News this Week

Science  25 Apr 1997:
Vol. 276, Issue 5312, pp. 528
  1. Cell Biology

    The Telomerase Picture Fills In

    1. Marcia Barinaga

    Researchers working with protozoans and yeast have identified a protein that appears to be the long-sought catalytic component of telomerase, the enzyme that synthesizes the chromosome ends

    When Vicki Lundblad and Tom Cech sat down to talk about their common interest last August at a conference in Hawaii, neither realized how fruitful that chat would be. They both were in hot pursuit of one of the most elusive prizes in cell biology: the key protein in the enzyme, called telomerase, that rebuilds the chromosome ends after each cell division and may play a role in both cancer and aging.

    Cech's and Lundblad's labs had each found telomerase proteins in one of two very divergent organisms—the ciliated protozoan Euplotes aediculatus and the brewer's yeast Saccharomyces cerevisiae. But neither knew whether they had found the crucial component of the complex enzyme. As a result of their talk, however, they swapped their protein sequences and quickly discovered that Cech's team at the University of Colorado, Boulder, and Lundblad's at Baylor College of Medicine in Houston had zeroed in on the same protein.

    Within months, the two teams had used a mix of biochemistry and yeast genetics to show that the protein has features expected of the telomerase's catalytic component, and that altering those features abolishes the enzyme's activity. Theirs is not the first candidate for the catalytic role. But the evidence, reported on page 561 of this issue, constitutes “the tightest case for having a catalytic activity that is part of the telomerase,” says Dan Gottschling, who studies telomeres at the Fred Hutchinson Cancer Research Center in Seattle.

    If the new protein is indeed the catalytic component of telomerase, the discovery will provide a boost to researchers trying to understand how the enzyme does its crucial job. Without telomerase, the chromosomes would shorten with every cell division, eventually disrupting the genes. Understanding telomerase activity and how it falters might lead to insights into aging, which some researchers have linked to telomere loss. The enzyme is also a potential target for cancer therapy, as its activity is high in many types of cancer cells, the continuous division of which may demand good telomere maintenance.

    The protein has been the biggest mystery in telomerase since Elizabeth Blackburn and then-graduate student Carol Greider, at the University of California (UC), Berkeley, showed in 1987 that the enzyme contains both protein and RNA. Blackburn's team, which moved to UC San Francisco (UCSF) during the course of the work, later showed that the RNA acts as a template from which the telomerase makes the DNA it adds to the chromosome tip. “That would make the enzyme officially a reverse transcriptase,” notes Titia de Lange, who studies telomeres at Rockefeller University in New York City. But while the RNA part of telomerase could be fished out of cells by searching for RNAs with sequences complementary to those of telomeric DNA, the search for the protein responsible for that reverse transcriptase (RT) activity proved difficult, mainly because the enzyme is in such scarce supply in most cells.

    To get around that problem, Lundblad took an approach that did not require isolating telomerase proteins directly. Three years ago, her team began a search for yeast mutants in which telomeres shorten over time—a defect that could result from a faulty telomerase. They found mutations in four genes, including one Lundblad had identified while a postdoc with Jack Szostak at Harvard Medical School. Yeast cells with mutations in any of the four genes, known as EST genes for ever-shorter telomeres, were indistinguishable from each other and from cells mutant in the telomerase RNA. That suggested the EST genes encode proteins necessary for telomerase function. “The problem came when we cloned the genes,” says Lundblad. They bore no resemblance to known genes in the database, and so there were no clues as to what their biochemical function might be.

    But a solution to this dilemma was coming—along a very different path. About the same time that Lundblad's team began its screen, Joachim Lingner, a postdoc in Cech's lab, began to purify telomerase from Euplotes, a protozoan they chose because it has a specialized nucleus called a macronucleus that contains 40 million tiny chromosomes. And because all those chromosomes have telomeres, Euplotes needs buckets of telomerase to maintain its chromosome ends.

    Lingner devised a way to snag the Euplotes telomerase RNA and purify the whole enzyme along with it. He found two proteins associated with the RNA, and fractions containing the RNA and proteins had telomerase activity, measured by their ability to add DNA to telomeres in the test tube. He couldn't isolate enough of the proteins to determine their amino acid sequences by the usual methods, however, so he and Cech teamed up with Matthias Mann at the European Molecular Biology Laboratory in Heidelberg, Germany.

    Mann had developed a way to sequence tiny amounts of proteins (as little as 10−15 mole) by digesting them with enzymes, squirting them through a highly charged needle, and separating the fragments by mass. Andrej Shevchenko in Mann's lab got partial sequences from the proteins, and Lingner used the sequences to make DNA probes that enabled him to clone the corresponding genes.

    Lingner searched the DNA databases for genes that resembled the proteins, and found a yeast gene that matched the larger protein, p123. No information was given about the identity of the matching yeast gene beyond its sequence. But Lingner, while poring over the p123 sequence, found scattered patches that match sequence motifs found in the active site of known reverse transcriptase enzymes. When he aligned the p123 and yeast-protein sequences, he found the RT motifs in the yeast protein, also. The parts the proteins had in common with RTs were small, and by themselves would not constitute a convincing match, but they were precisely the sequences that make up the active site of known RT enzymes.

    Lingner decided to knock out the yeast gene to see if it affected telomerase activity. But Cech and Lundblad's conversation in Hawaii spared him the work. Lundblad offered her unpublished EST gene sequences for comparison with those Lingner had found, and one of those genes, EST2, turned out to be the very gene that Lingner had pulled out of the yeast database. “That was very exciting,” says Cech, because of the Lundblad group's genetic evidence that EST2 is needed for telomere maintenance. It was looking like p123 and Est2 could be the catalytic components of their respective telomerases.

    The yeast connection was a lucky break for his lab, says Cech, because it enabled them to test that hypothesis genetically. “You can do genetics so easily in yeast,” he notes, “and in Euplotes you can't do it at all.” The teams collaborated to mutate sites in Est2 that correspond to the RT active site. They found that the telomeres of yeast cells containing the mutant enzyme shortened just as they did in cells missing the whole EST2 gene.

    Lingner subsequently showed that telomerase partially purified from those mutant yeast cells could no longer elongate telomeres. From that, “it was a reasonable conclusion,” says Lundblad, “that [the proteins] were the catalytic subunits of their respective enzymes.” Many in the field agree. “Looking at the whole package, I find it very convincing,” says Jef Boeke, who studies reverse transcriptases at Johns Hopkins University. “They mutated the [key RT] residues, and they killed an in vitro activity. That evidence is very strong.”

    The evidence also suggests that Est2 and p123 represent a new class of RT that is most closely related not to RTs from retroviruses like HIV, but to RTs coded for by transposable elements (segments of DNA that can move around in the genome) found in many cells. In that case, drugs that inhibit the HIV enzyme may not work on telomerase, Boeke says, so researchers wanting to develop antitelomerase drugs for possible anticancer applications may have to start from scratch.

    Despite the evidence that Est2 and p123 are telomerase catalysts, they still have rivals for that role: p95 and p80, telomerase-associated proteins purified and cloned 2 years ago from the ciliated protozoan Tetrahymena thermophilia by Greider, with Kathleen Collins and Lea Harrington, who were then postdocs in Greider's lab at Cold Spring Harbor Laboratory on Long Island. Collins proposed that p95 may be the catalytic component, based on resemblances it shows to enzymes that synthesize RNA or DNA. And in the past 3 months, Harrington's team at the University of Toronto and Fuyuki Ishikawa's at the Tokyo Institute of Technology have reported finding p80-like proteins in human, mouse, and rat telomerases.

    But, so far, no one has shown either p95 or p80 to be essential for telomerase activity, and the current discovery casts more doubt on the role of those proteins. “The most straightforward scenario,” says UCSF's Blackburn, is that the Tetrahymena and mammalian telomerases also have a p123-like protein as their catalytic unit, and it has somehow been missed. In that case, p80 and p95 may have an accessory role in the enzyme. “Given this [result], one should look very hard for RT motifs in all species of telomerases,” Blackburn says.

    Or it may be that either p80 or p95 will prove to be the catalytic part of Tetrahymena telomerase. But that would mean that the two ciliates, Tetrahymena and Euplotes, have evolved different enzymes to do the job—an idea that Gottschling describes as “sort of amazing, evolutionarily.” Most researchers say they are reserving judgment about these possibilities until some of the remaining gaps in the story are filled. However, most also agree with telomere researcher David Shore of the University of Geneva that “the burden of proof is now on those with p80 and p95 in their hands” to show that those proteins are essential parts of telomerase.

    Even if the new results do pinpoint the crucial actor in telomerase, the enzyme likely contains many other proteins, all with some role in its function, that vary from organism to organism. The picture looks confusing now, Greider says, because “we are interpolating between several different organisms. We need to solve the problem completely in each organism” before the true nature of telomerase is revealed.

  2. Gamma-Ray Bursts

    Visible 'Source' Teases Observers

    1. James Glanz

    Just when astronomers thought they might be solving one of their longest running mysteries, the story has taken another dizzying twist. The question is whether the flashes of gamma rays called gamma-ray bursts (GRBs) originate in or near our galaxy or billions of light-years away, at cosmological distances, which would make them the brightest outpourings in the universe.

    Astronomers thought they were on the verge of an answer when the Italian-Dutch satellite Beppo-SAX saw a fading source of x-rays that seemed to be the afterglow of a GRB the satellite had detected on 28 February. Because x-ray detectors have much better spatial resolution than those for gamma rays, that helped pin down the burst's position for further observation. Hopes shot even higher when ground-based telescopes aimed at the spot then fished out both a point of light and a faint fuzzy patch next to it—possibly the GRB source and its host galaxy in the distant universe (Science, 21 March, p. 1738). The cosmological alternative seemed poised to carry the day. But then the orbiting Hubble Space Telescope (HST) got into the act.

    The latest results of its scrutiny of the proposed source, reported on the Internet last week in International Astronomical Union (IAU) circulars, have thrown the debate wide open again. One group claimed that the pointlike burst source, if that's what it is, is moving noticeably across the sky and might be a nearby object. Another group—looking at the same data—saw nothing of the kind. Says Chryssa Kouveliotou of NASA's Marshall Space Flight Center in Huntsville, Alabama: “I'm more confused than anything.”

    This new chapter in what Princeton University's Bohdan Paczynski calls “the wonderful story of [gamma-ray burst] 970228” opened earlier this month when Kailash Sahu, Mario Livio, Larry Petro, and F. Duccio Macchetto of the Space Telescope Science Institute (STScI) in Baltimore, along with several collaborators including Kouveliotou, published a circular reporting HST observations of the optical counterpart 26 and 38 days after the burst. The observations confirmed the fading point source and the adjacent fuzzy patch. By prior agreement, the group immediately released its raw HST data to the entire astronomical community.

    Then, on 17 April, a separate group, including Patrizia Caraveo at the Istituto di Fisica Cosmica (IFC) in Milan, Italy, and several collaborators, posted another circular reporting their own analysis of the data. They had found something startling: The point source was moving across the sky. The angular motion was so quick, says one of the collaborators, Marco Tavani of IFC and Columbia University, that the object might have to be within a few hundred light-years of Earth, much closer than even the proponents of a galactic origin for GRBs have been suggesting recently. The fuzzy object could then be anything from a transient cloud of gas associated with the burst to a background galaxy, aligned by chance with the pointlike object.

    But the Sahu team's own analysis, completed after their first circular, shows that to within the error bars, the point source is stationary. “We cannot reproduce what Caraveo says in the IAU circular in spite of our best efforts,” Sahu told Science. Tavani suggests that the discrepancy is due to his own team's superior software for extracting apparent motion from the data, but the issue remains unresolved. So does the nature of the fuzzy patch. As Science went to press, several astronomers reported in an IAU circular that a so-far-uncorroborated measurement at the Keck 10-meter telescope indicated that the “host galaxy” might have faded, which would indicate that it isn't a galaxy at all.

    Resolving these issues, says Paczynski, is likely to require a third HST observation when the point source, now drawing close to the sun, reemerges from its glow in a few months. If the object is still visible, the next glimpse of it should settle the question of whether it is moving. And finding and studying optical counterparts for several more GRBs could finally solve the mystery—providing they yield their secrets more readily than this one has. In the meantime, says Peter Mészáros, a theorist at Pennsylvania State University, “It has been a bit of a roller coaster.”

  3. Physics

    Doubts Greet Claim of Cosmic Axis

    1. James Glanz

    Interesting—but probably wrong. That distills the reaction of most physicists and radio astronomers to an extraordinary claim that space itself might have an overall orientation. Aired last week amid a barrage of university press releases that culminated in front-page newspaper headlines, the claim was based on unaccountable differences in the way radio waves seem to propagate to Earth from galaxies at different spots in the sky. But other researchers say that the evidence behind the claim has serious shortcomings.

    Critics have not identified a single fatal error that could explain away the result, which Borge Nodland of the University of Rochester in New York and John Ralston of the University of Kansas published in the 21 April issue of Physical Review Letters. But astronomers say the two physicists relied on old and incomplete data, omitting perhaps 99% of recent observations of radio galaxies. Their analysis also depends on outdated ideas about the nature of the signals from those galaxies, says Kenneth Chambers of the University of Hawaii. Add the likelihood that the conclusions would—just for starters—overturn Einstein's theory of relativity, and other researchers are deeply skeptical. Michael Turner, a physicist at the University of Chicago, had one of the kinder assessments. “Extraordinary claims demand extraordinary evidence,” he said. “The evidence presented does not yet meet this standard.”

    Ralston says he is not surprised by the criticism: “It's always going to be there as long as somebody tries something new.” He and Nodland analyzed published observations of 160 radio-emitting galaxies scattered around the sky. The electric field in radiation from such galaxies often arrives at Earth with a distinct polarization, or orientation, like the silhouette of a baton held against the night sky. Nodland says he and Ralston thought data on these polarizations could provide a test of whether “certain [arrival] directions are special compared to other directions.”

    Like a baton hurled into the air, the polarization of radio waves rotates as they travel through space. Part of the rotation can be explained by the so-called Faraday effect, which results when the waves interact with charged particles and magnetic fields in space. Because the effect varies with frequency, Nodland and Ralston tried to estimate the Faraday rotation between Earth and any given radio galaxy by comparing the polarization of the signals at different frequencies.

    They then tried to estimate the signals' original polarization by assuming that each galaxy emits radio waves polarized at a specific angle relative to observed structures, like the energetic jets of plasma seen emanating from many radio galaxies. They subtracted the Faraday rotation from the polarization observed at Earth, then compared the result with their estimate of the original angle. The procedure revealed a residual twist—one that varied with the distance of the galaxy and its direction. The twist was greatest for galaxies lying near a single axis in the sky, running roughly in the direction of the constellation Sextans.

    That conclusion ended up making headlines last week after unrelenting promotion by the two universities' press offices. Its consequences would indeed be momentous. A preferred axis in space would violate physicists' cherished assumption that physical laws are the same everywhere in the universe, and it could revise cosmologists' picture of the big bang.

    But it also conflicts with the best evidence cosmologists have that the universe really is isotropic, or identical in all directions: the homogeneity across the sky of the microwave background radiation generated in the big bang. And the conclusion is weakened by problems in the selection and treatment of data, say other researchers. For one thing, the bulk of Nodland and Ralston's data set consists of observations from before about 1980, when the Very Large Array (VLA) near Socorro, New Mexico, came on line and began making far better observations of radio galaxies.

    For another, the assumption that each galaxy emits radio signals at a single, predictable polarization is decades out of date, says Chambers. Measurements at the VLA and elsewhere have shown that the emitted polarization varies widely over a single galaxy. Also, large and incalculable amounts of Faraday rotation often occur in the plasmas and magnetic fields near the galaxies themselves.

    Responds Ralston, “People have sent me e-mail saying there are newer data. Of course, there are newer data on radio galaxies in general,” but he says he didn't find any other observations that had all the features needed for his analysis. And he says the very fact that the analysis reveals a systematic variation in the twist shows the signals could not have been completely scrambled at the source.

    But most researchers think the effect would simply vanish in a larger, more modern data set. “I wouldn't look at this and say there's no chance it could be right,” says Ruth Daly, a physicist at Princeton University. “But it's not clear whether [the analysis is] telling us something interesting about the universe or about problems in the data.”

  4. Cancer Research

    Possible Function Found for Breast Cancer Genes

    1. Jean Marx

    Two of the most mysterious actors in cancer genetics may finally be sharing their secrets. These are the breast cancer susceptibility genes known as BRCA1 and BRCA2. Since their discoveries in 1994 and 1995, researchers have confirmed that the genes are potent agents of disease: Up to 80% of women who inherit mutated forms of either one will develop breast cancer in their lifetime, usually at a relatively early age, and women with BRCA1 mutations have a high risk of developing ovarian cancer as well.

    But researchers have been stymied in their efforts to find out just what the proteins made by these genes normally do—and why mutations in the genes have such serious consequences. One problem they have faced is that the proteins don't resemble anything in existing databases. “It's been really fascinating, but it's been frustrating, too,” says Andrew Futreal of Duke University School of Medicine, a co-discoverer of both BRCAgenes. Now, the veil of mystery may have begun to lift.

    In a paper that appears in the 24 April issue of Nature, a team lead by Allan Bradley of Baylor College of Medicine in Houston and Paul Hasty of Lexicon Genetics Inc. in The Woodlands, Texas, reports evidence indicating that the protein made by BRCA2 plays a critical role in enabling cells to repair their DNA when it is damaged. The group finds, for example, that BRCA2 binds to a known repair protein called RAD51. This result dovetails with work, reported by David Livingston of Harvard's Dana-Farber Cancer Institute and his colleagues just 3 months ago, indicating that BRCA1 also associates with RAD51—possibly putting both genes in the same DNA-repair pathway. What's more, the Texas team showed that embryonic mouse cells in which the murine version of BRCA2 has been inactivated can't recover from radiation damage. “This makes the idea that the BRCA genes are DNA-repair genes more believable,” says cancer gene expert Bert Vogelstein of Johns Hopkins University School of Medicine.

    If indeed they are, the results may help explain how BRCA mutations cause cancer. And the proposed mechanism is different from what most researchers expected, as Vogelstein and his Johns Hopkins colleague Kenneth Kinzler point out in a News and Views article also in the 24 April issue of Nature. The genes were originally considered to be classical tumor suppressors, which normally hold cell growth in check and which, if inactivated, can lead directly to cancer. But the new work suggests the mutations act indirectly, by disrupting DNA repair and allowing cells to accumulate mutations, including those that foster cancer development.

    The findings may also have therapeutic implications for women with BRCA mutations, who account for only few percent of total breast cancers but constitute a large number of patients, given that there are about 180,000 new cases of breast cancer every year in the United States alone. “If mouse cells depleted of BRCA2 are more sensitive to ionizing radiation than normal cells,” says Vogelstein, “it's a reasonable extrapolation” that breast cancers in which the gene has been inactivated may be especially good candidates for radiation therapy. The DNA-repair link may not be the full story of how BRCA mutations lead to cancer, however, as other recent evidence—some of it presented in a second Nature paper this week—points to additional functions for the BRCA proteins in regulating gene activity.

    In the absence of structural clues about BRCA2's function, Bradley and his team turned to a strategy researchers often rely on when they are trying to find out what a newly discovered gene does: making mouse strains in which the murine gene has been inactivated, or “knocked out,” and studying the resulting defects in the animals. When Bradley's group at Baylor tried this strategy on BRCA2, however, they found that all the embryos in which both copies of the gene had been inactivated died early in development. “They block around day 6.5 of embryogenesis—about when the gene comes on,” Bradley says. And other work suggested that inactivating the gene had an unexpected effect for a suspected tumor suppressor—halting rather than increasing cell division. Bradley's team found that when they tried to inactivate BRCA2 in cultured mouse embryonic stem cells, the cells simply wouldn't proliferate.

    The connection to RAD51, which came from the Lexicon group, provided a clue to what might be happening. The company's main interest at the time was this DNA-repair protein. To find out more about how RAD51 acts, Lexicon's Hasty was using a method known as the yeast two-hybrid screen to find proteins that interact with it in the cell. The screen consists of yeast cells engineered to express several cloned genes, one of which makes a protein—RAD51 in this case—that serves as “bait” for pulling out any of the other gene products it might interact with. In Lexicon's screen, the RAD51 bait fished out BRCA2. At that point, recalls Arthur Sands of Lexicon, they contemplated knocking out BRCA2 themselves, but on hearing of Bradley's work, joined forces with him instead.

    BRCA2's interaction with RAD51 suggested that it might be involved in DNA repair, and the Texas team went on to gather two additional lines of evidence in support of that idea. They found that the two genes have almost identical expression patterns in the tissues of the embryonic mouse—what you would expect, Bradley says, “if the interaction is real.” And perhaps most convincing, when the researchers exposed 3.5-day mouse embryos to ionizing radiation, they found that the radiation had little effect on embryos with either one or two functional copies of BRCA2—but it totally destroyed the embryonic cell mass of those in which both copies had been inactivated.

    To BRCA1 researchers, some of the new findings have a familiar ring. Other researchers had previously found that knocking out BRCA1 also leads to death early in embryogenesis. And work reported in the 24 January issue of Cell by the Livingston team points to an interaction between BRCA1 and RAD51. They found, for example, that the two proteins are located together in the cell nucleus in both ordinary cells and in cells undergoing meiosis, the specialized type of division that gives rise to the germ cells. Conceivably, then, mutations in either gene could lead to cancer by allowing cells to accumulate potentially dangerous mutations.

    A breakdown in DNA repair could also help explain the apparent antiproliferative effect of knocking the BRCA genes out in embryos. To keep damaged DNA from being transmitted to daughter cells, mechanisms called checkpoints either halt cell division so that the damage can be repaired before division occurs, or even kill the damaged cells. If the DNA is not repaired because an essential protein is missing, the cells might never pass the checkpoint, and so either fail to divide or simply die.

    That raises the question of how cells in the adult organism can keep proliferating—and form breast or ovarian tumors—when BRCA1 or BRCA2 is inactivated by mutation. One possibility, Bradley suggests, is that the checkpoint controls are much tighter in the embryo than the adult. Livingston proposes another: that checkpoint genes, too, have to get knocked out before cancer can develop.

    Causing defective DNA repair may not be the only way in which BRCA mutations lead to cancer, however. Both proteins are very large—BRCA1 contains 1863 amino acids and BRCA2 has 3418—and they may well have activities other than DNA repair. In the second Nature paper, Tony Kouzarides of the Wellcome/CRC Institute in Cambridge, U.K., and his colleagues present evidence that BRCA2 can activate gene transcription. They found that when they linked a particular region of BRCA2 to a known DNA binding sequence, it activated transcription of a so-called reporter gene in yeast. What's more, one BRCA2 mutation found in families with inherited breast cancer abolished the activity—an indication that its loss might be involved in development of the cancers. Other researchers have made similar observations with BRCA1. Still, the test systems used for all this work were artificial, and the results need to be confirmed—say, by identification of genes that the BRCA proteins activate normally.

    But even though the understanding of how BRCA1 and BRCA2 lead to cancer is tentative and incomplete, researchers feel that after years of frustration, they are finally making headway. Says Duke's Futreal, “Hopefully, we are moving toward [finding] a role for these things. It certainly looks like a trend.”

  5. Mathematics

    New Test Sizes Up Randomness

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

    Finding a random sequence of numbers is as easy as pi—or is it? Mathematicians often depend on irrational numbers likeπ, e (the base of natural logarithms), and √2 to give them an unpredictable stream of digits. But a paper in last week's Proceedings of the National Academy of Sciences is upsetting the conventional wisdom about randomness by showing that some of these numbers are far more predictable than expected. The finding is an early result of a new test of randomness that is also raising concerns in other fields where random-looking sequences crop up, such as cryptography. Ultimately, though, the new test could put those fields on firmer ground.

    Randomness has been hard to quantify. Any mathematician could tell you that 01101100 is more random than 01010101, but none could tell you just how much more random. Then, two researchers—Steve Pincus, a free-lance mathematician based in Guilford, Connecticut, and Burton Singer, a mathematician and demographer at Princeton University—created a method for measuring a sequence's “entropy,” or disorder. “[Their method] is one of those tools that makes you say, ‘Hey, that's a good one!’ and you put it in your tool kit,” says Max Woodbury, a mathematician at Duke University.

    Pincus and Singer built on the observation that all possible digits are represented about equally in a perfectly random stream of numbers. For example, the binary sequences 01101100 and 01010101—each with four 1s and four 0s—pass this test. But the researchers also noted that when the digits are taken two at a time, a random sequence should have an equal number of all possible pairs: 00, 01, 10, and 11, in this case. The sequence 01010101 fails this test miserably; there are no 00s or 11s at all. The same reasoning can be extended to larger groups of digits, taking them three at a time, four at a time, and so on. By comparing groups of digits to the expected frequency of those groups, Pincus and Singer come up with the “approximate entropy” (ApEn) of the sequence—a measure of its randomness.

    Pincus and Rudolf Kalman, a mathematician at the Swiss Federal Institute of Technology in Zurich, have now applied this tool to calculate the ApEn of various irrational numbers. Some, like √3 and √2, are “algebraic” numbers: They are the solution to a polynomial with a finite number of terms. Others are “transcendental,” or nonalgebraic, numbers like π and e. Because algebraic numbers are in a sense simpler than transcendental numbers, Pincus—like most other mathematicians—expected that when written out in decimal form, they would be less random than the transcendentals. He was wrong.

    “π is the most irregular,” says Pincus. “But I was very surprised that e was not next in line.” In fact, √2, an algebraic number, was more random than the transcendental number e. Mathematicians are still scratching their heads over this. “It's an interesting open question if the transcendental and algebraic numbers are mixed together” in order of randomness, says Kenneth Wachter, a mathematical statistician at the University of California, Berkeley.

    Pincus and Singer think other researchers should be taking note of this new tool, which they have incorporated into a computer algorithm. Cryptographers often try to make messages look like random sequences by adding a sequence of binary digits that is nearly random—preserving just enough order for the message to be retrieved. Given enough data, ApEn could tell the difference, distinguishing encoded messages from random noise. “Theoretically, you can bust them all,” says Pincus.

    Experiment designers could exploit ApEn as well, says Singer. In scientific experiments such as drug trials, researchers randomize the test subjects to avoid bias. But randomizing by coin toss or luck of the draw can occasionally produce an orderly pattern—with all the women assigned to the control group and all the men to the study group, to take an extreme example. ApEn, however, gives researchers an objective yardstick of randomness, so they can decide when the draw is too orderly and redo it. “[ApEn] allows you to increase the power of testing,” says Woodbury.

    ApEn may also provide a quick and easy way to screen data for randomness. Geriatricians and endocrinologists at two veterans' hospitals in Virginia, for example, sent Pincus the data from a hormone-sampling experiment. “We looked at the degree of disorderliness of the secretion of testosterone and luteinizing hormone in men,” says Thomas Mulligan, a geriatrician at the Hunter Holmes McGuire Veterans Affairs Medical Center in Richmond. Thanks to ApEn, Mulligan and his team found—and quantified—an effect of aging. “In older men, the disorderliness is markedly greater than in younger men,” he says.

    Pincus expects that his randomness test will uncover many more puzzles. “If I can bring nothing else to the party,” he says, “I want to ask a different set of questions.”

  6. Medical Research

    New Vaccines May Ward Off Urinary Tract Infections

    1. Robert F. Service

    What's almost as common as a cold, but, for most sufferers, far more uncomfortable? Urinary tract infections, or UTIs. Caused primarily by Escherichia coli, they send 1.5 million people (mostly women) to the hospital each year in the United States alone, and 7 million more to their doctors. In most cases, the infections are not life threatening: Standard antibiotics usually offer quick relief. But UTIs can recur frequently, and, when untreated, cause kidney damage and even death. A vaccine could reduce this toll, but until now, says Harry Mobley, a microbiologist at the University of Maryland School of Medicine in Baltimore, there has been “little successful work in UTI vaccines.”

    Bug blocker.

    A vaccine under development prevents adhesion proteins at the tips of spaghettilike pili on UTI-causing bacteria (above) from latching onto host cells.

    S. Hultgren/Washington Univ.

    On page 607 of this issue, however, researchers at MedImmune, a Maryland-based biotech company, and at Washington University in St. Louis report that they have developed a genetically engineered, injectable vaccine that prevents UTIs in mice. Meanwhile, researchers at the University of Wisconsin, Madison, are nearing completion of midstage human clinical trials of a vaccine delivered via a vaginal suppository that seems to offer at least short-lived protection. Mobley calls the new studies “very exciting.” Not only do they hold out the hope of reducing the number of UTI cases, he says, but the strategy followed by the MedImmune-Washington University group—targeting a single protein that enables the bacteria to latch onto their target cells—could prove to be valuable against other infections.

    These aren't the first efforts to combat UTI-causing microbes. For instance, one injectable cocktail of killed UTI-causing bugs, called Urovac, has been available in Europe since the late 1980s and provides short-lived protection. But natural toxins in the organisms often trigger painful inflammation around the injection site, among other side effects.

    To minimize inflammation, the Wisconsin group—led by urologist David Uehling—uses the same concoction of killed organisms as in the Urovac vaccine, but relies on another delivery method: a vaginal suppository. The researchers hoped that by allowing killed microbes to diffuse through the entire vaginal tract instead of injecting them into one small part of a muscle, a suppository would avert inflammation. They reasoned that the killed organisms would still trigger the production of a class of antibodies known as secretory IgA, which circulate in mucosal surfaces such as the lining of the urinary and reproductive tracts and block invading microbes.

    In preliminary trials of 25 women, the vaccine seems at least partially effective: Women who are prone to UTIs acquire infections less readily, and none have reported side effects. However, the vaccine's protective effect may diminish over time, says immunologist Walter Hopkins, a member of the Wisconsin team. If that is confirmed in the group's final-stage efficacy trials, says Hopkins, it could mean that women would have to readminister the vaccine, possibly as often as every few months.

    The MedImmune-Washington University team, led by MedImmune's Solomon Langermann and Washington University's Scott Hultgren, has adopted what may prove to be an even more promising strategy. Its vaccine triggers the immune system to produce antibodies that block the action of just a single, key protein on UTI-causing microbes. Administering only that protein does away with the need to expose patients to the whole organisms—and their side effect-producing toxins. The researchers targeted an “adhesion” molecule called Filamentous H, or FimH, present on E. coli. The microbes deploy FimH on the end of long, spaghettilike strands that extend from the cell body and latch onto sugar molecules on the surface of host cells. Says microbiologist Vince Fischetti of Rockefeller University in New York City, “If you block that interaction, you can prevent infection.”

    Researchers have tried to develop adhesion vaccines for other diseases such as gonorrhea, which is caused by organisms that rely on adhesion proteins. But in the past, adhesion vaccines “have not panned out very well,” says Mobley. When genetically engineered bacteria are coaxed into producing the large amounts of adhesion proteins needed for a vaccine, the proteins often become degraded or clumped together, losing their ability to provoke immune cells into making antibodies targeted to the protein.

    So, the MedImmune-Washington University team whipped up two separate vaccine formulations in the hope that at least one would yield suitable proteins. For the first, the researchers genetically engineered E. coli to express extra FimH, which they then collected and purified. As in earlier efforts to develop adhesion vaccines, these proteins ended up partially degraded. But fortunately, the part of the protein that triggers a protective antibody response remained intact. In the second formulation, the scientists modified the bacteria to express not only FimH, but also so-called chaperone proteins, which ensure that proteins fold into their proper conformation as they are made.

    The team injected separate groups of mice with the two vaccines and exposed them 9 weeks later to UTI-causing E. coli. Both groups of vaccinated mice were able to ward off UTIs for more than 7 months, the latest time point studied. Analyses of the animals' urine showed that both vaccines had elicited blood-circulating IgG antibodies, some of which leaked into the mucosal lining of the bladder and urinary tract. These antibodies, the researchers believe, bound to E. coli's natural FimH proteins, preventing the bacteria from binding to their target cells.

    Despite this early success, Mobley and others say that the MedImmune-Washington University team has a long way to go before proving that its adhesion-protein vaccine can make it to market. The researchers will have to demonstrate that the vaccine can block UTI-causing E. coli in humans while sparing another colony of E. coli: the beneficial intestinal flora that keep disease-producing bugs from proliferating in the gut. Says Mobley: “The current results are still quite preliminary.”

    In any case, Mobley and others agree, the new adhesion vaccine's initial success could pave the way for developing a host of other such vaccines for other diseases. Adhesion vaccines might just catch on.

  7. Human Genetics

    Putative Cancer Gene Shows Up in Development Instead

    1. Wade Roush

    Seven years ago, a team at Johns Hopkins University used a new chromosome-walking technique to locate a long-sought suspect in colon cancer. They announced in Science the identification of a gene called Deleted in Colorectal Cancer, or DCC (Science, 5 January 1990, p. 49). As the name implies, this gene is part of a long stretch of DNA on human chromosome 18 that was already known to be missing in tumor cells from many patients with colon cancer. Finding an actual gene in this stretch helped shore up the then-emerging theory that cancer results in part from inactivation of tumor-suppressor genes. And since then, nearly 150 papers have probed DCC's function and clinical significance. Today, its absence in tumor cells is used as an important diagnostic marker, helping to identify patients who need aggressive treatment, explains Ian Summerhayes, a cancer biologist at Beth Israel Deaconess Medical Center in Boston.

    But what exactly does normal DCC do in cells? Eric Fearon and Bert Vogelstein, the leaders of the Johns Hopkins team that identified DCC, hypothesized that it suppresses tumors by regulating cell growth or differentiation. In this week's issue of Nature, however, three independent groups report that DCC and newly discovered relatives in rats and mice have an entirely different function: They encode cell surface receptor proteins that interpret directional signals used by migrating cells or developing neurons. One group also reports that mice lacking DCC have brain and spinal cord defects, but no excess of gastrointestinal tumors. This raises doubts about whether DCC itself helps regulate normal cell growth or simply lies close to an unknown tumor-suppressor gene and is usually deleted along with it. “It could be useful as a marker and have nothing to do with [cancer] progression,” says Summerhayes. As one leading researcher jokes, DCC should perhaps be renamed “Deleted as a Cause of Cancer.”

    Still, closing arguments in the case are yet to come. Vogelstein calls the papers “terrific studies, very thorough,” and says they “very convincingly show that DCC [and its cousins] are important components of nervous-system development.” But the absence of tumors in mice with DCC knocked out, he says, “is a negative result, and unfortunately those are much more difficult to interpret than positive results.” And Fearon, now at the University of Michigan, suggests DCC could have other roles in addition to directing axonal growth. He remains convinced that it is the leading candidate for the role of tumor suppressor in cells lining the colon and rectum. It may be years before researchers map enough of the genetic sequence around DCC to flush out other candidate tumor suppressors, so many researchers say a faster way to incriminate or exonerate the gene may be to learn its exact function in development and whether disrupting that function can accelerate tumor growth.

    Developmental biologists have been citing their work's links to cancer in grant proposals for years, but DCC offers a striking case. Indeed, only one of the teams behind the trio of Nature papers originally set out to learn something about cancer. In 1990, Amin Fazeli, a Yale University medical student working in the laboratory of cancer biologist Robert Weinberg at the Whitehead Institute for Biomedical Research in Cambridge, Massachusetts, decided to delve further into Fearon and colleagues' discovery that 18q21, a region on the long arm of chromosome 18, is missing in more than 70% of human colon and rectal tumors. Hundreds of genes could fit inside the large 18q21 region, but DCC was the first one detected, and other studies rapidly began to link it to cancer. Fazeli put this to the test by genetically engineering several strains of transgenic mice, including one with a single working copy of the mouse version of DCC (labeled DCC+/−) and another with no working copies of the gene (DCC−/−).

    All the DCC−/− mice perished shortly after birth, although the problem was not in their gastrointestinal tracts, which were normal. Fazeli expected the DCC+/− mice to show an increase in gastrointestinal tumors as a result of spontaneous mutations in the lone working copy of DCC. But he was puzzled to find that of nearly 200 DCC+/− mice, only one developed such a tumor. “We could not come up with the smoking gun,” he says. “We were very disappointed, and surprised.”

    With Fazeli's studies stalled, the next batch of clues to DCC's character came from an unexpected source: the tiny worm known as Caenorhabditis elegans. In developing worms, as in most animal embryos, neurons and their threadlike axons—the cellular telephone lines that connect them to other neurons—travel quite a distance before settling into their assigned positions; their route is guided by nearby cells, which either attract or repel them. One vertebrate protein that cells secrete to attract axons is netrin-1, first discovered by developmental geneticist Marc Tessier-Lavigne and colleagues at the University of California, San Francisco. Mice lacking netrin-1, they found, exhibit nervous-system defects. Joseph Culotti at the University of Toronto, Ed Hedgecock at Johns Hopkins, and colleagues, meanwhile, discovered a C. elegans homolog for netrin-1, called UNC-6, that is secreted by neurons along the main trunk of the worm's nervous system; they also found that UNC-6 seems to pass along the attractant signal by binding to a receptor protein called UNC-40 on the axons' leading tips.

    Then, late last year, Culotti and Hedgecock announced a surprising discovery: The worm gene encoding the netrin receptor UNC-40 shares 30% of its amino acid sequence with vertebrate DCC (DCC's protein product), indicating that the two genes are distant evolutionary cousins and raising the possibility that DCC is also a netrin receptor, Culotti says.

    That's where the three new Nature studies come in. When Fazeli, now completing his Ph.D. in biology at the Massachusetts Institute of Technology, heard of the Toronto team's result, he reexamined his ill-fated DCC−/− mice, this time focusing on their nervous systems. “We were very pleasantly surprised,” he says. The mice had brain and spinal cord defects “strikingly similar” to those seen in the netrin-1 knockouts. Fazeli and the other Whitehead researchers teamed up with Tessier-Lavigne's group and, in the first Nature paper, report careful anatomical studies of DCC−/− embryos showing that without DCC, axons get lost. In normal cells, netrin-1 is emitted on the inner or ventral side of the developing spinal cord and attracts axons from the outer side. Without DCC, those axons never make their ventral journey (see photo). “That provided a very strong piece of evidence supporting the hypothesis that DCC is a receptor for netrin-1 in vertebrates,” says Fazeli.

    And that's not all. Culotti's group had previously found another netrin receptor in worms, called UNC-5, that apparently switches an axon's response, turning UNC-6 into a repellent rather than an attractant. Now, proteins homologous to this netrin receptor are turning up in vertebrates. The two other Nature papers report such homologs in rats and mice, and both are active in the brain and spinal-cord regions.

    All this suggests that netrins and their receptors make up a genetic module so powerful that it is conserved across worms, rats, mice, and even humans—and that DCC itself is a netrin receptor, expressed at the tips of growing axons and given the job of guiding them to the right position, says Culotti. The accumulating evidence makes it more and more difficult to see how DCC's absence in colorectal tumor cells could be the crucial factor permitting cancerous growth, says Ray White, a human geneticist at the University of Utah.

    Another gene in the 18q21 region could be the real culprit, with DCC as an innocent neighbor, says White. At least two other candidate tumor suppressors, called Smad2 and Smad4, have already been traced to the area; Smad4 loss is suspected as a leading cause of pancreatic cancer (Science, 19 January 1996, pp. 294 and 350). Both genes seem to modulate signals carried by a protein called TGF-β and its relatives, which among other functions are thought to direct the development of colon cells. All three genes, however, could merely be carried along for the ride when the entire 18q21 region is deleted. So, although all the candidate genes appear to have links to the colon, “somebody could find a [tumor-suppressor] gene that's an even better target than these,” explains Vogelstein.

    But Fearon notes that DCC might be more than just a receptor for netrins, perhaps interacting with other signaling molecules in some other pathway regulating cell movement or cell fate. DCC appears to be expressed at low levels in many cells that might need directional cues, Fearon says, including the colonocytes that slowly migrate up the lining of the colon, then slough off. “I think most of the data still weigh in for DCC as the most likely candidate,” he says.

    And the new data on DCC don't diminish its diagnostic usefulness as a marker for aggressive colorectal tumors, adds Beth Israel Deaconess's Summerhayes, because it seems clear that DCC is at least close to the actual tumor suppressor. Whichever gene proves to be the crucial target of deletion, DCC's story illustrates that developmental biology has a role to play in cancer research, says Johns Hopkins cancer researcher Scott Kern: “There's no better place to apply [the growing knowledge of development] than in human cancer biology.”

  8. Anthropology

    Back to Africa

    1. Ann Gibbons

    ST. LOUIS—More than 1000 anthropologists gathered here from 1 to 5 April for the 66th annual meeting of the American Association of Physical Anthropologists and associated groups. Researchers presented new genetic and fossil findings marking key milestones in long-running debates on such topics as the ancestors of apes (Science, 18 April, p. 355), the origins of modern humans, and the evolution of menopause.

    Since the mid-1980s, two diametrically opposed hypotheses for the origin of modern humans have been battling for primacy. One, called Regional Continuity, holds that our earliest ancestors arose in Africa and spread around the world more than 1 million years ago. Modern humans then arose in many different regions through separate evolution and interbreeding. The other—the favored contender—is the theory known as Out of Africa, which suggests that our ancestors arose in Africa and swept around the globe 100,000 years ago, completely replacing existing human populations on other continents. This model hasn't been proven, but a series of genetic and fossil studies have suggested to many researchers that, as Stanford University geneticist Neil Risch put it last year, “the rest of the world emerged from the northeast corner of Africa” (Science, 8 March 1996, pp. 1364 and 1380).

    Now, a middle ground may be emerging, as new data from two international teams of geneticists challenge the most extreme version of Out of Africa. That theory predicts that all genes in modern humans were inherited from a small number of Africans, but the new data suggest that some modern human genes come from ancestors in Asia, not Africa. The new evidence falls far short, however, of proving Regional Continuity; rather, it shows that both of these leading models of how modern humans emerged have been overly simplistic. “There's more than one migration out of Africa,” says Michael Hammer, a geneticist at the University of Arizona, Tucson. “And the direction is not just one way. Some are moving back to Africa.”

    Hammer came to that conclusion after studying DNA in 1500 males from 60 populations around the world. One region he focused on was a 2600-base-pair segment of the Y chromosome: the YAP region, which is passed from fathers to sons. This segment varies among individuals, but the sequences cluster in five major groups, known as haplotypes (shown in five colors; see map). The haplotypes occur in different frequencies in different populations, and Hammer's team found that one—YAP haplotype 3 (shown in red)—shows up far more often in Asians than Africans. Its sequence shows more diversity in Asians, implying that the haplotype had more time to acquire mutations in Asia than in Africa—and, therefore, that it arose in Asia. Its presence in some Africans hints that human ancestors migrated back from Asia into Africa at some point, says Hammer.

    What's more, Hammer's team thinks that this haplotype is ancestral to two of the most common ones found in Africa—haplotypes 4 and 5 (shown in blue and yellow)—because all three share a common sequence missing in other African haplotypes. In a paper in the March issue of the journal Genetics, Hammer and colleague Stephen Zegura conclude that “a major component of [modern] African diversity is derived from Asia.”

    But these findings come from just one gene in males, which may reflect the movements of a few men rather than entire populations. So, Hammer was pleased to hear similar findings at the meeting from another piece of nuclear DNA, the beta-globin gene on chromosome 11, which is inherited from both parents. Geneticist John Clegg of the Institute of Molecular Medicine in Oxford, England, along with population geneticist Rosalind Harding, and their colleagues, sequenced a 3000-base-pair region that included the beta-globin gene in 349 individuals from nine populations in Africa, Asia, and Europe. Using new computational methods, they built a genetic family tree to sort out the sequences, which fell into four major groups of haplotypes.

    The team then used mutation rates for human nuclear genes to calculate how long ago the sequences split from one ancestral haplotype. They found that the oldest version of the gene, haplotype B2, arose more than 800,000 years ago in Africa. Some time before 200,000 years ago, however, B2 gave rise to a set known as C haplotypes. These are common in Asians and very rare in Africans—implying that the haplotypes arose in Asia, says co-author and molecular anthropologist Malia Fullerton, of the University of Durham, England. Asians also have two ancient C haplotypes not found in Africans, and Africans have one C haplotype that appears to have been brought in recently. These findings again suggest that some Asian markers are older than African versions, challenging the complete replacement model, according to a paper in the April issue of the American Journal of Human Genetics. “We're seeing evidence of a widely dispersed human population about 200,000 years [ago], both in Africa and Asia. And both contributed to the gene pool” of modern humans, Fullerton says.

    Others have yet to be convinced. Pennsylvania State University postdoc Sarah Tishkoff, a co-author of a recent genetic study supporting Out of Africa, says that more sampling of DNA from Africans is necessary to make sure that these so-called Asian haplotypes aren't descended from an as-yet-unidentified African haplotype. But Tishkoff agrees that both leading models of modern human origins are too simplistic. “This tells us that we have to take into account more complex models,” says Tishkoff. “All of us in the field have oversimplified.”

  9. Anthropology

    Why Life After Menopause?

    1. Ann Gibbons

    ST. LOUIS—More than 1000 anthropologists gathered here from 1 to 5 April for the 66th annual meeting of the American Association of Physical Anthropologists and associated groups. Researchers presented new genetic and fossil findings marking key milestones in long-running debates on such topics as the ancestors of apes (Science, 18 April, p. 355), the origins of modern humans, and the evolution of menopause.

    Why do women live so long after they stop reproducing? Human females are the only ones in the primate family to live well beyond their last pregnancy—often as long as 40 years or more after menopause. Yet, evolutionary theory says that natural selection favors only traits that enhance reproduction—which implies that postreproductive women have no evolutionary reason to live.

    Grandmotherly love.

    Among the Hadza, the food that grandmothers provide is crucial to the survival of grandchildren.

    James F. O'Connell

    Now, a new study of African hunter-gatherers suggests a provocative answer to this riddle: Women live to a ripe old age to make sure their grandchildren eat. By provisioning grandchildren, grandmothers ensure the children's survival, boost their daughters' fertility—and improve the chances that their own genes are passed on. With grandmothers providing food, daughters can breast-feed infants for a shorter period and so bear more babies during their fertile years than primates without helpers do, say University of Utah anthropologist Kristen Hawkes and colleagues, who presented their paper at the Paleoanthropology Society meeting.

    This “grandmother hypothesis” suggests that natural selection favored menopause (because only grandmothers who are not busy feeding their own children have time to provision grandchildren), as well as long life and perhaps even close family ties. If the hypothesis is true, then grandmothers may be doing even more provisioning than male relatives, says anthropologist Sarah Blaffer Hrdy of the University of California, Davis. “This is the first serious challenge to the widely accepted view that the human family evolved because males were needed to provision mothers,” she says.

    Hawkes and colleagues James O'Connell of the University of Utah, and Nicholas Blurton Jones of the University of California, Los Angeles, learned of grandmothers' contributions when they spent a year studying a group of 300 Hadza hunter-gatherers in the rugged hill country southeast of Lake Eyasi in northern Tanzania. The Hadza move from season to season and survive almost entirely on wild resources. Men hunt and collect honey while women dig tubers or collect berries and other fruit.

    The researchers found that children's weight gains usually depended on how much time their mothers spent foraging. But when mothers gave birth to new babies, they had less time to find food for older, weaned ones. The “hardworking, incredibly fit” grandmothers, mostly in their 60s, took up the slack, says Hawkes. “What was striking was that these older women were spending more time foraging than younger mothers were,” she adds. As a result, the weight gain of children whose mothers were nursing depended on their grandmothers' foraging.

    Such provisioning by grandmothers may allow human mothers to have babies closer together than other apes can, and may also explain a suite of other distinctly human traits, says Hawkes, who has worked with University of Utah evolutionary biologist Eric Charnov on this theory. Unlike other primates, humans are weaned early, at a relatively small size, and have extended childhoods and high fertility. The “grandmother hypothesis” provides a “nice story” that could explain why all those features have been selected for, says Hrdy. And it challenges the idea that male hunting is the crucial factor allowing long, dependent childhoods in humans. Hawkes's group found that hunting was a less reliable source of food than the tubers grandmothers dug up.

    Unfortunately, the evidence comes only from the Hadza, and Hrdy isn't convinced that this group is a good proxy for the behavior of our ancestors. And although Hawkes's team is studying other hunter-gatherers and other primates, tests of their ideas may be hard to come by. Clues could come from fossils of early humans or the archaeological record, says O'Connell, but several paleoanthropologists at the meeting warned that it will be difficult to identify a unique signature of grandmother provisioning. “I like their paper, but it's going to be really hard to test,” says University of New Mexico paleoanthropologist Erik Trinkaus.

    For now, however, at least a few anthropologists have shifted their focus away from male hunters to a new group. “If there is a group that we have paid no attention to, it's old women,” says Hawkes. The bottom line, at least for the Hadza, she says, is a message worthy of a bumper sticker: “Grandmothers Matter.”

Log in to view full text

Via your Institution

Log in through your institution

Log in through your institution