G E T T I N G   I T   R I G H T

An unconventional geneticist explores
how fruit fly embryos fix mistakes to develop properly.

By Evelyn Strauss

ON THE WALL above Bill Sullivan's file cabinet hangs a winged tennis shoe. Trapezoid-shaped Plexiglas panels splay out from the side of the shoe, attached with metal hinges. Sullivan invented this contraption to help him negotiate a trail in Big Sur that meanders back and forth across a river. It's hard to swim wearing tennis shoes, he says, and he didn't want to take them off and put them on every ten minutes. The panels fold up for walking and down when he needs flippers.
      Sullivan's knack for solving problems many people never even ponder gives him a unique approach to scientific puzzles. His radically fresh perspective has led him to challenge and overturn some widely held assumptions in his field.
      Sullivan has applied this approach to ask how a fruit fly develops from a fertilized egg into the complex insect that makes a nuisance of itself in fruit bowls on hot summer days. In particular, Sullivan is asking how the fly's embryo grows at seemingly reckless speeds during the early stages of its development, and yet manage to produce an animal with exactly four wings, six legs, and two eyes, all of which function properly.
      The flies use unexpected methods to assure quality control during their initial growth period, Sullivan has found. These include techniques that organisms such as humans use later in life to protect themselves from cancer. His findings have opened up a new avenue for studying how cells normally behave as good citizens in the bodies of a wide variety of animals.

THE PROCESS of self-correcting development starts with a tiny fruit fly egg. After fertilization, the resulting cell duplicates its genetic blueprint. The two copies separate from each other, and each is sequestered in its own pouch, or nucleus, inside the cell. This process repeats as the fly embryo grows, to produce increasing numbers of nuclei.
      The nuclei synchronously zip through the first thirteen reproductive cycles at some of the fastest reproductive rates known. During this period, the duplicating nuclei inhabit one large sac. After seven initial divisions in the interior of the embryo, most of the nuclei begin to migrate to the surface. There they continue to double in concert, coordinating their movements as well as the timing of their divisions with choreography that resembles a Busby Berkeley dance scene in an MGM musical comedy. The appearance of the embryo reflects this precision: the nuclei that dot the ellipsoid sac are evenly spaced over the surface. Because so many nuclei are dividing quickly, simultaneously, and with scripted movements, young fly embryos lend themselves especially well to the study of regulated growth.
      At the fourteenth division, membranes form to separate the nuclei into distinct cells. Sullivan is asking some critical questions about this process. For example, until this point, how do the nuclei duplicate without bumping into one another? How do the embryos contend with defective nuclei that arise? These questions snagged Sullivan's curiosity because the nuclei divide at breakneck speed, closely packed, with no cell membranes to confine them to their own spaces.
      In most mature plants and animals, dividing cells ensure quality by inspecting their work at every step before they proceed. This process counteracts the defects that inevitably arise from slightly imperfect cellular machinery as well as harmful outside influences, such as environmental toxins. If a cell finds errors, it pauses to correct them. Cells check, for example, that they've finished copying their entire genetic blueprint before they split in two. That way, both offspring cells contain a complete set of plans.
      Young embryonic fly nuclei, in contrast, seem more focused on speed than accuracy. If they fail to copy their genetic blueprint completely in the time allotted, the nucleus divides anyway, without a full set of instructions for each resultant nucleus. Delays in individual nuclear divisions would disrupt the synchrony of the mass cycling and throw the process into chaos. "It's like in a race with thousands of people," Sullivan says. "If someone stops to tie his shoe, it'd be nice if everyone stopped--but that's not what happens." Like the person with an untied shoe, the disabled nuclei must stumble along with the crowd.
      But animals that accumulate flawed nuclei risk developmental problems and, after birth, cancer, so scientists reasoned that there must be some way for the embryo to correct its mistakes. Leaders in the field suggested that growing fly embryos tolerate imperfections until division fourteen. At that point, when nuclear duplication slows down, the embryo begins acting like an adult. Its cells then scrutinize nuclei and mend injuries, according to the theory.
      This idea bothered Sullivan. "They thought there was no mechanism to deal with damage [early during embryonic growth], but that didn't make sense," he says. " That's the most important time in development, so you'd think that it might be important to do it right."
      Sullivan wanted to see for himself what happened to faulty nuclei in developing embryos. He devised a way to look at chromosomes within dividing nuclei. Normal chromosomes pair up to form rows, which facilitate their organization during division. Sullivan was particularly interested in nuclei with jumbled, instead of neatly ordered, chromosomes. The nuclei containing these unruly chromosomes didn't divide properly, and lagged behind their neighbors.
      Sullivan saw a ruthless world, where nuclei that don't keep up get thrown out. Instead of stopping to fix problems, the embryo tosses defective nuclei into its middle, the "garbage can of the developing animal," he says, and there they die. "The fly makes a whole lot of nuclei and just throws out the bad ones," says Sullivan. In this way, the embryo maintains standards.
      A fruit fly embryo can lose as many as half its nuclei and still turn out right, Sullivan found. "The embryo has a tremendous ability to compensate," says Sullivan. "You can eliminate huge numbers of nuclei and somehow the embryo can keep track and replace what it got rid of."
      Beginning at stage fourteen, the animal seems to value each individual more, and rehabilitates misbehaving cells rather than sentencing them to death.
      Sullivan likens the initial stages of fly development to a basketball factory. "It's OK if two percent are messed up and you throw them out," he says. "But if you're making a Mercedes, you want to make each one perfect. It's too valuable to afford to lose one." Early on, when the nuclei are interchangeable, they are dispensable. Later, however, when they commit to futures as wing or antennae components, their worth increases.

SULLIVAN'S initial skepticism of the defect-toleration hypothesis has proved correct. The young embryo indeed monitors for damaged nuclei and expels them, thus assuring the integrity of its constituents. It does not, as had been suggested, allow sloppiness early and wait until stage fourteen to inspect.
      "Bill doesn't let the prevailing dogma affect his thinking at all," says Doug Kellogg, of the University of California, Santa Cruz. "He's not really bound by the way the rest of the scientific community looks at things."
      Sullivan's unconventional approach resonates with other aspects of his life. When he began working at UCSC as an assistant professor, his home was a boat. And as an undergraduate at the University of California, San Diego, he slept in his car for several years. "I think he didn't get around to making arrangements for living in campus housing," says Sullivan's sister Jeannie Sullivan. " It was easier to just buy a parking permit."
      At UCSD, Sullivan already knew he wanted to study science. He had become fascinated with the natural world as a child. Every week he tuned in the family TV to "The Undersea World of Jacques Cousteau."
      "At the end, the narrator would say something like 'and he solved the mystery of the Red Sea,' and I thought, 'By the time I'm old enough [to work as a scientist], Cousteau's going to have everything figured out,'" Sullivan recalls. "There got to be a point when I couldn't watch any more. At twelve, I was already worried about being scooped."
      In his office on the University of California, Santa Cruz campus, Sullivan sits at a desk with its drawers ajar. Piles of folders, a large glass flask, a box of computer disks, and a sweater litter his floor. A paper coffee cup and muffin wrapper lie on a tabletop supported by two sawhorses next to his desk. A thick black binder holds his scientific papers, neatly ordered and numbered.
      Sullivan, with the reds and blues of his different shirt arms peeking out from under one another, says he likes the formalism of genetics.
      "You have the analysis on paper and then do something simple like count the number of flies with a certain eye color, and you can learn something really profound," he says. "It's pretty amazing that Mendel got it all right--just by counting peas."
      More than 100 years ago, Gregor Mendel figured out the rules that govern the inheritance of many traits by mating plants of different colors and pea shapes.
      Sullivan sits up in his chair, nods his head, and leans forward as he talks. His eyes brighten. He seems a bit like a child, talking about his new Lego set.
      Anxious to get back to the lab, Sullivan glances at his watch, which is attached to its band with a strip of bright green lab tape.

AFTER Sullivan found out what happens to damaged nuclei in fly embryos, he dove deeper into the question of how they usually manage to divide correctly. He thought the answer would lie in the molecules that push and pull DNA, and package it into new nuclei. So he set out to find the embryonic components that shepherd the chromosomes.
      As his first step, he created strains of flies whose nuclei make more than the usual number of mistakes. These strains, he reasoned, should contain altered versions of molecules that normally make nuclei behave properly.
      This approach is similar to learning about how cars function by tying the hands of different workers in an automobile factory, Sullivan explains. The flawed cars that roll off the assembly line provide information not only about each person's job, but also about how the car works. When cars emerge without the round plastic devices on their front left-hand sides, one suspects that the incapacitated person installs steering wheels. And when those same cars are driven to the parking lot and fail to make the first turn in the road, one might conclude that the round plastic devices steer the car.
      In principle, by tying every worker's hands on different days and studying the emerging cars, one can match people with their jobs in the factory, and discover what the different components of the car do. In flies, each strain (or day in the factory) contains a defective gene (or worker with tied hands).
      "What's nice about [fruit flies] is that if there's a defect, you detect it because the nuclei are no longer evenly spaced in the embryo," says Kellogg. "There's an easy read-out."
      Sullivan looked at 76 strains of mutant flies. He found several that display an irregular pattern of too few nuclei on the embryonic surface and contain excess nuclei on the interior.
      His scheme succeeded, but, like many research projects, it easily could have failed. Daring experiments, however, don't daunt Sullivan, reports Kellogg. One Friday night about eight or nine years ago, Kellogg and Sullivan were walking up Tank Hill in San Francisco. "Bill stops in front of this big poison oak bush and tells me how he used to get these really bad cases of poison oak and he doesn't anymore and he's pretty sure he's become immune," says Kellogg. Sullivan reached out, grabbed the plant, and rubbed it on to the back of his hand.
      "Monday I come into work and Bill is sitting at the microscope. His eyes are nearly swollen shut, and he has big patches of poison oak all over his hands, his face, his arms," Kellogg says. "Bill really isn't afraid of trying radical experiments."

AMONG the fly genes Sullivan has found, one encodes a relative of a protein scientists already knew about. This protein resides in yeast cells, and gives them time to repair damaged DNA. When other molecules report problems, the protein pulls the brakes on cell division. Like mature animal cells, yeast correct errors, presumably because serious blunders are a matter of life and death for each of these single-celled organisms.
      Fly embryos containing a defective version of this gene contain clustered--instead of evenly spaced--nuclei. Because of the clumpy appearance, Sullivan named the gene "grapes."
      Embryos with a defective grapes gene contain higher than normal levels of DNA damage, Sullivan showed. A string of unbroken DNA usually composes each chromosome in a healthy cell. Although the chromosome is copied in small segments, the cellular machinery patches the resulting pieces together. Within the mutant embryos, almost five times the normal number of loose DNA ends exist. This result indicates that the nuclei don't have enough time to complete DNA replication before they start trying to divide, Sullivan says. Such DNA doesn't line up so it can divide properly, and these fragments of DNA pose a hazard to cells that contain them.
      Reasoning from these results and the resemblance of grapes to the yeast gene that suspends cell division, Sullivan proposed that the protein product of the grapes gene generates a pause that allows fly nuclei time to finish duplicating their chromosomes.
      "What's surprising is that the same genes are used [in yeast and flies]--just in different ways," Sullivan says. Yeast use their protein to counteract mistakes, while fly embryos use theirs as a normal part of nuclear division. The fly embryos don't really need the Grapes protein until cycle 11 because earlier, DNA replication occurs slightly more quickly and can keep up with the pace of nuclear division. In both yeast and flies, the relevant proteins generate a pause so the nucleus can ready the DNA for division.
      Once Sullivan decoded the fly grapes gene, Stephen Elledge's group at Baylor College of Medicine in Houston used the information to find its human counterpart. Although scientists had isolated the yeast gene four years earlier, their previous attempts to uncover the human version had failed because the two are too distantly related. "People have been looking for it [the human version] for a long time, but they couldn't pull it out," says Sullivan. Because humans genetically resemble flies more closely than yeast, Sullivan's gene provided the missing link that allowed scientists to nab the human gene.

GRAPES is one of two genes in yeast, flies, and humans that ensures that one portion of cell division is finished before the next one begins, says Sullivan. This means that scientists may be able to exploit flies to learn about humans. No one knows whether the grapes gene carries out similar jobs during human and fly development, but divisions early during embryonic development tend to differ from later ones in all higher organisms, so Sullivan's work may hold relevance for human pregnancy.
      In addition to informing scientists about embryonic growth, the grapes gene may provide information about the basic mechanisms of growth and tumor formation in a wide range of organisms. Genes such as grapes, which control the order and timing of events during cell division, have been implicated in the transformation of healthy cells into cancerous ones. "Cancer is a case of cells [growing] out of control," says Kent Golic, a geneticist at the University of Utah in Salt Lake City. "The cells don't know that they're not supposed to be dividing."
      The grapes gene may provide a powerful tool with which to investigate this process. Like yeast, flies are amenable to manipulation and experimentation. But their bodies, like those of mammals, are composed of specialized cells that carry out different jobs. Using flies to probe the function of grapes may well convey scientists to an understanding of humans that would not be possible by studying only yeast.
      Although Sullivan began this project searching for the components that shove DNA from one place to another within the cell, he wound up with a gene that plays a role in growth control.

IT FEELS familiar to Sullivan to start out with a general goal, but find himself in a place he never would have anticipated. It's the same approach he takes when he goes hiking in a new area. While some people pore over maps and meticulously plan their hikes, he just finds a trail head and starts walking, he says. "I've never been to any of the places and I'm sure they'll all be nice. The people who agonize over maps--they'll end up in a nice place too, but never where they hadn't planned. There are some types of science where you pick a trail and continue along that. If anything tries to lead you away, you don't let yourself detour."
      Jeannie Sullivan says she's surprised her brother wound up in the lab. "He never seemed to be a precision type of guy," she said. "He's not someone who would take apart an alarm clock and put it back together. He'd just take it apart."
      One day a few years ago, however, she started seeing the adult Bill in a different light. They had stopped to pick blackberries at the side of the road. "My brother was never neat, ever," she says. "We finished picking blackberries and my hands looked like I'd been mashing them. But his were completely clean. He somehow picked them so precisely that his hands didn't get dirty.
      At the same time, she says, "his socks don't match.
      "It has to do with putting your energy where it matters."