What does the axonal protein tau have to do with synapses? Quite a bit, some scientists say. As researchers continue to dig for the roots of Alzheimer’s disease, they are turning up more evidence that tau can have powerful effects on synaptic function, learning, and memory. In the December 22 Neuron, researchers led by Dezhi Liao and Karen Hsiao Ashe at the University of Minnesota in Minneapolis show that mutant tau accumulates in dendritic spines of mouse hippocampus early in disease. This misdirected tau leads to the loss of glutamate receptors and dampens synaptic signaling. Furthermore, it is the phosphorylation of tau at specific sites that causes the protein to stray, the authors find. Meanwhile, researchers led by Erik Roberson at the University of Alabama at Birmingham and Lennart Mucke at the Gladstone Institute of Neurological Disease focused on the interactions between wild-type tau, Aβ, and the kinase Fyn. In the January 12 Journal of Neuroscience, they show that reducing tau levels in AD mice restores normal synaptic activity and network firing, adding to the evidence that tau mediates many of the toxic effects of Aβ. Both papers heighten the growing luster of tau as a therapeutic target.

Tau’s role in synapses and in AD has increasingly drawn the spotlight. Previous work by Mucke and Roberson showed that tau reduction could protect AD mice from cognitive impairment (see ARF related news story on Roberson et al., 2007), suggesting that tau acts downstream of Aβ. Intriguingly, Mucke and colleagues found that Fyn also helped Aβ poison synapses (see Chin et al., 2004; Chin et al., 2005). Last year, work by Lars Ittner and Jürgen Götz at the University of Sydney, Australia, tied the two proteins together. In mice, they showed that endogenous tau targets Fyn to the N-methyl-D-aspartic acid (NMDA) receptor, allowing tau to mediate Aβ-induced excitotoxicity at the synapse (see ARF related news story on Ittner et al., 2010). Their work implied a physiologic role for tau in dendrites. Other studies have looked at the pathologic effects of too much dendritic tau. A recent study led by Eckhard and Eva-Maria Mandelkow at the Max-Planck-Unit for Structural Molecular Biology in Hamburg, Germany, demonstrated that in cultured hippocampal neurons, Aβ treatment caused tau to wander into dendrites, leading to the loss of synaptic spines, microtubules, and mitochondria (see ARF related news story on Zempel et al., 2010). This followed earlier work by the Mandelkows in which they overexpressed tau in cultured neurons and saw the axonal protein build up in dendrites and synapses disappear. In that case, the Mandelkows traced the cause to excess tau clogging up microtubule highways (see ARF related news story on Thies et al., 2007).

Liao and Ashe were interested in looking at some of the earliest changes caused by dendritic tau. First authors Brian Hoover and Miranda Reed used the rTgP301L mouse, which produces large quantities of human tau containing the P301L mutation. The authors demonstrated that these mice display spatial memory deficits and faulty long-term potentiation (LTP) in the hippocampus at ages as young as 4.5 months, prior to any loss of synapses. In order to make sure that these effects were due to the tau mutation and not simply the high levels of tau, Hoover and Reed also created a new mouse strain (rTgWT) that expresses wild-type human tau at comparable levels to rTgP301L. These animals showed mild synaptic and memory defects, but unlike the rTgP301L cousins, no neurodegeneration or worsening of memory with age, indicating that high levels of tau by themselves do not cause disease.

Hoover and colleagues wanted to test the idea that mutant tau mislocalizes into synapses. Using brain isolates from 4.5-month-old mice, they found high levels of tau in the post-synaptic preparations of P301L mice compared to both wild-type and rTgWT mice. To directly visualize tau movements, the authors switched to an in vitro setting, transfecting primary rat hippocampal neurons with DNA for green fluorescent protein/human tau chimeras. Mutant P301L tau populated the majority of dendritic spines, while wild-type tau only rarely popped up in spines, confirming that only mutant tau invades synapses in force.

To examine if this excess tau has synaptic effects, Hoover and colleagues measured post-synaptic currents in the transfected rat hippocampal cultures as well as in cultured hippocampal neurons from the transgenic mice. In both types of cultures, the presence of P301L tau in neurons correlated with dampened miniature excitatory post-synaptic currents (mEPSCs). Wild-type human tau had no such effect. To investigate the mechanism behind this decline, the authors labeled cultured neurons with fluorescently tagged antibodies specific for several glutamate receptor subunits. The labels revealed that neurons containing mutant tau had far fewer NMDA and AMPA receptors at post-synapses than did wild-type or tau-overexpressing neurons. The authors note that this drop in receptors could explain the lower mEPSCs (see Malinow and Malenka, 2002). A loss of glutamate receptors is also known to contribute to loss of spines (see McKinney et al., 1999; Richards et al., 2005; McKinney et al., 2010), suggesting that these early effects may precede synapse loss.

Finally, Hoover and colleagues demonstrated that tau’s ability to invade synapses depends on phosphorylation. The authors focused on 14 phosphorylation sites in a proline-rich region of tau, which are normally modified by proline-directed kinases. In Drosophila, these sites have been shown to modulate the neurotoxicity of tau and to govern the binding of tau to actin, a cytoskeletal protein that also occurs in spines (see Fulga et al., 2007; Steinhilb et al., 2007; Steinhilb et al., 2007). To prevent phosphorylation of these 14 sites, Hoover and colleagues mutated the serine and threonine residues to alanine or proline; to mimic phosphorylation, they changed the residues to glutamate. When cultured rat neurons were transfected with these constructs, glutamate-modified tau invaded synapses more heavily than did wild-type tau, while the alanine-proline modified tau shunned dendritic spines.

One limitation of these experiments for AD is that they say nothing about the interaction between tau and Aβ. In future work, Ashe said, they plan to use a combined APP/tau model such as APPswe/Tauvlw to examine the relationship between these proteins. Aβ has also been shown to have post-synaptic effects, blocking synaptic plasticity and causing synapse loss (e.g., see ARF related news story on Koffie et al., 2009 and ARF related news story on Lacor et al., 2007).

“I think it’s tantalizing that the site at which amyloid-β and tau cause neuronal dysfunction appears to be the dendritic spine, at least in the earliest stages,” Ashe said. Ashe and Liao will also investigate the mechanism behind the loss of AMPA and NMDA receptors. The presence of tau might be causing increased receptor internalization, or it might be blocking the trafficking of receptors to spines, Liao speculated.

Liao believes the tau phosphorylation and mislocalization mechanism they have uncovered could show great promise for the prevention or early treatment of AD. “For a long time, people have tried to treat the late stage of AD. We believe the best strategy is to treat it early,” Liao said. The AD field as a whole is moving in this direction, based in part on poor clinical trial outcomes (e.g., see ARF related news story and ARF related news story). Liao also points out that synaptic damage is now believed to be the key cellular mechanism underlying memory loss (e.g., see ARF related news story), and that by translocating to the synapse, tau might be involved. “If we can block this unwanted guest, we can potentially find a cure for neurodegenerative disease,” he said.

Michel Goedert of the MRC Laboratory of Molecular Biology in Cambridge, U.K., agrees this approach might have promise. “[The paper] confirms that tau hyperphosphorylation could be a good therapeutic target.” However, when relating these findings to the human condition, Goedert said, it will be important to know if the tau that enters spines is soluble or already filamentous. “From the human cases, there is evidence to suggest that tau aggregates early on, and it aggregates probably in the axon, and it could be that it aggregates before mislocalization.” (See, e.g., Braak and Del Tredici, 2010.) Another unanswered question, Goedert points out, is whether the mistargeting of mutant tau depends on the very high expression levels in these mice.

Eva-Maria Mandelkow notes that the phosphorylation sites examined by Liao and colleagues are not the sites that control attachment of tau to microtubules. It would be informative to also look at the effect of phosphorylation at these sites, she said, since detachment from filaments is a necessary precursor to tau aggregation and mislocalization.

Several commentators wonder how these experiments might relate to the behavior of endogenous, wild-type tau in AD. In that regard, Fred Van Leuven of Katholieke Universiteit Leuven, Belgium, finds it interesting that Liao and colleagues show that transgenic mice overexpressing wild-type human tau do show some cognitive deficits. “This is a major finding, important for the many primary tauopathies caused by wild-type tau, and not least for AD, the most frequent secondary tauopathy,” Van Leuven wrote in an e-mail to ARF. Van Leuven also points out that “we still do not know whether tau is a physiological constituent in spines and post-synaptic compartments or whether it wanders off there only in pathological conditions.” The answer to that question would also affect treatment approaches.

In the second paper, Roberson and Mucke examined the behavior of wild-type tau in relation to Fyn and Aβ. They crossed mice having a mild AD phenotype (hAPPJ9) with animals that overexpressed mouse Fyn to create a strain that demonstrates numerous defects characteristic of AD, such as memory problems, frequent seizures, hippocampal remodeling, and early mortality. When first author Roberson crossed these mice with tau knockout mice, the synaptic and network behavior returned to normal, adding to the evidence that the negative effects of Fyn and Aβ depend on tau. Likewise, the authors showed that tau reduction also normalized synaptic transmission, NMDA receptor function, long-term potentiation, and network excitability in hAPPJ20 mice. Significantly, Roberson and colleagues showed that tau reduction did not lengthen survival in a mouse model of amyotrophic lateral sclerosis, indicating that the protective effect of ablating tau is relatively specific for AD pathology.

“This paper brings a few ideas together,” Roberson said. It shows that tau reduction not only protects against seizures, but also normalizes many other aspects of electrophysiology in AD mice. It ties tau to network dysfunction, Roberson said, suggesting that tau plays a role in regulating neuronal activity and synchrony. In AD mice, “the inhibitory interneurons do not seem to be firing enough. This produces an imbalance between excitation and inhibition, which may drive a lot of the abnormal synchronization and abnormal firing rates of the neurons.” Dialing down tau prevents this imbalance.

In future work, Roberson said, he would like to use an inducible mouse model to control the timing and location of tau reduction. He hopes this approach will allow him to dissect the mechanisms behind tau’s contribution to pathology, and figure out what therapeutic approaches might work best. “Our paper provides further evidence that reducing tau or otherwise tapping into the effects of tau reduction, perhaps by targeting the tau-Fyn interaction, is potentially a powerful therapeutic approach to the disease,” Roberson suggested.—Madolyn Bowman Rogers


  1. This is nice work by Hoover and colleagues providing yet another piece of important evidence for the role(s) of tau in the post-synapse. While we showed before that endogenous mouse tau is associated with the post-synaptic density (PSD) (Ittner et al., 2010), this study now reveals that upon overexpression of human tau, even more tau is associated with the PSD (in particular, when the P301L mutations is present), which then impairs the normal function of synapses. I am particularly intrigued by the fact that (hyper)phosphorylation of tau specifically promotes its localization to dendritic spines.


    . Dendritic function of tau mediates amyloid-beta toxicity in Alzheimer's disease mouse models. Cell. 2010 Aug 6;142(3):387-97. Epub 2010 Jul 22 PubMed.

    View all comments by Lars M. Ittner
  2. I am intrigued by the findings regarding changes in transmission in the dentate gyrus of APP Tg animals. The authors find an increase in miniature inhibitory post-synaptic (IPSC) current frequency, but a decrease in "spontaneous" IPSC frequency (a question regarding "spontaneous" is that this is an isolated slice lacking cell bodies of many of the normal inputs—e.g., entorhinal cortex). I note that they confirm in the dentate gyrus what we have seen in the CA1 region: a reduced mini-excitatory post-synaptic current frequency. All these changes somehow contribute to changes in network activity.

    The question regarding which changes are primarily due to Aβ and which are compensatory is important. My feeling is that this question must be answered with more acute treatments: Making inferences regarding this question based on transgenic animals is almost impossible; it would require making causal inferences over time and development (from before birth to the time when the transgenic animal is analyzed) that would explain the phenotype.

    Given our (and others') findings that acute (minutes to days) Aβ treatment produces a reduction of excitatory transmission, it would seem that this should be considered a primary cause of the eventual phenotype observed in these animals. More careful assessment of Aβ effects on inhibitory neurons (transmission as well as spiking) is warranted.

    It will be also important to determine if the primary changes produced by enhanced Aβ lead to subsequent changes that produce a vicious (positive reinforcement) cycle that exacerbates the phenotype. For instance, increased Aβ produces a reduction in excitatory transmission, leading to hypoactivity of interneurons, which leads to increased network activity, which increases Aβ production/release. Interfering with this vicious cycle may prove therapeutically beneficial.

    View all comments by Roberto Malinow
  3. The paper from Lennart Mucke’s group demonstrates that Aβ-induced synaptic dysfunction depends upon cellular alterations in tau proteins. In contrast, our paper (Hoover et al., 2010) mainly focuses on how cellular alterations in tau proteins themselves impair synaptic functions. There is a possibility, although not proven, that the cellular mechanism unraveled by our study underlies the signaling steps downstream from the cellular alterations found by Mucke’s group. These two studies fit well to each other and both support this possible hypothesis.


    . Tau mislocalization to dendritic spines mediates synaptic dysfunction independently of neurodegeneration. Neuron. 2010 Dec 22;68(6):1067-81. PubMed.

  4. In this manuscript, Hoover and colleagues report that overexpressed tau distributes to dendritic spines, where it impairs synaptic responses. The results suggest that not only does tau lead to pre-synaptic dysfunction by impairing axonal transport, but it also causes post-synaptic dysfunction that contributes to behavioral deficits in mice overexpressing mutant tau. The importance of post-synaptic function through tau was also suggested by a previous report (Ittner et al., 2010). Therefore, tau may play a role in the post-synapse, directly or indirectly, leading to neural dysfunction in neurodegenerative disease. However, because the localization of tau in dendritic spines is shown only in a tau overexpression paradigm, we need to know whether or not endogenous tau also distributes to the dendritic spine in disease cases before we consider tau a therapeutic target.

    In AD or the other tauopathies, tau is not overexpressed in neurons, and it relocates from the axon to the somatodendrite when tau is hyperphosphorylated. In this manuscript, Hoover et al. show that when overexpressed, pseudophosphorylated tau localizes in the dendrite/dendritic spine. There are two possible mechanisms. One is that phosphorylated tau diffuses to somatodendrite, because it no longer associates with microtubules. The other is that phosphorylated tau is transported to the dendrite by some unknown active mechanisms. If tau is basically found in axons under normal conditions, then is phosphorylated tau in somatodendrites derived from axonal tau? If so, phosphorylated tau travels a long distance, and there must be some physiological reason for hyperphosphorylated tau to travel this long distance to the dendrite/dendritic spine. Alternatively, tau may localize in locations other than the axon, where it plays a physiological role under normal conditions, and accumulates in dendrites by hyperphosphorylation. Therefore, we may need to pay more attention to localization of endogenous tau (not overexpressed tau), and to other possible indispensable roles of tau besides microtubule stabilization, in normal conditions.

    Finally, I have a question about this result. Hoover and colleagues indicated that P301L tau accumulated in dendritic spines in transgenic mice, primary neuronal culture of the transgenic mouse, and when overexpressed in rat primary neurons. The accumulation of P301L tau reduces AMPA receptor levels, leading to memory deficit before synapse loss and neuron loss. If the accumulation of P301L tau in the synaptic region is a cause of synaptic dysfunction, why did the 1.3-month-old mouse not show learning deficit, while the 4.5-month-old mouse did? The tau expression level is similar in both.

    Also, in this week’s Journal of Neuroscience, Mucke’s group reported that reduction of tau rescued synaptic dysfunction in an APP Tg mouse, which again makes us more pay attention to the role of tau in synaptic function. Furthermore, the results make us consider not only NMDA signals, but also “homeostatic synaptic plasticity” to better understand development of dementia in AD.

  5. This paper by Erik Roberson, Lennart Mucke, and colleagues sheds new light on the susceptibility of APP transgenic mice to seizures. Accumulating evidence suggests that vulnerability to epilepsy is one aspect of Alzheimer’s disease where both amyloid and tau pathologies may be required. Mucke’s group earlier reported that mice expressing hAPP with the Swedish and Indiana mutations (either the high-expressing J20 line or the low-expressing J9 line sensitized to pathogenic Aβ effects by neuronal overexpression of Fyn) show reduced seizure threshold after systemic pentylenetetrazole (PTZ) administration, and aberrant calbindin and neuropeptide Y expression in the hippocampus similar to rodents with induced seizures (Palop et al., 2007). We reported similar histochemical changes in APPswe/PS1dE9 mice with generalized spontaneous seizures (Minkeviciene et al., 2009). The Mucke team also reported that reduction of endogenous tau by crossing hAPP mice with tau-/- mice prevents premature mortality and increases the threshold to kainic acid-induced seizures in J20 mice (Roberson et al., 2007). The present paper first confirms the protective role of tau reduction on early mortality in the hAPP/J9+ Fyn line and in another hAPP mouse line (TASD41, which expresses hAPP with the Swedish and London V717I mutations), but not in SOD1 transgenic mouse models of amyotrophic lateral sclerosis. Tau reduction also prevented increased susceptibility to PTZ-induced seizures and ameliorated the severity of spontaneous seizures in hAPPJ9/Fyn mice as well as in the hAPPJ20 mice.

    What might be the specific Aβ-tau interaction that renders amyloid-producing mice susceptible to epileptic seizures? Despite a large number of data, this key question is only briefly discussed in the paper. Of note in the present paper, the anti-convulsive effect of tau knockout was not restricted to hAPP mice, but was observed in Fyn transgenics or even wild-type mice. Similarly, tau reduction markedly suppressed bicucullin-induced epileptiform bursting in hippocampal slices from both non-transgenic and hAPPJ20 mice. However, some network effects of tau knockout were specific to hAPP mice. Dentate granule cells of hAPPJ20 mice showed increased frequency of miniature inhibitory post-synaptic potentials, while mini-excitatory post-synaptic potentials were increased compared to wild-type controls. Tau reduction prevented both types of changes in hAPP mice but had no effect in wild-type mice. Similarly, NMDA receptor-mediated currents in dentate granule cells were reduced, while AMPA-currents were unchanged in hAPPJ20 mice. The NMDA effect, too, was blocked by tau reduction in hAPP mice, while no effect of tau reduction was observed in wild-type mice.

    The change in NMDA currents links the current paper in an interesting way to a recent observation by Ittner and colleagues (Ittner et al., 2010). They crossed yet another hAPP mouse line (APP23) with tau-/- mice or mice expressing a truncated form of tau that prevented its interaction (via its amino-terminal domain) with the fyn kinase. Fyn phosphorylates the NMDA receptor NR2B subunit to facilitate its interaction with post-synaptic density protein 95, further linking the NMDA receptor to synaptic excitotoxic downstream signaling. Interestingly, both knockout or truncation of tau ameliorated premature mortality of APP23 mice, and reduced their susceptibility to PTZ induced seizures.

    Further evidence for the role of the tau-fyn interaction in premature mortality and seizure susceptibility of APP transgenic mice comes from fyn transgenic mice. These animals present with seizures and premature mortality (Kojima et al., 1998), and these are exacerbated in mice co-expressing transgenic APP (Chin et al., 2004; Roberson 2011). It remains to be seen whether the tau-fyn interaction can be linked to the initial seizure susceptibility of APP transgenic mice or only to the mortality associated with the seizures. Judged from the electrophysiological findings in the present paper, tau may also be involved in the network plasticity that changes the excitation-inhibition balance.


    . Fyn kinase modulates synaptotoxicity, but not aberrant sprouting, in human amyloid precursor protein transgenic mice. J Neurosci. 2004 May 12;24(19):4692-7. PubMed.

    . Dendritic function of tau mediates amyloid-beta toxicity in Alzheimer's disease mouse models. Cell. 2010 Aug 6;142(3):387-97. Epub 2010 Jul 22 PubMed.

    . Higher seizure susceptibility and enhanced tyrosine phosphorylation of N-methyl-D-aspartate receptor subunit 2B in fyn transgenic mice. Learn Mem. 1998 Nov-Dec;5(6):429-45. PubMed.

    . Amyloid beta-induced neuronal hyperexcitability triggers progressive epilepsy. J Neurosci. 2009 Mar 18;29(11):3453-62. PubMed.

    . Aberrant excitatory neuronal activity and compensatory remodeling of inhibitory hippocampal circuits in mouse models of Alzheimer's disease. Neuron. 2007 Sep 6;55(5):697-711. PubMed.

    . Reducing endogenous tau ameliorates amyloid beta-induced deficits in an Alzheimer's disease mouse model. Science. 2007 May 4;316(5825):750-4. PubMed.

    View all comments by Heikki Tanila

Make a Comment

To make a comment you must login or register.


News Citations

  1. APP Mice: Losing Tau Solves Their Memory Problems
  2. Honolulu: The Missing Link? Tau Mediates Aβ Toxicity at Synapse
  3. The Plot Thickens: The Complicated Relationship of Tau and Aβ
  4. Tau Roundup: Inducible Mice Accentuate Aggregation and More
  5. Spine Shrinkers: Aβ Oligomers Caught in the Act
  6. Aβ—Three Places, Three Ways of Wreaking Havoc
  7. Clinical Trials: Task Force Supports Earlier Intervention
  8. Early Events in AD Mice as Targets for Therapy
  9. New Orleans: Symposium Probes Why Synapses Are Suffering

Paper Citations

  1. . Reducing endogenous tau ameliorates amyloid beta-induced deficits in an Alzheimer's disease mouse model. Science. 2007 May 4;316(5825):750-4. PubMed.
  2. . Fyn kinase modulates synaptotoxicity, but not aberrant sprouting, in human amyloid precursor protein transgenic mice. J Neurosci. 2004 May 12;24(19):4692-7. PubMed.
  3. . Fyn kinase induces synaptic and cognitive impairments in a transgenic mouse model of Alzheimer's disease. J Neurosci. 2005 Oct 19;25(42):9694-703. PubMed.
  4. . Dendritic function of tau mediates amyloid-beta toxicity in Alzheimer's disease mouse models. Cell. 2010 Aug 6;142(3):387-97. Epub 2010 Jul 22 PubMed.
  5. . 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.
  6. . Missorting of tau in neurons causes degeneration of synapses that can be rescued by the kinase MARK2/Par-1. J Neurosci. 2007 Mar 14;27(11):2896-907. PubMed.
  7. . AMPA receptor trafficking and synaptic plasticity. Annu Rev Neurosci. 2002;25:103-26. PubMed.
  8. . Miniature synaptic events maintain dendritic spines via AMPA receptor activation. Nat Neurosci. 1999 Jan;2(1):44-9. PubMed.
  9. . Glutamate induces the rapid formation of spine head protrusions in hippocampal slice cultures. Proc Natl Acad Sci U S A. 2005 Apr 26;102(17):6166-71. PubMed.
  10. . Excitatory amino acid involvement in dendritic spine formation, maintenance and remodelling. J Physiol. 2010 Jan 1;588(Pt 1):107-16. PubMed.
  11. . Abnormal bundling and accumulation of F-actin mediates tau-induced neuronal degeneration in vivo. Nat Cell Biol. 2007 Feb;9(2):139-48. PubMed.
  12. . Tau phosphorylation sites work in concert to promote neurotoxicity in vivo. Mol Biol Cell. 2007 Dec;18(12):5060-8. PubMed.
  13. . S/P and T/P phosphorylation is critical for tau neurotoxicity in Drosophila. J Neurosci Res. 2007 May 1;85(6):1271-8. PubMed.
  14. . Oligomeric amyloid beta associates with postsynaptic densities and correlates with excitatory synapse loss near senile plaques. Proc Natl Acad Sci U S A. 2009 Mar 10;106(10):4012-7. PubMed.
  15. . Abeta oligomer-induced aberrations in synapse composition, shape, and density provide a molecular basis for loss of connectivity in Alzheimer's disease. J Neurosci. 2007 Jan 24;27(4):796-807. PubMed.
  16. . The pathological process underlying Alzheimer's disease in individuals under thirty. Acta Neuropathol. 2011 Feb;121(2):171-81. PubMed.

Other Citations

  1. rTgP301L mouse

Further Reading

Primary Papers

  1. . Tau mislocalization to dendritic spines mediates synaptic dysfunction independently of neurodegeneration. Neuron. 2010 Dec 22;68(6):1067-81. PubMed.
  2. . Amyloid-β/Fyn-induced synaptic, network, and cognitive impairments depend on tau levels in multiple mouse models of Alzheimer's disease. J Neurosci. 2011 Jan 12;31(2):700-11. PubMed.