Nude mice are born that way because they lack the genes that arm the immune system. That condition is called SCID - severe combined immunodeficiency disease. SCID mice make handy research animals because they can't reject the myriad molecules that scientists shove into their bodies.

Human SCID babies are another, more tragic, story. They are born defenseless against the whole world of infectious and antigenic molecules in the environment. Until gene therapy came of age some 13 years ago, these SCID infants were shielded inside plastic "bubbles" against the rare but fatal infections.

Then on Sept. 14, 1990, a team of pioneer gene therapists at the National Institutes of Health put back into the bloodstream of a 4-year-old SCID girl the gene that expresses adenosine deaminase (ADA), the enzyme she lacked. This first-ever gene therapy recipient enrolled in public school a year later, where she suffered no more infections than any of her normal classmates.

Meanwhile a second SCID patient, a girl of 9, began receiving the same gene-restoring treatment. On Sept. 14, 1992, both girls, now 6 and 11, were guests of honor at a two-year anniversary party, organized by the NIH gene therapy staff to celebrate the successful outcomes.

Then that bubble burst, and SCID children the world over began receiving the vector-delivered ADA gene, delivered for the most part by the Moloney murine leukemia virus (MLV). Clinical gene therapy results were favorable.

But then, in January 2003, the FDA placed a "clinical hold" on 27 gene therapy studies after two of 11 SCID children - one only 3 years old - being treated by French researchers developed a unique form of leukemia. Today's Science, dated June 13, 2003, updates the story in an article titled "Transcription start regions in the human genome are favored targets for MLV integration." Its senior author is geneticist/genomicist Shawn Burgess at the NIH's National Human Genome Research Institute in Bethesda, Md.

Crap-Shooting With Retroviral Vector Placement

"I think the punch line of this Science paper," Burgess observed, "is the assumption that the risk for using retroviruses as gene therapy vectors is relatively low, either because the gene integrations they delivered were random, or they were landing in places that would have a chance of not being very appropriate. We demonstrated for the first time that the genetically engineered mouse virus used in gene therapy trials tends to insert itself at the beginning of genes in the target cell, potentially disrupting the gene's normal function.

"When we looked at hundreds of integrations for the two most popular viral vectors, HIV-1 based and MLV based," Burgess recounted, "we realized that none of this high-throughput analysis can be done without the entire human genome being sequenced. We developed a simple laboratory trick for easily identifying the insertion site by capturing and identifying a small bit of the human genome immediately adjacent to where all these hundreds of integrations were landing.

"HIV-1 had a tendency to land in genes," he said, "and MLV preferred landing near the start of where the gene is located. In both cases, they liked to land in genes that were active at the time that the infection took place. I don't think HIV-1 has reached the clinic yet as a gene therapy vector," Burgess added, "but it's already the second most popular way to get genes into cells.

"What was significant," he continued, "was coupling the Moloney murine retrovirus with the French gene therapy trials for treating those SCID kids who came down with leukemia. In situations where they're taking a lot of cells and infecting with a lot of virus, its chances of landing in genes - and misregulating them in some way that is harmful - is much higher than they assumed originally. So there is a distinct and calculated risk that comes with using these vectors that is much greater than originally predicted. That's assuming random integration. Physicians now believe that the children in the French study contracted leukemia because the Moloney MLV inserted therapeutic genes next to a gene known to promote blood cancer.

"This cell-count map was originally done by me and my lab," Burgess went on, "not to understand where viruses land - although it became a direct outcome of this experiment. We used the viruses to infect zebrafish and then where we're mapping those integrations to create a library of knockouts in those preclinical animal models. Zebrafish are our main model, and we intend to use the MLV vector in particular to infect them, and generate mutations, with the opposite outcome of what people hope for in humans. We want to cause problems.

"We realized as we were working on this project that we had a valuable tool for people doing gene therapy, so we made sure we got that information out. We may go on and characterize integration preferences for other kinds of viruses, to see if we can start seeing trends in subclasses. Alternatively, we can actually find retroviral vectors that would be safer, because they avoid gene insertional mutagenesis - which is also possible."

Can Gene Therapy Prevail Over Odds?

"Our general conclusion as to the outlook for gene therapy in the light of these unfortunate and fortunate developments," Burgess reflected, "is that at the moment the retroviruses are still probably the best option we have in that they infect very efficiently. They're among the most efficient genes for DNA delivery we have."

But Burgess made the point that "there's a much stronger cautionary note now in that at least one of the gene therapy trials ended with a very negative outcome - the death of the patient. The cost-benefit analysis becomes much more tilted toward a conservative view where things should be tested more carefully before they go to the clinic. But I still think," Burgess concluded, "that these vectors are probably our strongest option right now."

NHGRI's director, Francis Collins, summed up: "Without the success of the Human Genome Project, knowing precisely where the retrovirus inserted would have been nearly impossible. These are the kinds of laboratory applications for which the finished genome sequence was intended - applications that will end up improving the practice of medicine."