WASHINGTON (AP) - 03/Fev/2000

A technique that allows insulin hormone to be stored in cells and then released as needed by a pill eventually may offer a treatment for diabetes that does not require daily injections, researchers say.

The experiments, thus far, have been performed only on mice, but researchers say a system using an implanted insulin gene may be ready for human testing within two years.

In a study to be published Friday in the journal Science, researchers at Ariad Pharmaceuticals in Cambridge, Mass., and at Memorial Sloan-Kettering Cancer Center in New York said the cell engineering technique was able to control diabetes in a group of laboratory mice and is now being tested on larger animals.

Tim Clackson, senior author of the study, said that the technique causes insulin, or some other protein, to clump inside a cell with another protein, forming a molecule that is too large to leave the cell. A drug, given as a pill, breaks up the clump, allowing the insulin to flow into the blood stream in a way that mimics the spurt of hormone normally secreted by the pancreas.

''The amount of protein (such as insulin) that gets released is directly related to the amount of drug that is given,'' said Clackson. ''The more drug you give, the more protein gets released into circulation.''

In diabetes, the technique theoretically would allow a patient to precisely control insulin levels in the blood by a pill. Many diabetics now must control insulin levels by injection.

A common type of diabetes is caused by the failure of the pancreas to produce an appropriate amount of insulin to metabolize glucose, or sugar, levels in the blood stream. Normally, the pancreas releases insulin in response to the detected level of glucose.

Dr. Richard Furlanetto, scientific director of the Juvenile Diabetes Foundation, said the experimental technique ''is very clever science'' but might fall short.

"To be truly useful, it would have to be coupled to a system that would release the hormone in direct response to the levels of glucose in the blood," said Furlanetto.

However, Furlanetto said the technique could be very useful in treating conditions that require periodic secretion, or pulsed release, of some needed protein, such as growth hormone.

In the experiment, Clackson and his colleagues inserted into laboratory cells genes that produce insulin and a protein that naturally clumps, or aggregates, with insulin. Once inside the cells, the genes produce the two proteins. They form clusters that are too large to pass through pores in the walls of the cell compartments.

The engineered cells were then injected into the muscles of mice that are diabetic and normally develop high levels of glucose in the blood.

When these mice were fed a drug that caused the protein clusters to split apart, insulin was released into the bloodstream and glucose levels dropped to normal.

In control mice, which had the engineered cells but were not given the oral drug, insulin did not appear in the bloodstream and glucose levels stayed high.

''The insulin stays in the compartments of the cell and has no toxicity or adverse effects. It just sits there,'' said Clackson. ''Only when the animal receives the drug do the aggregates break apart and then flow into the circulation.''

Clackson said the experiment was only a ''proof of concept'' for the technique. The next step is to transfer genes directly into body muscle cells, a common gene therapy technique. To do this, the target genes will be put into a virus that would be injected and deliver the genes into muscle cells, he said.

Although the initial target of the research is diabetes, Clackson said the protein clustering technique could also be used to deliver any protein that the body needs in spurts. For instance, the technique could be used to administer growth hormone, or proteins that would provide pain relief, appetite control or correct brain chemistry, he said."