The small, windowless space at the Children's Hospital of Philadelphia looks like any eye doctor's examining room, with an adjustable chair and half a dozen machines for testing vision. The 20-year-old patient, however, has not come all the way from Albuquerque to get new glasses. Alisha Bacoccini, who has short, blond-streaked hair and green eyes, was born with a disorder caused by a malfunctioning gene in her retina cells that has been diminishing her sight since birth. Now she sees only pale and blurry shapes. "If I look at you I can't see eye color or acne or your eyebrows, but I can see that someone's there," she says. Her seeing eye dog, Tundra, a black Labrador retriever, sits at her feet.

A month earlier, in an experimental treatment, researchers injected Bacoccini's right eye—the worse one—with billions of working copies of the retinal cell gene. Now they'll find out if the treatment has worked.

Jean Bennett, a physician and molecular geneticist, has Bacoccini rest her forehead against a small white machine that flashes light into one eye, then the other. This pupillometer will indicate how well Bacoccini's eyes respond to light. "OK, one, two, three, open," Bennett says, and repeats the procedure 16 times. On a computer screen in the darkened room, Bacoccini's pupils are two giant black circles that contract ever so slightly with each pulse of light. Another researcher escorts Bacoccini to the next testing apparatus. Half an hour later, Bennett says: "I just looked at your pupillometry results. Good improvement."

"That's good," Bacoccini says, though she sounds unsure. Since a few days after the injection, she has indeed seen more light out of that eye, she says, but things seem blurrier. When she tries to read a giant eye chart with her right eye, she does no better than before—she can pick out only a few two-inch-high letters from 16 inches away. Then again, her eye is still red from the surgery. Bennett's husband, Albert Maguire, is the retinal surgeon who operated on Bacoccini. He peers into her eye and says the surface hasn't yet healed, adding: "Hopefully, that's all it is."

The prospect of using gene therapy to treat diseases—particularly inherited diseases that involve one errant gene, such as sickle cell anemia and cystic fibrosis—has tantalized scientists for decades. If there were some way to give a patient a good version of an implicated gene, the thinking goes, it might repair or prevent damage caused by the inherited bad one. This seemingly simple idea has turned out to be unexpectedly complex in practice. There have been hundreds of human gene-therapy trials for many diseases, from hemophilia to cancer, in the past 18 years. But nearly all failed because of the difficulties of getting a working gene into cells without also causing harmful side effects.

Until last year, gene therapy had worked unequivocally against only one disease, the rare affliction called severe combined immuno-deficiency (SCID), which is caused by a flaw in any of a number of genes needed to produce white blood cells. The disease leaves the immune system unable to fight infections and usually leads to death in childhood. It is also called "bubble boy" disease, after one famous patient, David Vetter, who lived to age 12 in a sterile plastic bubble. Since the mid-1990s, European researchers have cured about 30 kids with SCID by inserting the appropriate functioning gene into their bone marrow. But even this success has been mixed with tragedy: five of the children developed leukemia and one has died. In those patients, who had a particular variant of the disease, the therapeutic gene accidentally turned on a cancer-causing gene after merging with the patients' DNA. Researchers are now testing ways to make gene therapy for SCID safer.

U.S. gene-therapy research was set back substantially after 18-year-old Jesse Gelsinger, who suffered from an inherited liver disease, died of multiple organ failure in 1999 while participating in a gene-therapy experiment at the University of Pennsylvania. News of the death prompted an uproar in the scientific community and hearings in Congress, with the teenager's father, Paul Gelsinger, and others accusing the Penn researchers of being too hasty to test the treatment in people. According to the Food and Drug Administration, the researchers had not sufficiently warned Gelsinger and his family of the experiment's risks. The lead researcher had also failed to disclose that he had a financial stake in a company that stood to gain if the treatment succeeded. "Those were the terrible days. The field bottomed out," says Leon Rosenberg, a Princeton University human geneticist, who performed early lab studies on the liver disease that Gelsinger had. "The integrity of science was damaged tremendously."

Bennett and Maguire joined the Penn medical school faculty in 1992. One of their colleagues is James Wilson, who oversaw the study in which Gelsinger died. Wilson was subsequently barred by the FDA from conducting human experiments. But Bennett and Maguire were not involved in that study. Their experimental gene-therapy trial began in 2007 after years of review by federal regulators, the Children's Hospital and Penn committees set up to address ethical and safety concerns raised by Gelsinger's death.

This past May, their team and a separate British group reported the first hopeful gene-therapy news in years: the technique could treat blindness. The patients in the study had a disease called Leber congenital amaurosis (LCA). The three patients whom Bennett and Maguire treated were able to read several more lines of an eye chart than they could before. One 26-year-old man even regained enough sight to walk through a maze. "I couldn't believe it," Bennett says. She made him walk the maze over again.

The study was small, and the patients are still legally blind, but their modest improvement and the apparent safety of the therapy have aroused the hopes of patients and researchers around the world. Now Bennett and Maguire are extending the research to more patients with LCA, including Bacoccini, to test whether patients can safely receive higher doses of the therapeutic gene.

Rosenberg says he is "delighted" for Bennett—who was a postdoctoral researcher in his lab in 1987—and the field. "I'm optimistic about gene therapy again and I haven't been for a while," he says. "Hopefully before the end of this decade there will be two or three other examples."

For both Bennett, 54, and Maguire, 48, science would appear to be in their genes. Bennett's mother, Frances, taught high-school literature and her father, William, was a popular physics professor at Yale who had co-invented the gas las­er in 1960 while working at Bell Labs in New Jersey. Bennett remembers being 6 years old and seeing her father race back to the lab after dinner to work until dawn; the device later led to compact disc players and supermarket price scanners.

She was more interested in biology than physics. "I loved creatures," she says, and spent many happy hours looking through her father's microscope at swamp water and leaves. After college at Yale, she went to the University of California at Berkeley to earn a PhD in developmental biology, using sea urchins, but she was drawn instead to then new research on inserting specific, foreign genes into mice and other animals—a forerunner of gene therapy. She spent several months in 1981 and 1983 at the National Institutes of Health in Bethesda, Maryland, in a lab where scientists were planning some of the first gene-therapy trials. "It was a glimmer that it was going to happen that got me excited. I wanted to be there as the field developed," Bennett recalls. To get the clinical background she needed, she went to Harvard Medical School, where she met her future husband in a first-year anatomy class.

Maguire also came from a family of scientists. His father, Henry, was a dermatologist and is now a cancer vaccine researcher at Penn, and his mother, Elise, worked there as a research assistant. Henry had cataracts and later retinal detachment. When Maguire was in high school, he administered his father's eyedrops—his earliest foray into ophthalmology treatment. In medical school, Maguire worked in a lab that studied retinal diseases caused by inherited gene defects. He remembers asking Bennett at the time if the bad genes could be fixed. "That makes sense," she told him. "Let's do it."

The eye is especially well suited to gene therapy. The cells into which a new gene must be inserted are limited to a small area; the retina contains only a few million cells. What's more, unlike most cells, retinal cells don't divide after a person is 3 months old, so researchers don't have to get the new gene into future generations of cells. That means they don't have to stitch the new gene into the cells' existing DNA, which is replicated when a cell divides. Keeping the therapeutic gene separate from the patient's DNA is safer; in the SCID patients who developed leukemia, the introduced gene was incorporated near a cancer-causing gene and accidentally switched it on. The eye is also immunoprivileged, meaning the immune system tends to ignore foreign material introduced there. A runaway immune response has been a problem in some gene-therapy trials and is what killed Jesse Gelsinger. "We're very lucky with our choice of target organ," Maguire says.

While Maguire trained to become a retinal surgeon, Bennett continued to specialize in research rather than clinical work, following her husband around the country for his internship, residency and fellowship. Complicating matters, they were traveling with toddlers. In their last year of medical school, the newlyweds had their first baby—"our senior project," they call it. Two more children soon followed.

In 1989, during one of Maguire's last training stops, in Royal Oaks, Michigan, Bennett set up makeshift labs in the building next door to the hospital and in the basement of their home. They conducted what they think was the first gene-therapy experiment involving the retina. Using mice and rabbits, they injected a gene for an enzyme found in bacteria. They used a dye to reveal whether the eye cells had built the enzyme, and the experiment succeeded: the animals' retinas turned blue for about two weeks.

At Penn, they published one of the first two papers showing that a virus endowed with a foreign gene could shuttle it into eye tissue, in this case in mice. (This strategy, common in gene-therapy experiments, essentially co-opts the virus' capacity to replicate by injecting its own genetic material into cells.) Bennett and Maguire later inserted therapeutic genes into the eyes of some Irish setters with inherited blindness. But Bennett thought the improvement in the dogs wasn't compelling enough to warrant a human trial. What they needed was a simple, slowly progressing form of blindness that was related to a disease that afflicts people. In 1998, they learned of a breed of briard dog in Sweden with an eye disease that, by a fluke, happened to be caused by one of the genetic mutations found in some patients with LCA.

About 3,000 people in the United States suffer from LCA, which encompasses several different blindness disorders that begin in childhood and are caused by mutations in any of several genes, one of which is called RPE65. It contains the instructions for an enzyme crucial to the retina's light-sensing cells, the rods and cones; the enzyme converts vitamin A into a form that the rods and cones use to make a necessary pigment, rhodopsin. In people who inherit a bad copy of the RPE65 gene from each parent, the rods and cones, deprived of rhodopsin, malfunction and eventually die.

Bennett, Maguire and co-workers used a virus called adeno-associated virus to insert a good copy of the RPE65 gene into three young briards. The AAV virus' two genes had been replaced with the RPE65 gene plus a string of DNA that switches the gene on. The dogs regained enough vision to navigate a maze. "It was fantastically exciting," Bennett says. One dog, Lancelot, became a kind of poster dog for gene therapy, shaking paws with people at press conferences and fundraisers.

On the sidewalk outside her office, Bennett shows off one of the more than 50 dogs they have treated. Venus, a medium-size briard with brown, wiry hair, strains at her leash and would clearly like to race away, but she sits to let a visitor pet her. "When she came here, she couldn't see a thing. She would crouch in a corner or in a cage baring her teeth at people," Bennett says. That was a year ago, before the Bennett/Maguire team treated the then 1-year-old in both eyes with gene therapy. Though still night blind, Venus can now hop over obstacles strewn along a corridor and catch a tennis ball. "Her behavior was transformed," Bennett says. "She's seeing well in both eyes. She's a very happy dog."

When they first tried the treatment in people, Bennett and Maguire didn't expect their patients to improve as much as the dogs. For one thing, the doctors were testing safety first and used low doses of the virus and gene. And their first patients, all from an eye clinic in Italy that tests blind patients for genetic defects, were 26-year-old twins and a 19-year-old. By that age, LCA patients don't have much retinal tissue left. Most are completely blind by age 40.

Still, Bennett and Maguire didn't know what to expect when they treated the first patient, one of the twins. He was anesthetized, then Maguire made several small incisions and replaced the gel-like vitreous material in the man's eye with a salt solution so that Maguire could maneuver a needle through it more easily. Using a microscope, he threaded a hair-thin needle through the white of the eye until it touched the retina. He injected 150 microliters of a solution (a drop the size of a pea) that contained 15 billion copies of the AAV virus with the RPE65 gene. The virus is considered an extremely safe delivery mechanism—it can't replicate on its own, and it doesn't cause disease in humans.

All the same, that night "I didn't sleep at all," says Maguire, who was worried about an immune reaction. To his relief, there wasn't any.

Even though the dose was low, Bennett says the first time she looked at the young man's pupillometry data was "a Eureka moment"—his eye could sense more light. A few months after receiving the experimental therapy, all three patients were seeing more light. Two who could see only hand motions before could read three or four lines of an eye chart.

All three patients' eyesight is still improving, Bennett says. The 19-year-old, who has returned to Italy, no longer needs help to walk around at night.

When Alisha Bacoccini was born, her mother, Eve Skidmore, could tell right away that something was wrong. Alisha seemed to focus only on bright light, like a window or lamp, Skidmore says. She thought her daughter might just need glasses, but the ophthalmologist said the little girl was going blind and nothing could be done. She was diagnosed with a progressive eye disease at 11 months, and a genetic test eventually identified the disease as LCA. As a child Alisha could still see well enough to play soccer with a white ball on green grass. "She was extremely fast," says Skidmore, which made up for not seeing the ball in the air. Around eighth grade, Alisha lost even that limited vision.

Today she can read text on a bright computer screen but not in a book. She works as a massage therapist. If she could see better, her dream job would be to work as a forensic pathologist—she devours Patricia Cornwell novels on tape. Skidmore wishes her daughter could regain enough sight "to see the stars in the sky and a rainbow, because she's never seen that."

Bacoccini says she realizes that her sight may not improve in the gene-therapy study, and could even get worse. She volunteered to take part so she could "help to figure out how to fix blindness," she says.

Three months after Maguire injected Bacoccini's eyes with the viruses carrying the retinal gene, her eyes were ten times more sensitive to light and her peripheral vision had improved, but she could not read an eye chart any better than before. Bennett says there are several reasons the treatment may not be working well for her—for example, the neural circuits between Bacoccini's eye and brain may no longer function properly.

Bacoccini is part of a second phase of the study that gave three LCA patients a larger dose of gene therapy than the first three volunteers received. One of the other patients in Bacoccini's group is a 9-year-old Belgian boy, who has shown some of the most dramatic improvement yet. He can see details of faces for the first time and no longer needs a special magnifying device to see the chalkboard at school. The younger the patient, Bennett and Maguire believe, the better chance the couple has of reversing blindness caused by LCA. Eventually they hope to treat babies.

Biomedical research often involves large teams of collaborators, but gene-therapy studies are an extreme case. Last year's paper in the New England Journal of Medicine announcing the initial success of gene therapy for blindness listed 32 co-authors, from the molecular biologists who designed the virus to the Italian doctors who found the patients. Bennett, the lead author, steers this group from a small office outside her laboratory. The space is crammed with notebooks and folders and decorated with thumbtacked photographs of her three kids, journal covers and a few pictures of Lancelot, now 8 years old and still seeing well.

Maguire claims that his role of giving patients injections is minor: "I just load the trucks." But he is, after all, one of the clinical experts. "With [inherited blindness] diseases, there's a huge emotional overlay," he says. "Doctors have always regarded them as incurable and told patients there is nothing we can do for you. The fact that this seems to be working is extremely exciting."

The success of the LCA trial has brought Bennett and Maguire a lot of attention—"an uncomfortable amount of attention," he says—including invitations from members of Congress to brief them on the work. But the duo seem to take it in stride. Bennett has been fielding a half-dozen phone calls and e-mails a day from blind patients or their parents who have heard about the LCA study. "I answer them all. All of these people are really, really upset about going blind or being blind," she says. To be sure, they are unlikely to fit into the LCA trial because they don't have the right genetic glitch. But she tells them to be tested for blindness genes anyway because a gene-therapy treatment for their disease may surface within a few years.

Soon Maguire and Bennett expect to begin experiments with Abyssinian cats with LCA caused by a gene mutation different from the one they've focused on so far. They're also planning a gene-therapy clinical trial for a form of Stargardt disease, or juvenile macular degeneration, which affects some 25,000 people in the United States and which they've successfully treated in mice engineered to have the disease. Now that it's been shown that gene therapy can be performed safely in the eye, companies are exploring ways to use the technique to treat diseases that aren't necessarily genetic in origin. For instance, introducing a gene that controls blood vessel growth might slow age-related macular degeneration, which afflicts more than ten million Americans.

Despite their high-flying medical successes, Bennett and Maguire drive to work in beat-up, ten-year-old cars. At home, she unwinds by gardening and playing her grandmother's grand piano, and he paints detailed, folk art-style farm scenes—rendering "every blade of grass," Bennett says. ("There's a little obsessive-compulsive disorder," Maguire explains about his hobbies.) Their youngest child has gone off to college, but they care for two dogs, an aquarium of fish and turtles and about 15 finches—Maguire's latest hobby is observing bird behavior. The family "has a high threshold for clutter," Maguire says.

Bennett stays up late at night writing reports and grant applications and planning more experiments. She is as driven as her father was when he worked on the gas laser. "There's this incredible excitement that you're about to break a barrier in something," she says.

Jocelyn Kaiser covers biomedical research and policy for Science magazine.
Stephen Voss recently photographed environmental degradation in China. Both live in Washington, D.C.