In the quest to link the two major hallmarks of Alzheimer's disease pathology, amyloid plaques and tau tangles, many studies have focused on the role of extracellular Aβ oligomers. Now, a new study suggests that a different fragment of the amyloid precursor protein (APP) influences tau. In the May 5 Cell Reports, researchers led by Frederick Livesey at the University of Cambridge, U.K., report that the amount of β-CTF correlated with total tau in neurons that were generated from human induced pluripotent stem cells. Supporting the finding, BACE inhibitors, which suppress production of β-CTF, lowered tau protein levels, whereas γ-secretase inhibitors, which allow β-CTF to build, raised tau levels. 

The results point to a previously unrecognized relationship between APP metabolism and tau, and highlight the utility of stem cell models, the authors wrote. “By studying human neurons and starting from defined mutations, you can discover new biology, which allows you to generate testable hypotheses,” Livesey told Alzforum. These young neurons did not contain pathological, insoluble tau, however, leaving it unclear how the findings relate to Alzheimer’s.

Neurons From Stem Cells.

Induced stem cells from a person with an APP V717I mutation differentiate into neurons (blue) that express markers for layer 5 (green) and layer 6 (red) cortical projection neurons. [Image courtesy of Cell Reports, Moore et al.]

Commentators called the results intriguing but preliminary. “This thoughtful study addresses some fundamental questions about how APP biology relates to tau pathology, which has been difficult to study in mouse models,” Asa Abeliovich at Columbia University, New York City, told Alzforum. “If the data replicate, they would support the model that fragments in addition to Aβ can play a role in Alzheimer’s.”

In cell cultures and transgenic mice, extracellular Aβ entices tau to wander into dendrites, where it mediates toxicity at synapses via the kinase Fyn (see Jul 2010 conference newsSep 2010 news). Other work links Aβ oligomers to tau phosphorylation (see Apr 2013 news). Few studies, however, have examined human neurons expressing endogenous levels of Aβ and tau.

To do this, first author Steven Moore differentiated neurons from induced pluripotent stem cells (iPSCs) made from patients with several genetic forms of the disease (see image above). These included three different presenilin 1 mutations, the V717I APP mutation, and an APP duplication. All mutant neurons produced similar amounts and ratios of extracellular Aβ40 and Aβ42, except for increased total Aβ from the APP duplication. However, to the authors’ surprise, neurons with an APP mutation or duplication accumulated about twice as much tau as did PS1 mutant or control neurons. Phosphorylated tau rose in tandem with total tau, and the ratio between the two did not change. The results agree with a previous iPSC study that reported higher tau and phospho-tau in neurons with APP duplications (see Jan 2012 news).

Why would only APP mutations drive up tau? Since Aβ production was the same in all the neurons bar those harboring three copies of APP, the authors wondered if other APP fragments might be responsible for the uptick in tau. Neurons with APP mutations and duplications contained more β-CTF than controls, whereas the PS1 mutants had less, suggesting this fragment could be the culprit. To investigate further, the authors treated control neurons with various compounds that affect APP processing. In the amyloidogenic processing pathway, BACE1 first snips APP, producing the β-CTF fragment (also called C99), and γ-secretase then cuts the fragment to release Aβ. Thus, BACE inhibitors prevent β-CTF formation, whereas γ-secretase inhibitors prolong its half-life. The authors found that a BACE inhibitor lowered tau and a γ-secretase inhibitor pumped it up in control neurons. The γ-secretase inhibitor boosted β-CTF and tau together in neurons from all mutant lines as well. The data suggested that more β-CTF somehow led to more tau.

What about γ-secretase modulators? GSMs are designed to shift the cleavage site of the β-CTF fragment rather than prevent its processing, so they should not increase the concentration of the fragment as γ-secretase inhibitors do. The authors tested one GSM on their cell lines and again found that effects on tau tracked with changes in APP processing. The compound amplified the Aβ40/Aβ42 ratio and suppressed tau, having its greatest effect on both markers in neurons with an APP duplication. Livesey suggested that GSMs increase turnover of the β-CTF fragment, although the authors did not directly quantify this. In one PS1 mutant line, by contrast, the GSM left the Aβ40/Aβ42 ratio unchanged, as expected, since GSMs poorly bind this presenilin. In this case, tau levels didn't budge.

How β-CTF would boost tau levels remains unclear. Tau transcription did not change, suggesting the increase might occur through enhanced translation or a more stable protein. Livesey noted that each tau molecule lasts for many days, meaning that even small changes in turnover can have large effects on the total amount of protein. He is investigating whether APP metabolism influences general protein degradation or tau turnover in particular. Some previous work has linked APP duplications and high β-CTF to a buildup of early endosomes, indicating a blockage in protein degradation (see Jan 2010 news).

Other researchers expressed caution in interpreting these results. “Whether the findings relate to human disease is unclear,” David Holtzman at Washington University in St. Louis wrote to Alzforum (see full comment below). Commentators noted that Alzheimer’s is characterized by hyperphosphorylated, aggregated tau, not simply high levels of normal protein. Abeliovich pointed out, “[These cell culture results] are not exactly what we see in patient pathology, but then again, we don’t know what to expect at the early stages of disease.”

Livesey believes these young neuronal cultures provide a glimpse into the immediate cellular consequences of APP and PS1 mutations. This allows researchers to trace downstream effects and could provide clues to early disease processes. In ongoing work, Livesey sees preliminary evidence that excess tau in APP mutant neurons strays into cell bodies. Mislocalization of tau has been tied to toxicity (see Zempel et al., 2010).

Do these in vitro findings translate to what happens in adult brains? Commentators noted that patients with presenilin mutations accumulate tau tangles, just as do patients with APP mutations and sporadic disease. “One would think that the mechanisms leading to tauopathy would most likely be the same in [people with] different APP or PS mutations,” Holtzman wrote. Livesey will collaborate with pathologists to examine postmortem brains from people with early stage disease and determine whether those with APP mutations built up more tau than those with PS mutations.

Others suggested more controls to ascertain the relationship. Hans Zempel at the German Center for Neurodegenerative Diseases, Bonn, noted that tau levels also serve as a marker for axonal growth, which could be occurring in the young neuronal cultures and might confound the correlation with C99. Future studies could rule this out by including measures of axonal growth and general neuron health, he suggested.

If the cell culture findings hold up and prove relevant to the human disease, the data would hint that BACE inhibitors and GSMs might have an additional mode of action in Alzheimer’s, lowering tau at the same time they suppress Aβ42 production, Livesey said. That would hint that GSMs and BACE inhibitors might tackle both main pathologies of AD at once, at least for people with certain familial mutations, Zempel wrote. “This paper describes an encouraging novel approach that will surely lead to a better understanding of the roles of Aβ and tau in neurons,” he concluded (see full comment below).—Madolyn Bowman Rogers


  1. In this study by Moore et al., the authors chose a very interesting approach to replicate familial AD (fAD) pathology in vitro. They converted fibroblasts of patients with different fAD mutations into induced pluripotent stem cells (iPSCs), which they then differentiated into excitatory forebrain neurons. As expected, these human neurons then showed different processing of APP, resulting in different levels and ratios of Aβ peptides, depending on the mutation. These results correlate nicely with previously published results on APP processing from various labs (e.g. DeStrooper, Haass, Sisodia labs and others), and it is quite a relief that human neurons do show roughly the same behavior as the cell lines and transgenic mouse models used in other studies.

    Interestingly, two of the fAD mutations tested (a duplication of APP and the V171I mutation of APP, both of which change total amounts of APP expression) resulted in higher levels of tau (approximately two- to fourfold) in these iPSC-derived human neurons. Phospho-tau, however, remained unchanged, when compared to total tau expression.

    This is certainly an interesting point that would require replication and a suitable sample size. The direct implication of this result, however, is unclear, because i) the increase in total tau is within the range produced by other mutations and the controls, ii) intraindividual controls are missing and iii) no other cellular parameters have been tested, thereby it is difficult to judge the general state of the cells. E.g.,  increased axonal growth due to higher trophic levels of a trophic APP fragment would also result in higher levels of tau.

    The authors also tested the effects of β- and γ-secretase inhibitors and modulators on tau levels, and found that these can change not just the processing of APP, but also the total tau levels, at least in the case of most samples analyzed. These experiments are unfortunately difficult to judge. These cells are obviously still growing/developing and  it is unclear how this is affected by the drug treatments. The drug dosing is also rather rough, with 1-10µM applied every two days for a period of up to 30 days, which might result in the accumulation of the drug and severe side effects. Finally,  phospho-tau levels only increased or decreased in parallel with total tau. Thus, even if one considers phospho-tau pathological, there seems to be no tau pathology  here.

    Tau phosphorylation and upregulation is a very normal process during development or axon extension, and of course tau is not only a marker for AD-like pathology, but also a neuronal or axonal marker, and thereby also a marker for proper axonal growth (see, e.g., Zempel and Mandelkow, 2014, for discussion). As these parameters (as well as general health markers, etc.) are not part of this study it is difficult to draw conclusions for therapeutic interventions. However, if these findings can be replicated and γ-secretase modulators are capable not only of decreasing pathogenic Aβ peptides, but also reducing tau levels without affecting axonal growth or general neuronal health, then one could tackle the two pathological drivers of AD at the same time, at least for some mutations. Whether this would be beneficial for all mutations of fAD or for other tauopathies is uncertain, therefore there is still a great need to develop tau-based therapeutics. However, this paper describes an encouraging novel approach that will surely lead to a better understanding of the roles of Aβ and tau in neurons.


    . Lost after translation: missorting of Tau protein and consequences for Alzheimer disease. Trends Neurosci. 2014 Dec;37(12):721-32. Epub 2014 Sep 12 PubMed.

    View all comments by Hans Zempel
  2. The results reported are interesting and certainly the use of iPSCs has a lot of promise in uncovering mechanisms. However, I think the findings reported are difficult to interpret in regard to any relationship they do or don’t have with fAD. A technical issue is that the authors never used technology to convert the mutations in the iPSCs back to normal. This is a much better control than iPSCs from separate individuals without the mutations.

    In terms of the scientific findings, the changes in tau reported with one APP mutation and in the APP duplication are not the kind of pathological tau seen in AD or in tauopathies. For example, there are no insoluble tau or tau fibrils demonstrated. Also, there is no evidence of Aβ aggregation or deposition in this culture system.

    Finally, PS1 mutations end up causing clear-cut tauopathy in human brain (after Aβ deposition), yet in this system, no tau “abnormality” is seen in the PS1 iPSC-derived neurons. One would think that the mechanisms leading to tauopathy would more likely be the same in different APP or PS mutations.

    Thus, while the changes in APP fragments seen in iPS neurons derived from APP duplication and other APP mutations are interesting, whether the changes in tau levels observed in the iPSC-derived neurons is related to the changes in tau seen in AD, in which there is hyperphosphorylated, aggregated tau with PHFs, is not clear.

    View all comments by David Holtzman

Make a Comment

To make a comment you must login or register.


News Citations

  1. Honolulu: The Missing Link? Tau Mediates Aβ Toxicity at Synapse
  2. The Plot Thickens: The Complicated Relationship of Tau and Aβ
  3. Tracing a Path from Aβ to Tau Leads Scientists to Lost Synapses
  4. Induced Neurons From AD Patients Hint at Disease Mechanisms
  5. APP in Pieces: βCTF implicated in Endosome Dysfunction

Mutations Citations

  1. APP V717I (London)

Paper Citations

  1. . Abeta oligomers cause localized Ca(2+) elevation, missorting of endogenous Tau into dendrites, Tau phosphorylation, and destruction of microtubules and spines. J Neurosci. 2010 Sep 8;30(36):11938-50. PubMed.

Further Reading

Primary Papers

  1. . APP metabolism regulates tau proteostasis in human cerebral cortex neurons. Cell Rep. 2015 May 5;11(5):689-96. Epub 2015 Apr 23 PubMed.