News this Week

Science  24 Dec 1999:
Vol. 286, Issue 5449, pp. 2430
1. RESEARCH MANAGEMENT

New NIH Rules Promote Greater Sharing of Tools and Materials

1. Eliot Marshall

As one of his final acts as National Institutes of Health (NIH) director, Harold Varmus last week approved controversial new guidelines that set ground rules for sharing research tools. But it will be up to his successor to reconcile the opposing views of buyers—biomedical researchers and large drug companies—who should be pleased with the increased access to new materials that it affords, and some biotech entrepreneurs, who could decide not to share materials on NIH's terms to avoid giving away the store.

NIH's TOOL-SHARING PRINCIPLES

NIH asks the scientists it funds not to agree to any terms for sharing materials that would give away “excessive” editorial control or delay publication of a paper more than 60 days.

Use Patents Appropriately

NIH discourages scientists from patenting and granting exclusive licenses on research tools—defined as products that are not likely to be submitted to the Food and Drug Administration for commercial use. If an exclusive license is necessary to attract investment, however, NIH asks that scientists insist on “reasonable” marketing terms.

Minimize Impediments to Academic Research

Whether as an initiator or respondent, NIH-funded scientists should try to simplify the paperwork involved in sharing research tools. NIH strongly discourages “reach-through” claims on inventions arising from shared tools.

Disseminate NIH-Funded Resources

NIH encourages the scientists it supports to share tools on generous terms both with nonprofit and profit-making research institutions.

Varmus, who next month takes up his new job as president of the Memorial Sloan-Kettering Cancer Center in New York City, is a longtime advocate of improving access to research tools, particularly transgenic mice. Soon after taking charge of NIH in 1993, he began to pressure universities and companies to refrain from patenting or imposing restrictions on the sharing of genetic data and research animals. In 1997, he commissioned a review of federal patent law and urged his advisers to find ways to take lawyers out of the picture. Last year, the group proposed ways to encourage materials sharing, and Varmus asked the NIH Office of Technology Transfer to develop new guidelines based on those proposals. A draft version released in May 1999 was largely welcomed by academics but criticized by officials of some biotech companies. Last week Varmus authorized the release of the guidelines on the Internet, although they're not expected to appear in the Federal Register for another week or two (www.nih.gov/od/ott/RTguide_final.htm). Whereas one university official called it a “good step” in resolving a difficult issue, a biotech executive decried it as “an unmitigated disaster” and derisively called it “Varmus's revenge.”

The guidelines attempt to meet two obligations: to share NIH-funded reagents and to commercialize any inventions under the Bayh-Dole Act of 1980. NIH's tech transfer officials say it's possible to do both by discriminating between inventions that ought to be controlled by tough legal agreements and those that should not. Clear ownership claims, including exclusive marketing rights, may be needed for discoveries that require additional investment and development, although NIH argues that such agreements should be executed in a way that guarantees “widespread” distribution of research tools, on “reasonable terms.” But any advance that can immediately be exploited “primarily as a research tool” should be disseminated without exclusive licensing. Purely commercial inventions such as drugs fall into a different category; they usually require patents and exclusive licensing.

NIH lays out four principles for handling such research tools. First, scientists who receive federal funds must avoid signing agreements that stifle academic communication. Any materials transfer agreements that impose “excessive” editorial control or might delay publication by more than 60 days are “unacceptable.” Second, scientists should not seek or agree to exclusive licenses on “research tools,” which are defined as inventions whose “primary usefulness” is “discovery” and not a product to be approved by the Food and Drug Administration. Third, academic scientists should “minimize administrative impediments” on materials exchanges by refusing “unacceptable conditions.” For example, NIH says, scientists should avoid using materials linked to “reach-through” legal provisions claiming broad rights to all future discoveries that might be linked to use of the materials. Fourth, academic institutions should be as flexible in dealing with others (including companies) as they would have others be with them.

University licensing officials generally support the goals, if not every detail, of the new policy. Joyce Brinton, director of Harvard University's technology licensing office, says the NIH principles are “a good step” because they may help academic institutions resist “unreasonable demands” from providers. But implementing the policy may be difficult, she warns. “Unless the for-profit sector is willing to lessen its demands” for control over research tools, Brinton wrote to NIH earlier this year, NIH's objectives “will not be met.”

A few large pharmaceutical companies—such as Glaxo Wellcome and Novartis—wrote NIH last summer in support of its efforts to free research tools from intellectual property constraints. But smaller biotech firms, whose survival may depend on selling such research tools, are not as enthusiastic. William Haseltine, president of Human Genome Sciences in Rockville, Maryland, asked NIH in August to “withdraw” the draft guidelines, saying they were “illegal” and would make it difficult for his company to work with NIH grantees. Haseltine declined to comment on the final text.

Richard Burgoon, vice president and general counsel of Arena Pharmaceuticals in San Diego, says many executives may not want to comment publicly on NIH's guidelines because they don't want to cause offense. Although he fears that some will not want to share new technology with NIH grantees without assurances that they can retain control, he remains “optimistic … that NIH will make a good-faith effort” to remain flexible.

2. CANCER RESEARCH

Bracing p53 for the War on Cancer

1. Elizabeth Pennisi

Sometimes called the “guardian of the genome,” the tumor suppressor protein p53 responds to DNA damage by either shutting down cell division or causing the cell to commit suicide. Either way, p53′s action helps short-circuit tumor formation by preventing cells that have suffered malignant mutations from continuing to grow. Yet the p53 gene itself is susceptible to damage, which is thought to contribute to the development of half of all cancers, including common ones such as skin, breast, and colon cancers. Now, researchers have identified a drug that may be able to restore the normal function of some mutated p53 proteins and might therefore point the way to a new kind of cancer therapy.

To halt cell division or trigger cell suicide, p53 needs to regulate the activity of various genes, which requires that it first bind to the DNA of the genes' regulatory sequences. And researchers have found that many of the mutations that disable p53 cause the protein to misfold, thereby producing a molecule without the rigid three-dimensional conformation it needs for this binding. In the new work, which is described on page 2507, cancer biologist Farzan Rastinejad and his colleagues at Pfizer Central Research in Groton, Connecticut, have come up with a molecular prosthesis that enables mutant p53 to fold correctly. With its proper posture restored, the aberrant p53 can put the brakes on cell division in both lab cultures and in tumors growing in mice, the researchers report.

Their results are just a first step on the long road toward making a drug that can be used in humans. Nevertheless, they represent “an exciting proof of principle of what promises to be a new form of therapy,” says Bert Vogelstein, a cancer biologist at The Johns Hopkins University School of Medicine in Baltimore, Maryland. What's more, Rastinejad adds, because misfolded proteins are implicated in other disorders, including Alzheimer's, cystic fibrosis, and the brain diseases thought to be caused by infectious proteins called prions, “this approach may pertain to a lot of diseases.”

Before searching for a molecule capable of bracing an aberrant p53 in the correct position to attach to DNA, Rastinejad and Pfizer cancer biologist Barbara Foster needed a quick way to tell whether a compound was working. They hit on the idea of using a well-known antibody that recognizes a part of p53 that is exposed only when the protein is in the right conformation. Ultimately, Rastinejad and Foster screened more than 100,000 compounds with the antibody, first identifying compounds that could increase its binding to normal p53, then testing the successful compounds on mutant p53 in the test tube. The compounds that passed that test then went on to the next phase, in which the Pfizer team looked to see which ones could correct a mutant p53 protein in cultured tumor cells.

The researchers found that a few did work in this assay, causing a fivefold increase in the amount of properly folded p53. The compounds also restored the ability of mutant p53 protein to activate genes. To monitor p53 activity, the researchers equipped the cells with the gene for luciferase, an enzyme that can make cells luminescent, linked to control sequences that would cause the gene to be turned on by p53. The p53-restoring compounds, they found, produced a 10-fold increase in the intensity of the luminescence. Because it takes about 5 hours to see these effects, Rastinejad thinks that the compounds aren't fixing existing p53 but rather are ensuring that new p53, which the cells constantly produce in high quantities, maintains its correct fold.

However they work, the booster molecules curbed cancer growth in mice. In animals that received daily injections of the best of these compounds for a week, tumors caused by injecting mice with human skin cancer cells that have mutant p53 genes grew to only half the expected size, and twice-a-day treatments reduced tumor growth even more—by 75%. Twice-daily treatment also completely prevented tumors from appearing in mice that had first been injected with human colon cancer cells. “These researchers seem to have hit on a way of making mutant p53 act like a normal protein,” notes Karen Vousden, a molecular biologist at the National Cancer Institute laboratory in Frederick, Maryland.

Vogelstein cautions that the doses required are too high for the compounds to be practical at this point. Still, Rastinejad says, these results suggest ways to make better compounds. He notes that the 300 compounds that worked have common features—a hydrophobic, or water-hating, end, which likely fits into a hydrophobic pocket in p53, and a positive charge at the other end, which likely attaches to a negatively charged spot on p53. “You [also] have to have just the right distance between the two ends of the molecule” for the molecular brace to fit right, Rastinejad adds. By designing new compounds with similar characteristics, he says, researchers can find molecules several orders of magnitude better at putting the guardian of the genome back on duty.

3. GERMANY

Agencies Tout Reforms, Seek Greater Support

1. Robert Koenig

Berlin—Stung by criticism from an international panel, the leaders of Germany's DFG basic research funding agency and the prestigious Max Planck network of research institutes last week outlined initiatives to make Germany's research system more flexible, better coordinated, and more open to fresh ideas and top researchers from abroad. The reforms also include plans for closer relations between Max Planck researchers and their university counterparts and increasing diversity among DFG peer reviewers. But DFG president Ernst-Ludwig Winnacker and Max Planck president Hubert Markl say that they can only do so much: The German government, they argue, must do its part by easing federal regulations and boosting spending.

The Max Planck and DFG initiatives, announced at an unprecedented joint press conference here, follow a separate move last month by the Helmholtz Association, which represents the country's 16 federally funded national research centers, to centralize some decision-making and foster work in six interdisciplinary areas. All three research organizations are responding in part to outside pressure. The Max Planck and DFG initiatives follow a report in May by an international evaluation commission that criticized the quality of German university research, branded the DFG as stodgy, and urged Max Planck to move more swiftly into new research areas (Science, 4 June, p. 1595). Helmholtz was recently jolted by the government's decision to shift one of its centers to the Fraunhofer applied science network, with hints of further changes. Together, the actions of these major research players signal an openness to reform as the country heads into the new millennium. “We want to become more flexible in pursuing fresh new research fields, more cooperative in our work with universities, and more attractive to top international scientists,” Markl told Science.

At Max Planck, the strengthening of a research planning panel and reforming the way institutes are evaluated should allow it to move more swiftly into new fields. Chagrined by the recent loss of three top scientists to international competitors, Markl also intends to strengthen overseas recruitment efforts. This summer Max Planck will make its first real foray into education by opening several international research schools that will offer advanced scientific degrees in tandem with German universities. Markl is also preparing to launch a program to create interdisciplinary teams from several institutes, starting with three projects in such areas as mixed-phase catalysts for fuel cells and materials analysis.

Winnacker would like to open up the DFG's peer-review system to more women and younger scientists by wrestling some control over appointments from Germany's scientific societies. In an effort to make the system more transparent, DFG will soon publish a first-ever list of its approximately 5000 outside reviewers. It also plans to attract foreign scientists by supporting focused research programs at several universities.

The reforms deviate somewhat from the recipe laid out last spring by the international panel, led by materials scientist Richard Brook, chief of Britain's Engineering and Physical Sciences Research Council. For example, Winnacker rejected the group's suggestion that the DFG use more of a “top-down” approach in setting research priorities for its university grantees. Such directed research, he says, should be limited to nascent fields such as bioinformatics. Markl and Winnacker also noted that the reforms need to be combined with larger changes to the research enterprise. “It's difficult to make reforms when you are handcuffed by overregulation and limited by a lack of funds,” said Winnacker, noting that Germany is falling behind the United States and Japan in the share of its economy devoted to research.

Although she has not responded directly to their challenge, federal Research Minister Edelgard Bulmahn has expressed support for reforming employment practices and has pledged to lobby for increased spending after the current round of budget cutting ends in 2003. “We plan to relieve research institutions and universities from some of these bureaucratic restrictions and strengthen their independence,” she said in a statement. An expert panel that includes Winnacker will begin a study early next year on how to solve such bureaucratic problems.

Some changes already are under way. At a recent meeting, Helmholtz president Detlev Ganten, who directs the Max Delbrück Center for Molecular Medicine, and the Helmholtz oversight board were given more power “to improve the coordination of research” among its facilities, which are intended to meet national needs. Although the six strategic areas cover such broad topics as the “structure of matter and basic physics” and “health research and life sciences,” Ganten says that the government has made it clear that “no center can do just what it wants to do anymore.” This spring the science council will conduct a systematic evaluation of Helmholtz and its centers, which range from Hamburg's DESY synchrotron to Potsdam's research in geosciences. Ganten predicts that the review “will lead to significant changes” in the association and its components.

4. ASTRONOMY

New Tragedy Hits French Observatory

1. Alexander Hellemans*
1. Alexander Hellemans writes from Naples, Italy.

Disaster has struck the Institute of Millimetric Radioastronomy (IRAM) in France for the second time this year. On 15 December, a helicopter en route to the mountain observatory crashed, killing at least three people on board. The passengers—one IRAM technician and three employees of a subcontractor—and the pilot were on a service mission to the observatory, which sits atop the 2552-meter Bure plateau in the French Alps. “We are still entirely in shock,” says IRAM astronomer Michel Bremer.

In July, a cable car servicing the observatory came loose from its cable, plummeting 80 meters to the ground and sending 20 IRAM staff and subcontractor employees to their deaths (Science, 9 July, p. 181). It's not clear what caused last week's crash.

The observatory, which is currently accessible only by helicopter because of wintry conditions, has been operated by a small crew since the July accident, says Bremer. Run jointly by the French basic research agency CNRS, Germany's Max Planck Society, and Spain's National Geographical Institute, the observatory is in the process of being linked to a 30-meter telescope at Pico Veleta in southern Spain; together, they will form a so-called very large baseline millimeter array, a virtual telescope the size of the distance between the two observatories. The accident is expected to slow completion of this effort even further, Bremer says.

5. HUMAN GENETICS

Checkpoint Gene Linked to Human Cancer

1. Michael Hagmann

As any driver knows, reliable brakes are every bit as important to safety as the gas pedal. The same can be said about cells when it comes to dividing. They have to know when to stop, say, when their chromosomes have been damaged, because if they don't the resulting mutations may propel them down the road to cancer. Over the past several years, a great deal of work, much of it in yeast, has identified a network of proteins, called “checkpoints,” that helps cells sense damage and put on the brakes. Now researchers have linked mutations in one of these checkpoint proteins to cancer.

On page 2528, a team led by cancer geneticist Daniel Haber of Massachusetts General Hospital (MGH) in Boston reports that mutations in a known checkpoint gene called hCHK2 cause some cases of Li-Fraumeni syndrome (LFS), a hereditary cancer susceptibility that leaves its patients prone to developing any of several cancers, including breast and brain cancers and certain leukemias. This is not the first gene linked to LFS. In 1990, Stephen Friend's team, also at MGH, found that inherited mutations in the well-known tumor suppressor gene p53 can cause the condition. Subsequent work showed that p53 mutations account for only about 75% of the cases, however. The new work provides an explanation for some, although not all, of the remaining LFS cases. And even though the number of LFS patients may be small—only about 200 families worldwide have been reported—the discovery of hCHK2 and additional LFS defects in the future may “help [us] to understand the molecular mechanisms of tumorigenesis” reaching far beyond LFS, says Friend, who is now at the Fred Hutchinson Cancer Research Center in Seattle, Washington.

The finger of suspicion already pointed at cell checkpoints as being important. Indeed, the p53 protein itself halts cell division in response to chromosomal damage. So Haber and his team studied several members of four LFS families that did not have p53 mutations, looking for mutations in the human counterparts of genes previously identified in yeast as playing a major role in checkpoint control. Most such genes turned up as perfectly normal. But in one family, three LFS patients had identical mutations in one copy of the gene encoding chk2, a kinase enzyme that passes on the stop signal in damaged yeast cells by attaching phosphate tags to other proteins. The protein produced by the mutated gene would be unable to perform this function, Haber says, because part of it, including its enzymatic center, is missing. A healthy relative, in contrast, had a normal hCHK2 gene.

Haber's team next looked at hCHK2 in 18 patients suffering from a related syndrome called variant LFS and in 49 cancer cell lines from a variety of nonhereditary human tumors. The researchers found one individual with a mutation similar to that in the first family. The gene from another individual and from one of the cancer cell lines had different mutations, changing one amino acid to another. Although Haber doesn't know for sure whether these “spelling errors” debilitate the kinase, he notes that his team failed to detect them in the gene from 50 healthy control individuals. “This suggests that these alterations are not simple sequence variants that are prevalent in the general population,” says Haber.

The results are likely to receive a warm welcome in the cancer community. “This is great. People have been searching for mutations to explain LFS [in families with intact p53] for almost a decade and [have] found absolutely nothing,” says Friend. But he adds that, because hCHK2 mutations turned up in only one of the four families studied, “there is a good likelihood that [other LFS families] will have mutations in other interesting genes.” Paul Russell, a yeast cell cycle expert at The Scripps Research Institute in La Jolla, California, notes that Haber's results point to the most obvious candidates. “One wonders whether different cancers could be explained by mutations in human versions of some of the other half-dozen or so yeast checkpoint genes around,” he says.

And because LFS patients with p53 or hCHK2 mutations are virtually indistinguishable, Haber thinks there may be a link between the two proteins. “The most fascinating possibility is that p53 is directly phosphorylated by chk2,” Haber speculates. For cell cycle expert Stephen Elledge of Baylor College of Medicine in Houston, Texas, a direct path from DNA damage via chk2 to p53 “makes perfect sense.” He notes that although yeast does not have a p53 gene, the organism makes other proteins that, when phosphorylated by chk2, induce a cell cycle stop, much as p53 does.

Still to be worked out are the details of where chk2 fits into the checkpoint control program in human cells. But, says Thanos Halazonetis of the Wistar Institute in Philadelphia, whose as yet unpublished results support a direct chk2-p53 link, “the interesting thing is that genes mutated in cancer fall in a very small number of signaling pathways, and the p53 pathway—including chk2—is likely the most important one.”

6. DEFORMED FROGS

Link to Parasites Grows Stronger

1. Jocelyn Kaiser

Philadelphia—As scientists labor to unmask the villain behind a rash of frog deformities across the United States, a suspicious character previously linked to this odd crime in California has now turned up in misshapen amphibians throughout the Northwest. The suspect—a parasitic flatworm, or trematode—has also been found in the Minnesota pond where the discovery of dozens of frogs with twisted, missing, or extra legs touched off a hunt for the perpetrator.

Linking trematodes to more crime scenes doesn't mean the case is closed—far from it. Abnormal frogs from some ponds still test negative for the parasite, sustaining the notion that chemicals or high doses of ultraviolet (UV) light might also be messing with frog development. “Without question there are other things that can cause [deformities],” says ecologist Pieter Johnson of Claremont McKenna College in Claremont, California, who described his team's trematode findings here last month at the annual meeting of the Society of Environmental Toxicology and Chemistry. But the circumstantial evidence suggesting that the worm is a major culprit has researchers worried that it is being nourished by a surfeit of nutrients, mainly chemicals in fertilizers, building up in U.S. watersheds.

Since students at a Minnesota middle school chanced upon misshapen northern leopard frogs on a field trip 4 years ago, deformities have been reported in more than 50 amphibian species in 44 states. Some scientists worry that the frogs are a “canary in a coal mine,” the earliest victims of a developmental poison that may end up harming humans—too much UV light penetrating the thinning ozone layer, for example, or pollutants such as pesticides.

In a step toward unraveling this mystery, Johnson and his colleagues reported last spring that the trematode Ribeiroia burrows into tissue around the pelvic area, where a tadpole's limbs begin forming. There, the parasites encase themselves in cysts that may influence limb development by pushing cells around or by secreting hormonelike chemicals. Besides finding the parasites in Pacific tree frogs with extra or missing legs in northern California, the team infected tadpoles in the lab with the trematode, raised them to metamorphosis, and observed deformities mirroring those seen in the field (Science, 30 April, p. 802).

Wondering if frogs outside California are also falling victim to the dread worm, the researchers spent last summer crisscrossing six northwestern states in a van, collecting frogs, toads, and salamanders from 103 ponds, including 42 ponds where deformities were found in six species at rates ranging from 5% to 90%. The misshapen amphibians at 40 of 42 ponds had Ribeiroia, while those from normal ponds almost never had the parasite. A brief search in Minnesota also turned up the trematode—including at the Ney pond, where deformed frogs were first spotted, and another hot spot. Bolstering its fieldwork, the team has shown that trematodes can cause deformities in the lab in a more terrestrial amphibian: the Western toad (Bufo boreas), another denizen of the afflicted ponds. “The fact that they can induce [deformities in] another species gives [the theory] more breadth,” says Andrew Blaustein, an ecologist at Oregon State University in Corvallis.

The findings do leave the chemical theory a leg or two to stand on. Although the northwestern waters tested free of pesticides, says Johnson, many of the ponds have a “long history of fertilizer input or cattle grazing.” He speculates that such nutrients could be an accessory to the crime by spurring algal growth, which in turn would boost populations of Ribeiroia's primary host, an aquatic snail. Others see a more direct role for chemicals. A group led by toxicologist Jim Burkhart of the National Institute of Environmental Health Sciences in Research Triangle Park, North Carolina, has found that the water itself from the Ney pond and other sites can cause deformities in parasite-free African clawed frogs, a widely used lab species. “It's not either-or,” says Burkhart. “There are factors in the water that contribute to malformations.” He believes that mixtures of unidentified hormonelike chemicals in the water, as well as the trematodes, each can trigger deformities. And they may work in concert, Burkhart says: Chemicals could be predisposing the frogs to trematode infections by weakening their immune systems.

So far, it hasn't been shown that trematodes are killing off significant numbers of frogs—they have only been blamed for deformities—so they don't appear to play a role in the worldwide decline of amphibians, notes parasitologist Peter Daszak of the University of Georgia, Athens. But Blaustein has a prime murder suspect: He's found that even low concentrations of nitrates from fertilizers can directly kill larvae of several Western species in decline, including the Cascades frog, one of the species with deformities. “The message on the effects of fertilizers is important,” Blaustein says. “Fertilizers are everywhere.”

Possible Clock Messenger Identified

1. Marcia Barinaga

A clock is useless unless it has an output—hands, a digital display, or an alarm. The same goes for the 24-hour molecular “clock” that ticks in organisms from bacteria to humans. To impose its rhythm on behaviors such as activity, sleep, and feeding, the oscillating molecules that make up the clock must communicate, through some kind of outgoing signals, to the brain areas that drive those behaviors. Now researchers working in fruit flies have for the first time put their hands on a good candidate for such a messenger.

In this week's issue of Cell, a team led by Paul Taghert at Washington University in St. Louis and Jeff Hall at Brandeis University in Waltham, Massachusetts, reports evidence that a peptide called PDF is a key outgoing clock signal. The researchers have shown that flies in which the pdfgene has been inactivated have functional clocks but lose their rhythmic activity patterns under certain conditions, suggesting that a signal from the clock that affects activity is missing.

The finding builds on earlier hints, including one reported in Cell 2 weeks ago by Michael Young of The Rockefeller University in New York City and postdoc Justin Blau, who found that the clock genes regulate PDF production, as would be expected if it were an output signal from the clock. If PDF does turn out to be a bona fide output signal, says clock researcher Paul Hardin of the University of Houston, it will help researchers to “ultimately map out a pathway” from the clock to behaviors that it controls.

The first clues that PDF (for pigment-dispersing factor) might be involved with the clock came from its resemblance to a peptide called pigment-dispersing hormone, which drives a daily rhythm of color changes in some crustaceans. Rhythm researcher Charlotte Helfrich-Förster at the University of Tübingen in Germany provided further evidence for that idea in 1993 when she discovered that PDF is made in certain of the so-called lateral neurons of the fruit fly: clusters of neurons on each side of the fly's head where its clocks are housed. What's more, she showed in 1998 that those neurons make direct connections to the part of the fly brain that appears to control the fly's activity level, and that those connections are essential to maintain daily activity rhythms. That strongly suggested that something made by the lateral neurons acts as the messenger controlling clock activity. PDF was a good candidate, but researchers had not yet knocked out the gene to test that idea.

Then last December, Taghert and graduate student Susan Renn were studying neuropeptide expression and in their lab stocks found flies that were missing PDF. Further testing showed that the flies had a mutation in the pdf gene. To check out whether the mutation affects the flies' clocks, Renn and Taghert joined forces with clock researchers Hall, Michael Rosbash, and postdoc Jae Park, who were already studying pdf gene expression. The researchers found that the animals have working clocks; for example, in normal light-dark cycles, the flies became active in anticipation of darkness, a sure sign that their clocks were running. But in constant darkness, their activity rhythms gradually faded away over several days. Because PDF is apparently not part of the clock itself, the fading suggests that it contributes to the output signal that controls activity rhythms, says Taghert. It can't be the only signal, because flies still have activity rhythms under some conditions without it, he notes, but it seems to dominate under constant dark conditions.

If PDF is such an output signal, it must be controlled by the clock. And that appears to be so. In their Cell paper, Blau and Young described the discovery of a new clock gene, vrille, and reported that both vrille and another clock gene, clock, regulate PDF: clock controls the expression of the pdfgene in the lateral neurons, and vrille controls the accumulation of the PDF peptide. Surprisingly, PDF production remains level throughout the day, but it might be released rhythmically from the lateral neurons, which would explain its rhythmic effects on activity.

The data also suggest another possible function for PDF, as a signal that helps to synchronize the pair of fruit fly clocks. If the two clocks in the lateral neurons were to get out of time with each other, the fly's rhythms would disintegrate into chaos. Normally light should synchronize the clocks, but in constant darkness, they could drift apart without a backup. In that case they could show a gradual loss of rhythmicity similar to what happened in the PDF mutants kept in the dark. Helfrich-Förster found that some of the PDF-containing lateral neurons crossed to the other side of the brain, where they could act as a backup synchronizer.

Right now researchers can't tell whether PDF exerts its effects on activity patterns directly, by synchronizing the clocks, or both. Researchers might be able to resolve that issue by selectively destroying the lateral neurons that cross to the other side of the brain, but not the ones that project to the activity- controlling area. And if they could find the receptor through which PDF exerts its effects, they should be able to identify the neurons that respond to it. That would enable them to confirm that it controls daily activity rhythms and perhaps find other clock-controlled behaviors as well, says Rob Jackson of Tufts University College of Medicine in Boston. Indeed, even though it is not yet certain that PDF really is the clock's output signal, the flurry of speculation about its possible roles has begun.

8. HUMAN GENETICS

mtDNA Shows Signs of Paternal Influence

1. Evelyn Strauss

Women have struggled to gain equality in society, but biologists have long thought that females wield absolute power in a sphere far from the public eye: in the mitochondria, cellular organelles whose DNA is thought to pass intact from mother to child with no paternal influence. On page 2524, however, a study by Philip Awadalla of the University of Edinburgh and Adam Eyre-Walker and John Maynard Smith of the University of Sussex in Brighton, U.K., finds signs of mixing between maternal and paternal mitochondrial DNA (mtDNA) in humans and chimpanzees. Because biologists have used mtDNA as a tool to trace human ancestry and relationships, the finding has implications for everything from the identification of bodies to the existence of a “mitochondrial Eve” 200,000 years ago.

The study “is pretty compelling and I can't think of good alternative explanations,” says Richard Hudson, a population geneticist at the University of Chicago. Anthropologists agree that if the study holds up, it could trigger a major shake-up in their field. “There is a cottage industry of making gene trees in anthropology and then interpreting them,” says Henry Harpending, an anthropologist at the University of Utah, Salt Lake City. “This paper will invalidate most of that.”

Yet not everyone is ready to grant a role to fathers in mtDNA inheritance just yet. Hudson and others caution that the new result “changes our view dramatically enough that we have to continue to think of other ways to explain it.” Fathers contributing mtDNA to their offspring “implies several different novel and important biological phenomena that no one's ever seen before,” such as contact between maternal and paternal mtDNA, adds Neil Risch, a human geneticist at Stanford University School of Medicine.

Researchers have assumed that mtDNA passes only through the mother, in part because experiments have shown that eggs destroy sperm after fertilization, and that mitochondrial traits, including a variety of inherited disorders, seem to come only from mothers. But some mtDNA sequences didn't fit neatly into a tree of maternal descent (Science, 5 March, p. 1435), so the scientists decided to look for signs of mixing between paternal and maternal mtDNA.

Such mixing—the usual fate of DNA in the cell nucleus—is called recombination, and it takes place when a piece of a DNA strand from one parent crosses over and pairs up with a strand from the other parent. The process generates a novel DNA molecule with features donated by each parent.

To probe whether recombination occurs in the mitochondrial genome, the team analyzed DNA variations. DNA in different individuals varies at many positions, so each new mutation arises on a distinctive genetic background. Unless the DNA can reshuffle itself, the new mutation will stick with the variations already on the same chromosome as it is passed on. But recombination, which mixes up pieces of the DNA, should gradually destroy such nonrandom linkages between DNA variations. The farther apart two sites lie on the chromosomes, the faster recombination can eliminate the linkage. Thus, if recombination is operating, specific variations are less likely to be found together if they are far apart on the chromosome than if they are neighbors.

Using mtDNA from humans and chimpanzees, the researchers tallied how often specific mutations at different sites tended to occur together; they also noted the distance between the mutation sites. In four out of five human data sets and one chimp set, nonrandom mutations at distant sites were less likely to be linked than nearby mutations—implying recombination between maternal and paternal DNA, says Eyre-Walker.

Such recombination could be a blow for researchers who have used mtDNA to trace human evolutionary history and migrations. They have assumed that the mtDNA descends only through the mother, so they could draw a single evolutionary tree of maternal descent—all the way back to an African “mitochondrial Eve,” for example. But “with recombination there is no single tree,” notes Harpending. Instead, different parts of the molecule have different histories. Thus, “there's not one woman to whom we can trace our mitochondria,” says Eyre-Walker.

What's more, over time, recombination mixes up genomes so that they become more homogeneous. That “makes even distantly related people look more similar to each other,” says Eyre-Walker, and causes past events to seem more recent than they really are. Our last common female ancestor, for example, would be older than the mtDNA implies. But not every mtDNA study would be invalidated by recombination, Eyre-Walker notes. “The major impact will be on the timing of those events and our basic understanding of mtDNA evolution,” he says.

Even so, many researchers aren't ready to accept these data as ironclad evidence of recombination. Other genetic processes might create a similar pattern, says evolutionary biologist Rebecca Cann of the University of Hawaii, Manoa. Some researchers have proposed models in which one mutation is more likely to occur close to another. “It's not yet clear whether there aren't explanations other than recombination,” agrees Vincent Macaulay, a mathematical geneticist at the University of Oxford.

Skeptics and supporters alike note that how recombination could be happening remains a mystery. Recombination requires physical contact between egg and sperm mtDNA, for example, and it's not clear when or how these molecules touch. In any case, it's possible to square previous observations of mtDNA inheritance with “a little bit of paternal leakage,” adds Jody Hey, an evolutionary geneticist at Rutgers University in Piscataway, New Jersey. Just how much leakage might take place is a critical question in practical as well as research uses of mtDNA, such as identifying human remains. “If the sequences are identical, the chances are very good that that's the woman's son or daughter,” says Eyre-Walker. “If you get a one-base-pair mismatch, do you say ‘This is not your child?’”

9. PLANETARY SCIENCE

Galileo Catches Lava Fountain on Io

1. Robert Irion

San Francisco—Astronomers are galvanized by a new image of what may be a curtain of lava spewing above a volcano on Jupiter's moon Io. The picture, snapped by the Galileo spacecraft during its daredevil dive past Io on Thanksgiving and released last week at a meeting of the American Geophysical Union, also reveals a complex, jagged cliff arcing near the volcano—further evidence of the moon's geologic turmoil.

Planetary scientists have long known that Io is the most volcanically active body in the solar system, thanks to constant gravitational tugs from Jupiter and its other moons that churn Io's interior. Galileo had previously revealed surface flows within vast volcanic wounds called calderas. The Galileo team hoped they also might find ribbons of fire called “lava fountains,” similar to those that spurt skyward from long, narrow gashes in the Kilauea volcano on Hawaii. However, Galileo's camera hadn't yet pointed in the right place at the right time to see one.

That changed during a moment of “dumb luck” on 25 November as Galileo passed within 300 kilometers of Io, says Galileo team member Laszlo Keszthelyi of the University of Arizona, Tucson. The camera was aimed near Io's north pole at a nondescript volcanic feature that the team now calls Tvashtar, after an Indian sun god. Heat flowing from a 25-kilometer-long stripe within Tvashtar's caldera was so intense that it blinded part of the camera's electronic detector and created a white blotch. However, details of parts of the blotch—especially its wavy top edge and a linear section in the middle—convinced Keszthelyi and his colleagues that they were seeing a curtain of lava at least 1.5 kilometers high. Key support for that scenario came from NASA's Infrared Telescope Facility on Mauna Kea, Hawaii, which happened to detect the Tvashtar hot spot from Earth just 3 hours later. According to planetary scientist John Spencer of Lowell Observatory in Flagstaff, Arizona, Tvashtar was so close to Io's horizon at the time that the telescope would not have seen the hot material unless it was jetting above the surface.

Still, the lava fountain scenario is “pretty speculative,” says planetary scientist Gerald Schubert of the University of California, Los Angeles. “Because of the bleeding problem [in the image], it's very uncertain exactly what they're seeing,” he says. But Spencer and his colleagues think they may be able to dispel those doubts by determining the three-dimensional shape of the hot spot from earlier observations of Io on the same day, when the view was more direct.

10. CLONING

Mice Cloned From Cultured Stem Cells

1. Gretchen Vogel

In an experiment that links the hot-button fields of stem cells and cloning, scientists announced this week that they have cloned mice from embryonic stem (ES) cells kept in culture. These cells are capable of developing into any type of tissue but ordinarily can't grow into a complete organism on their own. The technique is still inefficient, but if researchers can improve it, it could provide an easier way to create genetically modified mice.

Although sheep, goats, cows, and mice have all been cloned, in most cases the genetic material came from cells that had not spent much time in culture. In contrast, researchers report in the current Proceedings of the National Academy of Sciences that they produced live-born mice from two ES cell lines that had each gone through more than 30 cell divisions in the lab. Although cells in culture are easy to work with, they often accumulate mutations. Those mutations can impede embryonic development, so scientists weren't sure whether long-cultured cell lines would be candidates for cloning.

In the hands of reproductive biologist Teruhiko Wakayama, who created the clones in the laboratory of Ryuzo Yanagimachi of the University of Hawaii, Honolulu, the cultured cells seemed to work as well as cells from a living animal. In collaboration with Peter Mombaerts of The Rockefeller University in New York City, the researchers extracted a nucleus from an ES cell and transferred it into a mouse egg that lacked a nucleus. Like all cloning, the procedure was hit or miss, however. In the team's most successful experiment, it took more than 1000 cloning attempts to obtain 13 mice that survived to adulthood, notes developmental biologist Davor Solter of the Max Planck Institute for Immunobiology in Freiburg, Germany.

If scientists could boost the technique's efficiency and use it to go directly from a genetically manipulated cell to a live animal, “that would be fantastic,” says embryologist Brigid Hogan of Vanderbilt University in Nashville, Tennessee. Cloning mice from ES cells could streamline the rather cumbersome procedure used today to produce tens of thousands of so-called knockout mice a year.

Typically, scientists modify genes in an ES cell and then insert the cell into an already-developing embryo, producing a chimeric mouse that has descendents of the ES cell in most of its tissues, including the reproductive organs. When the chimeras mate with normal mice, some of their offspring carry the mutation in all their cells. Those mice are then bred to create knockout mice. If scientists could clone mice directly from particular cells, they could create knockouts in one generation instead of three, Mombaerts says.

But at the moment, cloning mice is far too difficult to rival the traditional procedure, several mouse researchers say; although other labs have tried, no one else has reported cloning a mouse. “Whether [cloning] is going to be a better way of generating mice than chimeras remains an open question,” says mouse geneticist Allan Bradley of Baylor College of Medicine in Houston, Texas. “I don't think we're going to drop everything overnight and start trying to clone ES cells.” Indeed, Yanagimachi's team reports that they tried to clone mice from ES cells modified to contain a specific mutation, but they produced only one cloned knockout mouse that died soon after birth.

The new work does suggest that in some respects cloning may be easier technically than researchers had thought. Earlier reports seemed to suggest that successful cloning requires starting with cells at the resting phase of the cell cycle; Dolly the sheep was cloned from a so-called “quiescent,” starved cell, and other cloning experiments have used cells in the resting phase. But ES cells don't survive long if they are starved, and they divide rapidly, so it is tough to catch one at the resting stage, says Mombaerts. The team found that cell stage didn't seem to matter: They produced live-born mice from cells in several stages of the cell cycle, including some that presumably had just finished replicating their DNA and were about to divide.

The next crucial step is to standardize the procedure. Mombaerts says he and Wakayama, who has recently moved to Rockefeller University, hope to do their part: They're planning a detailed guide on “how to clone a mouse”–a book that may make many researchers' lives a lot easier.

11. SPACE SCIENCE

Forecasting the Storms and Showers of Space

1. Gary Taubes

As the sun climbs toward its 11-year activity peak, clouds of plasma and bursts of radiation are buffeting the planet. Reliable predictions may be on the way too

A solar storm can be a fearsome thing. A billowing cloud of plasma and magnetic field, millions of kilometers across, descends from the sun. Far out in space, the envelope of charged particles trapped in Earth's magnetic field shudders under the impact. Nearer to Earth, particles showering from space can expose astronauts and passengers on commercial jets to high radiation levels and disrupt some radio communications. Currents surging through the atmosphere and the ground play havoc with technology, knocking out satellites and even entire electric power grids. In March 1989, for example, a particularly nasty solar storm took out the electricity in the entire province of Quebec for 9 hours.

That storm took place during the last peak of the sun's 11-year activity cycle. Next year, the cycle will reach another peak, and the frequency of solar storms will again rise from several each month to several each week. Most of these storms will have only minor consequences, but the most potent could match the 1989 storm or even surpass it. And with the growth of electric power grids and the increasing number of satellites in orbit (800 by the end of next year), not to mention the astronauts who will be building the international space station, society is more vulnerable than ever to the heaviest weather from space. “It's almost a given that bad things are going to happen due to space weather events. Satellites will fail; power grids will go down,” says Richard Behnke, head of the upper atmosphere research section at the National Science Foundation (NSF) and co-chair of the National Space Weather Program.

Just as a storm warning gives time for boats to be secured and windows checked, space weather warnings could allow engineers to avert some damage. Power networks, for instance, can be run at less than capacity to absorb power surges caused by the storms; companies running communication satellites can prepare for potential satellite failures or malfunctions. “It can make a dramatic difference if they have even a short warning,” says Ron Zwickl, deputy director of the Space Environment Center (SEC) at the National Oceanic and Atmospheric Administration (NOAA) in Boulder, Colorado. For power companies, “even an hour will make a dramatic difference.”

Yet space weather experts are no better at predicting solar storms than meteorologists in the 1960s were at forecasting weather on Earth, say experts. For 35 years, the SEC in Boulder has issued gradually improving 24-hour advance predictions of a phenomenon associated with space weather: solar flares, which are explosions in the sun's atmosphere apparently triggered when pent-up magnetic energy is released. But the only reliable warnings of the storms that do the most damage come just an hour out, when a cloud passes the Advanced Composition Explorer satellite, known as ACE, which sits 1.5 million kilometers sunward of Earth. “We are pretty good with 1-day forecasts of solar flares,” says Ernie Hildner, director of SEC, and “we can be 90% certain with predictions based on 1-hour advance in situ measurements upstream.” But he adds, “We are not very good with multiday forecasts of solar storms.”

Fortunately, “tremendous improvements” in predictive capability are on the way, says Hildner, although not in time for next year's solar max. NASA is planning to send up new research satellites to track solar storms from the sun to Earth, and a partnership between NOAA and the U.S. Air Force will put new instruments on weather satellites to provide critical information for improving predictions. The data will feed into a program that has been under way since 1995, when NSF, NASA, NOAA, and the Departments of Defense, Energy, and the Interior launched the National Space Weather Program with the goal of getting accurate and reliable space weather forecasts within a decade. Two years ago, these agencies began to form the Community Coordinated Modeling Center, run out of NASA's Goddard Space Flight Center in Greenbelt, Maryland, to create a single comprehensive computer model that can predict the eruption of storms on the sun and their effects on satellites and power networks on the ground. When complete, participants hope, the model will do for space weather what the computer models of the National Weather Service do for more mundane weather on Earth.

Windy; showers possible

Three solar phenomena drive most space weather. The least serious of the three are coronal holes, features of the sun's corona, or upper atmosphere, that appear darker than the surrounding regions if viewed in extreme ultraviolet or x-rays. Solar coronal holes are places where the sun's local magnetic field connects its surface directly to interplanetary space, allowing the particles of solar wind to blast outward unimpeded—like fire hoses, says Hildner—rather than being slowed, as elsewhere, by coiled loops of field. As the sun rotates, these coronal holes rotate with it, occasionally sweeping the fire hose of outflowing plasma across Earth's path. Earth's magnetosphere—the vast region of charged particles trapped in the magnetic field—contracts and expands as the fire hose hits and then passes by, causing electromagnetic disturbances on Earth.

Solar flares, phenomenon number two, bombard the upper atmosphere with a shower of energetic particles, x-rays, and extreme ultraviolet light, which can be 1000 times more intense than the levels emitted by the sun during quiescent periods. The x-rays and ultraviolet light knock electrons off atoms in the upper reaches of Earth's atmosphere. The resulting low-energy electrons can charge up the surface of a spacecraft, in a phenomenon equivalent to what happens “when you walk across a shag run in your socks,” says Zwickl. When the static buildup discharges, it can damage satellite electronics.

High-energy protons and heavier nuclei arrive from 20 minutes to a few hours after the x-rays, bringing more trouble. They can have a million times as much energy as the ions of the usual solar wind. Fortunately, most are deflected by Earth's magnetic field, but a fraction are able to penetrate into the magnetosphere. Satellites or astronauts without adequate shielding can be in danger. “These energetic particles can increase more than 10 million times above background in the largest events,” says Zwickl. “This presents a radiation hazard for anything flying in space, man or machine.”

The roughest space weather is triggered by the third phenomenon, coronal mass ejections. CMEs are often associated with solar flares, although not always. CMEs are erupting bubbles of solar gases containing tens of millions of tons of solar material as well as a portion of the solar magnetic field, and they expand quickly as they blow out into space. “The current belief is that they are magnetically driven,” says Hildner, “and the sun finds it energetically favorable to reduce its energy by kicking off a blob of its own atmosphere that is permeated by magnetic field.”

When the expanding, magnetized bubble of a CME crashes into Earth's magnetosphere, a few days after leaving the sun, “all hell breaks loose,” says Zwickl. Researchers have assembled a portrait of the havoc from magnetometers and radiation meters riding aboard satellites. Depending on how the bubble's magnetic field is oriented relative to Earth's, the collision can produce a massive distortion of Earth's field. Because a moving magnetic field induces an electric current, the thrashing field lines create a current in the ionosphere—the ionized region of Earth's upper atmosphere—which in turn heats up and expands, increasing the drag on any spacecraft in low orbits. The recoiling magnetic field can also accelerate electrons in the ionosphere, turning them into so-called killer electrons, which can bore deeply into a satellite and permanently disable it.

On Earth's surface, the recoiling field induces current surges in power lines. The greater the magnetic field in the storm, the more rapidly it varies over the power lines, and the longer the power lines on Earth, the greater the induced current becomes. This direct current can unbalance the alternating current on the grid, so that voltage regulators shut down, circuit breakers trip, and the whole system crashes, as happened in Quebec a decade ago.

Sun-gazing

When NOAA got into the space weather business during the Apollo years, researchers tried to forecast these events by looking for erupting solar filaments—bright ribbons of gas rising from the solar disc. These filaments occasionally detach and vanish into the upper reaches of the solar atmosphere, an event that is thought to indicate a change in the structure of the solar magnetic field and that sometimes coincides with a CME. When a filament vanished, says Joe Kunches, lead forecaster at the SEC, the NOAA space weather forecasters concluded that a CME had been emitted—although they couldn't say whether it was heading our way. Kunches describes the prediction as akin to a “wild guess.”

Today, the SEC's Web site (http://www.sec.noaa.gov/) posts outlooks of space weather conditions a week in advance, forecasts a day in advance, and alerts and warnings of space weather storms that are imminent or under way. Underlying these bulletins are data from instruments on weather satellites and ground-based devices, as well as from two key spacecraft: the Solar and Heliospheric Observatory (SOHO), launched in late 1995, and ACE, which went up 2 years later. Neither was dedicated to space weather, but both have instruments that are now used for that purpose.

The predictive power of SOHO, which also hovers 1.5 million kilometers sunward of Earth, comes mainly from a telescope called the Large Angle Spectrometric Coronagraph (LASCO), which blocks out the body of the sun with a small occulting disc, creating an artificial eclipse, and then observes the sun's much fainter corona. Through LASCO, CMEs look like bright bubbles emerging from the corona, explains Simon Plunkett, a solar physicist with the Naval Research Laboratory who works with the SOHO satellite. If a CME is aimed either directly toward or directly away from Earth, it appears as an expanding white ring, a “halo CME.” The catch is that LASCO alone can't tell whether the CME is coming or going.

To help distinguish between the two directions, the SOHO scientists use a second instrument, called the Extreme Ultraviolet Imaging Telescope, which observes the sun in ultraviolet wavelengths, looking for signs of a CME heading this way. For instance, a depletion in ultraviolet intensity suggests that hot material has left the corona and is moving outward. If the dimming is seen near the sun's center, it implies that the radially moving ejected material is moving Earth-ward. “It gives us a reasonably good prediction that we have a CME that might impact Earth,” says Plunkett. “By combining the capacity to image the disc and outer corona, we can tell where the CME comes from and how it's progressing as it moves outward.” Almost always, Hildner says, an ejection seen as a halo CME causes at least some reaction in Earth's magnetic field when it arrives. But predicting an extreme storm requires knowing whether or not the cloud's magnetic field is aligned opposite to Earth's, which allows them to couple. And SOHO can't determine the alignment.

That task falls to the ACE spacecraft, an hour upstream, which sounds the alarm if a CME really is going to hit and indicates whether it will hit hard. ACE also measures the density, composition, and velocity of the passing cloud. “Since we're sitting out in front of the Earth at a distance of about a million miles, we can give some advanced warning that a huge pressure wave is approaching Earth,” says Ed Stone, director of the Jet Propulsion Laboratory and ACE principal investigator.

Weather eye

To improve space weather forecasting, researchers are studying the sun. They hope to learn how to predict when CMEs are about to erupt and whether, as Plunkett says, “they are going to be big, fast ones or small, slow ones of very little consequence.” One way to do so is to study the strong magnetic fields on the sun's surface. “If you have opposite magnetic fields, very strong and very close to each other, you can imagine that they will try to annihilate and there will be the potential for catastrophic energy release,” says Hildner. Researchers can track the strength and position of the fields with a device called a magnetograph, which measures the polarization of sunlight caused by the magnetic field. If the field is strong enough and sufficiently contorted, it is a good bet that it will, in effect, unravel, generating a CME in the process.

Last year two groups—one led by solar physicist Dick Canfield at Montana State University in Bozeman and the other by Ron Moore at NASA's Marshall Space Flight Center in Huntsville, Alabama—reported new ways of analyzing the solar magnetic field to improve CME prediction. Canfield's team used x-ray data from the Japanese Yohkoh satellite, launched in 1991. Because patches of corona with high magnetic fields are filled with hotter plasma than the surrounding areas, they emit more x-rays, an effect that highlights both the strength and orientation of the field. Canfield and his collaborators noticed that S-shaped fields, whose kinked structure suggests that they are wound up and ready to explode, were two or three times more likely to lead to CMEs than hot spots without any obvious structure—although, as Moore says, “it was not guaranteed.” Moore and his colleagues noticed a similar effect when they used a magnetograph that could measure both the strength and three-dimensional orientation of the sun's field.

To improve the data, NOAA is planning to put an x-ray imager on a future weather satellite. “Currently, we get a few x-ray images a day from the Yohkoh,” says Zwickl. “The new x-ray instruments will deliver images every few minutes, 24 hours per day. For the first time, we will have continuous coverage of the sun's corona in x-rays, which will allow forecasters to see exactly the location of a flare and the start of any activity near active regions, as well as changes in the corona.”

Once a CME is emitted, a pair of new satellites known as Stereo should be able to track it to Earth. One will lead Earth in its orbit around the sun and the other will follow. The twin spacecraft will observe the Earth-sun environment stereoscopically, using identical instruments. “Ideally, we will see clouds coming toward Earth, measure their velocity, and get all the information we need” to predict their effects at Earth, says Boston University space physicist George Siscoe.

Knowing that a CME is on its way and how big it is, forecasters' next step is to input that information into computer models of the magnetosphere and the ionosphere and generate specific predictions of the effects on and near Earth. The task, however, is extraordinarily complicated because the magnetosphere and ionosphere themselves are. “There is a tremendous variability in properties, ranging from the ionosphere, where it's very cold and dense with strong magnetic fields, to the outer region of the magnetosphere where it meets the onrushing solar wind. We have this huge volume of space and these vastly different properties of gases and magnetic fields that permeate the different regions,” says Dan Baker, an atmospheric and space scientist at the University of Colorado, Boulder.

Some models are designed to skirt these complexities by simply predicting the impact and consequences of a solar storm based on what happened under similar conditions in the past. Others, however, model space weather from first principles, portraying the solar wind and the magnetosphere as interacting magnetized fluids. Both approaches have drawbacks: For some storms, there may be no good precedents to base predictions on, while the complex physics of space weather hampers efforts to simulate it from first principles. The goal of the recently formed Community Coordinated Modeling Center is to take the assorted models now being developed and try to combine them into a single model with the predictive power that models of terrestrial weather now provide. Michael Hesse, who directs the center, believes it can be done, but says actual numerical and localized predictions are still “a long, long, long way away.”

Even when such a model is up and running, Baker adds, there will be a serious catch to its predictive power. As with terrestrial weather, the hardest things to forecast will always be the rarest, most extreme events. These are the ones that do the most damage—the space weather equivalent, for instance, of the massive flooding that resulted from Hurricane Floyd last October. “We're on a good path in space science toward the same kind of predictive capability as the meteorologists,” Baker says. “But extreme events will probably still elude us to a significant degree, because they're so exceptional we either don't have any previous analogous events to base them on, or the system will get driven into some nonlinear state where we don't understand how it's behaving at all.”

12. BIOCHEMISTRY

DNA Cuts Its Teeth--As an Enzyme

1. Elizabeth Finkel*
1. Elizabeth Finkel is a writer in Melbourne, Australia.

DNA enzymes can inactivate genes in laboratory tests by binding to and cleaving their RNA messages. These molecules may be moving toward the clinic

Each year some 1 million people worldwide with blocked coronary arteries opt for a procedure called balloon angioplasty. A surgeon inserts a catheter into the clogged artery and inflates a balloon that smears the atherosclerotic plaque against the vessel wall like peanut butter. It's simple and often effective. But 30% to 40% of these procedures fail because the cells of the artery wall react badly to the trauma the procedure inflicts: They furiously repair the damage, often totally clogging the artery in the process.

Molecular biologists have tried to stifle the key genes involved in this overzealous repair process by inactivating or destroying the RNA messages they produce. They have had varying degrees of success in animal models. Now, a team of researchers led by Levon Khachigian at the University of New South Wales in Sydney, Australia, has attacked the problem with a promising new gene-inactivating technology: an enzyme made from DNA.

The work, reported in the November issue of Nature Medicine, was an early test of the therapeutic potential of DNA enzymes, which until now have been mainly laboratory curiosities. Compared to other molecules that can inactivate genes, DNA enzymes are more stable and cheaper to make, and they may be more discriminating in choosing their targets, which could reduce side effects. And in Khachigian's test, they appeared to be effective. His DNA enzyme, designed to bind to and cleave the RNA made by a damage-sensing gene called Egr-1, seemed to keep ballooned arteries from closing up in rat models of heart disease. “This is an exciting new use of a powerful technology,” says pathologist Tucker Collins of Harvard Medical School in Boston.

DNA enzymes are the able apprentices of RNA enzymes—RNA molecules that can catalyze chemical reactions, such as cleaving other RNA chains. Ribozymes, as they are called, were discovered in the early 1980s in living organisms; scientists soon made new variants in the laboratory, with different abilities, and then began to wonder whether DNA, RNA's chemical cousin, could follow the act.

No DNA enzymes are known in nature, but in 1994 Gerry Joyce of The Scripps Research Institute in La Jolla, California, and his then-postdoc, Ron Breaker, decided to try to create one using a laboratory version of natural selection. They hoped to produce a DNA molecule that could match the most elementary catalytic ability found in ribozymes: snipping the RNA phosphoester backbone. The starting ingredients for their primordial soup were trillions of random variants of a 50-base-long DNA chain, tethered to a matrix by a sequence that included an RNA nucleotide. When they washed the matrix with a lead solution, molecules capable of using the metal ion to help snip the tethering RNA nucleotide “selected” themselves into the column washings. The researchers then repeated the process several times, starting each round of selection with the population of molecules that had emerged from the previous round.

After 3 days, Breaker had evolved an efficient RNA-cleaving DNA enzyme. And later work in Joyce's lab by graduate student Steve Santoro produced the prize specimen, known as 10–23, for the 10th generation and 23rd clone. 10–23 had a catalytic efficiency higher than that of any known ribozyme. “It's the best we've ever seen; it will target any RNA you want,” boasts Joyce.

To target a specific RNA molecule, 10–23's RNA-cleaving catalytic domain can be fitted with stretches of DNA that bind to target sequences in the RNA. When the 10–23 catalytic domain is within range, it cleaves the RNA by speeding up the background reaction that makes RNA inherently unstable: the attack by an oxygen in one of RNA's ribose sugars on one of the bonds in the backbone. The so-called hammerhead ribozyme, originally found in plant viruses, works the same way, but 10–23 is less choosy about where it cleaves the sequence. Whereas the hammerhead prefers to bite where it finds a sequence of the nucleotides guanine, uracil, and cytosine—a target that may not be accessible on a tangled RNA molecule—10–23 can sink its teeth into any junction between the two kinds of nucleotides, purine and pyrimidine. Such a sequence is found at the start site of RNA messages, which usually dangles freely. Says Breaker, who is now at Yale University, “That's the beauty of 10–23; a target sequence is always accessible.”

Because 10–23 is made of DNA, it has some other powerful advantages over ribozymes. DNA lacks the inherent instability of RNA. “We have seen our DNA enzymes last at least 48 hours in serum,” compared to minutes for ribozymes, says Lun-Quan Sun, a biochemist at Johnson and Johnson Research in Sydney, which owns the patent for commercial applications of the 10–23 DNA enzyme. DNA is also easier and cheaper to make than RNA. And because it takes a more precise match for a DNA sequence to bind to RNA than for RNA to bind to another RNA molecule, DNA enzymes may be better than ribozymes at selecting their targets and ignoring similar sequences.

One place DNA enzymes could not supplant ribozymes, however, is in gene therapy. A gene for a therapeutic ribozyme could be inserted into the body, which would then make the ribozyme continuously; a DNA enzyme, in contrast, has to be synthesized and given as a drug.

Santoro and Joyce published the 10–23 sequence in the April 1997 Proceedings of the National Academy of Sciences, putting RNA-cleaving DNA enzymes in the hands of any researchers who wanted to copy the catalytic sequence. Explains Breaker, “These would be the simplest enzymes anyone can make. Just send an e-mail to a company that makes DNA, and you can have it the next day.”

In the last few months, researchers have reported promising test tube results with customized variants of the 10–23 DNA enzyme. Sun and colleagues, for example, designed a DNA enzyme to target the growth-stimulating gene myc, which effectively froze the growth of smooth muscle cells in a culture dish. And Kazunari Taira's group at the University of Tokyo and the National Institute of Advanced Interdisciplinary Research at Tsukuba Science City in Japan deployed a DNA enzyme against an abnormal gene message made in certain leukemic cells. The abnormal protein keeps the leukemic cells from self-destructing when normal cells do, leading to uncontrolled growth. The DNA enzyme destroyed the abnormal messenger RNA and triggered the self-destruct circuitry in the leukemic cells.

Khachigian's work with DNA enzymes is a first step toward the clinic. His group altered 10–23 so it would target the RNA from Egr-1, a wound-repair gene identified by Khachigian in Collins's lab at Harvard in 1996. Egr-1 acts like a chief of disaster operations. Undetectable in healthy arteries, its protein appears on the scene within minutes of injury, recruits a crew of tissue repair factors, and disappears a few hours later. Because Egr-1 acts early in wound repair, Khachigian thought it might be a strategic target. “It's a case of shooting the messenger,” he says.

Six hours before ballooning and during the procedure, the researchers applied the DNA to the outer surface of a rat's carotid artery, relying on chemical carriers to ferry it into the smooth muscle cells in the vessel wall. After 2 weeks, the new layer of cells lining the artery was twice as thick in untreated rats as it was in rats that had been treated with the DNA enzyme. Next, Khachigian plans to try this approach in pigs and then, if all goes well, in human patients.

But DNA enzymes modeled on the RNA-cleaving 10–23 molecule are just the beginning. In laboratory evolution experiments like the one that spawned 10–23, catalytic DNA molecules sporting bizarre new structures have emerged, some with four strands rather than the usual one or two, others co-opting amino acids—the building blocks of proteins—for extra catalytic power. Over the last year, these modified DNA variants have proved able to catalyze a whole new array of reactions in the test tube—for example, cutting, rejoining, and chemically modifying DNA strands. And Breaker thinks new DNA enzymes as potentially powerful as 10–23 could target proteins as well as RNA and DNA. “We must now consider the possibility that other enzymes like 10–23 are out there, with even greater catalytic power, just waiting to be discovered.”

Says Breaker, “My goal is to develop the therapeutic warheads of the future.”

13. MATERIALS RESEARCH SOCIETY MEETING

Building the Small World of the Future

1. Robert F. Service

Boston, Massachusetts—Nearly 4400 researchers gathered here from 29 November to 3 December to speed the way to future materials for everything from electronics to medicine. Highlights included new schemes for wiring molecular electronic devices and genetically engineered proteins that assemble materials.

Wiring Up the Nanoworld

In computer technology, if not in space flight, smaller almost always means faster, cheaper, and better. But researchers at the frontiers of miniaturization, who are fashioning experimental switches and storage devices from single molecules, have outrun their ability to wire these devices together into working systems. At the MRS meeting, scientists described several schemes—including one that uses DNA and another involving electric fields—that could help them link molecular circuitry with nanoscale wires. Although no working molecular devices have been rigged up yet, experts say the approaches could help make dreams of molecular electronics a reality.

“This was really nice work,” says Zhenan Bao, an advanced electronics specialist at Lucent Technologies' Bell Laboratories in Murray Hill, New Jersey. “It's the future direction the field [of molecular electronics] will take.”

On today's computer chips, the smallest features are 250 billionths of a meter, or nanometers, across. Although that's vanishingly small, the devices can still be made and wired up with photolithography, the industry's workhorse patterning technology, which shines light through stencils to direct the etching of fine features on silicon chips. Molecular-scale devices, however, can measure just a few nanometers in some dimensions, and light simply can't be focused tightly enough to lay out the patterns for the fine wiring needed to connect them. Researchers have to take another tack, such as first making the minuscule wires, and then positioning and connecting them. But although they can already make sufficiently small wires—one technique condenses metal atoms in the pores of membranes to form tiny metal rods—achieving the connections is another matter.

Working with nanorods that have platinum shafts and gold tips, a team of researchers at Pennsylvania State University in University Park led by electrical engineers Theresa Mayer and Thomas Jackson, along with chemists Thomas Mallouk and Michael Natan, has now come up with two speedy ways to solve this problem. One assembly strategy exploits the ability of DNA strands to seek out and bind to strands with matching sequences. The researchers first used small organic molecules called thiols to link single-stranded DNA to the rods' gold tips. Next, they decorated a separate gold surface with single-stranded DNA whose sequences were complementary to those attached to the rods. When the team mixed the rods with a solution containing the DNA-coated gold, the complementary strands bound to one another, linking the rods to the gold surface.

In a second strategy for assembling nanorods, the Penn State team turned to electrical attraction. Here they were linking a pair of electrodes resembling combs, positioned so their teeth interlaced. The researchers insulated these electrodes with a layer of silicon dioxide and placed a tiny gold pad halfway down each tooth, so that the pads on the adjacent teeth lined up in a row. They then immersed the apparatus in a solution containing their tiny metal rods and applied a voltage between the electrodes.

The voltage created an electric field that triggered two very different electrical effects to first attract and then bind the metal nanorods between adjacent gold pads. Initially, the field created a long-range electrical attraction that reeled in the nanorods from afar. The rods then bridged adjacent gold pads, as the full electrode assembly tried to maximize its capacitance, or its ability to store charge. Capacitors store charge as an electric field between two separated conductors harboring opposite charges. In this case, the overall electrode assembly contained two types of capacitors: The first consisted of the large electrodes and the gold pads separated by the insulating silicon dioxide, and the second of adjacent gold pads, separated by the solution.

The small size of the gold pads limited the capacitance of the overall device, which can only store as much charge as its smallest capacitors. But nanowiring lifted this restriction. By linking the gold pads, the nanowires opened a continuous electrical connection between the pads. That disrupted their ability to store charge, allowing the larger capacitors to take over. The bottom line, says Mayer, is that “we can use the field to drive the wires where we want them to be.”

That's not all. When the rods were shorter than the gap between the gold pads, the system's tendency to maximize capacitance caused several rods to line up end to end to form a bridge. The researchers then “welded” these rods into continuous wires by enriching the solution with gold ions. The ions filled in the gaps, producing longer rods that could conduct current, as the researchers confirmed.

“It's still somewhat unclear” how these techniques could be adapted for building molecular computers, says Mayer. “These are just baby steps,” adds Jackson. But it's the first steps that are the hardest to take.

Protein Patterns for Electronic Devices?

Biological organisms aren't just master builders of soft and squishy organic materials. They also do a pretty decent job of assembling rocklike inorganics—witness the strength of the everyday clam shell or your own bones and teeth. Of course, the secret behind such synthetic feats is the soft, squishy proteins that both clams and people have evolved to help organize inorganic molecules into intricate and useful patterns. Now materials researchers are trying to use the same trick as well.

At the Boston meeting two independent research teams, one led by materials scientist Mehmet Sarikaya of the University of Washington, Seattle, and the other by Angela Belcher, a chemist at the University of Texas, Austin, reported that they had used a laboratory version of evolution to create genetically engineered proteins that could bind to tiny semiconductor and metal particles and assemble them into larger clusters. If the same protein-engineering techniques can generate molecules capable of organizing and patterning a wide variety of materials, proteins could become invaluable tools for crafting transistors, wires, and other electronic devices with components hundreds of times smaller than those on current computer chips.

“The whole idea of merging biology with materials synthesis is very important,” says Chad Mirkin, a materials researcher at Northwestern University in Evanston, Illinois. “Organic systems have molecular recognition abilities that have had a long time to evolve. [They] far surpass what we can do readily in the lab.”

Both Sarikaya and Belcher hoped to exploit the abilities that the proteins critical to forming bones, shell, and teeth display: They have the selectivity to bind only certain inorganics, seeding and organizing their growth into desired patterns. Abalone, for example, use separate proteins to organize calcium carbonate into different mineral phases: iridescent mother-of-pearl, or aragonite, for the shell's inner layers and rock-hard calcite for the shell's outer surface. Such naturally occurring proteins don't work well with many industrially important materials such as metals and semiconductors, however. So the Washington and Texas researchers decided to see if they could improve matters.

For their part, Sarikaya and his Washington colleagues set out to coax bacterial proteins into binding to gold, which is used widely in the electronics industry. They started with multiporin, a cell membrane protein from the bacterium Escherichia colithat does not bind gold in its natural form. They then cloned the multiporin gene to make millions of copies. From each copy, they snipped out a section coding for a segment of the protein that forms a loop projecting from the E. coli membrane. That's where the protein would bind gold if its chemical makeup allowed it to do so.

To alter the makeup of the loop, the researchers replaced the snipped-out gene segment with random DNA sequences produced by an automated DNA synthesizer. They then introduced the mutated genes back into bacteria, grew the bacteria, exposed them to gold particles, and—in a set of steps analogous to natural selection—they washed off poor binders and regrew the better ones, eventually identifying the colony that did the best job of binding gold. Finally, they purified multiporin from these bacteria and attached the protein to the outer surface of both tiny plastic spheres and flat surfaces in solution. When they then spiked their mixture with a small amount of gold, the protein picked up the flecks, decorating either the outside of the spheres or dotting the surface.

Belcher's group, meanwhile, took a different approach to evolving proteins that could bind to semiconductors, such as zinc selenide and gallium arsenide, which are also widely used in electronics. The team used an off-the-shelf kit containing 109 random DNA sequences, which they inserted into copies of a gene that codes for the outer coat of a bacterial virus called a phage. They then infected bacteria with the modified phages, allowed the phages to multiply, and exposed the viruses to a solution containing semiconductor particles to select the viruses best able to bind the semiconductor.

Thus far, Belcher reported, the technique has worked beautifully. Her team has identified proteins that can discriminate between similar semiconductor alloys, such as gallium-arsenide versus aluminum-gallium-arsenide, and can even discriminate between different faces of the same semiconductor crystal, which have different arrangements of the atoms on the crystal surface. Down the road, she says, her team is planning to pattern the semiconductor-binding proteins on surfaces and use them to nucleate the growth of tiny semiconductor crystals in controlled arrangements. That's just what researchers around the globe are trying to do, in an effort to create ultrasmall transistors and other computing devices. And if Belcher and Sarikaya have their way, proteins may be just the handle they need to get there.

14. TAIWAN

Science Staggers Along After Deadly Earthquake

1. Dennis Normile

Chung Hsing University sustained heavy damage in the 21 September earthquake that left 100,000 homeless and caused massive disruptions to Taiwan's economy

Taichung, TaiwanWorking in the Food Science Building at Taiwan's National Chung Hsing University here is not for the faint of heart. The cracks and spalls in the building's concrete beams and columns are a constant reminder of the devastating earthquake that struck on 21 September, claiming 2300 lives and leaving more than 100,000 homeless. The toll included two students at Chung Hsing, one of Taiwan's major research universities. But 3 months later science goes on, as researchers and grad students step around scaffolds and reach over braces as they struggle to make up for the deadly interruption. “I don't feel safe,” says grad student Lai Li-An. But with her master's thesis due this spring, she says, “we have to keep working.”

Chung Hsing has so far received from the government only about one-fourth of the estimated $12.5 million needed for emergency repairs, and its request for another$34 million to replace the most heavily damaged buildings is pending. And it is far from the only supplicant. The quake wreaked havoc on educational facilities of all types, and drove students and faculty at National Chi Nan University, a small, relatively new university in Nantou, into borrowed classrooms in Taipei. The Ministry of Education estimates that it will cost $800 million to replace and repair more than 800 school buildings throughout the damaged area. The toll on university equipment and instruments has not even been calculated. Some things may be impossible to replace, such as biological culture collections. Yet it could have been worse. The campus, 40 kilometers from the epicenter, was spared the full brunt of the magnitude 7.6 earthquake that struck near the central mountain town of Puli, 160 kilometers south of Taipei. More than 50,000 structures in the region were destroyed, and downed power lines led to outages and rationing that lasted for several weeks. Total economic losses have been estimated at$9 billion, or about 3.3% of gross domestic product.

At Chung Hsing, the 30-year-old Food Science Building and the library, now closed for repairs, sustained the most damage. But most buildings “performed as expected,” says Lin Chi-Chang, chair of the Department of Civil Engineering, despite the fact that the lateral seismic loads imposed on the buildings were much greater than they were designed to withstand. The cracks that run through interior walls in many buildings, he adds, do not threaten their structural integrity.

Equipment is another matter, however. The quake destroyed the magnet in the chemistry department's prized 400-megahertz nuclear magnetic resonance (NMR) spectrometer, and the department doesn't have the $100,000 it would cost to buy a replacement. Instead, researchers are using older NMRs, which are less powerful and therefore slower. “People are having to wait for machine time,” says department chair Gau Han-Mou. A key component in a high-performance liquid chromatograph was also knocked out of kilter in the quake and sent off for repairs. “We keep calling, but they haven't yet figured out what's wrong,” says Gau, who was using it to study the chirality, or handedness, of molecules. That work has been suspended until the instrument can be fixed. The Institute of Molecular Biology's entire collection of culture samples was lost when a weeklong power outage shut down freezers. “We had emergency generators, but they were thrown off their supports in the earthquake,” says Tseng Yi-Hsiung, an institute professor. The lost samples represent years of efforts to identify and isolate agriculturally important strains of bacteria and enzymes. “I can't even estimate the loss,” Tseng says. Fortunately, his own work on a bacterium that damages vegetable crops has continued thanks to samples from other labs. Officials have also tried to ease the pain of those who suffered the greatest losses. The parents of one of the fallen students, a popular senior in the horticulture department, added their own funds to a collection taken up for their daughter and gave it back to the school for repairs. But university officials instead have combined it with other contributions and created a scholarship fund for students from families that lost a breadwinner in the earthquake. Some of those survivors say it has been hard to return to work. “The experts are more optimistic than we are,” says Tseng. Nights are particularly bad. “People used to work here until very late,” Tseng says. “Now we're all afraid to be in the building after dark.” For a long time, however, going home wasn't much comfort, either. Tsen Hau-Yang, a professor of food science, says that he and his family worried for weeks about the safety of their two-story house. “So I slept in my car for the first month after the quake,” he says. Although Tsen is back to sleeping in his bed, many faculty members expect repercussions from the quake for years. Tseng is particularly worried that the uncertain state of repairs will drive away top-quality applicants to the university and its graduate schools. “That could be the most serious problem of all,” he says. 15. RESEARCH FACILITIES UCSF on the Move to New Mission Bay Campus 1. Marcia Barinaga The faculty look to bright new labs at Mission Bay but worry that they will lose the cohesiveness that has made UCSF so successful The University of California, San Francisco (UCSF), is legendary for turning its liabilities into strengths. In 1976, USCF promised its neighbors that it would cap the growth of its campus in Parnassus Heights, near Golden Gate Park, west of downtown San Francisco. But the number of new faculty members didn't stop growing. As a result, researchers are packed tightly into existing buildings, their labs sometimes cobbled together from former bathrooms and closets. One might think faculty members would view this as a big disadvantage, but on campus the rallying cry is that the crowded conditions serve as a crucible for interaction and creative collaboration, which in turn have helped UCSF maintain its position near the pinnacle of medical research universities. Soon, however, the UCSF faculty will have to find a new rallying cry. On 25 October, UCSF broke ground for a new campus at Mission Bay, a 20-minute drive across town from Parnassus Heights. There, on 17.4 hectares of old railroad yards on the city's eastern waterfront with a view of the downtown skyline, UCSF plans to build 12 research buildings, a recreation and community center, student housing, and parking garages—a complete second campus that will double UCSF's research space and house 500 labs when it is completed in 10 to 20 years. “We are the first major medical school that I know of to undergo binary fission during adulthood,” quips UCSF virologist Donald Ganem. Robert McGhee, architect for the Howard Hughes Medical Institute, who has advised UCSF on the project, agrees. The building of a geographically separate campus on this scale, he notes, is unprecedented for a major American medical university. Despite all the talk about the benefits of cozy quarters, no one at UCSF disputes the need for expansion. The lack of space at Parnassus has prevented UCSF from riding some new waves in research, such as genomics and brain imaging, and the new campus will provide unprecedented opportunities for restructuring and adding programs. But the project is also fraught with hazards. Can the university raise the necessary funds—more than$1 billion over 10 to 15 years? And how can a research community as tightly interwoven as that at UCSF be split? “We're all nervous,” says one faculty member who requested anonymity. “People ask whether UCSF is going to lose that thing that is so special about it.”

Fissioning the university without tearing things apart will be a big challenge for UCSF's leaders. Nervous researchers are taking some comfort from the fact that the top duo are both longtime members of the UCSF research community, who understand what makes it work so well. The chancellor, oncogene researcher and Nobel laureate J. Michael Bishop, has been on the faculty since 1968, and neuroscientist Zach Hall, Bishop's vice chancellor for research, has been at UCSF since 1976. “The good thing is the leaders are people we have confidence in,” says Ganem.

Bishop and Hall are proceeding in characteristic UCSF style, encouraging faculty involvement in shaping the plans and programs for both new and old campuses. That input began in 1996 when medical school dean Haile Debas asked biologist Keith Yamamoto to chair a 16-member faculty committee to come up with a plan for how to populate the first research building at Mission Bay. Yamamoto says the committee strongly felt that no one should move to Mission Bay until a critical mass for a research community could be achieved, and the panel quickly realized that the one building originally proposed for phase I of the new campus wouldn't do it.

So Yamamoto reconvened an expanded committee of 44 faculty, which planned a much larger first phase that includes two research buildings, with a total of 30,150 square meters of space, a community center with a gym and pool, student services and meeting rooms, and 500 units of housing for students and postdocs, at a projected cost of $400 million. The project is expected to be completed by early 2003, and then “within a 6-month to a year period, we will move down 65 to 70 principal investigators [and their labs],” says Hall. The area ringing the campus will be a hotbed of new construction in coming years as well. The California-based Catellus Development Corp., which donated the Mission Bay site to UCSF, owns an additional 105.2 hectares surrounding the new campus on which it plans to build sites for biotech and pharmaceutical companies, as well as housing, shopping areas, a hotel, and a waterfront park. Catellus has just begun marketing the first wave of planned biotech space, and CEO Nelson Rising says there is considerable interest but no deals signed yet. Bishop expects that having biotech next door will “further enliven the intellectual climate” of the new campus and provide new potential for industry-academic collaborations. Indeed, the prospect of a whole new campus with a new set of potential interactions affords “a tremendous opportunity to configure a research facility from the ground up,” says Yamamoto. So, rather than just recommending “picking up Parnassus and putting it down at Mission Bay,” his committee created a general plan in which basic research will be concentrated at the new campus while Parnassus Heights will emphasize disease-related and clinical research. Under that plan, Mission Bay will be the seat of the Program in Biological Sciences, the more basic research-oriented of the school's two graduate programs. It will include areas such as cell and molecular biology and structural biology that are already well represented at UCSF, and some new programs, such as human genetics. There are also plans for a center for biological systems analysis, which would be peopled by a type of faculty member rare at medical schools—physicists and mathematicians—whose research focuses on the analysis of complex biological systems such as the numerous and interlocking signaling pathways that together control the cell division cycle. “A lot of people at UCSF are hungry for this sort of thing,” says UCSF cell biologist David Morgan, who will co-direct the new center. But those new opportunities come with a trade-off: Researchers who move to Mission Bay will be distanced from the clinical research at Parnassus Heights. That can be a drawback, notes cholesterol researcher and Nobel laureate Joseph Goldstein of the University of Texas (UT) Southwestern Medical Center in Dallas, because these days basic researchers can never tell when their work may take an unexpected turn toward clinical applicability. Yamamoto agrees that “this would be about the worst time imaginable for UCSF to make a decision to separate its basic scientists from its clinical scientists.” For that reason, he says, the long-term plan is to have some clinical presence at both campuses. Two independent disease-related institutes associated with UCSF, the Ernest Gallo Clinic and Research Center, which focuses on the biology of alcoholism and other addictions, and the Gladstone Institutes, which conduct research in Alzheimer's disease, AIDS, and cardiovascular disease, will be moving to their own buildings at Mission Bay in the first wave, bringing some human-disease emphasis. The second phase will include a human brain-imaging center and some related clinical neuroscience research. And San Francisco General Hospital, a part of the UCSF system, is only 5 minutes from Mission Bay, so new alliances are likely to grow there. Still, the Parnassus Heights campus will remain the seat of most of UCSF's clinical research and home to the Biomedical Science program, the more human disease-oriented of the school's two graduate programs. But the old campus will not be exclusively devoted to clinical research. The faculty of the immunology department has voted to stay permanently at Parnassus Heights, for example, and researchers in virology, vertebrate development, and other areas with links to disease-related research are electing to stay as well. In addition, a committee headed by Ganem has made several recommendations to integrate disease-related basic research more closely with clinical research at Parnassus. Recommendations include launching new programs in areas such as infection and immunity that span clinical and basic research departments; forming tissue and DNA storage banks, coordinated with patient data collected at the clinical research facilities, to aid geneticists, cancer biologists, and others who want to look at factors that affect disease prevalence; and in general making clinical research centers available to basic researchers. Those ventures may be aided by the self-selection that is occurring as researchers decide whether to go or stay, Ganem says: “The people who are choosing to stay are the people who really, really want to have a close integration with patient-based researchers.” How soon the big plans for both campuses can be fully implemented depends on how quickly the necessary funds can be raised. Bishop says that although$190 million of the estimated $400 million needed to build and outfit the first phase of Mission Bay will come from sources such as state funds and borrowing by the campus, gifts must provide the remaining$210 million. UCSF is nearly halfway to that mark: In its first 6 months, the campaign raised $50 million, which was doubled when the biotech firm Genentech of South San Francisco donated$50 million toward the first building at Mission Bay as part of its settlement of UC's lawsuit over the patent on human growth hormone (Science, 26 November, p. 1655).

Despite this promising start, some observers have wondered whether a public university without a large undergraduate alumni base to draw on can pull off the kind of sustained fund raising necessary to complete such a massive project. But David Glen, associate vice president and director of principal gifts at Stanford University, gives UCSF a good chance of success. “Medical fund raising is generally much less alumni driven,” he says, than other types of university projects.

A big uncertainty is timing. “We are at the beginning of a huge project,” says Hall, “and what we are unable to judge is the pace.” If the fund raising is “terrifically successful,” he says, “we could begin planning the first building of phase II in a year or two, and have [the transition between phases] be seamless.” But he admits that is optimistic.

The potential for a funding-forced hiatus between phases weighs heavily on some faculty members. For example, the neuroscientists will see their community split by phase I, with those with close ties to cell and molecular biology going to Mission Bay while the majority remain behind until the first building of phase II is completed. “I don't think any of the neuroscientists want the group to be split even for a year,” says UCSF neurophysiologist Steve Lisberger.

Even when Mission Bay is complete, UCSF will be a community divided by a 20- to 30-minute drive through city traffic. “No matter what we say about having fast shuttle buses,” says cell biologist Morgan, “that distance is going to really cut us apart.” Yamamoto suggests that separation pangs may be vital to success: “We want to maintain in everyone's mind the real drive to be communicating with their friends across town. We don't want to develop a two-institution mentality.”

To help reach that goal, he says, “both campuses are being wired to the teeth” in preparation for telecommunications equipment that in some cases isn't even developed yet. “I want to be sure that students at Parnassus can ask faculty members at Mission Bay to be on their thesis committee without a second thought, knowing that if they can't slip across town for a half-hour meeting, they can call them up on video conference.”

But will such assurances be enough to maintain UCSF's appeal to top faculty and students? UT Southwestern's Goldstein is unconcerned. “Their draw will outweigh the negatives,” he predicts, adding that the excitement of being part of such a large endeavor is sure to add to the attraction for some. Ganem's response to the challenge supports Goldstein's prediction. “There are no models for how to do this,” he says, “But that to me is the coolest part.”