They are no match made in heaven, but in neurons, the two proteins responsible for amyloid plaques and neurofibrillary tangles—amyloid-β (Aβ) and tau—might interact temporarily. So conclude Pat McGeer and colleagues in an advanced online publication in this week’s PNAS. They report that Aβ and tau bind to each other with considerable avidity in vitro, and that the liaison promotes phosphorylation of tau.

Though plaques and tangles are the major hallmarks of AD, establishing a solid link between them has been difficult. While mutations that lead to increased production of Aβ cause AD, tau mutations that promote neurofibrillary tangles cause a different form of disease, frontotemporal dementia. This suggests that the two proteins move in distinct circles. And yet there are also reports that these circles may overlap at times. Recent evidence that the two proteins conspire pathologically has come from Frank LaFerla’s lab at the University of California, Irvine. His work has shown that antibodies that mop up Aβ also reduce neurofibrillary tangles (see ARF related news story and Oddo et al., 2006). Other evidence also points to common pathways involving microtubule trafficking (see ARF related news story) and signal transduction through the Cdk5 (see ARF related news story), GSK3 (see ARF related news story), and Akt kinases (see ARF related news story). But if confirmed, the finding by McGeer and colleagues at the University of British Columbia, Vancouver, and at the Tokyo Institute of Psychiatry, Japan, could put some icing on the Aβ/tau cake.

First author Jian-Ping Guo and colleagues used simple Western blots to show that tau and Aβ formed complexes in vitro that survive detergent (SDS) and boiling. When Aβ40 or Aβ42 was added to recombinant, full-length tau, the complexes could be detected with either tau (tau 12, recognizing both phosphorylated and unphosphorylated protein) or Aβ (4G8, which detects either Aβ40 or Aβ42) antibodies. Furthermore, adding Aβ to the mix stimulated phosphorylation of tau by GSK3β, an indication that Aβ might directly promote the formation of neurofibrillary tangles in vivo (see also ARF SfN news story).

To hone in on the tau-Aβ binding site, the authors incubated Aβ with membrane-bound tau peptides and vice versa. This revealed that binding occurs between exons 7 and 9 of tau and the mid to C-terminal end of Aβ. These tau exons harbor threonine 212 and serines 214, 356, and 396, all subject to phosphorylation. Indeed, Guo showed that phosphorylation at T212 completely eliminates Aβ binding. Coupled with Aβ’s ability to enhance phosphorylation of tau, this suggests that the relationship between the two proteins may be short-lived because Aβ would probably change tau and drive the pair apart.

The paper does not answer the question of whether the two proteins interact in this fashion in vivo. Because Aβ is cleaved from its precursor protein (AβPP) on the luminal side of vesicles or extracellularly in the case of the cell membrane, it is not obvious how the two proteins meet. Several labs have detected intraneuronal Aβ deposits (see ARF related conference story and ARF conference story), and Guo and colleagues detected Aβ/tau complexes in brain tissue samples using an ELISA test that employed a tau antibody for capture and an Aβ antibody for detection; the signal from AD brain tissue was slightly higher than normal. They also report some data for co-localization of Aβ and tau in human neurofibrillary tangles in neurons of the entorhinal cortex. Guo et al. hypothesize that “an initial step in the pathogenesis [of AD] may be the intracellular binding of soluble Aβ to soluble non-phosphorylated tau, thus promoting tau phosphorylation and Aβ nucleation.” If true, then preventing this marriage could, in one fell swoop, prevent both neurofibrillary tangles and amyloid plaques.—Tom Fagan


  1. Guo et al. have found that tau and amyloid-β can interact in vitro. The authors used a mix of full-length recombinant tau and synthetic Aβ40 or Aβ42 to show that SDS-stable complexes are formed between tau and Aβ peptides after 5 hours' incubation time. Phosphorylation of tau by GSK3β weakened the interaction. A synthetic peptide spot array revealed three presumed binding sites for recombinant tau within the Aβ sequence, Aβ11-16, Aβ27-32 and Aβ37-42. The N-terminal site has a histidine residue at position 13 that is not conserved in rodents. The other two candidate sites are localized to the C-terminal region of Aβ, which adopts a β-sheet conformation before or during the aggregation of Aβ into amyloid.

    Surprisingly, immunostaining of tau and Aβ revealed a colocalization of tau and Aβ with varying degrees. I would expect Aβ to be associated with vesicular compartments and tau being associated with the cytoskeleton. So far, we do not know if the staining exclusively reflects Aβ staining, since the antibody 4G8’s epitope is also found in β-stubs. Also, a release of soluble hetero-oligomers of Aβ and tau by cells should lead to the formation of Aβ-tau aggregates. But so far, tau staining of Alzheimer plaques has rarely been reported, although there is a spatial relationship between plaques in the terminal fields of tangle-bearing neurons.

    In addition, the authors used the BIACORE technique to further prove that Aβ and tau interact in vitro. Recombinant tau showed a 1,000-fold stronger binding to immobilized Aβ (possibly a mixture of monomers, oligomers, and ADDLs) than to itself. Since the authors did not show binding at saturating conditions, it is difficult to judge how the self-interaction of tau influences tau binding to Aβ. One could even assume that tau-tau interactions might be strengthened in the presence of Aβ.

    Thus, the experimental approaches described by Guo et al. reveal an in-vitro scenario of how Aβ and tau could interact. It is not clear if, how, and where both molecules interact in vivo. So far, the hypothesis of the "amyloid cascade," assuming that tangle formation is a direct consequence of amyloid plaque formation, is still valid for both “Baptists” and “Tauists” until the spatial and temporal relationships have been solved.

  2. Patrick McGeer and colleagues showed that the Aβ species form SDS-resistant complexes with tau protein in vitro, and the binding promotes tau phosphorylation. Hence, they suggest, the intraneuronal interaction between Aβ and tau is crucial for the formation of neurofibrillary tangles.

    This conclusion would be more strongly supported by the demonstration of Aβ/tau complexes with immunoblotting in soluble fractions of AD brain. In addition, the immunocytochemistry is not convincing. We showed with immunoEM that Aβ and tau are associated only in the "ghost," extracellular tangles (Tabaton et al., 1991). In this case, the association reflects a "nucleation effect" of the core of PHF-tau on the soluble Aβ. Therefore, the association does not have a functional role in PHF formation. In the present paper, the authors do not convince that Aβ and tau coreactivity occurs intracellularly.


    . Ultrastructural localization of beta-amyloid, tau, and ubiquitin epitopes in extracellular neurofibrillary tangles. Proc Natl Acad Sci U S A. 1991 Mar 15;88(6):2098-102. PubMed.

  3. The identification in this paper of a complex formation between Aβ and tau by Patrick McGeer’s group (Aβ and tau form soluble complexes that may promote self-aggregation of both into the insoluble forms observed in Alzheimer disease) by Western blotting, surface plasmon resonance, and ELISA, is a fascinating piece of work. I like in particular the data obtained with the peptide array of 214 peptide spots covering the entire tau sequence and with 11 specific peptides, with or without phosphorylated serine and threonine. Not surprisingly, phosphorylation of the “physiological” AT8 epitope S202/T205 of tau does not interfere with Aβ binding, whereas phosphorylation of the “pathological” epitopes T212/S214 (AT100) and S422, among others, does.

    In light of the co-occurrence of Aβ plaques and Lewy bodies in a range of dementing disorders, it would be very interesting to determine in a follow-up study the interaction of α-synuclein and Aβ. If affinities there are much lower than for tau/Aβ, it may also partly explain why plaques and NFTs co-occur more often than plaques and Lewy bodies.

    Also, it would be interesting in a follow-up to address the interaction of tau and the reversed peptide of Aβ42. That is because, under some experimental conditions, where the reversed peptide is used as a control for Aβ42, biological effects are seen that have not been found when using PBS as control.

  4. Guo et al. address a longstanding enigma about the signature lesions of AD, that is, the senile plaques (SPs) formed by extracellular deposits of Aβ amyloid fibrils and the neurofibrillary tangles (NFTs) formed by intraneuronal accumulations of tau amyloid fibrils known as PHFs or PHFtau.

    Despite an abundance of hypotheses to account for how Aβ fibrils may cause formation of PHFtau and NFTs, or how PHFtau might induce formation of Aβ-rich SPs, there is little experimental data to support the predictions of these hypotheses. Circumstantial evidence from studies of sporadic and familial AD has been interpreted to imply that Aβ amyloid causes tau amyloidosis and NFTs in AD, and yet studies of Guam tauopathies and Niemann Pick Type C disease (NPC) lend support to the notion that the accumulation of SPs results from the earlier accumulation of NFTs in these diseases. However, there is little experimental evidence to support inferences based on these and other types of circumstantial evidence.

    Indeed, most transgenic mice engineered to overexpress human Aβ develop SPs but not NFTs, and transgenic mice engineered to express mutant or wild-type human tau develop NFTs but not SPs. While double and triple transgenic mice have been generated to model the formation of both AD-like SPs and NFTs, other approaches are needed to address this issue. For example, earlier experimental animal studies of injections of PHFtau into rodent brains implied that the release of PHFtau from dying neurons or their processes could interact with Aβ, thereby impeding the clearance of Aβ and leading to Aβ deposits (Shin et al., 1993; Shin et al., 1994.

    However, in the Guo et al. paper, the authors used novel in-vitro methods to study interactions of Aβ and tau, and these authors report that tau and Aβ form complexes in vitro. The authors suggest that these complexes may be mechanistically linked to the initial steps in pathways leading to the formation of SPs and NTFs in AD. These in-vitro experiments are complemented by studies of AD brain tissues that demonstrate the intraneuronal colocalization of Aβ and PHFtau. Based on these findings, the authors suggest that the perikarya and processes of neurons are the plausible locations where tau and Aβ could interact and bind to one another with high affinity to form the Aβ-tau complexes the authors observed in their in-vitro experiments.

    The findings of Guo et al. are provocative. They open up new avenues for experimental research which, hopefully, will stimulate further efforts to elucidate the enigmatic mechanisms underlying the coaccumulation of SPs and NFTs in AD as well as in other diseases such as Guam tauopathies and NPC.


    . Alzheimer disease A68 proteins injected into rat brain induce codeposits of beta-amyloid, ubiquitin, and alpha 1-antichymotrypsin. Proc Natl Acad Sci U S A. 1993 Jul 15;90(14):6825-8. PubMed.

    . Aluminum modifies the properties of Alzheimer's disease PHF tau proteins in vivo and in vitro. J Neurosci. 1994 Nov;14(11 Pt 2):7221-33. PubMed.

  5. The concept of this study could be important, as it may offer an explanation for tau overphosphorylation in AD. On the other hand, it is not clear how this theory explains the independent distribution of the amyloid plaques and neurofibrillary tangles (NFTs) in the AD (or even normal) brain, or the fact that NFTs can form in the absence of Aβ deposition. Also, since soluble Aβ is found in all people and all brains (and thus far solid evidence for a specific increase in any form of a soluble Aβ in sporadic AD is lacking), what keeps soluble Aβ from catalyzing tau overphosphorylation in normal brains?

  6. Reply by Pat McGeer to commentary above
    John Trojanowski, Juergen Goetz, Gerd Multhaup, Massimo Tabaton, and Nikolaos Robakis have each made thoughtful and pertinent comments about our paper.

    Our demonstration of a strong interaction between Aβ and tau is such a simple and easily replicable experiment that it may seem strange that it was not identified long ago. We suggest the reason is the widespread acceptance of the standard APP-Aβ model. It hypothesizes that Aβ is produced at the cell surface and then secreted into the extracellular fluid. Even the names given to the enzymes responsible for Aβ production reinforce this concept: α-secretase, β-secretase, and γ-secretase.

    But intraneuronal production of Aβ also occurs. There is a rich but overlooked literature on this subject. We hope our paper will draw attention to that literature, as well as stimulate productive new experiments involving intraneuronal Aβ. We did not cover papers involving transgenic mice, but a report of particular interest is that of Billings et al. (2005), who found that memory deficits in 3xTg-AD mice correlate with intraneuronal accumulation of AD before there is any evidence of plaque and tangle pathology.

    Dr. Trojanowski makes the important observation that our findings need to be followed up in other enigmatic diseases such as the Guam disease and Niemann-Pick disease, type C. Dr. Tabaton finds our immunohistochemistry unconvincing, but at the time of his writing did not have the benefit of seeing our additional evidence published as supporting information online because it had not appeared yet. Immunochemistry, including a movie showing colocalization of tau and Aβ inside the neuron, is now available for those wanting more detailed information. Dr. Tabaton discounts intracellular Aβ-tau complexes because he did not identify them by electron microscopy (EM). But he would not find intracellular Aβ-tau complexes by EM, since such complexes occur in the soluble state prior to aggregation and therefore will be washed away during EM preparation.

    Dr. Goetz makes the exciting suggestion that there might be an interaction between Aβ and synuclein. Already it is known that there is an interaction between tau and synuclein, and this possibility deserves exploration. Dr. Multhaup notes that the monoclonal 4G8 antibody may be detecting Aβ stubs. We have immuno-stained with a panel of antibodies recognizing N- and C-terminal epitopes, which give identical findings. We have observed intraneuronal Aβ aggregates in various morphological forms in AD brain, and this will be the subject of a subsequent publication.

    Dr. Robakis considers that our explanation does not account for the separate accumulation of plaques and tangles, nor for the accumulations of tau but not Aβ in the various tauopathies. The coexistence of intraneuronal Aβ and tau aggregates shows that they are not necessarily separated. Evidence in the literature suggests that multiple mutations in tau can cause insoluble tau aggregations without causing Aβ aggregations, but that excess Aβ, at least in humans, can cause tau aggregation. Therefore, tau-Aβ complexes must be only one of multiple mechanisms that can lead to tau aggregation. It should be noted that soluble forms of Aβ and tau, as well as their soluble complexes, will not be observed by immunohistochemistry. Our data indicate that it is the intraneuronal, soluble Aβ that is most closely associated with AD initiation.


    . Intraneuronal Abeta causes the onset of early Alzheimer's disease-related cognitive deficits in transgenic mice. Neuron. 2005 Mar 3;45(5):675-88. PubMed.

  7. Tau-Amyloid Interactions: A Timely Revival?
    The long-awaited direct connection between amyloid-β and tau may have finally arrived with the publication of this paper. The question is, given the experimental simplicity, why did it take so long? The simple answer is that it did not. Certainly, it was well established by three groups (Smith et al., 1995; Giaccone et al., 1996; Islam and Levy, 1997) that amyloid-β precursor protein (AβPP) is able to interact with tau and that this promoted fibril formation. Earlier still, it was established that amyloid-β (and, in fact, the entire AβPP protein) was associated with intraneuronal (Hyman et al., 1989; Allsop et al., 1990; Perry et al., 1992; Perry et al., 1993) and extracellular neurofibrillary tangles (Smith et al., 1990; Tabaton et al., 1991), leading one of us to the “outrageous” (at the time) conclusion that neurofibrillary tangles were the nucleation point for senile plaques (Perry, 1993).

    The difference? Earlier reports were perhaps not published in such a high-profile journal, perhaps not so elegantly direct and, perhaps most importantly, ahead of their time. Then, much less so than now, the Tauists and Baptists reigned supreme (Perry et al., 2004) and a collaborative partnership was unthinkable. These days, the swing from fact to belief (fibrils to oligomers) may have dissatisfied the proletariat sufficiently to necessitate interaction.


    . Neurofibrillary tangles in some cases of dementia pugilistica share antigens with amyloid beta-protein of Alzheimer's disease. Am J Pathol. 1990 Feb;136(2):255-60. PubMed.

    . beta PP and Tau interaction. A possible link between amyloid and neurofibrillary tangles in Alzheimer's disease. Am J Pathol. 1996 Jan;148(1):79-87. PubMed.

    . A4 amyloid protein immunoreactivity is present in Alzheimer's disease neurofibrillary tangles. Neurosci Lett. 1989 Jul 3;101(3):352-5. PubMed.

    . Carboxyl-terminal fragments of beta-amyloid precursor protein bind to microtubules and the associated protein tau. Am J Pathol. 1997 Jul;151(1):265-71. PubMed.

    . Neuritic plaques in Alzheimer disease originate from neurofibrillary tangles. Med Hypotheses. 1993 Apr;40(4):257-8. PubMed.

    . Beta protein immunoreactivity is found in the majority of neurofibrillary tangles of Alzheimer's disease. Am J Pathol. 1992 Feb;140(2):283-90. PubMed.

    . When hypotheses dominate. The Scientist. 2004 Dec 8;18(23):6.

    . Immunocytochemical evidence that the beta-protein precursor is an integral component of neurofibrillary tangles of Alzheimer's disease. Am J Pathol. 1993 Dec;143(6):1586-93. PubMed.

    . Demonstration of an A4 epitope in some extracellular neurofibrillary tangles in Alzheimer's disease. Neuropathol Appl Neurobiol. 1990;16:269.

    . Tau protein directly interacts with the amyloid beta-protein precursor: implications for Alzheimer's disease. Nat Med. 1995 Apr;1(4):365-9. PubMed.

    . Ultrastructural localization of beta-amyloid, tau, and ubiquitin epitopes in extracellular neurofibrillary tangles. Proc Natl Acad Sci U S A. 1991 Mar 15;88(6):2098-102. PubMed.

    View all comments by Xiongwei Zhu

Make a Comment

To make a comment you must login or register.


News Citations

  1. Tackling Alzheimer’s from the Outside In
  2. Tau Accused of Blocking Transport, Causing APP to Linger and Nerve Processes to Wither
  3. Aiding and Abetting, Hyperactive CDK5 Gives Mouse Tangles
  4. Lithium Hinders Aβ Generation, Buffing Up GSK as Drug Target
  5. SfN: Where, How Does Intraneuronal Aβ Pack Its Punch? Part 4
  6. SfN: Amyloid Oligomers—Not So Elusive, After All? Part 2
  7. Philadelphia: The Enemy Within—Neurodegeneration From Intraneuronal Aβ
  8. SfN: Where, How Does Intraneuronal Aβ Pack Its Punch? Part 1

Paper Citations

  1. . Temporal profile of amyloid-beta (Abeta) oligomerization in an in vivo model of Alzheimer disease. A link between Abeta and tau pathology. J Biol Chem. 2006 Jan 20;281(3):1599-604. PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . Abeta and tau form soluble complexes that may promote self aggregation of both into the insoluble forms observed in Alzheimer's disease. Proc Natl Acad Sci U S A. 2006 Feb 7;103(6):1953-8. PubMed.