Brave New Pharmacy

By MICHAEL D. LEMONICK

Inside an old factory building in Cambridge, Mass., a remarkable machine with the improbable name Zeus is hard at work. Flexing its two robotic arms, the computer-driven device reaches again and again into a storage area the size of a toddler's crib, where thousands of individual samples of genetic material sit in tiny wells etched into plastic plates, each one identified by a unique bar code. One by one, Zeus searches for a particular code, dips into the corresponding well with a fine, quill-like probe and picks up a minuscule droplet of liquid DNA.

Then Zeus transfers each precious droplet to a nearby sheet of nylon, moistens a designated spot and pivots back to the glass plates to find the next sample on its list. When Zeus is done, the nylon sheet will be spotted with a grid of about 1,000 droplets, forming what researchers call a microarray. Once the machine has created a few dozen of these arrays, they will be rolled up, inserted into glass tubes and doused with radioactive dye and genetic material from a range of human tissue types--from normal, healthy cells to diseased cells representing breast, prostate, lung or colon cancer. Emerging from this experiment will be a set of data points, glowing with eerie phosphorescence, that may someday lead scientists to a new cure for one of the deadliest scourges known to man.

When the human genome was sequenced last year, scientists finally gained access to the full text of God's reference manual: the 3 billion biochemical "letters" that spell out our tens of thousands of genes. These genes, strung out along the 46 chromosomes in virtually every human cell, carry the instructions for making all the tissues, organs, hormones and enzymes in our body.

Once scientists have decoded these instructions--a process already well under way--they should have a better understanding of precisely what happens, down to the molecules within individual cells, when the body malfunctions. And, says Francis Collins, director of the National Institutes of Health's Human Genome Research Institute, "if you understand the genetic basis of a disease, then you can predict what protein it produces and set about developing a drug to block it."

Here in Cambridge, a new industry is quietly taking shape that proposes to do that on a grand scale, as companies with names like Biogen, Genzyme, Genetics Institute and Millennium Pharmaceuticals--Zeus' home--prepare to change forever the way doctors fight disease. They're not alone: spurred by the prospect of scientific glory and enormous profit, big pharmaceutical firms and university and government labs have been joined by scores of new companies, not just in Cambridge but in Montgomery County, Md., Silicon Valley and other high-tech hot spots around the nation. It's a virtual gold rush to mine the mountain of potentially valuable data the genome contains.

The result could be a medical revolution. Until now, doctors haven't actually been fighting illnesses like cancer, stroke and heart disease. Instead they've been intervening at the level of symptoms--the last, visible step in a complex cascade of biochemical events. And they have done it largely by trial and error--finding new medicines in exotic plant extracts, for example, or looking for chemical compounds that resemble existing drugs. The process is so woefully inefficient that the drugs currently available target only 500 or so different proteins in the body, out of the 30,000 or so we're made of. Says Collins: "We've beaten those targets to death."

Even when they have the drugs in hand, doctors have to guess which ones might work for a given patient. To treat high blood pressure, for example, physicians must choose from six different classes of medications--and it's the rare patient who hasn't had to work his or her way through several of them before finding a medicine that works.

But in the new era of genomic medicine, this halting, inefficient approach should give way to something much more rational and systematic. Doctors will treat diseases like cancer and diabetes before the symptoms even begin, using medications that boost or counteract the effects of individual proteins with exquisite precision, attacking sick cells while leaving healthy cells alone, and they will know right from the start how to select the best medicine to suit each patient.

Sifting through the human genome for therapeutically useful gems, though, requires a well-designed search strategy combined with powerful technology. At Millennium, housed in a factory that once stamped out heart-shaped candy boxes for Valentine's Day, that strategy is embodied in Zeus, whose job is to find the handful of genes among the genome's tens of thousands that are key to individual diseases--and thus key to making effective medications.

To make this search as easy as possible, Millennium chief scientific officer Dr. Robert Tepper has chosen to focus on the low-hanging fruit--going first for the most obvious targets. In looking for anticancer drugs, for example, his researchers are concentrating on monoclonal antibodies, a type of biological "smart bomb" that targets cancer cells and leaves normal cells alone. Like all antibodies, these man-made cancer missiles seek out particular receptors--molecules on the cancer cell's surface that help the cell recognize and react to nearby enzymes and proteins. Almost a dozen such drugs are already on the market, including one called Herceptin. It zeroes in on the HER-2/neu receptor that sits on the surface of some breast-cancer cells, blocking the binding of growth factors. For the 30% of tumors involving the receptor, the drug may be helpful.

But Tepper's group wants to go a step further, identifying the one or two or three receptors common to all the major cancers--breast, prostate, lung and colon--and thus create a one-stop superdrug. Before the genome was available, this would have been almost impossible. Now Millennium scientists can take known genetic fragments of cancer-cell receptors and plug them into the genome database posted on the National Institutes of Health's GenBank website, searching for sequences in the genome that match and eventually getting to the genes that regulate cell-surface receptors. Almost immediately, they were able to discard as irrelevant some 23,000 of the genome's 30,000 or so genes.

Subsequently the researchers at Millennium had only 7,000 genes to sift through for those specifically active in cancer cells. For that they needed to compare the gene sequences with living cancer cells. That's where Zeus came in: after its custom-made microarrays had marinated for 18 hours in the genetic stew from human tissue cells, technicians scanned them to see which bits of DNA lighted up the brightest with radioactive dye. By comparing the cancer-covered arrays with those immersed with normal cells, the scientists could see which receptors were active in all the cancers yet inactive in normal cells--in this case, just 200 of the original 7,000. "These are experiments that we could only dream of but could never do before the genome," says Tepper.

But they still had too many targets for drug designers to deal with. To narrow the possibilities further, Millennium scientists took breast-cancer cells from two dozen patients and ran additional array screenings to get a better idea of how prevalent a particular receptor was on breast-cancer cells in the population at large. Then they focused on the most widespread and active among them. That brought the hundreds of choices down to just a few dozen, among which are a handful that are expressed in more than 80% of patients.

In just three months, Millennium had finished a winnowing process that would once have taken five or 10 years. Says Tepper: "Drug discovery could never be done this way before. You wouldn't know that a drug was effective or potentially effective in a given percentage of your patient population until very late in clinical development."

Once genomics has identified a potential target protein on cancer cells, scientists still have to find or create a compound--the monoclonal antibody--to lock onto that target and block its normal activity, or at least stick a red flag on it to make it vulnerable to destruction by the body's immune system. At this point, Millennium's process finally begins to look like the "wet lab" that drug companies have relied on for decades. To come up with a monoclonal antibody to fight cancer, Tepper's group uses a strain of mice whose immune systems are genetically engineered to generate human antibodies. Choosing whichever receptor protein Zeus has found for them, the scientists inject the mice with it, then extract the antibodies the animals create to fight the invader.

The antibodies then go through testing to make sure they will bind to cancer cells with the designated receptor, that they can be absorbed by the body and that they won't have toxic side effects. Some of these studies can be done in the lab, but they quickly move into animal and finally human subjects. Already, Millennium has 40 potential targets for monoclonal-antibody drugs against various cancers, and Tepper's goal is to generate 10 to 12 new ones each year.

Access to the genome has drastically improved the efficiency of another traditional drug-finding strategy--and again, Millennium's approach typifies what other firms are doing. Drug companies have often found new medicines by seeking compounds similar to ones they already know, and since most pharmacologically active compounds are based on proteins--that is, on chemicals manufactured naturally from genetic instructions--at least some of those genes should be hidden in the genome.

In 1998, Tepper's team used this reasoning to try to improve on the popular blood-pressure-lowering drugs known as ACE inhibitors. These compounds inhibit an enzyme called angiotensin-converting enzyme (ACE), which is responsible for making the muscle cells in blood vessels contract, which drives blood pressure up. By interfering with the activity of this enzyme, ace inhibitors keep blood vessels relaxed and pressure down.

But the ACE inhibitors currently on the market don't work on everyone, and Millennium figured that the genome might help them find a better version. So researchers sat down at their computers, plugged in some genetic sequences found in the gene for ACE and came up with 10,000 genes that might have comparable activity.

Then they used Zeus to set up microarray analyses and winnowed the 10,000 down to one promising protein they call ACE-2. Testing the enzyme on tissue cells from different organs in the body, the scientists showed that whereas the original ACE acts broadly on many tissues in the body, ACE-2 is particularly active in heart and kidney cells, where it might be more effective in controlling high blood pressure. Because they already knew on the molecular level exactly how ace worked, Tepper's team also knew precisely which lab tests would determine whether ACE-2 had the same effects.

It did, so they moved quickly to develop a compound that inhibits ACE-2. Scientists combed through Millennium's library of 700 different classes of compounds for molecules whose chemistry made them candidates to clamp down on ACE-2 activity. Then, with the help of protein-modeling software (see Bioinformatics box), they manipulated the chemical structure of their new inhibitor to give it optimal binding affinity with the ACE-2 receptor. In about two years, Millennium had created a new blood-pressure-drug candidate that is now being tested in animals.

The last step for the ACE-2 inhibitor, as for any drug, is human clinical trials. Because the Food and Drug Administration requires such rigorous testing, this is by far the most expensive part of drug development. So for human trials in some cases, Millennium has formed partnerships with large pharmaceutical companies that have the necessary resources and will share in any eventual profits.

Everyone looking for new drugs, whether genomically or in more traditional ways, wants to reduce the cost of bringing a medication to market--now estimated at $500 million. One way to do it is to limit trials to those people most likely to respond to a given drug. This too is governed by genetics. Says Ira Herskowitz, a biochemist and biophysicist at the University of California, San Francisco: "We're all different, we have different hair color and different features, right? How can we not metabolize drugs differently?"

That's why Herskowitz and his colleagues have launched a project to unravel exactly what--at the genetic level--makes some people benefit from drugs and others not. They suspect that one major factor is a class of proteins called membrane transporters. These proteins act as molecular gatekeepers, deciding which foreign substances in the bloodstream will be taken into and which rejected by individual cells. If, for example, people lack the gene for an inactivating enzyme, says Herskowitz, "a standard dose of a drug will be more potent. If they have an extra copy of the gene, a standard dose will be inadequate."

To get a handle on how these proteins vary from one person to the next, members of the Pharmacogenetics of Membrane Transporters project are focusing on 25 different transporters already known to play a role in drug absorption and elimination. The first step is to look at the genes for those transporters in DNA samples from 250 ethnically diverse people and see how they vary from one individual to the next. "Identifying the variants is rather easy," says Kathleen Giacomini, the project's principal investigator and UCSF's chairwoman of biopharmaceutical sciences. "The really hard part is in looking at whether the variants have significance for drug response."

That requires working with living cells. The researchers insert different versions of a given gene into a cell and see how its response to a particular body chemical--serotonin, for example, a neurotransmitter implicated in clinical depression--varies. Then they bathe the cells in Prozac, for instance, which works by modifying serotonin levels in the brain, and see how that response changes. "If there's a difference," says Giacomini, "I'll know that maybe your transporter interacts with the drugs a little differently from mine."

As of this month, UCSF researchers have done about 20% of the initial DNA analysis and have found more than a dozen variants, which are now being screened in cells. The scientists on tap to look for variants that haven't been analyzed yet, says Herskowitz, "are chomping at the bit, saying, 'When is my gene going to be done?'"

Clinicians, meanwhile, are assembling a list of 1,500 patients being treated for depression, whose varying responses to medication will be carefully documented. Eventually the clinical data will be combined with the genetic studies. Says Herskowitz: "It's interesting to see the changes to the cell, but what you really want to know is how someone with that change would respond differently to Prozac, or to an anticancer compound. That's more elaborate, which is why this clinical aspect is exciting stuff."

Promising as all these projects seem, they're really only the first stage of the revolution in genomics-based drug discovery. The ultimate payoff of genomics will be a drug resulting from an entirely novel, as yet undiscovered class of compounds. And that will come about only when scientists have assembled a road map laying out not just the functions of individual genes but the dizzyingly complex network of enzyme reactions, receptor interactions and protein-binding patterns that result--not just the building blocks of human life, in short, but the entire working machine. "When we understand that in great, gory detail, we'll be someplace," says Alfred Gilman, a pharmacologist at the University of Texas Southwestern Medical Center at Dallas and winner of the 1994 Nobel Prize for his work on cellular signaling mechanisms.

Gilman and some 50 investigators at 20 different universities have banded together to form the Alliance for Cellular Signaling, whose goal is to trace the maze of chemical pathways in working cells and then use that knowledge to create a "virtual cell" inside a computer. This electronic cell will, in theory, allow researchers to test potential drugs for safety and effectiveness with much less need to resort to mouse, monkey or human subjects. Says Gilman: "You'll be able to take a library of millions of hypothetical chemical compounds and let the computer watch them interact with the theoretical models of drug targets. It will be a fantastic drug-discovery engine for the future."

The task, he admits, is extraordinarily daunting. A typical cell has perhaps 50 different receptors, and the cell doesn't pay attention to just one receptor at a time. "How," asks Gilman, "does it know how to interpret the signal from one hormone when it's listening to 45 other ones at the same time? How does the whole signaling system work as a network? That's what we want to find out."

Leading-edge genomics firms aren't waiting until all the answers are in. Companies like Myriad Pharmaceuticals in Salt Lake City, Utah, Human Genome Sciences in Rockville, Md., and the British company GlaxoSmithKline, along with dozens of others, are moving equally aggressively to plumb the genome for whatever secrets it's ready to reveal.

For its part, Millennium--both on its own and in collaboration with its partners--has identified 121 potential targets, developed 17 drugs currently being tested in animals and moved six drugs into Phase I human trials, four into Phase II and one into Phase III. Says company CEO Mark Levin: "We're going to understand the mechanism of diseases much better, so eventually, obesity, asthma and schizophrenia will be seen not as single diseases but as a subset of 10 to a dozen conditions. That means breakthrough products that have better efficacy and therefore more value to patients than the drugs we have today."

The drugs will have more value for drugmakers as well. With its largely automated, computer-driven searches for new medicines and the cost saving from tightly targeted human trials, Millennium chief technology officer Mike Pavia estimates, the company can cut the cost of developing a new medicine from about $500 million to $200 million while shaving the development time from more than 10 years down to six or seven. These savings, the firm hopes, will translate into pure profit for investors.

That same hope is echoed by dozens of other companies that have jumped into the race to perfect genome-based drugs--and nobody seems to doubt that it will eventually happen. While the genomics revolution hasn't touched most of our lives yet, the day when it will may not be far off. "When it starts to happen, it will happen quickly," predicts Adrian Hobden, president of Myriad Pharmaceuticals. "There will be a few brave pioneers who believe in it, and the vast majority will carry on as they've done before. Then over five years it will become an accepted standard of care, and everyone will be doing it."

--Reported by Dan Cray/San Francisco, Alice Park/Cambridge, Cathy Booth Thomas/Dallas, and Dick Thompson/Washington

TIME.com ON AOL Log on to AOL Wednesday at 7 p.m. E.T. to chat with Michael Lemonick. Keyword: TIME

With reporting by Dan Cray/San Francisco, Alice Park/Cambridge, Cathy Booth Thomas/Dallas, and Dick Thompson/Washington