News this Week

Science  27 Mar 1998:
Vol. 279, Issue 5359, pp. 2038

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


    North America's Wars

    1. Heather Pringle
    1. Heather Pringle is a science writer in Vancouver.


    Most archaeologists have long viewed war as a disease of civilization. Only kingdoms and states with great armies, the theory goes, slaughtered opponents for economic ends and left battlefields littered with corpses. Tribal societies, such as those that flourished across prehistoric North America, were thought to have fought mainly for sport and to have halted hostilities after only a few deaths because they lacked resources for extensive battles. Political correctness, too, has favored the idea that native peoples lived in harmony. But new studies of prominent Southwestern cultures clash with this vision of a peaceful past.

    Clever new ways to read the subtle marks of periodic warfare in such features as the arrangement of villages and the placement of wells, plus direct evidence of massacres (see sidebar), are persuading archaeologists that ancient North American societies made war as fiercely as any nation states. In order to acquire scarce resources, particularly when the climate turned harsh, combatants slaughtered women and children, razed settlements, and inflicted stunningly high casualty rates. The turnaround in thinking is most dramatic in the Southwest. There, cultures once idealized as peaceful, such as the Hohokam of southern Arizona and the Anasazi of the Colorado Plateau, now seem to have been shaped by warfare, researchers argued this week at a special symposium at the Society for American Archaeology (SAA) meetings in Seattle.

    Defensive buildup.

    A massive wall, now highlighted by green growth, protects a Hohokam platform mound east of Phoenix.


    “There's been more and more evidence coming forward that the levels of violence in prehistoric times were quite high,” says Jon Driver, an archaeologist at Simon Fraser University in Vancouver and SAA program chair. Such studies are changing many anthropologists' minds about war in tribal societies, he says. “Now that we're seeing different types of warfare around the world, I think people are opening up more to warfare” as an explanation for prehistoric events.

    Not everyone, though. Skeptics such as Linda Cordell, director of the University of Colorado Museum in Boulder, argue that particularly in the Southwest, warfare advocates have yet to present convincing proof that ancient hostilities claimed many lives. “I'm looking for good syntheses and good reports with lots of bodies,” she says. Much of the new evidence for war is open to less sensational interpretations, agrees Charles Adams, an archaeologist at the University of Arizona, Tucson. Warfare proponents, he notes, have “taken a body of information and mixed all sorts of stuff in there. It could be the result of warfare, but they haven't demonstrated it.”

    Even those arguing for tribal warfare, such as Hohokam expert Glen Rice, director of Cultural Resource Management at Arizona State University in Tempe, admit that their ideas are “still controversial.” But for a growing number of scientists, the evidence can no longer be ignored. Says Rice, one of the new converts, “I've flopped recently from being very unconcerned about warfare to being very concerned that it is an important factor.”

    From farmers to warriors

    Although there are new signs of prehistoric warfare across North America, some of the newest and most controversial data come from the river valleys of Arizona. The Hohokam, who flourished there from A.D. 300 to 1450, have long been idealized as peaceful maize farmers who cooperated to build canals. But now Rice argues that not only did the Hohokam fight, conflicts over water actually defined their society.

    The Hohokam cremated their dead, so Rice can't rely on broken bodies to make his claim. Rather, he makes a more inferential argument based on what seems to be the Hohokam social system. From architecture and other clues, Rice argues that Hohokam society was organized to require constant readiness for battle, in an arrangement also seen in highly warlike societies like Sudan's Nuer and Central America's Quiche Maya. This structure, known to social scientists as “segmentary organization with complementary opposition,” is more simply described as “me against my brother, me and my brother against my father, and me and my brother and my father against my uncles,” says Neal Ackerly, an archaeologist with Dos Rios Consultants in Silver City, New Mexico. He and other experts agree: If the Hohokam had this structure, then they must have been warriors.

    Rice's ideas are based on his studies of major Hohokam irrigation canals east of Phoenix, some of which stretch as long as 15 kilometers along valley floors and sustained as many as seven separate Hohokam communities each. He and others wondered how water was parceled out among the communities during droughts, when canal water levels dropped.

    Some researchers theorized that a powerful centralized authority lived in the village nearest the canal head gate on the river and peacefully controlled water and economic life along the canal. But in Hohokam settlements east of Phoenix, Rice found no trace of centralized wealth and authority. Instead, in the architecture atop large earthen mounds at the center of Hohokam communities, he found public council rooms that were in paired and opposing arrangements. This, plus traces of distinct ceremonial regalia, indicated the presence of two or more opposing elites, he says. And some communities contained two or more of these mounds, each with their own sets of elites. That implies that there were modular segments of society that could be organized into bigger units—a hallmark of segmentary organization, explains Jeffrey Dean, a Southwest specialist at the University of Arizona, Tucson.

    With no controlling central authority, reasoned Rice, canal-end communities could only obtain water by resorting to occasional force and constant threats against upstream communities. Yet they cooperated with those same communities during times of canal maintenance. “It doesn't have to be a constant state of war, just a constant preparedness to go to war,” he explains. If so, each end settlement should be more populous—thereby ensuring greater military strength—than its neighbors upstream. In this way a fragile balance of power could be struck.

    Using the size and number of platform mounds as indicators of community size, Rice tested this theory on data from five canal systems, gathered by his own group and others. Just as predicted, the biggest settlement on each canal was the one farthest from the head gate. Moreover, in four of the canal systems, the end community was as large as all the others combined, giving it clear military superiority. And one canal-end community yielded many potsherds in a foreign style characteristic of the Lower Colorado Plateau. This suggested that people from as far as 260 kilometers away had joined the canal-end community, reinforcing its fighting forces, says Rice.

    Elsewhere, in the Hohokam hinterlands away from the largest canal systems, the evidence also suggests a volatile social system capable of exploding into hostility at the first sign of threat. At the SAA meetings, David Wilcox, an archaeologist at the Museum of Northern Arizona in Flagstaff, and Jerry Robertson, a former U.S. infantry rifle company commander, reported that Hohokam sites on a high mesa north of Phoenix have previously unnoticed defensive features such as lookouts. And Theodore Oliver, one of Rice's graduate students, found that sites in the Tonto Basin area east of Phoenix show signs of intentional destruction and catastrophic abandonment in the late 1300s. In one site, archaeologists sampled 32 of 80 rooms and found that 84% of them had been set ablaze—probably not accidentally, for adobe and cobble walls are difficult to burn. And 63% of the excavated rooms still contained a complete inventory of household objects.

    Oliver also sees a trend in Hohokam village size that he interprets as an effect of endemic warfare. About A.D. 1250, Hohokam farmers in the Tonto Basin area lived scattered in 98 settlements, averaging less than 10 rooms each. But soon after, they started clustering into larger enclaves. By the 1400s, just 13 Hohokam settlements remained, averaging 43 rooms each. It seems that the Hohokam gathered into larger and more defensible settlements as the threat of war increased. “Eventually what happens,” says Rice, “is that the system seems to destroy itself through warfare.”

    But the evidence of burnt houses does not persuade critics such as Adams. Razed homes need not mean fighting, he says: Historic groups in the Southwest often burned dwellings when a resident died of natural causes. So far, Dean adds, Rice's evidence falls short of traditional anthropological definitions of warfare. “I don't really think that there were armies wandering around out there, nor is there any evidence for what you'd call conquest.” Yet he and others agree that Rice's work is the first to make sense of the sometimes conflicting evidence seen in Hohokam sites. It's “a brilliant stroke … a major breakthrough,” says Dean. It's “the type of work that needs to be done” to eventually settle the question of war, says Adams.

    War among the pueblos

    The case for war doesn't stand or fall on the Hohokam, however. Clearer signs of battle come from the Anasazi of the Four Corners region. These prehistoric pueblo people were envisioned as peaceful farmers for years, but recently controversial evidence pointing to conflict has turned up, including human bones that bear traces of apparent cannibalism (Science, 1 August 1997, p. 635).

    In fact, warfare was not only an integral part of life in the Four Corners for centuries, but it decimated Anasazi and Mogollon corn farmers during the late 13th and early 14th centuries, argues archaeologist Stephen LeBlanc, of the University of Southern California in Los Angeles, in a major study to be published this fall. In stark contrast to popular perceptions of modern Pueblo peoples, he concludes that of the 27 pueblo groups that flourished along the Colorado Plateau at the beginning of this period, just three—the Zuni, the Hopi, and the Acoma—survived intense warfare among themselves. “The pueblo people you see today are basically the victors.”

    Some of LeBlanc's most compelling evidence is osteological—the remains of massacres found at five major Anasazi sites. In the early 1990s, for example, Bruce Bradley and other archaeologists at Crow Canyon Archaeological Center in Cortez, Colorado, found human skeletons, primarily men and children, abandoned with no funerary ceremony at two southern Colorado pueblos dating to between A.D. 1250 and 1285. Many bore smashed skulls and other signs of violence. When LeBlanc extrapolated the numbers of dead from the sampled locations to the sites as a whole, he concluded that 50% of the 500 inhabitants of Sand Castle Pueblo and 62% of 80 inhabitants of Castle Rock Pueblo were slaughtered in vicious massacres.

    At the same time, the design and location of Anasazi and Mogollon dwellings changed dramatically. In the early 1200s, Anasazi farmers lived in small, single-story room blocks arranged in an L or straight line and situated as much as 0.4 kilometers from the community well. But by about 1300, at every Anasazi site, they had moved into large, two-story pueblos built around central plazas. They also constructed wells on the plaza or within 4 meters of the outer walls. The higher roofs made better fighting platforms for residents warding off an attack, says LeBlanc. “And my suspicion is that this massive desire to have the drinking water very, very close to the pueblo is a way of protecting the women,” who would otherwise risk attack while getting daily water.

    The murals found in some pueblos of this time also reflect a preoccupation with warfare. Along the walls of a kiva at Pottery Mound near Albuquerque, New Mexico, for example, researchers have found paintings of what appear to be warriors armed with shields to deflect enemy arrows.

    No one knows just what ignited such an intense war on the Colorado Plateau, but LeBlanc notes that the 500-year-long global cold spell popularly known as the Little Ice Age began about 1300 and may have caused crop failures, famine, and hostilities. High-altitude pollen studies, he notes, revealed that tree lines in southern Colorado dropped in the 1300s as the climate turned colder and wetter. Whatever the cause, however, he argues that the evidence is overwhelming that warfare was an integral part of Anasazi life, shaping the very structure of their settlements. “The idea that you can understand Southwestern prehistory and pretend there was no warfare is just silly,” he concludes.

    But the Anasazi work too has its critics. “I can construct a model, based on ethnographic evidence, that takes into account virtually all of the warfare data and explains them in terms of ritual behavior,” says Adams. Prehistoric pueblo peoples in the Four Corners region, for example, once killed, then burned and mutilated the bodies of people they suspected to be witches. And the Anasazi may have built pueblos on seemingly defensive hilltops simply because they, like many modern tribal groups, viewed high places as sacred. Even depictions of violent acts don't necessarily mean that war was common: Artists often portray mythical battles fought by deities or shamans. “I just don't think we can explain a lot of the archaeological record in terms of conflict,” says Adams.

    Warfare researchers counter that their opponents are demanding unreasonable standards of proof. “If your only evidence of warfare is going to be bodies all over the place,” says Oliver, “then you're not going to be able to find it. Even in modern warfare, only a small percentage of the people involved in a conflict die.” These researchers think it's high time that anthropologists accept that life in the ancient Southwest could be very nasty, brutish, and short. “War is something like trade or exchange,” says Lawrence Keeley, head of the Department of Anthropology at the University of Illinois, Chicago. “It is something that all humans do.”


    Crow Creek's Revenge

    1. Heather Pringle
    1. Heather Pringle is a science writer in Vancouver.

    In A.D. 1325, prehistoric warriors stormed a palisaded village in South Dakota along the Missouri River, razed its earth lodges, and slaughtered many of its estimated 800 inhabitants. The 550 dead, mainly men and children, were later tossed into a mass grave. Nearly 95% of the intact skulls bore cut marks from scalping.

    Now known as Crow Creek, the village and its grave were discovered in the 1970s; the unfortunate villagers have been identified as the ancestors of a historic tribe called the Arikara. But for years, with prehistoric America seen as a peaceful place (see main text), Crow Creek was generally thought to be an extreme and isolated case of violence. Now, new analyses suggest that this massacre was merely one act in a long-running life-and-death struggle between Crow Creek and the ancestors of two neighboring tribes: the historic Mandan and Hidatsa.

    Battle scars.

    A broken skull at Tony Glas shows intent to kill.


    In research still in progress, Douglas Owsley, a physical anthropologist at the National Museum of Natural History (NMNH) in Washington, D.C., has found clear traces of massacres at two 14th century villages within striking range of Crow Creek. In a village known today as Tony Glas, built by the ancestors of the Mandan and Hidatsa, he found the remains of at least 50 people in one earth lodge. Most died between the ages of 15 and 25, and most were women. There were fractures caused by blows to the head and mouth and breakage of skulls after death, including “cut marks and intentional fragmentation of bone, intentional mutilation,” says Owsley.

    A nearby village called Helb, also inhabited by ancestors of the Mandan and Hidatsa, shows signs of a similar disaster—burned houses, an unfinished palisade, and a scattering of human remains, says Marvin Kay, an archaeologist at the University of Arkansas, Fayetteville, who excavated the site. “Those [remains] have cut marks from being scalped,” notes Owsley. “The evidence is consistent with a massacre.”

    To both Owsley and Kay, the evidence points to a desperate struggle between the two groups of farmers. Kay thinks the climate cooling of the 14th century led to crop failures, sparking battles over the narrow zone of arable land in the Missouri River Trench, which was “on the fringe or northern limits for prehistoric agriculture,” he says. Like many bitter wars before and since, it seems that the Crow Creek massacre and its companion slaughters were fought for land.


    A Hint of Unrest at Yucca Mountain

    1. Richard A. Kerr

    Geologists studying southern Nevada's Yucca Mountain have always looked to the past to see what the future might hold for the mountain, which is the leading candidate to become the long-term U.S. repository for highly radioactive wastes from nuclear power plants. Because the mountain has been so quiet for millennia, researchers concluded that the risks of earthquakes and volcanoes striking it are low. But on page 2096 of this issue of Science, a team of geologists and geophysicists reports that the crust at Yucca Mountain is stretching at least 10 times faster today than it has on average over geologic time.

    If so, the area could be suffering a bout of rapid crustal deformation that would boost the chances of a disaster such as a volcanic eruption during the 10,000-year life of the repository. In fact, the authors, led by Brian Wernicke of the California Institute of Technology in Pasadena and James Davis of the Harvard-Smithsonian Center for Astrophysics (CfA) in Cambridge, Massachusetts, suggest that geologists have underestimated the hazards at Yucca Mountain by a factor of 10.

    Safe spot?

    Tripod-mounted GPS receiver measures movement atop Yucca Mountain.


    For now, other researchers are intrigued but not yet alarmed by the new findings, which stem from a satellite-based Global Positioning System (GPS) survey. “It's an interesting and provocative idea,” says geologist Bruce Crowe, of Los Alamos National Laboratory in New Mexico, who has long worked on volcanic risks at Yucca Mountain. “It has to be looked at carefully.” But he and others caution that Wernicke's group has actually measured only a few millimeters of stretching in 6 years, which is near the limit of what GPS can reliably detect. “This is testing GPS,” says geophysicist Robert Smith of the University of Utah. “I don't think the answer is in.”

    Geologists agree that the best way to get that answer is to continue the GPS survey, which Wernicke and his colleagues have been conducting since 1991. As they report in Science, the team measured the positions of five benchmarks set along a 34-kilometer line stretching roughly east-west across Yucca Mountain. By comparing radio signals from several GPS satellites overhead, the researchers could determine the position of a benchmark with millimeter accuracy.

    Wernicke's team checked the benchmark positions seven times between 1991 and 1997. They found that their survey line was lengthening at a rate of 1.7 ± 0.3 millimeters per year, giving a total stretching of about a centimeter over the 6 years of the study. Although that's only one-quarter the rate of crustal deformation seen in active areas like the San Andreas fault, it's about 10 times faster than geologists had inferred for the Yucca Mountain region, based on how much nearby faults have slipped over hundreds of thousands of years. Wernicke and his colleagues suggest that the Yucca Mountain area may be undergoing a geologically brief episode of rapid crustal stretching, perhaps driven by magma rising beneath it.

    That could change hazard estimates based on the low long-term deformation rate recorded on faults. Previous analyses had put the risk that, say, a new volcano would pierce the repository, due to open in 2010 at the earliest, at 1 chance in 10,000 during the next 10,000 years (Science, 8 November 1996, p. 913). But if the rate of crustal deformation is upped by a factor of 10, so are the geologic risks, says Wernicke.

    So far, researchers such as Crowe who have helped to evaluate geologic hazards at the repository are cautious about changing their risk estimates. And others such as geophysicist Wayne Thatcher of the U.S. Geological Survey in Menlo Park, California, wonder whether the minute stretching the group observed is real. “The changes they're looking at are really small; that's a little worrisome. In this business, we don't usually look at such small changes,” he says. He and others also find the error bars surprisingly small; if they were larger, the observed motions would shrink. But co-author Richard Bennett of CfA finds no sign of additional errors in the data.

    Even if the crust is moving as fast as Wernicke's team finds, “I would expect the hazard to go up, but not by an order of magnitude, “says Crowe. A magmatic intrusion might first trigger new volcanoes near the youngest nearby volcano, Lathrop Wells, and so would pose less of a threat to Yucca Mountain, he says.

    Some scientists also wonder whether the apparent crustal extension is being driven by anything as threatening as magma intrusion. Thatcher suggests that at least part of the apparent stretching may have been caused by a magnitude 5.4 earthquake that struck Little Skull Mountain in 1992, just 8 kilometers from the eastern end of the GPS line. What Wernicke's team may have measured, he says, is the crust's readjustment to the strain released by the earthquake, which had built up over centuries or millennia. But Bennett says he calculates that such quake-driven effects are quite small, much smaller than the motions observed via GPS.

    Thatcher remains cautious, saying that the new work may be showing subtle movement that other studies missed—or errors or fleeting effects may be misleading everyone. Another 5 years of GPS surveying, he says, should suffice to decide whether recent history or the deep geologic past is the best guide to Yucca Mountain's future.


    No-New-Neurons Dogma Loses Ground

    1. Marcia Barinaga

    For many decades, both popular and scientific wisdom have held that adult brains can't make new neurons, so the ones we form when we're young have to last a lifetime. But last week, a research team from Princeton University, Rockefeller University in New York City, and the German Primate Center in Göttingen provided a challenge to this conventional dogma. In the 17 March issue of the Proceedings of the National Academy of Sciences (PNAS), they report that adult marmoset monkeys make new neurons in the hippocampus, a part of the brain associated with learning and memory.

    Previous work had shown that lower species, including birds and rodents, produce brain neurons throughout their lives. But some neuroscientists have hailed the new result as a potential breakthrough, because it is the first evidence that the same may be true for primates—perhaps even humans, says neuroscientist Ron McKay, of the National Institute of Neurological Disorders and Stroke. If further work confirms that the adult human brain can make new neurons, and if these cells join existing functional networks in the brain—both of which are in the realm of speculation at this point—it may open doors for enhancing neurogenesis, as new neuron formation is called, to repair brain damage from disease or trauma.

    Newborn neurons.

    BrdU marks the nuclei of 3-week-old neurons (arrows) in the dentate gyrus of an adult marmoset.

    E. GOULD

    Recent studies even suggest that stress reduction or an enriching experience can boost neurogenesis in some cases. In the current work, Elizabeth Gould from Princeton and her colleagues Bruce McEwen at Rockefeller University and Eberhard Fuchs at the German Primate Center found that stress decreases the rate of birth of new neurons in marmosets, while recent research on rodents by Fred Gage of the Salk Institute in La Jolla, California, and his colleagues has shown that an enriched environment can increase the neurogenesis rate. Together, the results “suggest that behavioral stimuli can regulate neuronal replacement rates,” says McKay. “That is what I find so exciting.”

    Other researchers in the field caution that it's a big leap from marmosets to humans, especially when one considers work that both the PNAS paper and a New York Times report on it failed to mention: a set of experiments published in the mid-1980s by neuroscientist Pasko Rakic and his colleagues at Yale that found no evidence of new neurons being born in the hippocampus of adult rhesus monkeys. Rhesus monkeys are more closely related to humans than marmosets are, and neuroscientists have interpreted those experiments as indicating that higher primates, and probably humans as well, lack the ability to generate new brain neurons.

    Gould and her colleagues chose to do their experiments on marmosets—a New World primate native to South America—because these animals are more readily available and less expensive than Old World primates like rhesus monkeys or chimpanzees, which are more closely related to humans. For their studies, the team injected adult male marmosets with bromodeoxyuridine (BrdU), which labels dividing cells. Two hours later, they sacrificed half of the animals and examined their brains for new cells that had picked up the BrdU. They found lots, in a part of the hippocampus called the dentate gyrus. Indeed, Gould calculated that “thousands of cells per day” were being born there. Three weeks later, when the team examined the brains of the remaining marmosets, they found that 80% of the BrdU-labeled cells looked like neurons and were making a neuron-specific protein. “A significant number of [the cells] survived,” says Gould, “and the majority of those … became neurons.”

    In a separate experiment, the researchers put adult male marmosets into the home cages of other adult males—a situation that is very stressful for the so-called “intruder” animals—and then injected the intruders with BrdU. The stress apparently decreased the number of new cells by 30%, suggesting it has a chilling effect on neuron replacement.

    Gould's group had shown last year that tree shrews, close relatives of primates, grow new neurons in their hippocampuses, and marmosets represent another step up the evolutionary ladder. To Gould, the combined discoveries of neurogenesis in the adult brains of rodents, tree shrews, and marmosets suggests that this is a common phenomenon, which she expects to occur in humans as well. But some in the field caution against making that leap from these data until there is an explanation for Rakic's negative result in rhesus monkeys. “The idea that they can extrapolate on to humans is blocked in part” by Rakic's results, says Salk's Gage.

    Gould counters that Rakic used a different—and less sensitive—method for detecting new neurons and thus might have missed their formation. Rakic acknowledges that his group could have overlooked a low level of neuron production but says “certainly we would not miss several thousand a day.”

    The issue will remain open, says Harvard neuroscientist Connie Cepko, until more tests on primates, including a reexamination of rhesus monkeys with the newer methods, are performed. “If one monkey species does it and one doesn't, we can't extrapolate to [humans],” she says, “but if all monkeys do it,” such an extrapolation would seem a safer bet. Some researchers aren't waiting to make their wagers: “I'll make a bet,” says McKay. “This is going to happen in humans. The question is under what circumstances, and what difference does it make?”

    Indeed, “one of the challenges for this field is to try to come up with some biological significance for this late-stage neurogenesis,” says Gage. That significance, says Stanford neuroscientist Susan McConnell, hinges on the question, “What is the actual result of putting new neurons into a circuit? Does it, for example, help you learn or remember better?” If it does, psychologists and neuroscientists will have a field day putting these results to practical use.


    Receptor Links Blood Vessels, Axons

    1. Wade Roush

    Veins gather blood and return it to the heart, Leonardo da Vinci mused in one of his notebooks, in the same way that rivers channel water back to the oceans. But a recent discovery suggests that da Vinci's analogy was more impressionistic than accurate. Rather than following random paths like water trickling downhill, it appears that blood vessels may instead use detailed chemical cues to navigate to prearranged addresses—much as nerves do.

    In the 20 March issue of Cell, a group led by biochemist Michael Klagsbrun at Children's Hospital in Boston reports that it has identified a new receptor molecule on the endothelial cells that line blood vessels. The receptor, called neuropilin, picks up a signal from VEGF, a protein made by tissues—including cancerous tumors—that need new blood vessels. VEGF was already known to stimulate blood vessel growth, or angiogenesis, but most researchers thought the actual paths of the vessels are as random as da Vinci's watercourses. The new receptor, however, turns out to be the same one that nerve axons use to detect Semaphorin III, a protein that helps steer axons to their proper destinations in the developing nervous system.

    Double duty.

    Neuropilin is present both on neurons, where it's involved in axonal guidance, and on endothelial cells, where it aids angiogenesis.


    “It's a very exciting discovery,” says Marc Tessier-Lavigne, a developmental biologist at the University of California, San Francisco, whose laboratory, along with one in Baltimore, identified the semaphorin receptor last year. The receptor's dual role, he says, raises the intriguing possibility that angiogenesis, far from being random, is in fact as highly scripted as axonal pathfinding, which is regulated by a whole menagerie of extracellular messengers in addition to Semaphorin III. He and Klagsbrun also note that if neuropilin allows VEGF to act on neurons and semaphorins to act on endothelial cells, this might indicate an unexpected level of coordination between the developing nervous and circulatory systems. The discovery, finally, could even provide researchers with a new target for drugs that might fight cancer by blocking tumor growth.

    Shay Soker, an instructor in Klagsbrun's lab and the Cell paper's lead author, cleared the way for the recent advance in experiments he performed 2 years ago. Using a biochemical technique called crosslinking, Soker found that VEGF forms two different kinds of complexes with molecules from the surfaces of endothelial cells. Because two VEGF receptors, designated KDR and Flt, had already been discovered, he assumed that one of these complexes contained VEGF and KDR and the other consisted of VEGF and Flt. But one of the complexes had a molecular weight too low to be either—implying that it contained an entirely new VEGF receptor.

    Soker, Klagsbrun, and co-workers Seiji Takeshima, Hua Quan Miao, and Gera Neufeld set out to clone this receptor. They first made a “library” of DNA clones representing the genes active in a line of human cells that lack both KDR and Flt yet are still able to bind VEGF. They then introduced small subsets of these DNAs into cells that can't bind VEGF and looked for cells that acquired that ability. These cells had to be carrying sections of the gene encoding the VEGF-binding segment of the new receptor. Those gene segments, the group discovered, were identical to sequences from the human gene encoding neuropilin. “It was a big surprise,” Klagsbrun says, “that we were dealing with a molecule that's well known in the developmental neuroscience field.”

    Neuropilin's role in angiogenesis seems to be different from its role in axonal guidance, however. When Semaphorin III binds to the receptor on the growing tips of neurons, it repels the cells, keeping them from getting off track. But on developing blood vessels, neuropilin seems to work in concert with KDR to stimulate vessel growth toward a VEGF source; cells to which the researchers added the genes encoding both receptors were four times more effective at binding VEGF than were cells carrying KDR alone.

    Although the researchers still have a lot to do to uncover neuropilin's exact role in angiogenesis, their work so far clearly shows, says Klagsbrun, that endothelial cells have more ways of sensing and responding to navigational signals than was previously supposed. Tessier-Lavigne adds, “I certainly think it's conceivable” that blood vessel growth follows patterns similar to those governing neuronal development. He says researchers will also want to know “how much cross talk there is” between developing nerves and blood vessels, a question Klagsbrun's group is already investigating.

    On the medical front, Klagsbrun has found that neuropilin is abundant on some kinds of cancer cells, suggesting that VEGF doesn't merely attract blood vessels but also feeds back on tumor cells themselves, perhaps helping them to stay alive. Developing a drug that blocks neuropilin-VEGF binding, therefore, might be a double whammy against tumors, depriving them of their survival factor while also deafening blood vessels to VEGF's siren song. And that's a benefit that Leonardo, a practical engineer as much as a perceptive artist, would have appreciated.


    For Island Lizards, History Repeats Itself

    1. Gretchen Vogel

    It is one of evolutionary biologists' favorite thought experiments: If one could start with similar organisms in similar environments, would evolution repeat itself to produce the same results?

    Some biologists say no. They think that even though organisms follow the rules of natural selection, historical accidents play a large role in their ultimate fate—who wins and who goes extinct. In this view, we live in the age of mammals in part because a comet or asteroid happened to wipe out the dinosaurs. But other scientists have held that natural selection is powerful enough to shape organisms to similar ends, no matter what the vagaries of history. Now a natural version of this experiment suggests that selection is stronger than chance—at least some of the time.

    On page 2115, evolutionary ecologist Jonathan Losos of Washington University in St. Louis and his colleagues report that anole lizards on four different islands independently evolved into strikingly similar creatures. Although examples of convergent evolution, such as wings on bats and birds, are well known, “what's remarkable here is the degree of similarity that has evolved on all four of the islands,” says Douglas Futuyma of the State University of New York, Stony Brook. Many evolutionary biologists welcome the finding. “People have been arguing this for a long time, and I think the case is finally solved,” says evolutionary ecologist Dolph Schluter at the University of British Columbia.

    Dozens of species of anole lizards thrive on the islands of the Greater Antilles—Cuba, Hispaniola, Jamaica, and Puerto Rico—and these relatively isolated island ecosystems offer a good place to test theories of evolution. “It's the natural equivalent of a replicated experiment,” says Futuyma.

    In the 1970s, evolutionary biologist Ernest Williams of Harvard University was the first to notice that lizards from different islands living in similar environments also look similar. Anoles that live in the tops of trees, for example, have large toe pads and short legs, while anoles that live on the ground have long, strong hindlegs. Williams divided the dozens of species into six “ecomorphs” and theorized that these forms had arisen independently on each island, but he had little genetic data to back his claim.

    Losos and his colleagues from Washington University, the National Museum of Natural History in Washington, D.C., and the Institute of Ecology and Systematics in Havana, Cuba, have now tested Williams's ideas. To establish that the ecomorphs are distinct groups, the team measured six characteristics that are linked to habitat, including mass, size of toe pads, and length of body, tail, and legs, in about a dozen lizards from each of 46 species. Williams's ecomorphs held up—each lizard species was indeed most like those living in an identical habitat on other islands.

    Next, the team analyzed mitochondrial DNA from 55 species to determine the relationships among the lizards. They found that members of the same ecomorphs on different islands are only distantly related, while species from the same island are closely related. They conclude that although the original lizard immigrants were likely different for each island, similar evolutionary pressures shaped them into similar ecomorphs. “There are certain ways a lizard can make a living on these islands,” Losos says. “In this case, the power of natural selection is so strong that it overwhelms any differences between the islands and what has gone on there before,” or what lizard began the process. In Cuba, for instance, so-called trunk-crown dwellers seem to have arrived first, while twig-dwellers were the first to inhabit Jamaica. “It's incredible,” says Schluter. “It's almost as though you start from different beginnings and end up in the same place.”

    But not everyone is completely convinced. While she praises the work, evolutionary biologist Joan Roughgarden* of Stanford University notes that in the Lesser Antilles, which stretch from Puerto Rico to Venezuela, the lizards have evolved several alternative ways of dividing their territory.

    Futuyma also points out that the natural experiment didn't produce exactly the same results on every island. Two ecomorphs are missing from Jamaica and one from Puerto Rico. Losos does not deny that chance still has a heavy hand. But, says Schluter, it seems that at least in some cases “history can repeat itself over and over.”


    Double Helix Doubles as Engineer

    1. Sunny Bains
    1. Sunny Bains is a science writer based in the San Francisco Bay area.

    A marriage of optics and electronics could produce a new generation of computers many times faster than today's. But like many unions, this one is threatened by some serious incompatibilities. Many of the best lasers, detectors, light modulators, and other optical devices are made from semiconductors such as gallium arsenide and indium phosphide, whereas conventional electronic devices are made of silicon. As a result, the two kinds of devices have to be made separately, then mated. Although integrating one or two devices is relatively easy, assembling hundreds, thousands, or millions into a single array would defeat conventional “pick-and-place” technology.

    Now, a team of researchers at the University of California, San Diego (UCSD), and Nanotronics Inc., also in San Diego, has come up with a novel way to create these hybrid devices. Like so much of the mating game, it involves DNA, which in this case serves as a selective glue for sticking the devices to the surface of the chip. Described at a meeting of the International Society for Optical Engineering early this year in San Jose, California, the work has intrigued experts in the field. Electrical engineer Joseph Talghader at the University of Minnesota, Minneapolis, for example, calls it “an exciting technique and one that merits a great deal of future work.”

    A strategy developed by Talghader and others was actually the starting point for the San Diego team, which is led by UCSD's Sadik Esener. In the earlier technique, known as fluidic self-assembly, the optical devices are fabricated as geometric shapes (“pegs”) that can then slot into similarly shaped “holes” etched in the silicon substrate. The pegs are suspended in a liquid and spilled out over the substrate, with luck sliding into the right hole and sticking there thanks to weak van der Waals forces.

    The San Diego researchers were looking for a way to help the right peg find its hole, and they settled on DNA. The chemical bases that make up DNA—cytosine, guanine, adenine, and thymine—will bind to each other only in particular pairings: C with G and A with T. Hence, a single strand made up of the bases ATTTGC will bind strongly with its complementary strand, TAAACG, and not with any other sequence. The researchers set out to exploit this selectivity by attaching short complementary strands of DNA to the pegs and substrate to help the devices find their correct positions.

    In their first experiment, the team coated a substrate with a particular short strand of DNA. They then covered parts of the substrate with a mask and exposed it to ultraviolet light. The light chemically altered the DNA in exposed areas so that it could no longer bind to complementary strands. The researchers then coated some microbeads—which acted as dummy devices—with strands of DNA complementary to those on the substrate. When a fluid carrying the coated beads was splashed over the substrate, the beads successfully bound only to those areas that had not been exposed to UV light. One drawback of the technique is that it worked only for small devices, several hundred micrometers across, that would flow easily and not block other devices.

    In a second experiment, designed to show that several varying kinds of “devices” could be deposited at once, the group used masks to deposit four different types of DNA strands onto a substrate and then attached complementary strands to four different fluorescent molecules. When the labeled molecules were splashed onto the substrate, the pattern of fluorescence showed that they had bound only to the appropriate regions of complementary DNA. In a real system, this would mean that four completely different types of devices could be attached to many selected sites on a chip.

    The researchers realize, however, that just providing the glue is not going to be enough. They are now looking for more active ways to guide the devices to their correct positions. One possibility is to add extra chemical groups to the DNA on the devices to give them an electric charge, then create electric fields on the substrate to attract the charged devices to “landing sites.” The team is also investigating other techniques, such as creating currents in the fluid that would sweep the tiny devices to the right places.

    An even bigger challenge will be creating an electrical connection between the devices and their host semiconductor. The team is looking at the possibility of putting the DNA glue on the top of devices and bonding them, upside down, onto a dummy substrate. Once all the devices are in position, the dummy could be flipped over and pressed down on the real substrate. The substrate might be coated with molten solder, which would add an electrical bond to the mass marriage.


    Weighing DNA for Fast Genetic Diagnosis

    1. Joseph Alper
    1. Joseph Alper is a free-lance writer in Boulder, Colorado

    The modern doctor's little black bag, already overflowing with high-tech diagnostic devices, may soon have to make room for another advance. To diagnose a disease, judge future risks, or design a treatment, doctors will one day want to know which disease-related genes a patient carries. And they will want this diagnostic verdict to be as fast and accurate as a cholesterol or blood chemistry test today. As Charles Cantor, director of Boston University's Center for Advanced Biotechnology, puts it: “You need a detection system that can identify the gene sequences that you are looking for with high specificity, quickly, and in large volumes. The best analytical tool for doing this,” he adds, “is mass spectrometry.”

    Borrowed from chemistry, this technology is a sharp departure from current methods, which identify a gene sequence by allowing it to bind to a matching probe, either on a gel or a chip. Instead, a mass spectrometer vaporizes the DNA and accelerates the molecules through a vacuum chamber with the help of an electric field. Tiny differences in the time it takes the DNA fragments to reach the detector reveal small differences in their mass, and hence their sequence.

    The basic technique used for biomolecules is one with an unwieldy name, matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry, but a harmonious acronym, MALDI-TOF. It is now a decade old, but recent improvements have made it a hot commodity among companies hoping to commercialize DNA analysis. “With today's technology, MALDI-TOF can analyze hundreds of DNA samples … in a matter of a few minutes,” says Daniel P. Little, who directs mass-spectrometry development at Sequenom Inc., a San Diego-based company hoping to be generating diagnostic products within 6 months.

    The standard way to distinguish different variants of a gene is to chop the DNA into fragments, separate them on a gel, and apply probes labeled with fluorescence or radioactivity, which bind to fragments with a particular sequence and light them up. But the process is slow and the gels can be hard to interpret. Newer techniques embed an array of different DNA probes on a single chip, allowing researchers to test for many gene variants at once. These so-called DNA chips can screen DNA quickly. But, as Cantor explains, the probes sometimes bind to sequences they don't completely match, which can limit the chips' accuracy.

    Mass spectrometry may combine the DNA chip's speed with exquisite accuracy. The technique has long offered chemists a fast way to sort small molecules that vaporize naturally or can be coaxed into a vapor with bursts of energy from a laser or ion beam. But vaporizing large biomolecules while keeping them intact once seemed impossible. A decade ago, however, Franz Hillenkamp and colleagues at Westfälische Wilhelms University in Münster, Germany, found a way to do so with proteins: Cocrystallize them with certain small molecules, collectively called matrices. When a nanosecond laser pulse vaporizes the matrix, the resulting puff of material gently lifts the ionized biomolecule as well.

    DNA was a tougher problem. But in 1993, Christopher Becker, then at SRI International in Palo Alto, California, and now at GeneTrace Systems in Menlo Park, California, found a simple matrix compound, 3-hydroxypicolinic acid, that worked with DNA sequences 20 to 25 bases long. By trial and error, MALDI practitioners have come up with several new matrices that work with DNA fragments as long as 100 bases.

    The latest MALDI-TOF machines allow the cloud of matrix molecules to dissipate before applying an electric field. The field accelerates the charged DNA fragments toward a detector, and the differences in time of flight can reveal mass differences as small as 0.03%. If the DNA sequences from a gene have the same length—as they do if they have been produced by the polymerase chain reaction—any departure from the mass of the normal sequence reflects a mutation that has deleted or added bases or substituted others that have a different mass. “The results are an absolute indicator of the presence or absence of specific DNA sequences,” says Sequenom's Little. MALDI-TOF can distinguish gene variants that differ by as little as a single base pair, and it can also analyze microsatellites—stretches of two-, three-, or four-nucleotide repeats often used as markers for locating disease-causing genes.

    Besides offering unmatched precision, MALDI-TOF is inherently fast. The DNA forms a vapor and flies to the detector in fractions of a second; even repeating the process several times with the same sample to boost the sensitivity takes as little as 2 seconds. By preparing the samples in a grid and having the laser scan each spot in turn, a MALDI-TOF instrument can analyze 100 samples or more in a matter of minutes.

    The combination of speed and accuracy could give the technique a role in genome sequencing as well as diagnosis. Standard, Sanger-type DNA sequencing generates many partial copies of a DNA sequence, each one starting at one end of the sequence and ending with a different one of the constituent bases. To determine the original sequence, biologists need to know the final base on each partial copy, together with the copy's length. Doing so now requires reading hundreds of bands on gels. But by sending the mixture through a mass spectrometer, biologists could quickly read off the fragments' lengths and—from the mass differences between successive fragments—the final base on each one. Investigators at both GeneTrace and Sequenom have published sequences determined with MALDI-TOF, the latest one, from Sequenom, appearing in the April Nature Biotechnology.

    For practical gene sequencing, however, MALDI-TOF would have to work with DNA fragments much longer than the current 100-base capacity. Becker reportedly has discovered a new proprietary matrix that he expects will extend MALDI-TOF's reach to 1000-base sequences. “If you can really do upward of 1000 bases using this technique, and if it is indeed faster and cheaper, then this would be a big breakthrough for high-throughput sequencing,” says Jeffrey Polish, who works in Mark Johnston's sequencing laboratory at Washington University in St. Louis.

    In the meantime, the technology has no shortage of applications. Sequenom has shown, for example, that it can discriminate among 30 of the mutations that cause cystic fibrosis and pick up polymorphisms in the apolipoprotein E gene, which have been linked to familial hyperlipidemias, heart disease, and Alzheimer's disease. GeneTrace has developed a mass spectrometry-based system that can analyze which genes are being expressed in cells by identifying expressed sequence tags, short stretches of DNA copied from the messenger RNA made by active genes. Knowing which genes are active in a tissue can help pharmaceutical companies determine which ones are good drug targets.

    With MALDI-TOF instruments running about $125,000 each—less than a standard clinical chemistry analyzer—these systems may also end up in large diagnostic labs. “Diagnostics at the level of the gene is something that we know is valuable, but is difficult, slow, and expensive today,” says David Cooper, chief scientific officer at Nichols Institute Reference Laboratories, a division of Quest Diagnostics, one of the big 3 national reference laboratories. MALDI-TOF, he says, could be just the right medicine.


    Making a Bigger Chill With Magnets

    1. James Glanz

    LOS ANGELES—Refrigerator magnets are best known for holding shopping lists and old postcards onto refrigerator doors. But in a few years, much more powerful magnets could be the key to keeping food cold in so-called magnetocaloric refrigerators, which would be more energy efficient and less polluting than standard models. Now a new class of magnetocaloric materials, announced here last week at a meeting of the American Physical Society, could make these magnetic refrigerators more practical and versatile.

    The magnetocaloric effect works when strong magnetic fields align quantum-mechanical “spins” of electrons within atoms. This transition reduces one aspect of the randomness, or entropy, of the atoms. But according to laws of thermodynamics, some other aspect of randomness has to increase in compensation, so the atoms increase the randomness of their velocities—vibrating and heating up. Once this heat is carried away by a coolant such as water, the field is removed and the effect works in reverse, chilling the material and cooling a refrigerator. To date, the peak performance has been with the element gadolinium.

    By adding various amounts of silicon and germanium to gadolinium's crystal lattice, Vitalij Pecharsky and Karl Gschneidner of the Ames Laboratory at Iowa State University discovered a new class of materials that can chill two to six times further in a single magnetic cycle, meaning that the refrigerators could operate with weaker magnetic fields or less material. Depending on the germanium-to-silicon ratio, the new materials also operate from about room temperature all the way down to −253 degrees Celsius. The cold end of the range would allow magnetocaloric freezers to liquefy hydrogen or natural gas for use in clean-burning power plants or future automobiles.

    To come up with the new compounds, the team followed up on hints that magnetocaloric materials containing gadolinium and either silicon or germanium—but not both—prefer a different range of temperatures than gadolinium alone. “We're not trying to come up with exotic new compounds out of the pure blue sky,” says Gschneidner. The surprise, he says, was that the magnetocaloric effect turned out to be far larger when both germanium and silicon were added to the material.

    “These new materials give you a lot more flexibility in designing magnetocaloric [refrigerators],” says Carl Zimm, a senior scientist in magnetic refrigeration at Astronautics Corporation of America in Madison, Wisconsin. The team is still working on making enough of the material to try it out in Zimm's prototype gadolinium-based refrigerator, which has been running for about a year. The test should take place “within a couple of months,” says Gschneidner.

Stay Connected to Science