# News this Week

Science  25 Jan 2002:
Vol. 295, Issue 5555, pp. 598
1. SMALLPOX

# WHO Puts Off Destruction of U.S., Russian Caches

1. Richard Stone

GENEVA— Humankind's worst public health enemy, on death row for more than 2 decades, has won another reprieve. Last week the World Health Organization's (WHO's) governing board agreed to delay destruction of the last known samples of smallpox, now kept on ice at two high-security facilities in Russia and the United States. The decision is a “victory for common sense,” says Lev Sandakhchiev, director-general of the State Research Center of Virology and Biotechnology, which houses Russia's smallpox facility.

The decision reflects a new consensus that the stocks may be needed to defend humanity against the possible use of smallpox as a bioweapon, fears heightened in the wake of last fall's World Trade Center attack and anthrax-tainted letter campaign. “We regard the potential release of smallpox as a critical national and international security issue,” says Kenneth Bernard, special adviser for national security, intelligence, and defense at the U.S. Department of Health and Human Services.

In staying an execution scheduled for this December, WHO's board has handed a dramatic victory to researchers hoping to develop drugs and a better vaccine. “There's been a sea change in thinking—and that's very good news,” says virologist Peter Jahrling of the U.S. Army Medical Research Institute of Infectious Diseases in Fort Detrick, Maryland, whose team is developing a potential monkey model for the disease. The board acted on a recommendation from WHO Director-General Gro Harlem Brundtland, who based her decision on a report last month from a scientific advisory committee.

One sticking point, however, is whether WHO should set a new date to destroy the stocks. The two countries holding all the publicly acknowledged smallpox cards—the United States and Russia—favor an open-ended research program. Setting a deadline “would make it impossible to carry out some research,” insists Yuri Fedorov, chief of the Russian health ministry's emerging disease unit. However, China, Cuba, and several other nations are expected to lobby hard for a deadline out of fear that an open-ended program increases the risk that terrorists could steal the virus or that the virus could escape in a lab accident. Observers speculate that the World Health Assembly (WHA) could set a deadline of 2005 or 2006 to destroy the stocks when it meets in May.

Smallpox is thought to have claimed hundreds of millions of lives in a reign of terror that began with the first human settlements. But Variola major, which kills nearly one in three people it infects, has an Achilles' heel: Humans are its only hosts. That weakness allowed WHO to mount a successful global immunization campaign that led to its eradication in 1980. All nations with declared stocks of live smallpox complied with a WHO request to incinerate these samples, with the Soviet Union and the United States permitted to hold on to live smallpox for research.

These stocks were slated for destruction in 1993, but two developments helped persuade WHA to delay that order. A well-placed defector revealed that the Soviet Union amassed tons of weaponized smallpox virus after the country had lobbied hard for the disease's eradication and had signed a 1972 treaty outlawing bioweapons development. And after the Gulf War, an Iraqi researcher admitted to United Nations inspectors that he had done research on camelpox, a close cousin of smallpox that does not harm humans. Analysts suggested that the work was a surrogate for smallpox research, says Jonathan Tucker of the Monterey Institute of International Studies in Washington, D.C., who described these concerns in Scourge: The Once and Future Threat of Smallpox (Atlantic Monthly Press). The allegation heightened concerns about clandestine smallpox stocks in other countries as well.

In 1999, WHO's variola advisory committee proposed a research program to extract as much information from the virus as possible before putting it to death at the end of 2002. Working at the U.S. repository, the Centers for Disease Control and Prevention (CDC) in Atlanta, researchers have sequenced 10 strains. The sequences are highly conserved, particularly in regions coding for proteins essential for replication. Such proteins would be good targets for potential drugs.

One intriguing development is a potential animal model for smallpox, which could be important for testing drugs and vaccines. “The grand old gentlemen of smallpox eradication have been claiming for years that it was impossible to create smallpoxlike disease in primates, and thus there was little reason to keep the virus around,” says Jahrling. And indeed, he says, “our initial attempt to infect primates at CDC was a dismal failure.” But in a presentation last month to WHO's variola panel, Jahrling described how his team succeeded at infecting cynomolgus macaques after switching to a strain very similar to one that the Soviet Union had weaponized. Nearly all the animals died within a week from a condition that included skin pustules and other hallmarks of smallpox.

The model still has several shortcomings, however. Jahrling's group injected the macaques with large amounts of smallpox, whereas humans would normally contract the disease through the air. The disease was also more deadly than what's observed in humans. Although critics say this suggests that the animal model would be a poor surrogate, Jahrling says that he expects to refine the model by testing lower doses and alternate infection routes. The Russian repository has won funding to ramp up its smallpox effort this year, and it too hopes to vet the monkey model.

Some countries are troubled by an open-ended research effort. “A final date for destruction should be determined, and no excuses should be given for further delay,” says Sha Zukang, China's Permanent Representative to the United Nations in Geneva. But China, which is not on the governing board, is unlikely to find many allies to press that point. An Indian representative, for example, sat quietly throughout the discussion at the WHO board meeting, although his country had until recently advocated swift destruction of the stocks.

The heightened concern about bioterrorism has led some health experts to question the central tenet that stocks of any microbial killer should be destroyed once it is eradicated in the wild. But proponents of eradication say that steps are also being taken to address a bioterror threat. With respect to polio, “efforts have been under way for some time to inventory laboratory stocks and to develop a framework for specimen storage and future research,” says James Hughes, director of the CDC's National Center for Infectious Diseases. The fact that the debate is taking place at all, however, represents another example of the expanding legacy of last fall's tragic events.

2. PROFESSIONAL ETHICS

# Data Hoarding Blocks Progress in Genetics

More than a quarter of U.S. geneticists say they can't replicate published findings because other investigators won't give them relevant data or materials. And the rejections are more than a breach of professional etiquette; they say that data hoarding actually retards progress in the field.

The results of a new survey, led by researchers at Massachusetts General Hospital in Boston, tarnish what has traditionally been a badge of honor among scientists: the sharing of information that allows others to replicate or disprove the original finding. “That's a pretty big deal,” says Robert Cook-Deegan, a science policy analyst at the Kennedy Institute of Ethics at Georgetown University in Washington, D.C. “And it's getting in the way of reliable science.”

The survey team, led by David Blumenthal and Eric Campbell of the hospital's Institute for Health Policy, compared the responses of 1240 geneticists with 600 other life scientists from the 100 universities that receive the most funding from the National Institutes of Health (NIH). The results appear in the 23/30 January issue of the Journal of the American Medical Association.

The survey explores a bread-and-butter issue: 84% of the geneticists report that they have asked another researcher to provide information, data, or materials related to published research. But almost half (47%) said that at least one request had been denied in the previous 3 years. The rejections had a significant impact on their work: 28% say that they had been forced to end a collaboration, and 21% had abandoned a promising line of research. The most likely requests to be thwarted were for biomaterials such as mice or viruses (35% had been denied such a plea), followed by sequence data (28%), findings (25%), phenotypes (22%), and lab techniques (16%).

Despite the widespread rejections, the survey found that naysayers were a distinct minority. Only 12% of geneticists reported that they had denied a request. This number may be an underestimate, Campbell explains, because researchers don't like to admit they resisted sharing their data. The most common reason cited for denying a request was the amount of effort required to produce the data (see table). Indeed, the more requests received, the more likely the scientist was to say no. Those engaged in commercial activities were also more likely to deny requests.

Geneticists say this proprietary behavior is having a negative impact on their field. Some 73% felt that withholding of data slowed progress in genetic research in general, and 58% said it had limited their own work. About the same fraction reported that it hindered the training of students and postdocs. More than twice as many scientists (35% to 14%) thought that withholding had risen rather than fallen over the last decade, although a bare majority (51%) said they hadn't noticed any change.

Campbell and his colleagues suggest that researchers might be more forthcoming if funding agencies provided money to defray the costs of meeting requests. Another step, they say, would be to make material transfer agreements more user friendly. “It's a legitimate cost of doing research,” agrees Wendy Baldwin, NIH's deputy director for extramural research, adding that researchers could either list the cost in their grant application or apply for a supplemental award.

NIH could also put more pressure on researchers to behave civilly, says Cook-Deegan, including a better system to track who's being uncooperative. “There's no shaming strategy available here,” he says.

3. NEUROSCIENCE

# Genes Keep Neurons' House in Order

1. Gretchen Vogel

As any homeowner knows, timely maintenance is vital for keeping a building functioning properly long after construction is finished. The same is evidently true for the complex architecture of the nervous system —at least in the roundworm. On page 686, neuroscientists Oliver Hobert, Oscar Aurelio, and David Hall describe a new family of proteins that help keep the wiring of the worm's nervous system tangle free.

Scientists have spent decades teasing apart the complex signals that guide axons—the long extensions that allow neurons to communicate with distant cells—to their correct destinations and help them make the right connections. But the discovery of a separate, later-acting maintenance mechanism is “really quite surprising,” says neuroscientist Joseph Culotti of the Samuel Lunenfeld Research Institute at Mount Sinai Hospital in Toronto. Developmental neuroscientist Barry Dickson of the Institute of Molecular Pathology in Vienna says the find makes sense. “You don't just have to make sure you wire up the nervous system properly in the first place, but you also have to make sure that the wires don't get tangled up as the animal grows and moves about,” he notes.

Hobert and Aurelio of Columbia University and Hall of the Albert Einstein College of Medicine, both in New York City, did not set out to look for the worm's maintenance molecules. Rather, they were examining the expression patterns of unknown genes in the so-called immunoglobulin superfamily, several members of which are known for their roles in neural development. Six genes stood out in the screen. They appeared on the scene later than others—in the larvae and the adult, after the upheaval of embryonic development is complete. “They're expressed after all the excitement is over,” Hobert says.

The genes, which the team dubbed the ziggenes, are expressed in a neuron called PVT in the larval worm's ventral nerve cord. This neuron plays a central role in the nervous system's development. It has an axon that is among the first to blaze a trail through the developing worm. The axon extends the entire length of the worm's body and secretes proteins that help guide other axons to the correct place in the growing nervous system. But most developmental biologists assumed that the neuron's guidance tasks were complete once the worm reached the larval stage.

The timing of the appearance of these newfound guidance-like molecules prompted the team to question that assumption. Aurelio used a laser to kill PVT neurons in early- larval-stage worms. When he examined the animals' nervous systems 2 days after surgery, he found that in nearly a third of the treated worms, axons had wandered across the worm's midline to the wrong side of the nerve cord.

To check whether the zig genes keep axons in place, the team examined a strain of worms that lacks zig-4. In those worms, the team found, development is normal during the embryonic stage, but once the worm develops into a larva, a subset of axons wanders across the midline—resembling the aberrant axons in the surgically treated worms.

It seems the molecular restraints of the ZIG proteins might be crucial during the early larval stage, when the worms' movements might jostle the still-fragile alignment of axons: When the scientists placed larval worms lacking PVT on a substance that paralyzes them, they observed no wayward axons. Hobert isn't sure what zig genes do in the adult worm, but he suspects that they keep axons in place in other parts of the body.

Dickson predicts that similar maintenance molecules will turn up in other animals— perhaps even in humans. “It could be that this only applies to a few axons in the worm nerve cord that are in particular danger of being jostled about as the worm writhes along,” he says. “But you can bet it is going to be a lot more general than that. If keeping the wires neat and tidy matters for a worm, it's going to matter for higher animals, too.”

4. HUMAN CLONING

# Report Backs Ban; Ethics Panel Debuts

1. Constance Holden,
2. Jocelyn Kaiser

Cloning and stem cells are once again on the nation's front burner after a 4-month hiatus in the aftermath of 11 September. Last week the National Academy of Sciences (NAS) released a report* calling for a legal ban on human reproductive cloning, and the president's new Council on Bioethics held its first meeting.

The academy panel, led by adult stem cell researcher Irving Weissman of Stanford University, confined itself to scientific and medical issues raised by reproductive cloning. It concluded that the high rate of abnormalities and other problems with animals cloned since Dolly the sheep was in 1997 indicate that such an effort in humans “is dangerous and likely to fail.” Weissman also said that testimony presented this summer from individuals with plans to clone humans raised serious questions about safety and monitoring (Science, 17 August 2001, p. 1237). Even so, the panel concluded that any ban should be revisited in 5 years because of probable research advances in related fields.

Funded by the National Academies themselves, the panel seconded an NAS study issued last fall that strongly endorsed so-called therapeutic cloning—making an embryo that can supply genetically tailored stem cells by inserting the DNA of a person's body cell into an enucleated egg. But it suggested that the procedure be labeled “nuclear transplantation to produce stem cells,” rather than cloning, if the blastocyst is not to be implanted in a uterus.

One day before the panel issued its call for a “broad national dialogue” on ethical and societal aspects of a reproductive cloning ban, the president's bioethics council began a 2-day discussion on the larger issues surrounding human cloning, including therapeutic cloning. The newly appointed 18-member group includes three biologists and a clutch of doctors, lawyers, and public thinkers (www.whitehouse.gov/news/releases/2002/01/20020116-9.html).

Its chair, University of Chicago bioethicist Leon Kass, said that the group hopes to go beyond influencing public policy and stimulate a national debate about bioethics. “One feels a palpable increase in America's moral seriousness” since the terrorism attacks, Kass said in welcoming the group to Washington, D.C. He signaled a scholarly approach to the subject by leading a discussion of “The Birthmark,” a story by Nathaniel Hawthorne in which a husband kills his wife while trying to make her perfect.

Calling cloning “the hot topic in bioethics circles today,” Kass said that the panel “would be remiss not to try to clarify the subject and place it on the most solid moral ground. Public concern stems from the intuition that what's at stake here is what it means to be a human being.”

Although council members all seemed to take a dim view of cloning for reproduction, they are clearly divided on the virtues of therapeutic cloning. For example, Michael Gazzaniga, a neuroscientist at Dartmouth College, thought harvesting cells from a blastocyst was no more problematic than harvesting an organ from a brain-dead patient. Others, including Kass himself, believe that a complete ban on all related work is needed even if the goal is only to prohibit the implantation of cloned embryos.

Kass warned reporters that the group plans to move with deliberate speed. The council will meet every couple of months and will convene next month to talk more about cloning. He said to look for a report, with policy guidance, by the summer.

5. FUSION POWER

# Spherical Tokamaks Are on a Roll

1. Andrew Watson*
1. Andrew Watson is a writer in Norwich, U.K.

Results from two fusion experiments, one in the United States and one in the United Kingdom, suggest that making a reactor, or tokamak, spherical with a hole through the middle—like a cored apple—may be more efficient than the traditional doughnut shape. The two machines both managed to confine a hot plasma of hydrogen ions in the dense, calm state used by traditional machines—an important first step toward fusion. Spherical reactors “may in the end be a better bet for a fusion reactor than the conventional … tokamak,” says Geoff Cordey, a plasma physicist at the Joint European Torus (JET) in Culham, near Oxford, the world's largest conventional tokamak. Still, he cautions, it's not yet a new ball game: “It's very, very early days.”

Nuclear fusion—the process that powers the stars—promises almost limitless energy with little nuclear waste. But researchers must first find a way to squeeze atomic nuclei together against their electromagnetic repulsion, close enough that pairs of them fuse into a new species of nucleus the mass of which is less than the combined starting masses. The missing mass emerges as energy. In tokamaks, heat does the squeezing: A whirling gas of hydrogen ions, or plasma, is held inside a big vacuum chamber by magnets and heated to millions of degrees by passing electrical currents through it or firing beams of atoms into it.

Doughnut-shaped tokamaks, such as JET and Princeton's Tokamak Fusion Test Reactor, have managed to achieve fusion, but the amount of energy put in to keep the reactor running far exceeds the amount of energy produced. To reach or even exceed the breakeven point and produce excess energy, researchers say they will have to build the biggest tokamak so far, the $4.2 billion International Thermonuclear Experimental Reactor (ITER). Governments around the world are currently considering whether to go ahead and build the machine. Proponents of spherical fusion, however, think it can be done more simply and cheaply. A brace of papers in this week's issue of Physical Review Letters shows that the Mega Amp Spherical Tokamak (MAST), also in Culham, and the National Spherical Torus Experiment (NSTX) in Princeton have both made first base by achieving high plasma confinement mode, or H-mode. Like a river that regains its composure downstream from white-water rapids, the H-mode is a smooth, dense flow of plasma that is twice as good as the lower density, more turbulent flow at retaining heat. “All the tokamak reactor proposals use H-mode,” says Cordey. Alan Sykes, head of the physics team at MAST, adds: “It wasn't obvious that spherical tokamaks would be able to access this higher mode of confinement.” Interest in spherical tokamaks began in the early 1980s when Martin Peng and Dennis Strickler of Oak Ridge National Laboratory in Tennessee suggested reshaping the torus. Their work attracted little attention until experimenters at Culham cobbled together a baby spherical tokamak. “That was very successful indeed, and the world saw that spherical tokamaks seem to produce good, high-quality plasmas,” says Sykes. The upshot was a whole new generation of spherical machines, of which MAST and NSTX are the largest. The potential advantage of the spherical tokamak is that it takes less magnetic field to give the same level of plasma control. This is because the magnetic field lines in a tokamak spiral like a helical spring down around the central hole before looping back from bottom to top via the outer reaches of the container. Crucially, it's the spiral around the hole that imparts stability to the plasma. Because a spherical tokamak turns the central hole into a long, narrow tube, the magnetic field lines can not only wrap round more tightly, but they do so over a much greater distance. Hence spherical tokamaks make much more efficient use of their magnetic fields and are better able to resist the plasma's urge to break free, says Rajesh Maingi, who led the NSTX search for the H-mode. This improved efficiency translates into as much as nine times the fusion output of the corresponding doughnut tokamak, according to Masa Ono of the Princeton Plasma Physics Laboratory, a co-leader with Peng of NSTX. What's more, it's “simpler and smaller engineering construction,” says Sykes. Although the performance achieved by these first attempts is promising, “they are quite a long way off from a reactor,” says Cordey. The spherical tokamaks must increase their temperatures 10-fold to reach that in JET and ITER—about 150 million degrees Celsius—while still keeping the plasma stable, he says. Sykes worries that such a small spherical machine may require impossible power densities when working as a real reactor. ITER remains the main focus for fusion researchers. Delaying or rejigging this$4.2 billion project “would be a big mistake,” says Cordey. The hope is to develop ITER and spherical tokamaks in tandem. “It may be that after ITER, when utilities want to build a fusion power plant, they find that the spherical tokamak is a more economical way of doing it,” says Sykes.

6. HIGH-ENERGY PHYSICS

# Atom Smasher Probes Realm of Nuclear 'Gas'

1. Charles Seife

“Oh, that this too too liquid nucleus would evaporate.” If Hamlet were a nuclear physicist, he might be feeling a bit more cheerful. Strange as it may seem, atomic nuclei do sometimes act like liquids, and when blasted apart at high enough energies they can sizzle into gas. Now scientists working at Brookhaven National Laboratory in Upton, New York, have charted the conditions under which gold nuclei make that leap, information that might help unravel the secrets behind the birth of a neutron star.

The work builds on a model that physicists cooked up in the 1930s to explain the fission of uranium. A neutron striking a nucleus more than 200 times its mass doesn't just knock off a chip or two; it splits the nucleus neatly in two. Physicists realized that the uranium nucleus is behaving like an oversized drop of water. When it is struck by a neutron, the nucleus oscillates, stretches out, and then blurps into two roughly equal parts (throwing off a few smaller fragments, such as neutrons, in the process). “Everyday garden-variety nuclei behave like a liquid,” says Victor Viola, a physicist at Indiana University, Bloomington. “It's a very successful description.”

Viola and colleagues decided to take the liquid analogy one step further by determining the nucleus's equation of state–the relations between pressure and temperature that govern when the nucleus behaves like a gas and when it behaves like a liquid. At Brookhaven, they shot protons, pions, and antiprotons at thin gold foil, adding energy that brought the gold nuclei to a boil. Meanwhile, a device called the Indiana Silicon Sphere (ISiS)–a beach ball-sized sphere studded with 450 detectors –kept careful track of the size and energy of the particles that flew off.

The physicists analyzed the readings in two different ways. The first starts with the distribution of the sizes of chunks that fly out of the nucleus. “In boiling water, you don't get individual water molecules coming off,” says Viola. “You get dimers, trimers, tetramers. The temperature of the vapor is related to the relative numbers of those clusters.” By comparing the energy added to the nucleus (hence its “temperature”) with the relative abundances of fragments, the physicists figured out the properties of the nuclear “liquid,” including its critical temperature: the point above which the liquid phase can no longer exist, which they calculate at about 7 million electron volts (MeV). The second analysis directly models the breaking and making of nuclear bonds and comes up with a slightly higher critical temperature, slightly above 8 MeV.

“I do think it's a really nice piece of work they've done,” says Joseph Natowitz, a physicist at Texas A&M University in College Station, who thinks that physicists will resolve the discrepancy once they get a better grip on how the nucleus expands and breaks up after the collision. “I have some ideas.”

Even though wrinkles need to be ironed out, the results have given physicists a new tool for understanding the “evaporation” of nuclei. They might also shed light on the reverse process, the condensation of nuclei from smaller parts. “It's relevant to what happens in the formation of neutron stars,” says Viola. If so, the work is likely to be a hit–a palpable hit.

7. SUPERCONDUCTIVITY

# Perplexing Compounds Rejoin the Club

1. David Voss

If you want to start a fight in a roomful of physicists, ask them how high-temperature superconductors (HTSCs) work. The compounds, which are based on layers of copper oxides, lose their electrical resistance at temperatures as high as 138 kelvin—almost 100 degrees warmer than the best conventional superconductor. By rights, they should be prime candidates for a unified theory. Yet 15 years after the discovery of HTSCs, every theorist seems to have a different explanation for their strange properties.

Now results of a laborious experiment, published online this week (http://www.sciencexpress.org/), have solved a longstanding puzzle about the superconductors: why some, but not others, appeared to show a fundamental fingerprint of magnetic spin. Physicists have debated the reason vigorously since 1991, when researchers shooting beams of neutrons onto HTSC superconductors at the Institut Laue-Langevin in Grenoble, France, discovered an unusual pattern in their scattering data. Neutrons are like tiny bar magnets, carrying no electrical charge but a small amount of magnetic spin, a property they share with electrons. So the way these miniature magnets bounce off a superconductor can reveal what the material's electron spins are up to.

The French team discovered a faint peak that suggested the spins were conspiring in some collective resonant interaction, like a sea of compass needles all wiggling in unison. And because the resonant peak grew large and sharp when the material was superconducting, many thought that magnetic interactions might help solve the mystery of HTSC superconductivity.

The catch was that the resonance was seen only in HTSC materials the crystal structures of which had two or more layers of copper oxide; single-layer compounds such as lanthanum-strontium-copper-oxide, the first HTSC ever discovered, seemed exempt. Some physicists believed that meant spin resonances were a red herring that they could ignore.

Now the herring is back, and it's real. A collaboration between the Max Planck Institute for Solid State Research, two Atomic Energy Commission (CEA) labs in France, and the Institute of Solid State Physics in Chernogolovka, Russia, reports that the resonance occurs in a single-layer HTSC compound after all. Because the material they studied, a thallium-barium-copper oxide compound, hasn't been grown in crystals big enough for neutron scattering, the scientists had to devise a painstaking technical workaround. “We aligned several hundred small crystals so that they behave like one large crystal,” says co-author Bernhard Keimer. With this composite specimen, they were able to carry out the neutron study. When they analyzed the data, the resonant peak was there. “This proves that the resonant mode is a generic property of these superconducting materials,” Keimer says.

“This is a tour de force,” says John Tranquada, an experimentalist at Brookhaven National Laboratory in New York. “Preparing and aligning 300 crystals was a tremendous task, and the measurements required considerable patience.” Tranquada believes the data will stand up to scrutiny. Michael Norman, a theorist at Argonne National Laboratory in Illinois, agrees: “Now it's clear that this resonance is the rule rather than the exception.”

Less clear is how theory will accommodate the new observations. “This is where the real debates start,” Norman says, “and it's a mine field.” Physicists are stepping lightly, because each theory has a different idea of what makes the HTSCs tick and no theorist is going to yield ground easily. All superconductors work because the electrons (or holes) become glued together in pairs; in conventional materials the pairing is due to one electron's distorting the crystal lattice and attracting another—like two bowling balls on a mattress. The pairs then waltz through the material without resistance. Most theorists believe some other kind of “glue” will be needed for the HTSC materials. Boosters of theories that invoke magnetic or spin effects to glue the charge carriers together will likely gain the most encouragement from the new data.

Keimer stresses that linking HTSCs through spin resonance is a first step, not a knockout punch. “Our experiment will not end the debate about a final theory of superconductivity,” he says, “but it may help tilt it in a specific direction.”

8. GENE THERAPY

# Blood Test Flags Agent in Death of Penn Subject

Exactly what killed Jesse Gelsinger, the first volunteer to die in a human gene therapy trial, remains a mystery, but last week researchers in Germany fingered a feature of his immune system as a prime suspect. They also believe that a simple blood test might be able to prevent similar tragedies in future gene therapy trials.

In September 1999, 18-year-old Jesse Gelsinger took part in a trial designed to test the safety of using a form of adenovirus to transport new genes into patients. Adenovirus normally only causes mild colds. Nonetheless, within hours of the injection of the virus “vector,” Gelsinger's immune system went into overdrive. Four days later he died of multiple organ failure. James Wilson, leader of the trial and head of the Institute for Human Gene Therapy at the University of Pennsylvania in Philadelphia, initially suggested that another viral infection or undetected genetic condition might have triggered the harsh immune response to the adenovirus that investigators concluded had killed Gelsinger (Science, 17 December 1999, p. 2244, and 12 May 2000, p. 951). After further studies in monkeys, he pointed to the proteins in the coat of the vector as a possible source of the immune response revolt. Wilson was unavailable for comment on the new findings.

Günter Cichon of the Max Delbrück Center for Molecular Medicine in Berlin and his colleagues sought to find out how adenovirus provokes the body's defenses. They mixed blood samples from 18 individuals with adenovirus that was “externally identical” to the one used in Wilson's trial. The virus set off a forceful response from the complement system, a natural and powerful defense against invading pathogens, but only in samples that already contained antibodies against adenovirus. Reporting in the current issue of Gene Therapy, the team concludes that a viral dose comparable to the one Gelsinger received raised the concentration of a key component of the complement system to a level that could start a damaging immune reaction.

Cichon notes that Gelsinger was known to have “suffered a chest infection some time before the trial,” so his complement system might have been sensitized already. In the bloodstream, the proteins of the virus coat would combine with antibodies, forming complexes that activate the complement system. This can cause inflammation in the vessel walls of liver, lungs, and kidney, and ultimately multiple organ failure. “Exactly the same symptoms were observed in the case of Gelsinger,” says Cichon.

Gene therapist Prem Seth of Des Moines University in Iowa thinks that complement activation could indeed cause some of the adverse reactions observed in gene therapy trials with adenovirus vectors. Several years ago, he observed that the coat proteins of the virus initiate a strong immune response in human blood. “I have always argued that the virus should only be applied locally, not into the bloodstream,” he told Science. And he agrees with Cichon that complement activation should be measured in blood samples to see if the test can predict which patients are likely to suffer strong adverse reactions. “All patients should be screened for their complement response,” he says. “It is an easy test.”

Phil Noguchi, director of the Food and Drug Administration (FDA) division for gene therapy, agrees that this finding is “a new piece in the puzzle” but emphasizes that the fatal trial probably had “multiple sources.” He says that the FDA is considering how to use a complement-sensitivity test in gene therapy trials.

9. EXTRASOLAR PLANETOLOGY

# Jupiters Like Our Own Await Planet Hunters

1. Richard A. Kerr

Astronomers have had plenty of luck lately finding planets circling other stars. But they've had no guarantees that the greatest prizes—planetary systems like our own, with a potential for life—are out there to be found. Due to limitations of the searches so far, the 77 newly discovered extrasolar planets either are gas giants orbiting much closer to their stars than Jupiter or are far more massive than Jupiter. No one can yet detect the most prominent hallmark of our solar system: planets resembling Jupiter in mass (at 70% of the solar system's total planetary mass) and in orbital distance (five times Earth's). But two new studies—one extrapolating from the oddball planets discovered so far and another modeling the way planetary systems form—give added hope that systems like ours are out there in abundance. Astronomers could start finding Jupiter-like exoplanets within a few years (also see p. 616).

Astronomers have found 77 exoplanets so far by watching how each parent star wobbles, pulled by an unseen planet orbiting it. The wobble shows up as a rhythmic variation in the star's color due to Doppler shifting; the more massive the planet or the closer it orbits the star, the bigger the wobble. In a paper posted on the preprint server astro-ph (0201003v1) on 2 January, physicists Charles Lineweaver and Daniel Grether of the University of New South Wales in Sydney, Australia, identified a subset of 44 exoplanets circling stars that had been monitored long enough to discover all the bodies orbiting at least as close as Mars orbits the sun and as large as or larger than Jupiter.

Extrapolating from trends in mass and orbital distance within this more representative subset, the Australian physicists find that “Jupiters are probably very typical” of the as-yet-unobserved exoplanets, says Lineweaver. They predict that 22 new Jupiter-like exoplanets—as big as Jupiter or larger, orbiting from just beyond the distance of Mars to a bit beyond the distance of Jupiter—will be found orbiting around the 1000 or so stars that have been monitored for more than 3 years. The prospect of familiar-looking planetary systems is more encouraging than earlier extrapolations had suggested, says Lineweaver, because by focusing only on well-studied stars they reduced observational bias and they also included the latest discoveries. (Extrapolation to the abundance of exoplanets as small and distant as Saturn is not yet advisable, the pair says.)

A second group has come up with a similar result by taking a more theoretical approach. In a paper submitted to Astronomy and Astrophysics, astronomer David Trilling of the University of Pennsylvania in Philadelphia, cosmochemist Jonathan Lunine of the University of Arizona in Tucson, and astrophysicist Willy Benz of the University of Bern in Switzerland consider how planetary systems created in a computer model compare with the exoplanets observed so far. Their model of planetary system evolution has a swirling disk of gas and dust of the sort from which planets agglomerate. But the disk can also destroy newly formed planets by interacting gravitationally with them and driving them into their star.

Planetary system formation is thus an old-fashioned cliffhanger, with planets heading for the furnace unless someone turns off the conveyor belt in time. Survival in real planetary systems depends on the mass of the planet, the mass of the disk, the lifetime of the disk, and the inherent ability of disk material to drag on the planet. Trilling and his colleagues varied these properties one at a time over plausible ranges in a series of simulations. In each they inserted a single planet into the disk at the distance of Jupiter and noted its fate by the time the disk had dissipated.

To judge by their simulations, “Planet formation is an 'easy come, easy go' business,” they write, “with many planets created and many planets destroyed. …” Two-thirds of all model planets migrate too fast and are consumed by their stars before disk dissipation. Ten percent to 30% of the surviving planets come to a stop close to their star. That fits the discoveries so far if about 30% of sunlike stars form planets. But 70% to 90% of surviving giant planets, according to the modeling, should remain too far out to have been found by past searches. Jupiters should be among them. And in such systems, unlike those observed to have giant planets near their stars, there could be room for small, rocky planets like Earth at a comfortable, habitable distance from the star.

Astrophysicist Scott Tremaine of Princeton University sees these results and Lineweaver and Grether's extrapolation as reasonable quantifications of trends hinted at by the discoveries so far, and he looks forward to coming discoveries. As some monitoring records approach the requisite 12 years, Doppler detection of extrasolar Jupiters may not be far off. And searches are in the works for terrestrial-sized planets by looking for planets passing in front of their stars. But Tremaine remains cautious about what these searches will turn up. Speaking as a theorist, he notes that “most every prediction by theorists about planetary formation has been wrong.”

10. CELL BIOLOGY

# Cell Migration Research Is on the Move

1. Marina Chicurel*
1. Marina Chicurel is a writer in Santa Cruz, California.

One of cell biology's first mysteries is coming under renewed scrutiny as new techniques allow researchers to follow in cells' steps

Anton van Leeuwenhoek was fascinated by the movements of the tiny creatures he discovered in rainwater in 1675. In a letter published by the Philosophical Transactions of the Royal Society of London, he described living atoms that put forth little horns, extended and contracted, and had pleasing and nimble motions.

Cell-sized movements still draw the attention of biologists today. “The cool thing about cell motility is it's one of the first research programs in [cell] biology,” says Thomas Stossel of Brigham and Women's Hospital and Harvard Medical School in Boston. No less appealing is its central involvement in a wealth of biological processes. Migrating cells shape organs and tissues in developing embryos, hunt down infectious agents, generate new blood vessels, and close wounds. And cell migration gone awry lies at the root of many diseases, including metastatic cancer, arthritis, and neurological birth defects.

Yet cell migration has not enjoyed the popularity of some of its sister fields. “The irony is that cell migration predates by hundreds of years the molecular biology revolution, yet it has been somewhat of a sideshow,” says Stossel. Part of the problem is that cell migration has been a fractured discipline, composed of many subfields. “There are so many interacting pathways and processes that it's very difficult for any one lab to work on all of them,” says Alan Hall of University College London. “It's kind of mind-boggling to try and put it all together.”

But powerful new tools are beginning to help assemble a more unified picture of cell migration. Stripped-down model systems and methods to search for migration-related genes and proteins on a large scale are enabling researchers to identify the multiplicity of players and track their interactions. Techniques to monitor and manipulate mechanical forces in living cells are providing clearer views of the dynamics of cell movement. And computer models are helping researchers design experiments, test hypotheses, and integrate the wealth of data.

“Finally, cell migration is starting to emerge as an integrated, whole field,” says Rick Horwitz of the University of Virginia in Charlottesville. Indeed, in September 2001, the National Institute of General Medical Sciences (NIGMS) in Bethesda, Maryland, awarded an $8 million “glue” grant to the Cell Migration Consortium, a multidisciplinary, multisite group led by Horwitz and Thomas Parsons, also at the University of Virginia. Over the next 5 years, NIGMS expects to award$38 million to this diverse group of biologists, chemists, physicists, and computer scientists, who plan to develop new technologies and approaches for tackling cell motility.

## The crawl

Modern theories of cell migration can be traced to the early 1970s, when Michael Abercrombie of University College London published a series of five articles in Experimental Cell Research. Based on studies using video and electron microscopy of cells crawling on solid surfaces, Abercrombie suggested that cells move by extending protrusions and forming new contacts with the surface. The cell body is then drawn up to the new adhesions by a “system of contractile filaments.”

This largely mirrors the current understanding of cell migration. Most researchers agree that eukaryotic cells extend protrusions to push their membranes outward by harnessing the growth of filaments of a protein called actin. Cells then attach their protrusions to the extracellular matrix, a mesh of proteins that surrounds cells and holds tissues together. To grab hold, cells use integrins, proteins that span the cell membrane and recruit signaling and structural proteins, including actin fibers, to form adhesions. Adhesions serve as points of traction over which the cell can pull the bulk of its body forward. Proteins that tug on actin filaments to generate force do the towing; they're called myosins and are similar to proteins that allow muscles to contract. Finally, the cell releases its rear attachments, marking the end of a full cycle of motion.

## A problem of numbers

Although this general description appears simple, it belies a complicated and often mysterious set of processes. Each step involves many players, and “there are a lot of subprocesses, each of which involves scores of molecules,” says Horwitz. One of the main goals of cell motility research is to identify these supporting actors and determine their function and relative importance.

Some researchers are tackling this multiplicity by developing models that strip away at least part of the complexity. Marie-France Carlier of the National Center for Scientific Research in Gif-sur-Yvette, France, for example, has assembled an actin-based moving machine in a dish that recreates many of the characteristics of how crawling cells extend protrusions called lamellipodia. Carlier monitors the ability of the components to produce movement by watching bacteria called Listeria and Shigella. These microbes infect cells and enlist several of the components that cells normally use to extend lamellipodia to power the bacteria's own travels within the host's cytoplasm. In experiments reported in Nature in 1999, Carlier succeeded in making the bacteria move as if they were inside a cell by adding only a handful of purified components: actin, four proteins that regulate actin assembly, and ATP, a molecule cells use as a source of energy.

More recently, she has stripped down the system even further by replacing the bacteria with latex beads coated with a small set of known proteins, producing an artificial system whose components are completely defined and that in many ways mirrors lamellipodial protrusion. Ultimately, Carlier hopes to make her system even more lamellipodium- like, adding regulatory proteins to enable it to respond to external signals and wrapping a cell membrane around the components. “We are trying to make an artificial lamellipodium from scratch,” she says.

Other researchers are making parts lists to track the various players in cell motility. “The idea is to produce an accounting of the proteins and their modifications at defined places in the cell,” says Mark Ginsberg of the Scripps Research Institute in La Jolla, California. His team has constructed baits with which to fish out proteins that interact with the portions of integrins that reside within cells. Because integrins are composed of two protein subunits, the researchers make their baits by using small protein modules to splice together fragments of the two subunits and mimic integrins' configuration inside cells. They then attach the engineered proteins to beads and expose them to cellular extracts. Proteins that normally interact with integrins stick to the beads and can be analyzed. “We've been doing this with success,” says Ginsberg, who first described the technique in 1998. But until recently, researchers have been limited to tracking the binding of a few proteins at a time.

Now new techniques are poised to help Ginsberg and others searching for interacting protein partners scale up their searches dramatically. “Mass spectrometry is probably the ideal way of getting at this,” says Donald Hunt of the University of Virginia. Hunt has designed mass spectrometers—instruments that can determine the molecular weights and amino acid sequences of protein fragments—that enable researchers to rapidly identify thousands of proteins in a mix with great sensitivity.

And Hunt has recently taken this capability a step further by designing a system, which relies on an instrument known as a Fourier transform mass spectrometer, that can compare the protein content of two different mixtures. It can pick out proteins that are differentially expressed in two types of cells, or in the same types of cells under different conditions—say, proteins present in mobile cells but not sedentary cells—and selectively identify them.

Researchers are also casting genetic nets to do large-scale searches for genes that participate in cell migration. A team led by Joan Brugge at Harvard, for example, is screening genes expressed by highly mobile cells derived from invasive breast tumors for their ability to boost the mobility of normal breast cells. So far, the researchers have fingered two genes coding for proteins that, although previously identified, were not known to play a role in migration. And using a new vector for delivering genes into cells, developed by George Daley of the Massachusetts Institute of Technology, Brugge thinks her search is about to speed up. The vector enables researchers to insert genes into cells efficiently and then retrieve them easily—making it possible to sequence the genes quickly once they've proven their abilities, in this case, by facilitating migration. “This will just save months and months of work,” Brugge says.

## The right place at the right time

Besides having a large cast of characters, cell migration involves precisely timed performances at precise locations. Biologists have scores of tools for grinding up cells and studying the identities and abundances of cellular components. But such techniques fall short when studying migration. It relies on processes that are restricted to certain areas of the cell, such as the lamellipodia or the rear end, and that change over time as a cell proceeds through its cycle of extension, attachment, towing, and detachment.

To monitor the coming and going of molecules as they perform their duties in living cells or simplified in vitro systems, cell biologists have long relied on fluorescent labels. By following labeled actin monomers under the microscope, for example, researchers have witnessed the polymerization and depolymerization of actin filaments in real time. But in the last 2 years, new advances have greatly expanded the capabilities of fluorescence microscopy, enabling researchers to get a better view of when and where proteins involved in cell migration do their jobs.

Advances in technologies to track single molecules, for example, allowed Kurt Amann, working in Thomas Pollard's lab at Yale University, to watch how a protein complex found in lamellipodia, called the Arp2/3 complex, regulates actin branching. In work published in the December 2001 issue of the Proceedings of the National Academy of Sciences, Amann used a technique called total internal reflection fluorescence microscopy (TIRFM) to follow the growth of isolated actin filaments in the presence of the Arp2/3 complex. Like many before him, Amann used fluorescent actin monomers to track growth. But by using TIRFM, he was able to monitor single actin filaments—a trick normally beyond the power of traditional microscopy. To avoid noise from other labeled monomers, TIRFM shines a laser beam on the specimen at an angle that is completely reflected by the interface between the glass slide that holds the specimen and the solution in which the specimen is immersed. In this way, only fluorescent molecules that are within 100 nanometers of the interface are excited.

Amann's TIRFM observations shed light on how the Arp2/3 complex assembles actin branches—an important issue because branched networks are thought to provide a sturdy scaffold to push the cell membrane forward during extension. “He found that activated Arp2/3 complex forms branches at random sites along the length of actin filaments,” says Pollard. But, cautions Carlier, “the [TIRFM] technique is not devoid of artifacts.” Branches that bump into the nearby glass, for example, probably don't grow normally. Based on other studies, Carlier and others think Arp2/3 complexes form branches from the ends of actin filaments, not their sides. Despite numerous experiments addressing this controversy, including two structural studies of the Arp2/3 complex (Science, 28 September 2001, p. 2456, and 23 November 2001, p. 1679), the question remains unresolved.

To track filaments in living cells, researchers such as Clare Waterman-Storer at Scripps have opted for loading cells with very low amounts of fluorescent monomers to reduce background noise. This strategy has the added advantage of creating irregularly labeled, or “speckled,” filaments that have a distinctive pattern and can thus be followed under a microscope as they move within the cell.

Now Waterman-Storer is trying to inject new power into the technique by collaborating with computer scientists and engineers to automate speckle tracking. Even trained experts find it hard to accurately measure filament movement, assembly, and disassembly by following speckles with their eyes. With computers, however, it should be easier to monitor the direction and magnitude of movement and to extract a greater fraction of the data generated by each experiment. Says Waterman-Storer, “My vision is this will be the quantitative tool of choice to allow you to more easily measure” actin and other proteins' dynamics in vivo.

## Tracking turn-ons

Following the behaviors of many proteins involved in motility also requires tracking their activation status. Knowing where in the cell and at what time signaling proteins become activated is key to understanding how cells coordinate movement.

A Scripps team led by Klaus Hahn recently did this using a technique called fluorescence resonance energy transfer, or FRET (Science, 13 October 2000, p. 333). The researchers created a biosensor to track active Rac, a signaling protein that initiates the formation of lamellipodia, by attaching a fluorescent label to a fragment of a protein that binds Rac when it is activated. In addition, they linked Rac to another fluorescent tag, called green fluorescent protein. GFP fluoresces at a wavelength that excites the biosensor label. So when the two proteins are in close proximity, as occurs when Rac is active, researchers can see the biosensor shine. By introducing both proteins into cells and tracking GFP fluorescence, the researchers showed that stimuli that induce migration turn on Rac at the cell's leading edge.

Although FRET is not a new technology, its use has been limited because it's difficult to get it to work. “Engineering things so that you actually get useful information and don't perturb the biological system is what's hard,” says Martin Schwartz, who collaborated with Hahn and also works at Scripps. Schwartz hopes to create biosensors to track the activities of several proteins involved in migration.

## Knock-in, knock-out

Light-sensitive molecules are useful not only for following proteins but also for manipulating them. “We can either selectively activate or knock out certain molecules in a very precise spatial and temporal way,” says Ken Jacobson of the University of North Carolina, Chapel Hill. Jacobson's team made migrating cells turn by shining a spot of ultraviolet light on one side of their lamellipodia, the researchers report in the May 2001 issue of the Journal of Cell Biology. The trick was that the cells were loaded with a protein that sops up actin monomers and that the researchers had chemically modified to make light-sensitive. The protein remained inactive, or “caged,” until stimulated by ultraviolet light. When the spot of light “uncaged” the protein in a small region of the cell, it created an imbalance of actin monomers, causing the cell to turn about the irradiated spot. Jacobson is now working on caging other molecules that alter actin dynamics. The approach, he hopes, will clarify the roles of potential migration proteins that have been studied extensively in vitro but whose in vivo roles remain uncertain.

A complementary technique called chromophore-assisted laser inactivation (CALI) may help dissect the functions of other proteins involved in migration by allowing researchers to knock out specific proteins. Although CALI is not a new technique, Jacobson thinks it now has a particularly bright future based on his as-yet-unpublished studies showing that GFP, which is readily available and easily linked to cellular proteins, can be used as CALI's hit man. “It could be a tremendous boost in the usefulness of GFP,” says Jacobson.

## Pushing and pulling

Few cellular behaviors are as visibly shaped by mechanical forces as cell migration. Yet much remains unknown about how movement results from the integrated effects of cell adhesion, contraction, and the physical characteristics of the substrates cells migrate on. Taking advantage of several recent advances in technologies to apply and measure mechanical forces, however, researchers are beginning to put the pieces together.

By adapting photolithographic methods similar to those used in the semiconductor industry, Donald Ingber's group at Harvard Medical School and Children's Hospital in Boston has examined the role of mechanical stress in determining a cell's choice of direction. The team created surfaces with circular and square islands similar in size to a single cell. Their unpublished results show that when the islands are coated with extracellular matrix proteins, cells spread out to assume the shape of the island, regardless of whether the island is a square or a circle. Round cells extend lamellipodia in random directions, the team found, but square cells send out extensions primarily from their corners. “Imagine you have a little water balloon,” Ingber explains. “In a circle, it's spreading out equally in all directions. But if you keep trying to pull it out on a square, the corners get stretched more than the sides. So by changing the shape, you've constrained where the cell puts its adhesions. And that constrains where it applies stress, which apparently dictates where it migrates.”

Benjamin Geiger's team at the Weizmann Institute of Science in Rehovot, Israel, has also used lithography to generate patterned substrates. But they used the substrates to measure forces rather than control them. Previous experiments had shown a give-and-take between adhesions and traction forces. Mature adhesions generate traction by gripping the extracellular matrix and triggering contraction. But at the same time, mechanical forces, including contraction, shape the strength and assembly of adhesions.

In a study published in the May 2001 issue of Nature Cell Biology, Geiger's group placed cells sporting fluorescently labeled adhesion proteins on flexible substrates featuring arrays of tiny pits or bumps whose exact positions could be tracked under the microscope. They were then able to monitor in real time how the cells gripping the substrate deform the arrays and consequently obtain precise measurements of the local forces the cells exerted at the adhesion sites. The researchers found that the larger the force, the larger the associated adhesion complex. And when they applied a drug that blocks contraction, the adhesions shrunk within seconds, suggesting that contraction affects adhesion assembly almost instantaneously, perhaps by directly deforming or rearranging adhesion molecules and their associated actin filaments.

## Virtual movement

Until recently, most cell biologists had shied away from theoretical approaches, often arguing that the complicated nature of cell behavior was not amenable to mathematical descriptions. But a growing number of studies are showing how computer modeling can help answer questions ranging from how actin pushes the cell membrane forward to how cells use chemical gradients to steer their crawling. More biologists are welcoming the technique as a valuable asset. “Our intuition about how things work is frequently faulty and can often only be appreciated after making a mathematical model,” says Pollard, a biologist who has long collaborated with modelers.

Despite modeling's growing popularity, several hurdles block its widespread use. Most agree that to be truly informative, theory and experiments must go hand in hand. Yet biologists often lack the mathematical training to create their own models, and setting up multidisciplinary collaborations is not always easy. One approach that may give biologists a hand is being developed by Alex Mogilner of the University of California, Davis, and Leslie Loew of the University of Connecticut Health Center in Farmington.

“What we hope to do is to make a computational tool that will allow ordinary biologists to simulate moving cells on the computer,” says Mogilner. “When we model, we do a lot of esoteric stuff that people who are not trained in mathematics cannot do. But if we're successful, then basically any scientist with only an elementary training in mathematics will be able to do it.”

Based on software developed by Loew, called Virtual Cell, the researchers hope to create a program in which users can test hypotheses about how different parameters—including cell shape, the number and distribution of cellular components, ion and protein concentrations, and reaction rates— influence cell motility. For example, a researcher who suspects that the concentration of actin monomers at the leading edge helps determine cell speed could use the model to obtain an estimate of the range of concentrations that are likely to affect speed in a real cell and then design experiments accordingly. “Of course, a lot of [the parameters are] unknown,” explains Mogilner. “But that's the role of the model: You play with the concentrations and rates and see what the model gives you as an outcome.”

The study of cell migration seems poised for integration, with researchers using a variety of approaches beginning to find common ground. “It's kind of like the railroad in the 1800s: people working from two coasts and meeting around Salt Lake City,” says Jacobson. “We hope the top-down approaches meet the molecular approaches and give us some understanding of how these processes are integrated.”

11. SOUTH KOREA

# Institute Helps Spread Use of Vaccines in Asia

1. Mark Russell*
1. Mark Russell writes from Seoul.

The International Vaccine Institute hopes that research collaborations will make vaccines more available in the developing world

SEOUL, SOUTH KOREA— Vietnam is providing Indonesia with technology to produce a low-cost oral cholera vaccine. And China is sharing with India and Vietnam its technology on a typhoid fever vaccine. These and other collaborations are fruits of a new program, called Diseases of the Most Impoverished (DOMI), dedicated to improving the health of nearly half the world's population. The program is the largest undertaken by the fledgling International Vaccine Institute (IVI), which takes a novel approach to stimulating research, training, and technical assistance on vaccines in developing countries.

IVI's goal is to escape the Catch-22 that hinders health research throughout the developing world: Cash-strapped governments don't have enough information about infectious disease killers to know how best to fight them, but they are reluctant to spend scarce resources to get that information because the payoff isn't clear. “We're trying to evaluate the magnitude of the problem and, just as importantly, show that there is a market for vaccine producers,” says IVI director John Clemens, a U.S.-trained epidemiologist who has spent 20 years working on vaccines for the developing world. Collaboration also offers member countries a chance to advance more quickly than if they worked on their own. “Typhoid fever and cholera vaccines have been available for 15 years,” says Clemens, “but they were not used in most developing countries. We're telescoping that introduction process.”

The institute is the creation of the United Nations Development Programme; it set up shop here in 1997 (Science, 16 April 1999, p. 410). South Korea has made room for it on the campus of Seoul National University and is building a five-story laboratory and headquarters that the small but growing scientific staff hopes will be ready by the end of the year.

IVI's first project, begun in 1999, tackled bacterial meningitis in Vietnam, Korea, and China. But DOMI, a 5-year, \$40 million effort funded by the Bill and Melinda Gates Foundation, is by far its largest initiative. Seven countries—China, India, Indonesia, Bangladesh, Pakistan, Vietnam, and Thailand—have declared war on cholera, typhoid fever, and shigellosis, which together kill nearly 2 million people a year and afflict 200 million in the seven countries. Last year DOMI supported more than 40 projects, including disease-burden studies, vaccine demonstration projects, training, and socioeconomic surveys.

In addition to the formidable scientific and technical challenges they face, scientists must also grapple with obstacles less visible but just as crippling. “There is a large-scale denial of the existence of cholera as a major public health problem in Pakistan,” says Zulfiqar Bhutta of the Aga Khan University in Karachi. The same is true for shigellosis, he says. A lack of data initially blunted his attempt to start a typhoid fever vaccination program, he recalls. But once he had enough information for a presentation to policy-makers, “the reaction was a mixture of surprise and shock.”

Clemens says that Bhutta's tale is not unusual in DOMI countries, where the concerns of the poor take a back seat to those of the wealthy and politically influential. “If these diseases affected the middle class,” he says, “we wouldn't have a problem.” DOMI helps scientists level the playing field by offering policy-makers what Clemens calls “a constellation of different kinds of evidence.”

The first steps are disease-burden studies. Simply determining how many people are suffering from these diseases is a challenge for countries with large gaps in their health care systems. Poor populations are highly mobile (especially refugee populations), are not well enumerated, and lack the resources to help themselves.

The next step is to evaluate the effectiveness of existing vaccines and develop new strains tailored to the local population. “The results from several countries using standardized protocols will have a much greater impact on regional policy than would individual studies in individual countries,” Clemens says. The third step is helping each country gain reliable access to the vaccines. Throughout the process, social scientists are needed to educate the public about the cost-effectiveness of the vaccines and to deal with people's concerns.

Although DOMI was launched just 2 years ago, it has already made an impact on health care in the region. Bio Farma, Indonesia's only vaccine producer, has built facilities for local production of Vi typhoid fever vaccine, a subunit of a live, attenuated oral vaccine, and a killed oral cholera vaccine developed in Vietnam that's less expensive than the Western version. Shantha, a major vaccine producer in India, is receiving help from China's Lanzhou Institute to produce the Vi vaccine for both the public and private sectors. Vietnam's National Institute of Hygiene and Epidemiology in Hanoi has taken out a low-interest loan from South Korea to build modern facilities, with design help from DOMI, that can produce the oral cholera vaccine under internationally recognized good manufacturing conditions.

“There were collaborations before IVI,” says Vietnam's Thu Van Nguyen. “But each institute has its strengths and weaknesses. IVI spreads the strengths around more widely then ever before,” he adds, through the meetings, workshops, and collaborations it sponsors. Wang Bing-rui of the Lanzhou Institute is quick to agree: “DOMI has helped us increase vaccine yield, with improved quality.”

IVI's success has also caught the eye of scientists from the industrial world, who are eager to work more closely with large afflicted populations. “I came here because of their experience with clinical trials in the Third World,” says Hans G. Kreeftenberg of the National Institute for Public Health and the Environment in the Netherlands, who visited Seoul recently for a workshop on technology transfer. “These diseases have a major public health impact, but they don't get priority from drug companies because of the low return on investment.”

IVI can also play a positive political role, says Bhutta: “It recognizes that the diseases are the real enemy.” And it's a foe that can't be ignored. “Officials often say that these problems will go away when a country gets good clean water,” says Clemens. “But that might take 50 years. We need to do something now.”

12. CANCER RESEARCH

# Taking Garbage In, Tossing Cancer Out?

1. Ken Garber*
1. Ken Garber is a science writer in Ann Arbor, Michigan.

Working in a novel way—blocking the cell's garbage disposal or proteasome—a new class of compounds is showing promise in clinical trials

The proteasome is the cell's garbage shredder, a barrel-shaped enzyme that sucks in damaged or short-lived proteins and dismembers them for eventual disposal or recycling. It's absolutely essential for survival. So when ProScript, a tiny, privately held company leasing a basement office in Cambridge, Massachusetts, discovered a drug in 1995 that could treat cancer by blocking the proteasome, the idea met with almost universal skepticism: The treatment seemed likely to kill patients along with their tumors. The company persevered, however, getting encouraging results in animals and eventually persuading the National Cancer Institute (NCI) to fund clinical trials of its drug, PS-341. “Our budgets were strapped,” recalls Julian Adams, ProScript's former chief scientist and now senior vice president for drug discovery at Millennium Pharmaceuticals, also in Cambridge. “We were down to fumes.”

ProScript's tenacity may be paying off. First, Millennium acquired the company in 1999, providing badly needed capital. And now, clinical results in multiple myeloma, presented at December's meeting of the American Society of Hematology (ASH), demonstrate that blocking the proteasome may actually work. “There has been very impressive antimyeloma activity even … when no other available therapy is effective,” says oncologist Ken Anderson of the Dana-Farber Cancer Institute in Boston, who is directing a phase II trial of the drug. “That accounts for our excitement.” But side effects remain an issue, and Dave McConkey of the M. D. Anderson Cancer Center in Houston points out that PS-341's mechanism of action is complex and still poorly understood. “I don't think it's the magic bullet,” he says.

PS-341 is generating interest largely because it works in a completely new way. Cancer drugs, with few exceptions, go after predictable targets: DNA replication (cisplatin), microtubules (Taxol), growth signaling pathways (Gleevec), and angiogenesis (for example, endostatin). Proteasome inhibitors are “a new group of agents aimed at a novel target,” says John Wright, a senior investigator in NCI's Cancer Therapy Evaluation Program. Still, NCI took a gamble in funding clinical trials, because protein breakdown, or proteolysis, is so central to normal cell function. “I think luck has to be on your side” for this approach to work, says British cell biologist Paul Nurse, co-recipient of this year's Nobel Prize in physiology or medicine.

How does PS-341 work? When the proteasome is blocked, proteins, instead of disintegrating, build up in the cell. This is ultimately fatal, because constant protein degradation, or “turnover,” is necessary for proper cell function. Proteins called cyclins, for example, must disappear for the cell cycle to proceed through cell division. Adams, a medicinal chemist and veteran of industry giants Merck and Boehringer Ingelheim, originally expected that his drug, by allowing cyclin buildup, would arrest cell division and halt tumor growth. PS-341 does seem to do this, but it seems to kill tumors in a variety of other ways as well.

“The most important mechanism is likely to be NFκB inhibition,” says McConkey. NFκB is a transcription factor that triggers expression of proteins that promote tumor and tumor blood vessel growth. It also blocks apoptosis, or programmed cell death, perpetuating cancer cells. By forcing the buildup of a protein that prevents NFκB activation, PS-341 seems to starve tumors of their blood supply and growth stimuli, thereby promoting their self-destruction.

On a cellular level, blocking the proteasome generally stresses cancer cells by jamming them with proteins. Adams believes that cancer cells may be selectively vulnerable to PS-341 because they can't handle the stress of the protein buildup as easily as normal cells can. This stress causes “catastrophic signaling events, which drive the tumor cell to die,” explains Adams. “A normal, untransformed cell can withstand the stress response, at least for short periods of time.” For that reason, intermittent dosing—once or twice weekly, for a limited time—is crucial. The rest periods are designed to allow the proteasome in normal cells to recover. “We keep patients' dose below a level of 80% proteasome inhibition,” says Adams. “There's a stress to the host, but a tolerable stress.”

The clinical results, from a multicenter trial sponsored by Millennium and headed by Dana-Farber, announced at ASH seem to bear this out. Of 54 myeloma patients, more than half experienced major tumor shrinkage, and the drug halted tumor growth in most of the others. Because myeloma is currently incurable, oncologists are jubilant, although they express some misgivings. “We have something to whack myeloma with we didn't have a year ago,” says James Berenson, director of myeloma programs at Cedars-Sinai Medical Center in Los Angeles. “That's pretty cool.” But, he adds, “this is not an easy drug.” Some patients experience such severe pain after taking PS-341 that they refuse to continue treatment. But enough have benefited that more than 30 separate clinical trials are now under way, sponsored by the NCI and by Millennium, in a wide range of cancers, including breast, colon, lung, and prostate.

Even if acute side effects can be managed, the long-term effects of partial, periodic proteasome inhibition in humans are unknown. “We definitely need to understand the effects of partial inhibition before we can make sweeping statements about why it's not toxic to normal cells,” says McConkey. Primate studies have only gone out 3 months.

Already, it's clear that PS-341 is not the ideal proteasome inhibitor, because the drug indiscriminately raises levels for hundreds of proteins without regard to their anticancer effect. Millennium is now trying to develop inhibitors upstream of the proteasome, by tagging proteins for survival even before they're sent to the proteasome for destruction. If a drug could inhibit specific enzymes that attach ubiquitin to individual proteins (ubiquitin chains mark proteins for destruction in the proteasome), it could, in theory block degradation of only those proteins thought to have a direct anticancer effect—for example, tumor suppressor gene products.

So proteasome inhibition, in all its guises, has arrived as an anticancer strategy, although no one, including Millennium, is quite sure how best to apply it. “They have a golden nugget, but they're going to have to figure out how to make it into a golden ring,” says Berenson. Only time will tell if PS-341 ultimately proves useful in the clinic, but the drug has already shown that playing with garbage has its rewards.

13. PRIMATE ORIGINS MEETING

# New Fossils and a Glimpse of Evolution

1. Anne Simon Moffat*
1. Anne Simon Moffat is a freelance writer in Chicago.

CHICAGO— In an all-too-rare occurrence, paleontologists and molecular biologists met here 13 to 15 December 2001 to share their data and their often very different perspectives. The gathering, the “First-Ever International Conference on Primate Origins and Adaptations: A Multidisciplinary Perspective,” was organized by the Field Museum and Northwestern University. Hot topics included descriptions of six exceptionally well-preserved fossils of archaic primates and of how primates evolved color vision.

## Fresh Look at Primate Ancestors

Small, furry, with large eyes, grasping hands, and a fondness for hunting insects: That's one of several popular images anthropologists have painted of the ancestors to primates. The first undisputed primates appear in the fossil record about 55 million years ago in the Eocene. The problem, however, has been a dearth of fossils before that time, in the Paleocene, to back up conjectures about primate ancestors. Vast movements of rock, earth, and water over tens of millions of years crushed the fragile fossils, usually leaving only scattered bones and teeth as evidence. Worldwide, less than a half-dozen incomplete Paleocene primate skeletons have been described.

Now, vertebrate paleontologist Jonathan I. Bloch and undergraduate Doug M. Boyer of the University of Michigan, Ann Arbor, and their colleagues have added considerably to this data bank. At the meeting they described their full cache of six exceptionally well-preserved, complete skeletons, dissolved out of freshwater limestone in the brightly colored badlands of the Clarks Fork Basin of Wyoming—world-renowned fossil beds of the Paleocene and Eocene eras of 65 million to 53 million years ago.

The stone blocks, although stored in the museum basement for a decade or so, have been painstakingly probed for fossils only in the last 3 years. Already, the new fossils are forcing paleontologists to reexamine and expand their earlier conceptions about archaic primates. The newfound fossils represent four of the 13 families of plesiadapiforms, or archaic primates, dating to about 56 million years ago: Carpolestes simpsoni, a new species of Ignacius, Plesiadapis cookei, and an as yet unnamed genus and species of micromomyid. The skeletons “are amazingly complete, a rarity for fossils that old,” says Northwestern University evolutionary morphologist Matt Ravosa. “They blew people away.”

Here's what the new evidence suggests:

Carpolestes simpsoni was a committed arborealist, weighing about 100 grams and capable of moving on large vertical supports and of grasping slender supports with its hands and feet. It sported a nail on its big toe, indicating specialized tactile and grasping abilities. “This fossil documents the first evidence of strong pedal grasping with an opposing big toe in a Paleocene primate,” says Bloch. Equally important, the morphology of the ankle bones, showing marked mobility, suggests that Carpolestes was not a leaper. And although some plesiadapiforms may have been gliders, closely related to the modern flying lemur, Carpolestes's lack of slender limbs and relatively short fingers on this fossil show that it was not.

With its flexible back, the 400-gram Ignacius could scamper both on the ground and up trees, as do modern squirrels. Judging from its teeth, Ignacius ate a generalized diet of insects and fruits, with a large helping of tree sap. Instead of gliding, it probably had a bounding gait, propelled by its hindlimbs.

Plesiadapis cookei, the only skeleton of this species ever unearthed, spent most of its time in trees, unlike its European relatives, which may have lived on the ground like marmots. It weighed about 4 kilograms, was a relatively slow and deliberate climber, and may have suspended itself from trees, its limb proportions suggest.

Tiniest of them all was the micromomyid, weighing in at about 20 grams. Earlier fossils of this animal included only isolated specimens of bones, jaws, and teeth, including an oddly shaped premolar, leading paleontologists to suggest that it ate fruits and insects. This first known skeleton suggests a highly arboreal creature that also may have hung from trees.

“Before these discoveries, plesiadapiforms were thought to have a limited range of movement,” says Ravosa. “This work shows they were diversifying in ways mirrored by later primates.” Bloch goes one step further, suggesting that these plesiadapiforms are a sister group to primates, an idea that has been spurned in recent years.

## Seeing Red

Back in the late Cretaceous, early placental mammals saw the world in limited colors, much like humans with red-green color blindness do. Then, about 35 million years ago, after New World monkeys set off on their own evolutionary path, the common ancestor of Old World monkeys and apes evolved full trichromatic vision. For more than 100 years, evolutionary dogma was that primates acquired Technicolor vision to enable them to find ripe fruit. But in the past few years, Peter Lucas and Nathaniel Dominy of the University of Hong Kong and their colleagues have argued that the driving force was leaf color, not fruit color—and they brought new evidence to the meeting to buttress their case.

The first blow to the common wisdom came in 1996, when Gerald Jacobs of the University of California, Santa Barbara, found that both the males and females of one species of New World monkeys—howler monkeys (genus Alouatta)—also had trichromatic vision, showing that they had evolved this advantage independently of Old World monkeys. Lucas was intrigued, as these monkeys feed mainly on young leaves rather than fruit. Because tropical foliage often emerges red, not green, Lucas and his colleagues suggested in 1998 that early Old World primates evolved trichromatic vision to better search for these young, reddish leaves, which are remarkably tender and high in protein. Primates able to forage for valuable foliage not clearly visible to other mammals would have a clear nutritional advantage over other leaf-eaters. “If fruits are first in importance in the diets of most primates, then leaves run a close second for many larger species and dominate in some,” says Lucas. And a simple gene duplication increasing the number of retinal cone pigments, he adds, was all that it took to transform a dichromatic primate into a trichromatic one.

Some evidence to support the idea that trichromatic vision evolved to help primates better search for reddish leaves, published last year in Nature, came from a survey of four Old World primates—chimpanzees, black-and-white colobus, red colobus, and red-tailed monkeys—at Kibale National Park in Uganda. Lucas's team observed that these trichromatic monkeys preferred tender, reddish younger leaves to mature green ones. They also asserted that trichromacy didn't evolve to allow primates to see ripe fruits better, because most fruit-eating Neotropical primate individuals are dichromatic.

The reaction to this paper, says Dominy, was mixed. “Lucas has presented a new idea, one that deserves to be taken seriously,” says vision scientist Daniel Osorio of the University of Sussex, U.K. Others were less enthusiastic: “We were questioning dogma dating from Victorian times,” explains Dominy.

At the meeting the team presented as yet unpublished evidence on primates to solidify their case. Some animals studied were dichromatic, some trichromatic, and some polymorphic (in which trichromacy is sex-linked, much as it is in humans). Lucas's colleagues, led by Kathy Stoner and Pablo Riba, gathered data on Costa Rican trichromatic howler monkeys and polymorphic spider monkeys, whereas Nayuta Yamashita studied the more primitive, dichromatic ring-tailed lemurs and polymorphic sifakas—another type of lemur—in Madagascar. Their collected data show, as before, that primates with full color vision prefer meals of redder leaves, and that diet can affect physical character in ways that are subtle but profound.

14. AMERICAN ASTRONOMICAL SOCIETY MEETING

# Latest Observations Bring the Unseen Into View

1. Govert Schilling*
1. Govert Schilling is an astronomy writer in Utrecht, the Netherlands.

WASHINGTON, D.C.— A mere 8.33 light-microseconds from the White House and the Capitol, 2300 astronomers at the American Astronomical Society's 199th meeting gathered on 6 to 10 January to discuss cosmic issues ranging from adaptive optics and brown dwarfs to black-hole companions and galactic coronas.

## Exoplanet Pix Coming Soon?

To capture their first-ever image of an extrasolar planet, astronomers may not need to wait for expensive space-based observatories. Incredibly detailed observations presented at the meeting have convinced some scientists that existing ground-based telescopes can do the job on their own. One researcher, Ray Jayawardhana of the University of California, Berkeley, says he and colleagues are monitoring a number of candidate objects and may produce an exoplanet snapshot within the year. “It's a big prize to win,” he says.

Although astronomers have deduced the existence of exoplanets from the way they tug at their mother stars, no one has ever seen one. Even at infrared wavelengths, in which the glare of a sunlike star is less severe, a giant planet like Jupiter would be only a billionth as bright—far too faint to be seen. But young planets still glowing with the heat of their formation may be 10,000 times as luminous as Jupiter. “They can be detected by 10-meter-class telescopes using adaptive optics,” says Jayawardhana.

Adaptive optics is the revolutionary technique in which ultrarapid undulations of computer-controlled flexible mirrors compensate for image distortions caused by turbulence in Earth's atmosphere. Using adaptive optics, ground-based telescopes can see just as sharply as the Hubble Space Telescope. But large instruments such as the 10-meter Keck telescope and the 8.1-meter Gemini North telescope, both located on Mauna Kea, Hawaii, collect much more light than Hubble's 2.4-meter mirror, so they are more sensitive to faint stellar companions.

At the meeting, Jayawardhana and Michael Liu of the University of Hawaii, Honolulu, presented striking proofs of the technique's power. Jayawardhana announced that he and Kevin Luhman of the Harvard-Smithsonian Center for Astrophysics in Cambridge, Massachusetts, had detected a dust disk around one member of a newborn quadruple star system 900 light-years from Earth in the constellation Aries. Liu and his collaborators found a faint brown dwarf—a “failed star”—orbiting close to a sunlike star 58 light-years from Earth in the constellation Sagitta. “Both discoveries are marvelous,” says Alan Boss of the Carnegie Institution in Washington, D.C., “but these are really just tantalizing appetizers for things to come.”

For example, Boss says, astronomers aren't sure whether the 2-million-year-old dust disk contains enough mass to form large planets like Jupiter. But if it doesn't, others will. “This is just one object—one example,” Jayawardhana says. “The ultimate goal is to trace the whole process of planet formation by studying disks at various ages.”

As for the brown dwarf, Liu says the low-mass companion could be as close to the central star as the planet Uranus is to the sun. “It's the closest brown dwarf companion found so far,” he says, but its true distance from the star is unknown. Future observations will shed light on the brown dwarf's orbit and on the abundance of these objects around sunlike stars.

If adaptive optics can make discoveries like those, a young Jupiter-like planet around a nearby star should be easy to spot, says Jayawardhana, who is carrying out an adaptive optics survey of 150 sunlike stars within a distance of 300 light-years. Jayawardhana says he already has detected “a few candidates”—faint companions to stars whose identities he won't reveal yet—although more observations are needed to confirm that they are real planets and not unrelated background objects. That may take anywhere between 6 and 18 months, he says.

Boss agrees that the first image of a newly forming protoplanet will probably be announced within the next few years, maybe sooner. Such a planet would mark an important step toward showing that “our solar system is not the fluke in the universe,” he says. Meanwhile, Jayawardhana and Luhman are planning follow-up observations on their candidate objects. “There's a race going on,” Jayawardhana says. “If we don't succeed, someone else will do it.”

## Black Holes: Mass Matters

For years, astronomers have puzzled over the nature of a class of strange cosmic middleweights: black holes far less massive than the monstrous beasts that lurk in the cores of many galaxies, but apparently much heftier than the run-of-the-mill stellar black hole that remains after a massive star explodes as a supernova. These intermediate-mass black holes have been known only by the x-rays they emit—until now. Astronomers at the University of Michigan, Ann Arbor, have spotted light from stars whirling around them, an important first step toward determining their true character. And other studies suggest that these objects could be surprisingly numerous.

Black holes do not emit radiation, so they can't be seen directly. Instead, astronomers observe the x-rays from hot gas clouds that swirl around the hole before they are sucked into oblivion. Common wisdom among astrophysicists has it that more-massive black holes suck in more gas, so they emit more- powerful x-rays.

If that simple relation holds, the ultraluminous x-ray sources discovered a few years ago in the outer regions of some spiral galaxies must be black holes hundreds or even thousands of times the mass of the sun. No single supernova explosion could ever produce such massive black holes, so astronomers assume that they have grown fat by merging with others (Science, 23 April 1999, p. 566; 29 June 2001, p. 2426).

To check this scenario, astronomers need to have an independent—and more accurate—way to determine the mass of the presumed middleweights. Finding the so-called optical counterparts of the x-ray sources is a crucial step toward that goal. By carefully matching observations from NASA's Chandra X-ray Observatory with photos made by the U.S.-European Hubble Space Telescope, Joel Bregman, Jifeng Liu, and Patrick Seitzer identified the faint optical counterparts of a handful of middleweight black holes in the galaxies M81 and M51.

The optical light must come from a normal star that orbits the black hole while slowly but surely being devoured by the hole's strong gravitational pull. Measuring the orbital velocities of these stars, using spectroscopy at large ground-based telescopes such as the Keck telescope in Hawaii, would provide astronomers with a solid mass determination for the black hole. According to Liu, such velocity measurements are already planned.

“Getting velocities would be great,” agrees John Tomsick of the University of California, San Diego. “It would be the best evidence” for the high mass of the black hole. But Tomsick points out that the black holes for which the counterparts have been found are not extraordinarily bright in x-rays. As a result, “their masses may be relatively low,” he says. “It would be more interesting to find a counterpart to the ultraluminous x-ray source in the galaxy M82,” which shines about 1000 times as brightly in x-rays.

Meanwhile, other researchers are finding that middleweight black holes may be more common than astronomers have believed until now. Theoretical studies by Kayhan Gultekin, Coleman Miller, and Douglas Hamilton of the University of Maryland, College Park, also presented at the AAS meeting, show that such black holes may form through gravitational interactions and mergers of smaller black holes in the cores of globular clusters—great spherical aggregates of old stars that populate the halo of the Milky Way galaxy.

Another possible birthplace is the dense central region of a smaller cluster of newborn stars, according to detailed computer simulations by Steve McMillan of Drexel University in Philadelphia and Simon Portegies Zwart of the University of Amsterdam. “There may be at least 10 intermediate-mass black holes in the Milky Way alone,” Gultekin says.

## Milky Way Gains a Crown

Using NASA's Far Ultraviolet Spectroscopic Explorer (FUSE), astronomers have discovered that the Milky Way galaxy sits in the middle of a huge bubble of hot gas. Temperatures in this tenuous “corona” are about 1 million degrees, but the average gas density is less than a billionth of a billionth of an atmosphere.

The existence of hot gas around our galaxy was first predicted in 1956 by the late Lyman Spitzer of Princeton University. Spitzer realized that supernova explosions in the Milky Way's disk would energize the interstellar medium, which would then expand into a bloated halo, much as Earth's upper atmosphere expands when it is struck by solar flares. FUSE confirmed Spitzer's general idea a few months after its launch in 1999 by glimpsing the halo. But the extended corona that the satellite has now discovered is much larger than Spitzer imagined and may have been formed by a different mechanism. It may extend as far as the Magellanic Clouds, the closest neighboring galaxies of the Milky Way, says team leader Ken Sembach of the Space Telescope Science Institute in Baltimore.

Working with Blair Savage, Bart Wakker, Philipp Richter, and Marilyn Meade of the University of Wisconsin, Madison, Sembach mapped the corona by studying clouds of gas that rain down on the Milky Way. Like meteors penetrating Earth's upper atmosphere, their surfaces are heated by friction between the clouds and the corona. By observing these hot cloud boundaries, the team was able to deduce the properties of the corona.

The gas clouds, which consist mainly of neutral hydrogen, are known as high-velocity clouds because they fall toward the Milky Way with speeds of a few hundred kilometers per second. According to Hugo van Woerden of the University of Groningen, the Netherlands, some high-velocity clouds may consist of gas that has been blown out of the plane of the Milky Way in a “galactic fountain” and now falls back. Others may represent genuine infall from intergalactic space—evidence that the Milky Way is still accreting mass.

Oxygen atoms in the outer parts of the high-velocity clouds are heated so much that they lose five of their eight electrons. This ionized oxygen absorbs radiation from more distant objects. Sembach and his colleagues studied the far-ultraviolet radiation of dozens of distant quasars and found the telltale absorption features at velocities that matched those of the high- velocity clouds.

Sembach says that the original, smaller halo and the new, extended corona are probably two different things. “The corona is not produced or maintained by events in the disk of the galaxy,” he says. Van Woerden agrees: “Such an extended corona cannot be produced by supernovas. It must be a relic of the formation and the continuous evolution of the Milky Way.”

Looking farther from home, a team led by Edward Murphy of the University of Virginia in Charlottesville studied a smaller halo of hot gas above and below the central plane of the galaxy NGC 4631, also known as the Whale galaxy because of its apparent shape. This starburst galaxy, 24.5 million light-years from the Milky Way, produces many supernova explosions and is known to be embedded in a cloud of x-ray- emitting gas. “But the x-rays represent less than 1% of the total supernova energy,” Murphy says. FUSE solved the mystery by showing that the gas releases its energy in the far ultraviolet. “We are seeing the gas cooling off” before it falls back into the galaxy, Murphy says.

FUSE has been out of order since 10 December 2001 because of a problem with one of its reaction wheels, which are used to point the telescope. Flight controllers hope to resume science operations within a few weeks, and scientists are keeping their fingers crossed.