Oligomers of Amyloid β Prevent Physiological Activation of the Cellular Prion Protein-Metabotropic Glutamate Receptor 5 Complex by Glutamate in Alzheimer Disease*

The dysfunction and loss of synapses in Alzheimer disease are central to dementia symptoms. We have recently demonstrated that pathological Amyloid β oligomer (Aβo) regulates the association between intracellular protein mediators and the synaptic receptor complex composed of cellular prion protein (PrPC) and metabotropic glutamate receptor 5 (mGluR5). Here we sought to determine whether Aβo alters the physiological signaling of the PrPC-mGluR5 complex upon glutamate activation. We provide evidence that acute exposure to Aβo as well as chronic expression of familial Alzheimer disease mutant transgenes in model mice prevents protein-protein interaction changes of the complex induced by the glutamate analog 3,5-dihydroxyphenylglycine. We further show that 3,5-dihydroxyphenylglycine triggers the phosphorylation and activation of protein-tyrosine kinase 2-β (PTK2B, also referred to as Pyk2) and of calcium/calmodulin-dependent protein kinase II in wild-type brain slices but not in Alzheimer disease transgenic brain slices or wild-type slices incubated with Aβo. This study further distinguishes two separate Aβo-dependent signaling cascades, one dependent on extracellular Ca2+ and Fyn kinase activation and the other dependent on the release of Ca2+ from intracellular stores. Thus, Aβo triggers multiple distinct PrPC-mGluR5-dependent events implicated in neurodegeneration and dementia. We propose that targeting the PrPC-mGluR5 complex will reverse aberrant Aβo-triggered states of the complex to allow physiological fluctuations of glutamate signaling.

Alzheimer disease is the most common form of age-related dementia. The Amyloid hypothesis of Alzheimer disease postulates proteolytic processing of the amyloid precursor protein (APP) 2 to be disease-relevant (1). This process releases hydro-phobic Amyloid ␤ (A␤) peptides that aggregate in the form of characteristic Amyloid plaques in the brain of patients (1). Aggregation of A␤ is a multistep process, with several intermediate forms of A␤ being generated. The A␤ species that correlates best with the severity of dementia is an oligomeric species of A␤, referred to as A␤o (2)(3)(4)(5). Several independent lines of research have revealed an A␤o-toxic function at synapses (6 -10). Thus, a better understanding of intracellular signaling induced by extracellular A␤o could reveal novel ways to prevent neurotoxicity.
Our previous work described cellular prion protein (PrP C ) as a high-affinity cell surface receptor for A␤o (11), capable of mediating deficits in plasticity, synapse density, and memory (12)(13)(14)(15). A␤o binding to PrP C triggers intracellular signaling via metabotropic glutamate receptor 5 (mGluR5) (12,16,17), which includes downstream Fyn activation and synapse loss (12,13,18). We recently demonstrated that PrP C is linked to intracellular proteins via mGluR5 and that this coupling is modulated by extracellular A␤o (16). Critically, our work further revealed that the interaction between PrP C and intracellular protein mediators is perturbed in Alzheimer disease (16). Here we focus on comparing how the physiological glutamate analog 3,5-dihydroxyphenylglycine (DHPG) affects the PrP C -mGluR5 complex in the absence and presence of A␤o. We investigate a multiprotein complex consisting of at least PrP C , mGluR5, Homer1b/c, protein-tyrosine kinase 2-␤ (PTK2B, Pyk2), and calcium/calmodulin-dependent protein kinase II (CamKII). Interestingly, Homer proteins as well as Pyk2 have been associated with susceptibility to Alzheimer disease by imaging quantitative trait loci studies and genome-wide association studies (19 -21). These studies provide a genetic link between the PrP C -mGluR5 multiprotein complex and Alzheimer disease. Furthermore, Homer proteins are known to be important postsynaptic protein mediators that link mGluR5 to protein kinases and are able to alter mGluR5-induced signaling independent of glutamate-induced activation (22)(23)(24)(25).
Importantly, A␤o has been shown to cluster at excitatory synapses that stain positively for Homer1b/c as well as for Cam-KII (26,27). This clustering of pathological A␤o is mGluR5and PrP C -dependent (26). Renner et al. (26) further observed that A␤o alters the synaptic trafficking of the otherwise laterally mobile mGluR5 within primary neuronal membranes. These data suggest that pathological A␤o bound to PrP C scaffolds mGluR5 into a pathological conformation within the plasma membrane, which might disrupt physiological glutamate signaling.
The aim of this study is to understand whether pathological A␤o alters physiological glutamate-induced stimulation of mGluR5 in neurons. Our data reveal that isolated A␤o as well as APP/PS1 transgene-dependent species prohibit DHPG-induced association changes in the PrP C -mGluR5 multiprotein complex. Furthermore, DHPG is unable to enhance phosphorylation of Pyk2 and CamKII in the presence of A␤o. We conclude that pathological A␤o traps the PrP C -mGluR5 complex in a pathological state that does not allow glutamate-induced regulation of the complex. Thus, the PrP C -mGluR5 complex is a potential target for disease-modifying therapy for Alzheimer disease. Our work advances the understanding of the regulation of the PrP C -mGluR5 complex by physiological versus pathological ligands. This knowledge is important for the development of therapeutics targeting the complex in the context of Alzheimer disease to counteract the A␤o-dependent disruption of physiological glutamate signaling at this metabotropic receptor.

Results
The PrP C -mGluR5 complex isolated by anti-PrP C immunoprecipitation associates with the intracellular protein mediators Homer1b/c, Pyk2, and CamKII in crude synaptoneurosome preparations from acute brain slices in a PrP C -dependent manner ( Fig. 1A and Ref. 16). We found that DHPG triggers an enhanced association between PrP C and Homer1b/c but a reduced association between PrP C and Pyk2 as well as between PrP C and CamKII (Fig. 1). Contrarily, A␤o treatment enhances the association of mGluR5 as well as CamKII with PrP C , whereas A␤o stimulation dissociates Homer1b/c as well as Pyk2 from the complex (Fig. 1). We have previously characterized these phenomena to be fully dependent on mGluR5 and PrP C using genetic null brain preparations as well as a HEK cell overexpression system (16). Here we further show that DHPG and A␤o both cause increased mGluR5 but reduced Pyk2 signals in the anti-PrP C precipitates, similar to our previous results showing parallel actions of DHPG and A␤o for Fyn activation and Ca 2ϩ signaling (12). However, DHPG and A␤o have opposite effects on the association of Homer1b/c and CamKII with the complex, demonstrating unique actions of these two ligands. Most critically, DHPG-triggered association changes between PrP C and Homer1b/c or PrP C and CamKII fail to appear in slices pretreated with A␤o for 15 min (Fig. 1, B, D, and F).
The brain slice incubations with A␤o occur over a short time period relative to the course of Alzheimer disease. We sought to understand whether chronic accumulation of A␤o in APP/PS1 transgenic brain prevents DHPG-induced association changes of the complex similar to acute A␤o exposure. In aged wild-type preparations, we found DHPG-enhanced association between PrP C and Homer1b/c but reduced association between PrP C and Pyk2 as well as between PrP C and CamKII (Fig. 2). Interestingly, DHPG-triggered alterations of the PrP C -mGluR5 complex are not detectable in APP/PS1 brain slices (Fig. 2). Thus, the APP/PS1 genetic background prevents DHPG-induced association changes in the complex.
Alterations in protein kinase activation might participate in Alzheimer disease pathogenesis. Pyk2 and CamKII are two kinases associated with the PrP C -mGluR5 complex that are activated by acute A␤o stimulation in a PrP C -and mGluR5-dependent manner (16,18). Here we determined that DHPG as well as ionomycin treatments activate both kinases similarly to A␤o-triggered phosphorylation of Pyk2 and CamKII (Fig. 3). A combination of DHPG and A␤o does not lead to a greater increase in phosphorylation of Pyk2 or CamKII compared with either treatment alone (Fig. 3).
Based on these acute A␤o exposure results, we considered whether chronic APP/PS1 transgene-dependent expression of A␤o similarly prevents DHPG-induced phosphorylation of Pyk2 and CamKII. Importantly, DHPG-triggered activation of the mGluR5-associated kinases Pyk2 and CamKII is not observable in aged APP/PS1 brain slices (Fig. 4). Thus, acute A␤o treatment as well as the APP/PS1 background prevent DHPGinduced activation of Pyk2 and CamKII. Of note, acute A␤o stimulation triggers phosphorylation of CamKII, whereas chronic exposure to A␤o in APP/PS1 mouse brain induces a slight deactivation of CamKII. Our previous work described a biphasic temporal effect of A␤o on CamKII by characterizing the activation of CamKII after different A␤o incubation periods (16). Those experiments documented activation of CamKII after acute A␤o stimulation (e.g. 30 min after stimulation) but deactivation of CamKII after longer A␤o incubation periods (e.g. 6 h after stimulation), representing a more chronic state of A␤o exposure (16). The key point here is that chronic A␤o exposure prevents DHPG regulation of Pyk2 and CamKII activation.
It is striking that DHPG and A␤o induce opposite association changes of CamKII with the PrP C -mGluR5 complex. In contrast, DHPG and A␤o both trigger dissociation of Pyk2 from the PrP C -mGluR5 complex. The differential effects suggest separate mechanisms underlying activation of Pyk2 and CamKII. Although both Pyk2 and CamKII activation are regulated by intracellular Ca 2ϩ (28,29), we sought to assess whether stimulation of Pyk2 and CamKII in acute brain slices might depend on different sources of Ca 2ϩ .
Our data demonstrate that removal of extracellular Ca 2ϩ for 30 min prevents activation of Pyk2 at Tyr-402 in acute brain slices (no significant difference of Pyk2 activation between vehicle treatment and DHPG/A␤o/DHPGϩA␤o treatment in the absence of ext. Ca 2ϩ ; Fig. 5, A and B). In contrast, A␤oinduced activation of CamKII at Thr-286 is independent of extracellular Ca 2ϩ , and DHPG as well as DHPGϩA␤o stimulation follow the same pattern (Fig. 5, A and C). Importantly, DHPG/A␤o/DHPGϩA␤o treatment activates Pyk2 independent of endoplasmic reticulum Ca 2ϩ store depletion by thapsigargin for 30 min (Fig. 6, A and B). In contrast, thapsigargin pretreatment of acute brain slices completely abolished the ability of DHPG/A␤o/DHPGϩA␤o treatment to activate Cam-KII at Thr-286 (Fig. 6, A and C). Thus, removal of extracellular Ca 2ϩ and thapsigargin-induced endoplasmic reticulum Ca 2ϩ depletion differentially affect the activation pattern of the intracellular kinases Pyk2 and CamKII.
We sought to further segregate pathways involved in the activation of Pyk2 and CamKII through PrP C -mGluR5 complexes. As shown previously, Fyn kinase is an additional mediator of A␤o-induced neurotoxicity (13, 30 -32). Preincubation of acute brain slices with AZD0530, a Src family inhibitor that fully blocks Fyn activation, suppresses basal phospho-Pyk2(Tyr-402) levels and prevents activation of Pyk2 at Tyr-402 induced by DHPG or A␤o (Fig. 7, A and B). In contrast, AZD0530 does not alter the activation of CamKII at Thr-286 (Fig. 7, A and C). Thus, both Fyn kinase activity and the source of intracellular Ca 2ϩ distinguish the activation pathways of Pyk2 and CamKII.

Discussion
We have shown previously that PrP C and mGluR5 function as a complex in the brain and that PrP C interacts with the intracellular protein mediators Homer1b/c, Pyk2, and CamKII via mGluR5 in an A␤o-regulated manner (12,16,17). Importantly, FIGURE 1. DHPG and A␤o differentially regulate the association between PrP C and intracellular protein mediators in acute mouse brain slices. A, representative immunoblots showing co-IP between PrP C and mGluR5, Homer1b/c (Homer), Pyk2, and CamKII in acute brain slice crude synaptoneurosomal preparations. Co-immunoprecipitation between PrP C and mGluR5, Homer1b/c, Pyk2, and CamKII is dependent on the genetic presence of prnp. B, wild-type brain slices were treated as indicated prior to preparation of crude synaptoneurosomes. DHPG treatment, 15 min; A␤o treatment, 30 min; A␤oϩDHPG treatment, 15 min of A␤o pretreatment followed by 15 min DHPG treatment. Veh, vehicle. C-F, densitometric analysis of the immunoblots from B, analyzed by one-way ANOVA with Fisher's LSD post hoc pairwise comparisons test (C and F) or post-hoc Tukey's multiple comparisons test (D and E). Data are mean Ϯ S.E. (n ϭ 6 independent experiments for vehicle/DHPG treated slices, n ϭ 14 independent experiments for vehicle/A␤o treated slices, and n ϭ 3 independent experiments for A␤oϩDHPG treated slices). Each experiment includes 6 -8 slices/treatment condition. C, A␤o as well as A␤oϩDHPG treatment of brain slices significantly increases the mGluR5 signal in anti-PrP C immunoprecipitates (*, p Ͻ 0.05). D, co-IP between PrP C and Homer1b/c is significantly enhanced after DHPG treatment (*, p Ͻ 0.05) and significantly reduced after A␤o treatment (***, p Ͻ 0.001). Enhanced DHPG-triggered interaction between PrP C and Homer1b/c is significantly reduced by A␤o pretreatment (*, p Ͻ 0.05). E, co-IP between PrP C and Pyk2 is significantly reduced after DHPG treatment, A␤o treatment, as well as DHPGϩA␤o treatment (**, p Ͻ 0.01; *, p Ͻ 0.05). F, co-IP between PrP C and CamKII is significantly reduced after DHPG treatment but significantly enhanced after A␤o as well as DHPGϩA␤o treatment (**, p Ͻ 0.01; *, p Ͻ 0.05).
our previous study demonstrated a pathological reduction of the PrP C -Homer1b/c association as well as the PrP C -Pyk2 association in transgenic mouse Alzheimer model brain and in human Alzheimer patient brain compared with age-matched healthy controls (16). Thus, a better understanding of the extracellular modulation of association changes in the PrP C -mGluR5 multiprotein complex could provide important insights into Alzheimer disease pathogenesis and might be relevant for therapeutic development. The major finding of this study is that pathological A␤o prevents DHPG-induced changes in the composition of the PrP C -mGluR5 signaling complex. Specifically, A␤o fully activates the intracellular kinases Pyk2 and CamKII. Maximal phosphorylation of these two kinases precludes further activation by the glutamate analog DHPG because of a ceiling effect. Of note, A␤o is a pathological ligand not present in healthy brain, and A␤o exposure in AD is chronic without rapid fluctuation during the synaptic cycle. This study suggests that A␤o exposure traps the PrP C -mGluR5 complex in a conformation that renders it non-responsive to physiological glutamate-induced activation. Moreover, the characteristics of the A␤o-triggered state appear to be distinct from that induced by the glutamate analog DHPG (Fig. 8).
Our previous work showed reduced association of Homer1b/c with the PrP C -mGluR5 complex upon acute A␤o treatment as well as in an Alzheimer disease transgenic model and patient brain (16). Contrary to the A␤o results, we show FIGURE 2. The APP/PS1 transgene prevents DHPG-triggered association changes in the PrP C -mGluR5 complex. A, representative immunoblots of co-IP between PrP C and mGluR5, Homer1b/c, Pyk2, and CamKII from acute brain slice crude synaptoneurosomal preparations. The genotype of brain slices and treatment with 100 M DHPG for 15 min is indicated above each lane. Slices were collected from WT and APP/PS1 mice at 12-16 months of age. B-E, densitometric analysis of the immunoblots from A, analyzed by oneway ANOVA with Fisher's LSD post hoc pairwise comparisons test (B, C, and E) or post-hoc Tukey's multiple comparisons test (D). Data are mean Ϯ S.E. (n ϭ 3 independent experiments). B, the APP/PS1 transgene significantly enhances the mGluR5 signal in anti-PrP C immunoprecipitates (*, p Ͻ 0.05). C, co-IP between PrP C and Homer1b/c is significantly enhanced after DHPG treatment in WT slices (*, p Ͻ 0.05). DHPG treatment of APP/PS1 slices does not significantly alter the association between PrP C and Homer1b/c compared with vehicle-treated APP/PS1 slices (ns, p Ͼ 0.05). D, co-IP between PrP C and Pyk2 is reduced by acute DHPG treatment of WT brain slices as well as the APP/PS1 transgene (*, p Ͻ 0.05). DHPG treatment does not significantly modulate the co-IP between PrP C and Pyk2 compared with vehicle-treated APP/ PS1 slices (ns, p Ͼ 0.05). E, DHPG treatment significantly reduces the co-IP between PrP C and CamKII in WT slices (*, p Ͻ 0.05) but not in APP/PS1 slices (ns, p Ͼ 0.05). here that DHPG stimulation triggers enhanced association between Homer1b/c and the PrP C -mGluR5 complex. Interestingly, pretreatment of brain slices with A␤o followed by DHPG stimulation prevents DHPG-induced enhanced association of Homer1b/c with the PrP C -mGluR5 complex, revealing a dominant effect of prior A␤o exposure.
More importantly, DHPG is unable to alter the association state between PrP C and Homer1b/c in Alzheimer disease transgenic model brain slices. Pathologically enhanced A␤o in human patients might similarly create a chronic state where fluctuating glutamate levels are unable to alter Homer interactions with mGluR5. It has been shown that preventing the interaction between Homer proteins and mGluR proteins prohibits mGluR-dependent synaptic plasticity, and A␤o are known to be potent synaptotoxins (6,33). Thus, we hypothesize that A␤o could prevent signaling mechanisms underlying synaptic plasticity by pathologically altering the conformational relationship between mGluR5 and Homer proteins.
Associated with the intracellular C-terminal tail as well as the intracellular loop 2 of mGluR5, CamKII is well known to be another important mediator of synaptic plasticity (34,35). Thus, CamKII could be a further candidate that might be involved in mechanisms underlying A␤o-triggered inhibition of synaptic function. Jin et al. (35) revealed DHPG-induced dissociation of CamKII from mGluR5 in primary striatal neurons (35). Our previous work as well as other studies demonstrated A␤o-induced enhanced association of CamKII with the PrP C -mGluR5 complex (16,34). Here we reveal abrogation of DHPG-  CamKII (Thr-286) phosphorylation levels after exposure to DHPG (15 min), A␤o (30 min), or DHPGϩA␤o (15 min of A␤o pretreatment followed by 15-min DHPG treatment). 30 min prior to treatment, slices were moved into a fresh bath of Ca 2ϩ -free artificial cerebral spinal fluid (ϩ 1 mM EDTA) as indicated. DMSO was used as vehicle, and the DMSO content was constant in all lanes. B and C, densitometric analysis of the immunoblots from A, data are mean Ϯ S.E. Under normal calcium conditions, n ϭ 11 independent experiments for vehicle or DHPG treatment, n ϭ 21 independent experiments for A␤o treatment, and n ϭ 8 independent experiments for DHPGϩA␤o treatment. In absence of extracellular (ext.) Ca 2ϩ , n ϭ 2 independent experiments for vehicle, DHPG, or DHPGϩA␤o treatment, and n ϭ 3 independent experiments for A␤o treatment. 3 slices/treatment condition and experiment were averaged. B, Pyk2 phosphorylation at Tyr-402 is enhanced by all treatment conditions under normal ext. Ca 2ϩ levels (**, p Ͻ 0.01 by two-way ANOVA with post-hoc Tukey's multiple comparisons test). Pyk2 phosphorylation at Tyr-402 is not significantly altered by any treatment condition in the absence of ext. Ca 2ϩ (ns, p Ͼ 0.05). C, CamKII phosphorylation at Thr-286 is enhanced by all treatment conditions under normal ext. Ca 2ϩ levels (**, p Ͻ 0.01 by two-way ANOVA with Dunnett's multiple comparisons test). In the absence of ext. Ca 2ϩ , CamKII phosphorylation at Thr-286 is enhanced by A␤o-stimulation (*, p Ͻ 0.05 by two-way ANOVA with Dunnett's multiple comparisons test).
induced dissociation of CamKII from the PrP C -mGluR5 complex in the presence of A␤o. Importantly, CamKII is known to dissociate from mGluR upon long-term potentiation induction (35). After dissociation from mGluRs, the kinase associates more strongly with NR2B subunits of NMDA receptors (35). Phosphorylation of NR2B at Tyr-1472 by Fyn kinase is critically involved in long-term potentiation and A␤o toxicity (12,18,36). A␤o-induced inhibition of long-term potentiation may depend in part on CamKII being trapped within the A␤o-PrP C -mGluR5 complex and thus prevent regulation of NMDA NR2B subunits by CamKII.
Activation of intracellular kinases could contribute to neurodegenerative phenotypes, and as such, Fyn kinase inhibition is currently under evaluation as an Alzheimer disease therapeutic in a clinical setting (30,37,38). In this study, we focus on A␤oinduced activation of Pyk2 and CamKII. We find that acute DHPG as well as A␤o treatment of brain slices equally activates  Pyk2 and CamKII. In the presence of both ligands, DHPG and A␤o, no further stimulation is achieved. We tested whether this observation is due to a ceiling effect achieved by either DHPG or A␤o. We used the calcium ionophore ionomycin to achieve a maximal supraphysiological intracellular calcium increase and thereby activate both kinases maximally. The levels of Pyk2 and CamKII phosphorylation in the presence of ionomycin are not significantly different from DHPG-or A␤o-activated Pyk2 and CamKII levels. Thus, our data indicate that activation of either kinase by a first ligand (A␤o) precludes further phosphorylation by a second ligand (DHPG) because both kinases are already fully activated by the first ligand.
Our results are supported by earlier studies reporting DHPG-induced stimulation of Pyk2 in primary cortical neurons as well as dissociation of Pyk2 from mGluR1 upon mGluR stimulation in HEK-293T cells (39,40). Interestingly, we observed a strong correlation between Pyk2 phosphorylation and dissociation from the PrP C -mGluR5 complex. Thus, Pyk2 phosphorylation at Tyr-402 might participate in the dissociation of Pyk2 from the complex. Alternatively, conformational changes induced during dissociation of Pyk2 from the PrP C -mGluR5 complex might enhance Pyk2 autophosphorylation at Tyr-402.
Our results do not show a correlation between CamKII phosphorylation and CamKII dissociation from the PrP C -mGluR5 complex. These differences in the structural relationship between the PrP C -mGluR5 complex and either Pyk2 or CamKII led us to investigate different mechanisms of Pyk2 and CamKII activation. We assessed intra-and extracellular Ca 2ϩ compartments to determine how Ca 2ϩ modulates the activation of Pyk2 or CamKII. We found that removal of extracellular Ca 2ϩ completely abrogated the activation of Pyk2. These results suggest that the activation of Pyk2 is most likely dependent on extracellular Ca 2ϩ influx through NMDA receptors. In contrast, stimulation of CamKII was not affected by removal of extracellular Ca 2ϩ but was fully abolished by using an endoplasmic reticu-lum Ca 2ϩ ATPase inhibitor. Importantly, mGluR5 activity is directly linked to endoplasmic reticulum Ca 2ϩ levels, which, in turn, regulate the expression of stromal interaction molecule 2 (STIM2) (41). STIM2 is an endoplasmic reticulum-localized protein that controls neuronal store-operated calcium entry (nSOC) and thereby regulates CamKII activity and stabilization of mushroom spines (42). Interestingly, previous work demonstrated that the STIM2-nSOC-CaMKII pathway is dysregulated in neurons from Alzheimer mouse models as well as in Alzheimer patient brains (41,42). Of note, thapsigargin-induced endoplasmic reticulum Ca 2ϩ depletion hyperactivates the nSOC pathway via STIM proteins, which leads to reduced activity of CamKII (42,43). Thus, our results demonstrating the absence of CamKII activation in the presence of thapsigargin might be due to thapsigargin-dependent opening of nSOC with reduced CamKII activity because of enhanced nSOC. Alternatively, activation of CamKII could directly depend on the release of Ca 2ϩ from endoplasmic reticulum stores. Previous workdemonstratedsimilarresultsforDHPG-inducedERKphosphorylation in rat striatal neurons. In those studies, intracellular Ca 2ϩ store depletion by thapsigargin completely abolished DHPG-triggered activation of ERK (25). Further work is necessary to ultimately clarify the role of different Ca 2ϩ channels in CamKII activation.
We further characterized activation of Pyk2 to be fully dependent on Fyn kinase activity, whereas activation of CamKII is not affected by Fyn kinase inhibition. Taken together, it is clear that two separate signaling pathways trigger the activation of Pyk2 and CamKII (Fig. 7).
Our data reveal signaling mechanisms that might contribute to Alzheimer disease pathogenesis. It is well established that glutamate-induced activation of metabotropic intracellular signaling is essential for physiological brain function. Our data demonstrate that Alzheimer disease transgenic model mice are unable to respond to stimulation by the glutamate analog DHPG. We speculate that A␤o-induced pathological confor- FIGURE 8. Alzheimer disease A␤o prevents physiological activation of the PrP C -mGluR5 signaling complex. A, schematic demonstrating the interaction between GPI-anchored PrP C with Pyk2, Homer1b/c, and CamKII via mGluR5 in postsynaptic spines. The interaction is modulated by physiological glutamate as well as by pathological A␤o. B, under physiological conditions, the glutamate analog DHPG triggers enhanced association between Homer1b/c and the PrP C -mGluR5 signaling complex. Contrarily, DHPG stimulation initiates dissociation of Pyk2 as well as CamKII from the complex, which coincides with autophosphorylation of both kinases. In presence of acute A␤o, Homer1b/c as well as Pyk2 are released from the complex, whereas CamKII shows enhanced association with the complex. DHPG and acute A␤o triggers phosphorylation of Pyk2 as well as CamKII. Activation of Pyk2 is dependent on ext. Ca 2ϩ and Fyn kinase activity, whereas phosphorylation of CamKII is dependent on the release of Ca 2ϩ from intracellular stores. Chronic A␤o affects Pyk2 similarly to acute A␤o. Contrarily, CamKII is not altered by pathological chronic exposure to A␤o. Interestingly, association changes in the PrP C -mGluR5 complex induced by Alzheimer disease A␤o prevent physiological DHPG-induced signaling events. 1 indicates enhanced binding between the protein of interest and the PrP C -mGluR5 complex or enhanced phosphorylation. 2 indicates reduced binding between the protein of interest and the PrP C -mGluR5 complex or a decrease in phosphorylation. Ϫ indicates the absence of changes in co-immunoprecipitation or phosphorylation of the protein of interest. The column DHPG in presence of A␤o compares the effect of DHPG after A␤o pretreatment with the effect of DHPG in the absence of A␤o; Ϫ implies no further change induced by DHPG stimulation after A␤o pretreatment.
mations of the PrP C -mGluR5 complex could participate in the development of synaptic and behavioral abnormalities in transgenic mice (14, 44 -47). We conclude that physiologically fluctuating levels of glutamate might be unable to modify the interaction between mGluR5 and intracellular signaling mediators in a pathophysiological A␤o-dependent disease state. Similar pathophysiological association states between mGluRs and Homer scaffolding proteins have been proposed in the context of several neurological diseases, including fragile X syndrome and Angelman syndrome (48,49). Our data suggests that preventing A␤o-induced activation of the PrP C -mGluR5 signaling complex has potential clinical implications for Alzheimer disease. Our previous work revealed PrP C -directed antibodies and mGluR5-directed compounds to potently modulate the interaction between PrP C and mGluR5 (17). Future studies will further explore their potential as Alzheimer disease therapeutics.

Experimental Procedures
Soluble A␤  Oligomer Preparation-The Keck Large Scale Peptide Synthesis Facility (Yale University) synthesized A␤  peptide. The preparation of A␤ 1-42 oligomers (A␤o) has been described previously (13). Concentrations of A␤o are expressed in monomer equivalents, with 1 M total A␤ 1-42 peptide corresponding to ϳ10 nM oligomeric species (11).
Mouse Strains-Mice were cared for by the Yale Animal Resource Center, and all experiments were approved by the Institutional Animal Care and Use Committee of Yale University. Wild-type and APPswe/PS1dE9 (APP/PS1) mice (50) were purchased from The Jackson Laboratory and maintained on a C57/Bl6J background. Mice were group-housed (2-5 mice/ cage) in a 12-h light/dark cycle. Male and female mice were used equally.
Preparation of Acute Mouse Brain Slices-Mouse brains were dissected after rapid decapitation and immediately dispersed in ice-cold artificial cerebral spinal fluid (119 mM NaCl; 2.5 mM KCl; 1.3 mM MgSO 4 ; 26.2 mM NaHCO 3 ; 11 mM D-glucose; 1.25 mM NaH 2 PO 4 ). 400-m coronal brain slices were cut in icecold artificial cerebral spinal fluid using a 1000 Plus vibratome with steel razor blades. Brain slices were incubated at room temperature in artificial cerebral spinal fluid plus 2.4 mM CaCl 2 under constant oxygenation with 95% O 2 and5% CO 2 . After a 2-h recovery period, slices were treated with A␤o and/or DHPG. For acute A␤o/DHPG experiments, 4-to 8-week-old wild-type mice were used. For transgenic studies, wild-type and APP/PS1 brain slices were prepared from mice at 12-16 months of age, when memory deficits, synaptic loss, and A␤ accumulation are established (12-14, 18, 51).
Crude Synaptoneurosome Preparation-Acute brain slices were homogenized in buffer A (0.32 M sucrose, 20 mM HEPES (pH 7.4), 1 mM EDTA, 1ϫ PhosSTOP-Roche, and 1ϫ cOmplete mini protease inhibitor mixture (Roche)). Homogenates were centrifuged for 10 min at 875 ϫ g at 4°C. The supernatant was again centrifuged for 10 min at 16,000 ϫ g at 4°C to obtain a cytosolic fraction (supernatant) and a crude synaptoneurosomal fraction (P2 pellet). P2 pellets were resuspended and sonicated in buffer A prior to use.
Immunoprecipitation-For each immunoprecipitation (IP) experiment, crude synaptoneurosomal fractions from 6 -8 mouse brain slices/treatment condition were combined prior to IP (n ϭ number of separate experiments). The protein concentration in crude synaptoneurosomal fractions was determined by Bradford assay (Bio-Rad protein assay), and fractions were precleared from endogenous antibodies for 4 h at 4°C. Crude synaptoneurosomal fractions were then incubated overnight with capture antibody (1 g/1 mg of homogenate protein) at 4°C. The capture antibody used was Saf32 (Cayman, 189720, mouse anti-prion protein), which was validated by comparing immunoprecipitation from wild-type to prnp-null brain. The preformed antibody-antigen complexes were then incubated with PureProteome protein A/G mix magnetic beads (Millipore, LSKMAGAG10) for 1 h at 4°C under gentle rotation. Beads were washed five times in buffer A prior to elution of proteins in SDS-PAGE sample loading buffer. The immunoprecipitated complexes were then resolved by SDS-PAGE and immunoblotted.
Analysis of Protein Activation States after Treatment of Acute Mouse Brain Slices-Acute mouse brain slices were prepared and treated as described above. A␤ was oligomerized as described previously in a vehicle composed of DMSO in glutamate-free F-12 medium (13). The vehicle used for DHPG, thapsigargin, as well as AZD0530 was double-distilled H 2 O. Ionomycin was prepared in DMSO and used at a concentration of 1 M. DHPG was used at a concentration of 100 M, and thapsigargin as well as AZD0530 were used at a concentration of 2 M. For the experiments in Figs. 3 and 5-7, the DMSO content in all lanes was constant.
After incubation of brain slices with compounds as indicated in the figure legends, slices were homogenized in radioimmune precipitation assay lysis buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 0.1% SDS, 0.5% deoxycholic acid, 1ϫ PhosSTOP-Roche, and 1ϫ cOmpletemini protease inhibitor mixture (Roche)) and centrifuged at 21,000 ϫ g for 20 min at 4°C. The protein concentration in the radioimmune precipitation assay-soluble fraction was determined by Bradford assay (Bio-Rad protein assay). The radioimmune precipitation assay-soluble fraction was then mixed with SDS-PAGE sample loading buffer, and proteins were resolved by SDS-PAGE followed by immunoblot. Experiments consisted of averaged data from three slices (n ϭ number of separate experiments). No experiments were excluded from the final data.
Immunoblots-Proteins were electrophoresed through precast 4 -20% Tris/glycine gels (Bio-Rad) and transferred with an iBlot TM gel transfer device (Novex-Life Technologies) onto nitrocellulose membranes (Invitrogen). Membranes were blocked (blocking buffer for fluorescent Western blotting, Rockland, MB-070-010) for 1 h at room temperature and incubated overnight in primary antibodies at 4°C. The following antibodies were used: rabbit anti-actin (Sigma-Aldrich, A2066, 1:3000), goat anti-CaMKII (Santa Cruz Biotechnology, sc-5392, 1:500), rabbit anti-Homer1b/c (Santa Cruz Biotechnology, sc-55463, 1:500), rabbit anti-mGluR5/1 (R&D Systems, PPS079, 1:1000), rabbit anti-phospho-CamKII (Abcam, ab5683, 1:1000), rabbit anti-phospho-Pyk2 (Cell Signaling Technology, 3291, 1:1000), mouse anti-Pyk2 (Cell Signaling Technology, 3480, 1:1000), and Saf32 (Cayman, 189720, 1:200, mouse anti-prion protein). Secondary antibodies were applied for 1 h at room temperature (Odyssey donkey anti-mouse or donkey anti-rabbit conjugated to IRDye 680 or IRDye 800, LI-COR Biosciences), and proteins were visualized with a LI-COR Odyssey infrared imaging system. Quantification of band intensities was performed within a linear range of exposure. Actin was used as loading control, and Actin levels were not affected by any treatment condition. For phosphorylation studies, signals detected by phosphoepitope-specific antibodies were normalized to the total protein level as detected by the non-phospho-specific antibody of the target protein. In the case of co-immunoprecipitation experiments, the level of the co-immunoprecipitated target protein was normalized to the input level of that protein. This normalized level was further normalized by the normalized level of the precipitated protein (PrP C level). Only validated antibodies were used. Validation information can be found on the website of the company or on Antibodypedia.
Statistics-All results are presented as mean Ϯ S.E. Prism 6 software was used for statistical analysis. Data were analyzed using either one-way or two-way ANOVA (analysis of variance), followed by post-hoc Tukey's multiple comparisons test, Dunnett's multiple comparisons test, or Fisher's LSD post hoc pairwise comparisons test, as specified in the figure legends. p Ͻ 0.05 was considered statistically significant.