Mice with severe spinal cord injuries regained much of their ability to walk normally after getting injections of stem cells taken from the brains of human fetuses, scientists in California reported yesterday.

The work strengthens recent evidence that various kinds of stem cells -- including some from human embryos and others from fetuses -- have the capacity to nurse injured nerve cells back to health and in some cases even become replacement neurons themselves.

Scientists cautioned that the approach was not ready for testing in patients with spinal cord diseases or injuries. "This is a first step in what has to be a long series of steps to get to anything clinical," said Aileen Anderson, a neuroscientist at the University of California at Irvine, who led the latest work with colleague Brian Cummings.

But at least three companies are racing to become the first to inject their neural stem cells into patients, and some researchers say the first of those studies could begin within the next nine months.

Yesterday, StemCells Inc. of Palo Alto, Calif., whose cells were used in the new mouse study, filed an amended application to the Food and Drug Administration asking permission to start injecting the cells into the brains of infants with Batten disease, a fatal, inherited syndrome that destroys the central nervous system.

The new research, described in the Sept. 27 issue of the Proceedings of the National Academy of Sciences, tracked mice injected with a kind of human stem cells called neurospheres. They are the laboratory-grown progeny of human cells retrieved from the brains of 16- to 18-week aborted fetuses.

Nine days after getting identical spinal cord injuries, each animal received about 75,000 neurospheres in four injections around the injury.

Within a day, the team reported, the cells began to migrate into the injured spinal cord. After 16 weeks, the mice were given tests of agility and leg coordination, and compared with two other groups. Mice that had received the stem cells scored significantly better than similarly injured mice that had not -- and also better than those injected with ordinary skin cells, a test to see whether just any kind of cellular injection might trigger healing. Researchers who scored the tests did not know which mice had received the injections.

The differences were "obvious to the untrained eye," Anderson said, with improvements both in terms of how many weight-bearing steps the mice could take and their ability to place their rear feet precisely where needed to cross a ladderlike bridge.

Microscopic analysis showed that most of the injected cells had turned into two different kinds of cells around the injury, said Anderson, who does not have a direct financial stake in the company but whose team included two members who do. Some became oligodendrocytes, which wrap themselves around injured nerve cells to help them transmit electrical signals. Others turned into neurons themselves.

Very few turned into a third kind of central nervous system cell, astrocytes, which contribute to scar formation and are generally undesirable around injuries.

Moreover, the neurospheres that became new neurons appear to have made connections with nerve cells that survived the initial injury -- a crucial development if those new nerves are really to help.

In a test to see whether the new human cells were really key to the animals' recovery, the team gave some of the recovering mice injections of a toxin that selectively kills human cells. The mice that got the injections regressed in their ability to walk, while those not injected continued to improve.

Two other U.S. companies also say they are close to the goal of testing human neural stem cells as therapies.

Earlier this year, Hans Keirstead and his colleagues, also at the University of California at Irvine, reported that rats with disabling spinal injuries could walk nearly normally again after getting injections with human embryonic, rather than fetal, cells developed by Geron Corp. of Menlo Park, Calif.

Those cells were initially harvested from days-old human embryos and then cultivated under special laboratory conditions that forced them to become immature oligodendrocytes. Once injected into injured spinal cords, the cells matured and wrapped themselves around injured neurons, which often lose those natural coverings as a result of injury-induced inflammation, leaving even intact neurons unable to function properly.

Geron has said it hopes to begin clinical trials in patients next year.

A third company, NeuralStem Inc. of Gaithersburg, is also in the race.

In unpublished research, rats with spinal cord damage improved significantly after getting injections of human fetal spinal cord cells, said neuroscientist Martin Marsala of the University of California at San Diego, who led the studies with NeuralStem's cells. The animals had ischemic paraplegia, a paralysis of the lower body and rear limbs caused by a temporary blockage of blood flow to the spine.

Patients with this syndrome, which can occur when one of the body's large arteries bursts, are not only paralyzed but also suffer from spastic twitches because of the loss of a kind of neuron that normally suppresses those movements. In rat and pig studies, about one-third of the human fetal cells morphed into exactly that type of neuron, resulting in far less spasticity, Marsala said.

NeuralStem has been talking with the FDA with the aim of getting the go-ahead to begin human testing next year.

The FDA has said several questions will have to be answered before such tests can go forward, including whether some stem cells might turn into the wrong kinds of cells after being injected.