The seventh-floor pulmonary ward at the National Institutes of Health in Bethesda, Maryland, looks much like any normal hospital ward. But if you look above the door to Room 7S-255, you'll see some unusual ductwork; beside the door you'll see a pressure gauge. The room has been modified so that the pressure inside is lower than the pressure in the rest of the building. Air can easily flow into the room, but it can leave only through a series of filters that trap even the smallest floating viruses.
In that room, on Sunday, April 18, 1993, pulmonary specialist Ron Crystal and his colleagues at the NIH dripped a virus into the lungs of a person with cystic fibrosis. The virus was carrying a normal version of the gene that's defective in CF patients; the researchers hoped that the virus would infect the lungs' cells and drop off the new gene, and that the new gene would, in turn, cure the cells by correcting their genetic defect. The filters and ducts were in place to ensure that no one else was exposed to these genetically altered bugs. "It was very much like Neil Armstrong and his colleagues coming back from the moon," recalls Crystal.
With that experiment, the researchers took their first, albeit tentative, step toward bringing gene therapy to these desperately ill people. In this early stage of testing, the scientists are more concerned with showing that the therapy is safe than with showing that it can cure. That will have to wait until the next few rounds. But while the experiment is far from over, and its outcome is anything but clear, already it's being considered one of the most important gene therapy experiments to date.
Gene therapy is one of the biggest potential payoffs from the past decade's explosion of research in molecular biology. Until now, however, its hype has far exceeded its accomplishments. In September 1990, researchers from the NIH performed the first gene therapy experiment, on a young girl with a rare but lethal immune system disorder called severe combined immunodeficiency, or SCID. They took blood from the child, then established a culture of her white blood cells, which were genetically unable to manufacture a crucial enzyme; next they inserted a new gene into the white cells; finally they transfused back into the patient her own, modified cells. The process required weeks of labor, not to mention thousands upon thousands of dollars, to treat one patient, one time. Since then, experiments have begun on 18 other diseases. So far, however, not one has resulted in a complete cure.
Crystal and other CF researchers think their strategy may finally allow gene therapy to reach that goal. For one thing, it's much less labor- intensive: the virus does all the work of delivering the genes to the cells, and does it quickly; the other therapies require that at least some cells be removed and cultured so that genes can be placed into them. The new approach is also simple. All the researchers have to do is drip a liquid containing the genetically altered virus into a patient's lungs. If the technique works, the procedure could one day be performed in a doctor's office or even in the patient's home.
Cystic fibrosis is the most common lethal inherited disorder in the United States. Approximately 30,000 people in this country have inherited two defective copies of the CF gene and thus the disease; approximately 1 in 20 Americans has just one copy of the gene and is a carrier of the disease.
People with cystic fibrosis have all sorts of medical problems, but by far the most serious is the buildup of mucus in their lungs. The mucus is a perfect breeding ground for bacteria, and CF patients are particularly vulnerable to respiratory infections. The mucus and the infections cause inflammation in the lungs; over time, the lung tissue loses its elasticity, and its tiny air sacs become increasingly unable to extract oxygen from the air. The average life span of a CF patient is 29 years.
Karen Ferri is 23. Although her cystic fibrosis was diagnosed shortly after birth, her respiratory problems didn't start until she was 12. Some CF patients are just lucky that way. But when the problems hit Karen, they did so with a vengeance. Throughout her adolescence she spent three, sometimes four hours a day thumping herself on the chest to break up the thick globs of mucus, and then inhaling a mist filled with drugs designed to open the air passages and get more oxygen into her blood. Then more thumping and more inhaling. Even with all that, she would wind up in the hospital several times a year with infections that required intravenous antibiotics.
She was too sick to attend classes regularly at Winter Park High School outside Orlando, Florida, so she graduated through the homebound program--near the top of her class, she likes to point out. Then college. It had to be a university with a good CF center nearby, and the University of North Carolina fit the bill. "I worked out a deal with my doctors," she says. "They'd give me a pass so I could go to class, and then I'd come back
But cystic fibrosis was taking its toll on her lungs. She dragged around an oxygen tank wherever she went. Last summer, Karen was nearing the end. Her lungs had all but stopped functioning. On September 16, after 14 months on a waiting list for a transplant, Karen got a new pair of lungs, and a new chance at life. She was lucky: most CF patients will never get her chance, since there simply aren't enough lungs to go around.
Only recently have researchers begun to understand Karen's disease. Cystic fibrosis was--and remains--a somewhat mysterious illness. It affects not only the lungs but also a host of other organs. In the pancreas, the tiny tubes through which digestive enzymes normally flow become blocked with secretions; as a result, CF patients are often condemned to a lifetime of enzyme supplements so they can digest their food. In the liver, secretions can clog ducts and cause the organ to fail. And nearly all males with cystic fibrosis are infertile because their sperm-carrying duct--the vas deferens--becomes obstructed. "It was really hard to understand what might be the defect," says Michael Welsh, a CF researcher at the University of Iowa College of Medicine. "What kind of genetic disease has the pancreas involved, the lungs involved, the male genital tract?"
The answer to Welsh's question is that all these organs have one thing in common: they rely on a type of tissue called epithelium. "The key feature about an epithelium is that it's a sheet of cells," says Welsh, "and it can separate compartments. So in the airways, epithelium separates the air side from the inside of you. In the intestine it separates the stool from the rest of you. In the pancreas it provides the small tubes that the pancreatic fluids flow through."
But what was the problem with the epithelium? Finding that answer took a lot of sweat. Literally. Researchers knew that the sweat of CF patients was unusual: parents of CF babies had mentioned that their children tasted salty when they kissed them. A series of experiments in the 1980s suggested that the problem lay in the mechanisms by which water and salt ions--chloride ions, in particular--make their way through the membranes of the epithelial cells. In the sweat glands, the hypothesis goes, chloride ions are not properly reabsorbed into cells; thus the salty- tasting skin. In the lungs and the rest of the organs, the problem seems to be that too few ions are escaping the cells. The concentration of ions outside the cell helps determine the amount of water entering and leaving: if there are lots of ions concentrated outside the cell, water should rush out to dilute them. But if there are too few, the water stays in and the organs' normal secretions become sticky and dehydrated. "If there's a defect in chloride transport, the fluid secreted by all those organs might be abnormal and might lead to dysfunction," says Welsh. "Clearly there's too much chloride in the sweat, the secretions from the pancreas are thick and sticky, and you get plugging and pancreatic failure; you get failure of the vas deferens in the male genital tract, and, because of the same process, you get lung disease."
Once they knew they were dealing with a chloride transport problem, researchers began to look at the cellular mechanisms that allow ions to travel across membranes. Did the problem lie with some active ion pump, or with some more passive ion channel? The only way to know would be to find the defective gene controlling these transport mechanisms.
By the mid-1980s the search was in full swing. It took a few years, but in September 1989 a team led by Francis Collins at the University of Michigan and Lap-Chee Tsui and Jack Riordan at the University of Toronto finally nabbed the CF gene. It was sitting in the middle of chromosome 7. The researchers dubbed the protein made by their newfound gene the cystic fibrosis transmembrane conductance regulator, or CFTR for short, because they knew it probably had something to do with getting things from one side of the cell's membrane to the other. But just what did this protein look like? What was its structure? What was its function?
Welsh, for one, was convinced that the protein was a chloride transport channel. "It was quite controversial at the time," he recalls, "because, at least when you look at the amino acid sequence, CFTR doesn't look like any other channel." To this day it's still not clear just how a problem with the chloride channel causes all a CF patient's suffering. But Welsh was sure that by correcting this defect he could restore a normal ion flow in cells. He began inserting a normal version of the CF gene into the nucleus of human lung cells, taken from a CF patient, in laboratory dishes. As the cells grew and divided, the ion concentrations stabilized: the cells' genetic machinery was using the normal gene to make a healthy chloride channel. "That was really satisfying," says Welsh. "It tied together the molecular defect with what we were studying physiologically."
It also showed that by fixing the genetic problem, you might fix the physiological problem--the basic concept behind gene therapy. And with cystic fibrosis, that should be relatively easy. "You don't need to express very much of the CF protein to correct the cell," notes Richard Boucher, who heads the CF gene therapy center at the University of North Carolina, one of nine NIH-sponsored CF gene therapy centers around the country. The biggest problem for gene therapists has been getting enough of a normal gene's protein to the place where it's needed. "In contrast," says Boucher, "with CF cells you probably need only between 10 and 100 CFTR molecules per cell to fully restore function of that cell type. So relatively inefficient systems have a chance to work."
But how do you get a gene inside even a fraction of all the billions of cells in the body--inside, say, those cells lining the airways in the lungs? And just where in each cell do you put it?
"There are two ways of thinking about gene transfer in the lung for CF," says Boucher. The first is what's called integrative gene transfer, in which the DNA that makes the normal protein is put directly into a chromosome--ideally, in the spot where the defective gene sits--and allowed to take up residence. Then, when the cell divides, the corrected gene should be passed on to the daughter cells. But lung epithelial cells die after a few months and are shed, so in order for the integrative strategy to correct the problem for life, it would somehow have to target the epithelium's stem cells--essentially immortal cells that give rise to new epithelium.
Getting the gene into the right cell and into the right spot in the cell is the ultimate goal. But with today's technology it's just not possible. For one thing, when you insert a gene directly into a human chromosome, it's almost impossible to know where it will land. "The worry," says Welsh, "is that it's going to set down next to an oncogene and turn it on, or it's going to set down in the middle of a tumor suppressor and turn the tumor suppressor off." Either of those possibilities would trigger a chain of events that could lead to cancer. "And that's a bad deal."
That's why researchers are embracing the less-than-ideal: a temporary fix in which genes are transferred into the cell's nucleus but not into the chromosomes. In this case, the nuclear machinery will read the gene's instructions and make its protein, but the gene won't be copied or passed on to future generations of that cell. "This is the so-called transient strategy," says Boucher, "where you know you're going to have an effect only for periods of weeks to months."
Crystal remembers the day he came up with a way to make use of that strategy. It was in the spring of 1989, and he was in his office on the sixth floor of the NIH Clinical Center when he got a call from Andrea Pavirani, a colleague in France. Pavirani wanted to tell Crystal about his research on a virus for transferring genes into people. The virus was an adenovirus.
That afternoon, as Crystal was heading into Rock Creek Park for his daily run, his thoughts drifted back to his chat with Pavirani, and he started to get excited. "I knew that adenoviruses cause respiratory infection," he recalls. And he knew that cystic fibrosis affects the lungs. "Suddenly it hit me that we might be able to cure cystic fibrosis by putting the CF gene into an adenovirus." The virus would naturally infect lung cells, and once there it could deliver the normal gene. "Viruses exist to remake themselves. The way they do this is by infecting cells and taking over their genetic machinery. Thus a virus is an ideal delivery truck for a new gene."
Once a virus has entered a cell, injected its genes into the cell's nucleus, and used the cell's enzymes to make more copies of itself, the copies can go on to infect other cells. But it wouldn't do to infect people with a virus that might grow wildly once it got inside the body. So Crystal and his colleagues made a defective virus. "You take out a part of the adenovirus that controls the virus's ability to remake itself," he explains. The idea was to remove enough of the adenoviral genome so that the virus wouldn't reproduce--but not so much that it was no longer infectious. "It brings the genetic information--in this case the CF gene-- to the cell, takes it to the nucleus, and deposits it," Crystal explains. "It does this very, very efficiently. But it can't remake itself."
Crystal began work on his ideal delivery truck several months before he even had a gene to transfer. But as it turned out, removing the viral reproduction genes helped solve a problem that cropped up when the cf gene was found: at first, the normal cf gene--at 4,500 bases--was too big to fit into the virus. By the fall of 1991, Crystal's lab had a virus-gene package that would deliver the normal gene to the cells it infected, both in laboratory test tubes and in live animals.
Still, there were safety issues. Crystal and his colleagues spent hours playing the "what if" game. What if there was viral genetic information--reproductive information--left over in the patient from a prior adenovirus infection? Would that allow the disabled virus to start growing again? What if they treated the patient, and then the patient became infected with an adenovirus that just happened to be floating around? Could the viruses recombine in some way? What if the virus caused too much of the protein to be produced? Would that be harmful? What if the virus got into other organs? What if it got into the gonads? And what if the genetically engineered virus managed to escape the patient and infect someone else? After all, normal adenoviruses are pretty infectious. Just walk into a room where someone's sneezing and you can catch one. "We wanted to infect patients at the clinical center at NIH," says Crystal, "but not to infect Bethesda."
Animal experiments eventually satisfied both Crystal and the Food and Drug Administration that the engineered virus would not remake itself, even when a normal adenovirus was present. There also didn't seem to be any harmful effect from making too much protein. But the only way to tell whether the genetically engineered virus could escape from a human patient was to try it and see.
On April 17, 1993, that's exactly what they did in the carefully designed Room 7S-255. They gently squirted the modified adenovirus with its normal CF gene into the nose of a young man with cystic fibrosis and checked to see if he was going to have a bad reaction to the virus. He didn't. So the next day they threaded a flexible tube called a bronchoscope into his lungs and dripped in the virus-gene package. Medical history was being made, and yet the whole process seemed somehow routine. For the patient, it was just another day in the hospital, except that he got pizza for dinner.
The reasons for safety testing soon became clear, however. In September, patient number 3--a woman--was given the highest adenovirus dose to date. "The patient developed a fever first," says Crystal. "It was handled with aspirin or Tylenol, the usual kinds of medications. But then we also began to see a slight decrease in blood pressure, and what we now realize was inflammation in the lungs."
Her symptoms cleared up in a matter of days, but it was an important lesson for Crystal and his team. "We knew that we'd begin to see toxicity at some point, but we didn't know at what level," says Crystal. "We've put a hundredfold higher, even a thousandfold higher in animals. So what this is helping us do is define where that window of efficacy and toxicity is. If you have a headache and you take one-tenth of an aspirin and it doesn't go away, the dose is not effective. You take two aspirin, it's effective and your headache goes away. If you take a hundred aspirin, it could be lethal. The same thing is going to happen in gene therapy. And what we're doing now is defining that therapeutic toxicity window, as we would for any drug."
Two months later, Crystal treated a fourth patient; there was no repeat of patient 3's scary symptoms. He then moved his lab to Cornell University Medical Center in New York, and he's begun another round of trials. Each of the patients will receive two doses--one in the nose, one in the lungs.
Three other teams--led by James Wilson at the University of Pennsylvania, Boucher at the University of North Carolina, and Welsh at the University of Iowa--have also begun using a modified adenovirus to deliver a normal CF gene to cystic fibrosis patients. Each has a slightly different approach to disabling the adenovirus and making the virus-gene package.
Welsh's team, for one, is taking the conservative approach and sticking with treating the epithelium of the nose. "We chose the nasal epithelium for three reasons," he explains. "First, it's easy to get to. You don't have to do bronchoscopy, which can have potential problems of its own. Second, the histology and physiology are the same in the lining of the nose as down in the lung, and most important, the CF defect is the same. And the third reason to go into the nasal epithelium is that this is an unknown risk; it's a new approach. And we thought by going to the nasal epithelium we could minimize any risk to the people who volunteered for this, yet at the same time get critical data that we needed to go forward."
Welsh's initial plans called for only three subjects: he started in midsummer last year and had finished by the end of September. In the nose, and at the doses they were using, there were no toxic side effects from the virus. But even better, Welsh was able to show that the therapy was apparently doing what was intended. "What we found is that the virus carried the DNA into the cell. And the thing we were most excited about is that it corrected the function." Nasal epithelial cells in all three patients began pumping out ions at normal, healthy levels.
Despite their enthusiasm for the potential of adenoviruses as a gene delivery system, the researchers know there are potential drawbacks. The procedure will always have to be repeated: lung epithelial cells are shed about every two to three months, for one thing, and Welsh has found that the new genes' efficacy seems to last only a few weeks. The modified virus will always look like a foreign invader to the immune system, so after repeated exposures, a patient's immune system could learn to drive off the virus before it delivers its load.
So researchers are experimenting with other delivery systems. In England and the United States, they're studying liposomes--a kind of DNA- filled soap bubble that fuses with the cell membrane and can deliver a gene to the nucleus. The advantage of liposomes is that they are not likely to cause inflammation or an immune response. The disadvantage is that they're not very good at getting DNA into cells. At the University of Cincinnati, Jeffrey Whitsett is experimenting with attaching a normal CF gene to a lung protein called surfactant. Physicians are already good at getting surfactants into lung cells--they're routinely given to premature infants to help their lungs develop. Other groups are working with ways to deliver genes through the blood, in the hope that they can cure all a CF patient's problems, not just the respiratory ones.
Some groups are sticking with viruses. Several labs are experimenting with adeno-associated virus, a virus that can insert DNA permanently into a cell's chromosomes--the integrative approach. Researchers say the normal virus inserts its DNA into the same place in the chromosome every time--obliterating worries about turning on tumor genes-- but so far, when modified to carry the CF gene, it appears to lose that ability. Other labs are testing retroviruses--viruses that carry RNA rather than DNA--as a way to transfer the normal CF gene to patients. They haven't had much luck yet.
Karen Ferri has more than a passing interest in all this research. Her new lungs are working fine, but she remembers all too clearly how the old ones felt. Besides, she still has to take enzymes to help her digest her food, and she has friends who are struggling to stay alive.
So she's decided to join the fight against cystic fibrosis. Though still an undergraduate, she's doing a research project in Boucher's CF center. She's working with a newly developed mouse model of cystic fibrosis in order to speed research on the disease. "I'm a very take-charge kind of person," she says.
Crystal thinks Karen's newfound energy is well directed. "Do I think we're going to have a cure for cystic fibrosis?" he asks. "I think we absolutely will. I can't predict when it will happen. I can't predict whether it will be with the systems we envision today. But it will happen."