September 28, 1998

Technology Puts DNA to Work to Fight Disease-Causing Genes

By LAWRENCE M. FISCHER

Of all the emerging technologies in the biotech firmament, few have been so controversial as antisense, in which small snippets of DNA can serve as drugs by blocking the action of specific genes. With the Food and Drug Administration's approval last month of the world's first antisense drug, this technology took a big step forward in critical acceptance and therapeutic significance.

Antisense is an elegantly simple concept: By knocking out the genes that instruct cells to produce disease-causing proteins, one ought to be able to treat nearly any infection, inflammation or cancer with minimal side effects. But producing and testing antisense molecules has proven both difficult and costly, and the field has been marked by repeated failures. Many scientists questioned whether the technology would ever work at all.

So for Isis Pharmaceuticals Inc., a tiny biotechnology company based in Carlsbad, Calif., the F.D.A.'s approval of its drug on Aug. 24 is a sweet vindication. The F.D.A. approved fomivirsen for the treatment of CMV retinitis, a viral infection that often causes blindness in AIDS patients. Improved AIDS therapies have made this a relatively rare condition, but Isis has other antisense drugs in clinical trials for far more common illnesses, including rheumatoid arthritis, various cancers and the devastating intestinal illness Crohn's disease.

"I think of this as the end of the beginning," for antisense, said Dr. Stanley T. Crooke, a cancer specialist who left a position as head of Smith Kline & French to found Isis in 1989.

Nearly 4,000 people have safely taken antisense drugs as part of trials, he said, and Isis has accumulated a vast library of animal data showing activity against virus, inflammation and cancer, with minimal side effects. "In that context, I don't think any rational person can conclude anything but that antisense works," he said.

Accounts differ as to who came up with the term antisense and when, but it derives from the landmark discovery of the structure of DNA by Dr. James D. Watson and Dr. Francis Crick in 1953. They showed that the genetic code is stored in the nucleus of a cell in a two-stranded form known as a double helix, and that this information is carried to the outer, protein-producing regions of the cell via a single-stranded messenger called mRNA, known as the "sense" strand. Most diseases occur when a pathogen or a gene defect causes a cell to make aberrant proteins.

Watson and Crick showed that DNA consisted of four repeating chemical bases, or nucleotides, often referred to among scientists by letter: A, for adenine; G, guanine; T, thymine, and C, cytosine. They also showed that these bases bound with each other in consistent and repeatable ways -- A to T, G to C -- a process known as Watson/Crick hybridization.

It did not take long for scientists to theorize that a mirror-image sequence of DNA, an oligonucleotide, would bind with and thus neutralize the sense strand -- thus "antisense" -- preventing the production of the encoded protein. But it was not until 1978 that Dr. Paul Zamecnik at Harvard University demonstrated that thus intervening at the mRNA level could produce an antiviral effect. And in the absence of efficient means for synthesizing DNA, his work was not reproduced for many years.

Antisense might have remained a scientific curiosity but for the invention and commercialization of automated DNA synthesis in 1987. Molecules that had required months of skilled lab work could now be obtained in large number at the push of a button. Academic researchers began ordering oligonucleotides by the score to test against various genes, and a handful of start-up companies received venture capital to pursue antisense technology.

But experts now say that the sudden broad availability of oligonucleotides led to a lot of dubious science. Academic researchers "would order two or three antisense oligos and maybe they would order a control and maybe not," said Arthur Krieg, a professor of internal medicine at the University of Iowa and editor of the journal Antisense and Nucleic Acid Drug Development. "The ones that didn't work they'd call a control, and the ones that did maybe worked one time in eight," he said. Just 5 to 10 percent of papers published during this time yielded reproducible results, he said.

At the same time, the fledgling antisense companies were finding the science far more challenging than they had anticipated.

The hurdles were many. Ordinary DNA could not be used as a drug because the body is full of enzymes that break it down, so oligonucleotides had first to be chemically modified, typically by adding sulfur ions. But these sulfurized oligos, called phosphorothioates, are sticky, binding promiscuously to enzymes, proteases and proteins as well as mRNA, so further modification was needed.

No one knew whether oligos would penetrate cells, and once inside, what part of the mRNA they should bind with. A typical sense strand is 2,000 bases in length, an antisense oligonucleotide about 20 bases; pick the wrong complementary sequence for the oligo, and while it may bind, it may not accomplish anything. Isis's scientists found that targeting was crucial, screening dozens of possible sequences to find the one that best suppressed the gene's activity.

In frustration, two other antisense companies, Gilead Sciences Inc. and Genta Inc., shifted their efforts to less-risky technologies, while Hybridon Inc., the company founded by Zamecnik, poured millions of dollars into an H.I.V. therapy only to cancel the project because of dose-related side effects. Isis's first drug, a treatment for the human papilloma virus that causes genital warts, had to be given so often to be effective that it could not be a viable commercial product and was withdrawn from clinical trials.

At times it seemed as though antisense would fade away but for Crooke's Ahab-like fixation on it. "I don't think people ever adequately estimated the complexity of tackling an entirely new drug discovery. They ought to have expected multiple failures," he said. "In truth, in my own lab at Smith Kline we had tried to reproduce some of the early data, and we had failed," he said.

What kept him committed was the promise of the technology to treat chronic diseases, like inflammation and cancer, that are ill-served by conventional drugs. Most drugs are indiscriminate, killing healthy cells along with diseased cells. "I came to the conclusion that if we were to make major advances against chronic diseases, we needed a quantum advance in specificity," he said. "Chronic diseases are by nature minimal deviation diseases because if biology deviates a lot you get acutely ill and die," he said. "We needed drugs that could affect one or two members of a multigene family and not the others."

Some scientists say it is not clear that fomivirsen proves that antisense works, for two reasons: one, because it is injected directly into the eye, it circumvents some of the questions about the ability of oligonucleotides to reach a disease site; two, because there is no way short of performing an autopsy to measure gene or protein expression in the eye, there is no proof that it is working by antisense and not some other unknown mechanism of action. Crooke said the company's animal trials have shown unquestionable antisense activity, and that the drugs now in clinical trials for cancer and inflammatory diseases, which are administered systemically, should lay remaining doubts to rest.

In any case, fomivirsen's success as a drug has rekindled interest in the field. An antisense conference in London next month sold out months in advance. It features talks by representatives of more than a dozen companies and academic institutions. Work on so-called second or third generation antisense molecules, which are based on different chemistry and could be given less frequently or in oral formulations, proceeds at Isis, Hybridon, and Inex Pharmaceuticals.

Perhaps the strongest boost to antisense will come from the Human Genome Project, the effort to spell out the sequence of every gene in the body. As the public and private genome centers crank out newly discovered genes, antisense is often the most efficient way to gauge their function and significance.

"This is a technology that fits perfectly with genomics," said Robert Glazer, a professor of pharmacology at Georgetown University Medical Center, who has directed early trials of Isis's anticancer drugs. "The more they know about genes and how they function in the malignant process, the easier it will be to design antisense drugs. When the whole of the human genome is known, there ought to be targets for every malignancy," he said.

Sometimes antisense will be the right drug to hit those targets, and sometimes it will point the way to a different drug, said Paul L. Herrling, director of research for Novartis A.G., the Swiss pharmaceuticals giant, which has both partnered with Isis and is developing its own antisense program. "Antisense is simply the way to knock out the expression of a gene whenever you want, and you can actually turn it on again when you stop treatment," he said. "It can be a tool to find out what that gene does, and in the end it can be the drug, too."