WASHINGTON - AIDS researchers reported on Monday they had designed a vaccine that they believe may do what no other vaccine has done before -- protect people from infection with the virus.

So far the team at the Institute of Virology at the University of Maryland has only tested monkeys. And they note that people trying to design a vaccine against the AIDS virus have repeatedly failed.

But they think their design, based on the mechanism the virus uses to attach to the immune cells it targets, is the best yet.

"In several animals, including monkey, we were able to generate neutralizing antibodies that are not type-specific but broadly cover various types of HIV," Dr. Robert Gallo, who helped discover the AIDS virus and who heads the institute where the work was done, said in a telephone interview.

Making a vaccine against HIV is difficult because the virus integrates itself into cells, and because it attacks the very immune cells that are normally stimulated by a vaccine.

Although dozens of vaccines are in various stages of development, no one thinks a truly protective vaccine has been designed. At best, doctors now hope that one or more of the vaccines may simply help people to live a little longer with the virus, or perhaps to reduce the ease with which it is transmitted from person to person.

But with 40 million people infected, and no cure in sight, the need is dire. AIDS has already killed 25 million people.

The earliest vaccine approaches used the so-called envelope protein gp120 found on the surface of HIV. The hope was that if the body's immune system could be trained to recognize gp120, antibodies would attack the virus whenever it appeared and neutralize it.

But it did not work well. VaxGen has a vaccine based on this approach in phase III trials, the last stage of development before a vaccine is licensed, but AIDS experts doubt it will protect large populations against infection.

QUICKLY MUTATING VIRUS

Gallo said one problem is that such vaccines work very specifically against the type of virus used in the lab. HIV mutates quickly and studies suggest the gp120 is very different across strains -- different enough to resist immunization.

His institute's team, writing in the Proceedings of the National Academy of Sciences, said they tweaked this approach.

Their vaccine uses gp120, but with a twist.

"When HIV infects a cell, gp120 latches on to CD4," Gallo said. CD4 is a cell receptor, a kind of cellular doorway, found on the immune system T-cells that HIV infects.

"It changes shape when it does that."

The team, led by Timothy Fouts, have broken the gp120 off the rest of the virus as it attaches to CD4. They combined the two joined pieces, CD4 and gp120, to make the vaccine.

"The belly of the beast may be exposed at this point in time," Gallo said.

Tests in macaque monkeys showed they produced antibodies when inoculated with the vaccine. In laboratory dishes, these monkey antibodies neutralized most of the main strains, or clades, of HIV, including HIV-A, B, C, D and E.

The subtypes differ from one another in about 30 percent of their genetic sequence. Subtype B is found mostly in Europe, the Americas and Japan while A, C, D and E are spreading in Africa and Asia, so a vaccine designed in the United States may not work to protect Africans, who are the hardest hit by AIDS.

The next step is to vaccinate monkeys and then see if they are protected from infection by the monkey version of HIV, called SIV, or from an engineered version of the two viruses called SHIV.

"The strategy designed here has worked in laboratory test systems and I have no reason to believe that man should be an exception but one doesn't make claims without data," Gallo said.