurons (time constant 20 ms) receiving three oscillatory inhibitory spike trains 7 12. O’Keefe, J. The Hippocampus Book (eds Andersen, P., Morris, R. G. M., Amaral, D. G., (Poisson processes with rate 50 1 30cos(2pft), in which frequency (f) varies Bliss, T. V. P. & O’Keefe, J.) 475–548 (Oxford Univ. Press, 2006). 13. Boccara, C. N. et al. Grid cells in pre- and parasubiculum. Nature Neurosci. 13, around 8 Hz according to running velocity, with a peak inhibitory synaptic con- 987–994 (2010). ductance 14 of 14 pS) and a noisy persistent excitatory current sampled from 14. Dayan, P. & Abbott, L. F. Theoretical Neuroscience 180–183 (MIT Press, 2001). N(m,2m), in which m 5 336 nA for low firing rates and m 5 436 nA for high rates (mean peak rates are 0.48 Hz and 5.11 Hz, respectively). Author Contributions C.B. analysed and collected experimental data; D.B. performed simulations; N.B. supervised analyses and simulations; and N.B., C.B., J.O. and D.B. contributed to writing. Caswell Barry 1,2,3 ,DanielBush 1,2 ,John O’Keefe 4,5 & Neil Burgess 1,2 1 Competing Financial Interests Declared none. UCL Institute of Cognitive Neuroscience, 17 Queen Square, London WC1N 3AR, UK. doi:10.1038/nature11276 Yartsev et al. reply REPLYING TO C. Barry, D. Bush, J. O’Keefe & N. Burgess Nature 488, http://dx.doi.org/nature11276 (2012) Barry et al. propose that it is impossible to detect theta rhythmicity in focused solely on ‘first generation’ oscillatory interference models, 1 bat grid cells because of their slow movement velocities and low firing ignoring our disproval of ‘second generation’ models. We thus uphold rates; hence, they posit that our findings do not refute the oscillatory our original results and interpretation . 2 2 interference models of mammalian grid cells. To support this Barry et al. analysed a data set of 85 rat grid cells, of which only claim, they use a data set of rat grid cells of which only 58% were 58% were significantly theta-modulated to begin with (although theta modulated, and constrained their analysis to periods of near oscillatory interference models require 100% of cells to be theta- immobility in the rat, a behavioural state in which theta is known modulated). The strength of theta modulation in their data was lower 3 to be absent . Despite these biases, we argue that their own analysis than in the much larger data set publically available from the Moser showed that down-sampled rat cells were substantially more theta- laboratory (a median theta index of 10.9 in Barry et al. compared to 4 modulated than real grid cells from bats, and we demonstrate 14.23 in data from the Moser laboratory ), which may lower the further that the bat data have adequate statistical power to detect theta detectability of theta rhythmicity after the data of Barry et al is rhythmicity—if it was present in bat grid cells. Finally, Barry et al. down-sampled. E 2 |N A T U R E| V O L4 8 8 |2A U G U S T 2 0 1 2 ©2012 Macmillan Publishers Limited. All rights reserved

BRIEF COMMUNICATIONS ARISING Barry et al. proposed that bats’ slow crawling velocity reduces theta network-level theta oscillations 9,10 . This was contradicted by our bat rhythmicity in bat grid cells. Consequently, they selectively removed data, in which brief theta oscillations occurred very rarely in the local- 2 portions of the rat data, retaining short periods, down to 500 ms in field potential , and multi-unit firing (reflecting network activity) 2 duration, until a median speed of 3.7 cm s 21 was achieved. This pro- never showed any theta oscillations . cedure has several flaws. In conclusion, we feel that the analysis by Barry et al. fails to support First, although these velocities correspond to an active movement their main argument, namely that the statistical power of the bat data state of bats, they are equivalent to nearly complete immobility in does not allow detecting theta rhythmicity. To resolve this debate, we 5,6 rats . Barry et al. thus compared a state in rats in which theta is not propose to make use of a large, unbiased, publically available data set 3 expected (near immobility), to a state in bats at which theta, if existing, of rat grid cells, such as that on the Moser laboratory website, which 3 should be most prominent (active movement). Furthermore, in their would allow transparency in analysis techniques and in the baseline model,fortheconstantb(whichdeterminesthevelocitymodulationof rat data being used. Furthermore, we expect that neural recordings dendritic inputs) Barry et al. used a value derived exclusively from rat from single units in flying bats, in which movement velocities and 7 data . However, different mammalian species would probably have neuronal firing rates are expected to be much higher, will provide different velocity dependences in dendritic inputs; for example, when another key approach. modelling grid cells in cheetah versus sloth, it would not make sense to usea bvaluetaken from rat,and thesamegoesfor the modellingofgrid Michael M. Yartsev , Menno P. Witter &NachumUlanovsky 1 2 1 cells in bats. Thus, we contend that movement speed should not be 1 Department of Neurobiology, Weizmann Institute of Science, Rehovot matched, neither when simulating a model nor when down-sampling 76100, Israel. data from one species to mimic another. email: [email protected] Second, Barry et al. used short portions of near-immobility data, as 2 Kavli Institute for Systems Neuroscience and Centre for the Biology of short as 500 ms in duration, creating very intermittent, unrealistic Memory, Norwegian University of Science and Technology, NO-7489 spike trains; this tapers down the oscillatory cycles (because they Trondheim, Norway. estimated 1000-ms autocorrelations using 500-ms data epochs) and could induce an unwarranted statistical bias downwards in detect- 1. Barry, C., Bush, D., O’Keefe, J. & Burgess, N. Models of grid cells and theta ability of theta rhythmicity. oscillations. Nature 488, http://dx.doi.org/nature11276 (2012). Third, Barry et al. found that when firing rates were matched to 2. Yartsev, M. M., Witter, M. P. & Ulanovsky, N. Grid cells without theta oscillations in the entorhinal cortex of bats. Nature 479, 103–107 (2011). those of bats while movement speed was left untouched, 24% of their 3. Buzsa ´ki, G. Theta oscillations in the hippocampus. Neuron 33, 325–340 (2002). theta-modulated grid cells retained significant theta rhythmicity 4. Sargolini, F. et al. Conjunctive representation of position, direction, and velocity in (14% of 58% 5 24%); substantially higher than the 4% (1/25) theta- entorhinal cortex. Science 312, 758–762 (2006). 2 modulated grid cells in bats . Notably, when Barry et al. analysed the 5. Jeewajee, A., Barry, C., O’Keefe, J. & Burgess, N. Grid cells and theta as oscillatory interference: electrophysiological data from freely moving rats. Hippocampus 18, top 51% of their theta-modulated rat grid cells (51% of the 58% 1175–1185 (2008). modulated cells 5 25 neurons)—which are the most relevant cells 6. Wills, T. J., Cacucci, F., Burgess, N. & O’Keefe, J. Development of the hippocampal toconsider fortheir model(especially given the weak thetarhythmicity cognitive map in preweanling rats. Science 328, 1573–1576 (2010). 7. Burgess, N., Barry, C. & O’Keefe, J. An oscillatory interference model of grid cell in their data set)—they found that when firing rates are matched to firing. Hippocampus 17, 801–812 (2007). those of bats and velocities are left untouched, the large majority (72%) 8. Giocomo, L. M., Zilli, E. A., Franse ´n, E. & Hasselmo, M. E. Temporal frequency of of rat grid cells retained significant theta rhythmicity. Thus, down- subthreshold oscillations scales with entorhinal grid cell field spacing. Science 315, 1719–1722 (2007). sampled rat grid cells were markedly more oscillatory than our bat 9. Giocomo, L. M., Moser, M.-B. & Moser, E. I. Computational models of grid cells. 2 grid cells, supporting our original analysis and interpretation . Neuron 71, 589–603 (2011). Last, Barry et al. considered only single-cell, first-generation 10. Zilli, E. A. & Hasselmo, M. E. Coupled noisy spiking neurons as velocity-controlled 7–9 oscillatory interference models of grid cells , which have been oscillators in a model of grid cell spatial firing. J. Neurosci. 30, 13850–13860 (2010). 9 criticized as theoretically problematic . Some of these problems have 9 been rectified in recent second-generation versions of these models , Author Contributions All authors contributed to writing this reply. which used networks of coupled oscillators and explicitly predicted doi:10.1038/nature11277 2A U G U S T 2 0 1 2|V O L4 8 8 |N A T U R E | E 3 ©2012 Macmillan Publishers Limited. All rights reserved

NEWS & VIEWS PALAEONTOLOGY An insect to fill the gap A complete insect fossil from the Devonian period has long been sought. The finding of a candidate may improve our patchy understanding of when winged insects evolved. See Letter p.82 WILLIAM A. SHEAR point speaking to its importance. We can perceive only what the fossil 300 nsects are, in terms of species number, record permits us to perceive. From our the most successful group of animals current viewpoint, the diversification of Carboniferous Iever to have lived. But their evolution- insects and of our own terrestrial verte- ary origins are a source of controversy, brate ancestors seems to have occurred 9 and will continue to be so until the fossil 325 in two evolutionary bursts . Between record finally yields up unequivocal evi- Palaeodictyoptera 425 million and 385 million years dence of insect beginnings. On page 82 of ago, both groups probably originated 1 this issue, Garrouste et al. claim to have and underwent an initial evolution- found precisely this. Although it can 350 Hexapoda ary radiation as they began occupying hardly be described as well preserved, gap the newly available subaerial realm. the fossil shows a six-legged thorax, long Time (millions of years ago) There then follows a long period, called single-branched antennae, triangular Romer’s gap (360 million to 345 mil- jaws and a 10-segmented abdomen (see 375 lion years ago) for the vertebrates and 1 Fig. 2 of the paper ). Insects are the only Strudiella devonica the longer Hexapoda gap for insects known arthropods (joint-legged inver- (385 million to 325 million years ago), tebrate animals) with this anatomical Devonian during which few, if any, fossils of these combination, allowing the authors to groups can be found (Fig. 1). Then, 400 make a strong case for the fossil’s insectan with apparent suddenness, an explosive identity. appearance of many new forms takes The 8-millimetre-long fossil, which Rhyniella praecursor place in the second round of diversi- the authors named Strudiella devonica, fication. For the insects, large winged 425 was found in a small rock slab excavated species of the major groups (mayflies, at a quarry in Belgium. Strudiella is proto-dragonflies and others, includ- dated to approximately 370 million years ing extinct types) show up, seemingly old, which places it late in the Devo- Figure 1 | Winged beginnings. The fossil record provides ample without precursors. The insects were examples of winged insects from around 325 million years ago, nian period (Fig. 1). This was the time such as the order Palaeodictyoptera from the Carboniferous off and running on their way to world when terrestrial ecosystems were first period. However, there is little evidence of insect evolution before domination. assembling from their aquatic progeni- this time; only a handful of fossils, including 402-million-year-old These gaps, and the two bouts of evo- 2 tors — the first forests were established Rhyniella praecursor, which appears similar to extant collembolan lution that they create, may or may not and the earliest four-legged vertebrates arthropods, have been found. But none of these few examples be real. There is evidence that a period were crawling out from freshwater pools comes from the period between 385 million and 325 million years of low atmospheric oxygen concentra- onto land. So far, only suggestive traces ago, which is referred to as the Hexapoda gap (striped region) tion coincided with the gap period, and of insects have been found in rocks of because of the lack of insect evidence. Now, however, Garrouste this could have suppressed the rate of 1 10 this age. The famous Rhynie chert, a et al. report the finding of Strudiella devonica, a fossil dated to appearance of novel anatomy . But a 370 million years ago that shows multiple anatomical features that sedimentary deposit in Scotland that are characteristic of insects. more parsimonious explanation is sim- is about 402 million years old, contains ply that we have not yet found the right 3 fossils of collembolans , a class of animal that For example, the fossilized head of a wingless rock formations to reveal fossils that would fill 6,7 contains today’s ubiquitous springtails, and insect found in Canadian strata somewhat in the gaps. For example, most of the exposed which is regarded as closely related to insects. older than the New York deposits is almost cer- strata for this period in Europe and North The Rhynie chert has also yielded a pair of tainly a contaminant — a much more recent America are of marine, not land, origin. jaw fossils called Rhyniognatha, which may or contemporary insect lodged in a crack in This brings us to the second reason for 4 be from an advanced, winged insect . In New the rocks. And Leverhulmia mariae, also from the importance of Strudiella — it is dated to York state, some fossilized scraps of character- Scottish chert near Rhynie, could have been an a time smack in the middle of the Hexapoda istic cuticle and the framework of a single com- insect, a close relative, or neither — it seems gap (Fig. 1). According to Garrouste et al., 8 pound eye, perhaps from a primitive, wingless to have too many legs to be easily classified . this significantly narrows the gap. And if, as insect, have been found in 385-million-year So although its age makes it too late to be the authors suggest, the fossil came from the 5 old rocks . But these fragments more or less an insect ancestor, or even the earliest insect, young stage of an animal that would have had complete the picture of all that is known of Strudiella is nonetheless of great potential wings as an adult, their finding would mean insects at this crucial time in Earth’s history. significance as the oldest complete insect that winged insects originated much earlier There have also been some false alarms. fossil yet found. This is the first and primary than fossils have heretofore told us, and that 3 4 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

NEWS & VIEWS RESEARCH the sudden appearance of many winged kinds which are largely indistinguishable from living (eds Gensel, P. G. & Edwards, D.) 29–51 (Columbia 11 around 325 million years ago is deceptive. It forms . The beginnings of the insects are to Univ. Press, 2001). would also suggest that the Rhyniognatha be found in rocks much older even than those 3. Greenslade, P. J. & Whalley, P. E. S. In Proc. 2nd Int. Semin. Apterygota (ed. Dallai, R.) 319–323 (Univ. fossils could indeed be the mandibles of that enclosed Strudiella, but almost no one is Siena, 1986). a winged insect, and that the diversification looking for them. The paltry few insect fossils 4. Engel, M. S. & Grimaldi, D. A. Nature 427, 627–630 of winged species could have taken place contemporary with Strudiella — and indeed (2004). at a much more leisurely pace, over some Strudiella itself — were serendipitous, not 5. Shear, W. A. et al. Science 224, 492–494 (1984). 6. Labandeira, C. C., Beall, B. S. & Hueber, F. M. 45 million years. deliberate, finds, and the Hexapoda gap still Science 242, 913–916 (1988). Considering the crucial role of insects in looms large. ■ 7. Jeram, A. J., Selden, P. A. & Edwards, D. Science present-day ecology, the number of people 250, 658–661 (1990). 8. Fayers, S. R. & Trewin, N. H. Palaeontology 48, engaged in studying their fossil history is dis- William A. Shear is in the Department 1117–1130 (2005). mally small. Furthermore, current specialists of Biology at Hampden-Sydney College, 9. Labandeira, C. C. & Sepkoski, J. J. Science 261, focus mostly on events from the Mesozoic Hampden-Sydney, Virginia 23943, USA. 310–315 (1993). period — the ‘Age of Dinosaurs’ — which e-mail: [email protected] 10. Ward, P., Labandeira, C. C., Laurin, M. & Berner, R. A. Proc. Natl Acad. Sci. USA 103, 16818–16822 began some 70 million years after the time of 1. Garrouste, R. et al. Nature 488, 82–85 (2012). (2006). the first known winged insect fossils, or on 2. Shear, W. A. & Selden, P. A. In Plants Invade the Land. 11. Grimaldi, D. A. & Engel, M. S. Evolution of the Insects even more recent amber-preserved insects, Evolutionary & Environmental Perspectives (Cambridge Univ. Press, 2005). EARTH SCIENCE of soil organic matter caused by rising global 8 temperatures . The balance of the past 50 years, Ballantyne et al. used a strikingly To assess variations in carbon sinks over the simple approach: they calculated the annual carbon budget changes in the global atmospheric CO 2 inven- tory from a worldwide network of long-term observations. By subtracting the annual total amount of anthropogenic CO 2 emissions from Careful analysis reveals that the global uptake of anthropogenic carbon dioxide these changes, they quantified the net CO 2 emissions by carbon sinks has doubled during the past 50 years — but the uptake by the land and oceans each year. This fractions of this absorbed by land and by sea remain unclear. See Letter p.70 fundamental strategy reminds us that from time to time we should step back and carefully consider the basic inputs and outputs of the INGEBORG LEVIN afforestation or increased biomass produc- carbon cycle at a global level. Although similar 2,9 tion by plants. These carbon sinks have been work has been done before , Ballantyne et al. recisely quantifying the fate of man-made efficiently damping increases in anthropo- go one step further by quantifying an increase carbon dioxide is vital for reliably esti- genic CO 2 (Fig. 1), but will not necessarily in global CO 2 uptake as worldwide emissions Pmating future atmospheric CO 2 levels continue to do so. Possible mechanisms that are increasing. and the contribution of this greenhouse gas could result in reduced carbon uptake include To determine the uncertainties of their esti- to global climatic change. There has been changes in carbon chemistry and mixing in mated carbon budget, the authors combined 7 much debate about whether currently active the oceans , and potential feedback from land the most comprehensive, high-accuracy carbon sinks are likely to change drastically in ecosystems such as increased decomposition measurements of atmospheric CO 2 available 1 the near future , or might already with three inventories of estimated have weakened during the past global fossil-fuel CO 2 emissions, 2,3 few decades . On page 70 of this 500 and with three inventories of 4 issue, Ballantyne and colleagues emissions estimated to have been clarify matters. By calculating the caused by land-use change. The worldwide increase in the atmos- 450 inventories had been made inde- pheric CO 2 burden during the past CO 2 pendently of each other, and were 50 years from precise global obser- Total anthropogenic CO 2 accumulation obtained using different methods. in sinks vations, and by carefully account- CO 2 (parts per million) 400 Bringing together all these data ing for anthropogenic carbon allowed the authors to reliably cal- sources and their uncertainties, culate the errors of their different they find that total global carbon 350 Atmospheric CO budget terms. Their finding that sinks have not declined during this 2 the net carbon sink is increasing is period. Rather, they have increased therefore robust, and proves that more than twofold — although fos- 300 ocean and land sinks have not both 1960 1970 1980 1990 2000 2010 sil-fuel CO 2 emissions have risen Year decreased in the last half century. 5,6 almost fourfold over this time. How the total sink was divided Roughly half of the anthropo- between oceans and land in the genic CO 2 emissions caused by Figure 1 | Partial absorption of anthropogenic carbon dioxide. The past, and whether the oceans’ CO 2 - graph compares atmospheric levels of carbon dioxide since the 1960s the burning of fossil fuels and by 11 uptake rate has decreased, as sug- (measured at Mauna Loa, Hawaii ) with the levels that would have occurred 2 land-use change (such as deforest- as a result of the accumulation of anthropogenic CO 2 emissions in the gested by some modelling studies , ation) are currently taken up by the absence of carbon sinks (as calculated by Ballantyne et al. ). The amount of remain open questions. 4 oceans or re-enter the terrestrial CO 2 that has accumulated in sinks is represented by the difference between Why is it so important to know biosphere — for example, through the two curves (green arrow). where half of the carbon that we are 2 A U GUST 2012 | V O L 488 | N A TURE | 3 5 © 2012 Macmillan Publishers Limited. All rights reserved

NEWS & VIEWS RESEARCH abc Antisense oligonucleotides DMPK gene Mutant DMPK 50 Years Ago RNase H MBNL protein The importance of lunar natural mRNA resources for the future of space exploration can scarcely be Toxic mRNA exaggerated. Lunar resources will not only play an important part in the establishment of a lunar base by providing life support Nucleus materials and vehicle fuels but will also be an important, and perhaps a limiting, factor in the logistics of interplanetary space Cytoplasm exploration. Certainly, only the most cursory exploration of the solar system could be conducted Figure 1 | How to silence a toxic RNA. a, Most messenger RNAs, such as that encoded by the DMPK using either existing or planned gene, are processed by splicing proteins and then rapidly exported from the nucleus into the cytoplasm, where they are translated into proteins. MBNL splicing proteins interact with many mRNAs, propulsion systems so long as the including the DMPK mRNA. b, Expansion of a repeat region in DMPK mRNA causes myotonic rocket vehicles must lift all their dystrophy type 1 by sequestering MBNL proteins and retaining them in the nucleus, thereby affecting fuel from the surface of the Earth. 5 the splicing and expression of many cellular RNAs. c, Wheeler and colleagues describe the successful A lunar fuel source, on the other use of antisense oligonucleotides (short RNA-like molecules) to ameliorate the disease’s symptoms in a hand, would provide an extremely mouse model. The oligonucleotides bind to the mutant RNA and selectively induce its destruction convenient low-gravity refuelling in the nucleus by the enzyme RNase H. station in space. From Nature 4 August 1962 and correction of RNA splicing defects and the are surmountable — although significant resultant myotonia. Astonishingly, the most hurdles remain regarding safety and delivery effective oligonucleotides continued to confer to affected tissues other than skeletal muscle, 100 Years Ago some benefit up to a year after treatment had such as the heart and brain. The authors’ find- been discontinued. ings also suggest that gapmer-based strategies Animal tests of RNA-directed therapies for might be suitable for the treatment of other A note bearing on the much- muscle diseases such as DM1 have had lim- disorders caused by expansions of repeated debated question of the age of the 7–9 ited success so far . So, why this apparent DNA sequences (such as amyotrophic lateral earth is given in the Proceedings breakthrough? It comes down to the therapy’s sclerosis and frontotemporal dementia 10,11 ), of the Tokyo Mathematico- probable site of action: the nucleus. Most provided that the mutant RNA tends to remain physical Society by S. Suzuki. The mRNAs are synthesized and spliced in the in the cell’s nucleus longer than the normal calculation refers to the time taken nucleus, then rapidly exported to the cyto- RNA. Furthermore, appropriately designed for the present crust of the earth to plasm. But gapmer oligonucleotides induce gapmer oligonucleotides may aid researchers solidify. A result is obtained on the degradation of RNA by RNase H, which is in defining the functions of specific nuclear supposition that the heat of fusion enriched in the nucleus and almost absent non-coding RNAs, some of which have key liberated by the solidification of from the cytoplasm. Wheeler and colleagues, roles in regulating gene expression. the crust supplies the heat lost by and a second research group working indepen- However, as promising as these findings are radiation, and it is further assumed 9 dently , reasoned that the nuclear retention for the prospect of DM1 therapeutics, they also that the effect of the curvature of expanded-repeat RNAs could make them serve as a cautionary tale for the applications of the earth’s surface may be good targets for RNase H. Consistent with this of antisense oligonucleotides. First, given our neglected. According to these idea, the authors describe how oligonucleo- limited understanding of the roles of nuclear hypotheses the calculated time tides designed to target RNAs that are rapidly non-coding RNAs and the likelihood that their varies between 30 and 300 million exported to the cytoplasm were ineffective sensitivity to this technology is enhanced, care years, according to the kind of rock at decreasing their expression in muscle. By must be taken in oligonucleotide design to (gneiss, basalt, or granite) assumed contrast, oligonucleotides targeting a nuclear avoid potentially deleterious off-target effects. in the calculations. The difficulty RNA (the long non-coding RNA Malat1) dem- Second, developing similarly potent therapies is, of course, our imperfect onstrated similar efficacy to that seen when for target mRNAs that are rapidly exported knowledge of the experimental targeting the expanded-repeat RNA. from the nucleus may require the use of oligo- data on which the conclusions What are the implications of Wheeler and nucleotides that do not act through RNase H. are based. colleagues’ results? They inspire optimism Third, therapeutic success in a mouse model [Editor’s note: Latest estimates give that previous challenges faced by researchers is still a long way from effective application in Earth’s age as 4.5 billion years.] looking at antisense oligonucleotide therapies humans. However, the path to success now From Nature 1 August 1912 for DM1 and other neuromuscular diseases seems clearly visible. ■ 2 A U GUST 2012 | V O L 488 | N A TURE | 3 7 © 2012 Macmillan Publishers Limited. All rights reserved

RESEARCH NEWS & VIEWS Peter K. Todd and Henry L. Paulson are 1. Brook, J. D. et al. Cell 68, 799–808 (1992). 7. Wheeler, T. M. et al. Science 325, 336–339 (2009). in the Department of Neurology, University 2. Harley, H. G. et al. Nature 355, 545–546 (1992). 8. Mulders, S. A. et al. Proc. Natl Acad. Sci. USA 106, 13915–13920 (2009). of Michigan, Ann Arbor, Michigan 48109, 3. Buxton, J. et al. Nature 355, 547–548 (1992). 9. Lee, J. E., Bennett, C. F. & Cooper, T. A. Proc. Natl 4. Renoux, A. J. & Todd, P. K. Prog. Neurobiol. 97, USA. Acad. Sci. USA 109, 4221–4226 (2012). 173–189 (2012). 10. Renton, A. E. et al. Neuron 72, 257–268 (2011). e-mails: [email protected]; 5. Wheeler, T. M. et al. Nature 488, 111–115 (2012). 11. DeJesus-Hernandez, M. et al. Neuron 72, 245–256 [email protected] 6. Mankodi, A. et al. Science 289, 1769–1773 (2000). (2011). ALZHEIMER’S DISEASE In their search for rare APP mutations with significant effect on the risk of developing A protective mutation Alzheimer’s disease, Jonsson and colleagues used a unique collection of genotypic, medical and genealogical records of Icelandic people gathered by the company deCODE Genet- A rare gene variant has been found that decreases the peptide deposition seen in ics. The authors screened the full genome the brains of people with Alzheimer’s disease. The mutation may also slow the sequences of 1,795 Icelanders for variants normal cognitive decline that occurs with age. See Letter p.96 in and outside the APP gene and used com- putational methods to predict these variant sequences in approximately 370,000 compa- BART DE STROOPER & THIERRY VOET substitution in APP (alanine to threonine; triots. This combination of sequence and rela- 8,9 A673T) close to the site at which the protein tionship information enabled the authors to lthough our understanding of the is cleaved by the enzyme β-secretase (Fig. 1). test the association of the APP variants they biological processes causing Alzheimer’s The authors show that fewer amyloid-β (Aβ) identified with common late-onset Alzhei- Adisease remains hazy, one clear defin- peptides are generated in cultured cells that mer’s disease. They found that the protective ing feature of the disease is an accumulation express this gene variant. It is worth noting A673T variant was significantly more com- in the brain of deposits of amyloid-β peptides. that a different amino-acid substitution mon in a group of over-85-year-olds without These form following cleavage of the amyloid at exactly the same site (alanine to valine; Alzheimer’s disease (the incidence was 0.62%) 7 precursor protein (APP), and mutations in the A673V) can cause Alzheimer’s disease , — and even more so in cognitively intact over- APP gene can affect this cleavage and/or the emphasizing the pathological relevance of 85-year-olds (0.79%) — than in patients with 1 biophysical properties of the peptides , leading this amino-acid residue. Alzheimer’s disease (0.13%). Although these to their aggregation into toxic peptide assem- Fascinatingly, Jonsson and colleagues also findings were largely based on predicted geno- 2,3 blies that propagate in the brain and promote found that, in a cohort of people over the age types, the authors demonstrated the accuracy 4,5 neurodegeneration . All APP mutations of 80, those who were heterozygous for the of their calculations by subsequently testing identified so far increase levels of these toxic A673T variant (one of their two copies of the for the A673T gene variant in thousands of amyloid-β species, but might APP mutations APP gene was mutated) performed better the individuals they studied. exist that act in the opposite way, by reducing in a test of mental capacity than did control Jonsson and colleagues also used in vitro peptide accumulation and protecting against subjects. The authors derive from this obser- studies to show that the mutated APP pro- neurodegeneration? On page 96 of this issue, vation the rather bold conclusion that the tein undergoes about 40% less cleavage by 6 Jonsson et al. report finding the proverbial identified variant not only protects against β-secretase compared with wild-type APP. needle in a haystack: a rare variant of the APP Alzheimer’s disease but also against the normal Together, β-secretase and γ-secretase are gene that protects against Alzheimer’s disease. mild cognitive decline that is associated responsible for the cleavage of Aβ peptides 6 The mutation causes an amino-acid with old age . from APP (Fig. 1); the enzymes are currently being assessed as targets for the treatment 10 of Alzheimer’s disease , and the finding of reduced Aβ accumulation following impaired Amyloid precursor protein β–Secretase Aβ peptide γ–Secretase β-secretase cleavage is encouraging for this line of inquiry. At first glance, one is tempted to draw the exciting conclusion that a lifelong, moderate lowering of Aβ is all that is needed to postpone Alzheimer’s disease and to maintain good cognition. Indeed, researchers have long Alanine to threonine: protective been struggling with the tantalizing question Alanine to valine: causative of to what extent levels of Aβ must be lowered to be effective in the treatment or prevention 11 of Alzheimer’s disease . However, it remains to be seen how these in vitro observations translate in the human brain, where the protective A673T variant is expressed together with a wild-type copy of Figure 1 | Amyloid precursor protein and Alzheimer’s disease. The amyloid precursor protein (APP) the gene. If the translation is linear, then the is cleaved by the enzymes β-secretase and γ-secretase into amyloid-β (Aβ) peptides (scissor symbols represent the cleavage sites). An accumulation of Aβ peptides is seen in the brains of patients with 40% decrease in Aβ levels observed in vitro Alzheimer’s disease, and multiple mutations that cause changes in the amino-acid sequence of APP (blue would result in a 20% decrease in Aβ in people 6 bars) are associated with the disease 14,15 . By contrast, Jonsson and colleagues have identified a mutation in carrying the mutation. It would be extremely the APP gene that reduces Aβ accumulation and that protects against Alzheimer’s disease. The mutation instructive to test this prediction using blood, results in an alanine-to-threonine amino-acid substitution close to the β-secretase cleavage site (red bar). cerebrospinal fluid or cells from individuals 7 Interestingly, an alanine-to-valine substitution at the same site can cause Alzheimer’s disease . carrying the mutation. 3 8 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

NEWS & VIEWS RESEARCH Despite this promising prospect, the question Disorders, KU Leuven, 3000 Leuven, Belgium. 8. Kong, A. et al. Nature Genet. 40, 1068–1075 remains as to whether a reduction in Aβ B.D.S. is also at the VIB Center for the Biology (2008). 9. Marchini, J., Howie, B., Myers, S., McVean, G. & explains the protective effect of the gene vari- of Disease, Leuven. Donnelly, P. Nature Genet. 39, 906–913 (2007). ant identified by Jonsson and colleagues. Here, e-mail: [email protected] 10. De Strooper, B., Vassar, R. & Golde, T. Nature Rev. it is worth keeping in mind a previously iden- Neurol. 6, 99–107 (2010). 1. De Strooper, B. Physiol. Rev. 90, 465–494 11. Karran, E., Mercken, M. & De Strooper, B. Nature tified Alzheimer’s-disease-causing mutation, (2010). Rev. Drug Discov. 10, 698–712 (2011). 7 the A673V APP gene variant . This mutation 2. Haass, C. & Selkoe, D. J. Nature Rev. Mol. Cell Biol. 8, 12. Meinhardt, J. et al. Protein Sci. 16, 1214–1222 increases Aβ generation but causes dementia 101–112 (2007). (2007). only in people in which both gene copies are 3. Benilova, I., Karran, E. & De Strooper, B. Nature 13. Chávez-Gutiérrez, L. et al. EMBO J. 31, 2261–2274 Neurosci. 15, 349–357 (2012). (2012). mutated, not just one. It also affects not only 4. Eisenberg, D. & Jucker, M. Cell 148, 1188–1203 14. Wilquet, V. & De Strooper, B. Curr. Opin. Neurobiol. the amount, but also, and importantly for this (2012). 14, 582–588 (2004). discussion, the biophysical properties of the 5. Stöhr, J. et al. Proc. Natl Acad. Sci. USA 109, 15. Goate, A. & Hardy, J. J. Neurochem. 120, 3–8 Aβ that is generated. It seems that the mutated 11025–11030 (2012). (2012). 6. Jonsson, T. et al. Nature 488, 96–99 (2012). protein interacts with wild-type Aβ to prevent 7. Di Fede, G. et al. Science 323, 1473–1477 B.D.S. declares competing financial interests. the generation of toxic Aβ assemblies. Given (2009). See go.nature.com/uwtjbg for details. that another mutation at the same site in the APP protein also affects the aggregation prop- 12 erties of Aβ peptides , the possibility that Jons- QUANTUM OPTICS son and colleagues’ A673T mutation exerts its protective effects by altering Aβ aggregation should be considered. This more qualitative Strongly interacting concept of Aβ toxicity contrasts with the idea that only an increase in Aβ levels can cause dis- ease — and evidence supporting this insight is photons 13 rapidly mounting . Further work is certainly needed to verify whether the A673T mutation protects against A fine marriage between atomic and optical physics has produced a medium that is age-related cognitive decline. Jonsson et al. transparent to single photons but opaque to multiple photons. The finding heralds report that A673T carriers perform better in the development of devices such as single-photon switches. See Letter p.57 cognitive tests than do control subjects, but one wonders whether this can be confirmed by other measurements of cognitive function, THAD G. WALKER a particular photon depends on the proper- and whether confounding factors compli- ties — most notably the number — of other cate the interpretation of the reported result. an photons be made to interact strongly photons in its vicinity. Until recently, however, For example, although there were no known with each other? Until recently, mater- available nonlinear materials required large cases of Alzheimer’s disease in the control Cials with nonlinear optical properties numbers of photons to be present in order for population, many other conditions, such as could mediate photon–photon interactions them to noticeably affect each other. Peyronel 2 Parkinson’s disease or depression (which were that were weak at best. These weak interactions et al. combined several recent developments not excluded in this assessment), can also have previously been artificially enhanced in atomic and optical physics to produce a 1 negatively affect mental capacity. So Jonsson using devices known as optical cavities , which novel nonlinear medium that is transpar- 6 and colleagues’ proposal that “Alzheimer’s make the photons repeat their encounters ent to single photons yet opaque to multiple disease may represent the extreme of the age- thousands to millions of times. On page 57 of photons (Fig. 1). 2 related decline in cognitive function” may this issue, Peyronel et al. demonstrate a new The largest nonlinear optical effects achiev- yet prove to be a premature interpretation of material in which single photons propagate able in atoms occur when a light field renders their findings. freely, but interact so strongly with each other the atoms transparent. Consider a sample of Nevertheless, the identification of a pro- that when just two photons are present one atoms that have three energy levels: a ground tective APP gene variant is certainly exciting, is quickly absorbed. The result opens up the state E g , an excited state E r and an intermediate 2 and it will be interesting to watch for the iden- possibility of realizing concepts such as single- level E e (see Fig. 1b of the paper ). Photons of tification of other protective gene variants, photon switches, deterministic photon-based frequency f 1 that are directed into the sample for example, mutations in the gene encod- quantum logic, and quantum gases of strongly and obey the Bohr equation, hf 1 = E e − E g , will ing β-secretase that might inhibit its expres- interacting photons*. normally be absorbed. However, when a strong sion or its proteolytic activity. As with many We have known since the dawn of quantum ‘control’ laser of frequency f 2 is also shone on genetic findings, more years of hard work will physics a century ago that light consists of par- the sample, the atoms become transparent to be needed to assess the clinical and therapeutic ticles, called photons, of energy hf, where h is frequency f 1 if the condition h (f 1 + f 2 ) = E r − E g implications of such findings. But if the prelim- Planck’s constant and f is the light’s frequency. is satisfied. This electromagnetically induced inary — and quite spectacular — conclusions Photons usually interact extremely weakly transparency (EIT; ref. 3) puts the atoms and of Jonsson et al. regarding the mechanism of with each other, but strongly with the charged photons into collective excitations called action of the A673T mutation, and its impli- particles that comprise matter. In most mater- polaritons. In Peyronel and colleagues’ exper- cations for cognition, can be confirmed, then ials, the optical response is linear — a beam iment, the control laser changes the trans- a lifelong suppression of Aβ production by as comprised of many photons scatters and mission of the atomic gas from essentially little as 20% may one day become the ‘fountain moves from place to place in the same way that zero to 60%. The EIT condition is extremely of youth’ for the brain. ■ single photons do. Inside nonlinear materials, sensitive: photons that obey it are transmit- however, the optical response is altered when ted with high probability, whereas those Bart De Strooper and Thierry Voet are in multiple photons are present. The motion of that violate it are absorbed normally. This is 4 the Center for Human Genetics and Leuven *This article and the paper under discussion were basically an optically controlled switch . 2 Research Institute for Neurodegenerative published online on 25 July 2012. When the upper level r is a state of large 2 A U GUST 2012 | V O L 488 | N A TURE | 3 9 © 2012 Macmillan Publishers Limited. All rights reserved

RESEARCH NEWS & VIEWS demonstrated a quality single-photon light source that has a rate of emission in the mega- hertz regime, as Dudin and Kuzmich have 9 shown using a related approach. The key capa- 2 bility of this experiment — engineering strong Photon Atom icc gas Atomic gas photon–photon interactions at the two-pho- ton level — should also lead to various other new possibilities. For example, single-photon switches, photon detectors of high quantum efficiency, and non-destructive photon detec- tion can easily be foreseen as extensions of this work. The physics of strongly interacting Rydberg polariton photons has a bright future. ■ Thad G. Walker is in the Department of Physics, University of Wisconsin-Madison, Madison, Wisconsin 53706, USA. 2 Figure 1 | A stream of single photons. Peyronel et al. have directed a beam of overlapping photons e-mail: [email protected] into an atomic gas in which single photons are converted into collective excitations known as Rydberg polaritons. The polaritons, which can be thought of as spheres comprising many atoms and one photon, 1. Birnbaum, K. M. et al. Nature 436, 87–90 (2005). strongly absorb additional photons. On exiting the gas, the polaritons are converted back to individual, non-overlapping photons. 2. Peyronel, T. et al. Nature 488, 57–60 (2012). 3. Fleischhauer, M., Imamoglu, A. & Marangos, J. P. Rev. Mod. Phys. 77, 633–673 (2005). principal quantum number n (a Rydberg have a narrower frequency range would 4. Harris, S. E. & Yamamoto, Y. Phys. Rev. Lett. 81, 5 state ), the EIT condition can be easily vio- improve the EIT transmission to nearly 100% 3611–3614 (1998). lated by weak interactions between the atoms. and reduce the overlap of photons from the 5. Saffman, M., Walker, T. G. & Molmer, K. Rev. Mod. Phys. 82, 2313–2363 (2010). For an n of about 100, a single Rydberg atom single-photon source. A looming challenge is 6. Pritchard, J. D., Weatherill, K. J. & Adams, C. S. will cause a violation of the EIT condition for to reconfigure the experiment so that the two- Preprint at http://arxiv.org/abs/1205.4890 all other atoms within a ‘blockade radius’ of photon nonlinearity delays rather than absorbs (2012). 6 10 micrometres. This Rydberg blockade pro- excess photons . This type of nonlinearity, 7. Wilk, T. et al. Phys. Rev. Lett. 104, 010502 (2010). duces record nonlinearities, as shown recently which preserves the number of photons, would 8. Isenhower, L. et al. Phys. Rev. Lett. 104, 010503 6 by Adams and colleagues , and has been be extremely useful for quantum-information (2010). used to entangle neutral atoms separated by purposes. 9. Dudin, Y. O. & Kuzmich, A. Science 336, 887–889 7,8 micrometre-scale distances . A Rydberg In one respect, Peyronel and colleagues have (2012). polariton can be thought of as a 10-μm sphere containing many ground-state atoms and one Rydberg atom — or, equivalently, many SYSTEMS BIOLOGY atoms and one photon. Should other photons enter a volume already occupied by a Rydberg polariton, the blockade effect causes a viola- A cell in a computer tion of the EIT condition, so the photons are absorbed rather than transmitted. Note that if the atom density is low, as in previous The small genomes of some bacteria could provide the first complete 6 experiments , the absorption probability may understanding of a biological system. A new computer model brings still be small. this goal closer, by calculating every process in a dividing Mycoplasma cell. The final, essential ingredient needed to generate strong photon–photon interactions at the two-photon level is an atomic cloud of MARK ISALAN mutations involving gene deletions. such high density that when two or more pho- Mycoplasma genitalium is a urogenital tons enter a blockade volume, all but one are t has long been a dream in biology to push bacterial parasite that has only 525 genes, absorbed within that volume, leaving a single reductionism to the limit: to describe a cell making it one of the smallest genomes of any Rydberg polariton. This ‘photon blockade’ is Ias a set of interacting components and to independently dividing cell — for comparison, the novelty of Peyronel and colleagues’ study. capture whole-cell behaviour in a computer the gut bacterium Escherichia coli has around Their experiment reveals that a multi-photon model. A good model doesn’t simply recapitu- 4,000 genes. Because of their status as one of incident light beam is converted, within a few late the observed behaviours that are fed into the ‘simplest’ cells, Mycoplasma species are micrometres, into a beam of single photons, it. Rather, the aim is to predict the unknown rapidly becoming the most measured biologi- with a small (less than 0.09) probability that effect of any novel perturbation or mutation. cal systems in history, and full descriptions two photons will leave the atomic gas at the Such goals are very ambitious because of the of their molecular content, in terms of DNA, 2–4 same time. Interestingly, even though their challenge of attempting to obtain quantitative RNA, protein and metabolites, are available . sample is large enough for several Rydberg information on every one of the cell’s gene The cells are therefore considered to be the 5 polaritons to coexist, the authors find that products and metabolites. Nevertheless, Karr ideal target for whole-cell modelling . 1 (and explain why) only one photon at a time is et al. , writing in Cell, present the most com- What is striking about Karr and colleagues’ found within the entire sample. prehensive model of a bacterial cell cycle so far, modelis the sheer ambition of its scale and its An exciting feature of this experiment is built on the basis of individual molecules and attention to detail. The authors retrieved (and that there are several clear avenues towards their relationships. Impressively, the model in some cases retested) more than 1,900 exper- improving the properties of the medium. can predict gene-expression levels and cell- imentally derived cellular parameters, such as Cooler, denser atomic gases and lasers that replication times in the challenging context of enzymatic reaction rates and protein-binding 4 0 | N A TURE | V O L 488 | 2 A U GUST 2012 © 2012 Macmillan Publishers Limited. All rights reserved

NEWS & VIEWS RESEARCH 8 affinities, from around 900 publications. mutations variably . However, the Myco- They then combined these to make 28 plasma model can track such variability sub-models of cellular processes, such a Send variables and therefore has the potential to predict as metabolism, protein translation and these outcomes. DNA replication. They used sub-models The metaphor of gene networks being so that they could apply the appropriate 16 cell variables 28 cell-process connected in wiring diagrams is becom- modelling method for each process. In (for example, sub-models ing commonplace and, even though (for example, 9 computational biology, this requirement RNA content, DNA replication, such networks can be non-intuitive , mass) 6 has been neatly summarized as “Don’t metabolism) they are ideal for computer modelling. model bulldozers with quarks”. So the Nevertheless, one of the most excit- authors combined different modelling ing ideas in studies of gene regulation techniques involving varying levels Update variables Cell divided? is that network relationships may not Then STOP of detail, to allow different factors — every second always involve direct molecular inter- including dependence on deterministic b actions. For example, imagine a gene reactions, known constraints, prob- that is required for cell division — if ability and random variability — to be there is low gene expression, the cell applied where appropriate. will not divide and so gene-expression M. LLUCH-SENAR Crucially, the authors then used a levels will have longer to accumulate, computational trick to join up the sub- which can create a feedback loop, or models (Fig. 1a). Models calculate vari- gene expression ‘according to need’ 10,11 . ables — numbers that represent varying Fascinatingly, there is already a hint of system states. And variables change this in one example from Karr and col- according to sets of rules — the equa- leagues’ model. They find that cell-cycle tions and parameters used to describe time is affected by the concentrations of the system. The authors allowed each DNA nucleotides (dNTPs), which are sub-model to calculate independently required for DNA replication. When the values of a set of 16 variables at a dNTP levels are low, the cycle slows time-step of approximately one sec- at the point of replication initiation, ond. They then combined these results, Figure 1 | Looping calculations to model cell division. allowing dNTPs to build up, which then 1 a, Karr et al. have constructed a computer model that attempts which generated a new set of vari- to calculate every process in Mycoplasma genitalium cells. Their speeds up the rest of the cycle. ables, and the process was repeated in modelling strategy involves 28 independent sub-models of cell Extrapolating from such findings, a loop. Thus, all the sub-models ‘com- processes, each incorporating different methods and levels I can imagine that similar feedback municated’ with one another and the of detail. The sub-models communicate by combining their processes might exist for any cellular cell’s status was constantly updated calculations for 16 cell variables for approximately one second factor that contributes indirectly to and recalculated. Although this is an of the cell’s life cycle, and then calculating the next second. The reducing its own concentration. For approximation, because in reality all looping process culminates when cell division is induced. b, A example, factors for which transiently processes happen simultaneously, the scanning electron micrograph of M. genitalium cells, before (left) low concentrations reduce the activ- end results converged plausibly towards and during division (right). Scale bar, 0.5 micrometres. ity of pathways for cell division, pro- the decision to divide, which the authors tein secretion or protein degradation assessed as the moment that the bacterial cell found 8 to be within the limits predicted by might similarly self-regulate, restoring and membrane ‘pinched’ together to form two new the model. In some of these cases, the experi- buffering themselves over time. So perhaps cells (Fig. 1b). ments resolved discrepancies between the the most exciting thing about a whole-cell After some optimization, the model pro- model and published growth rates, identify- model is that it may allow us to look beyond duced estimates of metabolite concentrations, ing, for example, a previously undescribed the direct molecular ‘cogs and wheels’ metabolism rates, and messenger RNA and slow-growth mutant. that drive biology and into the emergent protein levels, that were similar to experimen- So, can the authors claim to have recreated properties of biological systems. ■ tal data. The model also allowed the authors to a cell in a computer? They themselves say that make several predictions about cell behaviour, the model should be compared to the first draft Mark Isalan is at the EMBL/CRG Systems including that 90% of the cell’s genes will be of the human genome, and be considered a Biology Research Unit, Centre for expressed in the first 2.5 hours of the approxi- work in progress. However, in modelling, dis- Genomic Regulation and UPF, mately 9-hour cell cycle. This prediction sug- crepancies between predictions and experi- 08003 Barcelona, Spain. gests that the chromosomes are ‘explored’ mental results are the key to improvements e-mail: [email protected] rapidly by gene-expression machinery. The key — they direct more detailed analyses and 1. Karr, J. R. et al. Cell 150, 389–401 (2012). test, however, was whether higher-level system model refinement, and ultimately lead to better 2. Gu¨ell, M. et al. Science 326, 1268–1271 (2009). properties, such as the time taken for the cell models. More challenging tests could be imag- 3. Yus, E. et al. Science 326, 1263–1268 (2009). to replicate itself, would be correctly predicted ined. For example, could the model predict 4. Kühner, S. et al. Science 326, 1235–1240 for bacteria carrying genetic mutations. When synthetic lethal mutants, in which the com- (2009). the researchers ran the model with each of the bination of two gene deletions will kill a cell, 5. Tomita, M. et al. Bioinformatics 15, 72–84 (1999). 6. Goldenfeld, N. & Kadanoff, L. P. Science 284, 87–89 525 genes individually deleted, they found that although either deletion alone permits sur- (1999). 284 of the genes are essential for cell survival vival? Furthermore, any model that attempts 7. Jelier, R., Semple, J. I., Garcia-Verdugo, R. & Lehner, B. 7 and 117 are non-essential. These numbers are to predict phenotypes (biological proper- Nature Genet. 43, 1270–1274 (2011). approximately 80% in agreement with experi- ties) from geno types (gene sequences) will be 8. Burga, A., Casanueva, M. O. & Lehner, B. Nature mental data for gene deletions that have been subject to the problem that even genetically 480, 250–253 (2011). assessed previously. identical cells do not always give the same 9. Isalan, M. BioEssays 31, 1110–1115 (2009). 10. Tsuru, S. et al. Mol. Syst. Biol. 7, 493 (2011). The authors also tested the growth rates of output. For example, random differences in 11. Kashiwagi, A., Urabe, I., Kaneko, K. & Yomo, T. 12 of the gene-deleted bacterial strains, and the amount of chaperone proteins can ‘buffer’ PLoS ONE 1, e49 (2006). 2 A U GUST 2012 | V O L 488 | N A TURE | 4 1 © 2012 Macmillan Publishers Limited. All rights reserved

ARTICLE doi:10.1038/nature11213 Novel mutations target distinct subgroups of medulloblastoma 1,5 1,4 1,4 1,2 Giles Robinson 1,2,3 *, Matthew Parker *, Tanya A. Kranenburg *, Charles Lu , Xiang Chen , Li Ding 1,5,6 , 1,4 1,2 1,2 1,2 1,2 Timothy N. Phoenix , Erin Hedlund , Lei Wei 1,4,7 , Xiaoyan Zhu , Nader Chalhoub , Suzanne J. Baker , Robert Huether 1,4,8 , 1,2 1,9 1,2 1,8 1,5 1,4 1,4 Richard Kriwacki , Natasha Curley , Radhika Thiruvenkatam , Jianmin Wang , Gang Wu , Michael Rusch , Xin Hong , 1,4 1,9 1,9 1,7 1,4 Jared Becksfort , Pankaj Gupta , Jing Ma , John Easton , Bhavin Vadodaria , Arzu Onar-Thomas 1,10 , Tong Lin 1,10 , 1,9 Shaoyi Li 1,10 , Stanley Pounds 1,10 , Steven Paugh 1,11 , David Zhao , Daisuke Kawauchi 1,12 , Martine F. Roussel 1,12 , 1,4 1,7 David Finkelstein , David W. Ellison , Ching C. Lau 1,13 , Eric Bouffet 1,14 , Tim Hassall 1,15 , Sridharan Gururangan 1,16 , 1,3 Richard Cohn 1,17 , Robert S. Fulton 1,5,6 , Lucinda L. Fulton 1,5,6 , David J. Dooling 1,5,6 , Kerri Ochoa 1,5,6 , Amar Gajjar , 1,7 Elaine R. Mardis 1,5,6,18 , Richard K. Wilson 1,5,6,19 , James R. Downing , Jinghui Zhang 1,4 & Richard J. Gilbertson 1,2,3 Medulloblastoma is a malignant childhood brain tumour comprising four discrete subgroups. Here, to identify mutations that drive medulloblastoma, we sequenced the entire genomes of 37 tumours and matched normal blood. One-hundred and thirty-six genes harbouring somatic mutations in this discovery set were sequenced in an additional 56 medulloblastomas. Recurrent mutations were detected in 41 genes not yet implicated in medulloblastoma; several target distinct components of the epigenetic machinery in different disease subgroups, such as regulators of H3K27 and H3K4 trimethylation in subgroups 3 and 4 (for example, KDM6A and ZMYM3), and CTNNB1-associated chromatin re-modellers in WNT-subgroup tumours (for example, SMARCA4 and CREBBP). Modelling of mutations in mouse lower rhombic lip progenitors that generate WNT-subgroup tumours identified genes that maintain this cell lineage (DDX3X), as well as mutated genes that initiate (CDH1) or cooperate (PIK3CA) in tumorigenesis. These data provide important new insights into the pathogenesis of medulloblastoma subgroups and highlight targets for therapeutic development. Medulloblastoma is the most common malignant childhood brain The genomic landscape of medulloblastoma tumour . The disease includes four subgroups (sonic hedgehog (SHH) To identify genetic alterations that drive medulloblastoma, we per- 1 subgroup, WNT subgroup, subgroup 3 and subgroup 4), defined formed whole-genome sequencing (WGS) of DNA from 37 tumours primarily by gene expression profiling, that show differences in and matched normal blood (discovery cohort). Tumours were sub- 2 karyotype, histology and prognosis . Studies of genetically engineered grouped by gene expression (WNT subgroup, n 5 5; SHH subgroup, mice show that these tumours arise from different cell types: SHH- n 5 5; subgroup 3, n 5 6; subgroup 4, n 5 19; ‘unclassified’ (profiles subgroup medulloblastomas develop from committed cerebellar not available), n 5 2; Fig. 1, Supplementary Figs 1–3 and Sup- 3,4 granule neuron progenitors (GNPs) in Ptch1 1/2 mice ; WNT- plementary Table 1). Validation of all putative somatic alterations subgroup tumours are generated by lower rhombic lip progenitors including single nucleotide variations (SNVs), insertion/deletions 5 8 (LRLPs) in Blbp-Cre;Ctnnb1 1/lox(Ex3) ;Tp53 flx/flx mice ; whereas (indels) and structural variations (SVs) identified by CREST , was subgroup-3 medulloblastomas probably arise from an undefined class conducted for 12 tumours using custom capture arrays and of cerebellar progenitors . The identification of medulloblastoma sub- Illumina-based DNA sequencing (Supplementary Table 2). Putative 6 groups has not changed clinical practice. All patients currently coding alterations and SVs were validated in the remaining 25 receive the same combination of surgery, radiation and chemotherapy. discovery cohort cases by polymerase chain reaction (PCR) and This aggressive treatment fails to cure two thirds of patients with Sanger-based sequencing. Mutation frequency was determined in a subgroup-3 disease, and probably over-treats children with WNT- separate ‘validation cohort’ of 56 medulloblastomas (WNT subgroup, subgroup medulloblastoma who invariably survive with long-term n 5 6; SHH subgroup, n 5 8; subgroup 3, n 5 11; subgroup 4, n 5 19; 2,7 cognitive and endocrine side effects . Drugs targeting the genetic unclassified, n 5 12; Fig. 1 and Supplementary Table 1). alterations that drive each medulloblastoma subgroup could prove WGS of the discovery cohort detected 22,887 validated or high- more effective and less toxic, but the identity of these alterations quality somatic sequence mutations (SNVs and indels), 536 validated remains largely unknown. or curated SVs, and 5,802 copy number variations (CNVs; 92% 1 2 St Jude Children’s Research Hospital, Washington University Pediatric Cancer Genome Project, Memphis, Tennessee 38105, USA. Department of Developmental Neurobiology, St Jude Children’s 3 4 Research Hospital, Memphis, Tennessee 38105, USA. Department of Oncology, St Jude Children’s Research Hospital, Memphis, Tennessee 38105, USA. Department of Computational Biology and 5 Bioinformatics, St Jude Children’s Research Hospital, Memphis, Tennessee 38105, USA. The Genome Institute, Washington University School of Medicine in St Louis, St Louis, Missouri 63108, USA. 6 Department of Genetics, Washington University School of Medicine in St Louis, St Louis, Missouri 63108, USA. Department of Pathology, St Jude Children’s Research Hospital, Memphis, Tennessee 7 8 9 38105, USA. Department of Structural Biology, St Jude Children’s Research Hospital, Memphis, Tennessee 38105, USA. Department of Information Sciences, St Jude Children’s Research Hospital, Memphis, Tennessee 38105, USA. 10 Department of Biostatistics, St Jude Children’s Research Hospital, Memphis, Tennessee 38105, USA. 11 Department of Pharmaceutical Sciences, St Jude Children’s Research Hospital, Memphis, Tennessee 38105, USA. 12 Department of Tumour Biology and Genetics, St Jude Children’s Research Hospital, Memphis, Tennessee 38105, USA. 13 Texas Children’s Cancer and Hematology Centers, 6701 Fannin Street, Ste. 1420, Houston, Texas 77030, USA. 14 The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario M5G 1X8, Canada. 15 The Royal Children’s Hospital, 50 Flemington Road, Parkville, Victoria 3052, Australia. 16 Duke University Medical Center, 102382, Durham, North Carolina 27710, USA. 17 The School of Women’s and Children’s Health, University of New South Wales, Kensington, New South Wales NSW 2052, Australia. 18 Siteman Cancer Center, Washington University School of Medicine in St Louis, St Louis, Missouri 63108, USA. 19 Department of Medicine, Washington University School of Medicine in St Louis, St Louis, Missouri 63108, USA. *These authors contributed equally to this work. 2A U G U S T2 0 1 2|V O L 4 8 8 |N A T U R E|4 3 ©2012 Macmillan Publishers Limited. All rights reserved

RESEARCH ARTICLE Age and sex Histology Stage Outcome Chromosome nCTTNB1 Cohort Figure 1 | The genomic landscape <5 yr Melanotic M0 Disease free Balance – WGS Wild type Focal del. of medulloblastoma. Top, clinical, >5 yr Classic M1 Progression Loss + Valid. Missense Homo del F Desmoplastic M2 Dead Gain ND Nonsense/FS Focal amp. histological, gross chromosomal, M Anaplastic M2 ND Splice site Microdel. nuclear CTNNB1 (nCTNNB1) and WNT SHH Subgroup 3 Subgroup 4 cohort (discovery or validation) ER 123 4 5 6 7 8 9 10 11 1 23456789 10 11 12 13 123456789 10 11 12 13 14 15 16 17 1 2 3 456789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 details of 79 medulloblastomas by Age NS Sex ** subgroup. ER, enrichment. Bottom, Histology *** Clinical Stage * genetic alterations detected in 27 Outcome NS Mono 6 *** genes of particular interest. Colour 7q * 9p * key at top right. ANOVA 9q * 10p NS (continuous) or Fisher’s exact test 10q * Chromosome 16q *** (categorical) P value is shown on 17p * 17q *** right. False discovery rate (FDR) Y NS nCTNNB1 *** IHC estimates of each mutation are Cohort NS Cohort shown on right. Slash indicates loss ER FDR or mutation of wild-type allele, CTNNB1 *** 3.3 × 10 –14 CDH1 * 0.02 including X chromosome in males. DDX3X *** 4.6 × 10 –12 ***P , 0.0005; **P , 0.005; SMARCA4 ** 6.7 × 10 –8 WNT CREBBP * NS *P , 0.05; NS, not significant. F, TRRAP NS NS MED13 NS NS female; M, male. amp., amplification; SUFU NS 0.06 PTCH1 * NS del., deletion; microdel., TP53 NS 3.7 × 10 –6 MLL2 NS 0.04 SHH microdeletion; valid., validation GABRG1 * 0.08 PTEN * - cohort. ND, not done. MYCN NS - MYC NS - OTX2 NS - DDX31 NS - KDM6A NS 4.6 × 10 –11 ZMYM3 NS 0.18 KDM1A NS NS KDM3A NS NS Subgroup 3/4 KDM4C NS 0.006 KDM5A NS NS KDM5B NS NS KDM7A NS NS CHD7 NS 0.0006 PIK3CA NS 0.12 concordant with 6.0 SNP mapping arrays; Supplementary Tables 3–6 with excessive SVs) encode potential in-frame fusion proteins and Supplementary Figs 4–7). In all but five tumours with the highest (Supplementary Fig. 15); none affect the same gene or signal pathway. mutation rates, .50% of SNVs were CRT/GRA transitions Therefore, fusion proteins are likely to be an uncommon transform- (Supplementary Fig. 8). The mean missense:silent mutation ratio ing mechanism in medulloblastoma. was 3.6:1 and 40% of all missense mutations were predicted to be Although germline mutations in TP53, PTCH1, APC and CREBBP deleterious, suggesting a selective pressure for SNVs that affect protein predispose to medulloblastoma 11–14 , only 23 mutations previously coding (Supplementary Table 5). Global patterns of total SNVs associated with cancer were detected in discovery cohort germ lines. and amplifications varied significantly among medulloblastoma Only one of these—in a known case of Turcot’s syndrome—was subgroups, even when corrected for age and sex, supporting the notion accompanied by a somatic mutation (germline APC Y935*/somatic that these tumours are distinct pathological entities (Fig. 1 and deletion; WNT subgroup no. 11; Supplementary Table 8). Thus, Supplementary Fig. 6). Custom capture-based analysis of the allele inherited forms of medulloblastoma seem to be rare in our cohort. frequency of all somatic mutations in 12 medulloblastomas allowed us to predict the ancestry of certain genetic alterations, suggesting Novel mutations in medulloblastoma subgroups that aneuploidy precedes widespread sequence mutation in medullo- Because SVs and CNVs are unlikely to drive most medulloblastomas, blastomas with highly mutated genomes (Supplementary Figs 9–11). we investigated whether recurrent (more than two samples) somatic SNVs and/or indels might target discrete genes and pathways. This Novel CNVs and SVs are rare in medulloblastoma analysis identified 49 genes, across all 93 tumours, which were tar- The repertoire of focally amplified or deleted genes seems to be very geted by non-silent, recurrent, somatic mutations; 84% (n 5 41/49) limited in medulloblastoma. We detected expected gains of MYC, have not yet been implicated in medulloblastoma (Supplementary 2 MYCN and OTX2 in subgroups 3 and 4, but no novel recurrent Tables 9 and 10). Several of these congregated in disease subgroups amplifications (Fig. 1, Supplementary Fig. 12 and Supplementary and converged on specific cell pathways (Fig. 1, Supplementary Fig. 8 9 Table 7). In keeping with recent reports , high-level amplification of and Supplementary Table 11). MYCNin subgroup-3sampleno.16(samplenumberingas inFig.1)was generated by chromothripsis; although chromothripsis was observed Histone methylation is deregulated in subgroups 3 and 4 infrequently (n 5 2/37 of the discovery cohort; Supplementary Fig. 13). TheH3K27trimethylmark(H3K27me3)represseslineage-specificgenes 15 Focal homo- or heterozygous deletions of genes previously impli- in stem cells (Supplementary Fig. 8). H3K27me3 is written by the cated in medulloblastoma were also detected (for example, PTCH1, polycomb repressive complex 2 (PRC2) that includes the methylase PTEN; Fig. 1) 10,11 but novel recurrent focal deletions were rare. Three EZH2(refs16,17)andiserasedduringdifferentiationbythedemethylase 18 subgroup-4 tumours (nos 11–13) and one unclassified tumour KDM6A . As H3K27me3 is erased, chromatin remodellers recruited to deleted DDX31, AK8 and TSC1 at chromosome 9q34.14 in concert H3K4me3promotedifferentiation,forexample,CHD7(refs19,20).This with OTX2 amplification, suggesting that these alterations are coop- process is tightly controlled during development and deregulated in 21 erative (P , 0.0005, Fisher’s exact test). The breakpoint in this cancers; EZH2 is mutated in lymphomas and upregulated in breast 22 deletion occurs in DDX31, and two samples contained a missense and prostate 23 cancer, while biallelic inactivation of KDM6A (chro- mutation (subgroup 4, no. 15) and complex rearrangement mosome Xp11.2) or KDM6A and its paralogue UTY (chromosome 24 (unidentified case SJMB026) in this gene, suggesting that DDX31 is Yq11), occurs in adult female and male cancers, respectively . the target of these alterations (Supplementary Fig. 14). Hypergeometric distribution analyses revealed selective muta- Over 50% of SVs detected by WGS broke the coding region of at tion of histone modifiers in subgroup-3 and -4 medulloblastomas least one gene, but less than 2% (n 5 6/314, excluding two tumours (Supplementary Table 11). Six subgroup-4, one subgroup-3, and 44 | N A T U R E | V O L 488 | 2 A U GUST 201 2 ©2012 Macmillan Publishers Limited. All rights reserved

ARTICLE RESEARCH one unclassified medulloblastoma contained novel inactivating muta- subgroup-3 and -4 medulloblastomas retain a stem-like epigenetic tions in KDM6A (Figs 1 and 2 and Supplementary Figs 8 and 16). The state by aberrantly writing (EZH2 upregulation) or preserving single female with a KDM6A splice-site mutation showed a deletion of (KDM6A-UTY inactivation) H3K27me3, or disrupting H3K4me3 25 the second allele that escapes X inactivation (subgroup 4, no. 15), and associated transcription (CHD7 and ZMYM3 inactivation). Indeed, 57% (n 5 4/7) of KDM6A-mutant male medulloblastomas deleted human and mouse subgroup-3 and -4 medulloblastomas contained chromosome Y, compared with only 6% (n 5 3/51) of male, significantly more H3K27me3 than did WNT- or SHH-subgroup KDM6A wild-type tumours (P , 0.005, Fisher’s exact test; Fig. 1). tumours (Fig. 2b). Thus, gain of EZH2 and loss of KDM6A probably Thus, a two-hit model of KDM6A-UTY tumour suppression seems maintains H3K27me3 in subgroup-3 and -4 medulloblastomas. to operate in subgroup-4 medulloblastomas. Notably, mutations in Finally, we looked to see if the differential expression of H3K27me3 six other KDM family members (KDM1A, KDM3A, KDM4C, among medulloblastoma subgroups reflects ancestral chromatin KDM5A, KDM5B and KDM7A) were detected exclusively in marking in the progenitors that generate these tumours (Fig. 2b). subgroup-3 and -4 tumours, implicating broad disruption of lysine Relatively low levels of H3K27me3 were detected in LRLPs and com- demethylation in these medulloblastomas (Fig. 1, Supplementary mitted GNPs, which generate WNT- and SHH-subgroup medullo- Table 11 and Supplementary Fig. 16). blastomas, respectively , potentially explaining why mutations that 3–5 Subgroup-3and-4medulloblastomasalsogained andoverexpressed preserve this epigenetic mark are absent from these tumours. We EZH2 (chromosome 7q35-34), which writes H3K27me3, and recently showed that subgroup-3 medulloblastomas arise from a rare 6 contained novel inactivating mutations in effectors and regulators of fraction of cerebellar progenitors . We are currently investigating 26 the H3K4me3 mark (Fig. 2a and Supplementary Fig. 8). Gain of whether these progenitors are found among the H3K27me3-positive chromosome 7q was significantly enriched among subgroup-3 and cells seen in the external germinal layer (Fig. 2b). -4 medulloblastomas (P , 0.005, Fisher’s exact test) and correlated directly with EZH2 expression. Indeed, EZH2 was the eighth most Novel mutations in WNT-subgroup medulloblastomas significantly overexpressed gene on chromosome 7 among subgroup-3 WNT-subgroup medulloblastomas contained mutations in epigenetic and -4 medulloblastomas that gained chromosome 7q relative to those regulators that are different to those seen in subgroup-3 and -4 disease. with diploid chromosome 7 (P , 0.005, Bonferroni correction). CTNNB1, the principal effector of the WNT pathway, forms a Nonsense and frameshift mutations were detected in CHD7 in four transcription factor with the T-cell factor/lymphoid enhancer factor 28 subgroup-3 and -4 tumours. ZMYM3 (chromosome Xq13.1), which (TCF/LEF) . The carboxy terminus of CTNNB1 then recruits a series participates in a protein complex with KDM1A to regulate gene of protein complexes that remodel chromatin and promote transcrip- 27 expression at the H3K4me3 mark , was targeted by novel frameshift, tion at WNT-responsive genes (Supplementary Fig. 8). These include: nonsense and missense mutations in three male subgroup-4 medullo- histone acetyltransferases (for example, CREBBP and TRRAP–TIP60 blastomas. All three tumours with mutations in ZMYM3 also mutated complexes) 28,29 ; ATPases of the SWI/SNF family (for example, 30 KDM6A (subgroup 4, nos 19, 20) or KDM1A (subgroup 4, no. 21), SMARCA4) ; and the mediator complex that coordinates RNA 31 suggesting that these alterations are cooperative. Remarkably, polymerase II placement (for example, MED13) . As expected, KDM6A, CHD7 and ZMYM3 mutations were confined to subgroups .70% (n 5 8/11) of WNT-subgroup medulloblastomas contained 3 and 4, and clustered in samples with sub-median EZH2 expression mutations that stabilize CTNNB1 (Fig. 1 and Supplementary Fig. 8; levels (Fig. 2a; P , 0.05, Fisher’s exact test). These data suggest that P , 0.0001, Fisher’s exact test) 32,33 . A single subgroup-3 case (no. 5) a WNT SHH Subgroups 3 and 4 3 4 5 7 8 9 10 11 2 3 4 5 6 7 8 9 11 12 13 22 32 28 29 7 16* 3 8* 33 31 9 1 6 2 12* 5 10 14 37 14* 7* 17* 36 21 8 16 17 27 4 25 1* 24 11 35 12 18 23 6* 34 10* 30 19 15 2* 3* 15* 13* 13 20 38 26 9* 11* 4* 5* Chr 7 p q P < 0.005 EZH2 expr. † KDM6A P = 0.05 CHD7 ZMYM3 H3K27me3 P = 0.001 (×10 3 a.u.) 10.6 10.6 9.7 13.5 8.6 13.2 9.3 17.0 15.2 15.3 12.3 20.0 17.7 17.5 20.0 15.8 12.7 25.7 22.1 14.1 14.0 17.6 –4 0 4 Chr 7 copy number score Data N/A –2 0 2 Mutant EZH2 exp. log ratio Wild type WNT7 SHH7 GP4-7 b Mouse medulloblastoma Mouse E14.5 hindbrain P7 CB (i) (ii) H3K27me3 WNT (ii) (i) ventricle LRL URL Choroid SHH Subgroup 3 Cerebellum Fourth Brainstem IGL EGL Brainstem Figure 2 | Deregulation of H3K27me3 in subgroup-3 and -4 human and immunohistochemistry(numbersindicate colorimetry,PvalueANOVA).GP4-7 mouse medulloblastoma. a, Top row, SNP profiles of chromosome 7 (Chr 7) indicates case subgroup-4, no.7. a.u., arbitrary units. N/A, not available. copynumberinmedulloblastomas(samplesasFig.1;asteriskindicatessubgroup- b, H3K27me3 expression in mouse Blbp-Cre;Ctnnb1 1/lox(Ex3) ;Tp53 flx/flx (WNT- 3 cases). Second row, expression of EZH2. Subgroup-3 and -4 tumours are subgroup), Ptch1 1/2 ;Tp53 2/2 (SHH-subgroup) and Myc;Ink4c 2/2 (subgroup-3) ordered left to right by expression level, dagger indicatesmedian expression point medulloblastomas (right) and developing hindbrain (left). High-power views of (Bonferroni-corrected P value of EZH2 expression versus chromosome 7 gain). E14.5 LRL (i) and upper rhombic lip (URL) (ii). EGL,external germinal layer; Third row, mutation status of KDM6A, CHD7 and ZMYM3 (P value, Fisher’s IGL, internalgranule layer. Scale bar, 50mm. White arrows in P7 cerebellum (CB) exact test mutations versus EZH2 expression). Fourth row, H3K27me3 pinpoint H3K27me3 cells in the EGL. 2A U G U S T2 0 1 2|V O L 4 8 8 |N A T U R E|4 5 ©2012 Macmillan Publishers Limited. All rights reserved

RESEARCH ARTICLE also showed a mutation in CTNNB1, but this mutation has not ab shRNA TCF c 100- T T T T Phase Olig3 been reported in cancer, did not upregulate nuclear CTNNB1 Ctnnb1* N/A + T T (Fig. 1) and is of unclear relevance. Remarkably, six WNT-subgroup medulloblastomas showed mutations in chromatin modifiers that + 75- Wnt1 None are recruited to TCF/LEF WNT-responsive genes by CTNNB1 (Fig. 1 and Supplementary Fig. 8). Four WNT-subgroup tumours Vector – Percentage expression of controls real-time qPCR 50- *** contained heterozygous missense mutations in the helicase domain of *** T SMARCA4 (P , 0.002, Fisher’s exact test), two samples, including one DAPI + 0.37 ± 0.02 * 0.77 ± 0.09 25- T *** *** T *** with a SMARCA4 mutation (no. 5), contained nonsense mutations in Cdh1 ** T *** T T CREBBP (WNT-subgroup enrichment, P , 0.02, Fisher’s exact test), + and missensemutations in TRRAP and MED13 were detected in a single Merge Ddx3x 0.19 ± 0.08 0 WNT-subgroup medulloblastoma each. Thus, in addition to stabiliza- + Cdh1 shRNA Smarca4 shRNA Ddx3x shRNA Mll2 shRNA Gabrg1 shRNA Kdm6a shRNA tion ofCTNNB1,thedevelopmentofWNT-subgroupmedulloblastoma Smarca4 0.27 ± 0.05 may require disruption of chromatin remodelling at WNT-responsive d genes. Dorsal brainstem A small number of WNT-subgroup medulloblastomas lack muta- e tions in CTNNB1 or APC, suggesting that alternative mechanisms Control PGN Ctnnb1 mutant Cdh1 shRNA Ddx3x shRNA Ddx3x T275M Ddx3x G325E Gabrg1 shRNA Kdm6a shRNA drive aberrant WNT signals in these tumours. Three WNT-subgroup PGN Control Mll2 shRNA medulloblastomas in our series contained wild-type CTNNB1 (nos 1, 0 P1 10 and 11; Fig. 1). Sample no. 11 inactivated APC as the sole case of Dorsal brainstem Dorsal brainstem 0.1 Turcot’s syndrome in our study, but this tumour and sample no. 10 Dorsal brainstem also contained novel missense mutations in CDH1 (R63G, V329F; 0.2 WNT-subgroup enrichment, P , 0.05, Fisher’s exact test; Fig. 1). Ctnnb1 mutant 34 CDH1 sequesters CTNNB1 at the cell membrane , and mutations 0.3 that disrupt this interaction promote WNT signalling in adult P1 cancers 35,36 . The functional consequences of CDH1(R63G) and Dorsal brainstem 0.4 CDH1(V329F) remain to be determined, but their restriction to Dorsal brainstem WNT-subgroup tumours, mutual exclusivity with CTNNB1 muta- Relative distance travelled 0.5 tions, and adjacency to residues mutated in breast cancer (http:// Cdh1 shRNA www.sanger.ac.uk/genetics/CGP/cosmic/), suggest they might pro- 0.6 P1 mote aberrant WNT signals in medulloblastoma. We showed previously in mice that mutant Ctnnb1 initiates WNT- Dorsal brainstem 0.7 subgroup medulloblastoma by arresting the migration of LRLPs from 5 the embryonic dorsal brainstem to the pontine grey nucleus (PGN) . Ddx3x shRNA 0.8 Therefore, to test whether disruption of CDH1 might substitute for PGN mutant CTNNB1 in medulloblastoma, we used short hairpin (sh)RNAs P1 E16.5 0.9 to knockdown Cdh1 in embryonic day (E)14.5 mouse LRLPs (Fig. 3a–c). Deletion of Cdh1 expression upregulated Tcf/Lef-mediated gene tran- Median 1.0 1,907 84 261 – 1,662 1,468 1,662 1,203 1,552 distance (mm): scription in LRLPs and more than doubled their self-renewal capacity Ddx3x T275M P-cell distance: **** **** NS NS NS NS NS (Fig. 3b). Furthermore, in utero electroporation of LRLPs with Cdh1 P-cell number: NS NS **** ** ** NS NS NS shRNAs impeded their migration from the dorsal brainstem to the PGN PGN Labelled cells (%) P1 PGN with an efficiency similar to that of mutant Ctnnb1 (Fig. 3d, e; 0 10 >20 see Supplementary Methods). These data support the hypothesis that Figure 3 | Genes mutated in WNT-subgroup medulloblastomas regulate CDH1 suppresses the formation of WNT-subgroup medulloblastoma LRLPs. a, b, Isolated Olig3 /Wnt1 LRLPs were transduced in b with mutant 1 1 by regulating WNT-signals in LRLPs. Ctnnb1 (above hashed line) or the indicated shRNA-RFP (red fluorescence WNT-subgroup medulloblastomas were also enriched for novel, protein) construct (below hashed line). LRLPs were also transduced (1) or not recurrent somatic missense mutations in the DEAD-box RNA helicase (2) with a Tcf/Lef-enhanced green fluorescence (Tcf) reporter. Numbers on DDX3X at chromosome Xp11.3 (P , 0.0001, Fisher’s exact test; Fig. 1). right show clonal percentage 29 to 39 passage neurosphere formation DDX3X regulates several critical cell processes including chromosome (6 standard deviation (s.d.)). N/A, not applicable. Scale bar, 10 mm. 37 38 segregation , cell cycle progression , gene transcription and trans- c, Knockdown of genes targeted by shRNA relative to control transduced cells. 39 lation . Previously reported cancer-associated mutations in DDX3X Data show mean 6 s.d. d, Immunofluorescence of P1 mouse hindbrains disrupt the ATPase activity of the protein, but seven of eight muta- electroporated in utero at E14.5 with GFP (to control for equivalence of electroporation between embryos control) and the indicated construct. High- tions identified in our series clustered in the DEAD-box domain power views of indicated areas are shown right. Cells targeted by Ddx3x shRNA (Supplementary Information and Supplementary Fig. 8). Structural are present 48 h after electroporation but ablated by P1. Scale bars, 200 mm. 1 1 modelling predicts that these mutations interfere with nucleic acid e, Heatmap showing the distribution of GFP /RFP cells in eletroporated mice binding, possibly altering specificity and/or affinity for RNA substrates, at P1. Median distance migrated by cells and P values of migration distance and rather than inactivating DDX3X (Supplementary Figs 17–22). Indeed, cell number relative to controls is shown. ****P , 0.00005; ***P , 0.0005; ** 25 the wild-type alleleofDDX3Xthatescapes X inactivation wasretained P , 0.005; *P , 0.05. Red and green text reports significant increase or by two of three DDX3X-mutant female medulloblastomas, and knock- decrease, respectively, relative to control. down of Ddx3x halved the self-renewal rate of mouse LRLPs, suggest- ing that this protein is important for the proliferation and/or no.8)onLRLPs.Remarkably,althoughDdx3xshRNAswere expressed maintenance of the LRLP lineage (Fig. 3b). abundantly in E14.5 brainstem cells within 48 h of electropora- To understand better the role of DDX3X in WNT-subgroup tion, #0.5% of Ddx3x-shRNA-positive cells were present by postnatal medulloblastoma, we used our in utero migration assay to assess day (P)1, confirming the critical importance of this gene to maintain the impact of Ddx3x shRNAs, mutant Ddx3x T275M (identified in the LRLP lineage (Fig. 3d, e). In contrast, mice electroporated with WNT-subgroup sample no. 9), or mutant Ddx3x G325E (WNT sample either mutant Ddx3x T275M or Ddx3x G325E consistently contained 46 | N A T U R E | V O L 488 | 2 A U GUST 201 2 ©2012 Macmillan Publishers Limited. All rights reserved

ARTICLE RESEARCH ,50% more labelled cells at P1 than did controls, although these cells drives aberrant SHH signals in the remaining cases remains unclear. migrated normally (Fig. 3d, e and data not shown). Thus, mutations in These tumours contained mutations in MLL2, TP53 and PTEN that 42 DDX3X may contribute to WNT-subgroup medulloblastoma by have been reported previously in medulloblastoma ; but these muta- increasing LRLP proliferation rather than perturbing the migration tions occur in other subgroups and are not known to activate SHH of their daughter cells. Notably, comparable knockdown in utero of signals. Two SHH-subgroup tumours (nos 11 and 12) contained Mll2, Gabrg1 and Kdm6a that were selectively mutated in non-WNT identical novel T48M mutations in the GABA A (c-aminobutyric acid, medulloblastomas had no apparent impact on LRLPs; supporting the subtype A) receptor, c1, which is predicted to be deleterious (Fig. 1 value of our assay for assessing WNT-subgroup specific mutations and and Supplementary Table 9). Disruption of GABA A receptors can 43 underscoring the importance of cell context for functional studies of enhance neural stem cell proliferation , suggesting that these muta- genes mutated in cancer subgroups. tions might deregulate the proliferation of GNPs that generate SHH- subgroup medulloblastomas. PIK3CA mutations promote WNT-subgroup medulloblastoma Cancer-associated, activating mutations in PIK3CA were detected in Discussion a single case each of WNT-subgroup (PIK3CA(Q546K)), SHH- We have identified several, new, recurrent, somatic mutations in spe- subgroup (PIK3CA(H1047R)) and subgroup-4 (PIK3CA(N345K)) cific subgroups of medulloblastoma. Alterations affecting EZH2, medulloblastoma (Fig. 1 and Supplementary Fig. 23). Although KDM6A, CHD7 and ZMYM3 seem to disrupt chromatin marking 40 PIK3CA mutations are common in adult cancers and reported in of genes in subgroup-3 and -4 tumours. Further epigenetic studies medulloblastoma , their role in tumorigenesis remains controversial. will be required to uncover the identity of these genes, but evidence 41 In particular it is not known if these mutations initiate or progress suggests these may include OTX2, MYC and MYCN 44,45 . As amplifica- cancer. To test this, we generated mice that express a conditional tion of these genes was detected almost exclusively in subgroup-3 and allele of the Pik3ca E545K mutation. Mice harbouring Pik3ca E545K -4 tumours that lacked mutations in KDM6A, CHD7 or ZMYM3,itis or Pik3ca E545K and Tp53 flx/flx were bred with Blbp-Cre, which drives tempting to speculate that these genetic alterations target common 5 efficient recombination in LRLPs . Blbp-Cre;Pik3ca E545K mice, with transforming pathways. A recent study detected recurrent mutations or without Tp53 flx/flx , survived tumour free for a median of 212 in three other chromatin remodellers in medulloblastoma : 42 days with no evidence of aberrant LRLP migration (Fig. 4a and SMARCA4, MLL2 and MLL3, but this study did not include details data not shown). In stark contrast, 100% (n 5 11/11) of Blbp- of tumour subgroup. Here, we show that mutations in SMARCA4, Cre;Ctnnb1 1/lox(Ex3) ;Tp53 1/flx ;Pik3ca E545K mice developed WNT- CREBBP, TRRAP and MED13 are enriched in WNT-subgroup subgroup medulloblastomas by 3 months of age; only 4% (n 5 2/54) medulloblastomas; thereby uncovering potential cooperative muta- of Blbp-Cre;Ctnnb1 1/lox(Ex3) ;Tp53 1/flx mice develop WNT-subgroup tions in chromatin remodellers and their binding-partner oncogene, medulloblastoma by 11 months (Fig. 4a, b). Pik3ca wild-type and CTNNB1. Thus, disruptions in the epigenetic machinery of medullo- mutant mouse medulloblastomas displayed similar ‘classic’ histologies blastoma are likely to be subgroup specific and may cooperate with 1 and nuclear Ctnnb1 , but Pik3ca E545K mutant tumours contained other oncogenic mutations. The low incidence of MLL2 mutations greater AKT pathway activity as measured by pS6 and p4EBP1 detected in our study relative to previous work probably reflects 42 immunostaining. Thus mutations in PIK3CA probably activate differences in study populations (see Supplementary Results). the AKT pathway to progress, rather than initiate, WNT-subgroup Although medulloblastoma is more prevalent in males, especially medulloblastoma. with subgroup-3 and -4 disease , the reason for this sex bias is 46 unknown. One potential explanationis thelocationofmedulloblastoma SHH-subgroup medulloblastomas oncogenes or tumour suppressor genes on chromosome X .Three of 47 Four of thirteen SHH-subgroup medulloblastomas contained the most recurrently mutated genes detected in our study are located on expected biallelic inactivating alterations in SUFU or PTCH1. What chromosome X, of which two (ZMYM3 and KDM6A) were observed almost exclusively in males. Mutation in these genes might explain a some of the male sex bias in medulloblastoma. The third mutated X Pik3ca +/flx n =11 chromosome gene, DDX3X, is more likely to be a WNT-subgroup 100 Tumour-free survival (%) 75 Ctnnb1 +/lox(Ex3) ;Tp53 +/flx ;Pik3ca +/flx n = 11 carried heterozygous mutations in DDX3X that escape X inactivation , Pik3ca +/flx ;Tp53 +/flx n =15 medulloblastoma oncogene. Three of four female medulloblastomas Ctnnb1 +/lox(Ex3) ;Tp53 +/flx n = 54 25 Ctnnb1 +/lox(Ex3) ;Tp53 flx/flx n = 55 and our functional data indicate that mutations in this gene provide a 50 proliferative advantage to LRLPs that generate these tumours. Our findings also have important implications for drug develop- 25 P < 0.0001 maintain H3K27me3—for example, EZH2 methylase—may be useful 0 0 50 100 150 200 250 300 350 400 450 500 550 600 ment. Inhibitors of the epigenetic machinery, especially those that Time (days) treatments for subgroup-3 and -4 disease. These tumours include the b H&E Ctnnb1 pS6(Ser 235/236) p4EBP1(Thr 37/46) most aggressive forms of medulloblastoma, for which treatment options are limited. Mutations that activate PIK3CA and DDX3X in Ctnnb1 +/lox(Ex3) Tp53 +/flx ;Pik3ca +/flx ; WNT-subgrouptumoursmightalsobetargetedwithnoveltherapeutic 48,49 strategies . Future clinical trials of drugs that target these mutant proteins must recruit the appropriate patient populations, as we demonstrate that mutations show subgroup specificity in medullo- Ctnnb1 +/lox(Ex3) ; Tp53 +/flx blastoma. Our accurate mouse models of WNT-subgroup, SHH- subgroup and subgroup-3 medulloblastoma should help with future studies of the biological and therapeutic importance of the novel Figure 4 | Pik3ca E545K accelerates but does not initiate WNT-subgroup genetic alterations described in this study. medulloblastoma. a, Tumour-free survival of mice of the indicated genotype. METHODS SUMMARY All mice carry the Blbp-cre allele. Log rank P , 0.0001. b, Haematoxylin and Human tumour and matched blood samples were obtained with informed eosin (H&E) and immunohistochemical stains of indicated tumours. consent through an institutional review board approved protocol at St Jude Scale bar, 50 mm. Children’s Research Hospital. WGS and analysis of WGS data were performed 2 A UGUS T 2 012 | V OL 488 | NA TUR E | 4 7 ©2012 Macmillan Publishers Limited. All rights reserved

RESEARCH ARTICLE 50 as previously described . Details of sequence coverage, custom capture and other 28. Mosimann, C., Hausmann, G. & Basler, K. b-Catenin hits chromatin: regulation of validation procedures are provided in Supplementary Information (Supplemen- Wnt target gene activation. Nature Rev. Mol. Cell Biol. 10, 276–286 (2009). tary Tables 12–15). Immunohistochemistry and immunofluorescence of 29. Hecht, A., Vleminckx, K., Stemmler, M. P., van Roy, F. & Kemler, R. The p300/CBP acetyltransferases function as transcriptional coactivators of b-catenin in human and mouse tissues were performed using routine techniques and primary vertebrates. EMBO J. 19, 1839–1850 (2000). antibodies of the appropriate tissues as described (Supplementary Methods). 30. Barker, N. et al. The chromatin remodelling factor Brg-1 interacts with b-catenin to Medulloblastoma mRNA and DNA profiles were generated using Affymetrix promote target gene activation. EMBO J. 20, 4935–4943 (2001). U133v2 and SNP 6.0 arrays, respectively (Supplementary Methods). Real-time 31. Carrera, I., Janody, F., Leeds, N., Duveau, F. & Treisman, J. E. Pygopus activates PCR with reverse transcriptase (RT–PCR) analysis of genes targeted in mouse Wingless target gene transcription through the mediator complex subunits Med12 and Med13. Proc. Natl Acad. Sci. USA 105, 6644–6649 (2008). LRLPs by shRNAs were performed as described previously . LRLPs were isolated 32. Thompson, M. C. et al. Genomics identifies medulloblastoma subgroups that are 32 and transduced with indicated lentiviruses in stem cell cultures or targeted enriched for specific genetic alterations. J. Clin. Oncol. 24, 1924–1931 (2006). in utero with shRNAs or mutant cDNA sequences by electroporation as 33. Kool, M. et al. Integrated genomics identifies five medulloblastoma subtypes with 5 described (Supplementary Information). Mice harbouring a Cre-inducible distinct genetic profiles, pathway signatures and clinicopathological features. Pik3ca E545K allele were generated using homologous recombination: a lox- PLoS ONE 3, e3088 (2008). puro-STOP-lox cassette was introduced immediately upstream of the exon con- 34. Orsulic, S., Huber, O., Aberle, H., Arnold, S. & Kemler, R. E-cadherin binding prevents b-catenin nuclear localization and b-catenin/LEF-1-mediated taining the initiation codon, exon 9 was replaced with an exon containing the transactivation. J. Cell Sci. 112, 1237–1245 (1999). E545K mutation. Pik3ca E545K mice were bred with Blbp-Cre;Ctnnb1 lox(Ex3)/lox(Ex3) 35. Risinger, J.I., Berchuck,A., Kohler, M.F.& Boyd, J.Mutationsofthe E-cadheringene and Tp53 flx/flx mice to generate progeny of the appropriate genotype and sub- in human gynecologic cancers. Nature Genet. 7, 98–102 (1994). jected to clinical surveillance. 36. Becker, K.-F. et al. E-Cadherin gene mutations provide clues to diffuse type gastric carcinomas. Cancer Res. 54, 3845–3852 (1994). 37. Pek, J. W. & Kai, T. DEAD-box RNA helicase Belle/DDX3 and the RNA interference Received 13 January; accepted 2 May 2012. pathway promote mitotic chromosome segregation. Proc. Natl Acad. Sci. USA 108, Published online 20 June 2012. 12007–12012 (2011). 38. Lai,M.C.,Chang, W.C., Shieh,S.Y.& Tarn,W.Y.DDX3 regulatescell growththrough 1. Central BrainTumor Registry of the United States. Statistical report:primary brain translational control of cyclin E1. Mol. Cell. Biol. 30, 5444–5453 (2010). tumors in the United States, 1995–1999. https:// http://www.cbtrus.org/reports/ 39. Schro ¨der, M. Human DEAD-box protein 3 has multiple functions in gene 2002/2002report.pdf (CBTRUS, 2006). regulation and cell cycle control and is a prime target for viral manipulation. 2. Taylor, M. D. et al. Molecular subgroups of medulloblastoma: the current Biochem. Pharmacol. 79, 297–306 (2010). consensus. Acta Neuropathol. 123, 465–472 (2012). 40. Samuels, Y. et al. High frequency of mutations of the PIK3CA gene in human 3. Schu ¨ller, U. et al. Acquisition of granule neuron precursor identity is a critical cancers. Science 304, 554 (2004). determinant of progenitor cell competence to form Shh-induced 41. Broderick, D. K. et al. Mutations of PIK3CA in anaplastic oligodendrogliomas, high- medulloblastoma. Cancer Cell 14, 123–134 (2008). grade astrocytomas, and medulloblastomas. Cancer Res. 64, 5048–5050 (2004). 4. Yang, Z.J.et al. Medulloblastomacanbeinitiated bydeletion ofPatchedinlineage- 42. Parsons, D. W. et al. The genetic landscape of the childhood cancer restricted progenitors or stem cells. Cancer Cell 14, 135–145 (2008). medulloblastoma. Science 331, 435–439 (2011). 5. Gibson, P.etal. Subtypesofmedulloblastomahave distinctdevelopmental origins. 43. Anda ¨ng, M. et al. Histone H2AX-dependent GABA A receptor regulation of stem cell Nature 468, 1095–1099 (2010). proliferation. Nature 451, 460–464 (2008). 6. Kawauchi, D. et al. A mouse model of the most aggressive subgroup of human 44. Pasini, D. et al. Coordinated regulation of transcriptional repression by the RBP2 medulloblastoma. Cancer Cell 21, 168–180 (2012). H3K4 demethylase and Polycomb-Repressive Complex 2. Genes Dev. 22, 7. Mulhern, R. K. et al. Neurocognitive consequences of risk-adapted therapy for 1345–1355 (2008). childhood medulloblastoma. J. Clin. Oncol. 23, 5511–5519 (2005). 45. Khan, A., Shover, W.& Goodliffe,J.M.Su(z)2 antagonizesauto-repressionof Myc in 8. Wang, J. et al. CREST maps somatic structural variation in cancer genomes with Drosophila, increasing Myc levels and subsequent trans-activation. PLoS ONE 4, base-pair resolution. Nature Methods 8, 652–654 (2011). e5076 (2009). 9. Rausch, T. et al. Genome sequencing of pediatric medulloblastoma links 46. Northcott, P. A. et al. Medulloblastoma comprises four distinct molecular variants. catastrophic DNA rearrangements with TP53 mutations. Cell 148, 59–71 (2012). J. Clin. Oncol. 29, 1408–1414 (2011). 10. Castellino, R. C. et al. Heterozygosity for Pten promotes tumorigenesis in a mouse 47. Spatz, A., Borg, C. & Feunteun, J. X-chromosome genetics and human cancer. model of medulloblastoma. PLoS ONE 5, e10849 (2010). Nature Rev. Cancer 4, 617–629 (2004). 11. Hahn, H. et al. Mutations of the human homolog of Drosophila patched in the 48. Lindqvist, L.et al. Selective pharmacological targeting of a DEAD box RNAhelicase. nevoid basal cell carcinoma syndrome. Cell 85, 841–851 (1996). PLoS One 3, e1583 (2008). 12. Malkin, D. et al. Germ line p53 mutations in a familial syndrome of breast cancer, 49. Engelman, J. A. Targeting PI3K signalling in cancer: opportunities, challenges and sarcomas, and other neoplasms. Science 250, 1233–1238 (1990). limitations. Nature Rev. Cancer 9, 550–562 (2009). 13. Hamilton,S.R.et al.Themolecular basis ofTurcot’s syndrome.N.Engl. J.Med.332, 50. Zhang, J. et al. The genetic basis of early T-cell precursor acute lymphoblastic 839–847 (1995). leukaemia. Nature 481, 157–163 (2012). 14. Taylor, M. D. et al. Medulloblastoma in a child with Rubenstein-Taybi syndrome: case report and review of the literature. Pediatr. Neurosurg. 35, 235–238 (2001). Supplementary Information is linked to the online version of the paper at 15. Mikkelsen, T. S. et al. Genome-wide maps of chromatin state in pluripotent and www.nature.com/nature. lineage-committed cells. Nature 448, 553–560 (2007). 16. Cao,R.et al.RoleofhistoneH3lysine27methylationinPolycomb-groupsilencing. Acknowledgements This research was supported as part of the St Jude Children’s Science 298, 1039–1043 (2002). Research Hospital, Washington University Pediatric Cancer Genome Project. This work 17. Czermin, B. et al. Drosophila Enhancer of Zeste/ESC complexes have a histone H3 was supported by grants from the National Institutes of Health (R01CA129541, methyltransferase activity that marks chromosomal Polycomb sites. Cell 111, P01CA96832 and P30CA021765; R.J.G.), the Collaborative Ependymoma Research Network (CERN), Musicians against Childhood Cancer (MACC), The Noyes Brain 185–196 (2002). 18. Agger, K. et al. UTX and JMJD3 are histone H3K27 demethylases involved in HOX Tumour Foundation, and by the American Lebanese Syrian Associated Charities (ALSAC).Weare gratefultoS.Templeforthe giftofreagents and thestaffofthe Hartwell gene regulation and development. Nature 449, 731–734 (2007). 19. Schnetz, M. P. et al. Genomic distribution of CHD7 on chromatin tracks H3K4 Center for Bioinformatics and Biotechnology and ARC at St Jude Children’s Research methylation patterns. Genome Res. 19, 590–601 (2009). Hospital for technical assistance. 20. Sauvageau, M. & Sauvageau, G. Polycomb group proteins: multi-faceted Author Contributions G.R., M.P., T.A.K., C.L., X.C., L.D., T.N.P., E.H., L.W., X.Z., N.Ch., R.H., regulators of somatic stem cells and cancer. Cell Stem Cell 7, 299–313 (2010). N.Cu., R.T., J.W., G.W., M.R., X.H., J.B., P.G., J.M., J.E., B.V., A.O.-T., T.L., S.Po., S.Pa., D.Z., D.K. 21. Morin, R. D. et al. Somatic mutations altering EZH2 (Tyr641) in follicular and and D.F.contributed tothe designandconduct ofexperiments and to the writing.S.J.B., diffuse large B-cell lymphomas of germinal-center origin. Nature Genet. 42, R.K., M.F.R., R.S.F., L.L.F., D.J.D., K.O. and E.R.M. contributed to experimental design and 181–185 (2010). to the writing. A.G., D.W.E., C.C.L., E.B., T.H., S.G. and R.C. provided clinical expertise. 22. Kleer, C. G. et al. EZH2 is a marker of aggressive breast cancer and promotes R.K.W., J.R.D., J.Z. and R.J.G. conceived the research and contributed to the design, neoplastic transformation of breast epithelial cells. Proc. Natl Acad. Sci. USA 100, direction and reporting of the study. 11606–11611 (2003). 23. Varambally, S. et al. The polycomb group protein EZH2 is involved in progression Author Information Sequence and SNP array data were deposited in dbGaP under of prostate cancer. Nature 419, 624–629 (2002). accession number phs000409 and in the Sequence Read Archive (SRA) under 24. van Haaften, G. et al. Somatic mutations of the histone H3K27 demethylase gene accession number SRP008292. Reprints and permissions information is available at UTX in human cancer. Nature Genet. 41, 521–523 (2009). www.nature.com/reprints. This paper is distributed under the terms of the Creative 25. Yang, F., Babak, T., Shendure, J. & Disteche, C. M. Global survey of escape from X Commons Attributions-Non-Commercial-Share Alike licence, and is freely available to inactivation by RNA-sequencing in mouse. Genome Res. 20, 614–622 (2010). all readers at www.nature.com/nature. The authors declare no competing financial 26. Christensen, J. et al. RBP2 belongs to a family of demethylases, specific for tri- and interests. Readers are welcome to comment on the online version of this article at dimethylated lysine 4 on histone 3. Cell 128, 1063–1076 (2007). www.nature.com/nature. Correspondence and requests for materials should be 27. Lee, M. G., Wynder, C., Cooch, N. & Shiekhattar, R. An essential role for CoREST in addressed to R.J.G. ([email protected]) or J.Z. nucleosomal histone 3 lysine 4 demethylation. Nature 437, 432–435 (2005). ([email protected]). 48 | N A T U R E | V O L 488 | 2 A U GUST 201 2 ©2012 Macmillan Publishers Limited. All rights reserved

ARTICLE doi:10.1038/nature11327 Subgroup-specific structural variation across 1,000 medulloblastoma genomes A list of authors and their affiliations appears at the end of the paper Medulloblastoma, the most common malignant paediatric brain tumour, is currently treated with nonspecific cytotoxic therapies including surgery, whole-brain radiation, and aggressive chemotherapy. As medulloblastoma exhibits marked intertumoural heterogeneity, with at least four distinct molecular variants, previous attempts to identify targets for therapy have been underpowered because of small samples sizes. Here we report somatic copy number aberrations (SCNAs) in 1,087 unique medulloblastomas. SCNAs are common in medulloblastoma, and are predominantly subgroup-enriched. The most common region of focal copy number gain is a tandem duplication of SNCAIP, a gene associated with Parkinson’s disease, which is exquisitely restricted to Group 4a. Recurrent translocations of PVT1, including PVT1-MYC and PVT1-NDRG1, that arise through chromothripsis are restricted to Group 3. Numerous targetable SCNAs, including recurrent events targeting TGF-b signalling in Group 3, and NF-kB signalling in Group 4, suggest future avenues for rational, targeted therapy. Brain tumours are the most common cause of childhood oncological Twenty-eight regions of recurrent high-level amplification (copy death, and medulloblastoma is the most common malignant paediatric number $ 5) were identified (Fig. 1d and Supplementary Table 7). brain tumour. Current medulloblastoma therapy including surgical The most prevalent amplifications affected members of the MYC resection, whole-brain and spinal cord radiation, and aggressive family with MYCN predominantly amplified in SHH and Group 4, chemotherapy supplemented by bone marrow transplant yields five- MYC in Group 3, and MYCL1 in SHH medulloblastomas. Multiple 1 year survival rates of 60–70% . Survivors are often left with significant genes/regions were exclusively amplified in SHH, including GLI2, neurological, intellectual and physical disabilities secondary to the MYCL1, PPM1D, YAP1 and MDM4 (Fig. 1d). Recurrent homozygous 2 effects ofthese nonspecific cytotoxic therapies onthe developing brain . deletions were exceedingly rare, with only 15 detected across 1,087 Recent evidence suggests that medulloblastoma actually comprises tumours (Fig. 1e). Homozygous deletions targeting known tumour multiple molecularly distinct entities whose clinical and genetic dif- suppressors PTEN, PTCH1 and CDKN2A/B were the most common, 3–6 ferences may require separate therapeutic strategies . Four principal all enriched in SHH cases (Fig. 1e and Supplementary Table 7). Novel subgroups of medulloblastoma have been identified: WNT, SHH, homozygousdeletionsincludedKDM6A, ahistone-lysinedemethylase Group 3 and Group 4 (ref. 7), and there is preliminary evidence for deleted in Group 4. A custom nanoString CodeSetwas usedto verify24 clinically significant subdivisions of the subgroups 3,7,8 . Rational, significant regions of gain across 192 MAGIC cases, resulting in a targeted therapies based on genetics are not currently in use for verification rate of 90.9% (Supplementary Fig. 5). We conclude that medulloblastoma, although inhibitors of the Sonic Hedgehog SCNAs in medulloblastoma are common, and are predominantly pathway protein Smoothened have shown early promise . Actionable subgroup-enriched. 9 targets for WNT, Group 3 and Group 4 tumours have not been identified 4,10 . Sanger sequencing of 22 medulloblastoma exomes Subgroup-specific SCNAs in medulloblastoma revealed on average only 8 single nucleotide variants (SNVs) per WNT medulloblastoma genomes are impoverished of recurrent focal 11 tumour . Some SNVs were subgroup-restricted (PTCH1, CTNNB1), regions of SCNA, exhibiting no significant regions of deletion and whereas others occurred across subgroups (TP53, MLL2). We pro- only a small subset of focal gains found at comparable frequencies in posed that the observed intertumoural heterogeneity might have non-WNT tumours (Supplementary Figs 4, 6 and Supplementary underpowered prior attempts to discover targets for rational therapy. Table 8). CTNNB1 mutational screening confirmed canonical exon The Medulloblastoma Advanced Genomics International 3mutationsin63outof71(88.7%)WNTtumours,whereasmonosomy Consortium (MAGIC) consisting of scientists and physicians from 6 was detected in 58 out of 76 (76.3%) (Supplementary Fig. 6; 46 cities across the globe gathered more than 1,200 medulloblastomas Supplementary Table 9). Four WNT tumours (4/71; 5.6%) had neither which were studied by SNP arrays (n 5 1,239; Fig. 1a, Supplementary CTNNB1 mutation nor monosomy 6, but maintained typical WNT Fig. 1 and Supplementary Tables 1–3). Medulloblastoma subgroup expression signatures. Given the size of our cohort and the resolution affiliation of 827 cases was determined using a custom nanoString- of the platform, we conclude that there are no frequent, targetable 12 based RNA assay (Supplementary Fig. 2) . Disparate patterns of SCNAs for WNT medulloblastoma. broad cytogenetic gain and loss were observed across the subgroups SHH tumours exhibit multiple significant focal SCNAs (Fig. 2a, (Fig. 1b and Supplementary Figs 3, 7, 8, 10 and 11). Analysis of the Supplementary Figs 12, 15, 16 and Supplementary Tables 10 and 11). entire cohort using GISTIC2 (ref. 13) to discover significant ‘driver’ SHH enriched/restricted SCNAs included amplification of GLI2 and 10 events delineated 62 regions of recurrent SCNA (Fig. 1c, deletion of PTCH1 (Fig. 2a, e, f) . MYCN and CCND2 were among Supplementary Fig. 4 and Supplementary Tables 4 and 5); analysis the most frequently amplified genes in SHH (Supplementary by subgroup increased sensitivity such that 110 candidate ‘driver’ Table 6), but were also altered in non-SHH cases. Genes upregulated SCNAs were identified, most of which are subgroup-enriched in SHH tumours (that is, SHH signature genes) are significantly over- (Fig. 1c–e and Supplementary Table 6). represented among the genes focally amplified in SHH tumours 2A U G U S T 2 0 1 2|V O L4 8 8 |N A T U R E | 4 9 ©2012 Macmillan Publishers Limited. All rights reserved

RESEARCH ARTICLE Pan-cohort Analysis by SHH WNT Group 4 Group 3 abc Group 4 Group 3 WNT SHH Loss frequency (n = 76) (n = 266) (n = 168) (n = 317) 1.0 0.5 0 1.0 0.5 0 1.0 0.5 0 1.0 0.5 0 analysis subgroup 1 1 2 100 2 3 4 3 5 80 4 5 6 7 8 Number of events 60 Chromosome 6 7 8 9 Chromosome arm (p,q) 10 9 40 11 12 20 13 15 1 1 10 14 0 16 12 17 13 18 P = 0.02402 14 19 Subgroup 15 20 16 WNT 17 21 SHH 18 22 19 Group 3 20 21 Group 4 22 01.00.50 0 0.5 1.0 0 0.5 1.0 0.5 1.0 Log 2 copy number ratio Non-specifc Gain frequency –1.0 0.0 1.0 Gain Signifcant gain Deletion Signifcant deletion (CN 1.0) (CN 2.0) (CN 4.0) d Recurrent high-level amplifcations Freq. e Recurrent homozygous deletions Freq. MYCN (2) (3.5%) PTEN (5) (0.7%) MYC (1) (2.9%) GLI2 (1) (1.3%) PTCH1 (5) CDK6 (3) ACVR2B (4) (1.1%) CDKN2A/B (8) (0.6%) MYCL1 (6) (0.6%) PPM1D (3) KDM6A (N/A) CCND2 (2) (0.5%) (0.3%) YAP1 (25) CD99 (N/A) ACVR2A (1) (0.4%) PPP2R3B (N/A) PTBP1 (11) SMC1A (N/A) EHMT1 (N/A) TNFRSF13B (2) MDM4 (1) (0.3%) IKBKAP (N/A) KIT (1) CDK4 (13) TSC1 (4) SMPX (N/A) PIM1 (N/A) RBMS3 (N/A) GPR87 (N/A) Subgroup Subgroup (0.2%) IGF1R (6) SOX12 (31) LMO4 (N/A) WNT WNT MAP2K4 (N/A) SHH TP53 (19) SHH MIR17HG (N/A) Group 3 (0.2%) DUSP21 (N/A) Group 3 TGFBR1 (4) PHF2 (N/A) Group 4 Group 4 KAT6A (N/A) Non-specifc NLGN4X (N/A) Non-specifc DLC1 (119) MAP2K4 (N/A) ZIK1 (N/A) 0 5 10 15 20 25 30 35 40 0123456789 No. of cases No. of cases Figure 1 | Genomic heterogeneity of medulloblastoma subgroups. a,The (d, segmented copy number (CN) $ 5) and homozygous deletions medulloblastoma genomeclassified by subgroup.b, Frequency andsignificance (e, segmentedCN # 0.7) inmedulloblastoma.Thenumberofgenes mapping to (Q value # 0.1) of broadcytogenetic events acrossmedulloblastomasubgroups. the GISTIC2 peak region (where applicable) is listed in brackets after the c, Significant regions of focal SCNA identified by GISTIC2 in either pan-cohort suspected driver gene, as is the frequency of each event. or subgroup-specific analyses. d, e, Recurrent high-level amplifications a SHH MYCL1 [6] bc 11 e LMO4 [9] Chr17 MB-7 MB-1040 MB-875 MB-698 MB-486 MB-643 MB-198 MB-391 MB-1129 MB-1249 MB-1236 MB-1133 MB-488 MB-749 MB-1276 Chr1 MB-413 MB-216 MB-1233 MB-539 MB-184 MB-749 MB-862 MB-1106 MB-1037 MB-903 MB-698 MB-347 MB-14 MB-1094 MB-103 MB-1267 MB-434 MB-488 MB-13 MB-875 SHH signalling: 18.0% (P = 0.0009) RTK/PI3K signalling: 10.0% (P = 0.0139) 1q21.3[11] MDM4 [14] Mb Mb 2p23.3[28] MYCN [2] PTCH1 PPP2R2B PDGFRA IGF1R KIT 5.4% (7.7%) 0.4% (1.5%) 1.5% (4.9%) 1.9% (4.1%) 1.1% (4.9%) 4p16.3[126] GLI2 [1] ACVR2B [18] ELF2 [3] GLI1 GLI2 3q25.1[8] 0.8% (0.8%) MYCN 5.3% (14%) 5q35.3[105] 8.3% (18%) IRS2 KIT [2] 7p22.3[82] PPP2R2B [1] 55.0–57.0 1.9% (7.5%) 7q11.23[83] 6p21.2[2] PPM1D 201.0–202.0 PIK3C2B/MDM4 miR-17/92 CDKN2A/B 8p12[3] 1.1% (7.5%) CCND2 1.5% (8.3%) 6q24.3[1] 3.4% (3.8%) PIK3C2G 9p23[3] CDK6 [7] 1.1% (1.5%) CDKN2A/B [18] 8q22.1[2] TP53 signalling: 9.4% (P = 0.0012) PTEN1 PIK3C2B PTCH1 [5] YAP1 [1] CDKN2A/B 1.5% (7.5%) 1.5% (8.3%) YAP1 KLF4 [10] 12p13.33[35] 1.5% (8.3%) 1.5% (3.4%) 10p14[4] CCND2 [2] 10p11.23[4] 12p12.1[4] MDM1 MDM4 AKT3 PTEN [5] 12q15[1] PPM1D PIK3C2B/MDM4 0.8% (0.8%) 1.5% (7.5%) 0.8% (6.8%) 15q15.1[24] IRS2 [3] d TP53 TP53 [4] IGF1R [6] 3.0% (26%) TSC1 % of cases >3.0 >3.0 17p11.2[26] 16p13.2[223] KLF4 PPM1D 1.1% (36%) Deleted 0 Amplifed 17p11.2[69] 1.1% (36%) 2.3% (6.0%) 20p13[109] PPM1D [3] g 10 –10 10 –6 10 –4 10 –2 0.25 10 –2 10 –5 10 –10 10 –20 10 –50 21q22.11[2] MDM4 MYCN amplifcation GLI2 amplifcation PTCH1 focal deletion FDR Q value 1.0 1.0 1.0 | f MYCN 0.6 0.6 GLI2 OS probability 0.8 OS probability 0.8 b 0.6 ELF2 0.4 OS probability 0.8 PTCH1 0.4 p 0.4 CCND2 0.2 MYCN 0.2 0.2 GLI2 TP53 Non-MYCN Non-GLI2 PTCH1 Non-PTCH1 0.0 P = 0.0022 0.0 P = 1.5 × 10 –5 P = 0.22 Coverage: 23% Exclusivity: 79% 0.0 –3 –2 –1 0 1 2 3 01224364860 01224364860 01224364860 Time in months Time in months Time in months Copy number state Figure 2 | Genomic alterations affect core signalling pathways in SHH RTK/PI3K signalling represent the core pathways genomically targeted in medulloblastoma. a, GISTIC2 significance plot of amplifications (red) and SHH. P values indicate the prevalence with which the respective pathway is deletions (blue) observed in SHH. The number of genes mapping to each targeted in SHH versus non-SHH cases (Fisher’s exact test). Frequencies of significant region are included in brackets and regions enriched in SHH are focaland broad(parentheses) SCNAs arelisted.f, Mutual exclusivityanalysis of shaded red. b, c, Recurrent amplifications of PPM1D (b) and PIK3C2B/MDM4 focal SCNAs in SHH. g, Clinical implications of SCNAs affecting MYCN, GLI2 (c) are restricted to SHH. d, Fluorescence in situ hybridization (FISH) or PTCH1 in SHH (log-rank tests). validation of MDM4 amplification. e, SHH signalling, TP53 signalling and 5 0 | N A T U R E | VO L 4 8 8 | 2 AU G U S T 2 0 1 2 ©2012 Macmillan Publishers Limited. All rights reserved

ARTICLE RESEARCH (P 5 0.001–0.02, permutation tests; Supplementary Fig. 9). Recurrent medulloblastoma oncogene OTX2 is a prominent target of TGF-b amplification of SHH signature genes has clinical implications, as signalling in the developing nervous system and TGF-b pathway 17 amplification of downstream transcriptional targets could mediate inhibitors CD109 (ref. 18), FKBP1A (refs 19 and 20) and SNX6 (ref. 14 resistance to upstream SHH pathway inhibitors . 20) are recurrently deleted in Group 3 (Fig. 3a, d). SCNAs in TGF-b 25 Novel, SHH-enriched SCNAs included components of TP53 sig- pathway genes were heavily enriched in Group 3 (P 5 5.37 3 10 , nalling, including amplifications of MDM4 and PPM1D, and focal Fisher’s exact test) and found in at least 20.2% of cases, indicating that deletions of TP53 (Fig. 2a–e). Targetable events, including amplifica- TGF-b signalling represents the first rational target for this poor tions of IGF signalling genes IGF1R and IRS2, PI3K genes PIK3C2G prognosis subgroup (Fig. 3d). Similarly, novel deletions affecting and PIK3C2B, and deletion of PTEN were restricted to SHH tumours regulators of the NF-kB pathway, including NFKBIA (ref. 21) and (Fig. 2a, c, e). Importantly, focal events affecting genes in the SHH USP4 (ref. 22) were identified in Group 4 (Supplementary Fig. 23), pathway were largely mutually exclusive and prognostically signifi- proposing that NF-kB signalling may represent a rational Group 4 cant (Fig. 2f, g). Many of the recurrent, targetable SCNAs identified in therapeutic target. SHH medulloblastoma (IGF1R, KIT, MDM4, PDGFRA, PIK3C2G, Network analysis of Group 3 and Group 4 SCNAs illustrates the PIK2C2B and PTEN) have already been targeted with small molecules different pathways over-represented in each subgroup. Only TGF-b for treatment of other malignancies, which might allow rapid trans- signalling is unique to Group 3 (Fig. 3e). In contrast, cell-cycle control, lation for targeted therapy of subsets of SHH patients (Supplementary chromatin modification and neuronal development are all Group Table 16). Novel SHH targets identified here are excellent candidates 4-enriched. Cumulatively, the dismal prognosis of Group 3 patients, for combinatorial therapy with Smoothened inhibitors, to avoid the the lack of published targets for rational therapy, and the prior resistance encountered in both humans and mice 9,14,15 . targeting of TGF-b signalling in other diseases suggest that TGF-b Group 3 and Group 4 medulloblastomas have generic names as may represent an appealing target for Group 3 rational therapies comparatively little is known about their genetic basis, and no targets (Supplementary Table 16). 7 for rational therapy have been identified . MYC amplicons are largely restricted to Group 3, whereas MYCN amplicons are seen in Group 4 SNCAIP tandem duplication is common in Group 4 and SHH tumours (Fig. 1d) . Indeed, MYC and MYCN loci comprise Although Group 4 is the most prevalent medulloblastoma subgroup, 3,4 the most significant regions of amplification observed in Group 3 and its pathogenesis remains poorly understood. The most frequent Group 4, respectively (Fig. 3a, b, Supplementary Figs 13, 14, 17–20 SCNA observed in Group 4 (33/317; 10.4%) is a recurrent region of and Supplementary Tables 12–15). Group 3 MYC amplicons were single copy gain on chr5q23.2 targeting a single gene, SNCAIP mutually exclusive from those affecting the known medulloblastoma (synuclein, alpha interacting protein) (Fig. 4a and Supplementary oncogene OTX2 (ref. 16) and were highly prognostic (Supplementary Fig. 24). SNCAIP, encodes synphilin-1, which binds to a-synuclein Fig. 21) 3,16 . Type II activin receptors, ACVR2A and ACVR2B and to promote the formation of Lewy bodies in the brains of patients with family member TGFBR1 are highly amplified in Group 3 tumours, Parkinson’s disease 23,24 . Additionally, rare germline mutations of 25 indicating deregulation of TGF-b signalling as a driver event in Group SNCAIP have been described in Parkinson’s families . Large insert, 3 (Fig. 3c–e and Supplementary Fig. 22). The Group 3-enriched mate-pair, whole-genome sequencing (WGS) demonstrates that 1q41[1] 2p25.3[1] a 3q23[3] Group 3 MCL1 [7] b Group 4 2p24.3[2] ESRRG [1] MYCN [2] 4p16.1[126] MYCN [2] 2p24.3-p24.1[7] ELF2 [3] ACVR2B [4] 4q31.3[3] 2p24.1[1] 2q37.3[81] CTNNB1 [1] ACVR2A [3] 3p21.31[7] 5q13.2[27] 5q14.3[1] 5q35.3[43] ACVR2B [6] 5q13.2[27] SNCAIP [1] 6q24.3[1] 5q35.3[106] 6q24.3[1] 6q13[16] 7p22.3[40] CDK6 [13] 7p22.3[3] 7p21.3-p11.1[9] 8q21.11[1] 8p12[3] 8p12[3] 7q21.11[1] E2F5 [3] CDK6 [3] DDX31 [2] 8q24.12-q24.13[2] 9p13.2[4] 7q31.1-q33[34] 10p14[4] 10p12.33[5] MYC/PVT1 [6] 8p11.21-p11.22[27] 10p11.23[4] 8q24.21-q24.23[8] 10p11.23[4] 8q23.1[1] 11p15.4[6] 10q11.21[3] TGFBR1 [3] MYC [1] 11q22.3[7] 12p11.21[3] 9p24.2[1] 12p13.33[5] 13q12.2[4] OTX2 [2] RB1 [3] 10p15.3[2] 14q13.1[8] SNORD113-1/7 [7] NFKBIA [14] CCND2 [13] CDH13 [3] SMYD4 [3] 12p12.3[8] [] 16q23.3- 22q12.2[3] OTX2 [3] q24.1[102] 22q13.31[103] 17q12[12] HIC1 [42] 10 –10 10 –6 10 –4 10 –2 0.25 10 –3 10 –7 10 –10 10 –30 10 –60 10 –20 10 –9 10 –5 10 –3 0.25 10 –3 10 –6 10 –10 10 –20 10 –50 BRCA1 [1] MAP2K4 [8] FDR Q value FDR Q value 17q22[12] KCNJ2 [3] 20p13[37] 18q23[1] c e Microtubules and 19p13.3[11] 21q22.11[2] Chr2 Mb MB-1340 MB-717 MB-582 MB-423 MB-1036 MB-1351 MB-586 MB-1309 MB-329 MB-359 MB-794 Chr3 Mb MB-763 MB-1063 MB-1124 MB-101 MB-887 MB-1278 MB-761 Chr9 Mb MB-1 199 MB-1023 MB-1016 MB-359 d cytoskeleton Response TGF-β signalling: 20.2% (P = 5.3 × 10 –5 ) processes CD109 Cell cycle to stimuli 5.4% (7.7%) Cancer TGF-β signalling pathways TGFBR1 ACVR2A ACVR2B 2.4% (7.7%) 1.2% (4.2%) 2.4% (3.6%) Mesenchymal proliferation TGFBR1 SNX6 Chromatin Cell 100.8–101.2 FKBP1A 4.2% (14%) signalling modifcation ACVR2B 38.3–38.7 3.0% (21%) SMAD2/4 ACVR2A Androgen 148.0–149.0 OTX2 receptor 7.7% (20%) Group 3 signalling Group 4 Cellular % of cases Digestive >3.0 0 >3.0 Neuronal Muscle tract ACVR2A ACVR2B TGFBR1 Deleted Amplifed Development Figure 3 | The genomic landscape of Group 3 and Group 4 ACVR2B) and type I (TGFBR1) activin receptors in Group 3. d, Recurrent 25 medulloblastoma. a, b, GISTIC2plots depicting significant SCNAsin Group 3 SCNAs affecting the TGF-b pathway in Group 3 (P 5 5.733 10 , Fisher’s (a) and Group 4 (b) with subgroup-enriched regions shaded in yellow and exact test). Frequencies of focal and broad (parentheses) SCNAs are listed. green, respectively. c, Recurrent amplifications targeting type II (ACVR2A and e, Enrichment map of gene sets affected by SCNAs in Group 3 versus Group 4. 2 A UGUS T 2 012 | V OL 488 | N A TU RE | 5 1 ©2012 Macmillan Publishers Limited. All rights reserved

RESEARCH ARTICLE a Mb Group 4 medulloblastoma (n = 317) b MB-374: tandem duplication of 315,249 bp 121996051 Reads in tumour c MB-1350 SNP6 profle SNCAIP Reads in tumour 121680803 121.0 –1 1 0 10 discordant read-pairs Paired-end read Copy number 4 3 2 1 Chr5 121.2 Log 2 ratio (read depth tumour vs control) 1 MB-398: tandem duplication of ~223,850 bp 0 Genomic location (Mb) 121.7 121.9 q23.2 q23.1 ~121689850 ~121913700 4 discordant 100 kb read-pairs 121.4 SNCAIP –1 0 1 0 MB-1350: Tandem duplication of 193,069 bp Scale 121650000 SNCAIP 121750000 RefSeq genes SNCAIP′ 121800000 121850000 chr5: 121850000 121800000 121750000 121700000 121900099 121707031 Duplicated SNCAIP allele 8 discordant Chromosome 5 121.6 121.8 –1 121400 kb LOC100505841 SNCAIP 122000 kb SNX2 SNX24 Germline SNCAIP allele Group 4 medulloblastoma (n =188) read-pairs f 1 1 LOX ZNF474 10 122200 kb 121800 kb 121600 kb 9 Genomic location on chromosome 5 Northcott series (n = 103) 122.0 d 12 P = 5.713 × 10 –16 e Cophenetic coeffcient = 0.9956 Log 2 SNCAIP expression 8 7 6 Group 4α 5 122.2 Log 2 SNCAIP expression 10 8 SNCAIP duplication: g Subtype: P = 9.314 × 10 –14 Group 4β Group 4α 21/79 Group 4α medulloblastoma (n =79) 122.4 6 11.0 10.5 122.6 WNT SHH Group 3 Group 4 SNCAIP duplication: Log 2 SNCAIP expression 10.0 Group 4β 9.5 Group 4-high Subgroup Value 1/109 Log 2 copy number ratio 10 –15 9.0 –10 –1.5 0.0 1.5 t statistics –10 0 –5 0 5 8.5 P = 0.02502 (CN 0.7) (CN 2.0) (CN 5.7) Genes Group 4-low 10 SNCAIP 15 SNCAIP: Duplicated Non-duplicated Figure 4 | Tandem duplication of SNCAIP defines a novel subtype of Group Group 4. e, NMF consensus clustering of 188 expression-profiled Group 4 4. a, Highly recurrent, focal, single-copy gain of SNCAIP in Group 4. b, Paired- tumours supports two transcriptionally distinct subtypes designated 4a and 4b end mapping verifies recurrent tandem duplication of SNCAIP in Group 4. (cophenetic coefficient 5 0.9956). 21 out of 22 SNCAIP duplicated cases belong 28 c, Schematic representation of SNCAIP tandem duplication. d, SNCAIP is a to Group 4a (P 5 3.123 10 , Fisher’s exact test). f, SNCAIP expression is Group 4 signature gene.Upper panel, SNCAIP expression acrosssubgroupsin a significantly elevated in Group 4a versus 4b (P 5 9.313 10 214 , Mann– published series of 103 primary medulloblastomas. Error bars depict the Whitney test). g, Group 4a cases harbouring SNCAIP duplication exhibit a minimum and maximum values, excluding outliers. Lower panel, SNCAIP ,1.5-fold increase in SNCAIP expression. f, g, Error bars depict the minimum ranks among the top 1% (rank, 39th out of 16,758) of highly expressed genes in and maximum values, excluding outliers. SNCAIP copy number gains arise from tandem duplication of a Group 3 tumours (n 5 13) identified two independent gene fusions in truncated SNCAIP (lacking non-coding exon 1), inserted telomeric two different tumours (MB-182 and MB-586), both involving the 59 to the germline SNCAIP allele (Fig. 4b, c and Supplementary Fig. 25). end of PVT1, a non-coding gene frequently co-amplified with MYC in Affymetrix SNP6 array profiling of patient-matched germline material Group 3 (Fig. 5a, b, Supplementary Fig. 37 and Supplementary Tables confirmed that SNCAIP duplications are somatic (Supplementary Fig. 17 and 18). Sanger sequencing confirmed a fusion transcript consisting 26), and subsequent whole-transcriptome sequencing (RNA-Seq) of of exons1 and3 ofPVT1 fused tothecoding sequence of MYC(exons2 select Group 4 cases (n5 5) verified that SNCAIP is the only gene and 3) in MB-182, and a fusion involving PVT1 exon 1 fused to the 39 expressed in the duplicated region (Supplementary Fig. 27). Analysis end of NDRG1 in MB-586 (Fig. 5a, b). 26 of published copy number profiles for 3,131 primary tumours and Group 3 copy number data at the MYC/PVT1 locus indicated that 947 cancer cell lines (total of 4,078 cases) revealed only four cases with additional samples might harbour PVT1 gene fusions (Fig. 5c). PCR 27 apparent duplication of SNCAIP, all of which were inferred as Group 4 with reverse transcription (RT–PCR) profiling of select Group 3 cases medulloblastomas(datanotshown).Weconcludethat SNCAIPduplica- confirmed PVT1-MYC fusions in at least 60% (12/20) of MYC- tion is a somatic event highly specific to Group 4 medulloblastoma. amplified cases (Fig. 5d and Supplementary Table 19). Fusion Re-analysis of 499 published medulloblastoma expression profiles transcripts included many other portions of chr8q, with up to four confirmed that SNCAIP is one of the most highly upregulated Group 4 different genomic loci mapping to a single transcript, a pattern remin- signature genes (Fig. 4d and Supplementary Fig. 28). Profiling of 188 iscent of chromothripsis 28,29 (Fig. 5d). WGS performed on four MYC- Group 4 tumours on expression microarrays followed by consensus amplified Group 3 tumours harbouring PVT1 fusion transcripts non-negative matrix factorization (NMF) clustering delineates two identified a series of complex genomic rearrangements on chr8q subtypes of Group 4 (4a and 4b; Fig. 4e and Supplementary Fig. 29). (Fig. 5e, f, Supplementary Fig. 38 and Supplementary Tables 20 and Strikingly, 21 out of 22 SNCAIP duplicated cases belonged to Group 4a 21). Chromosome 8 copy number profile for MB-586 (PVT1- 28 (P 5 3.123 10 , Fisher’s exact test). SNCAIP is more highly NDRG1) derived from WGS showed that PVT1 and NDRG1 are expressed in Group 4a than 4b (Fig. 4f), and 4a samples with tandem structurally linked, as predicted by RNA-Seq, and several adjacent duplication showed approximately 1.5-fold increased expression, con- regions of 8q24 were extensively rearranged (Fig. 5e, f and Sup- sistent with gene dosage (Fig. 4g and Supplementary Figs 35 and 36). plementary Table 21). Monte Carlo simulation suggests that this Group 4a exhibits a relatively balanced genome compared to 4b fragmented 8q amplicon arose through chromothripsis, a process of (Supplementary Figs 30–32), and several 4a cases harbour SNCAIP erroneous DNA repair following a single catastrophic event in which a duplication in conjunction with i17q and no other SCNAs (Sup- chromosome is shattered into many pieces (Supplementary Fig. 39). plementary Fig. 33). Importantly, SNCAIP duplications are mutually Further examination of our copy number data set revealed rare exclusive from other prominent SCNAs in Group 4, including MYCN examples of chromothripsis across subgroups (Supplementary Fig. 40), and CDK6 amplifications (Supplementary Fig. 34). with onlychr8 inGroup 3 demonstratingstatistically significant,region- specific chromothripsis (Q 5 0.0004, false discovery rate (FDR)- PVT1 fusions arise via chromothripsis in Group 3 corrected Fisher’s exact test). Among Group3 tumours, theoccurrence Although recurrent gene fusions have recently been discovered in solid of chr8q chromothripsis is correlated with deletion of chr17p (location tumours, none have been reported in medulloblastoma. RNA-Seq of of TP53; data not shown), in keeping with the association of loss of 5 2 | N A T UR E | V O L 4 8 8 | 2 A UG US T 2 012 ©2012 Macmillan Publishers Limited. All rights reserved

ARTICLE RESEARCH a MB-182 RNA-Seq profle c chr8 e Inversion Duplication 100 kb 1,048 kb MB-586 WGS profle Deletion Translocation chr8 129000 kb 129400 kb 6 (q24.21) Sample 128600 kb Chromosome 8 Status 128850000 128900000 128950000 129000000 129050000 129100000 129150000 D458 4 MB-524 Chromosome band 8q24.21 MB-1003 Log 2 ratio (read depth tumour vs normal) 2 MYC PVT1 MB-1377 0 MB-1240 MIR1204 MIR1205 MIR1206 MIR1207 –2 1 571 2,366 1 1,901 MB-182 D425 5′ 1 ATG 2 3 3′ 5′ 1 23456789 3′ MB-1338 MYC PVT1 MED8A 0 50 100 150 MB-586 Chromosomal coordinates (Mb) 1 628 Sanger sequencing 1,901 MB-591 5′ 13 ATG 2 3 3′ MB-548 PVT1-MYC MB-506 AK125310 BC106081 Paired-end reads 128.2 DQ515898 RefSeq MYC PVT1 MIR f 131.1 POU5F1 genes MIR MIR MIR 1208 GSDMC 1204 1205 1207 MB-586: Chr8 BC042052 b MB-586 RNA-Seq profle d D458 MYC chr8 2 Mb TMEM75 PVT1 (q24.21- MB-524 CCDC26 q24.22) 130000000 131000000 132000000 133000000 134000000 MB-1003 CR593328 129500000 130500000 131500000 132500000 133500000 MB-1377 129.1 Chromosome band 8q24.21 8q24.22 MB-1240 2.36 Mb 134.5 NDRG1 MYC PVT1 NDRG1 MB-182 1 1,901 3,081 1,962 1 D425 WISP1 AX746885 5′ 12 3 4 567 89 3′ 3′ 16 10 8542 1 5′ SLAP PVT1 NDRG1 MB-1338 129.9 SLA 1 273 Sanger sequencing 2,570 MED8A 144.5 TG 5′ 1 10 16 3′ MB-586 MAFA PVT1-NDRG1 MB-591 C8orf73 133.8 Paired-end reads RHPN1 144.8 138.0 138.6 MB-548 ZC3H3 PHF20L1 MB-506 EEF1D TIGD5 In-tandem type PVT1 NDRG1 chr8~128.0 Mb chr8~128.7 Mb BC034020 TSTA3 PYCRL Tail-to-tail type MYC chr8~127.5 Mb chr8~125.5 Mb chr8~136.7 Mb Head-to-head type Figure 5 | Identification of frequent PVT1-MYC fusion genes in Group 3. fusions (shown in d). Yellow box highlights the common breakpoint affecting a, b, RNA-Seq identifies multiple fusion transcripts driven by PVT1 in Group 3. the first exon/intron of PVT1, including miR-1204. d, Summary of PVT1 fusion Schematics depict the structures of verified PVT1-MYC (a) and PVT1-NDRG1 transcripts identified in Group 3. e, f, WGS confirms complex patterns of (b) fusion genes. c, Heat map of the MYC/PVT1 locus showing a subset of 13 rearrangement on chr8q24 in PVT1 fusion (1) Group 3. MYC-amplified Group 3 cases subsequently verified to exhibit PVT1 gene TP53 and chromothripsis recently described in medulloblastoma interference of miR-1204 had a pronounced effect on MED8A growth 28 (P 5 0.0199, Fisher’s exact test) . Whereas the PVT1 locus has been (Fig. 6b). A comparable reduction in proliferative capacity was suggested to be a genomically fragile site, we observe that the majority achieved with knockdown of MYC. Conversely, the medulloblastoma of MYC-amplified Group 3 tumours harbour PVT1 fusions that arise cell line ONS76 exhibits neither MYC amplification nor a detectable through a process consistent with chromothripsis. PVT1-MYC fusion gene, and knockdown of miR-1204 had no effect in PVT1 is a non-coding host gene for four microRNAs, miR-1204– this line (Fig. 6c). miR-1207. Previous studies have implicated miR-1204 as a candidate PVT1 has been reported previously in fusion transcripts with a oncogene that enhances oncogenesis in combination with MYC 30,31 . number of partners 30,32,33 . The most prevalent form of the PVT1- PVT1 fusions identified in this study involve only PVT1 exon 1 and MYC fusion in Group 3 tumours lacks the first, non-coding exon of miR-1204. Importantly, miR-1204, but not the adjacent miR-1205 and MYC, similar to forms of MYC that have been described in Burkitt’s miR-1206, is expressed at a higher level in PVT1-MYC fusion (1) lymphoma 34 (Fig. 5a, d). The PVT1 promoter contains two non- 31 Group 3 tumours compared to fusion (2) cases (P 5 0.0008, Mann– canonical E-boxes and can be activated by MYC . This indicates a Whitney test; Fig. 6a). To evaluate whether aberrant expression of positive feedback model where MYC can reinforce its own expression miR-1204 contributes to the malignant phenotype, we inhibited miR- from the PVT1 promoter in PVT1-MYC fusion (1) tumours. Indeed, 1204 in MED8A cells, a Group 3 medulloblastoma cell line with a knockdown of MYC alone in MED8A cells resulted in diminished confirmed PVT1-MYC fusion (Fig. 5d). Antagomir-mediated RNA expression ofboth MYCand miR-1204, suggesting MYCmay positively regulate PVT1 (that is, miR-1204) expression in medulloblastoma cells a (Supplementary Fig. 41). MYC-balanced/ MYC-amplifed/ MYC-amplifed/ 1.0 CTL MYC-siRNA fusion (–) fusion (–) fusion (+) 0.8 miR-1204-antagonist Discussion 14 Absorbance (490 nm) b 0.6 Medulloblastomas have few SNVs compared to many adult epithelial P = 0.0008 0.4 malignancies , whereas SCNAs seem to be quite common. 11 Relative expression 12 8 6 4 0.8 0 0244872 Medulloblastoma is a heterogeneous disease , thereby requiring large 0.2 P = 0.1640 10 7 MED8A cohorts to detect subgroup-specific events. Through the accumula- Time (h) tion of .1,200 medulloblastomas in MAGIC, we have identified CTL MYC-siRNA miR-1204-antagonist subgroup-restricted, highly supporting their role as driver events in 0 2 P = 0.4389 Absorbance (490 nm) c 0.6 novel and significant SCNAs. Many of the significant SCNAs are 0.4 their respective subgroups. 1204 1205 1207 1204 1205 1207 1204 1205 1207 0.2 0 ONS76 Expression of synphilin-1 in neuronal cells results in decreased cell 36 35 miR miR miR 0244872 doubling time , decreased caspase-3 activation , decreased TP53 Time (h) transcriptional activity and messenger RNA levels, and decreased 37 Figure 6 | Functional synergy between miR-1204 and MYC secondary to apoptosis . Synphilin-1 is ubiquitinated by parkin, which is encoded PVT1-MYC fusion. a, Quantitative RT–PCR of PVT1-encoded microRNAs bythe hereditary Parkinson’s disease gene PARK2 (ref. 24), a candidate confirms upregulation of miR-1204 in PVT1-MYC fusion (1) Group 3 tumours. tumour suppressor gene . Whereas patients with Parkinson’s disease 38 MYC-balanced/fusion (2), n54; MYC-amplified/fusion (2), n56; MYC- have an overall decreased risk of cancer, they may have an increased amplified/fusion (1), n58. Error bars represent standard error of the mean 39,40 (s.e.m.) and reflect variability among samples. b, c, Knockdown of miR-1204 incidence of brain tumours . As tandem duplications of SNCAIP are attenuates the proliferative capacity of PVT1-MYC fusion (1) MED8A highly recurrent, stereotypical, subgroup-restricted, affect only a single medulloblastoma cells (b) but has no effect on fusion (2) ONS76 cells (c). Error gene, and as SNCAIP-duplicated tumours have few if any other barsrepresentthestandarddeviation(s.d.)oftriplicateexperiments.CTL,control. SCNAs, SNCAIP is a probable driver gene, and merits investigation 2 A UGUS T 2 012 | V OL 488 | N A TU RE | 5 3 ©2012 Macmillan Publishers Limited. All rights reserved

RESEARCH ARTICLE as a target for therapy of Group 4a. Similarly, PVT1 fusion genes are 16. Adamson, D. C. et al. OTX2 is critical for the maintenance and progression of Shh- independent medulloblastomas. Cancer Res. 70, 181–191 (2010). highly recurrent, restricted to Group 3, arise through a chromothripsis- 17. Jia, S., Wu, D., Xing, C. & Meng, A. Smad2/3 activitiesare requiredfor induction and like process, and are the first recurrent translocation reported in patterning of the neuroectoderm in zebrafish. Dev. Biol. 333, 273–284 (2009). medulloblastoma. 18. Bizet, A. A. et al. The TGF-b co-receptor, CD109, promotes internalization and degradation of TGF-b receptors. Biochim. Biophys. Acta 1813, 742–753 (2011). We identify a number of highly targetable, recurrent, subgroup- 19. Wang,T.,Donahoe,P. K.& Zervos, A. S. Specificinteraction oftypeI receptorsof the specific SCNAs that could form the basis for future clinical trials (that TGF-beta family with the immunophilin FKBP-12. Science 265, 674–676 (1994). is, PI3K signalling in SHH, TGF-b signalling in Group 3, and NF-kB 20. Parks, W. T. et al. Sorting nexin 6, a novel SNX, interacts with the transforming growth factor-b family of receptor serine-threonine kinases. J. Biol. Chem. 276, signalling in Group 4). Activation of these pathways through alterna- 19332–19339 (2001). tive, currently unknown genetic and epigenetic events could increase 21. Bredel, M. et al. NFKBIA deletion in glioblastomas. N. Engl. J. Med. 364, 627–637 the percentage of patients amenable to targeted therapy. We also (2011). identify a number of highly ‘druggable’ events that occur in a minority 22. Xiao, N. et al. Ubiquitin-specific protease 4 (USP4) targets TRAF2 and TRAF6 for deubiquitination and inhibits TNFa-induced cancer cell migration. Biochem. J. of cases. The cooperative, global approach of the MAGIC consortium 441, 979–986 (2012). has allowed us to overcome the barrier of intertumoural heterogeneity 23. Engelender, S. et al. Synphilin-1 associates with a-synuclein and promotes the in an uncommon paediatric tumour, and to identify the relevant and formation of cytosolic inclusions. Nature Genet. 22, 110–114 (1999). 24. Chung, K. K. et al. Parkin ubiquitinates the a-synuclein-interacting protein, targetable SCNAs for the affected children. synphilin-1: implications for Lewy-body formation in Parkinson disease. Nature Med. 7, 1144–1150 (2001). METHODS SUMMARY 25. Marx, F. P. et al. Identification and functional characterization of a novel R621C mutation in the synphilin-1 gene in Parkinson’s disease. Hum. Mol. Genet. 12, All patient samples were obtained with consent as outlined by individual 1223–1231 (2003). institutional review boards. Genomic DNA was prepared, processed and 26. Beroukhim, R. et al. The landscape of somatic copy-number alteration across hybridized to Affymetrix SNP6 arrays according to manufacturer’s instructions. human cancers. Nature 463, 899–905 (2010). Raw copy number estimates were obtained in dChip, followed by CBS segmenta- 27. Barretina, J. et al. The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature 483, 603–607 (2012). tion in R. SCNAs were identified using GISTIC2 (ref. 13). Driver genes within 28. Rausch, T. et al. Genome sequencing of pediatric medulloblastoma links SCNAs were inferred by integrating matched expressions, literature evidence and catastrophic DNA rearrangements with TP53 mutations. Cell 148, 59–71 (2012). other data sets. Pathway enrichment of SCNAs was analysed with g:Profiler and 29. Stephens, P. J. et al. Massive genomic rearrangement acquired in a single visualized in Cytoscape using Enrichment Map. Fluorescence in situ hybridiza- catastrophic event during cancer development. Cell 144, 27–40 (2011). tion (FISH) was performed as described previously 8,10 . Medulloblastoma 30. Shtivelman,E.&Bishop,J.M.The PVTgenefrequentlyamplifies withMYC intumor subgroup was assigned using a custom nanoString CodeSet as described previ- cells. Mol. Cell. Biol. 9, 1148–1154 (1989). 31. Carramusa, L. et al. The PVT-1 oncogene is a Myc protein target that is ously . Tandem duplication of SNCAIP was confirmed by paired-end mapping overexpressed in transformed cells. J. Cell. Physiol. 213, 511–518 (2007). 12 28 as previously reported . RNA was extracted, processed and hybridized to 32. Shtivelman, E. & Bishop, J. M. Effects of translocations on transcription from PVT. Affymetrix Gene 1.1 ST Arrays as recommended by the manufacturer. Mol. Cell. Biol. 10, 1835–1839 (1990). Consensus NMF clustering was performed in GenePattern. Gene fusions were 33. Pleasance, E. D. et al. A small-cell lung cancer genome with complex signatures of identified from RNA-Seq data using Trans-ABySS. Medulloblastoma cell lines tobacco exposure. Nature 463, 184–190 (2010). 10 were maintained as described . Proliferation assays were performed with the 34. Hann, S. R., King, M.W., Bentley,D. L.,Anderson, C.W. & Eisenman,R. N. A non-AUG translational initiation in c-myc exon 1 generates an N-terminally distinct protein Promega CellTiter 96 Assay. Additional methods are detailed in full in whose synthesis is disrupted in Burkitt’s lymphomas. Cell 52, 185–195 (1988). Supplementary Methods. 35. Li, X. et al. Synphilin-1 exhibits trophic and protective effects against Rotenone toxicity. Neuroscience 165, 455–462 (2010). Received 29 February; accepted 14 June 2012. 36. Smith, W. W. et al. Synphilin-1 attenuates neuronal degeneration in the A53T alpha-synuclein transgenic mouse model. Hum. Mol. Genet. 19, 2087–2098 Published online 25 July 2012. (2010). 37. Giaime, E. et al. Caspase-3-derived C-terminal product of synphilin-1 displays 1. Gajjar, A. et al. Risk-adapted craniospinal radiotherapy followed by high-dose antiapoptotic function via modulation of the p53-dependent cell death pathway. chemotherapy and stem-cell rescue in children with newly diagnosed J. Biol. Chem. 281, 11515–11522 (2006). medulloblastoma (St Jude Medulloblastoma-96): long-term results from a 38. Veeriah, S. et al. Somatic mutations of the Parkinson’s disease-associated gene prospective, multicentre trial. Lancet Oncol. 7, 813–820 (2006). PARK2 in glioblastoma and other human malignancies. Nature Genet. 42, 77–82 2. Mabbott, D. J. et al. Serial evaluation of academic and behavioral outcome after (2010). treatment with cranial radiation in childhood. J. Clin. Oncol. 23, 2256–2263 39. Olsen, J. H. et al. Atypical cancer pattern in patients with Parkinson’s disease. Br. J. (2005). Cancer 92, 201–205 (2005). 3. Cho, Y. J. et al. Integrative genomic analysis of medulloblastoma identifies a 40. Moller, H., Mellemkjaer, L., McLaughlin, J. K. & Olsen, J. H. Occurrence of different molecular subgroup that drives poor clinical outcome. J. Clin. Oncol. 29, cancers in patients with Parkinson’s disease. Br. Med. J. 310, 1500–1501 (1995). 1424–1430 (2011). 4. Northcott, P. A. et al. Medulloblastoma comprises four distinct molecular variants. Supplementary Information is linked to the online version of the paper at J. Clin. Oncol. 29, 1408–1414 (2011). www.nature.com/nature. 5. Remke, M. et al. FSTL5 is a marker of poor prognosis in non-WNT/non-SHH medulloblastoma. J. Clin. Oncol. 29, 3852–3861 (2011). Acknowledgements M.D.T. is the recipient of a CIHR Clinician-Scientist Phase II award, and was formerly a Sontag Distinguished Scholar with funds from the Sontag 6. Northcott,P. A., Korshunov, A., Pfister, S. M. & Taylor, M. D. Theclinical implications of medulloblastoma subgroups. Nature Rev. Neurol. 8, 340–351 (2012). Foundation. Funding is acknowledged from the Pediatric Brain Tumour Foundation 7. Taylor, M. D. et al. Molecular subgroups of medulloblastoma: the current (M.D.T. and J.T.R.), and the National Institutes of Health (CA159859 to M.D.T., R.W.R. consensus. Acta Neuropathol. 123, 465–472 (2011). and B.W.), The Canadian Cancer Society, Genome Canada, Genome BC, Terry Fox 8. Northcott, P. A. et al. Pediatric and adult sonic hedgehog medulloblastomas are Research Institute, Ontario Institute for Cancer Research, Pediatric Oncology Group Ontario, Funds from ‘The Family of Kathleen Lorette’ and the Clark H. Smith Brain clinically and molecularly distinct. Acta Neuropathol. 122, 231–240 (2011). 9. Rudin,C. M.et al.Treatment ofmedulloblastoma with hedgehog pathway inhibitor Tumour Centre, Montreal Children’s Hospital Foundation, Hospital for Sick Children: Sonia and Arthur Labatt Brain Tumour Research Centre, Chief of Research Fund, GDC-0449. N. Engl. J. Med. 361, 1173–1178 (2009). 10. Northcott, P. A. et al. Multiple recurrent genetic events converge on control of Cancer Genetics Program, Garron Family Cancer Centre, B.R.A.I.N. Child, CIHR (grant no. ATE-110814); the University of Toronto McLaughlin Centre, CIHR Institute of histone lysine methylation in medulloblastoma. Nature Genet. 41, 465–472 Cancer Research (grant no. AT1 – 112286) and C17 through the Advancing (2009). Technology Innovation through Discovery competition (Project Title: The Canadian 11. Parsons, D. W. et al. The genetic landscape of the childhood cancer Pediatric Cancer Genome Consortium: Translating next-generation sequencing medulloblastoma. Science 331, 435–439 (2011). technologies into improved therapies for high-risk childhood cancer). Canada’s 12. Northcott, P. A. et al. Rapid, reliable, and reproducible molecular sub-grouping of Michael Smith Genome Sciences Centre is supported by the BC Cancer Foundation. clinical medulloblastoma samples. Acta Neuropathol. 123, 615–626 (2012). J.R. is supported by The Children’s Discovery Institute. P.A.N. was supported by a 13. Mermel, C. H. et al. GISTIC2.0 facilitates sensitive and confident localization of the Restracomp Fellowship (Hospital for Sick Children) and is currently a Roman-Herzog targets of focal somatic copy-number alteration in human cancers. Genome Biol. Postdoctoral Fellow (Hertie Foundation). Salary support for L.G. was provided by the 12, R41 (2011). Ontario Institute for Cancer Research through funding provided by the Government of 14. Buonamici, S. et al. Interfering with resistance to smoothened antagonists by Ontario. E.V.G.M. is supported by NIH grants CA86335, CA116804, CA138292, NCI inhibition of the PI3K pathway in medulloblastoma. Sci. Transl. Med. 2, 51ra70 contracts 28XS100 and 29XS193, the Southeastern Brain Tumour Foundation, and (2010). the Brain Tumour Foundation for Children. This study includes samples provided by 15. Yauch,R. L.et al.Smoothened mutation confers resistance to a hedgehog pathway the UK Children’s Cancer and Leukaemia Group (CCLG) as part of CCLG-approved inhibitor in medulloblastoma. Science 326, 572–574 (2009). biological study BS-2007-04. J.K. and S.P. were supported by a grant from the German 5 4 |N A T U R E| V O L4 8 8 |2A U G U S T 2 0 1 2 ©2012 Macmillan Publishers Limited. All rights reserved

ARTICLE RESEARCH 61 16 4 CancerAid(109252).WethankC.Lu,K.OtakaandTheCentreforAppliedGenomicsfor Malkin 94,96 , Steven C. Clifford , Steven J. M. Jones , Jan O. Korbel , Stefan M. technical assistance. We thank N. S. Devi and Z. Zhang for technical assistance. We Pfister 2,19 , Marco A. Marra 17,97 & Michael D. Taylor 1,3,25 thank D. Stoll for project management, S. Archer for technical writing, and P. Paroutis for artwork. The MAGIC project is part of the International Cancer Genome Consortium. 1 Developmental &StemCell Biology Program,TheHospitalfor SickChildren,101 College 2 Street, TMDT-11-401M, Toronto, Ontario M5G 1L7, Canada. Division of Pediatric Author Contributions P.A.N. and M.D.T. co-conceived the study. P.A.N., M.A.M., and M.D.T. led the study. P.A.N. planned and executed experiments and analyses, Neurooncology, German Cancer Research Center (DKFZ), Im Neuenheimer Feld 280, 3 supervised data acquisition, performed bioinformatic analyses, and extracted nucleic 69120 Heidelberg, Germany. Department of Laboratory Medicine and Pathobiology, acidsfor theMAGIC cohort. D.J.H.S. led the bioinformatics andperformed analyses.J.P. University of Toronto, Medical Sciences Buildings, 1 King’s College Circle, 6th Floor, 4 performed quantitative RT–PCR and Sanger sequencing of PVT1 fusions, expression Toronto, Ontario M5S 1A8, Canada. Genome Biology, European Molecular Biology 5 profiled PVT1-encoded miRNAs, and generated schematics for PVT1 fusion genes. L.G. Laboratory, Meyerhofstrasse 1, 69117 Heidelberg, Germany. CCU Neuropathology, performed the MYC and miR-1204 knockdown experiments. A.S.M. supervised the German Cancer Research Center (DKFZ), Im Neuenheimer Feld 220-221, Department of RNA-Seq and WGS experiments and performed data analysis. T.Z., A.M.S. and J.O.K. Neuropathology, University of Heidelberg, Im Neuenheimer Feld 224, 69120 Heidelberg, 6 performed the large insert paired-end sequencing and PCR verification of SNCAIP Germany. The Donnelly Centre, University of Toronto, 160 College Street, Room 602, 7 duplication samples. A.Ko. performed interphase FISH and immunohistochemistry for Toronto, Ontario M5S 3E1, Canada. Department of Cancer Biology, Dana-Farber Cancer 8 candidate genes. J.R. and G.D.B. led the pathway analyses and generated enrichment Institute, 450 Brookline Avenue, Boston, Massachusetts 02215, USA. Department of plots. S.E.S. and R.B. provided technical support with the GISTIC2 bioinformatic Medical Oncology, Dana-Farber Cancer Institute, 450 Brookline Avenue, Boston, 9 platform. D.W.E. performed interphase FISH for candidate genes. C.R.M., A.C.L. and Massachusetts 02215, USA. Department of Medicine, Harvard Medical School, 25 10 S.W.S. performed the SNP6 genotyping analysis, provided a database of normal copy ShattuckStreet, Boston,Massachusetts02115,USA. Department ofMedicine, Brigham 11 number variants, and the control dataset used to infer copy number in the tumour and Women’s Hospital, 75 Francis Street, Boston, Massachusetts 02115, USA. Cancer samples. S.M., A.D., F.M.G.C., M.K., D.T.W.J. and H.W. performed bioinformatic analyses Program, Broad Institute, 7 Cambridge Center, Cambridge, Massachusetts 02142, USA. 12 and provided technical advice. Y.Y. sequenced CTNNB1 in the WNT tumours. V.R., D.K., Center for Cancer Genome Discovery, Dana-Farber Cancer Institute, 450 Brookline 13 M.F.R., T.A., and P.D. performed functional assays for candidate genes. B.Lu. extracted Avenue, Boston, Massachusetts 02215, USA. Pathology, St Jude Children’s Research 14 nucleic acids, managed biobanking, and maintained the patient database. S.M. and Hospital, 262 Danny Thomas Place, Memphis, Tennessee 38105, USA. McLaughlin A.R. performed the drug database analysis. Xin W., Xiaochong W. and M.R. provided Centre and Department of Molecular Genetics, University of Toronto, 101 College Street, 15 technical support. R.Y.B.C., A.C., E.C., R.D.C., G.R.H., S.D.J., Y.L., A.L., K.L.M., K.M.N., J.Q.Q., Toronto, Ontario M5G 1L7, Canada. The Centre for Applied Genomics and Program in Genetics and Genome Biology, The Hospital for Sick Children, 101 College Street, A.G.J.R., N.T., R.J.V., I.B., R.A.M., A.J.M., R.H. and S.J.M.J. led the RNA-Seq and WGS TMDT-14-701, Toronto, Ontario M5G 1L7, Canada. 16 Michael Smith Genome Sciences experiments and performed data analyses. A.F.-L and A.M.K. provided the database of Centre, BC Cancer Agency, 100-570 West 7th Avenue, Vancouver, British Columbia V5Z SHH-responsive genes. R.J.W.-R., W.A.G., M.P.-P., C.C.H., O.D., S.S.R., F.F.D., S.S.P.-F., 4S6, Canada. 17 Michael Smith Genome Sciences Centre, BC Cancer Agency, 675 West B.-K.C., S.-K.K., K.-C.W., W.S., C.G.E., M.F.-M., A.J., I.F.P., X.F., K.M.M., G.Y.G., C.D.R., L.M., 10th Avenue, Vancouver, British Columbia V5Z 1L3, Canada. 18 Tumour Cell Biology, St E.M.C.M., N.K.K., P.J.F., J.M.K., J.M.O., R.G.E., K.Z., L.K., R.C.T., MKC, B.La., R.E.M., D.D.B., Jude Children’s Research Hospital, 262 Danny Thomas Place, Memphis, Tennessee A.F., S.A., N.J., J.C.L., S.B., N.G., W.A.W., L.B., A.Kl., T.E.V.M., T.K., T.T., S.K.E., J.R.L., J.B.R., 38105, USA. 19 Department of Pediatric Oncology, University Hospital Heidelberg, Im L.M.L., E.G.V.M., M.F., H.N., G.C., M.G., P.H., A.G.S., A.I., S.J., C.G.C., R.V., Y.S.R., S.R., M.Z., Neuenheimer Feld 430, 69120 Heidelberg, Germany. 20 Departments of Hematology and C.C.F., J.A.C., M.L.L., Y.-J.C., U.T., C.E.H., E.B., S.C.C. and S.M.P. provided the patient Immunology, University Hospital Heidelberg, Im Neuenheimer Feld 430, 69120 samples and clinical details that made the study possible. P.H.B.S., M.M., S.L.P., Y.-J.C., Heidelberg, Germany. 21 Pediatric Clinical Trials Office, Memorial Sloan-Kettering Cancer U.T., C.E.H., E.B., S.W.S., J.T.R., D.M., S.C.C., S.J.M.J., J.O.K., S.M.P. and M.A.M. provided Center, 405 Lexington Avenue, New York, New York 10174, USA. 22 Neurological Surgery, valuable input regarding study design, data analysis, and interpretation of results. Vanderbilt Medical Center, T-4224 MCN, Nashville, Tennessee 37232-2380, USA. P.A.N., D.J.H.S., J.P., L.G., A.S.M., M.A.M. and M.D.T. wrote the manuscript. M.A.M. and 23 Cancer Biology, Vanderbilt Medical Center, 465 21st Avenue South, MRB III 6160, M.D.T. provided financial and technical infrastructure and oversaw the study. M.A.M. Nashville, Tennessee 37232-8550, USA. 24 Sanford-Burnham Medical Research Institute, and M.D.T. are joint senior authors and project co-leaders. La Jolla, California 92037, USA. 25 Department of Surgery, Division of Neurosurgery and Author Information SNP6 copy number and gene expression array data have been Labatt Brain Tumour Research Centre, The Hospital for Sick Children, 555 University deposited at the Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/) Avenue, Hill 1503, Toronto, Ontario M5G 1X8, Canada. 26 Developmental & Stem Cell as a GEO SuperSeries under accession number GSE37385. Whole genome and Biology Program, The Hospital for Sick Children, 101 College Street, TMDT-13-601, 27 transcriptome sequencing data have been deposited at the European Toronto, Ontario M5G 1L7, Canada. Department of Pathology, The Children’s Memorial 28 Genome-phenome Archive (EGA; https://www.ebi.ac.uk/ega/) hosted by the EBI, Health Institute, Aleja Dzieci Polskich 20, 04-730 Warsaw, Poland. Department of under accession number EGAD00001000158. Reprints and permissions information Oncology, The Children’s Memorial Health Institute, Aleja Dzieci Polskich 20, 04-730 29 is available at www.nature.com/reprints. This paper is distributed under the terms of Warsaw, Poland. Institute of Neurology, Medical University of Vienna, AKH 4J, 30 the Creative Commons Attribution-Non-Commercial-Share Alike licence, and is freely Waehringer Gu ¨rtel 18-20, A-1097 Vienna, Austria. INSERM U 830, Institut Curie, 26 rue 31 available to all readers at www.nature.com/nature. The authors declare no competing d’Ulm, 75238 Paris Cedex 5, France. 32 Unit of Somatic Genetics, Institut Curie, 26 rue financialinterests.Readersarewelcometocommentontheonlineversionofthisarticle d’Ulm, 75238 Paris Cedex 5, France. Department of Pediatric Oncology, Institut Curie, at www.nature.com/nature. Correspondence and requests for materials should be 26 rue d’Ulm, 75248Paris Cedex 5, France. 33 Pediatric Hematology and Oncology, CHUV 34 addressed to M.A.M. ([email protected]) or M.D.T. ([email protected]). University Hospital, 1011 Lausanne, Switzerland. Department of Neurosurgery, Division of Pediatric Neurosurgery, Seoul National University Children’s Hospital, 101 Daehak-Ro Jongno-Gu, Seoul 110-744, South Korea. 35 Head of Pediatrics, Cnopf ´ sche Kinderklinik, Theodor-Kutzer-Ufer 1-3, 90419 Nuremberg, Germany. 36 Departments of Pathology, Ophthalmology and Oncology, John Hopkins University School of Medicine, 1 Paul A. Northcott 1,2 *, David J. H. Shih 1,3 *, John Peacock 1,3 , Livia Garzia , A. Sorana 720 Rutland Avenue, Ross Building 558, Baltimore, Maryland 21205, USA. 37 INSERM 4 5 4 1 Morrissy , Thomas Zichner , Adrian M. Stu ¨tz , Andrey Korshunov ,Ju ¨ri Reimand , 6 U1028, CNRS UMR5292, Centre de Recherche en Neurosciences, Universite ´ de Lyon, 7 13 Steven E. Schumacher , Rameen Beroukhim 7,8,9,10,11,12 , David W. Ellison , Christian 69336Lyon, France. 38 Centre dePathologie EST,GroupementHospitalier EST, Universite ´ 15 14 R. Marshall , Anath C. Lionel , Stephen Mack 1,3 , Adrian Dubuc 1,3 , Yuan Yao 1,3 , Vijay de Lyon, 69500 Bron, France. 39 Department of Neurological Surgery, University of 1 1 1 Ramaswamy 1,3 , Betty Luu , Adi Rolider , Florence M. G. Cavalli , Xin Wang 1,3 , Marc Pittsburgh SchoolofMedicine,4401PennAvenue,Pittsburgh,Pennsylvania15224,USA. 1 16 16 1 16 Remke , Xiaochong Wu , Readman Y. B. Chiu , Andy Chu , Eric Chuah , Richard D. 40 Departments of Neurosurgery and Cell and Developmental Biology, University of 16 16 16 16 16 Corbett , Gemma R. Hoad , Shaun D. Jackman , Yisu Li , Allan Lo , Karen L. Michigan Medical School, 109 Zina Pitcher Place, 5018 BSRB, Ann Arbor, Michigan 16 16 16 16 16 Mungall , KaMingNip , JennyQ.Qian , AnthonyG.J.Raymond , NinaThiessen , 48109, USA. 41 Department of Neurosurgery, University of Michigan Medical School, 16 16 16 16 Richard J. Varhol , Inanc Birol , Richard A. Moore , Andrew J. Mungall , Robert 1500 E. Medical Center Drive, Taubman Center, Room3552,Ann Arbor,Michigan 48109, 17 2 18 2 18 Holt , Daisuke Kawauchi , Martine F. Roussel , Marcel Kool , David T. W. Jones , USA. 42 Department of Surgery, Division of Neurosurgery, University of Alabama at 21 Hendrick Witt 19,20 , Africa Fernandez-L , Anna M. Kenney 22,23 , Robert J. Birmingham, 1900 University Boulevard, THT 1052, Birmingham, Alabama 27 25 24 26 Wechsler-Reya , Peter Dirks , Tzvi Aviv , Wieslawa A. Grajkowska , Marta 35294-0006, USA. 43 Pediatric Neurosurgery, Catholic University Medical School, 00186 31 28 29 30 Perek-Polnik , Christine C. Haberler , Olivier Delattre , Ste ´phanie S. Reynaud , Rome, Italy. 44 Department of Pediatric Oncology and Hematology, Erasmus Medical 34 34 33 32 François F. Doz , Sarah S. Pernet-Fattet , Byung-Kyu Cho , Seung-Ki Kim , Center, Dr. Molewaterplein 50, 3000 Rotterdam, The Netherlands. 45 Department of 35 34 36 Kyu-Chang Wang , Wolfram Scheurlen , Charles G. Eberhart , Michelle Neurology, Erasmus Medical Center, Dr. Molewaterplein 50, PO Box 2040, 3000 CA 41 37 39 40 38 Fe `vre-Montange , AnneJouvet , IanF.Pollack ,XingFan , KarinM.Muraszko ,G. Rotterdam, The Netherlands. 46 Department of Pathology, Erasmus Medical Center, Dr. 43 43 44 42 Yancey Gillespie , Concezio Di Rocco , Luca Massimi , Erna M. C. Michiels , Molewaterplein 50, 3015 GE Rotterdam, The Netherlands. 47 Clinical Research Division, 46 45 Nanne K. Kloosterhof 44,45 , Pim J. French , Johan M. Kros , James M. Olson 47,48 , Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue North, D4-100, Seattle, 49 22 52 Richard G. Ellenbogen , Karel Zitterbart 50,51 , Leos Kren , Reid C. Thompson , Washington 98109, USA. 48 Seattle Children’s Hospital, Seattle, Washington 98104, USA. 56 53 56 Michael K. Cooper , Boleslaw Lach 54,55 , Roger E. McLendon , Darell D. Bigner , 49 Neurological Surgery, University of Washington School of Medicine, Harborview 61 57 Adam Fontebasso , Steffen Albrecht 58,59 , Nada Jabado 57,60 , Janet C. Lindsey , Medical Center, 325 Ninth Avenue, Seattle, Washington 98104, USA. 50 Department of 64 63 64 62 61 Simon Bailey , Nalin Gupta , William A. Weiss ,La ´szlo ´ Bogna ´r , Almos Klekner , Pediatric Oncology, School of Medicine, Masaryk University, Cernopolni 9, 613 00 Brno, 66 65 66 67 Timothy E. Van Meter , Toshihiro Kumabe , Teiji Tominaga , Samer K. Elbabaa , Czech Republic. 51 Department of Pediatric Oncology, University Hospital Brno, 625 00 69 68 71 70 Jeffrey R. Leonard , Joshua B. Rubin , Linda M. Liau , Erwin G. Van Meir , Maryam Brno, Czech Republic. 52 Department of Pathology, University Hospital Brno, Jihlavska 20, 74 72 75 75 73 Fouladi , Hideo Nakamura , Giuseppe Cinalli , Miklo ´s Garami , Peter Hauser , 625 00 Brno, Czech Republic. 53 Department of Neurology, Vanderbilt Medical Center, 80 79 81 76 AliG.Saad , Achille Iolascon 77,78 , Shin Jung , Carlos G.Carlotti , Rajeev Vibhakar , 465 21st Avenue South, MRB III 6160, Nashville, Tennessee 37232-8550, USA. 82 Young Shin Ra , Shenandoah Robinson 83,84 , Massimo Zollo 77,78 , Claudia C. 54 Department of Pathology and Molecular Medicine, Division of Anatomical Pathology, 89 87 88 Faria 85,86 , Jennifer A. Chan , Michael L. Levy , Poul H. B. Sorensen , Matthew McMasterUniversity,Hamilton,OntarioL8S4L8,Canada. 55 DepartmentofPathologyand 90 94 8 91 Meyerson , Scott L. Pomeroy , Yoon-Jae Cho , Gary D. Bader 6,14,92,93 , Uri Tabori , Laboratory Medicine, Hamilton General Hospital, 237 Barton Street East, Hamilton, 95 25 94 Cynthia E. Hawkins , Eric Bouffet , StephenW. Scherer 14,15 , James T. Rutka , David Ontario L8L 2X2, Canada. 56 Department of Pathology, Duke University, DUMC 3712, 2 A UGUS T 2 012 | V OL 488 | N A TU RE | 5 5 ©2012 Macmillan Publishers Limited. All rights reserved