# News this Week

Science  04 Jan 2013:
Vol. 339, Issue 6115, pp. 14
1. # Around the World

1 - Washington, D.C.
Curiosity's Big Year
2 - Brussels
E.U.'s Big Budget Decisions
3 - Geneva, Switzerland
Is It the Higgs?
4 - Karachi, Pakistan
5 - Washington, D.C.
Will Obama's Team Re-Up?
6 - Atacama Desert, Chile
ALMA Fully Online
7 - Iperó, Brazil
Nuclear Reactor Ground Breaking
8 - Kourou, French Guiana
Milky Way Monitor Lifts Off
9 - Beijing
Stem Cell Guidelines
10 - Paris
French Research, Post-Sarkozy
11 - Geneva, Switzerland
Gauging the Global Greenhouse
12 - Kumamoto Prefecture, Japan
U.N. Mercury Treaty Nears Finish
13 - Bangalore, India
Mars-Bound Mangalyaan
14 - Washington, D.C.
Permanent Cliff Dwellers

## A Look Ahead for 2013

### Washington, D.C.

#### Permanent Cliff Dwellers

The Obama administration has promised to protect research from deep cuts in negotiations with Congress on reducing the federal deficit. As Science went to press, it looked like the U.S. scientific community might temporarily avoid the fiscal cliff. But threats of budgetary catastrophe promise to be a recurring theme in 2013. The next test may come in March, when a 6-month freeze on agency budgets expires.

2. # Death of a Star

1. Yudhijit Bhattacharjee

The discovery of a nearby supernova has brought astrophysicists closer to understanding a class of stellar explosions. Along with that success came an unexpected tragedy.

Peter Nugent Drove to Work on the morning of 24 August 2011, still oblivious to the faraway cosmic explosion that would consume him for weeks ahead. Walking toward the entrance of the National Energy Research Scientific Computing Center (NERSC) in Oakland, California, he stopped to watch news vans covering protests against a shutdown of cell phone service on the Bay Area Rapid Transit system. Then Nugent, a theoretical astrophysicist who could be mistaken for a football player if he came thundering down a hall, turned his attention away from earthly matters and dived back into the otherworldly pursuit of astrophysics.

That morning, he had an urgent task: fixing a glitch in a digital pipeline that feeds astronomical images from a 1.2-meter-diameter survey telescope on Mount Palomar to the computers at NERSC. The pipeline had crashed the night before, leaving thousands of pictures waiting to be uploaded into a database, where they would be scanned by software designed to identify potentially interesting events such as a gamma ray burst. Shortly after noon, when the pipeline had been restored and all the images uploaded and analyzed, Nugent sat down to see what the system had picked from the previous night's observations as candidates worthy of follow-up. He knew that his buddy, Joshua Bloom, was likely doing exactly that a few kilometers away at the University of California (UC), Berkeley, where Bloom is an associate professor of astronomy.

Bloom—who had helped write the software for the automated search—was in fact looking through the top picks of the night, having logged in at Berkeley. And as they often did, the two began chatting over Gmail while going through the results. At 12:40 p.m., Bloom messaged Nugent that he had found an object that looked like a supernova that had gone off 400 million light-years away. “That's great,” Nugent responded. Then, seconds later, Nugent spotted another candidate that looked like a supernova in the Pinwheel galaxy, only 21 million light-years away. Because this one was so much closer to Earth, it was potentially of greater value to astronomers. “I see your $20 and raise you$100,” Nugent joked. “Dang,” Bloom replied, checking it out.

Detecting a newly exploding star or gamma ray burst tends to quicken the pulse of astronomers. Because of the transient nature of stellar explosions—they can fade away within hours to days—the moment of discovery marks the beginning of a race against time to collect data about the phenomenon. And so, within minutes of finding the object that he had labeled PTF11kly, Nugent instant-messaged another colleague named Mark Sullivan at the University of Oxford in the United Kingdom, asking if Sullivan could arrange for a telescope to start observing the object immediately.

Luckily, Sullivan was still in his office even though the local time was past 9 p.m. “Reckon it's real?” Sullivan asked, to which Nugent replied: “It is.” Sullivan e-mailed operators of the Liverpool Telescope, a 2-meter-diameter robotic instrument on La Palma in the Canary Islands off the coast of Spain. Within an hour, the telescope had begun taking images and spectra of the supernova, and around 2:30 p.m. California time, Nugent and Sullivan were looking at the first spectroscopic results, puzzling over the kind of supernova it was. After a few minutes of studying its spectral features, Nugent excitedly messaged Sullivan that the object was a type Ia—a class of supernova that shines with such predictable luminosity that astrophysicists use it as a standard candle for measuring cosmic distances. Following convention, the researchers named the object SN 2011fe.

Because of their usefulness to cosmology, type Ias are valuable finds. The one that Nugent and his colleagues had discovered was even more valuable because it had been detected just 11 hours after the supernova went off, making it the youngest type Ia discovered to date and allowing astronomers to study the explosion from an earlier stage in its progression than any type Ia seen before. Nugent sent out an Astronomer's Telegram on the Web encouraging observers around the world to follow up on the object, which the researchers would later label “the supernova of a generation” and “an instant cosmic classic.”

Nugent had already called a colleague at UC Berkeley—a quiet, 42-year-old astrophysicist named Weidong Li—to start looking at archival images of the Pinwheel galaxy in an effort to determine the supernova's progenitor. Nugent knew that Li's expertise in astrometry, the measurement of the position and motion of stars, would be instrumental in tracing SN 2011fe's history. Over the next several weeks, Nugent, Bloom, and Li would devote themselves night and day to studying the supernova's present and past.

Along with the work of dozens of others, the efforts of these three researchers would lead to a new understanding of how type Ias originate and unfold. But not all three would be around to celebrate the insights gained from SN 2011fe, which would end up generating dozens of research papers. Months later, one of their lives would come to a tragic and unexpected end.

## Core compression

Nugent, Bloom, and Li had known each other for several years before SN 2011fe lit up the sky. They had come to astronomy by very different paths.

Nugent, the son of a lawyer, got hooked on space as a kid following NASA's lunar program. His interest turned to astronomy when his grandfather gave him a telescope at the age of 12. In college, he briefly considered becoming an English major but hated rewriting, so he chose physics instead. Later, while exploring graduate schools, he ran into David Branch—a supernova expert at the University of Oklahoma—who quickly convinced him that supernovae were the most interesting things to study in astronomy. In 1996, he joined Lawrence Berkeley National Laboratory for a postdoctoral fellowship that turned into a staff position.

Li's beginnings were a world apart from Nugent's. Born to a farming couple in a Chinese mountain village, he was the first person in his district to go to college. Like Nugent, he became fascinated by supernovae, which he would later describe as “the glorious explosive stage of stellar evolution.” After earning his doctorate from Beijing Normal University in 1995, Li began working at the Beijing Astronomical Observatory, where he set up China's first systematic supernovae search using a telescope at the observatory's Xinglong station. Within a year, the survey had found six new supernovae, thanks in part to scheduling software that Li had written to specify what parts of the sky the telescope should observe when.

The success brought Li to the attention of supernova researchers elsewhere, including Alex Filippenko at UC Berkeley, who was looking for somebody to lead a supernova search he was initiating with a robotic telescope at Lick Observatory in Mount Hamilton, California. In 1997, Filippenko hired Li as a postdoc.

Nugent met Li not long after he arrived at Berkeley, at a meeting of Filippenko's research group, where Li passed around sweets he'd brought from China. Nugent didn't care for the sweets (“Bean paste is still bean paste,” he would later joke) but was impressed by the smiling and soft-spoken Li.

Within months, Li had the Lick search up and running. The very next year, the survey yielded a rich haul of 20 nearby supernovae, marking the beginning of what was to become a consistently productive run. Li's role became pivotal to Filippenko's group. “If Weidong were to be run over by a truck,” Filippenko would remark at conferences, “my whole group would fall apart.”

The youngest of the three—Bloom—came to UC Berkeley as an assistant professor in 2005 after getting a Ph.D. from the California Institute of Technology (Caltech) in Pasadena with work on gamma ray bursts. He was 30, with sparkling eyes, a blond goatee, and a knack for deadpan humor that he deployed to spice up his talks and rib colleagues.

Bloom and Nugent became friends, and in 2008 they began working together on the Palomar Transient Factory (PTF)—a project led by Bloom's doctoral adviser at Caltech, Shrinivas Kulkarni. The project was an automated search for transient phenomena including gamma ray bursts and supernovae.

Astronomers had been conducting such automated surveys for more than a decade, but Bloom wanted to take them a step further. Until then, automated searches—such as the one that Li was in charge of at Mount Hamilton—used computers to schedule observations, control the telescope, and scan images of the sky for possible new supernovae or gamma ray bursts by comparing the images with older reference images. A human being, however, still had to inspect each candidate to determine whether it was a real astrophysical object or something spurious like a speckle.

Bloom and his colleagues developed algorithms to distinguish between fake and real candidates and to determine what kind of object a candidate might be: a gamma ray burst, a nova, a variable star, or something else. By inspecting various features of a candidate—such as its brightness and the brightness of its host galaxy—the algorithms could make a probabilistic statement about the candidate, for example, classifying it as a supernova with 80% probability. The algorithms, called Realbogus and Oarical, gave PTF the ability to sift through several kilobytes of astronomical data within hours and classify thousands of new candidates from every night's observations.

## Explosion

Death comes to mortals and stars alike. For stars that end as supernovae, however, it brings ultimate glory: a flash of splendor often more brilliant than the combined brightness of an entire galaxy. Type Ia supernovae like the one that set astronomers' hearts racing on 24 August 2011, make up a special class of such stellar explosions. What makes them special is that all of them produce nearly the same brightness.

Astronomers think type Ias arise in binary star systems in which a small, dense star known as a white dwarf has been steadily accumulating material dumped onto it by a companion star. If the white dwarf happens to be composed entirely of carbon and oxygen, something extraordinary happens after it has gained enough material to approach 1.4 solar masses. The stage is set for a runaway thermonuclear reaction in which carbon and oxygen atoms fuse into nickel. The nickel decays radioactively into cobalt, which then decays radioactively to iron, powering the supernova's incandescence.

Although this theoretical model is generally accepted, astronomers have been looking for empirical evidence to confirm many of the details. A fundamental question about the progenitor system is whether the star that explodes is indeed a carbon-oxygen white dwarf. Another question is what kind of star the companion has to be in order for the exploding star to result in a type Ia.

Those questions were on the minds of Nugent and his colleagues on the afternoon of 24 August as SN 2011fe burned in the sky, continuing its ascent in brightness. After calling Li—who began looking at archived images of the Pinwheel galaxy—Nugent sent a text message to Caltech astronomer Richard Ellis urging him to start observing the supernova with the Hubble Space Telescope. But the Hubble was booked; observations had to wait a few days until it became free.

As the sun was setting over the West Coast, Nugent drove to the astronomy department on the UC Berkeley campus and walked down to a room in the building's basement from where astronomers can conduct remote observations at Lick Observatory and the Keck telescopes at Mauna Kea in Hawaii. A couple of Filippenko's students were at the controls of the Lick Observatory. Geoff Marcy, the planet-hunting Berkeley astronomer, and a student were preparing to use one of the Keck instruments. Drop everything, Nugent told them: “You need to observe this at Lick—and you need to observe this at Keck.” For the next several hours, both telescopes held SN 2011fe in their gaze, obtaining spectra.

The days that followed turned into a blur for Nugent, Li, and Bloom as they analyzed data pouring in from various telescopes on the ground and in space. In the spectrum of the supernova, Nugent found the signatures of carbon and oxygen, suggesting that the exploding star was a carbon-oxygen white dwarf as models predicted. According to one leading model, the companion star should have been a red giant—a large, bloated star nearing the end of its life, with a reddish envelope and a relatively cool surface temperature. But Li's scrutiny of archival images of the galaxy found no such star near where the supernova was now blazing, effectively ruling out a red giant as the companion.

Nugent's calculations agreed. If the companion had been a red giant or other large star, they showed, the outermost shell of material ejected by the supernova would have slammed into that star within a day or so of the explosion. Because telescopes had seen no sign of such a collision, Nugent and others concluded that the companion star had been considerably smaller than a red giant—probably a star of sunlike size still in the middle of its life.

For the next 2 weeks, as SN 2011fe got brighter in the sky, Li and Nugent burned the midnight oil to prepare their results for publication. By 9 September, a day before the supernova reached peak brightness—allowing thousands of amateur astronomers to view it with backyard telescopes—both Li and Nugent had submitted their papers to Nature. Adam Riess, an astronomer at Johns Hopkins University in Baltimore, Maryland, who years earlier had lost a lopsided game of table tennis to Li at Berkeley, sent Li a congratulatory note: “I'll bet no one has submitted a paper on a supernova by the time it reached peak! You are even faster in your science work than in ping pong! I am in awe.”

Three days later, Bloom learned of a piece of data that would help confirm the nature of the supernova's progenitor. A 0.4-meter-diameter robotic telescope on the Mediterranean island of Mallorca had imaged the Pinwheel galaxy on the same night that the supernova was detected, some 7.5 hours before the detection was made at Palomar. The images from Mallorca showed no supernova in the patch of sky where the star had exploded, even though 4 hours had elapsed since the moment of explosion. To have remained undetectable for so long, Bloom and his colleagues calculated, the exploding star must have been at most 2% the diameter of the sun—a white dwarf.

## Brightness falls

Through the fall, SN 2011fe dropped in brightness as its nickel decayed to cobalt, and its cobalt to iron. As December approached, Nugent, Bloom, and Li e-mailed back and forth to finalize press releases their institutions were drafting to announce the two Nature papers, which were due out in the journal's 15 December issue.

In the second week of December, Nugent flew to Stockholm to participate in the festivities related to the Nobel prizes. Several astronomers, including Nugent and Filippenko, had been invited to celebrate the physics Nobel, which was being awarded for the discovery of the accelerating universe. In some sense, the prize was a celebration of type Ia supernovae, whose usefulness in measuring cosmic distances was the foundation of the discovery.

On the evening of 12 December, Filippenko returned to his hotel room after dinner and checked his e-mail. Among the dozens of messages in his inbox was one from Li. The subject line said: farewell. Filippenko, who had always worried about losing Li to a competing research group, clicked on the message with trepidation.

“Dear Alex,” the message began. “Please find a seat to sit down before reading this email. I am sure you will be shocked beyond belief. By now, I should have already committed suicide.”

Stunned, Filippenko called Berkeley, where it was midafternoon. It was too late, he learned. Li had already killed himself.

In his suicide note, Li indicated that he had taken his life because of a personal family difficulty. Li apologized to Bloom and Nugent for the inconvenient timing of his death—days before the publication of the Nature papers. Li's cell phone number was on the embargoed press releases issued the week before; reporters had been calling Li's number without getting a response.

Nugent had just returned to Berkeley from Stockholm and was pulling into his driveway when he got Filippenko's e-mail bearing the sad news. Bloom got the message sitting in a hotel room in Hong Kong.

Devastated, they combed through months of their e-mail exchanges with Li, looking for clues to depression that they might have picked up on. They found nothing. Li's death would likely remain a mystery to them. All they knew for certain was that a shining star had dimmed.

3. Cell Biology

# The Immune System's Compact Genomic Counterpart

1. Mitch Leslie

Small but powerful, piRNAs protect the genome and may have other functions as well.

Parasitic DNA has infiltrated our genome and threatens our future. As in most other animals, much of the human genome derives from self-serving DNA strands known as transposons. These genetic gypsies often jump to new chromosome locations, sometimes disabling genes and even triggering cancer. In the germ line—sperm and eggs and the cells that spawn them—a transposon hopping to a new position can lead to sterility, a disaster from a Darwinian point of view. “Failure to control transposons in most animals is the surest path to extinction,” says biochemical geneticist Phillip Zamore of the University of Massachusetts Medical School in Worcester.

For that reason, a specialized group of RNA molecules known as piRNAs (pronounced “pie-RNAs”) are the superheroes of animal genomes. Discovered in the past decade, piRNAs team up with certain proteins to shackle transposons in animal germline cells. Together, these protein-RNA combos create a molecular defense that scientists liken to an immune system for the genome. Like our immune system, piRNAs and their partners can tell friend from foe, mobilize a response, and adapt to new invaders. Similarly, our genome guardians have a memory, a record of past threats.

“The complexity of this [piRNA] pathway has exploded during evolution,” says Julius Brennecke, a developmental geneticist at the Institute of Molecular Biotechnology in Vienna. The number of piRNA varieties that humans produce isn't clear, but the total could be in the millions. “It's not often that you discover something that is so abundant and that was missed for so long,” Zamore says. “It's the perfect scientific problem.”

Researchers intrigued by this problem have begun to sketch out details of how these small RNAs keep transposons in check. “We are starting to learn what's in the [piRNA] toolbox,” says molecular biologist Ramesh Pillai of the European Molecular Biology Laboratory in Grenoble, France. But biologists still don't know how cells manufacture this type of RNA, or what piRNAs might do outside the cells of the germ line. “In mammals, the transposon-silencing function is just a small piece of what they do, but it's the only piece we understand,” Zamore says. One recent study, for instance, raises the possibility that these molecules are important for learning. Researchers are also stumped as to why mice generate hundreds of thousands, even millions, of piRNA varieties that have no known transposon targets. “These are exciting times” in the field, Pillai says.

## Discovery of a new RNA

When researchers first detected piRNAs in 2001, they were just beginning to grasp the importance of so-called small RNAs. These molecules, which are typically between 18 and 40 nucleotides long and don't code for proteins, were proving ubiquitous. “Small RNAs have been harnessed by almost every single life form we know,” Pillai says. Organisms deploy some small RNAs to turn down the activity of their own genes, albeit indirectly. Before a cell synthesizes the protein encoded by a gene, it first makes an RNA version of the gene, known as messenger RNA (mRNA). The best-known types of small RNAs—small interfering RNAs (siRNAs) and microRNAs—target these mRNAs, destroying them or preventing the cell from translating them into proteins. Many organisms also enlist small RNAs to defend against pathogens. In plants and nematodes, for instance, small RNAs help destroy viral RNA.

Eleven years ago, Alexei Aravin, then a graduate student at Moscow State University, and colleagues discovered several small RNAs that shut down a transposonlike gene in fruit flies. At the time, the only hint that the molecules belonged to an unrecognized group of RNAs was that they were slightly longer than siRNAs, says Aravin, who is now a molecular biologist at the California Institute of Technology in Pasadena. In a follow-up fruit fly study 2 years later, however, he and colleagues identified more than 170 unique small RNAs that target transposons, suggesting that the insects have a specialized type of RNA for this function.

Aravin would soon come across this new class of small RNAs again, though by following a different research tack. He and other biologists were looking into the workings of Piwi proteins, which studies had indicated are necessary for fertility in several kinds of animals. Piwi proteins are part of the Argonaute family. siRNAs and microRNAs work by consorting with non-Piwi Argonaute proteins that slice up RNA molecules. Some researchers speculated that Piwis also functioned by partnering with RNAs. “It was only logical to imagine that these similar family members would also bind to small RNAs,” says molecular biologist Gregory Hannon of the Cold Spring Harbor Laboratory in New York. In 2006, Aravin and colleagues, Hannon and co-workers, and two other groups independently confirmed this hypothesis, uncovering thousands of small RNAs that collaborate with Piwi proteins in mice. Researchers realized that these small RNAs resembled the ones Aravin and colleagues had initially identified in fruit flies and declared that all of them were an RNA family unto themselves, the Piwi-interacting RNAs, or piRNAs.

piRNAs differ from microRNAs and siRNAs in several ways (see table, p. 27). As Zamore's team first reported in 2006 in Science (21 July 2006, p. 320), cells don't need the enzyme Dicer to make piRNAs. However, Dicer is essential for the maturation of microRNAs and siRNAs. Also, unlike siRNAs and microRNAs, piRNAs are exclusive to animals, occurring even in ancient groups such as sponges.

The DNA sequences that code for piRNAs are bunched in a few so-called piRNA clusters. One of the field's big mysteries is how these clusters give rise to piRNAs, notes molecular geneticist Eric Miska of the University of Cambridge in the United Kingdom. “piRNA biogenesis is still very enigmatic.” Cells likely make an RNA copy of an entire cluster and then dissect it, hewing the fragments into piRNAs. But the details of this processing remain obscure. “More than 10 proteins are involved, but we know very little about what steps they are doing,” Aravin says.

## Detecting danger

Although the workings of the piRNA system differ from those of our immune system, these two defenses face many of the same challenges. Their first job is detecting danger. piRNA clusters are crucial for this function. They contain partial and complete transposon sequences, and they serve as the memory banks for the piRNA system. “It's the way animals write down which transposons have invaded their genome,” Zamore says. Each piRNA targets transposons that contain a matching sequence to its own RNA sequence. By making piRNAs that correspond to the transposon sequences stored in the clusters, animals can keep these selfish strands in check.

But what if an animal has to contend with a transposon that it hasn't encountered before? The piRNA system relies on a nifty trick in these situations. “It makes use of the only thing that [selfish] genetic elements have in common—they move around the genome,” says molecular geneticist René Ketting of the Institute of Molecular Biology in Mainz, Germany.

As a new transposon migrates from location to location, it should eventually land in a piRNA cluster. When that happens, the transposon becomes part of the memory bank, and the animal will begin producing complementary, or matching, piRNAs to thwart the genomic interloper. Each piRNA cluster “is kind of a trap,” Pillai says. “Once a transposon falls in, you have immunity.”

Thanks to their genomic immune system, animals can recover from “infection” by a new transposon, much as you get over the flu because your immune system defeats the influenza virus. For example, in a study published in the 23 December 2011 issue of Cell, molecular geneticist William Theurkauf of the University of Massachusetts Medical School, Zamore, and colleagues followed what happened to young female flies that inherited a transposon called the P element, which they hadn't tangled with before. At first, the transposon got the jump on the insects. They were infertile and produced scant piRNAs that had any ability to control the P element. The genomic invader also unleashed other transposons that had been lurking in the flies' chromosomes. But as the flies grew older, they began to rein in the P element, cranking out piRNAs that targeted it. Moreover, the researchers found that other transposons released by the P element began falling into piRNA clusters, presumably allowing the flies to make piRNAs to counter them as well. As a result, the flies regained some of their egg-producing capability.

But how does a fly or another animal tell a transposon from its own DNA? If the immune system mistakes “self” for microbial invaders, its responses can trigger autoimmune diseases. One way that piRNAs avoid triggering genomic autoimmunity is their specificity; they key on transposons with complementary sequences. But a study of nematodes indicates that the piRNA system might deploy a second mechanism to prevent self-directed attacks, suggests molecular geneticist Craig Mello of the University of Massachusetts Medical School. He shared the 2006 Nobel Prize in physiology or medicine for discovering RNA interference: the ability of small RNAs to shut down gene activity.

To move, transposons often make an RNA copy, or transcript, of themselves that's converted back to DNA in a new place. In the 6 July 2012 issue of Cell, Mello's team proposed a novel way that piRNAs can avoid mistaking this transposon RNA for a cell's vital RNA, such as messenger RNAs. “People thought that piRNAs would target ‘aberrant RNA,’ ” that is, any sequence that differed from the animal's own RNA sequences, Mello says. He has a different take: “Our findings suggest that a foreign sequence is recognized as foreign because it's never been expressed”—used to make protein.

The researchers drew this conclusion after equipping nematodes with a fragment of worm DNA that also included a foreign sequence—instructions for making the fluorescent protein GFP. Mello and colleagues observed a curious pattern in the resulting mutant nematodes. In some of the worms, piRNAs ignored the inserted DNA, treating it as if it were a normal gene. Those worms made GFP and lit up.

But other worms remained dark because they reacted to the introduced DNA sequence as if it were a transposon and shut it down, preventing the production of GFP. These differences remained steady from generation to generation, Mello notes. “The ones that are on stay on, and the ones that are off stay off.”

According to Mello, why some inserted DNA sequences are initially expressed and others are silenced is probably a matter of chance. But if the DNA snippet is accepted and used to make proteins, the animal thereafter treats it as “self,” Mello suggests. He and his colleagues hypothesize that worms have a molecular pathway that keeps track of which DNA sequences have been active and prevents piRNAs and Piwi proteins from interfering with them. The researchers haven't pinpointed which molecules perform this job, Mello says, but they suspect an Argonaute protein called CSR-1 is the ringleader. Pillai describes this potential recognition mechanism as “an interesting idea and plausible,” adding that the Mello group's paper is “the only one which might explain the available data.”

## Taking on transposons

Once immune cells meet an intruder, they counterattack. piRNAs do the same, using a variety of measures against transposons. Some piRNAs dive into the fray. They track down transposon RNAs, and the Piwi proteins they bring along slice up the rogue strands.

Some piRNAs let others do the dirty work. In the 3 August 2012 issue of Science (p. 574), Miska and colleagues described how piRNAs boost their power by enlisting siRNAs to stifle transposons. The reason, Miska suggests, might be that although piRNAs come in many varieties—more than 16,000 in nematodes—each germline cell harbors just a few copies of each one. “They can't do much on their own,” he says. In contrast, siRNAs are plentiful.

Other animals bolster their piRNAs directly, relying on what's called the ping-pong amplification loop. Hannon's team and a group led by molecular biologist Mikiko Siomi, now at Keio University School of Medicine in Tokyo, independently described this mechanism in flies in 2007, but mice and zebrafish also take advantage of a similar process. In the ping-pong loop, piRNAs and Piwi proteins slice up transposon RNA. The resulting fragments undergo modification and join with other Piwi proteins to cut up RNA transcripts of piRNA clusters, thus making new piRNAs (see diagram, p. 26). The loop “only amplifies the useful piRNAs,” those with a target available in the cell, says Brennecke, a co-author on one of the papers.

piRNAs can fight back even if a transposon remains quiescent and hides as a stretch of genomic DNA. Researchers have found that in mice, piRNAs spur germline cells to affix methyl groups to transposon DNA, preventing its transcription into RNA and thereby blocking the rogue strand's movement to a new location in the genome. Fruit fly piRNAs stymie transposon transcription using a similar mechanism that involves molecular modification of histones, the protein spools around which DNA coils, Brennecke and colleagues reported in the 21 November 2012 issue of Cell.

These strategies lead to long-term protection, and in that regard, piRNAs have our immune system beat. After you've recovered from an infectious disease, you often will be immune to the pathogen that caused it for life—but your children and grandchildren won't be. Animals' genomic guardians, by contrast, can suppress some transposons for multiple generations, Mello's and Miska's teams revealed in the 6 July 2012 issue of Cell. For example, this genomic resistance lasts for at least 20 generations in nematodes, Miska and colleagues showed. Persistent protection makes sense: Transposons in the germ line can reactivate each generation, so locking them down long-term is beneficial.

## More than defense?

piRNAs may do more than thwart transposons. Some scientists suspect that they, like siRNAs and microRNAs, help adjust gene expression. For example, Mello and colleagues suggest that the targets of many nematode piRNAs are some of the worm's own genes, not transposons. They have shown that about 1000 of the roughly 20,000 nematode genes are under piRNA control. Many of the genes are normally turned off but switch on in worms that lack one kind of Piwi protein, Mello and colleagues reported in the 6 July 2012 issue of Cell. One possibility, Mello says, is that these genes perform functions that are useful in certain environmental conditions, such as when the worms are under stress. When times are tough, a cell in the germ line of a parent worm might rein in piRNAs, allowing the genes to switch on and helping the offspring cope with adversity.

The idea that piRNAs are tweaking gene activity gets mixed reviews from other researchers. Some remain skeptical that piRNAs ever silence genes. Even if they accept that possibility, other scientists question whether animals other than nematodes avail themselves of this gene-controlling mechanism. Few transposons trouble nematodes, so the worms might have the freedom to divert their piRNAs to new roles. “In most systems, the evidence favors transposons being the targets” of piRNAs, and not genes, Ketting says.

Also unclear is whether piRNAs function in nongermline cells. Most scientists have dismissed the possibility, as siRNAs quash transposons in these cells. Moreover, only germline cells seem to make Piwi proteins, piRNAs' collaborators.

But a discovery from neuroscientist Eric Kandel of Columbia University and colleagues suggests that piRNAs are active in the central nervous system, helping create memories. In the 27 April 2012 issue of Cell, the team reported that they had identified piRNAs in neurons from the sea slug Aplysia. The piRNAs help block the production of a protein called CREB2, which inhibits memory formation in these animals. Testing piRNAs' role in learning in other creatures shouldn't be difficult, Ketting says. If they do have a role, deleting Piwi proteins in animals such as mice or flies should cause memory lapses.

The question that has researchers scratching their heads involves the pachytene piRNAs, which are named for the stage of meiosis—the process that produces eggs and sperm—in which they appear. Mammals generate a huge number of different pachytene piRNAs—one recent study estimated the total for mice at more than 800,000, but Zamore says that value is almost certainly too low. Yet the sequences of the pachytene piRNAs do not match those of any transposons, suggesting that they aren't targeting the rogue strands. “What these pachytene piRNAs are doing—nobody knows,” Aravin says.

As piRNA researchers delve into such mysteries, some also wonder if these genomic superheroes sometimes take the day off to help a species adapt. We and other animals are alive today because, over hundreds of millions of years, piRNAs helped our recent and distant ancestors tamp down transposons. But transposons aren't necessarily all bad. They also create genetic variation in the germ line that is the raw material for natural selection. A few researchers speculate that when conditions are rough, animals might inhibit their piRNAs to unleash transposons and trigger more mutations, speeding up their evolution. That idea is “a very attractive hypothesis,” Zamore says. “I'd like to think of a way to test it experimentally.”

4. Research Funding

# Europe's €2 Billion Bet on the Future

1. Kai Kupferschmidt

This month, the European Union will pick two futuristic research proposals and shower them with up to €1 billion each. But will it be money well spent?

As the global financial crisis erupted in 2008, Dirk Helbing, a physicist and mathematician turned sociology professor, had an idea: What if all the data on human behavior could be integrated into a computer model that could predict such financial calamities—or, for that matter, revolutions, wars, the effects of overfishing, or epidemics? “Everybody saw that the models we had weren't working. And we have so much more data available today,” Helbing says.

A researcher at the Swiss Federal Institute of Technology in Zurich, Helbing envisioned a model that would help predict the societal implications of political decisions such as using bioethanol as fuel, banning high-frequency trading, or letting Greece drop out of the Eurozone. To make that possible, he says, a “planetary nervous system” would collect data on an unprecedented scale and develop “socially aware software” to analyze it.

That vision may sound like science fiction or even a dystopian nightmare. But it's the kind of idea that the European Commission, the European Union's executive body, might just deem worth €1 billion. This month, the commission will pick two winners—from a shortlist of six—in its Future and Emerging Technologies Flagship Initiative, a new funding program that aims to reward bold initiatives that attempt to reach a “visionary goal” and provide “novel benefits for society.”

In 2011, a scientific panel picked the six proposals based on a very short outline of their idea from a list of 21, and then ranked them (see table); since then, the candidate teams, each comprising many dozens of labs and companies, have spent 18 months and €1.5 million to develop a research plan as well as a Web site with splashy videos. In December, a new group of experts evaluated these proposals, and in late January, the commission will announce which two get the go-ahead. It isn't obligated to pick the top two of the experts' unpublished ranking.

The original list of 30 projects included some off-the-wall ideas, such as Matrix redone, which would allow people to learn martial arts or ride a motorbike in a virtual reality world inspired by the sci-fi flick The Matrix; another, the Personal Fabricator Network, would allow everyone to manufacture their own smart devices at home.

But even some of the six shortlisted plans aren't run-of-the-mill. There are two other modeling projects: a controversial plan to simulate the entire human brain, and another to build a “virtual patient” on which to test therapies. There's also a project to build robots that might help you dress or make coffee, one aiming to use tiny wearable sensors to monitor your body and environment for health risks, and a plan to develop new, graphene-based electronics. The two winners stand to receive €1 billion each over the next 10 years.

But some European scientists worry that the ambitions are just too grandiose—the computer models in particular have come under fire—and the funding too much. Average research projects might get €300,000 in funding, says Peter König, a computational neuroscientist at the University of Osnabrück in Germany, and “nobody has been able to explain to me why replacing 3000 good projects by one project is a good idea.”

And although a billion euros can free researchers from many burdens, largesse at this scale can also lead to waste, says Stefan Hornbostel, a sociologist at Humboldt University in Berlin who heads the Institute for Research Information and Quality Assurance. “There is a danger that it does not follow the burning scientific questions anymore, but rather a bureaucratic logic that this money needs to be spent.”

Others praise the commission for daring to dream. Ernst-Ludwig Winnacker, who held various positions in German and European science funding agencies and currently heads the Human Frontier Science Program, invokes Lewis and Clark, who explored the American West in the early 19th century. “They had no idea what they would find,” he says—but 42 years later, the gold rush started. To stay ahead, the European Union should fund pioneering research rather than areas where the gold rush has already set in, he argues.

Whether all of the money will materialize remains to be seen. The two winning teams will together receive a total of €108 million in the first 2.5 years; after that, this would go up to €100 million annually each. But half is supposed to come from member states or research organizations, and how much they will actually give is unclear. The German ministry of research, for instance, says it will wait for the final selection before deciding what contributions it can make, but it is skeptical of the whole concept of “Flagships.” Germany prefers using “tried and tested instruments … instead of new complex funding mechanisms,” the ministry said in a statement that was sent to Science.

## Extravagant claims

There's no doubt that the projects are ambitious. IT Future of Medicine, for instance, aims to collate a patient's genome, blood parameters, medical history, and other data into a “virtual patient,” on which any therapy can be tried in silico. “Everything else that is complicated or dangerous we model on computers, whether designing a car or training pilots,” says project coordinator Hans Lehrach, a director at the Max Planck Institute for Molecular Genetics in Berlin. “But we still crash patients in real life.”

When the original proposal list was whittled down, the virtual patient ended up in fifth place; the top spot went to FuturICT, Helbing's world model. But scientists have criticized that project for underestimating the world's complexity. If nobody saw the Arab Spring coming, why assume that a computer model can predict a revolution? “People have tried to build such huge models of the world before, for climate, democracy, or violence,” Hornbostel says. “But even on a theoretical level, it is completely unclear how many of the basic components interact.” “We are not building a crystal ball,” Helbing retorts. “This is a tool to understand causal relationships, to see what makes some systems stable and others unstable.”

The Human Brain Project has faced vocal opposition, too (Science, 11 November 2011, p. 748). The idea is to model everything known about neurons—including genetics, chemistry, and electrical signaling—and then stitch virtual neurons together into networks resembling the human brain. But there are big knowledge gaps at each stage of the process, and many neuroscientists doubt that anything can be learned from such a model; some argue that it would divert money from other important areas in neuroscience.

And the Flagship program's review processes may not detect weaknesses, König says. Small research grants are critically evaluated and sometimes accepted only after substantial changes; with a billion euros up for grabs, the review should be orders of magnitude stricter, he says. “But the opposite is the case. Getting more money is actually easier.” And although the projects will be evaluated independently once under way, König says that they are essentially too big to fail. “These projects are so huge and there are so many people involved that everybody wants them to be a success,” he says. “And everybody wants them to look like a success even if they aren't.”

But Ewan Birney of the European Bioinformatics Institute in Hinxton, U.K., who's involved in the IT Future of Medicine project, believes that the models' critics may be shortsighted. “People tend to overestimate what they can do in 2 years and underestimate what they can do in 10 years,” Birney says.

## Emotional robots

Qualms about the models' feasibility might give the more practical competitors an edge. One of those is a coordinated research effort into graphene, a relatively new material made of carbon atoms arranged in a honeycomb lattice that conducts both light and electricity. Andre Geim and Konstantin Novoselov, two Europeans who won a physics Nobel Prize for their work on graphene in 2010, are involved in the project. Jari Kinaret of Chalmers University of Technology in Gothenburg, Sweden, who coordinates Graphene, says that flexible electronic devices, such as an e-reader that you could roll up, are among the low-hanging fruits of this proposal.

Replacing silicon transistors in chips with graphene will be more difficult because its electrical properties make it very hard to switch such a transistor off completely. “Some of the things we promise are going to happen, some of them are not going to happen,” Kinaret says. But Flagship's long-term funding is vital to cross the “valley of death” between research and application, he says. Now, Europe does more graphene research than other parts of the world, but it obtains fewer patents.

Another project called RoboCom aims to make robots affordable and pioneer their use in natural disasters, surgery, and housekeeping; its researchers also want robots to become more sensitive and more emotional. The sixth project aims to develop tiny sensors called “guardian angels.” Integrated into clothes, they could monitor your heart rate, blood alcohol level, or fatigue; sensors in cars could help avoid crashes. They would run on energy harvested from the sun, vibrations, or temperature changes.

## A feasible model?

Winnacker points out that the selection committee doesn't have to pick projects that seem closer to application. “I would always favor the risky projects,” he says. “In the other areas, the breakthroughs may have happened already.” But Marja Makarow, vice president for research of the Academy of Finland and a former chief executive of the European Science Foundation, wonders if the top-down approach of the Flagship initiative can deliver “disruptive technologies” at all. Such technologies usually emerge from high-class, basic research, she argues: “Nobody asked for the World Wide Web or light-emitting diodes.”

For now, the six candidates are sitting tight and awaiting who gets the nod. The final choice “will show just how much risk the E.U. is willing to take,” says Wolfgang Boch, who coordinates the initiative at the European Commission in Brussels. But in the end, the point isn't just to develop new robots or models, Boch says. “The Flagships weren't created to launch two projects, but to see whether this is a feasible funding model,” he says. And that's a question that—at least today—no model in the world can predict.