Researchers at the University of Tübingen and Novartis have demonstrated that changes in amyloid-β and tau protein levels in the cerebrospinal fluid of two mouse models of Alzheimer's disease track with observed changes in Aβ and tau levels in the CSF of human Alzheimer's patients.
The findings, presented in a paper published this week in Science Translational Medicine, suggest that these mice models could prove useful for testing drug response in early stage Alzheimer's patients and for discovery of new biomarkers for the disease, Luis Maia, a University of Tübingen researcher and author on the paper, told ProteoMonitor.
The study investigated two transgenic strains of mice commonly used in Alzheimer's research. Both strains are designed to overexpress human amyloid precursor protein, APP – one, the APP23 model, by expressing mutant human APP, and the other, the APP/PS1 model, by expressing mutant human APP and presenlin-1.
The two strains, Maia said, are generally considered to be good models of cerebral β-amyloidosis in preclinical Alzheimer's disease, the stage at which amyloid pathology is present in the brain but clinical symptoms have not yet appeared. However, he noted, little work had been done to establish whether these mice demonstrated the changes in Aβ and tau CSF levels characteristic of Alzheimer's disease.
Were it possible to show that these mice strains did demonstrate changes in CSF Aβ and tau mirroring those observed in human Alzheimer's patients, it could significantly increase their utility as translational tools. CSF Aβ and tau are among the most important protein biomarkers in Alzheimer's research, used for purposes including diagnosing the disease, predicting progression from mild cognitive impairment to Alzheimer's, and serving as surrogate endpoints in drug trials. Given this, the ability to study these markers in the APP2 and APP/PS1 mouse models could prove quite useful.
Several previous efforts had been made to investigate CSF Aβ levels in these mouse models, Maia said, although he added that due to the small number of time points used, the studies provided relatively little information on changes in CSF Aβ as a function of disease progression.
No previous work had been done looking at CSF tau levels in these models, Maia noted, suggesting this was perhaps due in part to a lack of good assays as well as the field's focus on amyloid as the primary cause of the disease.
"I think the major finding of our study is that we can clearly see that in CSF tau increases after Aβ deposition starts," he said. "And that occurs without having neurofibrillary [tau] tangles and also without significant neuronal loss."
This lack of neurofibrillary tangles and neuron loss – both characteristics of late-stage Alzheimer's patients – had led researchers to discount the APP2 and APP/PS1 mice as tools for studying tau CSF levels, Maia said. However, he noted, due to the lack of tools for imaging tau tangles in the brains of living human patients, researchers have relatively little information regarding the state of these structures in preclinical Alzheimer's.
"For humans you have pretty well established [positron emission tomography] imaging studies where you can visualize amyloid in vivo, but for tau that is not yet possible," he said. "So there is no clear answer [regarding tau tangles in the brains of preclinical patients.] At the preclinical stage of [Alzheimer's], people are healthy, so there are not many neuropathological studies in those subjects."
He suggested that, based on current theories of Alzheimer's disease, it would make sense for preclinical patients to exhibit amyloid pathology prior to the appearance of neurofibrillary tangles, as in the two mice models.
"If you think of the amyloid hypothesis for Alzheimer's, the Aβ deposition precedes tau in most of the cases," he said. "So I would presume that in that stage of the disease people would not have that much [tau] pathology but would have plenty of Aβ pathology."
In the study, the researchers measured CSF Aβ and tau levels in the APPPS1 at five time points across 18 months and in the APP2 mice at five time points across 30 months. In both cases, they found that Aβ levels decreased inversely to the amount of Aβ deposits in the brain, while tau levels increased – both in a manner that the authors noted "suggests they may follow the same dynamics as predicted in humans."
This finding "has two major implications," Maia said. "One is that if you treat these mice with disease-modifying compounds, you would be able to see how the [mouse] CSF changes in response to those treatments, and then for [human] clinical trials and preclinical patients predict how those biomarkers should behave to see if they can be used as clinical endpoints."
The other, he said, is that the models could be useful as tools for discovery of new Alzheimer's biomarkers. The mice models could be particularly attractive for this purpose because they offer a homogeneous population for biomarker discovery.
"We know what is happening in the [mouse] brain in terms of pathology," whereas in humans "it is very difficult to have a homogeneous population at this [preclinical] point," Maia said.
He and his colleagues now plan to begin testing various Alzheimer's drugs in these mice to observe what happens to their CSF Aβ and tau levels.
These drugs have "been developed and tested in mouse models," Maia noted, but, he said, "it's not known what to expect in terms of [their affect] on [CSF] biomarkers for the disease."
"If one wants to test these in preclinical [human] cases, then one should anticipate what is going to happen [to the CSF markers] and see what … markers one should look for in humans" to gauge drug response, he said.
The researchers have not established any pharma collaborations as of yet, but Maia said he expects the publication of the paper this week will generate interest within the industry.
The APP23 mouse strain is owned by Novartis, while the APPPS1 strain was developed by University of Tübingen researcher Mathias Jucker, senior author on the Science Translational Medicine paper.