Viruses are some of the most insidious creations in nature. They travel light: equipped with just their genetic material packed tightly inside a crystalline case of protein, they latch onto cells, insert their genes, and co-opt the cells’ energy-producing, gene-copying and protein-making machinery, using them to make thousands of copies of themselves. Once formed, the new viruses percolate to the cell surface, pinch off inside minuscule bubbles of cell membrane and drift away, or else they continue reproducing until the cell finally bursts. In any case, they go on to infect and destroy other cells, resulting in diseases from AIDS to the common cold.

Different viruses cause different diseases in part because each virus enters a cell by first attaching to a specific suction cup–like receptor on its surface. Liver cells display one kind of receptor used by one family of viruses, whereas nerve cells display another receptor used by a different viral family, so each type of virus infects a particular variety of cell. Cancer researchers have envied this selectivity for years: if they could only target cancer therapies to tumor cells and avoid damaging normal ones, they might be able to eliminate many of the noxious side effects of cancer treatment.

Some scientists, including ourselves, are now genetically engineering a range of viruses that act as search-and-destroy missiles: selectively infecting and killing cancer cells while leaving healthy ones alone. This new strategy, called virotherapy, has shown promise in animal tests, and clinical trials involving human patients are now under way. Researchers are evaluating virotherapy alone, as well as viruses “armed” with therapeutic genes that enable them to administer traditional chemotherapies solely to tumor cells. They are also developing methods to label viruses, or the cells infected by these viruses, to track the movement of the viral agents in patients.

One of the first inklings that viruses could be useful in combating cancer came in 1912, when an Italian gynecologist observed the regression of cervical cancer in a woman who was inoculated with a rabies vaccine made from a live, crippled form of the rabies virus. Physicians first injected viruses into cancer patients intentionally in the late 1940s, but only a handful appeared to benefit. Twenty years later scientists found that a virus that causes the veterinary disorder Newcastle disease shows a preference for infecting tumor cells and began to try to enhance that tendency by growing the viruses for generations in human cancer cells in laboratory culture dishes. Although critics countered that such viruses could be exerting only an indirect effect against cancer by generally activating an individual’s immune system and making it more likely to detect and kill cancer cells, reports continued to pop up in the medical literature linking viral infection and cancer remission. In the early 1970s and 1980s two groups of physicians described patients whose lymphomas shrank after they came down with measles.

The modern concept of virotherapy began in the 1990s, when researchers genetically modified viruses to selectively replicate in, and kill, tumor cells. Teams led by Frank McCormick of ONYX Pharmaceuticals in Richmond, Calif., and Daniel R. Henderson of Calydon in Sunnyvale, Calif., independently published reports showing they could target virotherapy to human cancer cells grafted into mice, thereby eliminating the human tumors. (ONYX is no longer developing therapeutic viruses, and Calydon has been acquired by Cell Genesys in South San Francisco, Calif.) Both groups used derivatives of an adenovirus, a cause of the common cold that has been intensively explored for virotherapy. (Other viruses under study include herpes simplex, measles, parvovirus, vaccinia, reovirus and the avian Newcastle disease virus.) Adenovirus is appealing in part because researchers understand its biology very well after years of trying to cure colds and of using the virus in molecular biology and gene therapy research. It consists of a 20-sided protein case, or capsid, filled with DNA and equipped with 12 protein “arms.” These protrusions have evolved over millennia to latch onto a cellular receptor whose normal function is to help cells adhere to one another.

Adenoviruses are distinct from the types of viruses usually used in gene therapy to treat inherited disorders. Gene therapy traditionally employs retroviruses to splice a functioning copy of a gene permanently into the body of a patient in whom that gene has ceased to work properly. Unlike retroviruses, however, adenoviruses do not integrate their DNA into the genes of cells they infect; the genes they ferry into a cell usually work only for a while and then are lost. Scientists have investigated adenoviruses extensively in gene therapy approaches to treat cancer, in which replication-defective derivatives of the viruses are exploited as vectors for transferring into cells genes that, for example, make cancer cells more susceptible than normal ones to chemotherapy. In general, tests involving adenovirus vectors have been safe, but regrettably a volunteer died in 1999 after receiving an infusion of adenoviruses as part of a clinical trial to test a potential gene therapy for a genetic liver disorder [see box on page 78].

Gene therapists have been working to tailor adenoviruses and other viral vectors, or gene-delivery systems, to improve their safety and reduce the chances that such a tragedy might occur again. It is perhaps even more essential for researchers such as ourselves, who are investigating virotherapy, to develop safer, more targeted vectors, because virotherapy by definition aims to kill the cells the viruses infect and thereby produce a new generation of infectious viruses, not just insert a therapeutic gene into them. Replicating in the wrong cells could be dangerous.

Homing In on the Target

VIROTHERAPISTS ARE DEVISING two main strategies to make sure their missiles hit their objectives accurately with no collateral damage. In the first approach, termed transductional targeting, researchers are attempting to adapt the viruses so that they preferentially enter, or transduce, cancer cells. The second method, called transcriptional targeting, involves altering the viruses so that their genes can be active, or transcribed, only in tumors [see box on pages 76 and 77].

Transductional targeting is particularly necessary because viruses have not evolved to infect and kill tumor cells. Unfortunately, adenoviruses bind more efficiently to the variety of normal tissues in the human body than they do to most tumor cells. We can reverse this pattern using specially generated adapter molecules made of antibodies that snap onto the arms of the virus like sockets on a socket wrench. By attaching carefully chosen antibodies or other molecules that selectively bind only to a specific protein found on tumor cells, we can render adenoviruses unable to infect any cells but cancerous ones. Once the antibody-bearing virus latches onto a targeted cell, the hapless cell engulfs it in a membrane sac and pulls it inside. As the sac disintegrates, the viral capsid travels to a pore in the cell’s nucleus and injects its own DNA. Soon the viral DNA directs the cell to make copies of the viral DNA, synthesize viral proteins and combine the two into thousands of new adenoviruses. When the cell is full to capacity, the virus prompts the cell to burst, releasing the new viruses to spread to other cells.

The viruses can also be engineered more directly. In this regard, Curiel’s group at the University of Alabama’s Gene Therapy Center has designed adenoviruses that bind to cellular proteins called integrins. These molecules help cells stick to the network of connective tissue, called the extracellular matrix, that organizes the cells into cohesive tissues. Although integrins are also made by healthy cells, cancer cells produce them in abundance as they become metastatic and begin to squeeze through tissue layers and travel throughout the body. The University of Alabama research group has had encouraging results using the engineered viruses in mice bearing human ovarian cancers. The viruses homed in on the ovarian tumor cells and killed them, ridding the treated animals of the disease.

Transcriptional targeting generally takes advantage of genetic switches (promoters) that dictate how often a given gene is functional (gives rise to the protein it encodes) in a particular type of cell. Although each body cell contains the same encyclopedia of genetic information, some cells use different chapters of the encyclopedia more often than others to fulfill their specialized tasks. Skin cells called melanocytes, for instance, must make much more of the pigment melanin than do liver cells, which have little use for the protein. Accordingly, the promoter for the key enzyme for making melanin gets turned on in melanocytes but generally is off in most other body tissues. In the deadly skin cancer melanoma, the gene encoding this enzyme is frequently fully functional, making the tumors appear black. We, and others, have engineered adenoviruses that have a promoter for the enzyme adjacent to genes that are essential for the viruses’ ability to replicate. Although these viruses might enter normal cells, such as liver cells, they can reproduce only inside melanocytes, which contain the special combination of proteins needed to turn on the promoter.

Researchers are currently tailoring adenoviruses with a variety of promoters that limit their activity to particular organs or tissues. In liver cancers, for example, the promoter for the gene α-fetoprotein—which is normally shut down after fetal development—becomes reactivated. Adenoviruses containing that same promoter hold promise for eradicating liver tumors. Scientists led by Jonathan W. Simons, now at the Prostate Cancer Foundation, have tested the approach in men whose prostate cancer recurred following treatment with radiation. The researchers used adenoviruses that had been engineered by Cell Genesys to contain the promoter for prostate-specific antigen, a protein made in abundance by prostate tumors. They administered the virotherapy to 20 men who received varying doses of the adenoviruses. In 2001 Simons and his colleagues reported that none of the men experienced serious side effects and that the tumors of the five men who received the highest doses of the virotherapy shrank by at least 50 percent.

Other Strategies

VIROTHERAPISTS might end up combining the transductional and transcriptional targeting strategies to ensure that the viruses kill only tumor cells and not normal ones. Adenoviruses engineered to contain the promoter for the enzyme that makes melanin, for instance, can also replicate in normal melanocytes, so on their own they might cause spots of depigmentation. And adenoviruses that are designed to bind to receptors on the surfaces of tumor cells can still invade a small proportion of healthy cells. But viruses altered to have several fail-safe mechanisms would be expected to be less likely to harm normal cells. There are no results at present, however, to demonstrate that a combination of approaches makes viruses more targeted.

A further strategy for targeting virotherapy makes the most of one of cancer’s hallmarks: the ability of tumor cells to divide again and again in an uncontrolled manner. Healthy cells make proteins that serve as natural brakes on cell division—notably the retinoblastoma (Rb) protein. As cells turn cancerous, however, the genes that code for one or the other of these proteins become mutated or otherwise inactivated. Certain viruses, including adenoviruses, interfere with the braking mechanisms of a normal cell by making proteins that stick to and inactivate Rb. They do this because they can replicate only in cells that are preparing to divide.

Several research groups and biotechnology companies have engineered adenoviruses that fail to make the Rb blockers. Normal cells, which make these blockers, will stall the replication of these viruses by putting the brakes on cell division. But these viruses will replicate in cells in which the Rb protein is already disabled—cancer cells—and kill them. Curiel and Alvarez are currently conducting a clinical trial of this approach using a virus that is also integrin-targeted, as described above. The first phase I study began in early 2007. In this trial, escalating dosages of a modified adenovirus are injected daily for three days into the abdomens of patients with recurrent ovarian or other gynecological cancers. The objectives are to determine the maximum dose that can be tolerated, the toxicity, the immunological response, and any antitumor effects of the adenovirus. To date, nine patients have been treated with dosages of up to 10 billion virus particles a day. No toxicity has been detected, and more patients are joining the trial.

Researchers are also arming therapeutic viruses with genes that make the cells they infect uniquely susceptible to chemotherapy. The technique involves splicing into the viruses genes that encode enzymes that turn nontoxic precursors, or “prodrugs,” into noxious chemotherapies. In one example, which was reported in 2002, André Lieber of the University of Washington and his co-workers designed adenoviruses to carry genes encoding the enzymes capable of converting innocuous prodrugs into the anticancer compounds camptothecin and 5-fluorouracil. The scientists engineered the viruses so that they could make the enzymes only in actively dividing cells, such as cancer cells. When they injected the viruses and the prodrugs into mice bearing implanted human colon or cervical cancer cells, they found that the viruses reproduced and spread in the tumors.

Such “smart” virotherapies are the vanguard of the future. But physicians will also need to track the activity of virotherapies in a patient’s body to best assess how well the strategies are working and to refine them further. Virotherapists are now teaming with radiologists to establish novel imaging technologies to easily measure how effectively a given virotherapy is replicating.

The imaging strategies involve inserting a gene that governs the production of a tracer molecule into a virus or virus-infected cell. The tracer can be either a fluorescent protein that can be observed directly or one that binds to, or activates and traps, the radionuclides used in standard radiological imaging techniques. The fluorescent protein might work best for cancers that are accessible by an endoscope, such as cancers of the larynx. Physicians could peer into the endoscope and see exactly where the viruses—and, therefore, cancer cells—are by looking for fluorescence. So far the approach has worked best with viruses that do not kill cells, however. Nevertheless, we are convinced that such sophisticated imaging technologies will enable scientists to draw more meaningful conclusions from future clinical trials of virotherapy.

In 1995 gene therapy pioneer W. French Anderson of the University of Southern California School of Medicine predicted in Scientific American that “by 2000... early versions of injectable vectors that target specific cells will be in clinical trials.” These trials indeed began on schedule, as well as some he could not have envisioned then. We envision a substantial role for viruses—that is, therapeutic viruses—in 21st-century medicine.


DIRK M. NETTELBECK heads a research group focusing on virotherapy at the German Cancer Research Center and at Heidelberg University Hospital. He was a molecular biologist and postdoctoral fellow of the German Research Association at the University of Alabama at Birmingham (U.A.B.) from 2000 to 2003. He received his Ph.D. in 2000 from Philipps University in Marburg, Germany. RONALD D. ALVAREZ is professor in and director of the division of gynecologic oncology at U.A.B., and he directs the Women’s Cancer Program of the U.A.B. Comprehensive Cancer Center. He received his M.D. from Louisiana State University in 1983. He has extensive experience in translational gene therapy research for ovarian cancer. DAVID T. CURIEL is director of the division of human gene therapy at the Gene Therapy Center at U.A.B. Curiel, who holds an M.D. and a Ph.D., is Jeanne and Anne Griffin Chair for Women’s Cancer Research at U.A.B. and a professor of gene therapy at the Free University of Amsterdam. He has an equity interest in Birmingham-based VectorLogics.