(FORTUNE Magazine) – Sweeping his arm as if parting a theater curtain, Richard Garr enters a College Park, Md., lab that's bustling with life--in more ways than one. White-coated technicians are busily dribbling nutrients from pipettes into petri dishes stored by the hundreds in small refrigerators. Seen through a microscope, the contents of some petri dishes look like the bottom of a fashion model's sink: discarded eyelashes floating in pink water. The contents of others look like smashed dragonflies. But the pink water is a bath of amino acids and growth factors, and the "dragonflies" are human nerve cells that have grown from the "eyelashes"--or what Garr, the CEO of NeuralStem Biopharmaceuticals, calls "the world's largest supply of human fetal neural stem cells."

The cells are being bred in Garr's lab, as they are at a growing number of biotechs, to replace neurons lost to otherwise incurable brain disorders ranging from Parkinson's disease to Alzheimer's. If Garr and others working in the field are right, the results could utterly change both the way medicine is practiced and the way medicines are produced in the 21st century.

Scientists have been talking about such a shift--away from conventional drugs toward what is called regenerative medicine--since the human embryonic stem cell was discovered in 1998. At the time, Harold Varmus, then director of the National Institutes of Health, told Congress, "This research has the potential to revolutionize the practice of medicine and improve the quality and length of life. There is almost no realm of medicine that might not be touched."

The reason for his optimism: The replicating embryonic stem cell is the most potent cell in existence. It develops soon after fertilization and then differentiates into about 210 types of tissue-specific fetal stem cells, which turn into the mature cells that make up all of the body's tissues and organs. (NeuralStem's cells, for example, can turn into neurons and the two other kinds of mature brain cells.) Only pockets of stem cells remain after birth, which is why many of our organs can't rejuvenate after serious disease or injury--they're largely made of mature, nonreplicating cells.

Stem cells may offer a multifaceted approach to problems that today's drugs, narrowly targeted as they can be, can't touch. Mature cells communicate by flipping thousands of tiny molecules at one another like organic pinballs, each molecule a different word in the cellular language. Most intractable diseases involve the breakdown of many molecular conversations at once. But the pharmaceuticals industry deals in one molecule, one pinball, at a time--enough, say, to pop open clogged blood vessels, staving off heart attacks for a while, but not enough to block all the converging molecular pathologies involved in heart attacks, let alone in cancer or age-related disorders.

Stem cells are not only entire cells; they are the exquisitely sensitive cells of early development. They speak the cellular language fluently, juggling many molecular messages at once, as they do when building the body. When transplanted, they appear to respond to molecular cries for help. They can react to heart-attack damage by forming both blood vessels and cardiac muscle. They react to neural damage either by changing into and replacing neurons that have died, becoming a seamless part of the brain's conversation with itself, or by issuing molecular instructions, reteaching the brain the language of rejuvenation. Either way, as Human Genome Sciences CEO William Haseltine says, stem cells seem to "remind the body it knows how to heal itself." Harvard neurobiologist Evan Snyder calls them "magic seeds."

Though scientists are years from bringing stem cells into mainstream medicine, how fast they reach this goal could depend as much on politics as on breakthroughs in the lab. Biotechs working on fetal and embryonic stem cells rely heavily on the shared research of academia, which is largely funded by the U.S. government, the world's biggest engine for basic research. Yet fetal stem cells are drawn from aborted fetuses, and embryonic stem cells come from in vitro fertilization clinics, putting stem cells in the middle of America's reproductive-rights debate. Federal funding for human fetal-cell research was banned from 1987 to 1993, and a five-year federal-funding ban on human embryonic-cell research was lifted only last summer. This year, just as a clinical trial involving brain cells grown from stem cells was getting under way, the Bush Administration raised the possibility of reinstating one or both bans. In April the NIH was ordered to cancel a meeting to review embryonic research applications; a decision on funding is expected soon. Although such bans wouldn't directly affect private industry, another one might: There's talk in Congress of criminalizing all human therapeutic cloning, a technique that some biotechs, like Geron, are using to grow neurons from embryonic stem cells.

Political uncertainty has some of America's best scientists threatening to move to Europe, where laws supporting stem-cell research are being passed, says NeuralStem consultant and Harvard University neuroscientist Ole Isacson. Irv Weissman, co-founder of Stem Cells, a biotech in Palo Alto, Calif., says bans would be "devastating." Many fear that U.S. research will be slowed and that millions of lives--and dollars--will be lost. Others are more sanguine, including some 70 members of Congress who are co-sponsoring bills supporting the research. "The data is too powerful," says Harvard's Snyder, "to put the genie back in the bottle."

Few understand the life-and-death stakes of stem-cell politics better than Garr. An unlikely character to help spearhead a medical revolution, Garr was, until six years ago, a commercial real estate developer in Maryland. Then his young son Matthew developed a brain tumor. One day he took Matthew for a play date with a schoolmate whose father, Karl Johe, is a scientist. Garr wanted to explain that Matthew sometimes had trouble keeping his balance. Johe, who had been in the lab all night, answered the door in his bathrobe. The men talked about Matthew's condition. "Someday," Garr recalls Johe saying, "we'll fix that."

Johe's confidence grew out of his research in the lab of NIH neuroscientist Ron McKay, who has published more work on neural stem cells in rodents than anyone in the world. McKay and others had isolated neural stem cells in rodent brains in the early 1990s; in 1993, after the first ban on federal funding was lifted, they began finding stem cells in the brain tissue of human fetuses.

The first human stem cells discovered were blood stem cells, but from the start it seemed that neural stem cells might be more commercially viable. For one thing, they are more easily replicated. And they would be implanted in the brain, which is semi-protected from the immune system. Many other stem cells may have to be either administered with a crippling, lifelong immunosuppressive drug regimen, like any organ transplant, or taken from the patient, replicated, and returned--the expensive way most blood stem cells are transplanted today to reconstitute the blood of chemo patients. But neural stem cells held the promise of being profitably sold off the shelf, to become what Thomas Okarma, CEO of Geron, calls the first "21st-century pill."

At first many neuroscientists were skeptical. They had long believed the adult brain couldn't regenerate. But as they discovered that stem cells can help heal a variety of brain diseases in rats, they began to change their minds. Last November's Society for Neuroscience conference was a stem-cell fest. Top universities bragged that, in adult mammal brains, transplanted neural stem cells could: aid in the repair of severed axons in spinal-cord injury; migrate to Alzheimer's plaques; and restore some movement after stroke and traumatic brain injury. It all represented, says Weissman, who isolated the first human blood stem cells in 1991, "the greatest change in medicine you could possibly imagine."

Even so, neural stem cells were so alive, so willful, that no one had figured out how to control their growth in petri dishes--that is, to turn them into the right kinds of neurons, in the right amounts, every time, a requirement for any commercial development. That kept big pharmaceuticals companies away, as did the fear of future bans, which also scared off some academics. But as early as 1995, Johe believed he had many answers. So he latched on to Garr, an entrepreneur impatient for something, anything, to cure his son. "In industry you don't take forever to study," Johe explains. "You get quickly to the bottom line." That year the men quit their jobs and started NeuralStem.

Everybody assumed there was only one neural stem cell," recalls Garr, 48, sitting in his office, a yellowed newspaper clipping about a child who died of brain cancer taped to the wall behind him. As a result, most researchers thought the trick to getting neural stem cells to differentiate into the various kinds of neurons that make up the human brain was to pelt them with different growth factors in the petri dish. But the right recipes--the ones used by the brain--were elusive.

Johe, now NeuralStem's chief scientific officer, believed he had a way around the problem. Perhaps there were many different stem cells, each with its own predilection. Perhaps the trick was to play fewer tricks. As Garr puts it, "The right cells do the right things on their own."

So instead of working with just one stem cell, Johe worked with many, from different regions of the brain. Sure enough, each became the neurons he wanted. At first, the three different kinds of brain cells would appear in unpredictable proportions. But as he toyed with the cells, he saw what others were starting to see--stem cells exchange cues that profoundly affect their development. "Stem cells talk," says Garr, his face rosy with excitement behind black-rimmed glasses. "They say, 'Hey, don't become a neuron--I'm one.'" So Johe separated them. Voila. More predictable proportions.

A tour of NeuralStem's lab reveals that these living products do more than talk. All around, they are reproducing themselves with the help of scientists continuously feeding them specialized diets of amino acids, vitamins, minerals, iron-transporting proteins, and growth factors. Many neurodegenerative disorders that NeuralStem wants to treat involve the loss of different kinds of neurons. So each petri dish contains a different kind of neural stem cell harvested from a different part of the fetal brain. Here, a dish of mid-brain stem cells to replace dopaminergic neurons lost to Parkinson's; here, a dish of spinal-cord stem cells to replace motor neurons lost to Lou Gehrig's disease or spinal-cord injury; here, a dish of hippocampal cells for Alzheimer's; here, a dish of cortical cells for Tay-Sachs.

All must be monitored and tweaked, Johe explains, or "they can shift in their nature." But NeuralStem believes it can control them. "We can make a number of different neurons, in virtually the same numbers, every time, in a dish," says Garr. The company has filed a patent claiming its cells can now double 60 times--a loose threshold for commercialization--making billions of cells from one.

NeuralStem, which is testing its cells on diseased rats, isn't the only biotech making progress with fetal neural stem cells. Some still believe there is only one neural stem cell and that Johe's are intermediary, or progenitor, cells. And academics won't comment on the company's claims, since NeuralStem hasn't published them in peer-reviewed journals. Such secrecy is not uncommon in industry, which is more concerned with protecting patents than with sharing research. Still, many object. "They may be right," says Guy McKhann, a neurologist at the National Institute of Neurological Disorders and Stroke. "But one would prefer information like that were published in more conventional ways, so if it's true, one could see exactly how they did it."

Johe's approach has generated excitement all the same. A peer-reviewed paper he published while still in McKay's lab, which backs up some NeuralStem claims, has been cited in more than 180 academic papers. Garr says he was called by "nearly everyone" after a July meeting with the Food and Drug Administration to discuss clinical trials, which are expected to begin in two years. And some NeuralStem claims are echoed by others: Three academic groups will soon report they've been able to replicate fetal neural cells enough times to make them commercially viable.

Talk of outlawing therapeutic cloning worries companies like Geron, which hopes to turn its patented embryonic stem cells into a veritable garden of designer organs and tissues. Geron has the theoretical ability to clone genetically tailored cells using the technique employed in creating Dolly the sheep. The process involves inserting the nucleus of, say, a skin cell from a patient's ear into an egg, letting the egg begin to develop, then plucking out embryonic stem cells genetically identical to the patient's and growing whatever cells he needs. This way there would be no fear of rejection or need for immunosuppressive drugs.

But even if a cloning ban were imposed, Geron could still commercialize its neurons, since human cells going into the brain may not need long-term immunosuppressive therapy. And CEO Okarma is confident Geron will take no direct hits, pointing to the scores of Congressmen lining up behind pro-stem-cell bills. "The momentum of the cells," he says in his Menlo Park, Calif., office, a sleepy Doberman guarding his desk, "is too strong."

Yet fetal and embryonic cells may remain controversial, which is why many scientists are hoping to find an alternative in the few stem cells we carry with us into adulthood. But only one notable company, NeuroNova in Sweden, is focused on adult neural stem cells. This is not just because they are rare but also because, for now, they are only harvested from patients operated on for brain disorders. (Although Stem Cells, in addition to its work with fetal tissue, is starting work on neural stem cells taken from cadavers.) Researchers elsewhere are seeking molecules--Prozac may be one--that stimulate the growth of neural stem cells inside us. And there may be another alternative, a strange phenomenon called trans-differentiation, in which one type of adult stem cell may be morphed into another.

There's been a flood of startling new work in this area. Neuronyx in Malvern, Pa., is turning adult bone marrow cells, once thought only able to turn into cartilage and bone, into neurons. Cryo-Cell, in Clearwater, Fla., has a procedure it claims is even better, since bone marrow stem cells are rare. It says that stem cells from the blood of umbilical cords can be turned into neurons with a growth factor found in the fetal brain. It isn't known how replicable adult bone marrow or cord-blood stem cells are. But this spring a Cedar Knolls, N.J., company called Anthrogenesis claimed that it could cull ten times more stem cells from placentas than from cord blood, and that the cells can form many tissues, including neurons. And two companies--Artecel Sciences of Durham, N.C., and StemSource of Los Angeles--say they have found the ultimate, limitless, noncontroversial source of stem cells: human fat. StemSource says it can turn these cells into muscle, bone, cartilage, and neurons.

But most adult stem-cell work is ten years behind fetal-cell research. Says Doros Platika, CEO of Curis, a Cambridge, Mass., company that has a $25 million deal with Aegera to turn adult and embryonic skin stem cells into neurons: "Extensive fetal research is absolutely central to making adult stem cells." Stem Cell's Weissman agrees: "The greatest mysteries of human development lie in the time interval between the early embryo and the much later point at which organs are formed. What is needed is a thorough understanding of these cellular and genetic events."

Stem cells may offer the promise of more-dramatic cures than molecule-based drugs, but they'll never roll as inexpensively off assembly lines as pills. And they won't soon net $700 million a year per disease--more like $200 million, Garr says. But growing diseased cells in petri dishes may nevertheless generate big bucks for companies like NeuralStem, enough to finance cell-transplant trials that could one day save people like his son.

In another room at NeuralStem, machines with Buck Rogers names like CEO 2000 XL and BioRobot 8000 are breaking down diseased neurons into data, dragging the letters ACGT across computer screens in seemingly infinite combinations. It is here, Garr says, that neural stem cells may find their first application: the marriage of two of science's hottest disciplines, stem cells and genomics. In addition to investing $8 million in the company, Gene Logic, a functional genomics company in Gaithersburg, Md., is paying NeuralStem $7.5 million to create diseased cells. The goal: to sell the genetic data to pharmaceuticals companies for drug-testing research. The money is NeuralStem's first revenue stream. Geron has a similar deal with Celera.

"Have you heard?" Garr asks, referring to recent scientific papers. "Neural stem cells, armed with toxic genes, may track and shrink brain tumors in rats." He glances at his watch, then disappears, off to accompany his son on a new round of radiation treatment. Matthew is now 14, and his cancer has come back.