Science Editor

Karen Hsiao Ashe, professor of neurology and neuroscience at the University of Minnesota, has developed two separate mouse models of Alzheimer's disease. But she is as skeptical of them as anyone.

"I believe that they really model what I would call endophenotypes of Alzheimer's disease - and endophenotypes are traits that represent certain features of the disease, but are insufficient to define the disease in the whole," she told the audience during a recent lecture at the National Institutes of Health titled "Molecular Basis of Memory Loss in Alzheimer disease: Implications for Cures." But "there currently is no model in mice that models the full human disease," Ashe said.

Despite all their limitations, the mouse models developed by Ashe as well as others have allowed key findings about the interrelationship between structural and functional brain changes and cognitive impairment.

Ashe said that for almost 100 years after Alzheimer's disease was first described, the main attention of researchers was focused on the structural consequences of the buildup of amyloid plaques and neurofibrillary tangles, the two main structural calling cards of Alzheimer's.

But in the last 10 to 15 years, due mainly to methodological advances, it has been possible to understand how the proteins underlying plaques and tangles, amyloid-beta and tau protein, respectively, affect the function of the brain, not just its structure. And with that ability has come an increasing realization that cognitive impairment precedes structural damage and actual neuronal loss by years, even decades.

For example, human subjects start having poorer recall on memory tests in their 40s. But Alzheimer's incidence does not really become significant until some 40 years later. Though early-onset cases certainly do exist, people start developing Alzheimer's at a significant rate when they are in their 80s.

Ashe cautioned that the relationship between factual recall on a memory test and a propensity, or lack thereof, toward developing Alzheimer's disease is "completely unclear right now." But what is becoming increasingly clear, she said, is that "Alzheimer's is a chronic disease, that the biological processes that culminate in dementia begin years to decades before any symptoms or signs appear By the time the disease is present, quite a lot of pathology and neurodegeneration has already occurred in the brain."

Ashe's laboratory has developed one mouse strain that is mutant in amyloid precursor protein and another that is transgenic for a version of tau protein that is prone to neurofibrillary tangles. In both, cognitive impairment begins before gross anatomical abnormalities do, causing Ashe and her colleagues, along with a growing number of other researchers, to hypothesize that it is soluble intermediates, not the plaques and tangles themselves, that are the problem. Ashe and her team have named these proteins A-beta* (A-beta-star) and tau*, respectively.

Ashe's beta-amyloidosis mice develop amyloid plaques, and also have what Ashe termed "subtle" memory deficits. But the animals develop no neurofibrillary tangles, and show no evidence of neurodegeneration. For that reason, Ashe considers them models of elderly individuals at high risk, but not mice with Alzheimer's disease.

In the beta-amyloidosis mice, memory problems precede plaque formation by six months, a finding that led Ashe and her team to question whether the former have anything to do with the latter.

But administering A-beta antibodies to the mice was successful at reversing preexisting memory problems in both Ashe's and another mouse strain with amyloidosis, leading the team on a search that ended with A - beta*, an oligomer of amyloid beta.

When Ashe and her colleagues injected A - beta* into healthy mice, those mice were able to find a platform in water, indicating no motor deficits. But they could not remember the location of that platform the next days, demonstrating that A - beta* has an effect specifically on memory.

Human studies, in turn, have indicated that different forms of the oligomers are prominent in different stages of memory dysfunction ranging from aging-related memory loss that is considered normal, to mild cognitive impairment, to outright Alzheimer's Disease, suggesting that different mechanisms may be at play in different phases of the illness.

The only drug that has been tested in all three phases is Cox inhibitors; based on the original observation that patients with rheumatoid arthritis have a reduced risk of developing Alzheimer's, a number of human and animal studies have investigated the effects of Cox inhibitors on dementia and neurodegeneration.

The results are complex, and made even more so by the fact that one major study ended early when Vioxx (rofecoxib, Merck & Co. Inc.) was taken off the market. For now, the upshot is that at least one Cox inhibitor, naproxen, reduces the rate of Alzheimer's development in the long run, though it increases the move from normal to mild cognitive impairment over shorter time periods, lending merit to the idea that Cox inhibitors continue to be worth exploring.

Ashe pointed out, however, that results from animal studies with the models that are currently available do not predict accurately what will happen in human studies. Celebrex (celecoxib, Pfizer Inc.) "doesn't work in humans, and yet it seems to work in mice" leading Ashe to muse that "maybe it's time for us to allow the human data to inform a new generation of mouse-making so that we can... in an iterative way, make some progress."

In the meantime, she said, "APP transgenic mice should be used to guide studies on AD prevention. I think that's legitimate. But I don't think it's legitimate to use them to guide studies of treatment. I think a whole different process has taken place by the time the disease actually develops, and these mice will give you a lot of false hope." n