The γ-secretase has come a long way since its original discovery as the protease responsible for the production of amyloid-β peptides during sequential cleavage of the amyloid precursor protein. The unusual intramembrane cleavage catalyzed by the secretase turns out to be a popular regulatory maneuver for many cells, and the list of γ-secretase substrates continues to expand, encompassing a myriad of functions both in and outside of the nervous system. A report in the May 4 Journal of Cell Biology adds a new substrate to the tally, one with a critical role in synapse formation and plasticity. In the paper, Eiji Inoue and colleagues at the KAN Research Institute in Kobe, Japan (part of the Eisai pharmaceutical company), propose that γ-secretase processing of the receptor tyrosine kinase EphrinA4 (EphA4) in synapses, and release of its intracellular domain fragment, promotes dendritic spine formation. Two γ-secretase mutants that cause familial early onset AD show decreased activity for Eph4A processing. The results add support to the idea that γ-secretase mutations induce disease via a loss-of-function mechanism, and suggest that multiple substrates could mediate the mutants’ effects. In the case of EphA4, the FAD mutations might lead to lower dendritic spine density, and synapse loss.

Using mouse hippocampal neurons as a model system, Inoue and colleagues first determined by immunofluorescent staining that the catalytic subunit of γ-secretase, presenilin1 (PS1), resides in synapses. The enzyme seemed to have a hand in the formation and maintenance of synapses, since treatment of cells with a γ-secretase inhibitor reduced the clustering of AMPA-type glutamate receptors, a step in synapse maturation, and also reduced the density of dendritic spines, the structures that harbor synapses, by 20-25 percent. Similar results were seen when both presenilin genes (PS1 and PS2) were silenced by RNAi knockdown. The results are consistent with previous work showing a loss of synapses in PS1/PS2 conditional knockout mice (see ARF related news story and Saura et al., 2004).

To identify the γ-secretase substrate(s) responsible for the synaptic effects, the investigators purified the cholesterol-rich lipid raft membrane fraction from crude synaptic membranes, where PS1 has been found to be colocalized with other substrates. They used mass spec to identify 324 proteins in the fraction. The batch included EphA4, a transmembrane receptor and a likely substrate candidate. EphA4 bears a consensus cleavage site for γ-secretase, and it is a relative of the known γ-secretase substrate EphB (Georgakopoulos et al., 2006). Moreover, EphA4 signaling has been implicated in dendrite retraction (see Murai et al., 2003 and ARF related news story on Fu et al., 2007).

Like other secretase substrates, EphA4 has an extracellular domain that needs to be cleaved before γ-secretase can get at the transmembrane region. That cleavage is accomplished in the neurons by matrix metalloprotease, the authors show. After that, processing of the remaining C-terminal fragment depended on γ-secretase, as it was blocked by the secretase inhibitor or by PS1/2 knockdown. In vitro experiments showed that γ-secretase could release an intracellular domain fragment (ICD) of EphA4 from membrane preparations of cells overexpressing the receptor. In contrast to other γ-secretase substrates including Notch and EphB, EphA4 did not require ligand binding to trigger processing. Instead, processing appeared to be regulated by synaptic activity and glutamate receptor activation.

To look for a functional effect of the ICD in cells, the researchers overexpressed that fragment only. They found the ICD was sufficient to increase spine number, and that the γ-secretase inhibitor no longer affected spine density in the ICD-expressing cells. “These results suggest that the processing of EphA4 by γ-secretase has a critical role in the formation of dendritic spines, and that EphA4 is one of the critical substrates of γ-secretase that regulates the morphogenesis of dendritic spines,” the authors conclude. In further support of this idea, they show that Eph4A knockdown in cultured rat hippocampal primary neurons decreased spine number, with a strong effect on the mushroom spines containing mature synapses. The γ-secretase inhibitor had no effect on spine density in these cells.

How does Eph4A cleavage regulate spines? In NIH3T3 fibroblasts, overexpressing the EphA4 ICD activated the Rac signaling pathway and led to actin cytoskeleton reorganization and the formation of lamellipodia. In neurons, Rac regulates dendritic spine morphogenesis (Tashiro and Yuste, 2004), and the authors showed that expression of a dominant-negative Rac inhibited increased dendrite number in response to the EphA4 ICD in cultured neurons. The ICDs of several γ-secretase substrates have been shown to move to the nucleus and regulate gene expression, but neither nuclear localization nor the kinase activity of EphA4 was necessary for Rac activation or the increase in lamellipodia in NIH3T3 cells. The results suggest a complex dual role for EphA4 in synapse formation. Where previous work showed that binding of Ephrin protein ligands to EphA4 results in dendrite retraction, the current study suggests that ligand-independent processing can promote dendrite formation.

To look at the effect of AD PS1 mutants, the researchers measured EphA4 processing by γ-secretase in membrane preparations from PS1/2 knockout cells that were engineered to express normal PS1, a catalytically dead mutant (PS1D385A) or one of two AD mutants (M146L or E280A). The AD mutant proteins made much less ICD than the wild-type PS1, indicating that processing of EphA4 was impaired by FAD-linked mutations. More work is needed to see if ineffective processing as a result of FAD mutations plays a role in synapse loss in AD, or presents another complication to the use of γ-secretase inhibitors as potential treatments.—Pat McCaffrey


  1. This work by Inoue et al. presents exciting evidence that the PS1/γ-secretase system regulates synaptic physiology through the cleavage of EphA4 receptor. The Eph receptors and the Ephrin ligands are very important multimember families of proteins known to play critical roles in synaptic function and integrity. The regulated cleavage of EphA4 receptor by PS1/γ-secretase is very interesting since this receptor participates in the morphological shaping of dendritic spines, induction of LTP and LTD, and the regulation of structural plasticity of synaptic connections in the mature brain.

    By downregulating PS in neurons and using γ-secretase inhibitors the authors show that presenilin/γ-secretase regulates the morphogenesis of dendritic spines and more specifically regulates the number and shape of synapses as well as the clustering of AMPA receptors. The molecular mechanism through which PS1/γ-secretase exerts this function involves the activation of Rac1 by the product of γ-secretase cleavage of EphA4 (peptide EphA4 ICD), which interacts with and activates Rac1, leading to cytoskeletal rearrangements. Inhibition of these functions in disease may thus lead to impaired cytoskeletal dynamics in neurons and eventually loss of synaptic connections in the brain. Interestingly, the authors show that mutations of PS1 found in familial Alzheimer disease inhibit the processing of EphA4, as it has been previously shown for cadherins, EphrinB ligands, and EphB receptors, confirming the observation that these mutations cause loss of function, potentially leading to neurodegeneration in AD (Marambaud et al., 2003).

    The regulation of EphA4 by PS1/γ-secretase is of particular interest since EphA4 receptors play a critical role in the development and function of neural connections in the brain and synaptic abnormalities correlate more closely with the degree of dementia than any other histopathological hallmark of AD.


    . A CBP binding transcriptional repressor produced by the PS1/epsilon-cleavage of N-cadherin is inhibited by PS1 FAD mutations. Cell. 2003 Sep 5;114(5):635-45. PubMed.

  2. Presenilin is the catalytic subunit of γ-secretase, which is responsible for the cleavage of amyloid-precursor protein (APP), leading to the generation of the amyloidogenic 42-residue β-amyloid peptide (Aβ) and formation of senile plaques that are hallmark features in Alzheimer disease (AD). Familial Alzheimer disease (FAD) is associated with mutations of the presenilin (PS) gene, but it is not clear how the dysfunction of presenilin accounts for the AD pathophysiology. One possibility is a toxic gain-of-function mechanism that involves increase in Aβ production and subsequent cell death. Alternatively, increasing evidence suggests that PS may have a crucial role at synapses under normal physiological condition, and loss-of-function mutations in PS1 gene may account for the impaired synaptic plasticity and memory formation associated with AD (Saura et al., 2004). This later hypothesis is supported by the recent paper by Inoue et al., which reveals a novel function of PS in promoting the clustering of AMPA receptors and formation of dendritic spines in hippocampal neurons via processing of the receptor tyrosine kinase EphA4.

    In this study, PS1 was shown to be enriched in the synaptic lipid raft of hippocampal neurons. Inhibition of γ-secretase reduces both the clustering of AMPA receptors and the density of dendritic spines. To understand the mechanisms underlying the reduction of synapse number upon inhibition of γ-secretase, the authors went on to identify its substrates by performing mass spectrometry of the synaptic lipid raft fraction. Among the ~300 proteins identified in the lipid raft is EphA4, which is cleaved by matrix metalloproteases (MMP) followed by γ-secretase to generate the intracellular domain (EphA4-ICD). The processing of EphA4 to generate the ICD is induced by synaptic activity (treatment with forskolin, rolipram, and bicuculline). Importantly, expression of EphA4-ICD in dissociated hippocampal neurons promotes the formation of dendritic spines through activation of the Rho family of GTPase Rac1. Moreover, the inhibitory effect of γ-secretase inhibitor on spine formation and AMPA receptor clustering is abolished either by the expression of EphA4-ICD or knockdown of EphA4 expression. Together these findings suggest that EphA4 processing and generation of ICD by γ-secretase may be crucial for activity-dependent spine morphogenesis during synaptic plasticity. In addition, the PS1 construct that harbors FAD-linked PS1 mutation fails to cleave EphA4, raising the possibility that impaired processing of EphA4 may underlie the memory formation deficit in AD patients.

    Studies from my laboratory and others have found that activation of EphA4 by its cognate ligand Ephrin results in spine retraction through activation of another Rho GTPase, RhoA, and inhibition of integrin signaling pathway (Murai et al., 2003; Fu et al., 2007; Bourgin et al., 2007). In contrast, the processing of EphA4 to produce ICD by γ-secretase and the subsequent activation of Rac1 do not depend on ligand binding, but instead can be enhanced by synaptic activity. The study by Inoue et al., therefore, raises an interesting notion that EphA4 can be differentially coupled to activation of different Rho GTPases in response to distinct stimuli (synaptic activity vs. Ephrin-binding), leading to opposing consequences in spine density. One important missing link in their hypothesis is the direct demonstration that increasing synaptic activity in either hippocampal slices or dissociated culture can indeed promote spine formation in a PS1- or EphA4-dependent manner. Further examination on EphA4 processing and spine morphogenesis in conditional PS1/2 double knockout mice (Saura et al., 2004) will also provide in vivo evidence to strengthen this interesting hypothesis. Nonetheless, this and other studies that reveal potential physiological roles of PS and γ-secretase in normal functioning of synapses might raise concerns about potential side effects of using γ-secretase inhibitors as a therapeutic strategy for AD.


    . The EphA4 receptor regulates dendritic spine remodeling by affecting beta1-integrin signaling pathways. J Cell Biol. 2007 Sep 24;178(7):1295-307. PubMed.

    . Cdk5 regulates EphA4-mediated dendritic spine retraction through an ephexin1-dependent mechanism. Nat Neurosci. 2007 Jan;10(1):67-76. PubMed.

    . Control of hippocampal dendritic spine morphology through ephrin-A3/EphA4 signaling. Nat Neurosci. 2003 Feb;6(2):153-60. PubMed.

    . Loss of presenilin function causes impairments of memory and synaptic plasticity followed by age-dependent neurodegeneration. Neuron. 2004 Apr 8;42(1):23-36. PubMed.

  3. The paper by Inoue et al. adds important new evidence to the synaptic function of EphA4, a receptor involved in the development and function of the CNS. In addition, it verifies reports that PS1 is localized at synaptic contacts (Georgakopoulos et al., 1999) and indicates that this protein may promote formation of dendritic spines and synapses. Interestingly, PS1 is reported to specifically affect clustering and localization of AMPA receptors, but it seems to have little or no effect on other synaptic markers like synaptophysin and PSD95. The lack of effects on these synaptic markers may be explained by the suggestion of specific effect on "silent" synapses. Interestingly, the effects of γ-secretase inhibitors on the size of synapses and PSD95 clusters may indicate involvement of synaptic cadherins, a class of proteins processed by γ-secretase and known to regulate synaptic structure and function.

    Importantly, the paper also reports inhibitory effects of PS1 FAD mutations on the processing of EphA4 protein. This result is in excellent agreement with earlier reports that the γ-secretase processing of many synaptic proteins, including cadherins, EphB receptors, and EphrinB ligand proteins, is inhibited by FAD mutations. Thus, the report by Inoue et al. adds to accumulating evidence that loss of PS and γ-secretase functions caused by FAD mutations may be involved in the mechanism by which these mutations promote neurodegeneration and AD (see also Marambaud et al., 2003).


    . Presenilin-1 forms complexes with the cadherin/catenin cell-cell adhesion system and is recruited to intercellular and synaptic contacts. Mol Cell. 1999 Dec;4(6):893-902. PubMed.

    . A CBP binding transcriptional repressor produced by the PS1/epsilon-cleavage of N-cadherin is inhibited by PS1 FAD mutations. Cell. 2003 Sep 5;114(5):635-45. PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. The Senility-Presenilin Connection Turned Upside Down
  2. What Drives Dendritic Spine Loss? Study Taps Cdk5

Paper Citations

  1. . Loss of presenilin function causes impairments of memory and synaptic plasticity followed by age-dependent neurodegeneration. Neuron. 2004 Apr 8;42(1):23-36. PubMed.
  2. . Metalloproteinase/Presenilin1 processing of ephrinB regulates EphB-induced Src phosphorylation and signaling. EMBO J. 2006 Mar 22;25(6):1242-52. PubMed.
  3. . Control of hippocampal dendritic spine morphology through ephrin-A3/EphA4 signaling. Nat Neurosci. 2003 Feb;6(2):153-60. PubMed.
  4. . Cdk5 regulates EphA4-mediated dendritic spine retraction through an ephexin1-dependent mechanism. Nat Neurosci. 2007 Jan;10(1):67-76. PubMed.
  5. . Regulation of dendritic spine motility and stability by Rac1 and Rho kinase: evidence for two forms of spine motility. Mol Cell Neurosci. 2004 Jul;26(3):429-40. PubMed.

Further Reading

Primary Papers

  1. . Synaptic activity prompts gamma-secretase-mediated cleavage of EphA4 and dendritic spine formation. J Cell Biol. 2009 May 4;185(3):551-64. PubMed.