# News this Week

Science  03 Jun 2011:
Vol. 332, Issue 6034, pp. 1132
1. # Around the World

1 - Rome
Manslaughter Trial Approved for Quake Scientists
2 - Berlin and Bern
Germany and Switzerland Nix Nuclear Energy
3 - Washington, D.C.
NASA Aims to Grab Asteroid Dirt
4 - Brasília
Proposed Brazilian Forest Law Causes Furor
5 - Germany
E. coli Outbreak Rages On

## Rome

### Manslaughter Trial Approved for Quake Scientists

Bogus predictions of earthquakes can cause a public panic. But seven scientists and technicians who analyzed seismic activity ahead of the devastating earthquake that struck the Italian town of L'Aquila on 6 April 2009 are about to find out whether the flip side of the coin—erroneously assuring citizens that they are not in imminent danger—could actually be a crime.

On 20 September, the seven will go on trial in an Italian court for manslaughter, a judge decided last week. Last year, it appeared that the charges would be based on a failure to provide residents with adequate warning of the magnitude-6.3 earthquake that killed 308 people. But the case, which the judge ruled had sufficient grounds to go ahead, may instead center on accusations that the defendants gave undue reassurances to residents about their safety in comments that prompted many to remain home despite a series of prequake tremors leading up to the main event. For more, see page 1135.

## Berlin and Bern

### Germany and Switzerland Nix Nuclear Energy

In the wake of the Fukushima disaster, the governments of Germany and Switzerland have announced plans to end the use of nuclear energy. On 30 May, German Chancellor Angela Merkel announced that the country would shut down all of its 17 plants by 2022. Switching to renewable sources without relying on nuclear power is “a great challenge, but above all it is a huge opportunity for future generations,” Merkel said. Meanwhile, following the largest antinuclear protests ever held in Switzerland, the Swiss government announced on 25 May that it plans to close the last of its five reactors by 2034. Parliament will discuss the plan next month, and there could be a referendum. Nuclear plants generate about 23% of Germany's and 40% of Switzerland's electricity.

Also last week, European Union countries agreed on the criteria for so-called stress tests to determine the safety of the E.U.'s 143 nuclear reactors. The tests were slated to begin on 1 June. http://scim.ag/_nix, http://scim.ag/swiss_nuclear

## Washington, D.C.

### NASA Aims to Grab Asteroid Dirt

NASA has announced the next medium-class science mission to explore the solar system. The winner of a three-way competition is a mouthful: Origins Spectral Interpretation Resource Identification Security Regolith Explorer (OSIRIS-REx). It is a spacecraft that would scoop as much as 2 kilograms of rocky soil off a 600-meterdiameter asteroid in 2020 and return the sample to the Utah desert in 2023 for laboratory analysis. While strictly speaking a science mission to better understand the nature and origins of a primitive solar system building block, OSIRIS-REx dovetails nicely with President Barack Obama's plans to send astronauts to an asteroid by 2025.

## Songs From the Sea

An underwater opera now playing in Berlin gives audiences a chance to hear from one of the most extreme environments on Earth. AquAria_PALAOA is being staged in the city's striking art deco Neukölln public swimming pool and incorporates scientists' recordings of the soundscape underneath Antarctica's Eckström ice shelf.

The stars of those recordings are Weddell seals. “They are some of the world's most accomplished underwater singers,” with more than 50 songs in their repertoire, says Lars Kindermann, a biophysicist at the Alfred Wegener Institute for Polar and Marine Research in Bremerhaven, Germany, which runs the Antarctic recording project.

Kindermann advised the opera's artistic director, Claudia Herr, on both the music and the story, which tells of a woman's search for immortality. He gave Herr, a swimmer-turned-opera singer, a spectrogram of seal vocalizations, which she and the composer were able to use “sort of like sheet music,” he says. “[They] were fascinated. She could almost sing from the spectrograms.” The vocalists even sing some songs underwater, wearing oxygen tanks and special microphones.

Weddell seals use their voices for survival, echolocating breathing holes in the ice and alerting other seals to the fresh air, Kindermann suspects. The libretto, too, tells “a story of survival in an extreme environment. It is about time, youth and age, love, finding community and loneliness,” he says. The 1 May premiere impressed audiences and critics; four more performances of the opera will run in mid-June and mid-September.

3. DNA Nanotechnology

# DNA Nanotechnology Grows Up

1. Robert F. Service

Once dismissed as molecular parlor tricks, techniques for piecing ultrasmall structures together with DNA are starting to prove their worth in serious research.

SNOWBIRD, UTAH—All scientists face rejection when their proposals are dissected and their papers picked apart. Ned Seeman's worst slapdown came after what he considered at the time his greatest success. Seeman, an x-ray crystallographer at New York University (NYU) in New York City, had spent more than a decade working out the details of how to use DNA not as the master genetic control, but as a construction material for making molecular beams, joints, and girders that could be programmed to weld themselves together in molecular triangles, squares, and other simple shapes. Finally, in 1991, Seeman and colleagues at NYU managed to forge a DNA cube, the first three-dimensional (3D) nanoscale object in which the position of each atom was programmed, defined, and known.

Seeman submitted his manuscript to (ahem) Science. But it was promptly returned with a comment from a reviewer who, as Seeman recalls, asked, “Where is the biology?” Today, Seeman is regarded as the founder of the burgeoning field of DNA nanotechnology, in which researchers arrange DNA's four building blocks—molecules of adenine (A), guanine (G), cytosine (C), and thymine (T)—so that they assemble themselves into whatever structures the scientists want to build. After cubes, researchers in the field went on to build octahedra, rafts, smiley faces, and even a collection of strands that look like a Rock 'Em Sock 'Em Robot. Many critics are still unconvinced. “People always say, ‘Yeah, that's cute. So what? What are you going to do with it?’” Seeman says. Or as Hendrik Dietz, a biophysicist at the Technical University of Munich in Germany, says he often hears it, “‘It's an amusing but pointless exercise.’”

“Pointless” may seem harsh for an endeavor that represents nanotechnology's closest thing to building materials atom by atom from the ground up. But until now, the critics had a point: The field has searched for relevance.

No longer. DNA nanotechnology has left its childhood behind and entered adolescence. Like a teenager who clings to parts of childhood, some DNA nanotech continues to be playful and impractical. (The molecular robot boxer can't actually throw a punch.) But the field is also growing in strength and power. It's now churning out applications that are helping researchers map the atomic structure of proteins and compute inside cells (see sidebar, p. 1142), and soon may even start tracking and curing diseases. “We're still playing around,” says Andrew Turberfield, a physicist at the University of Oxford in the United Kingdom. “But we've gotten good enough that we can do some interesting things.”

## Flying fish

Getting to this stage has been a slog. Thirty years ago, Seeman, then a young assistant professor, was struggling to make a mark in his field of protein crystallography, in which researchers interpret the way beams of x-rays bounce off copies of a protein aligned in a crystal to work out the molecule's structure. Many proteins rebuff efforts to force this order. And Seeman was struggling. “I was confronting this fatal progression of no crystals, no crystallography, no crystallographer,” he recalls.

Seeman wondered whether DNA might help. DNA is most commonly thought of as a linear chain of A's, T's, G's, and C's. That's true, of course. But thanks to the propensity of individual nucleotide bases of DNA to pair with one another (T's with A's and G's with C's), DNA in cells forms a double helix, in which two strands of complementary nucleotides zip together. Seeman and others knew that under special circumstances, such as when DNA is copied during cell division, double-stranded DNA can unzip and begin binding to other strands, forming DNAs with branch points, not perfectly linear molecules. By tweaking the DNA's base pair sequence, researchers quickly realized that they could make artificial DNA branches with four arms, six arms, or more.

It was these branched DNAs that gave Seeman his eureka moment. In a tale that has risen to legend in the field, Seeman was drinking a beer in a bar in Albany when an image of the M. C. Escher woodcut Depth popped into his head. The woodcut depicts dozens of flying fish soaring in formation, with head and tail, left and right fins, and top and bottom fins all oriented the same way. Seeman realized that artificial DNAs with six arms (front, back, up, down, left, and right) could be tailored to link up into a regular 3D cubic lattice with a large empty space at the center of each cube. And as a crystallographer, Seeman immediately envisioned that such an array could be used to trap copies of a single protein in the voids and get them to line up with the same orientation. In other words, he imagined a tool for determining the structure of virtually any protein at will.

In 1982, Seeman and colleagues laid out their ideas for creating such a lattice and other complex nanostructures in the Journal of Theoretical Biology. Actually making the structures was a lot harder. “It took a long time to figure out how to get these experiments to work,” Seeman says. Among the challenges were learning how to outfit double-stranded DNAs with single-stranded tails that would link up with a complementary tail on another DNA fragment and how to make normally floppy DNAs rigid enough to form stable structures.

A series of advances cleared these and other hurdles. Over the next few years, Seeman's lab turned out triangles, squares, and other shapes. Then came the 1991 cube, and by 1998 Seeman's team had figured out how to assemble such parts into an extended two-dimensional array. Even then the field remained small, as building new structures required painstaking effort to design and synthesize all the component DNA strands. And synthesizers could churn out DNAs only hundreds of base pairs in length, a constraint that limited the complexity of the final structures.

That changed 4 years ago when Paul Rothemund, a chemist at the California Institute of Technology in Pasadena, and colleagues developed a technique called DNA origami. He and colleagues started with a 7000-base-pair viral genome for which the entire sequence was known. Next, using a computer, they modeled how they would need to fold this single strand over and back upon itself to create a desired shape. Then they synthesized 250 short “staple” DNA strands designed to bind to sections of the DNA that ended up next to each other when folded, holding the structure in shape. Finally, the researchers added the staples to the viral genome, heated the brew, and cooled it down. Presto, the now iconic image of the DNA smiley face. “Seeing the pictures, my jaw dropped,” says William Shih, a DNA nanotech expert at Harvard Medical School (HMS) in Boston. “This was a game changer for the field.” Adds Dietz: “With origami, it's gone ‘whoosh!’ to a completely different scale. Now we can make structures in the megadalton range with 16,000 base pairs where the position of every base pair is at a precisely registered position.” And since Rothemund's achievement other groups have added versions of the technique to fold DNA origami in 3D, as well as computer-aided design programs able to automate the task.

## Construction boom

These advances haven't quieted all of the critics. At a Foundations of Nanoscience meeting held here in April,* a prominent U.S. chemist who asked not to be named said that he still thought the field lacks widespread utility. But Shih and other proponents argue that is beginning to change. “Some people see a DNA robot and say, ‘You can build a robot,’” Shih says. “Some people see a robot and say, ‘You can build anything.’”

Maybe not anything yet. But Seeman, for one, is now aiming to complete his original dream of using DNA nanostructures to help solve protein structures. In 2009, Seeman and a host of collaborators reported in Nature that they had stitched together a series of triangles into a rigid crystalline 3D lattice with rhombohedral voids. Then, using x-ray crystallography, they mapped out the structure of their crystal with a resolution of 4 angstroms. When he showed an image of the crystal at the meeting, Seeman exclaimed: “This is the most exciting slide I've ever shown. I know where every atom is.”

More or less, that is. Getting a picture of a molecule with a 4-angstrom resolution suggests that the arms are still wiggling around too much to determine the structure of copies of a protein held within. So Seeman's team is still working on getting its lattices to diffract at a higher resolution and on anchoring copies of a protein to be imaged inside each cell. If they succeed, they will be able to work out the shape of the protein by carrying out x-ray diffraction on the combination and then subtracting out the scaffolding holding the protein in place.

While Seeman's group tinkers with its protein scaffolds, researchers in other groups have been making progress with alternatives. At the meeting, for example, Shih reported that he and colleagues had for the first time used DNA nanotech tools to map the structure of a previously unsolved protein, using nuclear magnetic resonance (NMR) spectroscopy. The technique works by identifying the magnetic signature of atoms in proteins relative to their neighbors. By knowing each atom's neighbors, researchers can piece together the structure of an overall protein, in a process much like solving a complex jigsaw puzzle. But the technique works only for modest-sized proteins, for which sorting out the interactions between neighbors is manageable.

Tweaks to standard NMR have provided researchers with additional bits of information for helping solve their molecular jigsaws. In 1997, for example, researchers at the National Institute of Diabetes and Digestive and Kidney Diseases spiked a protein-containing NMR solution with a compound that spontaneously forms liquid crystals, materials that flow but have a regular molecular orientation like a crystal. They found that as the protein molecules tumbled around in solution and repeatedly banged into the walls of this soft crystalline material, they wound up spending ever-so-slightly-more time in one particular orientation than in others. The subtle preference biased the NMR results enough for the researchers to spot clues such as the angle between two atoms bonded in a protein. That information made it possible to solve structures for several proteins that reside in cell membranes and are nearly impossible to crystallize. A big drawback to the technique and later variations of it is that many cell-membrane proteins can stay in solution only with the help of detergents, which often tear apart the liquid crystals.

In 2007, Shih and colleagues replaced liquid crystals with origami-based DNA nanotubes that weren't affected by detergents. They showed they could solve the structure of a transmembrane protein domain of a T cell receptor. The structure had previously been determined by other methods, so it wasn't clear whether the technique could be used to solve unknown proteins. At the April meeting, Shih reported that he and his colleagues had used the origami nanotubes to solve the structure of a previously unsolved membrane protein known as UCP2, which helps govern insulin secretion in pancreatic cells.

That's not all. Turberfield's lab at Oxford reported online 10 January in Nano Letters that it had used related DNA nanotech tools to improve another protein structure determination technique called cryo–electron microscopy. In this case, Turberfield's team obtained high-resolution images of a hard-to-crystallize protein known as a G protein, the receptor that binds it, and the two paired together.

DNA nanotech's growth isn't limited to mapping proteins. At Kyoto University in Japan, chemical biologist Hiroshi Sugiyama has turned to DNA nanotechnology to help him watch protein catalysts carry out reactions in real time. Sugiyama and colleagues at Kyoto and the Japan Science and Technology Corp. in Tokyo reported online 15 January 2010 in the Journal of the American Chemical Society using DNA origami to stitch together what looks like a minuscule picture frame. They then spanned the frame with two nearly identical double-stranded DNAs tailored to interact with a protein called M.EcoRI, which adds methyl groups to specific sites in DNA as a key part of cellular development.

M.EcoRI works by bending its target DNA by as much 59° in order to insert its methyl group. To see this feat in operation, Sugiyama and his colleagues engineered two DNA strands and stretched them across the frame. The first, 64 bases long, was taut; the other, with 74 nucleic acid bases, was floppier. They then watched the protein with a fast rastering atomic force microscope. M.EcoRI had a far easier time stuffing its methyl group into the longer, relaxed strand, suggesting that DNA's structure plays a big part in how it is modified. Seeman, for one, says he's impressed by the results: “It's a chemist's dream to watch individual reactions happen.”

Biophysicists are also looking to DNA constructions to help them investigate molecules one at a time. At the meeting, Dietz reported that he and his colleagues are using DNA origami to improve a now-standard set of biophysics tools to see what happens to proteins and DNA as they are pulled apart. In the standard approach, researchers use lasers as molecular tweezers to trap plastic beads in a particular location in solution. Then they attach linkers to these beads and anchor a protein or other target molecule between the two linkers. By making slight adjustments to the lasers, researchers can pull the beads apart and bring them back together to see how the changes in tension affect a protein's ability to fold, among other properties.

But the technique has limitations. For starters, current linkers are floppy. This blurs the resolution of techniques used to track the captured protein. What's more, once the protein is pulled apart, the linkers drift away from one another, so the experiment cannot be run in reverse or repeated. And in many cases, the floppy linkers must be yanked so hard to get the proteins to move that they pull the protein apart.

So Dietz and his colleagues replaced the usual floppy linkers with stiff rods made from DNA origami containing as many as 18 helical tubes each. They haven't tried them on a working protein yet. But initial tests have shown that the rigid DNA bundles are better at transferring force to their targets, making it possible to move the pinned molecules with less than half the applied force. That should make tracking the effect on proteins easier. The rigid linkers should also stay in place once an experiment is run, allowing it to be reversed. “We're excited,” Dietz says. “If you can build on the scale of biological macromolecules, it opens up new areas of scientific exploration.”

Someday, DNA nanotechnology may also push past basic science to find real-world applications. Shawn Douglas, a postdoc in the genetics technology lab of George Church at HMS, is working on what he calls a DNA origami nanorobot designed to seek out and destroy cancer cells. Douglas's “robot” looks more like a hollow cylinder some 60 nanometers long and 25 nanometers across. He built it from DNA origami and stapled it closed using DNA strands called aptamers, which in this case were designed to bind specifically to molecules specific to cancer cells. He then loaded the cylinders with fluorescent immune-system proteins that bind to cancer cells and induce apoptosis and added them to cancer cells in an in vitro assay. The loaded cylinders bound to their targets, released their cargo, and killed up to 40% of the cells. The work has a way to go before it will be ready for patients. Nevertheless, “it's beautiful data,” says Tim Liedl, a condensed matter physicist at Ludwig Maximilian University in Munich, Germany. “It's getting us closer to the vision of something patrolling our own bloodstreams.”

Certainly DNA nanotech has not reached that level of maturity. But Shih and others say it's growing up fast. “In the next 5 years, this is where we're going to make a real contribution, getting control over small collections of individual molecules,” Shih says. By then the teenager may find itself a young adult.

• * 8th Annual Conference on Foundations of Nanoscience, 11–15 April, Snowbird, Utah.

4. DNA Nanotechnology

# Next Step: DNA Robots?

1. Robert F. Service

After years of trying—and failing—to use DNA's ability to store and manipulate information to build a DNA computer, the field is finally advancing by going back to DNA's biochemical roots.

DNA nanotechnology isn't only about using DNA as a set of tiny Tinkertoys. Researchers in the field have long sought to use DNA's ability to store and manipulate information to build a DNA computer. But after years of trying—and failing—to outdo conventional computer technology, the field is finally advancing by going back to DNA's biochemical roots.

The appeal of a molecular computer is easy to see. A single gram of dried DNA contains roughly as much information as 1 trillion compact discs. And the fact that life exists and evolves shows that this information can be stored, read out, manipulated, and copied. In short, it can be used for computation. In principle, the selective binding and unbinding of complementary DNA strands can even be used to process vast amounts of information in parallel.

In 1994, Leonard Adleman, a computer scientist at the University of Southern California in Los Angeles, harnessed that power to solve a so-called Hamiltonian path problem, a version of the well-known “traveling salesman problem” that computer scientists use as a benchmark for tough computations (Science, 11 November 1994, p. 1021). Later, Adleman and other researchers moved on to far more complex problems. The idea of using DNA to perform conventional computations, however, quickly fizzled out.

“We gave up on that back around 1998,” says Erik Winfree, a DNA computation expert at the California Institute of Technology (Caltech) in Pasadena. DNA computers, Winfree notes, are slow, error-prone, and difficult to scale up to perform millions of operations. That makes them impractical for tasks that microelectronics does well.

Instead, Winfree and others argue, DNA computers should play to their strength: processing information inside organisms or other wet environments where conventional computer chips can't go. “The real application is making slow, crappy computers that can talk to cells,” enabling them to do things such as diagnose and treat diseases, says William Shih, a biological chemist at Harvard Medical School in Boston. “You don't need [Intel's] Core 2 Duo processor to do that.”

A striking example of this approach came last year from the laboratory of Niles Pierce, one of Winfree's colleagues at Caltech. In an article published online 7 September 2010 in the Proceedings of the National Academy of Sciences, Pierce and Caltech colleagues described how they had used pairs of small synthetic molecules of DNA's short-lived cousin RNA to diagnose and kill tumor cells in vitro. The first RNA strand was designed to recognize and bind to an RNA unique to cancer cells. Latching on to the cellular RNA caused the strand to expose a binding site to which the second RNA could attach itself. That second binding, in turn, unleashed a chemical cascade that made the cancer cell think it had been infected with a virus and triggered the cell to kill itself. Non-cancer cells, meanwhile, were spared.

Pierce and his colleagues also reported online 31 October in Nature Biotechnology that a related strategy could be used to image RNA expression inside cells. To researchers in the field, such sensor-triggered step-by-step processes represent biological computer programs.

Chemists are also getting in on the act, using tiny DNA “walkers” to help them build molecules. DNA walkers are rudimentary DNA robots designed to move step by step down a linear track. The robots' legs are DNA strands that bind to specific complementary DNAs on a predesigned surface. Although there are many types of walkers, most take a step each time researchers add a specific DNA snippet, known as a “fuel” strand, to their brew. Each fuel strand acts like a computer command telling the walker what to do next. The first fuel strand binds the site on the track holding the back “leg” of a two-legged walker, causing it to unbind from its DNA partner on the surface, and then bind to another DNA sequence past the front leg. Another snippet is then added to move the second leg forward, and so on.

In 2010, three groups took the idea a big step forward by putting DNA walkers to work as construction crews. In the 13 May 2010 issue of Nature, Nadrian “Ned” Seeman, a DNA nanotech expert at New York University, and his colleagues described creating a fleet of DNA walkers, each with four legs and three DNA arms designed to pick up and move various pieces of molecular cargo. Seeman's DNA walkers linked their successive loads into a growing molecule, creating perhaps the world's smallest assembly line.

Groups led by Andrew Turberfield, a physicist at the University of Oxford in the United Kingdom, and David Liu, a chemist at Harvard University, built related construction robots a short while later. And earlier this year, Turberfield and his team reported in Nano Letters that they had programmed their cargo-carrying walkers to move in desired directions down a track with multiple branches that are preprogrammed to display synthesized DNA with binding sites specific to DNA stretches on the legs of the walkers. In the future, Turberfield says, researchers might be able to program multiple robots to construct a wide variety of different molecules simultaneously.

For now, these and other DNA machines are slow and simple. But they've already proved that they can manipulate information in ways that microelectronics may never be able to match.

5. Physics

# Possible Sighting of Dark Matter Fires Up Search and Tempers

1. Yudhijit Bhattacharjee

A second experiment may have spotted hypothetical dark matter particles called WIMPs, but its leader's take-no-prisoners attitude has competitors steaming.

It's not hard to imagine Juan Collar as a matador. He is Spanish, for one thing, and he certainly seems to relish waving a red flag in front of his rivals. Collar's arena doesn't involve charging bulls or flashing swords, however. The University of Chicago cosmologist is a contender in the intensely competitive race to find dark matter: the mysterious, invisible stuff that is thought to constitute 80% of the universe's mass. And in recent weeks, the 47-year-old researcher has emerged as the bad boy of dark matter detection.

Collar made waves in May at the American Physical Society meeting in Anaheim, California, when he announced preliminary results from an experiment that he has been running inside an abandoned mine in Soudan, Minnesota. The results suggest that Collar's group may have spotted a dark matter signal in their detector, which goes by the name Coherent Germanium Neutrino Technology (CoGeNT). If the results hold up to scrutiny and are confirmed by more observations from Collar's experiment, it will be the first corroboration of a similar signal that an Italian underground observatory called DAMA has been seeing for 10 years. That would considerably strengthen the case that researchers have indeed detected the weakly interacting massive particles (WIMPS) that some theorists have proposed as the main building block of dark matter.

But that's not the only reason Collar is drawing attention. In talks over the past month, he has taken his sword to a rival experiment called XENON, led by Elena Aprile of Columbia University, which has long challenged DAMA's assertion of having detected dark matter. Although Collar has himself expressed skepticism about DAMA's results in the past, he says Aprile and her colleagues are relying on shoddy science and flawed analysis to debunk DAMA's claim.

“I have not seen a sloppier job than what XENON is pushing,” says Collar, who seems to straddle a youthful rebelliousness and middle-aged curmudgeonliness. “They cannot fool all of the people all of the time.” Collar's stinging remarks—one of his slides describes a particular XENON result as “pure, weapons-grade balonium”—have been countered by scalding rebukes from Aprile, who calls Collar's attack on XENON a “Spanish Inquisition.” She says Collar should be more concerned with the fact that CoGeNT's result contradicts yet another, longer-running dark matter experiment located in the same mine, which has not seen a dark matter signal even though its detector uses the same target material as CoGeNT's: germanium.

Controversy is nothing new to the field of dark matter research. In fact, it is only to be expected in the search for anything that has been hypothesized but not yet detected convincingly. But Collar's taunting of XENON appears to have heated up the debate over DAMA's claim to unprecedented red-hot temperatures. “There's a little bit of Nobelitis going around,” says Daniel McKinsey, a particle physicist at Yale University who was once a XENON collaborator. “The stakes are high, and there are strong personalities involved.”

He's referring not just to Collar and Aprile, both of whom have a reputation for being pugnacious, but also to Rita Bernabei, leader of the DAMA project, who is known for her no-nonsense air and impatience with critics. The three scientists may be looking for WIMPs, but their interactions with one another have been anything but weak, and it's safe to say that none of them is a wimp.

## Complicated quest

The existence of dark matter was proposed decades ago when astronomers realized that the mass of stars, dust, gas, and other ordinary matter could not provide sufficient gravitational glue to hold a galaxy together as it spun around. There had to be invisible matter that endowed galaxies with most of their mass. Theorists hypothesized that this mysterious stuff was made of particles that were massive and barely interacted with the ordinary matter of the universe: WIMPs. Since then, rival hypotheses about what constitutes dark matter have faded in popularity (Science, 21 July 2006, p. 287), leaving WIMPs as the leading candidate.

Astronomers and particle physicists have launched a variety of attempts to detect dark matter through astrophysical observations in space and underground particle detectors. The search is complicated not just because WIMPs, if they exist, can barrel through a lot of ordinary matter without leaving a trace, but also because researchers don't know how light or heavy a WIMP might be.

In December 1998, the DAMA collaboration, led by Bernabei, a physicist at the University of Rome, reported that it had seen a dark matter signal. The collaboration's detector was an instrument equipped with nine 10-kg crystals of sodium iodide, a material that scintillates, or produces a tiny flash of light, when its nuclei or electrons recoil after being hit by an incoming particle. The detector, running inside the Gran Sasso mountain in central Italy, had witnessed a pattern of flashes that Bernabei's group claimed could have been produced only by WIMPs colliding with nuclei in the crystals.

The clinching evidence, the researchers said, was a seasonal fluctuation in the scintillation that was in line with a widely accepted picture of WIMPs. As the sun orbits the galactic center, the solar system swoops through our galaxy's “dark matter halo” like a car driving through rain. Meanwhile, Earth is also circling the sun, moving sometimes into the “rain” of dark matter particles—WIMPS—and sometimes away from it (see figure, p. 1147). As a result, the number of WIMPs striking a detector on Earth should peak during the summer and dip in winter. That's exactly the pattern that Bernabei's group reported seeing in its signal.

Many researchers were skeptical about DAMA's claim, in part because the group would not make all of its data available for scrutiny. An independent examination of both published and unpublished data was required, critics said, because of the complicated analysis involved in extracting a WIMP signal from the background noise of mundane events that produce similar flashes of light, such as incoming gamma rays.

Although DAMA reported seeing the annual modulation year after year, skeptics remained unconvinced, much to the frustration of Bernabei and her colleagues. At the top of its home page on the Web, DAMA posted part of Rudyard Kipling's poem If to convey the group's feeling of having been treated unfairly. It reads: “If you can bear to hear the truth you've spoken/twisted by knaves to make a trap for fools,/… you'll be a Man my son!”

## Hope for the WIMPs

The problem was that other detectors did not see a dark matter signal. One effort was led by Aprile, head of the XENON collaboration, which built a detector by hooking up electronics to a 15-kg vat of liquid xenon. The technology is able to measure both light flashes and electrical charge resulting from nuclear and electronic recoil in xenon atoms when they are struck by external particles. Aprile, like Bernabei an Italian, ran the experiment inside Gran Sasso National Laboratory between March 2006 and October 2007 and failed to see a WIMP signal, which was seen as a strike against DAMA's claim. A second version of the experiment, using 62 kg of xenon to provide much higher sensitivity, did not see any WIMPs either, Aprile reported in May in a talk at the Space Science Telescope Institute in Baltimore, Maryland.

Another blow to DAMA came from results of a U.S.-based experiment called the Cryogenic Dark Matter Search (CDMS), whose detector consists of germanium and silicon crystals cooled to near absolute zero. The experiment is run by a collaboration of 14 universities, including Stanford; the University of California, Berkeley; the California Institute of Technology, and others. The instrument offers the advantage of being able to distinguish between nuclear recoil events—likely to be caused by a WIMP striking the target material—and electron recoils, which are caused by non-WIMP events. Deployed in the Soudan mine since 2003, the detector has not seen WIMPs.

In December 2009, Collar and his colleagues began running the CoGeNT experiment in the Soudan mine, right next to CDMS's detector. Although CoGeNT cannot distinguish between electron and nuclear recoils the way CDMS can, it can measure the tiniest of charges produced in the germanium crystals. That allows it to detect very low energy events, which would be the case if WIMPs are light. “It makes CoGeNT the most sensitive instrument available to detect light WIMPs,” Collar says.

In 2010, Collar's group reported that the detector had seen some events suggesting evidence for light WIMPs. In April 2011, after a fire in the Soudan mine caused a temporary shutdown of both CoGeNT and CDMS, Collar and his colleagues analyzed the 15 months of data they had collected. They discovered an annual modulation in the signal that was consistent with DAMA's findings, although the statistical significance of 2.8σ was too low to support any chest-thumping assertions. And so, in talks describing that result, Collar has been careful to mute his excitement, noting that the modulation needs to be seen consistently over the coming years.

Collar says CoGeNT does not claim to have detected WIMPs, as DAMA has done. “We want to make 100% sure that nobody accuses us of making a claim based on these very limited statistics,” he says. “We just want to present the facts, share the data, and take two steps back. Even if we had a stronger statistical case, it would be one for the presence of a modulation, not the presence of WIMPs.”

CoGeNT's findings, which Collar was preparing to post on the preprint server arXiv this week, have given DAMA researchers a glimmer of hope that the tide of scientific opinion may at last be turning in their favor. “The results of CoGeNT are very intriguing, and we wait with great interest to read the scientific paper,” Bernabei told Science in an e-mail interview.

Rafael Lang, a postdoc at Columbia and a member of XENON, says the obvious question Collar needs to answer is why CDMS has not seen the events that CoGeNT has, especially as both detectors use germanium and are located in the same mine. And in Lang's view, the ability to distinguish between electron and nuclear recoils makes CDMS the detector to beat.

Collar says part of the difference arises because of CoGeNT's ability to detect low-energy events, and part of it lies in statistical methods used to analyze the data. He argues that CDMS researchers could have tossed out the dark matter signal from their data. Blas Cabrera, a physicist at Stanford University in Palo Alto, California, who is a spokesperson for the CDMS collaboration, says CDMS stands by its results. “We believe our paper is pretty solid,” Cabrera says, adding that CDMS's findings largely rule out a light WIMP of the kind that CoGeNT is entertaining.

Although Cabrera acknowledges that CoGeNT has the “world's best charge measurement capability,” he does not believe it gives CoGeNT an advantage over CDMS when it comes to detecting light WIMPs. He is reluctant to poke holes in CoGeNT's analysis, however, and prefers to stay out of the squabble. Collar's “approach has been like a ton of bricks that doesn't encourage proper scientific discourse,” he says.

## Above the noise

In Collar's view, DAMA's assertive claim combined with the refusal to show all of its data has made some in the field overly eager to prove the group wrong. He singles out XENON's researchers for special criticism.

Collar says XENON's Aprile and others should acknowledge that liquid xenon is not such a great target for detecting light WIMPs. Because xenon is a lot heavier than germanium or sodium iodide, he says, a xenon nucleus will recoil much less if a WIMP strikes it. As a result, it might produce a signal too small to be detected reliably over background noise.

In a critique of XENON published last year, Collar and Yale's McKinsey pointed out that XENON's analysis had relied on a flawed and incomplete understanding of how a xenon nucleus behaves when it has a small amount of energy deposited on it, as would be the case in an encounter with a light WIMP. XENON researchers have since redone some of the analysis, but Collar says Aprile and her colleagues are still a long way from drawing unbiased conclusions. “Whenever they face an uncertainty,” Collar says, “they choose the route that helps them obtain the best limits.” In other words, they choose an interpretation that rules out WIMPs. And Collar says XENON's researchers are not being transparent about all of their uncertainties and assumptions.

“Juan is losing his mind,” says Aprile, scoffing at the notion that XENON might be biasing its results deliberately or otherwise. She says Collar doesn't dare to bring up these criticisms in front of her—the two had a friendly exchange at a recent symposium at Princeton University, she says—because “I'd shut him up pretty quickly.” And she says there is a perfectly good reason she and other researchers have gone after DAMA's results in recent years. Their claim is so strong that it “calls for a response,” Aprile says.

Aprile laments the growing rivalries within the field and acknowledges that some of the friction goes beyond scientific disagreement to the clashing of egos. When she and Bernabei first met more than a decade ago, “Rita was very charming, extremely kind, praising me,” Aprile says. “Then, I got this proposal approved [to start the XENON experiment in Gran Sasso]. All of a sudden, she didn't even care to say hi anymore.”

Bernabei agrees that dark matter research is showing its dark side. “In my opinion over the past 10 years, some activities in the field [have] become more aggressive than competitive,” she said in an e-mail to Science. She will not comment on her rift with Aprile. All that matters in science, Bernabei says, is “serious, reliable and dedicated work with good detectors and methodological approach.” And in that regard, she says, “XENON has many questionable aspects both on experimental and interpretative parts.”

Some researchers are more conciliatory. XENON's Lang says the data from the different groups “may not be so irreconcilable” after all. “There's only one truth in science,” he says. “The conflict we are having could be the result of uncertainties in the astrophysics or uncertainties in the detectors. Maybe our understanding of dark matter has to be different. We could find ways to say that dark matter does not interact with xenon at all but interacts with germanium.” But Lang says he does not understand why the discussion has to be so heated.

Collar says he is taking stabs at XENON not for sport but for the sake of the science. “I certainly like for people to do a proper job, including myself,” he says. He is eager to resume the CoGeNT experiment after officials complete safety inspections of the mine this month.

Collar's criticisms of XENON—no matter how sharply worded—will help the field in the long run, says Dan Hooper, a theorist at Fermi National Accelerator Laboratory in Batavia, Illinois. “On one hand, Juan can come across as abrasive to some people, and I don't know if this style does him many favors,” Hooper says. “On the other hand, I am very glad to have people like Juan aggressively challenging claims being made by other scientists. He is an exceptional scientist, and the kind of scrutiny he brings to the table ultimately helps the scientific community to better understand the problems at hand.”