In this new world of often unbridled pessimism, it’s worth noting that nobody ever guaranteed you could make a living, let alone a good one, by pushing the limits of technology. Take Nils Lonberg, for instance. Lonberg compares his company’s story to Sleeping Beauty’s, but his analogy is not precise. Sleeping Beauty, after all, was lucky enough to sleep through her ordeal. Lonberg and his colleagues were wide awake through theirs.
Lonberg was just a few years out of graduate school in 1989 when he signed on with GenPharm International, a company developing a class of drugs known as monoclonal antibodies-souped-up versions of the proteins produced by the immune system to fight disease. The goal was to genetically engineer a mouse with a human immune system, one that could be used to generate “fully human” monoclonal antibodies.
His timing could hardly have been worse. A series of promising monoclonal-antibody drugs to cure everything from cancer to severe infections were about to turn out to be high-profile failures. One of these, a drug called Centoxin, made by Malvern, PA-based Centocor, was projected to earn billions as a treatment for often fatal septic shock before the U.S. Food and Drug Administration declined to approve its use in 1992. The FDA decision marked the seeming demise of monoclonal antibodies and, at least for a while, of the prospects of the entire biotech industry.
“The short story,” says Lonberg, “is we hunkered down.” The longer story, like Sleeping Beauty’s, does have a happy ending. In the past four years, Princeton, NJ-based Medarex, which purchased GenPharm in 1997, has blossomed. The monoclonal-antibody business now has some 300 employees, a brand new research facility in Milpitas, CA, and 60 hectares in New Jersey where it’s building a huge development center. It also has a half-billion dollars in cash to make ends meet while it develops its technology and its drugs. “We’re on our way,” says Lonberg, who is now senior vice president and scientific director of Medarex.
The same can be said for monoclonal antibodies in general, which are in the midst of a remarkable revival. Technologies to make monoclonal antibodies have finally come of age, and the drugs themselves are being touted once again as potential cures or treatments for the entire spectrum of human illness.
Since 1997, the FDA has approved 10 monoclonal-antibody drugs-constituting a quarter of all biotech drugs on the market-with combined sales of well over a billion dollars a year. And perhaps more telling, one in every five biotech medicines in development is a monoclonal antibody. Even biotech giant Genentech, a South San Francisco, CA, company founded with the goal of producing enzyme and hormone drugs, now finds monoclonal antibodies filling half its development pipeline.
Indeed, the story of monoclonal antibodies is more than a fairy tale; it’s a lesson in the values of persistence and patience. “When we started with monoclonal antibodies ten years ago, the prevailing wisdom in biotech was, Been there, done that, didn’t work,’” says Paul Carter, a researcher with Seattle-based Immunex who helped launch Genentech’s monoclonal-antibody research in 1990. “Now, everybody and their dog wants to get into antibodies.”
As monoclonal antibodies make a comeback, nobody’s claiming miracle cures anymore. Having passed through the cycle of “Holy Grail to dirty word,” says Lehman Brothers biotech analyst Rachel Leheny, monoclonal antibodies have become a working technology with an established set of strengths and weaknesses.
Monoclonal antibodies are designed by the immune system to bind only to specific target molecules, making them much more precise than typical small-molecule drugs, the relatively simple compounds that have been the staple of the pharmaceutical industry. And unlike other protein therapeutics, which can activate or block only one specific biological process each, monoclonal antibodies can be developed for any target protein or cell type imaginable. This combination of built-in precision and flexibility can mean faster development and lower toxicity. The risk of a potential antibody drug failing in clinical trials because of unwanted side effects is considerably less than it is for small molecules.
“The probability of success is much higher, and the time course of developing them is much quicker,” says Geoff Davis, chief scientific officer at Abgenix, a Fremont, CA-based antibody drug company. Considering that pharmaceutical companies now spend an average of 15 years and $800 million to bring a new drug to market, saving a few years in development and reducing the risk of a drug’s failing in clinical trials can translate into an enormous profit.
The latest boost to the monoclonal-antibody revival comes from the sequencing of the human genome and the burgeoning genomics industry. Suddenly, pharmaceutical and biotech researchers are deluged with genes, tens of thousands of them, many of which may be valuable drug targets. The result is a gold rush mentality, as researchers race to establish which of these genes and their accompanying proteins are the best targets for inhibiting disease processes. Here monoclonal antibodies-in the guise of laboratory tools that will bind to specific proteins and knock them out of action-represent one of the quickest ways to answer those questions. And once a viable target is nailed down, the low risk and precise targeting of monoclonals can make them the easiest drugs to get to market against it. “They cut right to the chase,” says Immunex’s Carter.
The object of the pursuit, the antibodies themselves, are Y-shaped proteins that constitute the immune system’s first line of defense. They will bind to anything the immune system finds unfamiliar and hence potentially dangerous-say a bacterium or virus-and then hold on tight, calling forth the full range of the immune system’s forces to neutralize or destroy the target (see “Mobilizing Immunity”).
Over the course of a lifetime, the human body generates roughly 100 billion different antibodies. In each case, the base of the Y is virtually identical; the arms of the Y differ from antibody to antibody, thus providing the vast variability that maximizes the possibility that the immune system will spot almost any conceivable invader.
Researchers have long envisioned inducing antibodies to cure or treat diseases that the immune system either ignores, such as cancer, or causes, like rheumatoid arthritis, lupus and other autoimmune diseases. Biologists have known for decades that if you immunize a mouse with a human cancer cell, or even a single protein from such a cell, the mouse will respond by generating its own antibodies to fight the foreigner off-in effect, an anticancer antibody. If you inject the same cancer cell over and over and over again, your mouse will generate antibodies exquisitely specialized to target the cancer cells you put in.
Illustration by John MacNeill
It was back in 1975 that Csar Milstein and Georges Khler, immunologists at the Medical Research Council Laboratory of Molecular Biology in Cambridge, England, developed a technology to mass-produce such antibodies and, in the process, launched an industry. Khler and Milstein fused the antibody-producing cells from mice-known as B lymphocytes-with tumor cells that would keep those B cells alive forever in a laboratory.
In the body, immature B cells reside in the bone marrow and contain a full complement of the gene segments that code for antibodies. As the cells mature and migrate into the bloodstream, the segments undergo rearrangement so that each mature B cell makes just one type of antibody-a monoclonal antibody (see “Genes to Antibodies”). Each of Khler and Milstein’s cell lines would pump out an uninterrupted supply of a single monoclonal antibody, depending on the B cell it arose from. In 1984, their discovery earned the pair a share of the Nobel Prize in medicine, by which time there were roughly a thousand companies trying to cash in on the technology. Most of them would fail.
The problem was inherent in the antibody sources available at the time. Antibodies could be obtained by immunizing mice, as with Raritan, NJ-based Ortho Biotech’s Orthoclone, designed to fight organ transplant rejection and the first antibody drug approved by the FDA. (Ortho Biotech has been a Johnson and Johnson subsidiary since 1990.) Or they could be harvested directly from the human victims of a particular disease, as with Centocor’s Centoxin.
The catch was that mouse antibodies are, well, not human. The human immune system still considers them foreign and does its best to fight them off-a response known as the “human antimouse antibody” response, which not only destroys the antibodies but can lead to kidney failure and death.
When Orthoclone came on the market in 1986, the pharmaceutical industry suddenly learned just how bad the response could be. While most recipients were fine (and, in fact, Orthoclone is still in use), some patients had severe reactions. “Here was a drug that was specifically designed to suppress the immune system, and you still got a strong response,” says Lonberg.
As for antibodies harvested directly from human patients, they simply don’t bind to their targets tightly enough to stem disease-or at least that was the case with Centoxin, and the reason its highly publicized clinical trials failed.
Centoxin’s failure, says Lonberg, “caused everyone, at least on Wall Street and big pharma, to throw up their hands in despair with antibodies and walk away.” Eli Lilly was a classic example. Lilly had purchased Hybritech, the original monoclonal-antibody company, in 1986 for nearly $500 million. After building Hybritech up to 1,400 employees, Lilly sold the company in 1995, post-Centoxin, for only a fraction of what it paid for it.
Illustration by John MacNeill
By then, however, salvation was already on the way. It arrived in the guise of antibodies that were considerably less mouse and increasingly human in origin and function. Researchers had begun developing methods to create these blends in the mid-1980s, and the technologies were just beginning to bear fruit around the time Lilly gave up on Hybritech. Chimeric antibodies came first, constructed back in 1984 by joining the gene that generates the constant region of a human antibody, the base of the Y, to the genes that generate the variable regions of mouse antibodies, the outer arms of the Y. The resulting chimera is about a third mouse and two-thirds human. It still binds tightly to the target for which it was designed, while avoiding the bulk of the human antimouse antibody response.
Next out of the research lab were humanized antibodies. Whereas chimerics were one-third mouse, humanized antibodies were less than a tenth. From the mouse came only the very tips of the arms of the antibody Y, just that part of the variable region that binds directly to its target. All the rest was human. But there was a problem: when the tips of a mouse antibody were genetically grafted onto a human framework, the antibody was often unable to grip its target tightly enough.
Cary Queen, a mathematician turned biologist then working at the National Institutes of Health, created an algorithm to analyze the fit between mouse tips and human framework and then figure out just which molecules in the human framework would have to be adjusted, and by how much, to leave the mouse region sitting comfortably atop the arms of the Y and binding tightly to its target. Queen patented the technology and cofounded Fremont, CA-based Protein Design Labs in 1986 to humanize antibodies for its own drug business and for anyone else who might employ its services.
Between chimeric and humanizing technology, antibody drugs started to make it to market, and the industry revival took off. In 1994, the FDA approved the first monoclonal-antibody drug since 1986, Centocor’s chimeric-antibody drug to inhibit clotting following cardiovascular surgery. (Johnson and Johnson acquired Centocor in 1999.) The agency then approved three more monoclonal-antibody drugs in 1997-including San Diego-based Idec Pharmaceuticals and Genentech’s chimeric Rituxan for non-Hodgkin’s lymphoma.
Rituxan and Genentech’s humanized Herceptin, approved the following year for breast cancer, may alone have pushed the antibody business back into the boom phase, simply because they were anticancer therapies and so fulfilled, at least modestly, one aspect of the mid-1980s hype. In any case, approval for five more monoclonal antibodies followed in 1998, and four more since. At the moment, nearly 50 humanized antibodies are in clinical trials, targeting the full spectrum of human diseases from psoriasis to heart disease and cancer.
But that’s just the beginning, as “fully human” antibodies are now hitting the pharmaceutical pipelines. There are two ways to develop these drugs. In the first, researchers remove the full complement of antibody genes from human B cells and transplant them into bacteria-specific viruses known as “phages.” The viruses promptly generate the appropriate antibodies from the newly acquired genes and then “display” the antibodies on their surfaces-one antibody per phage.
“Now we put all those phages into a test tube,” explains David Chiswell, cofounder of Cambridge Antibody Technology in Cambridge, England. “Anytime we want antibodies to a particular target, we essentially dip the target into the test tube-which can be thought of as a library of antibodies-and hook out just that subset of 100 billion antibodies that happens to bind specifically to the target.”
Method number two is the transgenic mouse with a human immune system-Lonberg’s original dream at Medarex, also pursued by competitor Abgenix. Put a target molecule into such a mouse, and you’ll get a human antibody out-no human antimouse antibody response to worry about. To make the mice, researchers cloned the gene segments responsible for generating the millions of possible human antibodies, put them all into mouse embryonic stem cells and grew the mice to maturity. It was a challenge, says Abgenix’s Davis, “because nobody had ever put that much DNA into a transgenic mouse.” They also had to inactivate the genes that produced mouse antibodies, which they did in another set of mice, and then bred the two lines together.
While the transgenic mice are the latest antibody technology to come to market, it’s still an open question whether they’re actually better than any of the other techniques for generating new antibody drugs. “All of these technologies probably produce antibodies that will behave essentially identically when used in people,” says Robert Kirkman, Protein Design Labs’ vice president for business development. “How do you get the best antibody against the given target is the ultimate question. And that is probably not always going to be the same technology.”
Whichever technology ultimately rules the field, it’s safe to say that antibodies will play a huge role in genomics and in the biotech industry. If nothing else, says Lonberg, bringing a monoclonal-antibody drug to market can give pharmaceutical companies a way to tackle diseases while they spend the extra years necessary trying to develop a small molecule that will do the same task.
Drug makers are pursuing this separate track because monoclonal antibodies still have two serious drawbacks compared to small-molecule drugs. One is that they are large proteins. This means that they have to be given intravenously rather than in pill form, so they won’t be chewed up in the digestive system-although researchers hope to soon solve that problem. (The flip side is that they last considerably longer in the human body, which means they can be administered perhaps once a month rather than once a day.) And the second is that they’re expensive to produce.
None of the methods used to generate chimeric, humanized or fully human antibodies can produce antibodies in commercial quantities. At the moment, companies are making antibody drugs by first inserting the gene for a specific antibody into cells culled from hamster ovaries and then growing the cells by the trillions in enormous vats, through a fermentation process not unlike that used to make beer.
The cells in each vat then secrete a single type of antibody that can be harvested from the surrounding fluid every few months. The process is expensive, however, so researchers have been pursuing dramatic biotech solutions to accomplish the same task. In particular, they’re creating genetically engineered fruits and vegetables-say corn, alfalfa or bananas-loaded with the desired antibodies, or even transgenic animals to serve as living monoclonal-antibody factories.
Somewhere in central Massachusetts is a “biopharm production facility” built by Genzyme Transgenics. While the company does not like to say exactly where it is, it says that it looks a lot like any other farm-except for the 2,000 or so resident goats that have been genetically engineered with human genes to express monoclonal antibodies or other large-protein drugs in their milk.
According to Jack Green, Genzyme Transgenics’ chief financial officer, a few monoclonal-producing goats lactating for a year can match the productivity of an entire fermentation vat of hamster cells. And the goats come with the built-in advantage that if a drug suddenly has a bigger market than anticipated, you don’t have to finance a whole new vat at the cost of a few hundred million dollars: you just breed more goats. “The generation time for a new goat is seven months to sexual maturity,” says Green, “so in a year you can get to however many goats and whatever scale of production you desire.”
After a quarter-century of struggle, monoclonal antibodies have survived to become a mature and exciting technology. And the skeptics are long gone. The question now is how much of 21st-century medicine will be dominated by these remarkable molecules. “Suddenly all this looks real,” Lonberg says.
“Monoclonal antibodies are no longer somebody’s fantasy.”
Antibodies on the Attack
Company Drug/Target Disease Stage Ortho Biotech (Raritan, NJ) Orthoclone/Heart, liver and kidney transplant rejection Approved June 1986 Centocor (Malvern, PA)/
Eli Lilly (Indianapolis, IN) ReoPro/Post-cardiovascular-surgery clotting Approved December 1994 Genentech (South San Francisco, CA)/
Idec Pharmaceuticals (San Diego, CA) Rituxan/Non-Hodgkin’s lymphoma Approved November 1997 Centocor Remicade/Crohn’s disease, rheumatoid arthritis Approved August 1998 Genentech Herceptin/Metastatic breast cancer Approved September 1998 Millennium Pharmaceuticals (Cambridge, MA) Campath/Chronic lymphocytic leukemia
Campath/Multiple sclerosis Approved May 2001
Phase II clinical trials ImClone Systems (New York, NY) Erbitux/Various cancers Phase II and III clinical trials Abgenix (Fremont, CA) ABX-CBL/Transplant rejection Phase II/III clinical trials Tanox (Houston, TX) AD-439/HIV, AIDS Phase II clinical trials GlaxoSmithKline (Middlesex, U.K.) SB-240563/Asthma, allergy Phase II clinical trials Medarex (Princeton, NJ) MDX-010/Malignant melanoma, prostate cancer Phase I clinical trials BioTransplant (Charlestown, MA) AlloMune/Non-Hodgkin’s lymphoma, Hodgkin’s disease Phase I clinical trials
Saudi Arabia plans to spend $1 billion a year discovering treatments to slow aging
The oil kingdom fears that its population is aging at an accelerated rate and hopes to test drugs to reverse the problem. First up might be the diabetes drug metformin.
Yann LeCun has a bold new vision for the future of AI
One of the godfathers of deep learning pulls together old ideas to sketch out a fresh path for AI, but raises as many questions as he answers.
The dark secret behind those cute AI-generated animal images
Google Brain has revealed its own image-making AI, called Imagen. But don't expect to see anything that isn't wholesome.
A quick guide to the most important AI law you’ve never heard of
The European Union is planning new legislation aimed at curbing the worst harms associated with artificial intelligence.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.