News this Week

Science  12 Sep 1997:
Vol. 277, Issue 5332, pp. 1602

You are currently viewing the .

View Full Text

Log in to view the full text

Log in through your institution

Log in through your institution


    Blunting Nature's Swiss Army Knife

    1. Charles Seife
    1. Charles Seife is a science writer in Riverdale, New York.

    Researchers are trying to develop inhibitors of a wide variety of proteases, the protein-cleaving enzymes that play a role in everything from the common cold to cancer

    A class of drugs with an unusual mechanism for fighting disease has breathed new hope into AIDS research during the past year. At least for some patients, these compounds, which block the activity of a protein-snipping enzyme that the AIDS virus needs to complete its life cycle, are turning what was once a death sentence into a chronic condition. But AIDS isn't the only disease that may be treatable by blocking the activity of these enzymes, called proteases. Researchers are now exploring the use of protease inhibitors against everything from infections, including the common cold and the parasitic disease schistosomiasis, to inflammatory conditions like asthma and rheumatoid arthritis, and even cancer.

    View this table:

    Proteases are such an inviting target for therapies because they play key roles in disease development. James McKerrow, a molecular biochemist at the University of California, San Francisco (UCSF), describes them as “Mother Nature's Swiss army knife. They have many different functions, even though they often have nearly identical structures.”

    When the AIDS virus infects cells, for example, the proteins it needs to reproduce itself are synthesized as one large precursor molecule. It's then the job of the HIV protease to split out the individual proteins. If a drug blocks the protease, the replication of the virus is also blocked. Proteases play similar roles in the life cycles of other viruses, including the coronaviruses, which cause about one-third of all cases of the common cold. Proteases are also important contributors to inflammatory damage, and by breaking down proteins that normally hold cells in place, they apparently help cancer cells escape from the primary tumor and spread to new sites in the body.

    These findings, combined with the encouraging precedent set by the AIDS antiprotease drugs, have spurred companies ranging from small start-ups like San Francisco's Arris Pharmaceutical to industry giants like Merck to try to develop therapies based on protease inhibitors. “A whole host … of protease inhibitors is beginning to appear,” says Les Hudson, Pharmacia & Upjohn's vice president for research.

    None of these has yet hit the market, and information on drugs in development can be hard to come by because of proprietary concerns, but preliminary clinical trials of at least a few protease inhibitors are under way. The hope is that these will provide therapies for currently untreatable conditions such as colds, or drugs with fewer side effects for conditions such as rheumatoid arthritis that are now treated but with potent, but unspecific, drugs.

    One of the protease inhibitors that has entered clinical trials is a drug called APC366, being developed by Arris as a possible treatment for asthma. It's directed against a protease called tryptase, released when immune cells known as mast cells are activated by an allergen, such as dust mite feces or pollen. Tryptase contributes to asthma symptoms both directly and indirectly.

    Tryptase's protein-splitting action accounts for its direct effects, among them increasing the permeability of the blood vessels. The fluids that consequently ooze into the tissue lead to swelling in the narrow passages of the lungs. Tryptase also promotes scarring in the lungs by cleaving and activating enzymes that foster the deposition of collagen, a protein found in scar tissue. In addition, the enzyme can indirectly cause lung damage by helping recruit eosinophils, white blood cells that are the bane of asthmatics. “Eosinophils damage the epithelium of the lung, … and tryptase is a potent chemoattractant for human eosinophils,” says Stephen Holgate, an immunopharmacologist at the University of Southampton in the United Kingdom.

    To block these effects, Arris chemists looked for a compound—preferably a small molecule that could be taken by mouth—that could fit into, and block, tryptase's active site, the relatively small part of an enzyme that binds its target molecules and performs the catalytic work. A strategy for fine-tuning candidate molecules to improve their protease binding sped up Arris's effort, says Heinz Gschwend, the company's executive vice president for research and preclinical development. He declined to divulge the details of the procedure, but it is apparently a variant of one of the new combinatorial methods that enable drug designers to generate and test many thousands of related compounds simultaneously (Science, 31 May 1996, p. 1266). Gschwend does say, however, that this “Delta technology,” as the company calls it, “is phenomenally simple, and it really shortens the time it takes to develop [protease] inhibitors.”

    APC366, one of the products of that procedure, proved to be a potent tryptase inhibitor. Once animal testing had indicated that the drug would be sufficiently safe to test in humans, the company began clinical trials, which have already produced some promising results. In one small study of 16 asthma patients that was completed last September, the drug reduced by 32% the patients' “early airway response”—the swelling caused by mast cells, which sets in right after exposure to an allergen. At several hours after administration, it increased lung capacity by 42%.

    At the same time, the treated patients apparently showed few side effects. That may give APC366 and other tryptase inhibitors an advantage over the steroids now commonly used to treat asthma, even though the inhibitors may be less effective at reducing inflammation, says Lawrence Schwartz, a microbiologist at Virginia Commonwealth University in Richmond. Still, the encouraging results of the APC366 trial, with its limited number of patients, will need to be confirmed in larger studies before the drug can be approved by the U.S. Food and Drug Administration.

    A host of targets

    Tryptase is not the only cellular protease being targeted by Arris and other companies. Another is the enzyme cathepsin K, which is made in large quantities by human osteoclasts, cells that are engaged in the remodeling that constantly goes on in normal bones. The osteoclasts destroy bone, presumably aided by cathepsin K's protein-digesting abilities, while another type of cells, called osteoblasts, build it up. In healthy adults, the two are equally matched, so bone density remains roughly unchanged. But if the osteoclasts get the upper hand, the bones weaken, as may happen, for example, in osteoporosis, a disease that afflicts more than 25 million people, most of them postmenopausal women, greatly increasing their risk of bone fractures.

    At least two companies, SmithKline Beecham and Arris, are attempting to develop drugs that inhibit cathepsin K, in hopes that they will be able to combat the bone weakening of osteoporosis. As a step toward that goal, both have determined the three-dimensional structure of the protease when it is complexed with smaller molecules—a compound SmithKline calls E-64 and one from Arris designated APC3328. (The results appear in the February issue of Nature Structural Biology.) By helping researchers tailor molecules to the geometry of the protease's active site, such structural information could aid in designing effective inhibitors.

    Proteases may also play a crucial role in the deadly activity of metastasizing cancer cells. By digesting away the proteins that hold cells together, the enzymes apparently help cancer cells burrow through tissue so that they can spread to new sites throughout the body. “There's overwhelming evidence that [proteases] are a major mechanism in invading normal tissue,” says Marc Shuman, an oncologist at UCSF. “If we can inhibit the proteolytic activity, we can prevent [metastasis] from happening.” For example, Shuman's group has identified a new protease that is expressed on the membranes of prostate cancer cells and may contribute to the cells' ability to metastasize. The team has also found a naturally occurring protein that inhibits the protease and can, at least in lab cultures, block the spread of prostate cancer cells, Shuman says.

    Shuman has no plans to attempt clinical trials with his inhibitor, but Lance Liotta's team at the National Cancer Institute in Bethesda, Maryland, has developed a family of protease inhibitors that is moving toward the clinic. These so-called “TIMPs”—tissue inhibitors of metalloproteinases—block the protease enzymes that several different types of cancer cells use to snip the collagen fibers in the extracellular matrix. This activity not only helps the cancer cells to spread, but also fosters the growth of the new blood vessels needed to supply blood and oxygen to growing metastatic tumors. British Biotech, plc., a company based in Oxford, United Kingdom, is beginning clinical trials of TIMPs in a number of cancers, including those of the pancreas, lung, and brain.

    Besides trying to stop specific cells, like osteoclasts or cancer cells, drug companies have also set their sights on members of a series of proteases—a so-called protease cascade—that act successively to bring about blood-clot formation. These companies hope that inhibitors of individual proteases could act as anticoagulant drugs more selective than the ones now given to heart attack, stroke, and other patients in danger of suffering life-threatening blood clots. Heparin, for example, inhibits many “factors” in the clotting cascade. A protease inhibitor might be able to break the cascade at only a single point, making it less prone to cause unwanted side effects.

    One promising candidate for such an inhibitor comes from biochemist Joanna Chmielewska of Pharmacia & Upjohn in Stockholm, Sweden, and her colleagues. It is directed against an early participant in the clotting cascade: factor Xa, which splits prothrombin, releasing another active protease, thrombin. Thrombin in turn cuts fibrinogen, ultimately producing fibrin, the raw material of clots.

    Chmielewska, like many of the other researchers designing protease inhibitors, built on a knowledge of her target protein's three-dimensional structure that had previously been determined by other investigators. “The combining site [of factor Xa] is quite well understood; you can predict the structure of a protein or chemical that will sit in the site,” says Pharmacia & Upjohn's Hudson. That enabled Chmielewska to come up with a small-molecule inhibitor that gums up factor Xa's active site. She says, however, that the inhibitor is still in an early phase of research.

    Parasites have proteases, too

    While many researchers are looking for inhibitors that can block the activity of proteases indigenous to the body, others have as their targets the proteases that invading pathogens deploy to help them establish infections. HIV is only one example. The cold-causing coronaviruses also rely on proteases to release their component proteins from a large precursor protein.

    The body makes its own inhibitors of these viral proteases. Neuroimmunologist Arlene Collins of the State University of New York, Buffalo, is investigating proteins called cystatins, found in saliva and tears, that jam the coronavirus protease. Unfortunately, the cystatins themselves can't be used to treat colds, for they are digested if given orally, making it difficult to get the drug to where it is needed. An inhibitor “won't work if you can't get it to the protease,” says Collins.

    Larger invaders, including the parasites that cause diseases such as schistosomiasis, also need proteases. The mature schistosome, which lives in the bladder, intestines, and other organs, uses a protease to break down hemoglobin in the blood for food and also to process proteins needed for egg production. It is a good target for antischistosomiasis drugs, McKerrow says, because blocking it will interrupt the parasite's life cycle even after it is established in the body.

    Arris has designed several low-molecular-weight compounds that resemble the particular linked amino acids split by the adult schistosomiasis protease and would thus be expected to bind to the enzyme's active site. In test tube studies, McKerrow's group has shown that these compounds inhibit the protease. What's more, McKerrow says, the compounds may work on related proteases from other parasites, like the one that causes Chagas' disease, which is a major cause of heart disease in Latin America. The inhibitors are about to be tested in dogs.

    Even protease inhibitors that are showing promise in early human trials won't make it to market anytime soon, however. They won't benefit from the streamlined drug approval that allowed the HIV protease inhibitors to be marketed quickly. But the research is unlikely to slacken. Says Arris's Gschwend, “There are plenty of targets in proteases to keep us busy for a long time.”


    Searching for Living Relics of the Cell's Early Days

    1. Gretchen Vogel

    CHAFFEY'SLOCKS, ONTARIO—Some 50 evolutionary biologists gathered here for an annual meeting held from 23 to 27 August by the Canadian Institute for Advanced Research. The meeting saw model organisms and cherished hypotheses rise and fall as participants discussed how to tease out clues to the emergence of the first complex cells.

    Model Organisms Unseated

    Because ancient fossils are scarce, scientists trying to look back billions of years to the early days of life turn to single-celled organisms that appear very primitive. But two findings presented at the meeting suggest that creatures thought to represent the earliest stages in the evolution of complex cells may not be such good models after all.

    Ancient baggage?

    Giardia may have gained and lost a mitochondrion.


    At stake is a glimpse of what eukaryotic cells—cells with nuclei—looked like before they acquired the energy-producing organelles called mitochondria. Only eukaryotes have mitochondria, which have their own small complement of genes, and most scientists believe that these organelles originated when a bacterium took up residence inside a primitive eukaryotic cell. Scientists had identified certain protists that have nuclei but no mitochondria—called Archezoa—as possible direct descendants of the first eukaryotes. But it now appears that mitochondria may have appeared well before these “ancestors.” The findings, says molecular evolutionist Geoffrey McFadden of the University of Melbourne in Australia, have “blasted apart our ideas about the earliest eukaryotic cell.”

    Other Archezoa have lost their ancestral status in recent years when closer examination of their DNA revealed genes that code for proteins typically associated with mitochondria in other organisms. The implication was that these organisms once had the organelles, then somehow lost them. Now Andrew Roger, a postdoctoral fellow at the Marine Biological Laboratory (MBL) in Woods Hole, Massachusetts, reports that another Archezoan, Giardia lamblia, may have also had a mitochondrion at one point. The organism causes giardiasis, often caught by hikers who drink unfiltered stream water—and it has been one of the primary models for early eukaryotes.

    Roger, MBL molecular evolutionist Mitchell L. Sogin, and their colleagues found that the parasite has a gene for a protein called chaperonin 60, which helps other proteins fold properly. This gene is thought to be a good tracer for mitochondria, because it is thought to have moved from the mitochondrion to the nucleus in other organisms.

    Another model organism was struck down by Martin Embley, a microbiologist from the Natural History Museum in London. He and postdoctoral fellow Robert Hirt found that an organism in the group called Microsporidia, microbes that can cause deadly infections in immune-compromised people, contain heat shock protein 70, which helps stabilize other proteins after heat exposure. It closely resembles a heat shock protein in other organisms that is thought to be derived from mitochondria. The good news, Embley says, is that data from other proteins suggest that the Microsporidia are closely related to fungi. So if the organisms once had mitochondria but lost them, they may be true fungi, susceptible to antifungal drugs in patients.

    One candidate for a truly premitochondrial eukaryote is still unchallenged, says Roger: the oxymonads, a little-studied group of organisms that live in the hindgut of termites. But data on those creatures—which have yet to be cultured in the laboratory—are scarce.

    If mitochondria had already taken up residence before the “model” organisms evolved, Roger says, these organelles must be far more ancient than scientists had thought. They may have originated at the same time or even earlier than the cell nucleus, implying that the mitochondrion played a crucial role in the development of the eukaryotic cell. That idea intrigued the other scientists at the meeting, although not everyone is convinced that mitochondria predate the surviving eukaryote ancestors. “It's still a debatable story,” says molecular evolutionist Michael Gray of Dalhousie University in Halifax, Nova Scotia. Still, participants agreed that the discoveries have created an opening for new theories of how complex cells were put together.

    Ancient Organelle Glimpsed

    Evolutionary biologists may be losing their models for the ancestral eukaryotic cell, before it acquired the energy-producing organelles called mitochondria. But they may be gaining a whole gallery of models for the steps that took place later, after specialized bacteria took up residence to form these organelles.

    At the meeting, molecular evolutionist Gertraud Burger of the University of Montreal presented genetic portraits of three mitochondria, from three different organisms, that may chronicle the transition from newly acquired bacterium to highly modified organelle. One of these organisms, from a recently discovered freshwater protist called Reclinomonas americana, provides the best link yet between the mitochondria of plants and animals and the bacterial precursor, says Burger's colleague, Michael Gray of Dalhousie University in Halifax, Nova Scotia. “You'd be hard pressed to look at it and say this wasn't a compressed bacterial genome,” he says. The other two may offer glimpses of later stages in the process.

    Burger and her colleagues suspected they had unearthed an ancestral mitochondrion in Reclinomonas when they sequenced its mitochondrial genes. As they reported in Nature in May, the sequence contained genes similar to every gene ever identified in any other mitochondrion, plus 18 new ones—many of which closely resemble those of bacteria. This complement of genes suggests, says Gray, that Reclinomonas inherited an ancestral mitochondrion but preserved it better than other organisms.

    At the meeting, Burger also presented data from two nearly completed sequences from relatives of Reclinomonas, called Jakoba libera and Malawimonas jakobiformis. Although their mitochondrial genomes also contain almost all the genes found in other mitochondria, Jakoba has only 10 of the bacterial genes identified in its cousin, and Malawimonas has just four. This suggests, says Burger, that they may provide a link between Reclinomonas and other, more derived mitochondrial genomes—and clues to how and why cells lose mitochondrial genes or transfer them to the nucleus.

    Now the team is preparing to sequence two more protists, one of which, called Macropharyngomonas salina, may have a mitochondrial genome three times larger than that of Reclinomonas, suggesting that it might be packed with bacterial genes—and may offer an even closer look at the ancestral organelle.


    First p53 Relative May Be a New Tumor Suppressor

    1. Steven Dickman
    1. Steven Dickman is a free-lance writer in Cambridge, Massachusetts.

    In the 20 years since its discovery, the p53 gene has become one of the most heavily scrutinized genes in history. Indeed, it's referenced in over 8000 papers in Medline, the online biomedical abstract service. The fascination is easy to understand: Loss or inactivation of p53, which is a so-called tumor-suppressor gene, is thought to contribute to the development of 50% of all human cancers. All that time, p53 was thought to be an only child, with no close relatives. Now, researchers have discovered a new gene, a long-lost cousin called p73, that bears a strong resemblance to p53.

    No longer alone.

    The p73 proteins, although longer, resemble p53 in three regions: the transcription activation (TA, 29% identical) and DNA binding domains (63% identical) and also the domain where p53 binds itself (OLIGO, 38% identical). The labeled amino acids indicate residues that are frequently mutated in p53 and are conserved in p73.


    It is being greeted with the same surprise as any newfound relative. “Given the intense interest in this area, the fact that [p73] slipped through the cracks is surprising,” says cancer biologist Tyler Jacks of the Massachusetts Institute of Technology (MIT). But the new gene should generate some intense interest of its own, because its protein not only resembles the p53 protein, but also seems to have similar activities. The p53 protein acts as a “security guard,” deployed when a cell's DNA is damaged to prevent the cell from becoming cancerous. It does this by either inhibiting cell growth until the damage is repaired or causing the cell to commit suicide through a process called programmed cell death or apoptosis. The p73 protein appears to share these growth-inhibiting and apoptosis-promoting effects, although what triggers them and exactly what its cellular role is are both unknown.

    Those findings, together with p73's location in a region of chromosome 1 that is often deleted in cancers including neuroblastoma, a malignant tumor of nervous tissue, suggest that it, too, may be a tumor suppressor. “This [discovery] will titillate a whole lot of people,” predicts cancer geneticist Bert Vogelstein of Johns Hopkins University School of Medicine, a pioneer of p53 research. Indeed, if p73 can stand in for p53 when that gene is lost, it might be possible to design new cancer drugs that work by turning on p73 in tumors lacking p53.

    Molecular biologist Daniel Caput and his colleagues at the pharmaceutical company Sanofi Recherche in Labege, France, identified the p73 gene while looking for something completely different, namely genes that respond to certain immune system regulators. When the French team sequenced the many potential targets their screen had turned up, they were shocked to find that one false positive had remarkable similarities to p53.

    As the researchers report in the 22 August issue of Cell, the proteins made by p73 are somewhat larger than p53. But they found that one section of p73 closely resembles the so-called “core binding region” of p53. Many of p53's activities depend on its ability to regulate other genes, and the core binding region is where the protein attaches itself to the DNA when exerting its effects. Of 177 amino acids in that region, 112 are identical. Additional similarities turned up in two other sections thought to be involved in p53 activity—one needed for its gene regulatory effects and another where it apparently binds to itself. These resemblances are enough to suggest that the two genes are the progeny of a gene that was duplicated in some ancient cellular event. Indeed, p73 may be the ancestral gene, because a gene found in squid that was supposed to be that species's version of p53 is actually more similar to p73.

    The structural similarities between p53 and p73 also suggested that the proteins might have similar roles in the cell. So Caput and his colleagues joined forces with their longtime collaborator, cell biologist Frank McKeon, who studies gene expression and cell division at Harvard Medical School in Boston, to look for parallels. One way p53 restrains cells that have damaged DNA is by triggering the production of a protein called p21, which blocks cell division. The Caput-McKeon team found that adding p73 to a line of neuroblastoma that lacks the gene also triggers p21 production, an indication that p73 inhibits cell growth through the same pathway used by p53.

    In a paper that appears in this week's issue of Nature, molecular biologist William Kaelin at the Dana-Farber Cancer Institute in Boston and his colleagues report similar findings with another tumor cell line. Kaelin's team also found evidence that p73 can mimic p53's ability to cause cell suicide. When overexpressed in these cells, p73 latched onto stretches of DNA to which p53 normally attaches itself when instructing a cell to self-destruct.

    Together, the findings suggest that p73, too, may be a tumor suppressor, an idea that is buttressed by its provocative chromosomal location. The Caput-McKeon team found p73 in a region near the tip of chromosome 1 that was already suspected of harboring one or more tumor suppressor genes, because the region is often missing in tumor cells.

    The teams did find one major point of difference between the two genes, however. Unlike p53 protein, p73 does not seem to be produced in response to DNA damage. That implies that the protein is not a cell “security guard” the way p53 is. Early results of experiments in which McKeon and Caput deleted the p73 gene in mice suggest another possibility: It may be “developmentally important,” he says, especially in the brain and immune system, although how remains to be clarified.

    If p73 is a tumor suppressor, it may behave somewhat differently than p53 and other previously discovered tumor suppressors. Classic tumor-suppressor genes require two “hits” to be inactivated—a partial or complete deletion of one of the two gene copies, for example, and another, lesser change that cripples the second copy. But Caput, McKeon, and their colleagues have evidence that one p73 copy is already inactive in normal cells—the apparent result of a mysterious process called imprinting. Its precise function isn't known, but during embryonic development, imprinting alters certain genes so that the copy inherited either from the mother or the father is specifically shut down.

    If one p73 copy has been silenced by imprinting, then only one hit—loss of the active copy—might be all that it takes to tip a cell into the uncontrolled growth of cancer. Says Kaelin, “p73 may be the first example of a new paradigm for how tumor-suppressor genes are involved in cancer.”

    Indeed, molecular biologist Rogier Versteeg of the Academic Medical Center in Amsterdam, the Netherlands, has evidence that an imprinted gene may be involved in neuroblastoma development. He has identified two sites of chromosome damage that contribute to neuroblastoma by knocking out as-yet-undiscovered tumor-suppressor genes. Both lie in the same region of chromosome 1 where p73 is located, and one illustrates “a strong bias” toward loss from the maternal copy of the chromosome in the cancer cells. This bias implies that this specific copy is the active one and must be lost to cause the cancer.

    Other work from the Caput-McKeon team suggests that this mystery tumor-suppressor gene may be p73. When they looked for the gene in neuroblastoma cell lines, they found that one p73 copy had been lost. And while they couldn't uncover any mutations in the remaining copy, most of the cell lines made no detectable p73 protein, implying that the second copy had been silenced by imprinting.

    In spite of the differences in the roles of p53 and its new cousin, both in normal cells and in cancer, the family resemblances may be strong enough for them to substitute for each other. If so, says MIT's Jacks, cancer might be treated by finding a way to switch on p73 in tumor cells that have lost p53. “Even if p73 is not normally involved in tumor suppression, maybe it could be recruited,” says Jacks. Now McKeon and Caput are searching for further family members. But the discovery of p73 is already certain to captivate their peers.


    HIV Gets a Taste of Its Own Medicine

    1. Jon Cohen

    In an attempt to fight fire with fire, researchers have engineered a virus that usually infects cattle to attack the AIDS virus in humans. The innovative approach has so far shown promise only in test-tube experiments, but it is attracting widespread attention among AIDS researchers. “It's really on the verge of a breakthrough,” says Nava Sarver, who oversees development of novel AIDS treatments at the National Institute of Allergy and Infectious Diseases (NIAID).

    Trojan horse.

    CD4 and CXCR4 receptors expressed by genetically engineered VSV bind to HIV's gp120 protein on surface of infected cell.


    Yale University virologist John Rose and co-workers describe in the 5 September issue of Cell how they have constructed a potential HIV treatment by modifying vesicular stomatitis virus (VSV), which farmers detest because it causes a mouth infection in cattle that prevents them from eating. As the Yale researchers' experiments show, their newfangled VSV selectively targets and destroys HIV-infected human cells. “It's a pretty interesting way of harnessing a virus for peaceful purposes,” says the University of Pennsylvania's Robert Doms. “It's a very clever approach.”

    The work builds on recent discoveries made by Doms and others about a two-part handshake between HIV and the cells it infects. After HIV binds to the CD4 receptor on a white blood cell, it also must link to another molecule found on the cell's surface, known as a chemokine receptor. Once these handshakes are complete, HIV gains entry, and shortly thereafter, new virus proteins make their way to the cell's outer coating, where they stick out like a flag of victory.

    Rose and colleagues reasoned that if VSV could be induced to express these receptors on its surface, they would bind to the HIV proteins displayed on infected cells, turning VSV into a kind of guided missile. To test this idea, the researchers stitched into VSV the genes that code for CD4 and one of HIV's favored chemokine receptors, CXCR4, and added their engineered VSV to a culture containing HIV-infected cells. The virus did, indeed, target just the infected cells, killing them rapidly. “VSV is so fast,” says Rose—much faster than HIV, he notes.

    A potential downside to this approach is that the modified VSV might kill cells that aren't infected by HIV. Rose believes that won't happen because he has stripped VSV of its own surface protein, which is what allows it to infect a broad range of cells. “Without its normal coat, it can't infect anything,” says Rose. But only animal tests will provide evidence of that, cautions NIAID director Anthony Fauci.

    Although Fauci has high praise for the concept's ingenuity, he is concerned that it might take an impracticably high dose of the modified VSV to make a real dent in a person's HIV levels. Another worry, says monkey researcher Ronald Desrosiers of the New England Regional Primate Research Center in Southborough, Massachusetts, is that the body will quickly develop an immune response against VSV, limiting its ability to attack HIV.

    Still, Sarver, Fauci, and others are anxious for Rose and colleagues to put their viral guided missile to more stringent test-tube and animal tests. Desrosiers already has begun working with Rose to test the concept in monkeys that have been infected with SIV, HIV's simian cousin. Desrosiers expects to have results in the next few months. Even if they are positive, however, human trials will require the approval of the Food and Drug Administration, which has shown great caution in the past about putting potentially therapeutic viruses into people.

    Rose's strategy is not limited to attacking HIV. NIAID's Sarver suggests that if researchers can swap different receptors into this “gutted” VSV, the precisely targeted viruses could be used in everything from vaccines to gene therapies to cancer treatments. “We're not there yet,” says Sarver, “but the potential applications are enormous.”


    Opening the Door to More Membrane Protein Structures

    1. Anne Simon Moffat

    If one picture is worth a thousand words, recent advances in x-ray crystallography methods are providing the equivalent of the Encyclopedia Britannica. Crystallographers are now churning out the three-dimensional (3D) structures of proteins at the rate of one or more per day. But so far one key group of proteins—those normally located in cell membranes—has been badly underrepresented in this ever-expanding book of structural knowledge. The problem is that researchers have to crystallize proteins in order to probe their structure with x-rays, and integral membrane proteins often can't withstand being removed from their normal environment. A solution to this problem may be at hand, however. A solution to this problem may be at hand, however.

    Clarified vision.

    The electron density map shows the locations of some of the amino acids in or near the bacteriorhodopsin proton pathway. The red dots indicate two water molecules located in a water pocket behind tyrosine 57 and aspartic acid 212.


    On age 1676 of this issue, Eva Pebay-Peyroula of the Institute of Biological Structure in Grenoble, France, and Gabriele Rummel, Jurg Rosenbusch, and Ehud Landau of the University of Basel, Switzerland, offer a newly detailed 3D structure of bacteriorhodopsin, a key protein enabling the salt-loving bacterium Halobacterium salinarium to convert energy from sunlight to chemical energy that the bacterial cells can use. The structure, with a resolution of 2.5 angstroms, offers the most detailed look yet inside this solar power plant.

    Bacteriorhodopsin absorbs photons of light and uses their energy to pump protons out of the bacterial cell, generating a chemical and electrical gradient across the membrane that can serve as an energy source. Water molecules located in the coils of the protein are thought to act as a kind of bucket brigade that helps the protein's amino acids achieve this proton transport. By showing the precise positions of eight water molecules, as well as of the amino acid side chains, the new structure is helping investigators understand how the bacteriorhodopsin choreographs this process. The structure may also provide a model for understanding the operation of other membrane proteins with similar overall architecture, which include several receptors for hormones and neurotransmitters.

    But even more important, the novel technique the Swiss-French team developed for crystallizing bacteriorhodopsin could make it much easier to grow crystals of membrane proteins generally, opening the way to direct analyses of their structures. The researchers grew their crystals within lattices of membranelike materials, created by mixing lipids and water under appropriate conditions. The technique, says Columbia University protein crystallographer Eric Gouaux, “offers a very creative approach to growing crystals of membrane-bound proteins that, in the past, have proved difficult to prepare.”

    The problem is that proteins embedded in membranes are surrounded mainly by fat molecules. When researchers then try to coax the proteins out of the membrane so that they can be crystallized, they tend to unfold and become disorganized in their new watery surroundings. In the past, this limited investigators to making two-dimensional bacteriorhodopsin crystals, consisting of just a single layer of well-ordered protein molecules. These could be analyzed by a technique called electron crystallography, which enabled pioneering researchers such as Richard Henderson and his colleagues at the Medical Research Council Laboratory in Cambridge, United Kingdom, to solve the bacteriorhodopsin structure to a resolution of 3.5 angstroms. But that resolution was not quite good enough to see all the atoms of the protein and its associated water molecules. Currently, only x-ray crystallography can deliver a view that is sharp enough. And x-ray crystallography requires 3D crystals.

    Biophysical chemist Ehud Landau reasoned that the best way to grow 3D crystals of bacteriorhodopsin would be to keep the protein in a congenial environment. He custom-built a lattice of lipids, fats similar to those in membranes, to house the bacteriorhodopsin during crystallization. “We have faked the natural environment,” says Rosenbusch. As a result, extracted membrane proteins that previously fell apart in a matter of hours were “happy and functioning for months,” yielding very small, 3D crystals.

    But growing 3D crystals of bacteriorhodopsin for the first time was not enough to get the job done. Because the crystals were tiny and fragile hexagonal plates, only 20 to 40 micrometers (millionths of a meter) in diameter and 5 micrometers in thickness, they tended to crack easily, like potato chips. Because this distorts the spatial arrangement of bacteriorhodopsin in the crystal lattice, such crystals diffract x-rays poorly. To get the desired high resolution, the team needed an x-ray beam narrow and bright enough to produce good diffraction patterns from the very small crystal fragments remaining. The team turned for help to physicist Christian Riekel at the European Synchrotron Radiation Facility (ESRF) in Grenoble, a powerful x-ray source. ESRF has recently opened a new microfocus beamline that can pack hundreds of billions of photons a second into a beam just 10 micrometers in diameter (Science, 29 August, p. 1217).

    The result was a high-resolution picture of bacteriorhodopsin that is broadly similar to the existing picture, based on the electron crystallographic analysis and on studies in which H. Ghobind Khorana of the Massachusetts Institute of Technology and his colleagues mutated amino acids to identify the ones associated with proton transport. But some critical differences may require researchers to modify their view of how that transport occurs.

    For example, mutational studies had identified aspartic acid 85 as a key component of proton transfer. But the new structure shows that this amino acid is too far from the schiff base attached to retinal, the pigment bound to bacteriorhodopsin that puts the proton pump in operation by absorbing photons, for direct proton transfer to occur. However, because the new structure locates water molecules in the bacteriorhodopsin proton pathway, researchers do have their first clues to how some water molecules may help shuttle protons through the bacterial membrane. “The issue is, are these the interesting water molecules?” says Henderson. Rosenbusch says they soon hope to answer that question, by introducing mutations into the bacteriorhodopsin molecule that alter proton transfer and may allow the effects of the water molecules to be seen.

    Crystallographers may soon be getting such intimate views of other membrane proteins, thanks to Landau's technique for mimicking the natural environment. Eventually, they may be able to design and build artificial membranes with different lattice sizes to snare and crystallize membrane proteins of varying sizes and shapes for structural studies. Riekel also predicts that further improvements in ESRF's microfocus beamline will allow analysis of crystals as small as 5 micrometers across. “Microdiffraction has a large future for difficult-to-crystallize substances,” he says. If so, protein chemists will soon see a large number of membrane proteins in their structural encyclopedia.


    More Signs of a Far-Traveled West

    1. Richard A. Kerr

    For 2 decades, earth scientists have argued over the proposal that chunks of North America's western edge migrated thousands of kilometers northward to their present positions. Now, a few exquisitely preserved fossils have given the theory an extra push. A team of geologists and geophysicists reports in this issue of Science that the superb condition of marine fossils from near Vancouver Island provides a key test of the evidence, which consists of traces of ancient magnetism in the fossil-laden rock. Seventy million years ago, they conclude, Vancouver Island was adjacent to Baja California, thousands of kilometers to the south.

    After so many years, the dispute (Science, 5 May 1995, p. 635) is not likely to be settled by a single finding. Indeed, critics are already identifying loopholes. Still, the study, reported on age 1642 by paleontologist Peter Ward of the University of Washington in Seattle, paleomagnetician Joseph Kirschvink of the California Institute of Technology in Pasadena, and their colleagues, “could influence the fence sitters,” says geologist Darrel Cowan of the University of Washington, who has written on possible geologic tests of the so-called Baja-British Columbia hypothesis.

    The hypothesis originated 20 years ago in studies of the magnetism locked in rocks from the so-called exotic terranes of the Pacific Northwest—large chunks of crust that seem to have formed elsewhere and migrated to their present positions. Because Earth's magnetic field is horizontal at the equator but vertical at the poles, the inclination of a rock's magnetism shows how far north it was when it formed and locked in Earth's field. Along the west coast of North America, researchers measured paleomagnetic inclinations smaller than they should be if the rock had formed in place, as part of North America. Many paleomagneticians took that to mean that the terranes had slid up the coast from far to the south, much as California west of the San Andreas fault is sliding now.

    Most geologists and some paleomagneticians disagreed. For one thing, they couldn't see the large faults that would have guided the rocks northward. Instead, they proposed that the terranes originated offshore at roughly their present latitudes and later docked with North America. The shallow magnetic inclinations were misleading, they argued, because most of these measurements came from great masses of frozen magma, which the tectonic jostling of the coast could easily have tilted from their original orientations over tens of millions of years.

    Sedimentary rock would solve that problem, because it is laid down in recognizable horizontal layers. But sedimentary rocks from the largest terrane—the Insular superterrane, which makes up much of the coastal crust from northern Washington state into Alaska—seemed to have been heated long after they formed, wiping them clean of their original magnetic signature.

    Kirschvink, however, realized from unrelated work he and Ward were doing in central California that temperature-sensitive fossils could be a marker for rock that hadn't been heated and magnetically altered. Ward, in turn, knew of fossils from the Texada and Hornby islands—part of the Insular superterrane off the east coast of Vancouver Island—that fit the bill. These fossils of extinct mollusks, called ammonites and inoceramids, retained the pearly luster of the living animals, implying that the paleomagnetic inclinations in the surrounding rock could be relied on. The magnetism was about 25 degrees shallower than expected at Vancouver Island's current latitude, implying that 70 million to 80 million years ago, when the rock formed, the terrane was 3500 kilometers to the south, off Baja California.

    Ward and his colleagues say the condition of the fossils also bears witness against another process that could have skewed the data: compaction shallowing. Sediment can compress by 50% or more as the weight of new sediment above it squeezes out the water between its grains. Compaction can reduce any existing paleomagnetic inclination as the magnetic grains tilt toward the horizontal under the compression.

    Ward and Kirschvink note that even if there had been compaction, the 10° of compaction shallowing typically found couldn't explain the 25° in their rocks. But the fossils argue against even that much compaction, says Kirschvink: “One of the indicators [of sediment compaction] is to look for compacted fossils, and they're not present in these sediments.” Further reassurance comes from the carbonate globules that tend to encase these fossils, he says. These concretions began to form soon after the fossils were buried, he says, and would have welded grains of rock in place, preventing compaction shallowing. That's “a good way to stop inclination error,” he says. “We can rule it out.”

    “Their argument is pretty good,” says paleomagnetician Kenneth Kodama of Lehigh University in Bethlehem, Pennsylvania, who has studied compaction shallowing in the lab. But, says Kodama, “they haven't thought enough about whether or not the concretions could have formed after a certain amount of compaction. As they say, it may not explain all of their shallowing of inclination, but it could explain part of it.” Sedimentologist Peter Mozley of the New Mexico Institute of Mining and Technology in Socorro agrees. “Typically, concretions are thought to form early, but late-stage concretions do exist,” he says.

    Kodama and Mozley recommend further analysis of the concretions to pin down just how much compaction actually occurred. Until then, says longtime Baja-British Columbia critic Robert Butler of the University of Arizona, Tucson, “the whole thing goes on without a definitive closure.”


    New Exotic Particle Points to Double Life for Gluons

    1. Gary Taubes

    As elementary particles go, the gluon is looking inordinately multitalented. It has long been known as an insubstantial “force particle” that flits between quarks, conveying the strong force that binds them together into protons, neutrons, and other composite particles. Now there's evidence that gluons can act as constituents of matter as well, contributing mass just as quarks and electrons do. Two experiments, one at Brookhaven National Laboratory in Upton, New York, and the other at CERN, the European laboratory for particle physics, have just weighed in with glimpses of a short-lived hybrid particle consisting of a quark, an antimatter quark, and a gluon.

    Forced union.

    Colliding a proton containing two up quarks and a down with a pion (an anti-up and a down quark) spawns an “exotic meson” containing either two quarks and a gluon or four quarks.


    “If they've really done it, it's a confirmation that the gluon is a constituent of matter with the same degree of respectability as a quark,” says University of Chicago physicist Jonathan Rosner. The particle, which the Brookhaven physicists call an exotic meson, wouldn't be the first clue that gluons can step into a more substantive role: For the last 2 years, researchers have been tantalized by evidence of glueballs, particles made up of nothing but gluons (Science, 15 December 1995, p. 1756). But the exotic meson, reported by the Brookhaven group in the 1 September Physical Review Letters and confirmed by the CERN group, would put gluons on an equal footing with quarks in a single particle. It would also vindicate theoretical predictions based on quantum chromodynamics (QCD), the theory that describes the behavior of quarks and gluons.

    Because QCD's complex equations break down at energies typical of ordinary matter, a reliable understanding of the interactions between quarks and gluons is hard to come by. In the early 1980s, however, theorists, including Frank Close, a physicist at Britain's Rutherford Appleton Laboratory, and Ted Barnes of the University of Tennessee and the Oak Ridge National Laboratory, recognized that QCD allowed for the existence of a hybrid quark-gluon particle.

    “However you modeled this thing,” Close explains, “it suggested such a state should exist—whether as a quark, an antiquark, and single gluon trapped together in a sort of bag … or as a quark and antiquark linked by a tube of flux, which would be a collection of gluons working together. If you excited the flux, like plucking a violin string, you would find such exotic states,” because the gluons would gain energy, the equivalent of taking on mass.

    The constituent gluon or gluons would also endow the hybrid with a set of quantum numbers—parameters such as spin and charge—that the equations forbid for more mundane mesons, consisting only of a quark and an antiquark. Because these exotic quantum numbers would manifest themselves in how the short-lived particles decayed, physicists could look for these signatures in the shower of debris created by slamming particles together in accelerators. They have been doing that, unsuccessfully, for 15 years. Because its signature is excruciatingly subtle and buried beneath a background of ordinary mesons, says Rosner, “the quark-antiquark-gluon hybrid is not the easiest thing in the world to identify.”

    In 1994, the Brookhaven physicists, a group some 51 strong from seven institutions, began a new search, using a newly refurbished 2-decade-old detector, known as the multiparticle spectrometer. Brookhaven's Alternating Gradient Synchrotron sent a beam of pions, a kind of ordinary meson, crashing into a liquid hydrogen target. The multiparticle spectrometer captured the resulting spray of secondary particles.

    In some 200 million collisions between pions and protons in the target, says Suh-Urk Chung, a spokesperson for the experiment, 40,000 yielded debris, suggesting that a new particle had formed in the collision and survived for all of 10−23 seconds. The charge, energy, and direction of the decay products, says Chung, imply that it has a mass of 1.4 gigaelectron volts and a set of quantum numbers that cannot be carried by a typical quark-antiquark meson.

    The physicists say they are heartened by the nearly immediate confirmation that came from the Crystal Barrel collaboration at CERN, which was looking in an entirely different system of particles. According to University of Bonn physicist Eberhard Klempt, a member of the collaboration, the CERN experiment studied collisions between antiprotons from CERN's Low Energy Antiproton Ring and a heavy hydrogen target. Among the collisions captured in the Crystal Barrel detector were ones in which an antiproton encountered a neutron in the heavy hydrogen. The meeting of antimatter and matter generated a flash of energy, briefly creating “a subsystem with exotic quantum numbers which fully agree with the Brookhaven numbers,” says Klempt. Researchers from a third experiment, this one at the Serphukov Laboratory in Russia, may also be seeing a similar phenomenon.

    But physicists aren't quite ready to hail the definitive arrival of the hybrid. As Close puts it, “There are two questions to be answered: First of all, have these people really discovered the first example of something with these exotic quantum numbers? And second, if so, what does it tell us about the deep structure of the object?”

    For starters, he says, the three groups measure subtly different masses, suggesting either the existence of three subtly different exotic mesons, or that one, two, or all three of the collaborations are somehow being fooled. The third possibility, says Close, “which I think is probably the likely one, is that there is a single object, and some complicated dynamics are causing it to give the impression of being at different masses.”

    But even if the particle is real and the quantum numbers truly exotic, the groups might be seeing not a quark-antiquark-gluon hybrid, but a particle made of two quarks and two antiquarks. That would make the exotic meson the first four-quark particle to be identified—an interesting finding, but not as interesting as a quark-gluon hybrid. Barnes notes, however, that “QCD predicts so many [four-quark particles] that if you take them seriously, you have to expect to see hundreds of them,” contrary to the results so far.

    If the experiments really have corralled a hybrid, physicists will have a new test bed for QCD: a system in which to isolate and study the behavior of gluons. “The way they'll behave should be different than the way quarks behave,” says Close. “The challenge will now be showing exactly what that difference is.”