Scientists are raising the first hope of recovery from an autism-like disease that leaves thousands of girls unable to talk or walk — with an experiment that erased symptoms of the disorder in genetically engineered mice.

Thursday's report challenges the long-held belief that the brain damage from Rett syndrome is permanent, but it comes with a big warning: Researchers do not know how to even try such treatment in people yet, making the discovery one of both optimism and continued frustration.

"The thing that keeps your feet on the ground with this study is it shows the principle of reversibility, but doesn't give you any clue about how to accomplish that," cautioned lead researcher Dr. Adrian Bird, a geneticist at Scotland's Edinburgh University.

Still, parents rejoiced at the research, saying it gave them hope that even if finding actual therapies takes another decade, their severely disabled daughters still may have a chance to improve.

"To think even 10 or 20 years down the road ... she could learn to talk and walk," said Jennifer Endres of her daughter Jillian, 4, who smiles delightedly as she watches other children play but can't even hold a toy in her own hands. "If it can happen in Jillian's lifetime, that means the world to us."

And the work, published Thursday by the journal Science, could affect more than Rett syndrome, because the same genetic culprit plays a role in certain forms of autism and other brain diseases, too.

"In this class of disorders, now we have a great hope that if we figure out a way to manipulate the function of the neurons ... we have a chance of recovery," said Dr. Huda Zoghbi of the Baylor College of Medicine. Zoghbi discovered Rett's genetic culprit in 1999.

But Rett is the most physically disabling of the family of autism diseases. Girls are born healthy but then the culprit gene somehow mutates, destroying speech and normal movement. A classic sign is hands that can do nothing but wring. Many children never walk; those that do have a stiff-legged gait. Symptoms usually appear between 6 and 18 months of age. Patients can live to adulthood, although many die of infections before then.

What goes awry: a gene called MECP2 that is supposed to switch off other genes involved in the maturation of neurons. When MECP2 shuts off in Rett syndrome, girls' brain cells don't die, but they don't keep developing.

The question is whether switching that gene back on could restore neuronal function. Even Bird did not think so — he was stunned by his own team's finding.

First the Scottish researchers switched off MECP2 in male mice by inserting a chemical roadblock into the gene that they could switch on and off with medication. When the gene was switched off, baby mice hardly moved and died within weeks.

The first attempt at switching the gene back on killed half the mice, apparently by flooding their cells with too much of MECP2's protein. So the scientists tried again, this time gradually increasing gene activity — and rescuing all but one mouse. Video shows the treated animals scurrying around just like normal rodents.

The real test was in female mice. Why? The mutated gene is carried on just one of females' two X chromosomes, giving them some normal gene function so they do not show symptoms until they are older. If even adult female mice could be saved, it would signal hope for girls — and it worked.

"We expected that at best we'd get a reduction in the severity of the symptoms. To have them go away virtually completely was a big surprise," Bird said.

The experiments could not test cognitive functions, such as speech, just physical ones.

But "the work is really beautiful," said Dr. Uta Francke of Stanford University, a co-discoverer of culprit MECP2 who has studied and cared for Rett patients for 20 years. "They have very convincing evidence" that damage once thought permanent is reversible.

The Rett Syndrome Research Foundation, a parents' group that co-funded the research, now will push scientists to develop human treatments.

"We were worried there was this narrow window of opportunity within which you had to intervene," said foundation co-founder Monica Coenraads, whose 10-year-old daughter Chelsea has Rett syndrome. The new study dispels that notion, and "gives scientists every reason to explore the next steps on every front, from gene therapy to drug discovery."

Gene therapy probably is a long shot, specialists agree. More practical would be to figure out what other genes MECP2 governs and how to use medications to manipulate those downstream effects, Zoghbi said.