Accumulation of amyloid β-peptides (Aβ) in the brain has been linked with memory loss in Alzheimer's disease and its animal models. However, the synaptic mechanism by which Aβ causes memory deficits remains unclear. We previously showed that acute application of Aβ inhibited long-term potentiation (LTP) in the hippocampal perforant path via activation of calcineurin, a Ca2+-dependent protein phosphatase. This study examined whether Aβ could also inhibit Ca2+/calmodulin dependent protein kinase II (CaMKII), further disrupting the dynamic balance between protein kinase and phosphatase during synaptic plasticity. Immunoblot analysis was conducted to measure autophosphorylation of CaMKII at Thr286 and phosphorylation of the GluR1 subunit of AMPA receptors in single rat hippocampal slices. A high-frequency tetanus applied to the perforant path significantly increased CaMKII autophosphorylation and subsequent phosphorylation of GluR1 at Ser831, a CaMKII-dependent site, in the dentate area. Acute application of Aβ1–42 inhibited dentate LTP and associated phosphorylation processes, but was without effect on phosphorylation of GluR1 at Ser845, a protein kinase A-dependent site. These results suggest that activity-dependent CaMKII autophosphorylation and AMPA receptor phosphorylation are essential for dentate LTP. Disruption of such mechanisms could directly contribute to Aβ-induced deficits in hippocampal synaptic plasticity and memory.
Alzheimer's disease (AD) is a neurodegenerative disease characterized by progressive memory loss. Amyloid β peptides (Aβ), the major constituents of senile plaques in AD brain, have long been implicated in the pathogenesis of the disease. The exact cellular mechanisms for Aβ action in AD, however, remain to be clarified. Earlier studies supported the hypothesis that amyloid fibrils, generated by self-aggregation of Aβ, drive the neurodegeneration cascade in AD (Hardy and Higgins 1992; Hardy and Selkoe 2002). More recent evidence indicates that soluble forms of Aβ, such as small oligomers and protofibrils, also have potent toxic effect on neurons (Klein et al. 2001; Walsh et al. 2002). This may explain why AD-like neurological deficits can occur in the absence of amyloid deposits in some strains of transgenic mice overexpressing human amyloid precursor protein (APP) (Van Dam et al. 2003; Yamaguchi et al. 1991). In other APP transgenic mice, impaired hippocampal long-term potentiation (LTP) and behavioral deficits are correlated with elevated brain Aβ level and plaques, but few have shown neuronal cell death or synaptic loss (Chapman et al. 1999; Giacchino et al. 2000; Larson et al. 1999). These findings suggest that early accumulation of Aβ, likely in its soluble forms, causes synaptic dysfunction that initiates cognitive decline prior to synapse loss and cell death. A key question to be answered is how Aβ at this stage affects synaptic plasticity in neuronal networks where memories are formed and stored. Understanding such mechanisms may provide important clues for therapeutic intervention of AD at its early stages.
We and others have shown inhibition of LTP in area CA1 or dentate gyrus of rat hippocampus by acute application of synthetic Aβ (Chen et al. 2000, 2002; Kim et al. 2001; Wang et al. 2002) or conditioned culture medium containing Aβ species secreted by cells transfected with human APP (Walsh et al. 2002). Subsequent studies have focused on elucidating the pivotal signaling mechanisms underlying Aβ disruption of synaptic plasticity (Mattson and Chan 2003; Rowan et al. 2003). For example, we showed in the dentate gyrus that Aβ inhibition of LTP was blocked by specific inhibitors for calcineurin, a Ca2+- and calmodulin-dependent protein phosphatase (Chen et al. 2002). This result indicates that increased calcineurin activity contributes to Aβ-induced LTP deficits. It also gives rise to a new question as to whether Aβ can also alter protein kinase activity thus further disrupting the dynamic balance between protein phosphorylation and dephosphorylation during LTP. This study examined, in the hippocampal dentate gyrus, the effect of Aβ1–42 on Ca2+ and calmodulin-dependent protein kinase II (CaMKII). Activation of this enzyme is essential for LTP generation (Malinow et al. 1989; Silva et al. 1992) and is under the negative control of a calcineurin-dependent phosphatase pathway (Blitzer et al. 1995; Lisman 1994). CaMKII-mediated phosphorylation of α-amino-3-hydroxy-5-methyl-4-isoxazloe proprionic acid (AMPA) type of glutamate receptors, a crucial step in LTP expression (Lee et al. 2000), was also measured following LTP and Aβ treatments. Our results show that Aβ1–42 attenuates LTP-induced activation of CaMKII and subsequent phosphorylation of AMPA receptors in the dentate gyrus.
Hippocampal slices were prepared from 30- to 60-day-old male Sprague-Dawley rats as described (Chen et al. 2000). Briefly, rats were decapitated under halothane anesthesia. The brain was removed and sectioned into 500-μm-thick slices using a vibroslicer. The slices were maintained at room temperature in a holding chamber containing oxygenated artificial cerebrospinal fluid (ACSF; in mM: 120 NaCl, 25 NaHCO3, 3.3 KCl, 1.23 NaH2PO4, 2 CaCl2, 1.0 MgSO4, and 10 d-glucose, at pH 7.4). After at least 1 h equilibration, slices were transferred into a submerged recording chamber and continuously perfused with 29–30°C oxygenated ACSF at 2–3 ml/min throughout the experiments. All protocols were in accordance with PHS Guidelines and were reviewed and approved by the Chancellor’s Animal Research Committee of the University of California at Los Angeles.
Field excitatory postsynaptic potentials (fEPSPs) were recorded in dentate gyrus using standard extracellular recording techniques as previously described (Chen et al. 2000). Briefly, fEPSPs were evoked by stimulating the medial perforant path in the middle third of the dentate molecular layer with 0.1-ms pulses delivered via a sharpened monopolar tungsten electrode and recorded from the same layer with a glass microelectrode filled with 2 M NaCl. Baseline responses were collected using 0.008-Hz test pulses that yielded 40–50% of the maximal fEPSP slope in ACSF. LTP was induced by a single high-frequency stimulation (HFS) of 100 Hz for 1 s at the same stimulus intensity used for baseline. After HFS, the stimulation was returned to the baseline frequency, and slices were harvested at different time-points as described. The amount of LTP was expressed as percent changes of fEPSP slopes from baseline after HFS.
Aβ application and experimental design
Aβ1–42 (Bachem, Torrance, CA) was prepared as 1 mM stock solution in H2O, stored in small aliquots at −20°C, and diluted with ACSF to 200 nM immediately before use. Samples of applied peptide solutions were visually inspected under a 400× phase-contrast microscope. No evident fibrillar or sheetlike aggregates were observed at this concentration. The Aβ solution was bath-applied to slices during the field potential recordings, starting 20 min before and ending immediately after the HFS. To measure Aβ-induced changes in protein phosphorylation, four slices from the same animal were used for different treatments: two receiving HFS with or without Aβ1–42 treatment (HFS, Aβ+HFS) and the other two as paired controls receiving matching treatment except no HFS (control, Aβ). The order in which these treatments were given to a set of slices was randomly decided at the beginning and rotated daily across animals. The dentate region of slices was rapidly dissected out after treatment. The mini-slices were frozen on dry ice and kept at −70°C until further processing for immunoblotting.
CaMKII or GluR1 phosphorylation was measured in the dentate mini-slices using immunoblot analysis as described previously with some modifications (Makhinson et al. 1999). Individual dentate slices were homogenized in 70 μl of ice-cold buffer containing 50 mM HEPES, 10 mM MgCl2, 1 mM EDTA, 1 mM EGTA, 10 mM benzamide, 100 ng/ml leupeptin, 100 ng/ml aprotinin, 0.01% TritonX-100, 0.08 mM sodium molybdate, and 2 mM sodium pyrophosphate at pH 7.4. Aliquots of the homogenate were taken to determine protein concentration using Protein Assay Reagent (Bio-Rad, Hercules, CA). A cold denaturing loading buffer was added immediately into the remaining homogenate to stop protein kinase and phosphatase activity. Samples were boiled for 5 min, and aliquots of 30 μg protein were electrophoresed on a SDS/PAGE gel containing 4% stacking and 12% resolving acrylamide. The proteins were transferred onto nitrocellulose membranes, blocked for 1 h with phosphate-buffered saline containing 4% nonfat dried milk (Blotto), and probed overnight at room temperature with primary antibodies. A monoclonal antibody against the α subunit of CaMKII phosphorylated at Thr286 (ABR, Golden, CO) was used at 1:1,000 to detect autophosphorylation of the kinase. Two polyclonal antibodies (Upstate, Lake Placid, NY) were used at 1:500 to recognize GluR1 phosphorylated at Ser831 or Ser845, respectively. Antibodies for total CaMKII (Chemicon International, Temecula, CA) and total GluR1 (Upstate) were used at 1:2,500. All primary antibodies were diluted in Blotto. Nitrocellulose membranes were further incubated for 1 h with horseradish peroxidase-conjugated secondary antibodies at 1:10,000. The proteins were visualized with enhanced chemiluminescence (Amersham ECL Western Blotting Analysis System). Densitometric data were obtained by processing the exposed films with a Molecular Dynamics Personal Densitometer SI (Sunnyvale, CA) using ImageQuant Software. For the loading control, membranes were stripped and reprobed with an antibody for β-actin (1:2,000; Sigma, St. Louis, MO). Values of phosphorylated and total proteins for either CaMKII or GluR1, measured with aliquots from the same slice sample, were normalized to β-actin.
Data are presented as group means ± SE. Percent changes in the protein bands due to Aβ or HFS treatments were calculated relative to the control band (ACSF alone) run in the same experiment using slices from the same animal. One-way ANOVA was applied to test for overall statistical significance across multiple group means, followed by Bonferroni post hoc test for pairwise mean comparisons. Student's t-test were used in LTP experiments for two-group comparisons. Statistical significance was defined as P < 0.05.
LTP-inducing stimulation increases αCaMKII autophosphorylation in dentate gyrus
A single HFS to the media perforant path induced significant LTP in the dentate gyrus of hippocampal slices, increasing fEPSP slopes to 132 ± 3% of the baseline level at 30 min after HFS (n = 17, P < 0.01). To determine LTP-associated changes in CaMKII autophosphorylation, slices were harvested at different time-points after HFS and measured for their phospho-CaMKII and total CaMKII levels. Along with the potentiation of fEPSP (Fig. 1D), the phospho-CaMKII level increased rapidly, reaching 131 ± 15, 158 ± 13, and 155 ± 14% of the control level at 5, 15, and 30 min after the HFS, respectively (n = 8 for each time-point, P = 0.01 at 15 and 30 min compared with unstimulated controls; Fig. 1, A and B). In contrast, the total CaMKII protein was not significantly altered by HFS during the same period of time (Fig. 1, A and C).
Aβ1–42 inhibits dentate LTP and HFS-induced CaMKII autophosphorylation
As reported previously (Chen et al. 2000), acute application of Aβ1–42 during HFS reduced LTP in the dentate gyrus (Fig. 2D). At 30 min after HFS, the fEPSP slope in Aβ-treated slices decayed to 100 ± 3% of baseline (n = 16), significantly smaller than that in the control slices (132 ± 3%, n = 17, P < 0.01). In parallel, immunoblot analysis revealed significant changes in dentate phospho-CaMKII levels following experimental treatments [F(3, 60) = 5.2, P < 0.01, one-way ANOVA]. Although Aβ treatment alone caused no evident change in the basal phosphorylation level of CaMKII (98 ± 9% of the controls, n = 16, P > 0.05), the peptide clearly reduced LTP-induced CaMKII phosphorylation when applied with HFS (Fig. 2, A and B). The phospho-CaMKII level in the Aβ + HFS group at 15 min after HFS was 107 ± 8% (n = 16) of the controls, significantly lower than the level in the HFS alone group (135 ± 10%, n = 16, P < 0.05) and indistinguishable from that in the Aβ alone group (P > 0.05). The total CaMKII protein level did not differ among all groups (Fig. 2C).
Aβ1–42 blocks HFS-induced GluR1 phosphorylation at a CaMKII-dependent site
Studies show that LTP is associated with an increase in CaMKII-mediated phosphorylation of GluR1 subunits of AMPA receptors (Barria et al. 1997b). We therefore further determined whether Aβ inhibition of CaMKII led to reduction in GluR1 phosphorylation. Using two phosphorylation site-specific antibodies, we distinguished GluR1 phosphorylation at a CaMKII site (Ser831) or a protein kinase A (PKA) site (Ser845). As shown in Fig. 3, HFS induced significant increases in Ser831 phosphorylation in the absence of Aβ (142 ± 12% of the control level at 15 min after HFS, n = 22, P < 0.01). In Aβ-treated slices, however, the HFS failed to increase Ser831 phosphorylation. At 15 min after HFS, the phospho-Ser831 level in the Aβ + HFS group did not differ from the Aβ alone group (108 ± 9 vs. 99 ± 7%, n = 22 for each group, P > 0.05) and was significantly lowered than the HFS only group mentioned above (P < 0.05). Furthermore, the effect of HFS and Aβ1–42 on GluR1 phosphorylation appeared to be specific for the CaMKII site, because phosphorylation at Ser845 was not significantly affected by Aβ (108 ± 18%, n = 13) or HFS treatment (115 ± 15%, n = 13; Fig. 4 ).
This study showed activity-dependent increases in CaMKII autophosphorylation and subsequent phosphorylation of AMPA receptor GluR1 subunits at a CaMKII-dependent site in the dentate gyrus of rat hippocampal slices. Both phosphorylation processes along with dentate early-phase LTP were blocked by acute application of Aβ1–42. These results suggest that inhibition of CaMKII-dependent protein phosphorylation may play a major role in Aβ-induced LTP and memory deficits. It is well established that generation of N-methyl-d-aspartate (NMDA) receptor-dependent forms of LTP in the hippocampus requires an initial postsynaptic Ca2+ influx and subsequent conversion of this transient signal into long-lasting enhancement in synaptic strength (Malenka and Nicoll 1999). CaMKII, a kinase highly concentrated at postsynaptic density and activated by Ca2+/calmodulin (Bennett et al. 1983; Kelly et al. 1984), is thought to have a key role in this conversion. In the CA1 region of hippocampal slices, NMDA receptor–mediated Ca2+ transients lead to autophosphorylation of CaMKII and its accumulation in the dendrites of pyramidal neurons after LTP induction (Fukunaga et al. 1995; Ouyang et al. 1997). Autophosphorylation at Thr286 enables Ca2+/calmodulin-independent, persistent CaMKII activity (Miller and Kennedy 1986; Soderling et al. 2001), which has been proposed to serve as a molecular switch to convert transient NMDA receptor activation into long-lasting biochemical changes underlying synaptic plasticity (Lisman et al. 2002). In support of this view, genetic or pharmacological manipulations that prevent CaMKII autophosphorylation block LTP induction in the Schaffer collateral-CA1 pathway (Giese et al. 1998; Malinow et al. 1989; Silva et al. 1992). It was unclear, however, whether synaptic plasticity induced in other hippocampal regions uses similar, CaMKII-dependent mechanisms. Here we show for the first time that CaMKII autophosphorylation may be necessary for LTP induction or expression in the perforant path–dentate gyrus pathway. The LTP-inducing tetanus triggered significant increases in CaMKII autophosphorylation in dentate area. The increase became evident within 5 min, peaked around 15 min, and lasted for ≥30 min after HFS, resembling the time course of CaMKII autophosphorylation following LTP in the CA1 region (Banke et al. 2000; Barria et al. 1997b; Giovannini et al. 2001). This rapid change in the phospho-CaMKII level was not accompanied by a significant increase in the level of total CaMKII protein. Thus the single HFS protocol we used to induce early LTP appeared to promote phosphorylation of preexisting CaMKII proteins rather than to induce synthesis of new kinases. There is in vivo evidence that levels of synaptically associated CaMKII mRNA and protein can be increased following LTP in the dentate gyrus (Davis et al. 2000; Havik et al. 2003; Roberts et al. 1998). However, these increases are restricted in the dendritic spines without parallel upregulation in the soma, indicating translocation of preexisting CaMKII mRNA to synapses rather than an overall increase of CaMKII expression in the whole tissue. Furthermore, dendritic CaMKII expression occurs at a much slower rate than protein phosphorylation, often taking 2–3 h to develop after LTP induction (Davis et al. 2000; Roberts et al. 1998) and therefore being more relevant to the late phase rather than the early phase of LTP (Miller et al. 2002). Based on the above, it is not surprising that we did not observe significant increases in the total CaMKII level when measuring the whole tissue homogenates at 15–30 min after induction of early LTP.
Application of Aβ1–42 attenuated both LTP and associated CaMKII phosphorylation in the dentate gyrus. These results indicate that preventing CaMKII autophosphorylation/activation could lead to impairment of long-term synaptic plasticity in this hippocampal subregion. Several distinct mechanisms may have contributed to Aβ inhibition of CaMKII activity (Fig. 5). First, at the concentrations effective for LTP blockade, Aβ1–42 inhibits NMDA receptor–mediated synaptic currents (Chen et al. 2002). This effect can reduce Ca2+ influx through the NMDA receptor and thus prevent activation of CaMKII. Second, CaMKII activity is regulated by a balance between PKA and calcineurin (phosphatase 2B). These two enzymes antagonistically regulate the activity of protein phosphatase 1 (PP1), which in turn serves as a gate controlling the phosphorylation state of CaMKII and hence synaptic strength (Blitzer et al. 1998; Lisman 1994). Blocking calcineurin activity reverses Aβ-induced deficits in the late-phase LTP in dentate gyrus, suggesting activation of calcineurin by Aβ (Chen et al. 2002). It is thus likely that enhanced calcineurin activity creates an imbalance between calcineurin and PKA activity, leading to activation of PP1 and thereby dephosphorylation of CaMKII.
How changes in CaMKII activity affect long-lasting synaptic potentiation in the hippocampus has been the subject of intense investigation. A major postsynaptic target of CaMKII activity is the AMPA type of glutamate receptors. AMPA receptors are heterooligomeric complexes assembled from GluR1-4 subunits. The most common subunit combinations of AMPA receptors in hippocampal neurons are GluR1/2 and GluR2/3 (Craig et al. 1993). LTP in CA1 neurons is associated with phosphorylation of GluR1 subunit at Ser831 by CaMKII (Barria et al. 1997a; Lee et al. 2000). This phosphorylation increases the single-channel conductance of AMPA receptors (Benke et al. 1998; Derkach et al. 1999), which can directly contribute to LTP expression at CA1 glutamatergic synapses. This study showed an increase in Ser831 phosphorylation following dentate LTP and attenuation of such changes by Aβ1–42. The peptide did not affect the basal phosphorylation level of Ser831, consistent with the earlier finding that Aβ1–42 was without effect on AMPA receptor-mediated baseline synaptic currents in dentate granule cells (Chen et al. 2002). LTP-induced Ser831 phosphorylation, however, was clearly blocked by Aβ, in parallel with the reduction of HFS-triggered CaMKII autophosphorylation. Studies show that the efficacy of CaMKII-mediated substrate phosphorylation is regulated by its autophosphorylation at Thr286. The autophosphorylated kinase translocates to the postsynaptic density (PSD) (Strack et al. 1997), where it binds directly to the NMDA receptor (Leonard et al. 1999; Strack and Colbran 1998). This process places the kinase at an ideal site to control synaptic function and provides a positive feedback for its further autophosphorylation (Lisman et al. 2002). Thus inhibition of the initial autophosphorylation of CaMKII could impede its translocation and synaptic targeting, leading to reduced AMPA receptor phosphorylation and impaired synaptic plasticity.
The effect of Aβ and LTP on GluR1 phosphorylation appeared to be specific for the CaMKII-dependent site, because another phosphorylation site on the GluR1 subunit, Ser845, was not altered by either HFS stimulation or Aβ1–42 application. Phosphorylation of Ser845 by PKA is known to increase the peak open probability of AMPA receptor channels (Banke et al. 2000). This mechanism is important for maintaining the strength of basal synaptic transmission, because dephosphorylation of Ser845 leads to long-term depression (LTD), a depressive form of synaptic plasticity (Lee et al. 2000). Calcineurin, a Ca2+-dependent serine/threonine phosphatase, is required for LTD induction (Mulkey et al. 1994) and thus likely mediates Ser845 dephosphorylation. Interestingly, acute application of Aβ1–42 facilitates hippocampal LTD (Kim et al. 2001). Whether under LTD conditions Aβ could facilitate Ser845 dephosphorylation via calcineurin is an interesting issue to be further investigated.
This work is supported by National Institutes of Health Grants P50DA-05010 and R01AG-17542 to C.-W. Xie and by State of California, Department of Health Service/Alzheimer's Disease Program (Agreement 01–15944) to J. B. Watson.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 by the American Physiological Society