News Feature | Published:

Therapeutic antibodies

Magic bullets hit the target

After decades of disappointment, antibodies are finally emerging as viable — if expensive — drugs. Trisha Gura finds biotech start-ups and pharmaceutical giants rushing to claim a piece of the action.

For more than a century, doctors have dreamt of using antibodies as 'magic bullets' to cure their patients. After all, these highly targeted proteins are among the immune system's key weapons. So if specific antibodies against proteins involved in disease could be produced in bulk, they should be ideal bespoke drugs.

The story so far has been a roller coaster of hope and disappointment. Despite advances in techniques for producing antibodies in the lab, the magic bullets have — until recently — performed poorly in the clinic. But thanks to developments in genetic engineering, cancer biology, immunology and genomics, antibodies might finally be on the brink of realizing their therapeutic potential. “Now everybody is going crazy to make monoclonals,” says Leonard Presta, director of protein and antibody technology at DNAX, a biotech company in Palo Alto, California.

Monoclonals are antibodies mass-produced in the lab to recognize an individual molecular target. To date, the US Food and Drug Administration (FDA) has approved 11 of them — the majority in the past four years — to treat cancer and transplant rejection and to combat autoimmune diseases such as rheumatoid arthritis (see table). At least 400 other monoclonal antibodies are in clinical trials worldwide.

Table 1 Antibodies in action: monoclonal therapies approved by the US Food and Drug Administration
Of mice and men: researchers have produced drugs based on (from the top) mouse, chimaeric, humanized or human antibodies. All have the same basic structure (bottom). Credit: P. JONES & G. WINTER

Antibodies are Y-shaped proteins, consisting of four polypeptides — two identical light and two identical heavy chains (see diagram). The arms of the Y identify and bind to the antibody's specific molecular target, or antigen. If that happens, the portions of the heavy chains that extend into the stem of the Y alert and recruit the other components of the immune system to attack the structure to which the antibody is bound.

The therapeutic appeal of antibodies can be traced back more than a century, when mice were first investigated as a potential source. By injecting mice with infectious agents, scientists aimed to stimulate the production of antibodies targeted against the infection. They hoped that they could then treat people suffering from the same condition by injecting them with the rodents' blood sera. But these crude preparations were ineffective, and the sera sparked adverse immune reactions in some unfortunate patients.

In 1975, however, Georges Köhler and César Milstein of the UK Medical Research Council's Laboratory of Molecular Biology (LMB) in Cambridge raised hopes with their invention of hybridoma technology1, later honoured with a Nobel prize, which for the first time allowed researchers to mass-produce individual antibodies. By fusing antibody-producing cells from immunized mice with antibody-secreting mouse cells derived from a type of cancer called myeloma, they generated hybrid cell lines that could be cloned and cultured indefinitely. Injected into mice, these immortalized cells grew into tumours that could produce large amounts of monoclonal antibodies. Today, improved cell-culture techniques mean that large quantities of monoclonals can be made without the need to grow tumours in mice.

Mouse antibodies worked well in rodent models of disease2. But when doctors started injecting mouse monoclonals into human patients, problems emerged. The patients' immune systems quickly recognized the mouse antibodies as foreign proteins, and generated 'human anti-mouse antibodies' that cleared the mouse proteins from the bloodstream before they had chance to work. In rare cases, the result was a fatal allergic response2.

Less is more

Researchers also soon realized that mouse antibodies could not function normally in people, because the structure of the proteins is subtly wrong. Even if the arms of a mouse monoclonal did manage to capture its corresponding antigen, the antibody's heavy chains could not signal the human immune system to attack the bound target3.

Indeed, only one mouse antibody made it through clinical trials. This monoclonal, called Orthoclone OKT3, was approved in 1986 and is used to help prevent the rejection of transplanted organs. It works by targeting a glycoprotein on the surface of T cells in the immune system that would otherwise recognize the organ as foreign. In effect, Orthoclone OKT3, marketed by Johnson & Johnson, shuts down one arm of the immune system — which helps to explain why it does not get cleared from the bloodstream quickly.

At the time, making completely human monoclonals was impracticable, because there was no human myeloma cell line suitable for making hybridomas. So scientists turned to genetic engineering of both mouse antibody-producing cells and hybridomas, mixing and matching DNA from mouse and human antibody genes in an attempt to make antibodies that would not be rejected by the human immune system. In the first of these chimaeras, genes encoding mouse antibody arms were simply grafted onto the genes for human antibody stems, producing antibodies that were roughly 30% mouse, 70% human. These chimaeric antibodies could communicate with the human immune system4. But problems continued — in many cases, the chimaeras were still sufficiently 'mousey' to be attacked by the immune system5.

The breakthrough came in 1986, when Greg Winter's group at the LMB shaved the mouse component of chimaeric antibodies down to only 5–10% (ref. 6). Winter knew that three loops of amino acids within a part of each antibody arm called the variable region acted as the 'glue' to bind antibody to antigen. So he replaced everything but the genes for these loops with human sequences. The resulting 'humanized' mouse monoclonals seemed to evade the human immune system. “All of a sudden, it looked like maybe these things would become drugs,” says Nils Lonberg, senior vice-president and scientific director of Medarex, a biotech company in Princeton, New Jersey.

As chimaeric-antibody technology and Winter's refined technique transferred to the biotech industry, a healthy pipeline of monoclonals lined up for FDA approval. In 1994, Eli Lilly of Indianapolis gained permission to market the first chimaeric antibody, ReoPro, which lessens the risk of blood clots in patients with cardiovascular disease by targeting a receptor protein on the surface of platelets. And in 1997, Roche of Basel, Switzerland, won approval for the first humanized monoclonal antibody, Zenapax — which combats organ rejection by binding to and inhibiting a receptor on activated white blood cells, which would otherwise stimulate tissue rejection.

The plethora of potential targets being identified by companies working in the fields of genomics and proteomics should fill the pipeline further. But so far, the main focus has been cancer, where a concentrated research effort has already identified many targets.

Solid start

Cooking up a storm: Genentech's new antibody production plant in Vacaville, California. Credit: GENENTECH

After first going after leukaemia and lymphomas, with some success, companies have moved onto solid tumours, which are generally harder to treat. Herceptin, marketed by Genentech of South San Francisco, targets and blocks a growth receptor on the surface of breast cancer cells. In a trial of 469 women with late-stage breast cancer who tested positive for the receptor, a combination of Herceptin and standard chemotherapy caused greater tumour shrinkage and extended the patients' lives by an average of five months, compared with chemotherapy alone7. That, in the world of oncology, is considered a success. “Any incremental therapy, we embrace,” says cancer researcher Rakesh Jain of the Massachusetts General Hospital in Boston.

The effectiveness of monoclonals against cancer might be improved by attaching toxins or radionuclides to the antibodies, to deliver a knockout punch to the targeted cells. In February, for instance, IDEC Pharmaceuticals of San Diego received approval for Zevalin, a radioantibody for use against non-Hodgkin's lymphoma. And some researchers are advocating the use of cocktails of monoclonal antibodies that will simultaneously target different molecules in multiple cell-signalling pathways. Jain has tried to do just that by testing Herceptin with another antibody, also made by Genentech, that inhibits the formation of blood vessels — needed to supply a growing tumour with oxygen and nutrients.

Intriguingly, Jain's team reported in March that Herceptin alone may do more than simply shut down the cancer cells' growth signal — it might also discourage the growth of blood vessels8. To some experts, findings such as these suggest that antibodies may have much broader applications than just inhibiting the function of particular proteins, or labelling them for attack by the human immune system. “You've got to think about antibodies as agents that change the biology of the cell,” says Lee Nadler, senior vice-president of experimental medicine at the Dana-Farber Cancer Institute in Boston.

In that vein, Nadler wonders if antibodies might be used to help boost the effects of chemotherapy or to sensitize cells to radiation. Monoclonals are also being explored for their potential to bind to a receptor and activate it.

In parallel with the clinical developments, techniques for producing and selecting antibodies have also been advancing, allowing the generation of completely human monoclonals. A technology called phage display, for instance, allows researchers to build libraries of human antibody genes and incorporate them into bacteriophages, viruses that infect bacteria. The phages reproduce in cultures of Escherichia coli, and researchers can fish out any desired antibody from the resulting phage 'soup' using the appropriate target antigen tethered to a surface9.

Human behaviour

Meanwhile, Medarex and another biotech company, Abgenix of Fremont, California, are working with mice engineered to have immune systems that are human, as far as their production of antibodies is concerned. Scientists first knocked out the rodents' ability to produce mouse antibodies by deleting regions of the rodents' heavy- and light-chain genes, and then added the equivalent human genes10,11. Now researchers just need to inject the rodents with the antigen of choice and the animals churn out completely human antibodies. “The beauty is that the mouse does everything for you,” says Geoff Davis, chief scientific officer at Abgenix.

At the same time, researchers have not given up on making human hybridomas. Abraham Karpas of the University of Cambridge, UK, has spent years patiently cultivating a human myeloma cell line that can withstand fusion with other antibody-producing cells and still grow properly. He reported success last year12. And in unpublished findings, Michael Neuberger at the LMB has come up with a way to speed up the generation of new antibodies using a line of antibody-producing cells that mutates its genes faster than normal.

But despite the current wave of enthusiasm, some problems remain. Investment in the antibody business recently took a hit as biotech company ImClone Systems of New York came under fire for inadequacies in its application to market a promising antitumour antibody (see 'Box 1 Giving antibodies a bad name'). But the biggest issue is cost. Although antibodies require much less investment in initial research and development than conventional small-molecule drugs, they are hugely expensive to manufacture.

This stems from the cost of scaling up antibody production to meet clinical demand. The proteins are usually generated by cell cultures in bioreactors that have a capacity of 40,000 litres or more. By the time the cells are nurtured, isolated and the antibodies they produce are purified, the cost can run as high as US$1,000 per gram — compared with $5 per gram for typical small molecules produced by chemical synthesis.

In addition, building a 100,000-litre facility, such as the one just completed by Genentech in Vacaville, California, can take five years and cost $400 million. “There is this tension that as more and more antibodies come on line, there is a real problem with making enough,” says Presta, who used to work for Genentech.

Right now, researchers are concentrating on trying to improve the efficiency of antibody production in cell culture. But they hope eventually to move to streamlined cell-free systems — a step that would surely place antibodies in the clinical mainstream. “We have a long way to go before the story is finished,” says Winter.


  1. 1

    Köhler, G. & Milstein, C. Nature 256, 495–497 (1975).

  2. 2

    Winter, G. & Milstein, C. Nature 349, 293–299 (1991).

  3. 3

    Lubeck, M. D. et al. J. Immunol. 135, 1299–1304 (1985).

  4. 4

    Brüggemann, M. et al. J. Exp. Med. 166, 1351–1361 (1987).

  5. 5

    Brüggemann, M., Winter, G., Waldmann, H. & Neuberger, M. S. J. Exp. Med. 170, 2153–2157 (1989).

  6. 6

    Jones, P. T., Dear, P. H., Foote, J., Neuberger, M. S. & Winter, G. Nature 321, 522–525 (1986).

  7. 7

    Slamon, D. J. et al. N. Engl. J. Med. 344, 783–792 (2001).

  8. 8

    Izumi, Y., Xu, L., di Tomaso, E., Fukumura, D. & Jain, R. K. Nature 416, 279–280 (2002).

  9. 9

    Winter, G., Griffiths, A. D., Hawkins, R. E. & Hoogenboom, H. R. Annu. Rev. Immunol. 12, 433–455 (1994).

  10. 10

    Choi, T. K. et al. Nature Genet. 4, 117–123 (1993).

  11. 11

    Fishwild, D. M. et al. Nature Biotechnol. 14, 845–851 (1996).

  12. 12

    Karpas, A., Dremucheva, A. & Czepulkowski, B. H. Proc. Natl Acad. Sci. USA 98, 1799–1804 (2001).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.