# News this Week

Science  23 May 1997:
Vol. 276, Issue 5316, pp. 1192
1. Biotechnology

# Antisense Aims for a Renaissance

ANTISENSE THERAPIES, ONCE PLAGUED BY SIDE EFFECTS AND MYSTERIOUS FAILURES, HAVE NOW SHED THEIR TROUBLED IMAGE, THANKS TO ENCOURAGING NEW CLINICAL RESULTS AND BASIC RESEARCH ON PAST PROBLEM

CAMBRIDGE, MASSACHUSETTS—The upscale stretch of the Charles River between Harvard University and the Massachusetts Institute of Technology could be renamed Biotech Brook. First, there is Genzyme's new $150 million facility, pointing skyward like a brick version of Chartres Cathedral. Downstream, the red-brick behemoth of Millennium Pharmaceuticals looms arrogantly over the river in a former Ford Motor Co. factory. But, for years, the next building, across the railroad tracks and down the river, was the neer-do-well of the neighborhood, a drab tire warehouse-turned-office-building with little architectural—or scientific—presence. Then last year, the building was renovated in a snazzy postmodern style, complete with penthouse and checkerboard-brick ornamentation. This architectural rebirth mirrors a scientific one, for the new occupant is Hybridon Inc., a major manufacturer of antisense oligonucleotides—short, synthetic stretches of DNA and RNA designed to block the action of specific genes by binding to their RNA transcripts. Back in 1992—after studies suggested antisense might be a promising source of precisely targeted therapies for cancer and other diseases—Science named antisense technology one of the 10 hottest research areas of the year. But its prospects have been in doubt ever since. By 1995, researchers faced depressing evidence that their antisense molecules just didn't seem to work as intended (Science, 27 October 1995, p. 575). In addition, the technology was experiencing menacing glitches, including unforeseen—and sometimes lethal—side effects in animals. Now, however, scientists are discovering that some of the problems stemmed from peculiarities of the first-generation antisense drugs, some of which bound to all sorts of molecules in addition to their targets, and second-generation antisense molecules are on the way. The result is a biotech boomlet, symbolized by Hybridon's new headquarters and the investor dollars flowing into the field; Hybridon netted$52 million in its initial public offering last year. Also, early results from new clinical trials of first-generation drugs seem promising. Just this week, at the American Society of Clinical Oncology meeting in Denver, researchers reported that antisense drugs shrank ovarian tumors in small clinical trials. Early trials of drugs against AIDS and Crohn's disease, an inflammation of the bowel, also look favorable. “It seems that many of the underlying problems that have been endemic to the field are on the way to being worked out,” says analyst Michael Sheffery of the New York biotech investment firm Mehta and Isaly. Biochemist Fritz Eckstein of the Max Planck Institute for Experimental Medicine in Göttingen, Germany, agrees, noting that “antisense has come of age.”

Indeed, a recent Cambridge antisense conference* turned into an “antisense lovefest,” according to Mark Matteuci, director of bioorganic chemistry at Gilead Sciences of Foster City, California, one of a handful of companies heavily invested in antisense. Even so, Matteuci and some others are still cautious about the technology. Despite the new mood of optimism, a number of serious wrinkles—such as the difficulty of getting enough antisense molecules into cell nuclei, and continuing doubts about whether they exert their effects through true antisense mechanisms—have yet to be ironed out, warns Matteuci.

## Making sense of antisense

The idea behind antisense is simplicity itself. In the first step of protein synthesis, the bases on one strand of DNA's double helix are transcribed into the complementary sequence in a messenger RNA (mRNA) molecule. As Harvard biochemist Paul Zamecnik and others first showed in the 1950s, the sequences of these “sense” molecules are translated into a series of amino acids, giving rise to proteins. Then, in 1978, Zamecnik showed that an oligonucleotide tailored to complement part of a particular mRNA—an “antisense” strand—could bind to that section of the mRNA, causing a cutting enzyme to home in on the mRNA and destroy it. This prevents protein expression and turns off a gene's activity. Zamecnik quickly saw that antisense drugs might forestall a host of diseases linked to abnormal gene activity, including cancers and viral infections, and he founded Hybridon in 1989 to exploit the idea.

Researchers had some early test-tube successes, for example, showing in cell cultures that antisense molecules could reduce replication in HIV, the virus that causes AIDS. But putting the antisense strategy into practice proved difficult. The first generation of drugs—the so-called phosphorothioates, whose nucleotide backbones carry extra sulfur atoms to slow their degradation in cells—behaved unexpectedly. Some compounds were apparently unable to get into their target cells, and control oligos that didn't even complement the target mRNA seemed to work almost as well as the antisense oligos, throwing a monkey wrench into the logic of antisense drug design. Such “non-sequence-specific” effects plagued Hybridon's antisense oligo called GEM91, designed to block expression of a crucial HIV gene, and many others. The antisense technique, so precise in theory, began to look like a biological blunderbuss in practice.

There were other worrisome signs as well. Researchers found that antisense seemed to trigger extreme—and sometimes dangerous—immune responses. A group led by University of Iowa immunologist Arthur Krieg, for example, reported in 1995 that antisense oligos carrying a certain brief sequence (a cytosine nucleotide followed by a guanine) somehow pump up the body's production of immune-system warrior cells—a vexing result, for Krieg had been trying to treat diseases such as lupus that involve unbridled immune activity. In addition, several monkeys at Hybridon died after injections of phosphorothioates.

But explanations—and solutions—have now emerged for many of these setbacks. For starters, researchers now realize that the added sulfur ions in the phosphorothioates may be responsible for some of the non-sequence-specific effects. These compounds carry a large negative electrical charge, which makes them chemically sticky, explains nucleic acid chemist Sudhir Agrawal, Hybridon's chief scientific officer. Hence, they are liable to hitch up willy-nilly not just with mRNAs, but with antibodies and intracellular messenger molecules, and so interfere with many biological processes.

As a consequence, antisense firms are developing a second generation of phosphorothioates that carry other clusters of atoms with less charge. Hybridon, for example, has found that antisense oligos bearing uncharged chemical groups have fewer unwanted side effects and survive longer inside cells, says Agrawal.

Another big stumbling block, researchers say, has been the naïve assumption that all mRNA sequences make equally attractive targets for antisense drugs. For some reason, antisense oligos bind to some sequences with much greater affinity than others, explains cancer researcher Stanley Crooke, founder, chair, and CEO of antisense leader Isis Pharmaceuticals, based in Carlsbad, California. “Randomly making antisense oligos gives random results, and when it works, it's luck,” says Crooke. Cy Stein, a Columbia University pharmacologist who has questioned whether the early positive results were due to true antisense mechanisms, agrees. “You can't just pull a reagent off the shelf—you have to be prepared to screen 30 or 40 molecules to find the ones that work.”

But synthesizing and testing large numbers of ultrapure oligos are expensive and time-consuming. Hybridon, which does commercial-scale manufacturing of antisense oligos, charges about $2000 per gram. Academic scientists can't afford such costs, but companies can. Isis, for example, has found success by sifting through dozens of oligos that complement slightly differing segments of the targeted mRNA, and selecting those that work. As a result, although academics are energetically applying antisense to research questions, both corporate and academic researchers agree that today many of the hottest antisense results come from companies rather than university labs. “Isis has brought sense to antisense,” says Iowa's Krieg. Hybridon researchers have also begun to make sense of their disturbing monkey fatalities. Giving monkeys a large, one-time injection of phosphorothioates triggered a systemic and lethal inflammation by activating complement, the part of the immune system responsible for organ rejection after transplants, says Agrawal. He says that giving the same drugs more slowly avoids the problem. Even the problem of slipping antisense compounds into the cell nucleus, where they can bind to their target mRNAs, looks a bit less daunting than before. The previous round of experiments was largely done in culture, where the molecules often failed to make it past the cell membrane. So, some scientists spent years trying to craft lipid-based costumes to disguise the antisense molecules and enable them to slip through. But it appears—again for reasons not well understood—that antisense molecules do better at reaching their targets in living animals than in culture, says Crooke. Data from Hybridon and Isis show that plain, uncloaked antisense molecules injected into live animals are taken up by many different organs, although questions remain about exactly where the molecules bind once they are inside cells. ## Success in the clinic While putting these old questions to rest is key to antisense's future, “Nothing is going to make a bigger difference than positive clinical data,” says analyst Sheffery. In recent months, even the first generation of antisense oligos has begun to yield encouraging results. This week, Stanford University oncologist and pharmacologist Branimir Sikic reported at the oncology meeting that ISIS 3521, a 20-base oligonucleotide developed by Isis, stopped the spread of ovarian cancer in three patients out of 17 and caused few toxic side effects in Phase I clinical trials. A promising antisense oligo that blocks replication of cytomegalovirus—the virus that destroys the retinas of many AIDS patients—is already in Phase III trials at Isis. The company hopes to file a new drug application with the Food and Drug Administration by early 1998. The largest crop of encouraging data came in late February, when Isis announced results from a Phase II trial of the antisense drug ISIS 2302 as a treatment for Crohn's disease. The inflammation response that causes Crohn's depends in part on a cell adhesion protein that helps inflammatory blood cells punch through the walls of blood vessels; the new antisense compound inhibits the synthesis of this protein. After 1 month of treatment, the disease had gone into remission in nearly half of the 15 patients treated, compared to zero out of five patients taking a placebo. That outcome “was much stronger than I expected it to be,” says Sheffery. Some optimistic researchers say that if certain lab experiments blossom into clinical applications, more successes could be on the way. At the meeting, pharmacologist Ryszard Kole of the University of North Carolina, Chapel Hill, for example, reported a potential breakthrough in the treatment of thalassemia, a hereditary form of anemia common in many parts of the world. In some individuals, this disease is caused by mutations in the gene encoding the hemoglobin subunit globin. These mutations cause part of an intron in the gene—a nucleotide sequence that interrupts the gene and is not intended to code for protein—to be left in the mRNA by mistake. The resultant mutant mRNA is unstable and degrades before it can be translated into protein. But Kole found that infusing mutant cells in culture with certain antisense oligos caused the intron to be snipped out, as it should be, allowing normal globin to be made. Reassuringly, mismatched antisense molecules injected as controls had no effect. To the optimists, this and many other such experiments indicate that the new antisense molecules are living up to the original vision, homing in accurately on their target mRNAs. But others note that the compounds still exhibit many non-sequence-specific effects—Hybridon's GEM91, for example, not only blocks HIV replication but inhibits its binding to cells. That suggests that antisense molecules bind profligately to unintended targets and so could create a new crop of unforeseen problems. To these researchers, the atmosphere at the Cambridge conference was too utopian. “A couple of years ago, some of the more negative issues got blown out of proportion, and I think that this year things went in the opposite direction,” says cell biologist Richard Wagner, director of Gilead Sciences. Yet, even skeptics such as Wagner and Columbia's Stein—who co-organized a much more negative gathering 2 years ago—agree that antisense is moving up in the world and deserves its renewed status as an up-and-coming research technology. “It's clear that things are a lot better,” says Stein. “A lot of the theoretical problems have been resolved. … There is really a chance of seeing what people have wanted to see.” • * “Antisense 97: Targeting the Molecular Basis of Disease,” 1-2 May. 2. Astronomy # It's Official: Gamma Bursts Come From Far, Far Away 1. Govert Schilling Ebullient astronomers think they may have solved one of astronomy's most durable puzzles: where in the universe the mysterious flashes of energy called gamma-ray bursts come from. Occurring at a random position in the sky about once a day, these seconds- to minutes-long pulses of gamma rays originate either from local sources in or near our Milky Way galaxy or from vastly more powerful events in the far reaches of the universe. Now, by detecting the afterglow of a burst on 8 May—Ascension Day—in the northern constellation Camelopardalis (the Giraffe) and measuring its distance, a team of observers has delivered a verdict: Gamma-ray bursts come from far beyond our galaxy, at cosmological distances. “I think this is a victory,” says Princeton astronomer Bohdan Paczyński, a longtime proponent of the cosmological view. Adds John Heise of the Space Research Organization Netherlands (SRON), the project scientist for Beppo-SAX, the satellite that was crucial to the discovery: “We finally got the proof. It's fantastic!” Some advocates of the local view are still holding out, but many astronomers are now considering a new puzzle: what kind of distant cataclysms could explain the bursts, which would have to be the most energetic events in the cosmos. The brief gamma-ray flashes themselves have yielded little information about their sources, so astronomers have been hunting for counterparts at other wavelengths. That search quickened with the launch of Beppo-SAX, an Italian-Dutch project, just over a year ago. Gamma-ray detectors have poor angular resolution, making it hard to know where to look for a visible-light counterpart, but Beppo-SAX also carries two wide-field x-ray cameras, built at SRON's Utrecht laboratory, that operate in tandem with the gamma detector. The sharper view of the x-ray cameras enables astronomers to narrow down the location of a gamma-ray burst, giving colleagues with optical and radio telescopes a smaller patch of sky to search. Beppo-SAX scored its first success when a burst detected on 28 February guided a Dutch team led by Jan van Paradijs of the University of Amsterdam and the University of Alabama, Huntsville, to an optical counterpart (Science, 21 March, p. 1738). The rapidly fading optical source lay within a smear of light that looked like a very distant galaxy. However, a couple of weeks later, a group led by Patrizia Caraveo of the Istituto di Fisica Cosmica in Milan, Italy, reported that Hubble Space Telescope observations seemed to indicate that the optical source was moving across the sky. Motion would be imperceptible at a distance greater than a few hundred light-years, so, they argued, the source must be nearby (Science, 25 April, p. 529). Now, the “Ascension burst” tips the scales strongly back in favor of the cosmological explanation. Notified by an Internet alert from the Beppo-SAX team, Howard Bond of the Space Telescope Science Institute in Baltimore discovered a point of light at the burst position less than 7 hours after the event, through a modest 90-centimeter telescope at the Kitt Peak National Observatory in Arizona. Over the next several days, the point brightened, then faded. Then, on 11 May, Mark R. Metzger and colleagues at the California Institute of Technology used the 10-meter Keck II telescope on Mauna Kea, Hawaii, to capture a spectrum of what had become known as Bond's star. They found that the light contained absorption lines—in effect, shadows from material in front of the source. And the lines, from the elements iron and magnesium, were drastically shifted toward the red end of the spectrum by the expansion of the universe. This redshift, of 0.835, indicated that the absorbing material lies several billion light-years away, and the optical source is at least that far off. British theorist Martin Rees of Cambridge University notes that “the important question is whether the optical event is related to the gamma-ray burst.” But he thinks “this seems quite probable.” Paczyński has fewer reservations: “It is a very direct proof that at least this one burst was at a considerable redshift.” The same would then be true for all other bursts, he says, because it is “very unlikely” that there could be two separate classes of gamma-ray bursts. Says Paczyński's co-worker, Ralph Wijers of Cambridge University: “We can finally forget about the local hypothesis.” Don Lamb of the University of Chicago, a vocal proponent of the local view, isn't convinced, saying that he has “increasing doubts that [Bond's star] has anything to do with the gamma-ray burst.” Lamb points out that the variable object shows all the signs of being a so-called BL Lacertae object, a kind of turbulent, energetic galaxy. Because efforts to link earlier bursts to such galaxies had failed, Lamb thinks the burst and the high-redshift optical variable are unrelated. “Time will tell,” he says; “that's the wonderful thing in science.” Caraveo, whose team had reported that the burst counterpart was moving and therefore nearby, is also doubtful. He sees “no reason to withdraw our claim,” saying that his team's motion-finding technique “has been tested with a source of comparable brightness seen by the same instrument of the Hubble Space Telescope.” Van Paradijs, however, is convinced that the Italians “were wrong. Other groups, working with the same data, were unable to confirm the reported motion. In my opinion, there is no question anymore” about the validity of the cosmological model. If gamma-ray bursts are indeed at cosmological distances, as most astronomers now seem to believe, the question is what kind of astrophysical mechanism could spark these extremely energetic events. The varying brightness of the visible afterglow may help theorists narrow the field of possibilities. Another clue came on 13 May, when astronomers at the Very Large Array radio telescope in New Mexico spotted a flaring source of radio waves at the position of Bond's star. Rees says that the observations so far agree with a scenario he has proposed, in which a titanic event—probably two neutron stars colliding—generates a blast wave that emits first gamma rays, then the lower energy optical and radio waves, as it slams into the surrounding interstellar medium. “The observed light curves are quite easy to explain,” he says. Observers and theorists alike are sure to have plenty to discuss when they gather on the Italian island of Elba for a gamma-ray burst workshop at the end of this month. 3. Physics # Hopes for Exotic New Particle Fade 1. James Glanz Nothing is surer to rouse the field of high-energy physics than hints of a new particle. That helps explain the intense interest in a recent announcement by researchers at the Deutsches Elektronen-Synchrotron in Hamburg, Germany, who had data from particle collisions that might have signaled the brief appearance of an exotic new beast called a leptoquark (Science, 28 February, p. 1266). This week, however, the field is standing down a bit. A group at the Fermi National Accelerator Laboratory has made an extensive search through its data for evidence of leptoquarks resembling the ones hinted at by the DESY results—and come up empty. The negative search does not strictly rule out the possibility of someday finding these particles, which would combine properties of the two basic kinds of matter, says Carla Grosso-Pilcher of the University of Chicago, a member of the multinational Collider Detector at Fermilab (CDF) collaboration. But Grosso-Pilcher, who reported on the analysis last week during a workshop at Vanderbilt University in Nashville, Tennessee, says that if leptoquarks do exist, they must either have a mass greater than Fermilab can detect, or decay into elusive particles like neutrinos, which the search could have missed. “The CDF [analysis] leaves almost no room for the simplest leptoquark solution,” says Herbi Dreiner, a theorist at the Rutherford-Appleton Laboratory in the United Kingdom. The original announcement came after two detectors at DESY's HERA accelerator—which smashes antimatter particles called positrons into protons—had seen more “hard,” or violent, collisions than expected under physicists current theory of the fundamental structure of matter, called the Standard Model. One possible explanation was that the collisions were spawning a particle that combines the properties of quarks—the building blocks of the proton—and leptons, such as the positron and the electron. By briefly materializing, then decaying in a spray of ordinary particles, a leptoquark might explain the seemingly violent collisions. “The HERA results came out, and we really moved fast,” says Henry Frisch of the University of Chicago, co-convener of CDF's “exotics” group. The group analyzed data on the debris from 3 trillion collisions in Fermilab's Tevatron accelerator, which smashes together protons and antiprotons, their antimatter counterparts. These collisions could sometimes produce pairs of leptoquarks, as the proton's building blocks tangled with their counterparts in the antiproton. The leptoquark pairs would then decay into ordinary particles, leaving a recognizable signature. That signature wasn't seen, says Frisch. That still leaves the HERA group mulling over the anomalous collisions. “Well continue to compare our … data to Standard Model predictions,” says Bruce Straub of Columbia University, a HERA collaborator. The answer, when it finally comes, could be as mundane as minor tweaks in physicists understanding of proton structure—or as electrifying as another new particle. 4. Materials Science # Nonlinear Molecules Trip the Light 1. Robert F. Service As high-speed information carriers, photons of light are hard to beat. They travel at, well, the speed of light, and can be packed close together without interacting, allowing many streams of information to be transmitted together. No surprise, then, that many of today's long-distance messages—phone calls, faxes, e-mails—are converted from electronic signals to pulses of light and beamed over long-distance optical fibers. But like expressways that end at sluggish intersections, fiber-optic systems rely on low-speed electronic components for switching, routing, sending, and storing the information. Visionaries foresee speeding up these intersections by controlling the flow of light with other light. These all-optical switches are still mostly on the drawing boards, however, because few optical materials allow one light beam to manipulate another. Now, on page 1233, an international team of researchers led by chemist Seth Marder at the California Institute of Technology (Caltech) reports a development that could bring the all-optics vision a step closer to reality. Marder and his colleagues have developed polymers with an enhanced version of an effect seen in practically every optical material: Beam in sufficiently intense light, and it will change the way the light itself or another beam propagates. Usually, this effect—called a third-order nonlinear optical (NLO) effect—is vanishingly small. But the Caltech team found that by manipulating the electronic character of small molecules embedded in the polymers, they could elicit unrivaled third-order effects—in one case 35 times better than ever before. “It's outstanding,” says University of Pennsylvania optical physicist Anthony Garito, who helped develop some of the basic principles behind the new work. “This will have implications for a huge number of applications,” ranging from optical switches to data storage. Garito, Marder, and others caution, however, that the new materials themselves are not likely to be technologically useful, because they break down under even modest light and heat. But they think that the success will show the way toward developing more robust materials with equally strong NLO properties. An optical material qualifies as nonlinear if an electric field or light itself can change its optical properties. Marder explains that in any optical material, light's oscillating electric field causes electric charges to vibrate, generating a secondary field. If this field is strong enough—or is supplemented by an external field—it can feed back to influence the vibration. More photons can then interact with the material in pairs (a second-order effect) or even trios (a third-order effect), producing light frequencies that are harmonic overtones of the original, much like the overtones that give violins their rich sound. These overtones can alter the original light's color. They can also change the material's refractive index or transparency—effects that can be used to manipulate a light beam. Change a material's refractive index, for example, and it can act as an optical switch, steering a light beam from one fiber-optic line to another. In second-order NLO materials, this switch can be tripped only by applying a voltage. But in third-order materials, one light beam can switch another. The third-order effect is inherently much weaker than its second-order cousin, however, as it is harder to coordinate the interaction of trios of photons than pairs. In recent years, hopes for improving third-order materials have rested on optically active organic molecules known as chromophores. Chemists can easily tailor these molecules and incorporate them into transparent host polymers. In 1989, Garito and his colleagues proposed a strategy for developing chromophores that have strong third-order NLO properties: create molecules that can sustain large separations of electric charge. These researchers knew that many chromophores can essentially shuttle negative charges to one end of the molecule, leaving the other end positively charged, and that this separation increases in response to light. They calculated that the greater this light-driven charge separation, the greater the material's light-changing NLO properties would be. As Marder's collaborator George Stegeman, a physicist at the University of Central Florida in Orlando, explains, the charge separation alters the electronic structure of the molecule, making it easier for trios of photons to interact in the material. In 1993, Marder and his colleagues improved a class of small chromophores to more than double their NLO properties. In the current experiment, the team turned to longer molecules whose charges could separate even farther. The researchers started with a batch of −carotene, a well-known third-order NLO chromophore consisting of a long, narrow chain of carbon atoms capped on either end by ring-shaped groups. One of the rings is an electron-hungry “acceptor” group that snags an electron from the other end of the molecule, giving −carotene its NLO properties, among the strongest yet observed. To enhance these properties, Marder and his colleagues replaced the −carotene's acceptor with successively stronger electron grabbers. When Stegeman and his colleagues at Central Florida added these chromophores to a polymer known as poly (methyl methacrylate) and spun it into films, they found that the films made with chromophores that had the strongest electron grabbers produced an NLO effect 35 times stronger than the starting molecule. For many applications, says Garito, that is “very close” to what's needed for commercial success. The chromophores break down when they are exposed to light and heat, ruling out their use in light-based communication devices, says Nasser Peyghambarian, an NLO expert at the University of Arizona, Tucson. But he and others believe that the same charge-separation strategy could improve the NLO properties of other types of molecules, among them more robust chromophores made from chains of linked rings or compounds containing stable metal complexes. “It's fairly wide open,” says Garito. If so, more and more communications visionaries are likely to see the light. 5. AIDS Vaccine # Looking for Leads in HIV's Battle With Immune System 1. Jon Cohen BETHESDA, MARYLAND—When officials at the National Institutes of Health (NIH) tapped Nobel laureate David Baltimore last December to head its newly formed AIDS Vaccine Research Committee, they hoped he would give a much-needed boost to a floundering field. It is too early to tell whether Baltimore's committee will live up to those expectations. But a 4-day meeting* on AIDS vaccine development held here at the NIH 2 weeks ago suggests that it will have plenty of new leads to follow. AIDS vaccine development has been limping along since 1994, when NIH decided not to push ahead with large-scale tests of the leading vaccine candidates because they didn't look promising enough (see sidebar). The organizers of last week's meeting—the ninth in a series of these annual gatherings—tried to give the field a booster shot by adding reports of cutting-edge basic research on how HIV causes disease to the standard AIDS vaccine fare. This strategy seemed to pay off. Several reports—including one by Baltimore—on the protective role played by the immune system's killer cells, called cytotoxic T lymphocytes (CTLs), sparked much discussion. A presentation on the possible capture of a long-sought factor that provides some protection against HIV sent a ripple through the meeting, as did a suggestion that a goat virus might provide the basis for an HIV vaccine. Although the 550 attendees heard no headline-grabbing talks, veterans in the field came away moderately encouraged. “Pessimism is destructive,” said Anthony Fauci, director of the National Institute of Allergy and Infectious Diseases (NIAID). “I'm convinced that we will have a vaccine and that we'll have it in a reasonable period of time.” For many AIDS vaccine researchers, the most troubling roadblock has been teasing out which immune responses a vaccine should trigger to protect a person from HIV. Most viral vaccines work by stimulating the immune system to produce antibodies that bind to a virus, preventing it from infecting cells. But it's not clear that antibodies offer much protection against HIV. “As I try to understand the role of antibodies, … I keep coming up with a blank,” said Baltimore, hence the focus on CTLs. Like smart bombs, CTLs search out and destroy cells that a virus has infected. They play a key role in the body's natural defenses, and many traditional vaccines stimulate their production along with antibodies. “Cytotoxic T lymphocytes are at least very important, if not the most important, thing [for protection from HIV],” said Baltimore. However, an AIDS vaccine that stimulates CTLs would face hurdles too, as Baltimore's own work shows. When a cell is infected by HIV, it typically puts a piece of the virus on its surface. CTLs are trained to pulverize any cells that display these viral peptides. Baltimore and his Massachusetts Institute of Technology colleague Kathleen Collins, collaborating with HIV CTL guru Bruce Walker of the Massachusetts General Hospital, have new evidence that HIV escapes the deathblows of CTLs by preventing cells from displaying viral peptides. Building on work first published in the March 1996 Nature Medicine by the Pasteur Institute's Olivier Schwartz and co-workers, the Baltimore group focused on the HIV protein Nef. As the Schwartz group showed, Nef can prompt cells to yank down from their surfaces a molecule known as the major histocompatibility complex (MHC), which displays viral peptides to the immune system. The group predicted that this “down-regulation” of MHC would make HIV-infected cells resistant to CTL killing—just what the Baltimore group's new data now show. Baltimore says these findings imply that if researchers hope to base a vaccine on stimulating CTLs, the timing is critical: Unless CTLs flood the bloodstream shortly after an infection occurs, HIV may undermine their utility by dispatching Nef. That may be a tall order, but one heartening talk at the meeting, by Duke University's Kent Weinhold, suggests that the down-regulation of MHC by Nef does not completely shut down CTL activity. “I don't think it's an all-or-none phenomenon,” says Weinhold. In contrast to the Baltimore group's finding, Weinhold and co-workers showed that CTLs could kill 10% to 40% of HIV-infected cells in a test-tube assay. And Weinhold had more good news about CTLs. Researchers have long worried that an HIV vaccine might only work against the particular viral strain from which it is made. But Weinhold's study of blood taken from people given experimental vaccines made from pieces of HIV showed that their CTLs could kill cells infected with a wide range of HIVs. “The extent to which he found cross-reactivity [to different HIV strains] with CTLs is surprising people,” said Patricia Fast, NIAID's associate director for vaccines and prevention. Still, many questions remain about how long these CTL responses will last, and whether they pack enough wallop to fend off a real infection. Mary Klotman of New York's Mount Sinai School of Medicine presented intriguing findings pointing to another possible criterion for a promising vaccine. Klotman's work addresses a finding by Jay Levy of the University of California, San Francisco, that some white blood cells carrying a CD8 receptor on their surface produce some kind of soluble factor that can suppress HIV. Levy called this mysterious substance the CD8+ Antiviral Factor (CAF). But despite years of trying to identify CAF, Levy has been stumped. Recently, Klotman and Arevik Mosoian of her lab isolated and characterized an “extremely small protein” that she thinks might be the elusive CAF. If so, vaccine developers may have yet another lead to analyze in their hunt for immune responses that correlate with protection. Perhaps the meeting's most unusual talk suggested that infection with a goat virus might protect humans from HIV. Angeline Douvas of the University of Southern California in Los Angeles reported that people infected with caprine arthritis-encephalitis virus—a distant HIV relative, common in Mexico, that appears harmless to humans—make antibodies that react with HIV. Bruce Weniger, an epidemiologist at the Centers for Disease Control and Prevention in Atlanta, went to the microphone and connected the dots. “This is really a remarkable finding that raises the wild speculation that you discovered the natural accidental vaccine for HIV … [in] the way cowpox was the natural accidental vaccine that works for smallpox,” said Weniger. Douvas replied that they hope to test this hypothesis by doing an epidemiologic study of people infected with the goat virus to see if they have a lower incidence of HIV infection. Will any of these exotic findings pan out? Who knows. But this meeting proved that AIDS vaccine research is at least bubbling with ideas. • * Conference on Advances in AIDS Vaccine Development, 4-7 May, sponsored by NIAID, Bethesda, Maryland. 6. AIDS Vaccine # Planned Tests in Thailand Spark Debate 1. Jon Cohen While researchers in the United States are looking for new approaches to developing AIDS vaccines (see main text), two first-generation vaccines are inching toward full-blown efficacy trials in Thailand. However, as plans for the trials advance, a debate is heating up. Critics within Thailand, using data supplied by U.S. researchers, contend that the vaccines are likely to prove worthless, while others worry that the furor could frustrate the best chance to determine once and for all whether these preparations can slow the epidemic. To date, more than two dozen AIDS vaccines have been tested in small-scale human studies to assess their safety and ability to trigger immune responses. None of these tests has as yet moved into an efficacy trial, which would involve several thousand people at a cost of$20 million or more. In June 1994, two vaccines, both made from genetically engineered versions of HIV's surface protein gp120, were set to take the plunge. But a panel convened by the National Institutes of Health (NIH) decided that the probability of the preparations efficacy was too low for the government to fund these trials. The panel's decision, however, applied only to the United States; countries facing more intense epidemics might deem it worth the risk.

Indeed, researchers in Thailand, facing a serious AIDS epidemic, pressed ahead with small-scale trials of these vaccines, originally made by Genentech of South San Francisco and Chiron Corp. of Emeryville, California. Thai researchers hope to begin these efficacy tests as early as next year. David Baltimore, the Nobel laureate who heads a committee that advises NIH on AIDS vaccines, says he recently discussed Thailand's interest in testing these vaccines with Natth Bhamarapravati, who chairs a similar group in Thailand. Baltimore says he thought their thinking “was all very reasonable.”

Last month, however, Praphan Phanuphak, who directs the Thai Red Cross Society's Programme on AIDS, wrote to the deputy governor of Bangkok that the gp120 vaccines are “not useful in preventing HIV infection” and that it was “not appropriate for Thailand to allow (approve) an efficacy trial of the mentioned vaccine.” Praphan based his doubts on data that he requested from Steven Wolinsky of Northwestern University and David Ho of the Aaron Diamond AIDS Research Center in New York City. However, these data, on “breakthroughs”—people who became infected despite receiving the gp120 vaccine in the small-scale U.S. trials—should not be overinterpreted, according to Wolinsky: “We'd be remiss if we didn't provide the Thais with that information. But it never was intended to stop or start a trial. Our colleagues in Thailand are very intelligent, and they don't need David Ho or Steven Wolinksy to tell them what to do.”

Praphan's criticisms are troubling Thai officials. William Heyward, an epidemiologist who heads the AIDS vaccine program of the Centers for Disease Control and Prevention in Atlanta—which has been helping Thailand stage AIDS vaccine trials—says the issue came up last week when a delegation visited from the Thai Ministry of Health. They were “very concerned” that Thai politicians “would not fully understand the debate and would back off [from efficacy trials],” says Heyward.

That worries the U.S. companies supplying the vaccines, as well as their academic and government collaborators. Chiron hopes to begin a 300-person study this summer that includes a new gp120 vaccine made with HIV subtype E, a strain found in Thailand. If all goes well, the company hopes to start efficacy trials around 2000. Donald Francis, who a few months ago started the company VaxGen to develop Genentech's gp120 vaccine, calls the attacks on the trials “myopic.” VaxGen has raised more than \$24 million during the past few months to stage efficacy trials of the Genentech vaccine, which Francis says could begin next year.

7. Marine Geology

# New Way to Hit the Hot Spot Hints at a Complex Pacific

1. Richard A. Kerr

Hundreds of volcanoes, some active like Hawaii, others long dead like Samoa and Bikini, poke above the blue waters of the wide Pacific. But these islands don't begin to suggest the profusion of more than 10,000 submerged volcanic seamounts now known to dot the Pacific sea floor. Many marine geologists have argued that these are the products of a dozen or so “hot spots,” fixed sites of deep-seated volcanism that left a trail of seamounts as the oceanic plate moved over them. But these hot spots are often elusive, smoldering unnoticed beneath the sea floor or burned out long ago. Now two marine geophysicists have come up with an elegant way to track them down.

Geophysicists already have a search technique, but it requires knowing seamount ages, which can be hard to come by. In this week's issue of Nature, marine geophysicists Paul Wessel and Loren Kroenke of the University of Hawaii describe their new method, which relies on simple geometry yet seems to pinpoint some hot spots more precisely than before. In fact, these first results raise questions about traditional hot-spot theory, hinting that the volcanism responsible for many of the Pacific seamounts may not be as deeply rooted in the mantle or as stationary as researchers had thought. “It's a really elegant method, very clever,” says marine geophysicist Seth Stein of Northwestern University. “It's a more objective way of testing a lot of things.”

In the traditional “backtracking” method, the volcanic track burned through the sea floor guides researchers to the hot spot. For instance, the Hawaiian Islands and the Emperor Seamounts stretch across the North Pacific in a broad V with the point toward the southwest; the bend in this chain formed 43 million years ago when the Pacific Plate abruptly shifted direction. By radiometrically dating rocks from at least a few seamounts and knowing the plate's motion from the shape of the seamount track, researchers can backtrack along the chain to where the hot spot should be—beneath the island of Hawaii, in the world's clearest example. But in other cases, it is hard to fathom just how an assortment of poorly dated seamounts might sort into a series of hot-spot tracks, especially if the hot spots themselves have gone dormant and no longer spark the telltale volcanism. In one small region of the Pacific, around the Line Islands, researchers have proposed up to five separate hot spots to explain the chaotic jumble of seamounts.

Wessel was using this backtracking approach in a study of Pacific volcanism when something went awry. “I was trying to write the backtrack code [on my computer],” he says, “but the first time I did, I made a mistake.” Instead of producing Hawaii's familiar V-shaped hot-spot track with the bend pointing to the southwest, the computer produced a mirror image, with the bend pointing to the northeast. Wessel realized that the computer had traced not the chain of seamounts seen today but the actual path of a single seamount over time, as plate motions dragged it away from the hot spot first in one direction and then another.

When Wessel plotted these “sea-floor flow lines” for several seamounts in the Hawaii-Emperor chain, he noticed that they formed two trends, tracking seamount paths before and after the shift in direction. The trends cross to form an X marking the one position every seamount had passed across: the site of the hot spot. This technique, Wessel realized, freed him from the need to know seamount ages. Hence, rather than being limited to a couple of hundred seamounts that were often poorly dated, he could instead seek hot spots for any seamounts whose positions were mapped. And thanks to a recent burst of declassification of military-satellite data—which reveal the tiny sea-surface bulges induced by seamounts' gravity (Science, 3 March 1995, p. 1260)—information was available on 8800 Pacific seamounts taller than 1 kilometer.

With this new technique in hand, Wessel and Kroenke set out in search of Pacific hot spots. In Hawaii, hot spotting initially came up with a big bright X 150 kilometers from current volcanic activity. So Wessel and Kroenke inserted a slight change in plate direction 3 million years ago—an adjustment other researchers had earlier suggested for independent reasons—and Hawaii appeared right on target. But the X marking another well-known feature in the far southern South Pacific, the Louisville hot spot, still fell 375 kilometers from its traditional location. In this case, however, it seems that the conventional methods are off and hot spotting is on the money. Seismic rumblings have been detected within 75 kilometers of the X, and last year, dredging at the site of seismic activity uncovered fresh lava within 100 meters of the surface, suggesting an active volcano. Hot spotting also puts a big X near, although not right on, the traditional site of the Rarotonga hot spot (see illustration).

In this preliminary analysis, however, no obvious X's appeared to mark the rest of the dozen or so other presumed Pacific hot spots. “I would have expected that there would be more very bright spots,” says paleomagnetician Gary Acton of Texas A&M University. Instead, “there were a lot of fuzzy regions.” Wessel and Kroenke say that problems such as inaccurate assumed plate motions could explain the fuzzy patches.

On the other hand, the problem may not be in the data but in traditional hot-spot theory, say Wessel and Kroenke. The idea has been that hot-spot volcanoes are fed by deep, hot plumes reaching to the very bottom of the mantle near Earth's core. Such deeply rooted hot spots should not move much, and indeed the Hawaii and Louisville hot spots have not moved detectably relative to each other, according to hot-spotting results. But hot spots that are drifting relative to each other because they have shallow or no roots could create the fuzzy pattern.

The hot-spotting results are “pointing up the real problems we have with plume theory,” says geophysicist Marcia McNutt of the Massachusetts Institute of Technology. Her fieldwork and that of others has suggested that some sea-floor volcanoes might form a chain not one at a time but all at once, as magma that had pooled just below a plate leaked up along a crack or weak zone. Still, the answers are not yet clear. Even Wessel agrees that “things aren't suddenly easy now that we have this technique.” However, with this new method, he says, the hunt for hot-spot origins is likely to heat up.

8. # Nitrogen Oxide Pollution May Spark Seeds' Growth

1. David A. Malakoff
1. David A. Malakoff is a science writer living in Bar Harbor, Maine.

Could a common air pollutant be fooling the seeds of some wild plants into germinating when conditions are deadly to the seedlings? That's the provocative question raised by a finding reported in this issue. Two ecologists have learned that the seeds of a common California wildflower can be prompted to germinate by exposure to nitrogen oxides, gases produced by both natural wildfires and motor vehicles and power plants.

Botanists have long recognized that many plants inhabiting fire-prone areas, such as the arid forests of the Southwest and the shrub-choked hillsides of southern California, grow best on freshly burned-over land. Scorched soils are often rich in nutrients, such as nitrogen and phosphorus released from burned vegetation, and they are free of shrubs that can otherwise shade sun-loving seedlings. For many of these phoenixlike plants, a fire's searing heat is what prompts them to rise from the ashes. Heat cracks the hard, outer coat of seeds that can lie dormant in the soil for decades. This allows water to seep in, spurring growth.

In the late 1970s, however, scientists began realizing that heat was not the only trigger. Exposing the seeds of some of these plants to charred wood was enough to prompt germination, and in the last few years, smoke has been recognized as a potent trigger. Researchers have identified dozens of smoke-germinated species around the world and isolated more than 70 compounds in smoke thought to be potential germination triggers. But they never identified exactly which compound induced germination in a particular plant.

Now, on page 1248, plant ecologists Jon Keeley and C. J. Fotheringham of Occidental College in Los Angeles report that nitrogen dioxide, one of the several nitrogen oxides found in wood smoke, spurs the germination of seeds from Yellow Whispering Bells, an annual herb that springs up after fires in California's chaparral and sage scrub plant communities. “This work documents a fascinating and novel mechanism for cueing seed germination to a fire event,” says William Schlesinger, a biogeochemist at Duke University in Durham, North Carolina, who has studied fire-related germination.

In their study, Keeley and Fotheringham, who are interested in how burned sites are recolonized by plants, collected dormant seeds from Whispering Bells growing in the Los Angeles Basin. The researchers then placed the seeds in small chambers and exposed them to either straight nitrogen dioxide or wood smoke made by burning a few sprigs of chamise, a common chaparral shrub, on an electric hot plate. Exposure to either smoke or nitrogen dioxide for as little as 1 minute triggered germination in every seed. The researchers saw similarly high germination rates after they exposed seeds to nitrogen dioxide vapors from sand, paper, and water that had been allowed to absorb smoke as long as 2 months before the tests. In contrast, seeds not exposed to smoke or nitrogen dioxide did not sprout. “The germination response of these seeds to even small quantities of nitrogen dioxide was remarkable,” says Keeley.

He cautions that nitrogen dioxide may not be a universal trigger: Although other smoke-triggered species he has tested do respond to the gas, a few appear to be responding to something else. Schlesinger adds that not even all populations of Yellow Whispering Bells rely on fire to germinate. In the early 1980s, he and a colleague demonstrated that germination in seeds from Whispering Bells collected in the Mojave Desert appears to be sparked by sand abrasion. The fact that fires are relatively rare in the desert would explain this different adaptation, he says.

A nitrogen oxide trigger “makes sense” for plants growing in fire-prone regions, Schlesinger says, because the germination signal would reach even seeds not directly exposed to smoke. He notes that rates of soil nitrification, in which soil bacteria convert ammonium, a common fire byproduct, into nitrogen compounds, can “absolutely skyrocket in the months after a fire.” Buried seeds may be responding to these changes in soil chemistry, which can elevate nitrogen oxide levels, rather than directly to the smoke, researchers say.

The seeds' sensitivity to nitrogen oxides raises what Schlesinger calls a “provocative but reasonable question”: Could nitrogen oxide from air pollution trick seeds from some smoke-sensitive plants into germinating before a wildfire has cleared the way for their growth? If so, hapless seedlings might sprout and quickly die in the shade of bushes, and the soil's natural bank of stored seed might become depleted, threatening future plant populations. “The concern is that after a fire you wouldn't see regeneration, because the seed reserves wouldn't be there,” says Keeley.

The question has particular relevance in the Los Angeles area, which is home to dozens of smoke-germinated species—and has the nation's highest levels of nitrogen oxide pollution. U.S. Forest Service researchers estimate that as much as 45 kilograms of airborne nitrogen are deposited per hectare in the Los Angeles Basin annually, mostly in the form of nitric acid. This potentially represents at least five times the amount of nitrogen oxides needed to trigger germination in Yellow Whispering Bells.

“The effect of air pollution on germination could be real but very hard to document,” says Schlesinger. One problem is that the soil and atmospheric chemistry of nitrogen compounds are complex and poorly understood, as is the exact process by which nitrogen oxides trigger seed germination. In addition, ecologists say that finding small plants that mistakenly sprout amid thick shrubs could be a daunting task—a task further complicated by hungry rodents that might eat the appetizing shoots long before any scientist finds them.