One of the earliest signs of amyloid-β (Aβ) toxicity is synaptic loss, and recent work from a number of labs has shown that Aβ causes depressed synaptic transmission due to removal of cell surface neurotransmitter receptors (see ARF related news story). In a new paper, Roberto Malinow and colleagues at Cold Spring Harbor Laboratory, New York, present a wealth of data showing that Aβ-induced internalization of synaptic AMPA-type glutamate receptors is the key event leading to synaptic depression, and ultimately, the disappearance of dendritic spines. The effects of Aβ turn out to have a lot in common with long-term depression, a mechanism of synaptic plasticity that results from increased phosphorylation and endocytosis of synaptic AMPARs. Based on their results, Malinow and colleagues offer stabilization of AMPARs as a potential therapeutic approach for Alzheimer disease.

Stabilization of AMPA receptors turns out to be a specialty of the synaptic protein PSD95, according to another new paper from Richard Huganir and colleagues at Johns Hopkins University School of Medicine, Baltimore, Maryland, and the Howard Hughes Medical Institute. Their results show that PSD95 knockout mice lose AMPA receptors, but only in a subset of synapses. PSD95 is a potential target for Aβ’s action on synapses, but Huganir’s results show that getting rid of PSD95 creates a different phenotype than Aβ treatment. In contrast to Aβ’s effects on spine density, the loss of AMPAR in response to PSD95 deletion did not affect spine morphology.

The publication of the Malinow paper follows first author Helen Hsieh’s presentation of some of the results at SfN 2005 (see ARF related news story). There, she showed that expressing APP in hippocampal neurons led to the depression of AMPA and NMDA receptor-related neurotransmission in a way that mimics LTD. Hsieh’s results showed that production of Aβ occluded LTD, which suggests the two use a common pathway. Like LTD, the internalization stimulated by Aβ required calcineurin and p38 MAP kinase signaling.

In their paper, they present all this data and add the link to dendritic spine loss. The system they use, organotypic cultures of hippocampal pyramidal neurons, is amenable to transfection experiments. They show that overexpression of APP and the ensuing production of Aβ, or treatment of cells with added Aβ, reduced spine density by 30 percent in the cultures. They used a variety of mutants in the AMPA receptor GluR2 subunit to probe the mechanism of spine loss. As they had shown for synaptic depression, spine loss required internalization of the AMPA receptor—neither occurred in cells expressing an endocytosis-resistant mutant of the GluR2 subunit of the AMPA receptor. These results suggested that the removal of AMPAR by endocytosis is necessary to see spine loss.

To test if AMPAR removal was sufficient to trigger synaptic loss, they used a phosphomimetic GluR2 mutant (R607Q;S880E) that is constitutively internalized. Phosphorylation of S880 is required to stimulate receptor endocytosis during LTD. Expression of the mutant led to a decrease in both AMPAR- and NMDAR-mediated neurotransmission and dendritic spine loss, showing that removal of AMPA receptors can deal a fatal blow to synapses. Tying this result back to Aβ, they show that in cells expressing GluR2, Aβ treatment results in a small but significant increase in phosphorylation at S880.

From this, Hsieh and colleagues conclude that the loss of synaptic AMPA receptors is driven by Aβ-induced phosphorylation of GluR2, and leads to spine loss. This loss, in turn, may account for the observed decrease in NMDAR-mediated synaptic transmission. “Our results indicate that Aβ drives the removal of synaptic AMPA receptors and this plays a key role in the toxic effects of Aβ on spines,” they write. Their findings suggest that strategies to stabilize AMPAR may be one way to treat AD.

How Aβ taps into the LTD pathway is unclear, but one idea that has been around for a while is that Aβ destabilizes synapses via its interaction with the synaptic protein PSD95. This protein, which regulates expression of both NMDA and AMPA glutamate receptors, has been shown by Bill Klein’s work to colocalize with Aβ in synapses (see Lacor et al., 2004). Work from Claudia Almeida and Gunnar Gouras earlier this year showed that in neuronal cultures from AD mice, loss of AMPAR was preceded by a decrease in the PSD95 protein.

Now, Huganir and colleagues show that in PSD95 knockout mice, there is a loss of AMPA receptors, but only in selected synapses. Previous PSD95 knockouts did not cause total loss of protein function, so first author Jean-Claude Beique and coworkers generated a true null allele. They found that AMPAR-mediated synaptic transmission was reduced in homozygous knockout mice, but there was no change in NMDA function. When they analyzed the function of individual synapses using a caged glutamate compound, they found that, curiously, the knockout had synapse-specific effects. Many synapses were unaffected by the knockout, while in others AMPAR-mediated neurotransmission was absent.

The loss of PSD95 did not affect the ability to enhance AMPAR numbers during LTP. On the contrary, the mice showed enhanced LTP compared to wild-type mice. The researchers speculate that this reflects the increased number of AMPAR-negative synapses at baseline in the knockout mice. The results indicate that PSD95 mainly plays a role in maintaining the stability of synaptic AMPARs.

In contrast to the Malinow observations, AMPAR-lacking synapses were found on normal, mature spines, and the knockout mice showed no changes in spine volume. These results suggest that while disruption of PSD95 or addition of Aβ can both downregulate AMPAR levels and synaptic function, the ramifications appear very different for the health of dendritic spines.—Pat McCaffrey.

Hsieh H, Boehm J, Sato C, Iwatsubo T, Tomita T, Sisodia S, Malinow R. AMPAR Removal Underlies Abeta-Induced Synaptic Depression and Dendritic Spine Loss. Neuron. 2006 Dec 7;52(5):831-43. Abstract

Beique JC, Lin DT, Kang MG, Aizawa H, Takamiya K, Huganir RL. Synapse-specific regulation of AMPA receptor function by PSD-95. Proc Natl Acad Sci U S A. 2006 Dec 5; [Epub ahead of print] Abstract


Make a Comment

To make a comment you must login or register.

Comments on this content

  1. Hsieh and colleagues provide exciting results on a major question in AD
    research: how β amyloid induces synaptic dysfunction.
    In well-controlled experiments, they show that the mechanisms of Aβ
    effects on synapses parallel those involved in LTD. They confirm that Aβ
    causes internalization of AMPA receptors, suggesting that this is required
    for subsequent alterations in NMDA receptors. Importantly, an endocytosis-defective GluR2 construct prevents the physiological alterations induced by Aβ.
    This work strengthens the idea that Aβ has a role in synaptic biology, although
    the molecular mechanisms whereby Aβ causes internalization of AMPA
    receptors and promotes pathology at synapses in AD remain unclear.

  2. Our findings reported in Neuron indicate that removal of synaptic AMPAR by Aβ leads to loss of spines and loss of NMDA responses. In other cases, removal of synaptic AMPAR does not lead to loss of NMDA responses (e.g., Shi et al., 2001; expression of GluR2 c-tail leads to selective decrease in AMPAR responses). In the Beique et al. paper, which is a very nice study, they find that in PSD95 knockout animals, there are some large spines with no AMPAR responses. So I would conclude that removal of AMPAR from synapses can lead to spine loss, but it appears to be dependent on how those receptors are lost. There may be AMPAR-associated molecules, which normally stabilize spines, that are also removed by Aβ but can persist in PSD95 KO animals.


    . Subunit-specific rules governing AMPA receptor trafficking to synapses in hippocampal pyramidal neurons. Cell. 2001 May 4;105(3):331-43. PubMed.

  3. The study by Malinow and colleagues provides compelling evidence for a direct link between the synaptic deficits associated with Aβ production and known cellular pathways for physiological synaptic plasticity. This work is important, as it points to several well-established molecular mechanisms of glutamate receptor trafficking as potential early mediators of amyloid-induced synaptic dysfunction.

    The study by Huganir and colleagues sheds new light on the fundamental molecular events regulating synaptic transmission and excitatory synapses in the brain. The authors generated mutant mice lacking a critical protein component of excitatory synapses, the scaffold molecule PSD-95. In brain slices from these animals, synaptic transmission was impaired, but plasticity was actually increased. These findings are important because they pinpoint the functional defects in synaptic transmission produced when synapses lack PSD-95. Together with the results of Malinow and colleagues, one can begin to envision an unraveling of the molecular mechanisms underlying synaptic failure in Alzheimer disease and other disorders of memory and cognition.

  4. This is a landmark paper on how Aβ peptides might induce synaptic depression by reducing the number of AMPA receptors at postsynaptic sites of hippocampal neurons. The authors propose that Aβ peptides, either generated endogenously from APP, or added exogenously, induce endocytosis mechanisms that are thought to underlie long-term depression. Two questions, which I imagine the authors are now poised to answer, are (i) the identity of the active Aβ peptide (is it a monomer dimer, oligomer, or Aβ*?) and (ii) how it acts on the postsynaptic membrane. My guess is that the active Aβs must be either monomers or dimers, since the effect is seen when Aβs are generated from endogenous APP, and there is no reason to expect large amounts of Aβ to be generated at concentrations needed to oligomerize. Synthetic Aβ peptides added to the medium of the hippocampal slices need not act only on the external surface of the membrane, since they were added at concentrations (micromolar) high enough to enter and cross the lipid bilayer, giving them access to the cytoplasmic surface. If present at high enough concentration, they might affect in some way the functioning of the cytoskeletal network which must play a role in the endocytosis process.

    Alternatively, they could partition within the lipid bilayer, as was proposed earlier (1). If a fraction of the Aβ peptide pool remained within the lipid bilayer, their hydrophobic domains, which contain GxxxG/A sequences, could conceivably interact with comparable domains in the TARP proteins, all of which have two hydrophobic and putative transmembrane segments that are similar but not identical to the Aβ domains. A comparison between Aβ and one of the T-M domains is shown below:


    Interactions between Aβ peptides and T-M segments of TARP proteins could affect the AMPA receptor trafficking mechanisms proposed earlier (2), and this would be an alternative way in which Aβ peptides might reduce the AMPA receptor population. A test of this idea would be to express Aβ alone in hippocampal neurons with a vector that incorporates an appropriate signal sequence along with the Aβ segment and some anchoring sequence at its C-terminus.


    . An alternative interpretation of the amyloid Abeta hypothesis with regard to the pathogenesis of Alzheimer's disease. Proc Natl Acad Sci U S A. 2005 Jun 28;102(26):9093-8. PubMed.

    . Auxiliary subunits assist AMPA-type glutamate receptors. Science. 2006 Mar 3;311(5765):1253-6. PubMed.

  5. The classical view of amyloid action in the pathogenesis of Alzheimer disease centers around the neurotoxic properties of aggregated peptides. Recent studies have, however, been challenging this as the exclusive mechanism, suggesting direct modulatory functions for Aβ; compelling evidence that supports this view is now provided by the paper by Hsieh et al., who offer some novel mechanistic insights. They use a clever combination of imaging (with synaptopHluorin tagged receptors) and electrophysiological (electrophysiological tagging of AMPA receptors) techniques. The authors show that application of synthetic Aβ or the overexpression of C99 causes endocytosis of GluR1 and Glur2 receptors from synapses. Furthermore, Hsieh et al. show the involvement of p38/MAPK and calcineurin in GluR2 endocytosis, and detect an increase in the phosphorylation of the cytoplasmic tail of GluR2 after Aβ treatment. Based on these findings, Hsieh et al. suggest that Aβ induces GluR2 endocytosis through a pathway that may be shared with LTD induction. In fact, they also show that Aβ can mimic and partly occlude mGluR1-induced LTD. Finally, Hsieh et al. bridge the difficult gap between electrophysiological and morphological plasticity and show that Aβ treatment, as well as C99 overexpression, leads to synaptic spine loss, an effect that can be mimicked by expressing AMPAR, which bears mutations that result in enhanced endocytosis; in doing the reverse experiment, they demonstrate that the same mutated AMPAR attenuates Aβ-induced spine loss. On the basis of previous data from this same group (Kamenetz et al., 2003) as well as from Cirrito et al. (2005), it is now suggested that Aβ cleavage is dependent upon synaptic activity which, together with the data from Hsieh et al., strongly backs the view of Aβ as modulator of synaptic activity. In summary, the studies by Hsieh et al. indicate that Aβ might operate in a negative feedback loop to homeostatically regulate synaptic strength.


    . APP processing and synaptic function. Neuron. 2003 Mar 27;37(6):925-37. PubMed.

    . Synaptic activity regulates interstitial fluid amyloid-beta levels in vivo. Neuron. 2005 Dec 22;48(6):913-22. PubMed.

  6. Hsieh et al. conducted carefully executed experiments addressing the mechanisms underlying Aβ-induced depression of glutamatergic synaptic transmission. Their work provides convincing evidence that Aβ causes internalization of AMPA receptors which results in synaptic dysfunction and dendritic spine loss. These results further support the link between Aβ and glutamate receptor function during pathological onset in AD.

    We wish to learn more about how AMPA receptor internalization leads to the loss of functional NMDA receptors. Specifically, the mechanisms by which loss of AMPA receptors leads to the loss of NMDA receptors remain unclear. There is undoubtedly a correlation between the two events because of their anatomical location and functional relation; however, there are several steps in the process that have yet to be revealed.


News Citations

  1. Amyloid-β Zaps Synapses by Downregulating Glutamate Receptors
  2. SfN: Where, How Does Intraneuronal Aβ Pack Its Punch? Part 2

Paper Citations

  1. . Synaptic targeting by Alzheimer's-related amyloid beta oligomers. J Neurosci. 2004 Nov 10;24(45):10191-200. PubMed.
  2. . AMPAR removal underlies Abeta-induced synaptic depression and dendritic spine loss. Neuron. 2006 Dec 7;52(5):831-43. PubMed.
  3. . Synapse-specific regulation of AMPA receptor function by PSD-95. Proc Natl Acad Sci U S A. 2006 Dec 19;103(51):19535-40. PubMed.

Further Reading


  1. . AMPAR removal underlies Abeta-induced synaptic depression and dendritic spine loss. Neuron. 2006 Dec 7;52(5):831-43. PubMed.
  2. . Synapse-specific regulation of AMPA receptor function by PSD-95. Proc Natl Acad Sci U S A. 2006 Dec 19;103(51):19535-40. PubMed.

Primary Papers

  1. . AMPAR removal underlies Abeta-induced synaptic depression and dendritic spine loss. Neuron. 2006 Dec 7;52(5):831-43. PubMed.
  2. . Synapse-specific regulation of AMPA receptor function by PSD-95. Proc Natl Acad Sci U S A. 2006 Dec 19;103(51):19535-40. PubMed.