Despite rampant plaque pathology, most transgenic mice that overproduce amyloidβ (Aβ) fail to capture other hallmarks of Alzheimer’s disease, such as neurofibrillary tangles and neuronal loss. Now, a study in the July 17 Science Translational Medicine suggests that some of these models may be better than previously thought. Scientists led by Mathias Jucker, University of Tübingen, Germany, report that changes in mouse cerebrospinal fluid (CSF) tau and Aβ look strikingly similar to those of people with sporadic or familial Alzheimer’s disease. These animals could help predict how human CSF Aβ and tau may shift in response to drugs in early-stage trials, and may help researchers develop new CSF-based markers, the authors wrote.

“The study suggests that animal models of amyloidosis can help us model aspects of human AD that we didn’t think they could before,” said David Holtzman, Washington University, St. Louis, Missouri.

First author Luis Maia presented this data at the Alzheimer's Association International Conference, which ran July 13-18 in Boston, Massachusetts. His presentation came hot on the heels of a well-received plenary talk by David Borchelt, University of Florida, Gainesville, entitled "Are our animal models good enough?" Borchelt noted that while potential AD therapeutics look impressive in various mouse models, none have translated into an approved treatment. He contended that since their cognitive deficits are subtle and they retain the vast majority of their synapses and neurons, many of these mice model prodromal Alzheimer's rather than symptomatic. Maia agreed in his talk, suggesting that the emerging biomarker data for APP/PS1 and APP23 mice fits the prodromal profile. Other researchers at the meeting challenged that view. For example, Kelly Dineley, University of Texas Medical Branch, Galveston, suggested that the cognitive decline in the mice was more akin to symptomatic AD.

Previous studies reported that CSF monomeric Aβ levels fall as transgenic mice age and develop Aβ plaques (see Kawarabayashi et al., 2001 and Liu et al., 2004). That recapitulates the pattern seen in people who develop AD. As more Aβ deposits in the brain, less of the soluble peptide winds its way into the CSF. However, few researchers thought to look for CSF tau changes, Jucker told Alzforum. This was partly because tau was not expected to show up in CSF since the animals have no neurofibrillary tangles and little neurodegeneration, and partly because the teeny volume of mouse CSF (about 5-10 uL) and the low concentration of tau made the protein difficult to measure. Maia and co-first author Stephan Kaeser used a highly sensitive antibody-based chemiluminscent assay developed by the company Meso Scale Discovery to detect both tau and Aβ.

The researchers used a method previously developed in Holtzman’s lab to almost double the yield of mouse CSF. It keeps anaesthetized animals alive long enough to drain the fluid twice. Since the protocol requires animal sacrifice, Maia and colleagues sampled five different groups of mice at different time points over their lifespan. In the APPPS1 mice, Aβ42 fell by half at six months of age and by 80 percent at 18 months. Aβ40 dropped, too, but to a lesser extent: only 45 percent by 18 months. In the APP23 mice, Aβ42 decline began around 16 months, and reached a 60 percent drop by 30 months. In both of these cases, CSF Aβ levels inversely correlated with rising Aβ deposits in the brain as seen by immunocytochemistry and Congo red staining.

To the researchers’ surprise, endogenous mouse tau rose in the CSF of both models shortly after Aβ levels began to drop. For APP/PS1 mice this was at six months, with tau surging to five times above baseline level by 18 months. The upshot was a 45-fold increase in the tau:Aβ42 ratio. In the APP23 mice, tau also rose about five-fold relative to baseline at 24 months, amounting to a 10-fold hike in the tau:Aβ42 ratio. In both types of mice, the rise occurred in the absence of global neuron loss, neurofibrillary tangle formation, or change in total tau in the brain. Many neuritic structures surrounding Aβ deposits contained phosphorylated tau, however. Meanwhile, levels of α–synuclein stayed the same, suggesting no overall release of cytoplasmic proteins from neurons.

These results challenge current assumptions about the cause of rising tau in human CSF, wrote the authors. If it is unrelated to neuron loss or neurofibrillary tangle formation, why does tau end up in the CSF? The authors suggested that neurons might release tau during normal physiological activity (see Pooler et al., 2013), which is stimulated by Aβ. Alternatively, tau could leak from neurites in response to Aβ–induced damage, suggested Jucker. Several other scientists voiced support for the latter hypothesis. Holtzman in particular said he has suspected such a tau-release mechanism for a while, as CSF tau levels tend to be stable in people with non-Aβ-related tauopathies (see Hall et al., 2012 ). If Aβ causes tau release, then anti-Aβ therapies might reduce CSF tau. On the other hand, tau therapies might not lower CSF tau because Aβ would continue to promote its release into the fluid.

CSF profiles in these mice do not exactly match those of humans with mild cognitive impairment or AD, noted Douglas Galasko, University of California, San Diego, in an email to Alzforum (see full comment below). He pointed out that CSF Aβ42 levels fall lower and tau levels rise higher in these mice, and that CSF Aβ40 does not typically fall in humans. “Nevertheless, the time course of these changes is highly relevant to evaluating the impact of treatment interventions, with the hope that predictions used from brain and CSF analysis in mice could help to guide translation into human studies,” he wrote.

The results give the first real clinical significance to these mouse models, Jucker said, because they can help foretell how drugs will affect key CSF biomarkers in people, especially those with the particular familial AD mutations carried by the mice. He plans to observe how CSF Aβ and tau levels change in these models in the presence of BACE inhibitors, γ-secretase inhibitors, and various immunizations. He also plans to develop an assay to detect phosphorylated tau, which his group was unable to measure in this study.

Longitudinal mouse studies would be highly informative. They would possible if researchers developed assays that require less CSF so that small volumes could be drawn from mice without being lethal, said Henrik Zetterberg, University of Gothenburg, Sweden. Zetterberg called for further research to figure out how amyloid stimulates the release of tau in these models.—Gwyneth Dickey Zakaib and Tom Fagan


  1. This is an excellent and carefully conducted study, with implications for extrapolating from transgenic mouse models to human Alzheimer's disease. By characterizing two different transgenic models, one with less-aggressive amyloid pathology, the authors could dissect the time course of changes in the cerebrospinal fluid and the brain. It appears that brain Aβ increases slightly before the drop in CSF Aβ42, and tau increases more or less at the same time. The authors did not present a time course of specific forms of Aβ (e.g., oligomers or plaques), and it would be interesting to see this superimposed. That tau is increased in mouse CSF is interesting, and suggests that there is a degree of damage, with release of tau from neurons or axons, i.e., tau may be a damage marker rather than a neurodegeneration marker—because of the absence of neurotic pathology or tangles in these models. It will be interesting to identify whether phosphorylated tau is also present in the CSF, and its time course.

    There are some differences between the findings and the typical CSF biomarker profile in AD (including familial AD). First, the CSF Aβ42 levels continue to decrease and the tau levels continue to rise to five-fold above baseline. This pattern is not seen to the same extent in mild cognitive impairment (MCI) or AD. CSF Aβ42 in MCI or AD shows relatively small serial declines over 1–2 years, for example. Second, a decrease in CSF Aβ40 is not typical of AD. Nevertheless, the time course of these changes is highly relevant to evaluating the impact of treatment interventions, with the hope that predictions used for brain and CSF analysis in mice could help to guide translation into human studies. The species of tau that are found in mouse CSF and their phosphorylation state will be important to investigate further. Also, the mechanism of tau release in the transgenic mice has not been determined.

Make a Comment

To make a comment you must login or register.


Research Models Citations

  1. APPPS1
  2. APP23

Paper Citations

  1. . Age-dependent changes in brain, CSF, and plasma amyloid (beta) protein in the Tg2576 transgenic mouse model of Alzheimer's disease. J Neurosci. 2001 Jan 15;21(2):372-81. PubMed.
  2. . Longitudinal observation on CSF Abeta42 levels in young to middle-aged amyloid precursor protein/presenilin-1 doubly transgenic mice. Neurobiol Dis. 2004 Dec;17(3):516-23. PubMed.
  3. . Physiological release of endogenous tau is stimulated by neuronal activity. EMBO Rep. 2013 Apr;14(4):389-94. PubMed.
  4. . Accuracy of a panel of 5 cerebrospinal fluid biomarkers in the differential diagnosis of patients with dementia and/or parkinsonian disorders. Arch Neurol. 2012 Nov;69(11):1445-52. PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . Changes in amyloid-β and Tau in the cerebrospinal fluid of transgenic mice overexpressing amyloid precursor protein. Sci Transl Med. 2013 Jul 17;5(194):194re2. PubMed.