If gene therapy ever gets off the ground, it will be becausethe therapists have learned to fire their transgenessquarely into the cells, tissues and organs they aim toinfluence.

That "ground" on which gene therapy stands today isspelled "ex vivo," which means removing cells from thepatient's body, culturing them with desired genesequences, then reinserting the package in hopes it willend up where needed.

Such was the approach in the very first application ofgene therapy, five years ago next month, whichsuccessfully supplied genes encoding adenosinedeaminase to a little girl with severe combinedimmunodeficiency (SCID).

"The ideal way to do gene therapy," said virologist InderVerma, of the Salk Institute for Biological Studies in SanDiego, "will be to put the gene directly into the cell wherethe product it encodes is required.

"How do you do that?" Verma asked. "How do youmanage to put a gene into a liver or heart or lung? Youcan open people up," he observed, "then inject them therewith the gene. That's one possibility."

That possibility, in fact, is being attempted right now atthe National Institutes of Health, where gene therapistsare giving brain tumors direct through-the-skull injectionsof genes programmed to produce tumor-killing cells,delivered by retroviral vectors.

This ploy is plausible because retroviruses can infectdividing cells but not non-dividing ones, such as braintumor cells.

The Future Of Retroviruses

Verma sees a more all-purpose future for retroviruses ingene-therapy. "We have set out to make a retroviralvector specific for the tissue or cell of choice," he toldBioWorld Today, and added, "To use a terrible analogy, itwill be like a radar-homing guided missile, with only onepurpose in life: to find that cell, to achieve truly targetedin vivo gene therapy."

In the current issue of the Proceedings of the NationalAcademy of Sciences (PNAS), dated Aug. 1, Vermareports "Generation of targeted retroviral vectors by usingsingle-chain variable [antibody] fragment: An approachto in vivo gene delivery."

This progress report, he explained to BioWorld Today, isonly part way toward the ultimate goal. So far, he hasconstructed a vector that equips the shell of a mouseretrovirus with the compact nose-cone fragment of amonoclonal antibody programmed to home in on anantigen found mainly in human liver cells _ the target ofhis demo model system.

That antigen is the receptor for human low-densitylipoprotein (LDL), an indicted co-conspirator incholesterol build-up in the body. Verma chose itsreceptor, "which is relatively unique to liver cells," todemonstrate how organ specificity, in this case thecholesterol-clearing liver, can con the construct to itsdestination.

Into this virus-propelled packaging cell line he inserted ademonstrator bacterial gene, which codes for E. coli's b-galactosidase enzyme.

By this approach, Verma predicted, "One day we willtake a receptor that is absolutely unique to hematopoieticcells, such as CD4," which is the T cell's receptor for theAIDS virus, HIV.

To harness the target receptor, Verma had two choices:"We could either take the ligand _ that is, the proteinthat binds to that receptor _ put it on the virus, and makethat construct look as if it is nothing more than the ligand.

"That," he said, "has been a difficult thing to do, becauseligands are very large in size. The retrovirus has verycomplex machinery, and if you start to stick additionalthings on it, it just doesn't take them."

Verma observed, "We have to constantly remindourselves that it took billions of years for these viruses toevolve to where they are."

So he opted for an alternative strategy, involvingantibodies rather than ligands. "But the whole antibody isalso enormous in size," he said, "so we generated arecombinant cDNA which codes for that small fragmentof the antibody directly involved in association with theLDL receptor."

Sticking this programmed binding site onto the retroviralenvelope "converts a normal virus that can go into allkinds of cells into a tissue-directed virus," Vermaexplained, "Now the virus doesn't quite know that `I amdifferent from the rest of myself,' " he added. "All it hasis a thing sticking out, which makes it specific for theliver cell containing the receptor for the antibody bindingsite."

Normally, Verma said, "The mouse virus can't infecthuman cells at all, but equipped with the human genesequence, it can now infect the human LDL receptor."

So far, so good, but "That's only half the story," Vermacontinued, "because retroviruses can't integrate in non-dividing cells. This means we have to do anothermodification." That next step, now under way, is muchmore complicated, because the AIDS retrovirus can infectnon-dividing cells. "How can HIV do it?" he demands."What's so special about them?

"We now have an idea," he said, "what's special aboutthem, though we're not yet at the stage of publication.We're generating viruses that can infect non-dividingcells, in vitro, and are at the point of trying to do it bydirect injection. That takes more time, but I am quitehopeful."

The Ultimate Vector

A combination of the two, of course," Verma concluded,"dividing and non-dividing, will be the ultimate vectorfor gene delivery."

Molecular biologist Dusty Miller, once a postdoc inVerma's lab, is now pursuing gene-therapy research as amember of the Fred Hutchinson Cancer Research Centerin Seattle.

Verma's PNAS paper, Miller told BioWorld Today,"Represents an important step and a reasonableimprovement." But he sees three still-unsolved problemsin the retroviral construct:

* "It's still providing a low titer of viral particles, 10,000reported, whereas a typical human vector might need amillion or 10 million.

* "The vectors will be sensitive to destruction by thehuman complement system, because they're made using amouse viral envelope.

* "But he'll still need that mouse envelope to get thisthing to work. The antibody fragment allows the virus toget close to the cell, then the second envelope, thefunctional one, has to allow them to fuse." n

-- David N. Leff Science Editor

(c) 1997 American Health Consultants. All rights reserved.