|Home | About | Journals | Submit | Contact Us | Français|
D.E.B. designed, coordinated and carried out the bulk of experiments, and contributed to the writing of the manuscript. C.D. performed the fluorescent PtdIns(4,5)P2/DAG probe experiments and image analysis. S.V.V. carried out the Western blot analysis, lipid measurements in Synj1+/− neurons and ATP measurements. H.Z. and O.A. contributed the electrophysiology experiments. L.B.J.M. and A.Z.M. prepared and analyzed synthetic and cell-derived Aβ species. A.S. and O.A. contributed the behavioral experiments. T.-W.K. and G.D.P. conceived and supervised the project, wrote the manuscript and contributed equally to this work.
Synaptic dysfunction caused by oligomeric assemblies of amyloid-β peptide (Aβ) has been linked to cognitive deficits in Alzheimer's disease. Here we found that incubation of primary cortical neurons with oligomeric Aβ decreases the level of phosphatidylinositol-4,5-bisphosphate (PtdIns(4,5)P2), a phospholipid that regulates key aspects of neuronal function. The destabilizing effect of Aβ on PtdIns(4,5)P2 metabolism was Ca2+-dependent and was not observed in neurons that were derived from mice that are haploinsufficient for Synj1. This gene encodes synaptojanin 1, the main PtdIns(4,5)P2 phosphatase in the brain and at the synapses. We also found that the inhibitory effect of Aβ on hippocampal long-term potentiation was strongly suppressed in slices from Synj1+/− mice, suggesting that Aβ-induced synaptic dysfunction can be ameliorated by treatments that maintain the normal PtdIns(4,5)P2 balance in the brain.
Alzheimer's disease is the most common form of late-onset dementia. Although the molecular mechanisms responsible for the development of the idiopathic cases of Alzheimer's disease (that is, without any obvious genetic basis) are unknown in the majority of patients, growing evidence indicates that cerebral elevation and accumulation of Aβ peptide mediate many aspects of Alzheimer's disease pathogenesis. Aβ peptide is produced by the sequential proteolytic cleavage of amyloid precursor protein (APP) by themembrane-bound proteases β- and γ-secretase1–3. Intramembraneous cleavage of the COOH terminus of Aβ peptide by presenilin, the active component of the γ-secretase complex, gives rise to two major species, Aβ1–40 and Aβ1–42. Aβ1–40 is the predominant cleavage product, although the longer and less abundant product Aβ1–42 (Aβ42) is more amyloido-genic and is thus believed to be a key cytotoxic agent in Alzheimer's disease. Although Alzheimer's disease brains generally contain amyloid plaques that consist of insoluble aggregates of Aβ, mounting evidence indicates that different conformations of Aβ, such as Aβ oligomers and fibrils, may contribute to Alzheimer's disease pathogenesis via distinct mechanisms at different stages of the disease2. Notably, accumulation of soluble oligomeric forms of Aβ in the extracellular space closely correlates with cognitive decline and/or disease progression in animal models and Alzheimer's disease patients2. Consistent with these studies, oligomers of Aβ, particularly Aβ42, have been shown to directly affect synaptic plasticity and trigger the loss of synaptic dendritic spines2,4–6. Although we are beginning to understand Aβ-induced synaptic alterations that lead to cognitive decline, the molecular mechanisms underlying these defects are still unclear. Important clues have been provided by studies showing that Aβ can modulate cell surface levels of AMPA and NMDA receptors and affect calcium homeostasis, suggesting that a fundamental mode of action of Aβ is to alter the signaling properties and permeability of synaptic membranes2,4–7. Whether Aβ directly alters lipid bilayers by forming pores that ultimately modify their conductance8–10 or acts through proteinaceous receptors (for example, α7 nicotinic acetylcholine receptor)11 is a matter of intense debate. Similarly, the intracellular signaling pathways perturbed by Aβ at synapses are under extensive investigation, as their elucidation may pave the way for the identification of suitable drug targets and the development of new therapeutic interventions.
Phosphoinositides (that is, phosphorylated derivatives of phosphatidylinositol) are major signaling phospholipids in cellular membranes that regulate many aspects of physiology. Of the seven known phosphoinositides, PtdIns(4,5)P2 and PtdIns(3,4,5)P3 are the most relevant for processes occurring at the cell surface, as they are concentrated at the cytosolic leaflet of the plasma membrane and control signal transduction, actin dynamics, exo-endocytosis and the permeability of ion channels12–15. Although a potential link between Alzheimer's disease and phosphoinositides was first suggested by a study that found reduced levels of inositol lipids in the brains of individuals with Alzheimer's disease16, our recent studies have demonstrated that familial Alzheimer's disease (FAD) mutations in the genes encoding presenilin 1 and presenilin 2 (PSEN1 and PSEN2, respectively) result in a perturbation of the metabolism of PtdIns(4,5)P2, which, in turn, affects Ca2+ currents through transient receptor potential channels as well as the biogenesis of Aβ17. More specifically, this study showed that downregulating the levels of PtdIns(4,5)P2, either indirectly by expression of the presenilin FAD mutants or directly by overexpressing a PtdIns(4,5)P2 5-phosphatase, enhances the production of Aβ42, suggesting that an imbalance in PtdIns(4,5)P2 may underlie some aspects of FAD pathogenesis17. However, the mechanism by which presenilin mutations destabilize PtdIns(4,5)P2 metabolism is unclear.
Because elevated Aβ levels are believed to underlie key pathogenic aspects of both idiopathic and FAD, we set out to expand our studies on the link between Alzheimer's disease and phosphoinositide metabolism by investigating whether Aβ per se can affect the metabolism of PtdIns(4,5)P2, and if so, whether this potential effect is pathophysiologically relevant, particularly with respect to Alzheimer's disease–linked synaptic dysfunction. We found that Aβ oligomers destabilize the metabolism of PtdIns(4,5)P2 both acutely and chronically in primary cortical cultures and that preventing this destabilization by genetic means suppresses some of the synapse-impairing actions of Aβ oligomers.
To test whether soluble oligomers of Aβ42 affect PtdIns(4,5)P2 metabolism in neurons, we prepared primary cultures from mouse neonatal cortices, allowed themto differentiate for 15 d and incubated them with a crude oligomer preparation made from synthetic Aβ42 peptides (oAβ42) (Supplementary Fig. 1 online). We determined the effects of this preparation after either an acute (0–120 min) or a subchronic treatment (72 h), under conditions where we observed no obvious cell death or changes in ATP levels (Supplementary Fig. 2 online). The levels of phosphoinositides, as well as a variety of other anionic phospholipids, including phosphatidic acid (PtdA), cardiolipin (or diphosphatidylglycerol, DPtdG) and phosphatidylserine (PtdS), were measured and quantified in neuronal extracts using high-performance liquid chromatography with suppressed conductivity detection17,18.
oAβ42, whichwas used at 200 nM(that is, a concentration that is lower than that typically used in neurotoxicity and cell-death studies), induced a rapid and progressive decrease in the levels of PtdIns(4,5)P2, which stabilized at approximately 60% of control (vehicle) levels after 120 min (Fig. 1a). After a trend toward a transient increase, the levels of PtdIns4P also decreased in response to oAβ42, but the effect was more subtle than that observed for PtdIns(4,5)P2 and was found only after a 120-min treatment (Fig. 1a). No changes were observed in the levels of most other lipids, although we found a trend for an increase in PtdA, but with a higher variability than all the other anionic phospholipids (Fig. 1a). We did not observe an effect on PtdIns(4,5)P2 or the other lipids when experiments were done using a control peptide that contained the inverse sequence (Aβ42Rev) or with a preparation of the shorter (1–38) and noncytotoxic Aβ peptide, Aβ38, which was processed for oligomerization similarly to oAβ42 (Fig. 1b). Notably, Aβ-induced PtdIns(4,5)P2 deficiency was rescued by preincubating oAβ42 with the antibody 6E10 (Fig. 1b), which has been shown to neutralize the toxic effects of Aβ oligomers19. The antibody alone had no effect on PtdIns(4,5)P2 levels (Fig. 1b). Oligomers of Aβ42 also exerted a sustained decrease in the levels of PtdIns(4,5)P2 after 3 (Fig. 1c) or 7 d (Supplementary Fig. 2) of treatment, suggesting that the effects of oligomers on this lipid are long-lasting. In contrast to the acute time-course experiments, prolonged exposures to oAβ42 failed to cause a decrease in the levels of PtdIns4P, indicating that the long-lasting effect was specific for PtdIns(4,5)P2. When Aβ42-containing medium was replaced with control medium after either 3 or 7 d of treatment, the levels of PtdIns(4,5)P2 returned to normal levels 3 d later (Fig. 1d and Supplementary Fig. 2), demonstrating that the peptide effect is reversible. Finally, oAβ42-induced alterations of the levels of PtdIns(4,5)P2 occurred in the absence of major changes in the expression levels of the two main enzymes regulating the synthesis and elimination of PtdIns(4,5)P2, PtdInsP kinase type 1–γ and synaptojanin 1, respectively (Supplementary Fig. 3 online).
In light of recent studies suggesting that soluble oligomers of Aβ42 are probably the most cytotoxic species of the Aβ family with respect to synaptic function, we next addressed whether different assembly states of Aβ42 differentially affect PtdIns(4,5)P2 metabolism.
We found that the effect of Aβ42 on primary cortical neurons was markedly selective for crude oligomer preparations, as crude monomer and fibril preparations did not affect PtdIns(4,5)P2 levels after a 3-d treatment (Fig. 2a and Supplementary Fig. 1). To further establish the role of oligomers as phosphoinositide-destabilizing agents, we capitalized on the existence of a naturally occurring hexahydroxy alcohol, referred to as scyllo-inositol, which has been shown to antagonize the neurotoxic effects of soluble Aβ42 oligomers19–21. Notably, we did not observe oAβ42-induced PtdIns(4,5)P2 deficiency when neurons were incubated for 3 d with both the peptide and 5 μM scyllo-inositol (Fig. 2b). The effect was specific for scyllo-inositol, as another inositol enantiomer that is devoid of Aβ42 oligomer-neutralizing properties, chiro-inositol, did not show any protective action on PtdIns(4,5)P2. When added alone, neither of the two compounds had any effect on PtdIns(4,5)P2 levels (Fig. 2b). Together, these results suggest that oligomeric assemblies of Aβ42 may be particularly potent at destabilizing PtdIns(4,5)P2 metabolism compared with the monomers and the fibrils.
Next, to test whether cell-derived Aβ also affects PtdIns(4,5)P2 levels, we used cells expressing the ‘Swedish’ mutant of APP (APPsw) because of the well-described increase in Aβ production associated with this mutant22. Primary neurons were prepared from the cortex of transgenic mice expressing APPsw (Tg(Appsw) neurons) under the control of the prion protein promoter22. After 2 weeks in culture, we observed a 58% decrease in the levels of PtdIns(4,5)P2 of Tg(Appsw) neurons compared with controls (Supplementary Fig. 4 online). This biochemical deficiency was also seen in the N2a neuroblastoma cell line expressing the same mutant (Supplementary Fig. 4). Although these data are consistent with an effect of Aβ on phosphoinositide metabolism, the overexpression of APP itself, along with other potential ‘intrinsic’ properties of the cells expressing the transgene, may affect phosphoinositide metabolism independently of Aβ. To specifically address the role of cell-derived secreted Aβ on PtdIns(4,5)P2 metabolism, we used conditioned media from control and APPsw-expressing N2a cells.
Immunoprecipitation of Aβ from concentrated media, followed by western blot analysis of the immunoprecipitated peptide separated by SDS-PAGE, showed that APPsw-expressing N2a cells secrete Aβ and that a fraction of this peptide in the media is present in the form of trimers and tetramers (Supplementary Fig. 1). These oligomeric assemblies comigrate with the better-characterized oligomer species secreted by Chinese hamster ovary (CHO) cells coexpressing human APP and the ΔE9 FAD mutation of PSEN1 (ref. 23), as well as with trimers and tetramers derived from synthetic preparations (Supplementary Fig. 1). Conditioned media from APPsw-expressing N2a cells (diluted 1:4) induced a decrease in the levels of PtdIns(4,5)P2 after a 3-d treatment of control neuroblastoma cells, whereas the media derived from control cells had no effect (Fig. 2c). The deficiency of PtdIns(4,5)P2 was triggered by cell-derived secreted Aβ, as preincubation of conditioned media from the Swedish cells with 6E10 antibody abolished this effect (Fig. 2c), as was previously shown in the synthetic peptide experiment (Fig. 1b). Because the levels of cell-derived Aβ measured in the media by ELISA were below the nanomolar range (data not shown), we conclude that naturally secreted Aβ is particularly potent at destabilizing PtdIns(4,5)P2 and that it can achieve this effect at concentrations that theoretically fall in the pathophysiological range.
Having observed an Aβ oligomer–induced PtdIns(4,5)P2 deficiency, we wanted to elucidate the signaling mechanisms underlying this effect. On the basis of evidence suggesting that soluble oligomers of Aβ trigger Ca2+ dyshomeostasis7,21, we tested whether Aβ-induced PtdIns(4,5)P2 deficiency is associated with Ca2+ mobilization. Cortical neurons were treated with oAβ42 for 60 min in the presence or absence of the cell-impermeable Ca2+ chelator ethylene glycol tetraacetic acid (EGTA, 2 mM). Chelation of extracellular Ca2+ prevented the oAβ42-induced decrease in PtdIns(4,5)P2 levels, whereas EGTA had no effect on the levels of this lipid in the absence of the oligomers (Fig. 3a). The cell-permeable Ca2+ chelator 1,2-bis-(o-aminophenoxy)-ethane-N,N,N′,N′-tetraacetic acid, tetraacetoxymethyl ester (BAPTA-AM) also blocked the action of oAβ42, although sequestration of both extracellular and intracellular Ca2+ by this compound led to a twofold increase in the levels of PtdIns(4,5)P2 in both vehicle-treated and oAβ42-treated cortical neurons (data not shown). In parallel experiments, treatment of neurons with the Ca2+ ionophores ionomycin and A23187 (2 μM) caused PtdIns(4,5)P2 levels to decrease, although the latter effects were more notable compared with those of oAβ42 (Fig. 3a). Together, these experiments suggest that Aβ-induced PtdIns(4,5)P2 deficiency involves Ca2+ dyshomeostasis.
Next, we used a pharmacological approach to test whether the actions of oAβ42 on PtdIns(4,5)P2 are mediated through specific Ca2+-permeable glutamate receptor channels. The NMDA and AMPA receptors are likely candidates on the basis of previous studies showing that Aβ42 decreases the cell surface levels of these channels, thereby promoting synaptic changes resembling long-term depression5,6,21. Additionally, Aβ-induced neuronal damage can be rescued by an NMDA receptor antagonist21,24. Blockade of the AMPA receptor with the selective inhibitor 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) had no effect on PtdIns(4,5)P2 in the absence of oAβ42 and did not prevent oAβ42 from destabilizing this lipid when neurons were incubated with the peptide (Fig. 3b). However, blockade of NMDA with the selective inhibitor D(−)-2-amino-5-phosphonovaleric acid (AP5) caused a partial rescue of the PtdIns(4,5)P2 deficiency (Fig. 3b), suggesting that oAβ42 in part requires functional NMDA receptors to exert its effects on PtdIns(4,5)P2.
A large pool of PtdIns(4,5)P2 is concentrated at the plasma membrane, as determined by studies of the localization of genetically encoded probes for this lipid, such as the pleckstrin homology domain of phospholipase C δl (PHPLCδ1)25 and of PtdIns(4)P5 kinases, the main PtdIns(4,5)P2-synthesizing enzymes12. To test whether oAβ42 induces a deficiency of PtdIns(4,5)P2 at the plasma membrane, we transfected the pheochromocytoma cell line PC12 with a construct encoding a GFP fusion of the PHPLCδ1. After 16–24 h, we visualized the transfected PC12 cells by confocal microscopy and observed that the fluorescence of GFP-PHPLCδ1 appeared as a rim that borders the cells and is thus concentrated at the plasma membrane, as expected25 (Fig. 4a). In minutes, treatment of cells with oAβ42 triggered a partial loss of probe fluorescence from the plasma membrane and a corresponding increase of fluorescence in the cytoplasm, which appeared as a more diffuse signal. This effect was mimicked by a treatment with ionomycin, suggesting that it reflects hydrolysis of PtdIns(4,5)P2 via activation of the PLC pathway at the plasma membrane (Fig. 4b, and see below). No change in the probe localization was observed when cells were exposed to Aβ42Rev or to Aβ38 (Fig. 4b,c). Similar to the experiments that we carried out in primary cortical neurons (Fig. 2b), preincubation of oAβ42 with scyllo-inositol, but not chiro-inositol, was able to prevent the disappearance of the PtdIns(4,5)P2 probe from the plasma membrane (Fig. 4d).
The PLC pathway, which involves a family of lipid enzymes that hydrolyzes PtdIns(4,5)P2 to diacylglycerol (DAG) and inositol-1,4,5-trisphosphate (Ins(1,4,5)P3), is a major pathway activated by elevated intracellular calcium. Thus, Aβ-induced PtdIns(4,5)P2 deficiency may be caused by an activation of PLCs. To address this, we treated PC12 cells with oAβ42 for 120 min in the presence or absence of the PLC inhibitors U73122 or edelfosine (0.5 μM). The decrease in PtdIns(4,5)P2 levels that was mediated by oAβ42 treatment was abolished in the presence of these compounds (Fig. 4c), suggesting that oAβ42 reduces the levels of PtdIns(4,5)P2, at least in part, by promoting its hydrolysis through the PLC pathway. When incubated with the drugs alone, PtdIns(4,5)P2 fluorescence at the plasma membrane was not affected (Fig. 4c).Metabolites derived from this reaction may be detectable in cells exposed to the peptide, further indicating that oAβ42 promotes PtdIns(4,5)P2 hydrolysis through an activation of PLC. One of these metabolites, DAG, can be recognized by a genetically encoded probe, the C1 domain of protein kinase Cγ (C1-PKCγ), which we fused to GFP25. Treatment of PC12 cells with oAβ42 led to a partial and rapid redistribution of this probe from the cytoplasm to the cell surface, indicating that the peptide application resulted in increased DAG levels at the plasma membrane (Fig. 4e). Together with the PLC blocker experiments and the disappearance of PtdIns(4,5)P2 from the plasma membrane, these data suggest that oAβ42 activates the PLC pathway.
If PtdIns(4,5)P2 deficiency is an important aspect of oAβ42-induced synaptic dysfunction, reducing PtdIns(4,5)P2 catabolism may diminish or inhibit the action of this peptide. To test this hypothesis, we used mouse Synj1 mutants, as previous work has shown that neurons lacking synaptojanin 1 have higher levels of PtdIns(4,5)P226. Because homozygous mutants are not viable26, we used heterozygous animals (Synj1+/−), which show a 50% decrease in the levels of synaptojanin 1 (Fig. 5a) and a reduced PtdIns(4,5)P2 phosphatase activity in the brain (S.V.V. and G.D.P., unpublished data), to produce primary cortical neurons. Incubation of Synj1+/+ neurons with oAβ42 for 2 h caused the usual decrease in the levels of phosphoinositides (Fig. 5b). In contrast, neurons derived from Synj1+/− cortices were resistant to oAβ42 treatment, as we did not observe the decrease in PtdIns(4,5)P2 levels under the same conditions. The reduced sensitivity of Synj1+/− neurons to oAβ42 appeared to be specific for PtdIns(4,5)P2, as the levels of PtdIns4P in heterozygous neurons were found to be decreased to virtually the same extent as was seen in wild-type neurons during the acute exposure to the peptide (data not shown). These data suggest that the haploinsufficiency of Synj1 may offer a resistance to the neurotoxic actions of oAβ42, potentially by compensating for the PtdIns(4,5)P2 destabilizing action of the peptide.
If the PtdIns(4,5)P2-destabilizing effect of oAβ42 is an important aspect of oAβ42-induced synaptic dysfunction, the haploinsufficiency of Synj1 may also confer a resistance against the peptide with respect to neurophysiology. To test this, we carried out experiments addressing the effects of oAβ42 on synaptic plasticity in adult hippocampal slices from wild-type and Synj1 heterozygous mice26. As a premise to these experiments, oAβ42 was tested for its ability to affect phosphoinositide metabolism in hippocampal slices under the same conditions as were used for the electrophysiology experiments. We saw that oAβ42 decreased the levels of PtdIns(4,5)P2 by approximately 40% compared with vehicle treatment (Supplementary Fig. 5 online). Consistent with the data that we obtained fromneuronal cultures (Fig. 1a), we observed a subtle downregulation (≈20%) of PtdIns4P in oAβ42-treated slices, with no changes being detected in any of the other anionic phospholipids that we analyzed by high-performance liquid chromatography (Supplementary Fig. 5). Basal neurotransmission was normal in Synj1+/−slices in the absence of peptide (Supplementary Fig. 6 online). Similarly, we induced paired-pulse facilitation in the CA1 region of the hippocampus through stimulation of the Schaeffer collateral pathway and observed no differences between the two genotypes over the whole interstimulus interval range (Supplementary Fig. 6). Furthermore, long-term potentiation (LTP) in Synj1+/+ slices was comparable to that obtained in Synj1+/− slices in the presence of vehicle (Fig. 5c). Notably, although oAβ42 partially impaired LTP in control slices, as has been previously reported27–29, the effect of this crude oligomer preparation on LTP was strongly suppressed in Synj1+/− slices (Fig. 5c). These results indicate that the haploinsufficiency of Synj1 may protect neurons against the deleterious effects of soluble oligomers of Aβ on synaptic function. Together with behavioral data indicating that the performance of Synj1+/− mice was indistinguishable from that of wild type mice in tasks assessing associative memory (that is, fear conditioning; see Supplementary Fig. 7 online), as well as spatial learning and loco motor activity (that is, the Morris water maze; see Supplementary Fig. 7), our data indicate that partial blockade of synaptojanin 1 function may confer neurophysiological and neurobehavioral benefits on elevation of brain levels of Aβ, particularly the oligomeric forms.
Our data provide robust evidence that Aβ affects the metabolism of phosphoinositides in neurons. Of the variety of anionic phospholipids that we tested, PtdIns(4,5)P2 and, to a lesser extent, PtdIns4P, appeared to be downregulated by acute applications of Aβ42. However, prolonged applications of Aβ42 (tested up to a week) resulted in the selective deficiency of PtdIns(4,5)P2, which was observed solely when the peptide was delivered as a low molecular-weight oligomer preparation, rather than as monomer- or fibril-enriched preparations. PtdIns(4,5)P2 deficiency does not simply reflect increased cell death or poor viability, as the effect of Aβ42 was reversible and was not accompanied by global changes in DNA pyknosis and ATP levels. Pharmacological dissection of the Aβ signaling pathway indicated that the destabilizing effect of the peptide on PtdIns(4,5)P2 requires extracellular Ca2+, full functionality of the NMDA receptor and at least two PtdIns(4,5)P2-consuming pathways: via PLC, which hydrolyzes PtdIns(4,5)P2 into DAG and Ins(1,4,5)P3, and the inositol-5 phosphatase synaptojanin 1, which dephosphorylates this phosphoinositide on the 5′ position of the inositol ring.
Taking into consideration the multiple effects of Aβ reported to impact cell physiology raises the fundamental question of whether the biochemical changes triggered by the peptide are pathophysiologically meaningful or are simply part of compensatory responses with little relevance for the phenotypic manifestations of Alzheimer's disease.We reasoned that a powerful approach to test for the relevance of the PtdIns(4,5)P2 change induced by Aβ was to exploit a genetic model in which a major pathway involved in the downregulation of phosphoinositides, and PtdIns(4,5)P2 in particular, is defective. We used mice that are haploinsufficient for Synj1. The use of the null mutant was precluded by the previously described early postnatal lethality of these animals26 and the critical importance of synaptojanin 1 in synaptic vesicle recycling26,30,31 and receptor internalization32. However, mice containing approximately half the amount of synaptojanin 1 showed no obvious abnormalities and performed similarly to wild-typemice in tasks assessing various forms of learning and memory, consistent with the lack of electrophysiological phenotypes observed in the hippocampus of Synj1+/− mice. When incubated with Aβ oligomer concentrations leading to, and previously reported to induce, a disruption of LTP in control hippocampal slices, preparations fromSynj1+/− mice showed normal LTP. Notably, neurons derived from heterozygous mice did not undergo substantial changes in PtdIns(4,5)P2 levels in response to Aβ, suggesting that Synj1 haploinsufficiency protects against the neurotoxic effects of the peptide. Because the protective effect was observed for PtdIns(4,5)P2, but not for PtdIns4P (which is also affected by Aβ on acute treatments and may be a physiological substrate of the NH2-terminal phosphatase domain of synaptojanin 1, Sac1), a plausible interpretation is that Aβ-triggered hydrolysis of PtdIns(4,5)P2 may be part of the signaling cascade that ultimately leads to the suppression of hippocampal LTP (see below). Whether changes in other phosphoinositides that have been previously implicated in neurophysiology and are potentially affected by Aβ (for example, PtdIns(3,4,5)P3) participate in the described effects remains unclear, however. Similarly, a dysregulation of the PtdIns(4,5)P2 metabolites Ins(1,4,5)P3 and DAG may also account for some of the observed effects.
A fundamental question that is raised by our study is how the haploinsufficiency of Synj1 exerts its protective effect on oAβ42-mediated downregulation of PtdIns(4,5)P2 and impairment of synaptic function. Under normal conditions, Synj1 haploinsufficiency does not appear to affect neurophysiology and learning behavior per se. However, under conditions where PtdIns(4,5)P2 is excessively catabolized through the PLC pathway (as in the presence of Aβ), reducing the function of synaptojanin 1, and thus the further catabolism of PtdIns(4,5)P2, may facilitate the regeneration of this lipid. The resulting normalization of PtdIns(4,5)P2 levels in the presence of Aβ may prevent some of the molecular and cellular changes that occur in response to the neurotoxic peptide at synapses, such as the internalization of glutamate receptors and the remodeling of actin6,27,29,33–36. It remains to be determined whether Aβ treatment promotes the activation of synaptojanin 1 (alongside PLC) as part of the neurotoxic signaling pathway that impairs synaptic function (for example, suppression of LTP and induction of long-term depression)6,27,29,33–35. Stimulation of synaptojanin 1 may occur via its dephosphorylation by calcineurin37–39, a major protein phosphatase that is regulated by Ca2+/calmodulin, which has been previously shown to be activated by Aβ5,21,40. Regardless of how Synj1 haploinsufficiency mediates its protective effect against Aβ, our study suggests that treatments using inhibitors of synaptojanin 1 and, potentially, specific PLC isozymes to prevent Aβ-induced PtdIns(4,5)P2 breakdown may improve synaptic function in individuals with Alzheimer's disease and offers a previously unknown avenue for therapeutic approaches to Alzheimer's disease.
Finally, our study may have important implications for trisomy 21, also known as Down syndrome. Indeed, individuals with Down syndrome develop the pathology of Alzheimer's disease after the third decade of their lives, at least in part because they overexpress the chromosome 21 gene APP41. Because Synj1 is also present on this chromosome, its overexpression in individuals with Down syndrome may contribute to their early Alzheimer's disease pathology by rendering neurons more sensitive to Aβ insults, in contrast to the protective effect conferred by its haploinsufficiency. Altogether, neurons in the brains of individuals with Down syndromemay undergo a dual ‘hit’ on PtdIns(4,5)P2, accounted for by Synj1 trisomy and Aβ elevation, which could accelerate the onset of cognitive decline in middle-aged adults with Down syndrome.
Synj1 wild-type and heterozygous mice (B6C3Sn background) of 3–4 months of age were used for behavior and electrophysiology. All experiments were in accordance with protocols approved by the Institutional Animal Care and Use Committee of Columbia University.
PC12 cells were maintained as previously described42. Transfections of PC12 cells with constructs encoding GFP fused with PHPLCδ1 or C1-PKCγ25 were carried out with Lipofectamine 2000 (Invitrogen). Primary cortical neurons were generated from newborn wild-type mice as previously described43. Some experiments were carried out using transgenic mice overexpressing the Swedish mutant of APP (Tg2576)22 and Synj1+/− mice26, along with their wild-type littermates. Treatments with Aβ42 were carried out after 15 days in vitro, and incubation time with oAβ42 was 60 min unless otherwise specified. Drugs were added to the cultures 30 min before the addition of Aβ. We preincubated 6E10 antibody (Signet Laboratories/Covance, 1:200 final dilution) with the oAβ42 solution for 2 h at 20–23 °C before adding it to the cultures in the neutralization assay.
To solubilize Aβ peptide, we allowed synthetic Aβ (1–42) peptide (American Peptide) to equilibrate at 20–23 °C for 30 min before it was resuspended and diluted to 1 mM in 1,1,1,3,3,3-hexafluoro-2-propanol. After evaporation for 2 h, peptide films were dried in a Speed Vac and stored at −20 °C. Peptide films were resuspended to 1 mM in dimethyl sulfoxide (DMSO) using bath sonication for 10 min. The solution, which was stored at −20 °C, was used within 2 weeks of dilution in DMSO. To form the oligomers, we diluted the 1 mM DMSO solution to 100 μM in cold phosphate-buffered saline (PBS), vortexed it for 30 s and incubated it overnight at 4 °C. Before use, the Aβ-PBS solution was further diluted in culture media. To form the monomers, we immediately diluted the 1 mM DMSO solution in culture media to the final concentration following bath sonication. To form the fibrils, we diluted the 1 mMDMSO solution to 100 μM in 10 mMHCl, vortexed it for 30 s and incubated it overnight at 37 °C. We diluted the solution in culture media and vortexed briefly before use44.
N2a cells and N2a cells stably overexpressing APP with the Swedish mutation (APPsw) were cultured as previously described45. CHO cells stably expressing human APP (7WD4, a kind gift from D. Selkoe, Harvard University) were transfected with a construct encoding Psen1 FAD mutant ΔE9 to generate the stable line, CHO-7WD4 Psen1ΔE9, that was used in this study. Conditioned media was collected after 4 d and centrifuged at 130g to remove cellular debris. Cleared media was concentrated about 40-fold with Amicon Ultra-3K nominal molecular weight limit device and subjected to immunoprecipitation as previously described27,34 with the 6E10 antibody. Immunoprecipitate was eluted in 2× NuPAGE sample buffer and 2× NuPAGE sample reducing agent (Invitrogen).
SDS-PAGE was carried out according to the manufacturer's protocols (Invitrogen).For the analysis of Aβ, samples of the final dilution in culture media were prepared with NuPage LDS sample buffer and loaded onto 4–12% Bis-Tris NuPage gels; separation and electroblotting were carried out using MES running and transfer buffers (Invitrogen). Samples were transferred to PVDF membranes (Bio-Rad). Membranes were boiled for 5 min in PBS before blocking in a solution of 50% commercial blocking buffer (Li-Cor) and 50% Tris-buffered saline. Membranes were incubated with the primary antibody 6E10 and visualized using an infrared imaging system (Odyssey Infrared Imaging System, Li-Cor). The immunoblotting of brain extracts from Synj1 mutant mice and of neuronal cultures was carried out as previously described26 using rabbit polyclonal antibodies to the COOH terminus of synaptojanin 1 (OCK) and the COOH terminus of PIPK1γ (Zola) (kind gifts from P. De Camilli, Yale University) and a mouse monoclonal antibody to α-tubulin (B-5-1-2, Sigma).
To analyze GFP-PHPLCδ1 domain dissociation from the plasma membrane, we incubated PC12 cells with 200 nM Aβ42 oligomers, 2 μM ionomycin (Sigma-Aldrich), 200 nM of Aβ42Rev (inverted peptide) or 200 nM of Aβ38 24 h after transfection with GFP-PHPLCδ1 domain. Cells were then washed in phosphate buffer and fixed with 4%paraformaldehyde (wt/vol). Confocal z-stack images (0.5 μm) of PC12 were obtained using a Nikon EZC1.2.30 confocal microscope with a 100× oil-immersion objective. We calculated GFP intensity using the ImageJ software (US National Institutes of Health). For each cell in a given image, a line intensity profile across the cell was obtained. The relative decrease in plasma membrane localization was calculated as the ratio between the plasma membrane fluorescence intensity and the average cytosolic fluorescence intensity. To analyze GFP–C1-PKCγ translocation to the plasma membrane, we incubated PC12 cells with oAβ42 for 10 min 24 h after transfection with GFP-C1PKCγ and fixed them. After 30 min in blocking solution (5% donkey serum (vol/vol) from SIGMA in PBS) cells were stained for 30 min with Wheat Germ Agglutinin–Alexa 594 (WGA, 5 μg ml−1). As for GFP-PHPLCδ1, the relative increase in plasma membrane localization was calculated as the ratio between the plasma membrane fluorescence intensity and the average cytosolic fluorescence intensity. WGA–Alexa 594 was used as a plasma membrane marker to select the coordinates for determining the external peaks for the quantitative analysis.
Transverse hippocampal slices (400 μm) were cut with a tissue chopper (EMS) and maintained in an interface chamber at 29 °C for 90 min before recording, as previously reported46. CA1 field-excitatory postsynaptic potentials were recorded by placing both the stimulating and the recording electrodes in CA1 stratum radiatum. LTP was induced using a θ-burst stimulation (four pulses at 100 Hz, with bursts repeated at 5 Hz and each tetanus including three 10-burst trains separated by 15 s). oAβ42 was applied for 20 min before the θ burst.
Statistical analysis of the biochemical experiments was carried out using ANOVA and a paired or unpaired Student's t-test (Prism Software). The electrophysiological data were analyzed by two-way ANOVA. The behavioral data were analyzed using two-way ANOVA with repeated measures, one-way ANOVA and a Bonferroni test for post hoc comparisons. The significance level was established at P < 0.05.
We would like to thank P. De Camilli and O. Cremona for providing the Synj1 mutant mice, K. Hsiao-Ashe for providing the App mutant mice, D. Selkoe for the gift of CHO APP(7WD4) cells, E. Micevska for technical help with the animals and the genotyping, P. Scheiffele, P. De Camilli, M. Wenk, A. Bhalla and B. Chang for critical reading of the manuscript and G. Thinakaran (University of Chicago) for providing the stable N2a cell line. This work was supported by grants from the US National Institute of Neurological Diseases and Stroke (NS043467 to T.-W.K., NS056049 to G.D.P. and NS049442 to O.A.), the US National Institute of Child Health and Human Development (HD047733 to G.D.P), the US National Center for Complementary and Alternative Medicine (AT002643 to T.-W.K.), SMART Biosciences (G.D.P.) and the Cure Alzheimer's Fund (T.-W.K.).
COMPETING INTERESTS STATEMENT
The authors declare competing financial interests: details accompany the full-text HTML version of the paper at http://www.nature.com/natureneuroscience/.
Note: Supplementary information is available on the Nature Neuroscience website.