Alzheimer’s disease-related phenotypes in mice can be rescued by blockade of either cellular prion protein or metabotropic glutamate receptor 5. We sought genetic and biochemical evidence that these proteins function cooperatively as an obligate complex in the brain. We show that cellular prion protein associates via transmembrane metabotropic glutamate receptor 5 with the intracellular protein mediators Homer1b/c, calcium/calmodulin-dependent protein kinase II, and the Alzheimer’s disease risk gene product protein tyrosine kinase 2 beta. Coupling of cellular prion protein to these intracellular proteins is modified by soluble amyloid-β oligomers, by mouse brain Alzheimer’s disease transgenes or by human Alzheimer’s disease pathology. Amyloid-β oligomer-triggered phosphorylation of intracellular protein mediators and impairment of synaptic plasticity in vitro requires Prnp–Grm5 genetic interaction, being absent in transheterozygous loss-of-function, but present in either single heterozygote. Importantly, genetic coupling between Prnp and Grm5 is also responsible for signalling, for survival and for synapse loss in Alzheimer’s disease transgenic model mice. Thus, the interaction between metabotropic glutamate receptor 5 and cellular prion protein has a central role in Alzheimer’s disease pathogenesis, and the complex is a potential target for disease-modifying intervention.
Biomarker studies of late-onset Alzheimer’s disease in conjunction with genetic studies of dominantly inherited early-onset cases support the hypothesis that accumulation of amyloid-β peptide triggers the disease. Amongst peptide conformations, soluble oligomers of amyloid-β impair synaptic function and can lead to subsequent tauopathy ( Lambert et al. , 1998 ; Kayed et al. , 2003 ; Lesne et al. , 2006 ; Shankar et al. , 2008 ; Jin et al. , 2011 ). An unbiased genome-wide expression cloning experiment identified cellular prion protein (PrP C ) as a cell surface binding site for amyloid-β oligomers ( Lauren et al. , 2009 ). Amyloid-β oligomers bind to PrP C with high affinity ( Lauren et al. , 2009 ; Balducci et al. , 2010 ; Zou et al. , 2011 ) and multiple studies support a role for PrP C in mediating Alzheimer’s disease-related pathology ( Lauren et al. , 2009 ; Chen et al. , 2010 ; Chung et al. , 2010 ; Gimbel et al. , 2010 ; Barry et al. , 2011 ; Bate and Williams, 2011 ; Freir et al. , 2011 ; Resenberger et al. , 2011 ; Zou et al. , 2011 ; Kudo et al. , 2012 , 2013 ; Larson et al. , 2012 ; Um et al. , 2012 , 2013 ; Fluharty et al. , 2013 ; Nicoll et al. , 2013 ; Ostapchenko et al. , 2013 ; Rushworth et al. , 2013 ; Dohler et al. , 2014 ; Hu et al. , 2014 ; Klyubin et al. , 2014 ; Walsh et al. , 2014 ). While two studies demonstrated amyloid-β triggered impairment of long-term potentiation (LTP) independently of PrP C ( Calella et al. , 2010 ; Kessels et al. , 2010 ), it has been suggested that different conformations of amyloid-β preparations explain variable outcomes in amyloid-β oligomer toxicity assays ( Reed et al. , 2011 ; Nicoll et al. , 2013 ). Critically, the major species of soluble amyloid-β oligomers derived from human Alzheimer’s disease brain bind to PrP C ( Um et al. , 2012 ) and require PrP C to impair LTP in vivo ( Barry et al. , 2011 ). Among the consequences of amyloid-β oligomers binding to PrP C , downstream activation of FYN kinase alters NMDA ( N -methyl D-aspartate) receptors and dendritic spine stability ( Um et al. , 2012 ). FYN also interacts with tau, such that FYN kinase inhibitors are being examined as agents with disease-modifying action in Alzheimer’s disease ( Nygaard et al. , 2014 , 2015 ; Kaufman et al. , 2015 ).
The altered function and loss of excitatory synapses is a central step in Alzheimer’s disease brain dysfunction ( Scheff et al. , 1990 ), and PrP C is essential for synaptic loss induced by amyloid-β oligomers in vitro or APP/PS1 + (amyloid precursor protein/presenilin 1) transgene in vivo ( Gimbel et al. , 2010 ; Um et al. , 2012 ). The metabotropic glutamate receptor 5, mGluR5 (encoded by GRM5 ), is a protein known to provide chronic control over the density and activity of such synapses in a broad range of conditions ( Bear et al. , 2004 ). The presence of amyloid-β oligomers alters the trafficking of mGluR5 in neurons, and inhibition of mGluR5 function rescues amyloid-β oligomer or APP transgene deficits in synaptic plasticity, synaptic density and memory function in vivo ( Renner et al. , 2010 ; Um et al. , 2013 ; Hamilton et al. , 2014 ; Hu et al. , 2014 ). mGluR5 is highly accessible to cytosolic binding partners at its large intracellular C-terminal tail, where members of the Homer family of scaffold proteins link mGluR5 to protein kinases and modulate mGluR5 signalling ( Brakeman et al. , 1997 ; Xiao et al. , 1998 ; Ango et al. , 2001 ; Mao et al. , 2005 ). One of the kinases associated with both mGluR5 and Homer (encoded by HOMER1 ) is calcium/calmodulin-dependent eukaryotic elongation factor 2 kinase (eEF2K, encoded by EEF2K ) ( Park et al. , 2008 ). eEF2K is released upon activation of mGluR5 to phosphorylate eukaryotic elongation factor 2 (eEF2, encoded by EEF2 ) at T56, a translation elongation controlling protein. Interestingly, amyloid-β oligomer-induced impairment of LTP is dependent on eEF2 phosphorylation at T56 ( Ma et al. , 2014 ). Importantly, phosphorylation at this residue is enhanced in the hippocampus of APP/PS1 + Alzheimer’s disease model mice as well as in human Alzheimer’s disease post-mortem brain samples compared to healthy controls ( Ma et al. , 2014 ). eEF2 phosphorylation at T56 inhibits translation of proteins globally ( Nairn and Palfrey, 1987 ), but initiates translation of specific mRNAs, one of which is calcium/calmodulin-dependent protein kinase II (CamKII, encoded by CAMK2A ) ( Scheetz et al. , 2000 ). CamKII has been reported to associate with mGluR5 in a Ca 2+ - and amyloid-β oligomer-dependent manner ( Jin et al. , 2013 b ; Raka et al. , 2015 ). Analysis of Alzheimer’s disease patient brain revealed that autophosphorylated Thr286-CamKII is selectively lost in the frontal cortex and hippocampus of Alzheimer’s disease patients compared to age-matched healthy controls ( Amada et al. , 2005 ). A further member of the mGluR5 signalling complex is proline-rich tyrosine kinase 2 (Pyk2, encoded by PTK2B ; also known as protein tyrosine kinase 2 beta or focal adhesion kinase 2) ( Nicodemo et al. , 2010 ). A single nucleotide polymorphism in Pyk2 was recently associated with susceptibility to late-onset Alzheimer’s disease in a genome-wide association study, emphasizing the potential significance of this kinase for Alzheimer’s disease ( Lambert et al. , 2013 ; Wang et al. , 2015 ).
While both PrP C ( Lauren et al. , 2009 ; Chen et al. , 2010 ; Chung et al. , 2010 ; Gimbel et al. , 2010 ; Barry et al. , 2011 ; Bate and Williams, 2011 ; Freir et al. , 2011 ; Resenberger et al. , 2011 ; Zou et al. , 2011 ; Kudo et al. , 2012 , 2013 ; Larson et al. , 2012 ; Um et al. , 2012 , 2013 ; Fluharty et al. , 2013 ; Nicoll et al. , 2013 ; Ostapchenko et al. , 2013 ; Rushworth et al. , 2013 ; Dohler et al. , 2014 ; Hu et al. , 2014 ; Klyubin et al. , 2014 ; Walsh et al. , 2014 ) and mGluR5 ( Um et al. , 2013 ; Hamilton et al. , 2014 ; Hu et al. , 2014 ) are required to mediate Alzheimer’s disease-related phenotypes, they may function together in a complex or, alternatively, act independently in Alzheimer’s disease pathophysiology. Our previous co-immunoprecipitation experiments suggested a co-dependent mechanism ( Um et al. , 2013 ), but extracellular regulation of mGluR5 by endogenous molecules other than glutamate is unique. Importantly, the previous in vivo functional tests blocked either mGluR5 or PrP C , and therefore could not distinguish their independent action from an essential role for amyloid-β oligomer/PrP C interaction with mGluR5 in vivo.
Here we show that coupling between PrP C and mGluR5 is critically involved in Alzheimer’s disease pathology. Our data reveal that PrP C associates with Homer1b/c, Pyk2, and CamKII selectively when mGluR5 is present. Lowering the gene dosage of both Prnp and Grm5 rescues amyloid-β oligomer-induced deficits in synaptic plasticity in vitro and both synapse loss and mouse survival in the APP/PS1 + Alzheimer’s disease transgenic model in vivo. These data indicate that complex formation between PrP C and mGluR5 has central significance in Alzheimer’s disease. Furthermore, partial blockade of both targets, or their interaction, suggests a promising therapeutic window for modifying the course of Alzheimer’s disease.
Materials and methods
Amyloid-β 1-42 preparation
Synthetic amyloid-β 1-42 peptide was obtained as lyophilized powder from Keck Large Scale Peptide Synthesis Facility (Yale University). Preparation and characterization of amyloid-β 1-42 monomers and amyloid-β 1-42 oligomers (amyloid-β oligomers) have been described previously ( Um et al. , 2012 ). Amyloid-β 1-42 fibrils were prepared as amyloid-β oligomers and incubated for 6 months at room temperature. Concentrations of amyloid-β oligomers are expressed in monomer equivalents, with 1 μM total amyloid-β 1-42 peptide corresponding to ∼10 nM oligomeric species ( Lauren et al. , 2009 ).
Mice were cared for by the Yale Animal Resource Center and all experiments were approved by Yale’s institutional animal care and use committee. Mouse strains were maintained on a C57/Bl6J background as described previously ( Lu et al. , 1997 ; Jankowsky et al. , 2003 ; Gimbel et al. , 2010 ). Mice were group housed (two to five mice per cage) with a 12-h light/dark cycle.
All experiments were conducted in a blinded fashion with respect to genotype and treatment. Groups were matched for age and sex. In addition, human brain groups were also matched for post-mortem interval and demographical data. The number of animals used for each subsequent experiment was calculated based on previous experiments to ensure adequate statistical power.
Preparation of acute mouse brain slices
Mouse brains from 4–12-week-old wild-type, heterozygote or knockout mice were dissected after rapid decapitation. Acute 400-μm coronal slices were cut in ice-cold artificial CSF (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 ) using a 1000 Plus Vibratome® with steel razor blades. Brain slices were then transferred to brain slice incubators filled with artificial CSF plus 2.4 mM CaCl 2 and incubated under constant oxygenation with 95% O 2 and 5% CO 2 . Slices were recovered for 2 h prior to incubation with amyloid-β oligomers for 30 min.
Crude synaptoneurosome preparation
To purify crude synaptoneurosomes, acute brain slices were homogenized in buffer A [0.32 M sucrose, 20 mM HEPES (pH 7.4), 1 mM EDTA, 1 × PhosSTOP ™ and 1 × cOmplete ™ Mini protease inhibitor cocktail (Roche)]. Homogenates were centrifuged for 10 min at 875 g at 4°C. The supernatant was collected and centrifuged for 10 min at 16 000 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.
The protein concentration in crude synaptoneurosomal fractions was determined by Bradford assay (Bio-Rad Protein Assay) prior to immunoprecipation. Crude synaptoneurosomal fractions were pre-cleared from endogenous antibodies for 4 h at 4°C and lysates were incubated overnight with capture antibody (1 µg/ 1 mg of homogenate) at 4°C. The capture antibodies used were Saf32 (Cayman, 189720) and Bar233 (Cayman, 10009036) for anti-PrP C immunoprecipitation or anti-Myc (Sigma Aldrich, C3956). 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 (sodium dodecyl sulphate-polyacrylamide gel electrophoresis) sample loading buffer. The immunoprecipitated complexes were then resolved by SDS-PAGE and immunoblotted.
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). Loading controls (actin) were run on the same gel. 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 primary antibodies were used: anti-actin (Sigma-Aldrich A2066; 1:3000), anti-CaMKII (Abcam ab89197; 1:1000), anti-eEF2 (Cell Signaling Technology #2332; 1:1000), anti-Homer1b/c (Santa Cruz Biotechnology sc-55463; 1:1000), anti-Interleukin-1 beta (R&D systems AB-401-NA; 1:1000), anti-mGluR5/1 (R&D Systems PPS079; 1:1000), anti-phospho-CamKII (Abcam ab5683; 1:1000), anti-phospho-eEF2 (Cell Signalling Technology #2331; 1:1000), anti-phospho-Pyk2 (Cell Signalling #3291; 1:1000), anti-Pyk2 (Cell Signalling #3480; 1:1000) and Saf32 (Cayman 189720; 1:200). Only validated antibodies were used. Validation information can be found on the company websites or on www.antibodypedia.com . Appropriate 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. For co-immunoprecipitates, the level of each precipitated protein is normalized to that protein’s input (e.g. Homer co-immunoprecipitate level/Homer input level). This normalized level is further normalized by the normalized PrP level (PrP immunoprecipitate level/PrP input level).
Human embryonic kidney-293T (HEK-293T) cells were maintained in Dulbecco’s modified Eagle medium, supplied with 10% foetal bovine serum, 1% L-glutamine (2 mM final concentration), 1% sodium pyruvate (1 mM final concentration) and 1% penicillin/streptomycin (100 U/ml). Cells were transfected using Lipofectamine® 2000 transfection reagent (Invitrogen). Expression vectors used for rat mGluR5, human PrP C , Homer1c, CamKIIα and Pyk2 are pcDNA3.1 or pEGFP. HEK-293T cells were harvested 48 h after transfection in buffer A. Cell membranes were prepared using the same protocol as described in the ‘Crude synaptoneurosome preparation’ section. Here, the P2 pellet was resuspended in radioimmunoprecipitation assay (RIPA) 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; 1 × PhosSTOP ™ -Roche; 1 × cOmplete ™ Mini protease inhibitor cocktail (Roche)] and centrifuged at 100 000 g for 1 h at 4°C. The protein concentration in the supernatant was determined by Bradford assay (Bio-Rad Protein Assay) and subjected to immunoprecipitation as described above.
Mouse brain tissue collection for biochemistry
All mice used for biochemistry were between 12 and 26 months old at the time of analysis with an average age of 644 ± 32 days at the time of sacrifice. Mice were euthanized by rapid decapitation and hippocampus, cortex and cerebellum of brains were rapidly dissected. Brain regions were weighed and homogenized in three times the brain tissue weight in TBS (Tris-buffered saline) containing PhosSTOP ™ (Roche) and cOmplete ± Mini protease inhibitor cocktail (Roche) to extract cytosolic proteins. After centrifugation for 30 min at 100 000 g and 4°C, the supernatants were collected as TBS-soluble fractions. The remaining pellets were dissolved in three times the original brain tissue weight in RIPA lysis buffer. After centrifugation for 30 min at 100 000 g and 4°C, the supernatants were collected as RIPA soluble fractions. The protein concentration in samples was measured by Bradford assay (Bio-Rad Protein Assay) prior to immunoprecipitation or direct separation by SDS-PAGE.
Human brain tissue collection
Human post-mortem brain tissue was collected as approved by the Institutional Review Board at Yale and stored at −80°C until experimental use. The neuropathological features of Alzheimer’s disease in patient’s brains were confirmed by classifying them as Braak stage V or higher with a ‘frequent’ CERAD (Consortium to Establish a Registry for Alzheimer’s Disease) neuritic plaque score. Healthy controls were classified as Braak stage 0–II with a CERAD neuritic plaque score of ‘none’ or ‘sparse’. Patient demographic data have been described previously ( Kostylev et al. , 2015 ). Parkinson’s disease without Alzheimer’s features was diagnosed by clinical neuropathological criteria.
Human frontal cortex was homogenized in three times the brain tissue weight in TBS containing PhosSTOP ™ (Roche) and cOmplete ™ Mini protease inhibitor cocktail (Roche). Homogenates were centrifuged for 30 min at 100 000 g and 4°C. The supernatant was again centrifuged for 30 min at 100 000 g and 4°C. The supernatant was collected as TBS-soluble fraction. Proteins were extracted from the remaining pellets by centrifugation in three times the original brain tissue weight in TBS + 1% Triton ™ X-100 for 30 min at 100 000 g and 4°C. The supernatant was collected and again centrifuged for 30 min at 100 000 g and 4°C. The protein concentration in the supernatant was measured by Bradford assay (Bio-Rad Protein Assay) and proteins were subjected to immunoprecipitation.
Analysis of protein activation states after treatment of acute mouse brain slices
Acute mouse brain slices were prepared and treated as described above. Slices were then homogenized in RIPA lysis buffer and centrifuged at 21 000 g for 20 min at 4°C. The protein concentration in the RIPA soluble fraction was determined by Bradford assay (Bio-Rad Protein Assay). The RIPA soluble fraction was then mixed with SDS-PAGE sample loading buffer and proteins were resolved by SDS-PAGE followed by immunoblot.
Coronal slices (400-μm) of adult 6–12-week-old mice were prepared as described above and allowed to recover for 1 h prior to recording. Slices were submerged in a recording chamber (BSC-PC, Warner Instruments) containing artificial CSF with 2.4 mM CaCl 2 that was continuously oxygenated (95% O 2 , 5% CO 2 ). Extracellular field recordings were performed in the stratum radiatum of CA1 neurons in the hippocampus by stimulating Schaffer collateral fibres using a bipolar tungsten electrode (TM33CCNON, World Precision Instruments). Extracellular field excitatory postsynaptic potentials (EPSPs) were recorded using a glass microelectrode (2–6 MΩ) (4878, World Precision Instruments) filled with artificial CSF, and placed in the stratum radiatum of CA1 neurons. Test stimuli were induced at 0.033 Hz, and the stimulus intensity was set at 50% of maximal field EPSP slope. A stable baseline was recorded for at least 20 min prior to induction of LTP, and slices were excluded if baseline could not be stabilized after 1 h. LTP was induced by theta burst stimulation (10 bursts of four shocks at 100 Hz, with an interburst interval of 200 ms) given at baseline intensity. Slices were excluded in post hoc analysis by pre-established cut-off if 10 or more data points fell below 75% of baseline level during post-theta burst stimulation recording. Field EPSPs were recorded using an Axon Instruments 700B amplifier and a Digidata 1440 A digitizer, with data analysis performed by pClamp 10 software (Molecular Devices) and Prism 6 software. Amyloid-β oligomers (1 µM), and vehicle were diluted into artificial CSF of the recording chamber and incubated for 30 min prior to LTP induction. Paired-pulse facilitation experiments were performed as described by Kiyota et al. (2009) by using a dual pulse at 40-µs duration with stimulus intensity set at 40–60% of maximal field EPSP slope. Pulses were delivered in 20-s intervals and six consecutive responses were averaged for each paired-pulse facilitation test with varying interpulse interval (25, 50, 100, 200, 300 ms). All slices maintained a steady baseline for at least 20 min before paired-pulse facilitation test. The ratio of paired-pulse facilitation was determined by the ratio of the second field EPSP slope divided by the first field EPSP slope in each pair. Input–output responses were recorded from slices that maintained a steady baseline for 20 min. Responses were recorded by increasing stimulus intensity by 0.5 V with 40-µs duration starting at 10–20% of maximal field EPSP.
Immunohistochemistry of mouse brain slices
All mice used for immunohistochemistry were between 12 and 26 months old at the time of analysis with an average age of 666 ± 35 days at the time of sacrifice. Mice were euthanized and immediately perfused with PBS (phosphate-buffered saline) for 3 min, followed by 4% PFA (paraformaldehyde) for 5 min. Brains were dissected out and fixed in 4% PFA overnight. Brains were sliced into 40 μm parasagittal brain sections using a Leica WT1000S vibratome. Proteins were permeabilized by three washes in PBS + 0.1% Triton ™ X-100 for 5 min each. Sections were blocked in 10% normal donkey or normal horse serum (Jackson ImmunoResarch Laboratories) in PBS + 0.1% Triton ™ X-100 for 1 h and then incubated with primary antibodies for 16–24 h at 4°C. The following primary antibodies were used: anti-β-amyloid (Cell Signaling Technology 2454; 1:250), anti-GFAP (glial fibrillary acidic protein; Abcam ab7260; 1:500), anti-Iba1 (ionized calcium-binding adapter molecule 1; Wako 019-19741; 1:250), anti-PSD95 (postsynaptic density protein 95; Invitrogen 51-6900; 1:250), and anti-SV2a (synaptic vesicle glycoprotein 2A; Abcam 32942; 1:250). Validation of antibodies is reported on either www.antibodypedia.com or on the company websites. Sections were washed three times in PBS and incubated with secondary antibodies (donkey anti-rabbit or donkey anti-mouse fluorescent antibodies, Invitrogen Alexa Fluor®; 1:500) for 2 h at room temperature. After three washes in PBS, the sections were mounted onto glass slides (SuperFrost ™ Plus, Fisher Scientific) and coverslipped with Vectashield (Vector Laboratories H-1200) antifade aqueous mounting medium. Sections used for thioflavin-S (Sigma T1892) staining were mounted onto glass slides (SuperFrost ™ Plus, Fisher Scientific), washed in PBS for 5 min and incubated in 0.1% thioflavin-S solution in 70% ethanol at room temperature for 15 min. Slices were washed twice in 70% ethanol and twice in distilled water prior to coverslipping.
Imaging and analysis of immunohistochemistry
For imaging of synapse density stained by anti-SV2a and anti-PSD95 antibodies, an UltraView Vox spinning disc confocal microscope (PerkinElmer) with a 60 × 1.3 NA oil-immersion lens was used. The area occupied by immunoreactive synaptic puncta from the molecular layer of the dentate gyrus was measured as described previously ( Gimbel et al. , 2010 ). For imaging and analysis of the tissue stained for Iba1, GFAP or amyloid-β, a Zeiss AxioImager Z1 fluorescent microscope with a 4×or 20× air objective lens was used. The hippocampus, frontal cortex, cortex dorsal to hippocampus and occipital cortex were imaged for each mouse and the area occupied by plaques or cell bodies was analysed by quantification using ImageJ. Statistical analysis was based on separate mice.
Immunocytochemistry of primary neuronal cultures
Cortical neurons were obtained from embryonic Day 17 wild-type or Prnp −/− mouse embryos. After digestion with papain (30 min at 37°C and 5% CO 2 ), the neurons were plated on poly-D-lysine-coated coverslips (BD Biosciences BD354086) placed into 24-well plates or 35 mm cell culture dishes. Neurons were cultured in Neurobasal® A medium supplemented with B-27, 0.5 mM L-glutamate, 1 mM sodium pyruvate, and 1 × penicillin and streptomycin (Invitrogen). At 20–21 days in vitro , neurons were treated with 1 µM amyloid-β oligomers or vehicle. Cultures were then fixed in a 4% PFA + 4% sucrose solution, permeabilized with 0.1% Triton ™ X-100 in PBS for 15 min and blocked with 10% normal donkey serum for 1 h. Immunostaining of neurons was performed overnight at 4°C. The following primary antibodies were used: anti-Homer (Santa Cruz Biotechnology sc-8921; 1:250), anti-MAP2 (Novus Biologicals NB300-213; 1:2000), anti-mGluR5 (Abcam ab76316; 1:250), anti-phospho-CamKII(T286) (Abcam ab5683; 1:1000), and anti-prion protein (Aves Labs chicken anti-prion protein; 1:1000). www.antibodypedia.com or the company websites offer validation information of all antibodies used in this study. Incubation of neurons with a 1:500 dilution of secondary antibodies (Alexa Fluor® 488, 568 or 633 conjugated donkey anti-rabbit, donkey anti-chicken or donkey anti-goat antibody) was performed for 1 h at room temperature. Coverslips were mounted onto glass slides (SuperFrost ™ Plus) with Vectashield (Vector Laboratories H-1200) antifade aqueous mounting medium. Images were acquired on an UltraVIEW Vox spinning disc confocal microscope (PerkinElmer) with a 60 × 1.3 NA oil-immersion lens.
Measuring the levels of total amyloid-β peptide and PrP C interacting amyloid-β oligomers
Total amyloid-β levels in TBS-soluble fractions were measured using a commercially available amyloid-β 1-42 detection kit (Life Technologies KHB3442) and all measurements were performed according to the manufacturer’s recommendations. A Victor 3 V plate reader (Perkin Elmer) was used to measure the absorbance at 450 nm. The absolute amount of amyloid-β was determined using the standard curve based on serial dilutions of a monomeric preparation of synthetic amyloid-β. PrP C -linked immuonosorbent assay PrP C -ELISA, which measures PrP C -interacting amyloid-β oligomers, has been described previously ( Kostylev et al. , 2015 ).
All results are presented as means ± standard error of the mean (SEM). IBM SPSS Statistics version 21 and Prism 6 software were used for statistical analysis. Data were analysed using one-way or two-way ANOVA, followed by post hoc Tukey’s multiple comparisons test or Fisher’s LSD post hoc pairwise comparisons test, as specified in the figure legends. In case of two column comparisons, Mann-Whitney test was used for analysis. Only two-sided tests were used and all data analysed met the assumption for the specific statistical test that was performed. Probability levels of P < 0.05 were considered as statistically significant.
mGluR5 couples cellular prion protein to intracellular protein mediators
If physiologically relevant brain PrP C /mGluR5 complexes exist, then PrP C is expected to associate via mGluR5 with a macromolecular assembly that includes multiple intracellular signalling partners. We screened multiple postsynaptic intracellular proteins for association with PrP C in acute brain slice crude synaptoneurosome preparations ( Fig. 1 A). Our data show that Homer1b/c, Pyk2, CamKII, and mGluR5 strongly and selectively associate with PrP C in acute brain slice crude synaptoneurosome preparations ( Fig. 1 ). Genetic deletion of Grm5 strongly reduced the co-immunoprecipitation signals of intracellular protein mediators in anti-PrP C immunoprecipitates ( Fig. 1 B–E). Interestingly, amyloid-β oligomer treatment of acute wild-type brain slices for 30 min dissociated Homer1b/c and Pyk2 from the PrP C association complex ( Fig. 1 B–D). Notably, amyloid-β oligomer stimulation enhanced the association between PrP C and CamKII as well as between PrP C and mGluR5 ( Fig. 1 B, E and F). Our data further show that association changes between the PrP C -mGluR5 complex and intracellular protein mediators is specifically triggered by the oligomeric form of amyloid-β, but not monomeric amyloid-β or amyloid-β fibrils ( Supplementary Fig. 1A–D ).
We also examined the acute amyloid-β oligomer-induced dissociation of Homer from the receptor complex histologically ( Supplementary Fig. 2 ). The co-localization of mGluR5 and Homer was extensive in cortical neurons with a punctate dendritic pattern, which largely overlapped with PrP C distribution. Exposure to amyloid-β oligomers significantly reduced Homer co-localization with mGluR5 in wild-type, but not in Prnp null neurons ( Supplementary Fig. 2 ). Thus, the proximity of Homer, mGluR5 and PrP C is disrupted by amyloid-β oligomers in a PrP C -dependent manner in neurons.
The synaptic structures in brain slices contain many components, therefore we sought to determine if the marked reduction of co-immunoprecipitation between PrP C and intracellular mediators in Grm5 −/− slices implies direct coupling of these molecules through mGluR5. We examined a simplified non-neuronal HEK-293T cell system expressing only the neuronal proteins in question to assess whether mGluR5 is not only necessary but also sufficient for complex formation between PrP C and intracellular signalling proteins. Cells co-expressing PrP C , Homer1c, Pyk2 and CamKIIα revealed minimal co-immunoprecipitation signals of Homer1c, Pyk2 and CamKIIα in anti-PrP C immunoprecipitates. This signal was strongly enhanced by co-expressing either Myc-mGluR5 or Myc-mGluR1 ( Fig. 2 A–D). We sought to recapitulate the amyloid-β oligomer-induced association changes in anti-PrP C immunoprecipitates we observed in acute mouse brain slices. Amyloid-β oligomer stimulation of HEK-293T cells co-expressing PrP C , Myc-mGluR5, Homer1c, Pyk2 and CamKIIα significantly reduced Homer1c and Pyk2 signals in anti-PrP C immunoprecipitates, similar to what we observed in acute brain slice crude synaptoneurosome preparations ( Fig. 2 A–C). In contrast, amyloid-β oligomer treatment enhanced the co-immunoprecipitation signal of CamKIIα in anti-PrP C immunoprecipitates ( Fig. 2 A and D). Amyloid-β oligomers failed to modulate the interaction between PrP C and Homer1c, Pyk2 or CamKIIα when Myc-mGluR1 was expressed instead of Myc-mGluR5, consistent with a selective role of mGluR5 related to Alzheimer’s disease ( Fig. 2 A–D). Notably, treatment of cells with amyloid-β oligomers for 30 min enhanced the association between PrP C and Myc-mGluR5 but not between PrP C and the closely related Myc-mGluR1 ( Fig. 2 A and E). We found similar results in the complementary anti-Myc immunoprecipitation. Amyloid-β oligomers were unable to modulate the interaction between Myc-mGluR5 and Homer1c, Pyk2 or CamKIIα in HEK-293T cells in the absence of PrP C ( Fig. 2 A and F–H). Moreover, amyloid-β oligomers were only able to selectively modulate the interaction between Myc-mGluR and Homer1c, Pyk2 or CamKIIα in cells co-expressing PrP C and Myc-mGluR5, but not PrP C and Myc-mGluR1 ( Fig. 2 A and F–H). Thus, mGluR5 links PrP C with a class of intracellular signalling proteins in an amyloid-β oligomer-regulated manner.
Having observed amyloid-β oligomer regulation of the physical association of PrP C with intracellular signalling proteins via mGluR5 in vitro , we assessed this phenomenon in the brain. For brain lysates from aged APP/PS1 + mouse model hippocampus with Alzheimer’s disease pathology, anti-PrP C immunoprecipitates showed significantly reduced Pyk2 and Homer1b/c signals compared to wild-type ( Fig. 3 A, D and F). Based on these positive findings from the mouse model, we examined evidence for altered coupling of PrP C with Homer1b/c and Pyk2 in human autopsy brain samples. We observed a similar reduction of Pyk2 and Homer1b/c signals in anti-PrP C immunoprecipitates from Alzheimer’s disease post-mortem frontal cortex compared to tissue from age-matched neurologically healthy controls or Parkinson’s disease subjects ( Fig. 3 B, G and H). Thus, the physical association of mGluR5-dependent signalling molecules with PrP C is regulated by amyloid-β pathology in mouse and human disease.
Genetic coupling between Prnp and Grm5 mediates acute alterations in protein activation states
Our study revealed that mGluR5 connects amyloid-β oligomer-PrP C to intracellular protein mediators. We sought to determine if there is a genetic interaction between Prnp and Grm5 with regard to amyloid-β oligomer-induced signalling within neurons. Previous studies proposed an important role of alterations in protein phosphorylation states in contributing to Alzheimer’s disease pathogenesis. We analysed whether amyloid-β oligomer stimulation of acute brain slices activates the specific mGluR5 signalling proteins found to be associated with PrP C . Thirty minutes amyloid-β oligomer stimulation of acute brain slices significantly increased phospho-Pyk2(Y402) and phospho-CamKII(T286) levels ( Fig. 4 ). We verified that activation of Pyk2 and CamKII is mediated by the oligomeric form of amyloid-β only, but not monomeric amyloid-β or amyloid-β fibrils ( Supplementary Fig. 1F–H ). We found that amyloid-β oligomer-induced activation of Pyk2(Y402) and CamKII(T286) was not significantly altered by removal of one allele of either Prnp or Grm5 ( Fig. 4 ). However, removal of one allele of both Prnp and Grm5 significantly reduced amyloid-β oligomer-induced phosphorylation changes (green bars compared to black bars; Fig. 4 ). Moreover, amyloid-β oligomer treatment failed to activate Pyk2 at Y402 and CamKII at T286 in Prnp−/− as well as in Grm5−/− acute brain slices (yellow and blue bars; Fig. 4 ). A two-way ANOVA with post hoc Tukey’s multiple comparisons test for Prnp and Grm5 was performed to verify a significant interaction between Prnp and Grm5 in triggering amyloid-β oligomer-induced aberrant phosphorylation states ( Fig. 4 ). Notably, the deletion of a single Prnp or Grm5 allele does not influence the reciprocal protein expression in heterozygote mouse brain ( Supplementary Fig. 3 ). Thus, the synthetic phenotype for amyloid-β oligomer signalling in the trans-heterozgous state demonstrates genetic interaction between PrP C and mGluR5 in Alzheimer’s disease-related signal transduction biochemistry.
Genetic coupling between Prnp and Grm5 mediates acute amyloid-β oligomer-induced impairment of synaptic plasticity
To provide further genetic evidence for the coupling between PrP C and mGluR5 in amyloid-β oligomer signalling, we investigated the effect of amyloid-β oligomers on LTP in the Schaffer collateral pathway between CA3 and CA1 neurons in acute brain slices. We found that amyloid-β oligomers significantly reduce the slope of the field EPSP in wild-type ( Fig. 5 A), Prnp+/− ( Fig. 5 B) and Grm5+/− ( Fig. 5 C) brain slices following theta burst stimulation. However, amyloid-β oligomers failed to inhibit LTP in Prnp+/−Grm5+/− brain slices ( Fig. 5 D). Comparing across genotypes without amyloid-β oligomer pretreatment, the extent of LTP was indistinguishable ( Fig. 5 E). In the presence of amyloid-β oligomers, removal of one allele of either Prnp or Grm5 did not significantly alter the magnitude of LTP ( Fig. 5 F). In contrast, deletion of one allele for both Prnp and Grm5 was able to significantly rescue amyloid-β oligomer-dependent inhibition of LTP seen in wild-type slices ( Fig. 5 F). A two-way ANOVA for Prnp and Grm5 confirmed a significant interaction between both genes in mediating amyloid-β oligomer-dependent deficits in LTP ( Fig. 5 G). We considered whether the lack of amyloid-β oligomer responsiveness in the double heterozygote state as due to altered baseline excitatory neurotransmission. We performed paired-pulse experiments at varying lengths to determine whether glutamate release probability was affected by Prnp or Grm5 heterozygosity ( Supplementary Fig. 4A ). Paired-pulse ratios approached 1.0 with increasing interval length, and did not vary significantly with respect to genotype. These results suggest glutamate signalling remains unchanged in single and double heterozygote Prnp and/or Grm5 brain slices. Furthermore, we measured input-output curves to determine baseline synaptic transmission in all genotypes. We recorded field EPSPs in the CA1 of the hippocampus of wild-type, Prnp +/− , Grm5+/− , and Prnp+/−Grm5+/− brain slices. Baseline synaptic transmission remained unchanged across genotypes ( Supplementary Fig. 4B ). Together these data demonstrate a genetic interaction between PrP C and mGluR5 in mediating acute amyloid-β oligomer-induced biochemical signalling and electrophysiological plasticity.
Genetic coupling between Prnp and Grm5 mediates alterations in protein activation states in Alzheimer’s mice
The preceding data show that mGluR5 physically and genetically couples PrP C to intracellular signalling proteins during acute exposure to amyloid-β oligomers. Next, we analysed the genetic coupling in a chronic Alzheimer’s disease model, APP/PS1 + transgenic mice. We have shown previously that acute amyloid-β oligomer treatment and TBS-soluble extracts from human Alzheimer’s disease patients, but not controls, trigger phosphorylation of eEF2 at T56 in cortical neurons ( Um et al. , 2013 ). Here we found that phospho-Pyk2(Y402) and phospho-eEF2(T56) levels were significantly enhanced in APP/PS1 + hippocampus (black bar compared to grey bar; Fig. 6 A, B and D). Enhanced phospho-Pyk2(Y402) and phospho-eEF2(T56) levels were not significantly altered by removal of one allele of either Prnp or Grm5 (black bar compared to red and purple bar; Fig. 6 A, B and D). Pyk2(Y402) and eEF2(T56) phosphorylation in APP/PS1 + hippocampus was significantly reduced by lowering the gene dosage of both Prnp and Grm5 as well as by full genetic deletion of either Prnp or Grm5 (green, yellow and blue bar compared to black bar; Fig. 6 A, B and D). While Pyk2 phosphorylation at Y402 was unaltered between wild-type and APP/PS1 + mouse cortex ( Fig. 6 C and Supplementary Fig. 5A ), eEF2 phosphorylation at T56 in APP/PS1 + mouse cortex was significantly enhanced compared to wild-type (black bar compared to grey bar; Fig. 6 E and Supplementary Fig. 5A ). Enhanced cortical phospho-eEF2(T56) levels were not significantly altered by removal of one allele of either Prnp or Grm5 (red and purple bar compared to black bar; Fig. 6 E and Supplementary Fig. 5A ). Phospho-eEF2(T56) levels in APP/PS1 + cortex were significantly reduced by lowering the gene dosage of both Prnp and Grm5 as well as by full genetic deletion of either Prnp or Grm5 (green, yellow and blue bar compared to black bar; Fig. 6 E and Supplementary Fig. 5A ). Thus, enhanced activation of Pyk2 and eEF2 in transgenic mouse brain is mediated by Prnp-Grm5 genetic interaction, being absent in transheterozygous or full knockout transgenic brain.
We further observed reduced phosphorylation levels of CamKII at T286 in hippocampus and cortex of APP/PS1 + mice compared to wild-type (black bar compared to grey bar; Fig. 6 A, F and G and Supplementary Fig. 5A ). Reduced phospho-CamKII(T286) levels in APP/PS1 + brain were significantly enhanced by genetic deletion of Prnp in hippocampus (yellow bar compared to black bar; Fig. 6 F and Supplementary Fig. 5A ) or genetic deletion of either Prnp or Grm5 in cortex (yellow and blue compared to black bar; Fig. 6 A and G). The cerebellum, a brain area largely unaffected in Alzheimer’s disease, did not reveal any changes in phosphorylation of Pyk2(Y402), eEF2(T56) or CamKII(T286) in APP/PS1 + mice compared to wild-type littermates ( Supplementary Fig. 5 ). Further, we did not observe changes in phosphorylation states of Pyk2(Y402), eEF2(T56) or CamKII(T286) in Prnp−/− and Grm5−/− brain in absence of the APP/PS1 + transgene ( Supplementary Fig. 6 ). Together, these phosphorylation studies provide genetic evidence that PrP C and mGluR5 cooperate in mediating altered neuronal signalling in selective Alzheimer’s disease transgenic brain areas.
Time dependence for CamKII activation after acute amyloid-β oligomer stimulation
Chronic exposure to the APP/PS1 + transgene revealed alterations of Pyk2 phosphorylation at Y402 that were similar to acute amyloid-β oligomer treatment. In contrast, we observed that amyloid-β oligomer triggered enhanced CamKII phosphorylation at T286 by acute amyloid-β oligomer treatment but reduced CamKII phosphorylation at T286 after chronic exposure to the APP/PS1 + transgene. To investigate differences in CamKII activation between short-term amyloid-β oligomer stimulation and chronic exposure to amyloid-β oligomers, we compared phospho-CamKII(T286) levels in acute wild-type brain slices exposed to amyloid-β oligomers for different periods of time. As amyloid-β oligomers are unstable in artificial CSF due to spontaneous fibrilization ( Supplementary Fig. 7A ), acute brain slices were exposed to a fresh bath of amyloid-β oligomers after every hour. We observed a significant interaction of CamKII phosphorylation at T286 between amyloid-β oligomer treatment and duration of treatment by two-way ANOVA with post hoc Tukey’s multiple comparisons test ( Supplementary Fig. 8A and B ). Treatment of slices with 1 µM amyloid-β oligomers for 0.5 h triggered an initial increase in CamKII phosphorylation at T286 ( Supplementary Fig. 8A and B ). Phospho-CamKII(T286) levels were significantly reduced 3 h and 6 h after treatment ( Supplementary Fig. 8A and B ).
Activation of Pyk2 at Y402 in acute brain slices, on the other hand, was consistently enhanced by treatment with amyloid-β oligomers from 0.5–6 h ( Supplementary Fig. 8A and C ). We sought to recapitulate activation of eEF2 as seen in the chronic APP/PS1 + Alzheimer’s disease model by amyloid-β oligomer treatment of acute brain slices. Amyloid-β oligomers triggered enhanced eEF2 phosphorylation at T56 6 h after stimulation ( Supplementary Fig. 8A and D ).
We used immunostaining for phospho-CamKII(T286) of dissociated 21 day in vitro primary cortical neurons as a second approach to analyse the time-course of CamKII activation at T286 induced by amyloid-β oligomer stimulation. As amyloid-β oligomers are stable in Neurobasal® A medium at 37°C and 5% CO 2 ( Supplementary Fig. 7B ), cortical neurons were exposed to 1 µM amyloid-β oligomers and incubated for indicated times. We found a significant interaction between treatment and duration of treatment by two-way ANOVA with post hoc Tukey’s multiple comparisons test ( Supplementary Fig. 8E and F ), similar to what we observed by amyloid-β oligomer stimulation of acute brain slices. Amyloid-β oligomers trigger an initial increase of phospho-CamKII(T286) intensity, which reverts to vehicle levels at 3 h of treatment and declines below vehicle levels 6 h after treatment ( Supplementary Fig. 8E and F ). Together, these experiments show that short-term exposure to amyloid-β oligomers activates CamKII at T286, but long-term amyloid-β oligomer treatment or chronic exposure to Alzheimer’s disease transgenes deactivates CamKII at T286.
Genetic coupling between Prnp and Grm5 mediates reduced survival in Alzheimer’s disease model mice
We questioned whether more complex phenotypes are also dependent on the genetic coupling between PrP C and mGluR5 in APP/PS1 + Alzheimer’s model mice. Mouse survival of wild-type, APP/PS1 + , APP/PS1 +Prnp+/− , APP/PS1 +Grm5+/− , APP/PS1 +Prnp+/−Grm5+/− , APP/PS1 +Prnp−/− and APP/PS1 +Grm5−/− mice was monitored over 600 days. After 600 days, the majority of APP/PS1 + mice died, compared to only 8% of wild-type mice (black line compared to grey line; Fig. 7 ). Full genetic deletion of either Prnp or Grm5 significantly improved mouse survival in APP/PS1 + mice (blue and yellow line compared to black line; loss of 7% and 17% after 600 days; Fig. 7 ). Removal of one allele of either Prnp or Grm5 did not benefit mouse survival (red line and purple line compared to black line; Fig. 7 ). Genetic removal of one allele of both Prnp and Grm5 significantly improved survival of APP/PS1 + mice (green line compared to black line; Fig. 7 ). We conclude that reduced survival of APP/PS1 + mice depends on the genetic interaction of PrP C and mGluR5.
Genetic coupling between Prnp and Grm5 mediates loss of synaptic markers in Alzheimer’s disease model brain in vivo
Synapse loss is one of the hallmarks of Alzheimer’s disease that is observable in Alzheimer’s disease transgenic mouse models ( Scheff et al. , 1990 ; Jacobsen et al. , 2006 ). Previous studies revealed that synapse loss of APP/PS1 + mice is dependent on the genetic presence of Prnp ( Gimbel et al. , 2010 ) and can be recovered by mGluR5 antagonism ( Um et al. , 2013 ). We sought to investigate whether or not synapse density in the dentate gyrus is mediated by a genetic coupling between Prnp and Grm5. Loss of synaptic markers of APP/PS1 + mice (black bar compared to grey bar; Fig. 8 ) was fully rescued to wild-type levels by genetic deletion of Prnp or Grm5 (yellow and blue bar compared to grey bar; Fig. 8 ). Notably, removal of one allele of either Prnp or Grm5 did not benefit synapse loss in APP/PS1 + brain (red and purple bar compared to black bar; Fig. 8 ). However, removal of one allele of both Prnp and Grm5 fully rescued loss of synaptic markers to wild-type levels (no significant difference between grey and green bar; Fig. 8 ). Thus, synapse loss in APP/PS1 + mice is mediated through an interaction between Prnp and Grm5.
Amyloid-β levels and enhanced gliosis are independent of the genetic presence of Prnp and Grm5
PrP C and mGluR5 are predicted to function as a complex mediating synaptic disruption in neurons downstream of amyloid-β oligomers, but not to alter amyloid-β accumulation or the responses of other cell types to amyloid-β pathology. We compared amyloid-β levels in wild-type, APP/PS1 + , APP/PS1 +Prnp+/− , APP/PS1 +Grm5+/− , APP/PS1 +Prnp+/−Grm5+/− , APP/PS1 +Prnp−/− and APP/PS1 +Grm5−/− brain. We observed highly enhanced amyloid-β levels by amyloid-β immunostaining as well as amyloid-β dense-core plaques by thioflavin-S staining in APP/PS1 + brain compared to wild-type brain (black bar compared to grey bar; Supplementary Fig. 9A–C ). We further analysed amyloid-β levels by amyloid-β ELISA and amyloid-β oligomer-dependent PrP C -linked immunosorbent assay and, again, found enhanced amyloid-β levels in APP/PS1 + brain (black squares compared to grey squares; Supplementary Fig. 9D and E ). However, we did not find differences in amyloid-β levels within APP/PS1 + groups in any of these measurements (black bars/squares compared to red, purple, green, yellow and blue bars/squares; Supplementary Fig. 9 ). Thus, the PrP C /mGluR5 complex does not regulate amyloid-β levels.
Alzheimer’s disease pathology is associated with inflammation, as marked by activated microglia and astrocytes, in areas severely affected with amyloid-β plaque deposition ( Itagaki et al. , 1989 ). We analysed gliosis by immunostaining of astrocytes (GFAP) and microglia (Iba1). Our data revealed an increase in gliosis in APP/PS1 + brain compared to wild-type (black bar compared to grey bar; Supplementary Fig. 10A–C ). However, there was no difference within APP/PS1 + groups (black bar compared to red, purple, green, yellow or blue bar; Supplementary Fig. 10A–C ). Moreover, we observed enhanced expression of anti-interleukin-1 beta in the hippocampus and cortex of APP/PS1 + mice compared to wild-type by immunoblot (black bar compared to grey bar; Supplementary Fig. 10D–F ). Again, we did not observe any differences in interleukin-1 beta expression within APP/PS1 + groups (black bar compared to red, purple, green, yellow or blue bar; Supplementary Fig. 10D–G ). Thus, Prnp-Grm5 genetic interaction does not alter the chronic response to amyloid-β accumulation in cell types other than neurons.
The major finding of the current study is that PrP C and mGluR5 function as a neuronal complex to mediate amyloid-β oligomer action and Alzheimer’s disease-related phenotypes. While previous studies have implicated both proteins in Alzheimer’s disease pathophysiology and revealed co-immunoprecipitation between them, the genetic and biochemical data here reveal an essential role for their coupling in Alzheimer’s disease-related models. Moreover, the phosphorylation state of these intracellular signalling partners of PrP C is specifically regulated by amyloid-β oligomers and Alzheimer’s disease transgenes through the genetic coupling of PrP C and mGluR5, such that transheterozygotes, but not single heterozygotes, are rescued from amyloid-β oligomer signalling. Chronic effects on mouse survival and loss of synaptic profiles also depend upon the genetic interaction between PrP C and mGluR5.
These data may also shed light on the potential physiological role of PrP C . Despite enormous advances in understanding the role of prion protein in prion disease, the function of PrP C remains elusive. Genetic deletion of Prnp in mice does not elicit a direct phenotype ( Bueler et al. , 1993 ). Thus, understanding the endogenous function of PrP C remains a priority. This can be achieved by identifying protein partners that interact with PrP C . We are the first to show that intracellular signalling partners associate with PrP C via mGluR5. Strikingly, PrP C and mGluR5 function as a neuronal complex to mediate amyloid-β oligomer pathophysiology in mouse and human brain. Despite this role in the disease state, glutamate-dependent signalling is not altered by the presence of PrP C under normal physiological conditions ( Um et al. , 2013 ). It is possible that unidentified physiological PrP C partners control mGluR5 function in a manner usurped by pathological amyloid-β oligomers in Alzheimer’s disease. The current findings suggest that mGluR5-dependent assays may serve as a tool to understand physiological PrP C functions.
mGluR5 couples PrP C to intracellular protein mediators
We report that PrP C is linked by mGluR5 to the intracellular protein mediators Homer1b/c, Pyk2 and CamKII via mGluR5. Co-immunoprecipitation between PrP C and intracellular proteins requires co-expression of either Myc-mGluR5 or Myc-mGluR1 in HEK-293T cells. In acute brain slice crude synaptoneurosome fractions, co-immunoprecipitation is essentially eliminated by genetic deletion of Grm5. This indicates that GPI (glycosyl-phosphatidyl-inositol)-anchored PrP C coupling to the here-analysed protein mediators within the cell requires mGluR5, or mGluR1. We further show that amyloid-β oligomers modulated the coupling between PrP C and intracellular protein mediators in HEK-293T cells and acute mouse brain slices. Notably, amyloid-β oligomers were only capable of altering the interaction between mGluR5 and intracellular protein mediators in HEK-293T cells and primary hippocampal neurons in the presence of PrP C . These results provide further evidence for PrP C as a functional cell-surface receptor site of amyloid-β oligomers. Interestingly, we found a pathologically reduced association between PrP C and Pyk2 as well as between PrP C and Homer1b/c both in Alzheimer’s disease mouse model brain and human post-mortem Alzheimer’s disease brain compared to age-matched controls. Similarly, a pathologically reduced association between mGluRs and Homer has been demonstrated in several neurological diseases, including fragile X syndrome and Angelman syndrome ( Giuffrida et al. , 2005 ; Pignatelli et al. , 2014 ). This pathological situation is predicted to prevent physiologically fluctuating levels of glutamate to regulate the interaction between mGluR5 and intracellular signalling proteins.
The simplest model for amyloid-β oligomer signalling is that Alzheimer’s disease-derived oligomers induce a conformational change in their PrP C binding site, which is transmitted via protein-protein interaction to mGluR5, and subsequently to its intracellular mediators. In addition, higher order mGluR5 protein assembly states and lateral mobility might contribute to signal transduction. It has been demonstrated that amyloid-β oligomers trigger clustering of mGluR5 in primary hippocampal neurons under conditions in which anti-PrP C antibodies reduce neuronal binding ( Renner et al. , 2010 ). Under physiological conditions, the stabilization of synaptic proteins within the plasma membrane is transient, with constant lateral movement of receptors such as mGluR5 ( Triller and Choquet, 2008 ). The presence of the amyloid-β oligomer ligand appears to generate a pathophysiological state where mGluR5 is scaffolded by the stabilized amyloid-β oligomer-PrP C complex. Prevention of the lateral movement of mGluR5 within the plasma membrane is likely to alter changes within the mGluR5 multiprotein complex that normally occur in response to glutamate stimulation. For example, amyloid-β oligomer-induced overstabilization of mGluR5 may alter intracellular Ca 2+ levels ( Renner et al. , 2010 ). Both conformational changes within the mGluR5 multiprotein complex and altered Ca 2+ levels may drive the amyloid-β oligomer-induced acute activation of Pyk2 and CamKII observed here.
PrP C interaction with mGluR5 has been observed in several situations in addition to our previous analysis of PrP C coupling to FYN kinase ( Um et al. , 2013 ). Amyloid-β oligomer-induced long-term depression (LTD) is dependent on both mGluR5 and PrP C ( Hu et al. , 2014 ). Also, amyloid-β oligomer-induced redistribution of CamKIIα into cytoplasmic puncta is dependent on the co-expression of mGluR5 with PrP C ( Raka et al. , 2015 ). Furthermore, studies of laminin signalling after PrP C binding demonstrate mGluR5 mediated increases of intracellular Ca 2+ levels ( Beraldo et al. , 2011 ). Thus, mGluR5 seems to transduce different signals from PrP C onto intracellular signalling mediators.
Association between mGluR5 and intracellular protein mediators
In the current study, PrP C is associated via mGluR5 with Homer1b/c, Pyk2 and CamKII. Previous work confirmed a physical interaction between the postsynaptic scaffolding protein Homer and the C-terminal proline-rich sequence of mGluRs ( Brakeman et al. , 1997 ). It has been shown that stimulation by agonist quisqualate enhances co-immunoprecipitation between Homer and class I mGluRs, whereas mGluR antagonism by MPEP reduces association ( Rong et al. , 2003 ). Amyloid-β treatment of cortical neurons is reported to decrease synaptic levels of Homer1b ( Roselli et al. , 2009 ). Here we show that amyloid-β oligomer treatment triggers dissociation of Homer1b/c from the PrP C -mGluR5 complex. Further, chronic exposure to amyloid-β oligomers in humans or mice causes a pathological state, where Homer1b/c is less associated with the PrP C -mGluR5 complex. We hypothesize that in this state, glutamate-dependent activation of the receptor is less capable of stimulating physiological changes in Homer1b/c. It has been shown previously that disruption of the interaction between mGluR5 and Homer prevents mGluR-LTD as well as mGluR-induced signalling cascades ( Ronesi and Huber, 2008 ). Thus, the pathological effect of amyloid-β oligomers could be due to blocking signalling events that depend on association changes between mGluR5 and intracellular mediators like Homer1b/c.
Other studies reported the association between CamKII and the intracellular C-terminal tail as well as the intracellular loop 2 of mGluR5 ( Jin et al. , 2013 a , b ; Raka et al. , 2015 ). Ca 2+ stimulation triggers dissociation of CamKIIα from mGluR5 in rat striatal neurons, which is hypothesized to facilitate LTP induction ( Jin et al. , 2013 b ). In contrast, amyloid-β oligomer treatment triggers enhanced association between CamKIIα and mGluR5 ( Raka et al. , 2015 ). Thus, the potent inhibition of LTP by amyloid-β oligomers may depend in part on CamKII being unable to dissociate from mGluR5 in the presence of amyloid-β oligomers. Our data further revealed acute short-term amyloid-β oligomer treatment to phosphorylate CamKII at T286, but chronic exposure to the APP/PS1 + transgene to dephosphorylate CamKII at T286. To address this difference, we analysed phospho-CamKII(T286) levels after varying times of amyloid-β oligomer treatment in acute brain slices and primary cortical cultures. We found that phospho-CamKII(T286) levels were initially increased by amyloid-β oligomers, returned to basal levels at 3 h of treatment and slightly declined below basal at 6 h of amyloid-β oligomer treatment. Similar to what we observed with acute amyloid-β oligomer treatment, glutamate stimulation of cortical neurons induces rapid phosphorylation of CamKII, which returns to basal levels 1.5 h after washout of glutamate, followed by a slight decline below basal levels 2.5 h after washout ( Hosokawa et al. , 2006 ). Under physiological conditions, activated phospho(T286)-CamKII dissociates from the mGluR5 complex. Phospho(T286)-CamKII then associates with and phosphorylates NMDAR GluN2B subunits ( Jin et al. , 2013 b ). We hypothesize that this physiological event is unable to occur in the presence of excess amyloid-β oligomers, which affects synaptic plasticity.
Further studies showed association between mGluR5 and Pyk2 in rat brain as well as co-localization of mGluR5 and Pyk2 in cortical neurons ( Nicodemo et al. , 2010 ). Pyk2 is known to dissociate from mGluR1 after treatment with quisqualate in HEK-293T cells ( Nicodemo et al. , 2010 ), similar to what we observed in HEK-293T cells and acute mouse brain slices after amyloid-β oligomer treatment.
Genetic coupling between Prnp and Grm5 alters protein activation, plasticity, synaptic density and survival
We found that the genetic interaction between Prnp and Grm5 is significantly involved in mediating amyloid-β oligomer-induced phosphorylation of intracellular proteins and inhibition of LTP acutely. Our data further confirmed a genetic coupling between Prnp and Grm5 in mediating alterations in protein activation states selectively in APP/PS1 + transgenic hippocampus and cortex. Notably, a selective vulnerability for brain areas such as neocortex and hippocampus towards Alzheimer’s disease pathogenesis exists ( Morrison and Hof, 1997 ). Our results show that intracellular protein mediators are selectively altered in hippocampus and cortex of Alzheimer’s disease transgenic model mice compared to wild-type. All members of the here-studied signalling cascade are present in the cerebellum. Yet, no activation changes of intracellular protein mediators were observed in transgenic cerebellum compared to wild-type. We speculate that regional differences in the interaction between proteins of the here-studied signalling pathway contribute to differences in amyloid-β oligomer-dependent protein activation states. This regional vulnerability to amyloid-β oligomers might make hippocampal and cortical neurons more prone to degeneration compared to cerebellar neurons during Alzheimer’s disease progression.
Our previous work and other studies demonstrated amyloid-β oligomer/PrP C -mediated activation of the non-receptor tyrosine kinase FYN ( Larson et al. , 2012 ; Um et al. , 2012 ). Furthermore, mGluR5 is implicated in this pathway ( Um et al. , 2013 ). FYN phosphorylation of NR2A and NR2B subunits alters trafficking of NMDA receptors, which regulates synaptic plasticity ( Grant et al. , 1992 ; Suzuki and Okumura-Noji, 1995 ; Nakazawa et al. , 2001 ). Critically, amyloid-β oligomer binding to PrP C also triggers tau phosphorylation ( Larson et al. , 2012 ). Evidence exists for a genetic interaction between FYN and tau to regulate synapse density and behaviour in APP transgenic mice ( Chin et al. , 2004 , 2005 ; Roberson et al. , 2011 ). Notably, amyloid-β oligomer-induced phosphorylation of eEF2 in cortical neurons is dependent on FYN signalling and the kinase inhibitor sarcatinib fully blocks amyloid-β oligomer-induced eEF2 phosphorylation ( Um et al. , 2013 ). The data indicate that the amyloid-β oligomer/PrP C /mGluR5 complex activates different signalling pathways, some of which are dependent on FYN kinase activation. Moreover, eEF2 signalling occurs downstream of FYN signalling and is dependent on the genetic presence of Prnp and Grm5 ( Um et al. , 2013 ). Based on the findings of this manuscript, we hypothesize that FYN activation is also dependent on the genetic interaction between Prnp and Grm5. It is important to note that acute activation of FYN via amyloid-β oligomer/PrP C /mGluR5 leads to compensatory changes including increased STEP phosphatase ( PTPN5 ), that return phospho-FYN levels to baseline ( Um et al. , 2012 ), and prevent assessment of FYN during chronic activation in vivo. The relative roles and interplay of FYN, eEF2, Pyk2, CamKII and calcium levels as mediators of the pathophysiological amyloid-β oligomer/PrP C /mGluR5 complex will require future studies.
Critically, we also found that amyloid-β oligomer-PrP C coupling to mGluR5 is responsible for reduced survival and synapse loss in APP/PS1 + mice compared to wild-type. Notably, these phenotypes are recovered without altered gliosis, β-amyloid plaque load or amyloid-β oligomer levels. This suggests that the recovery in APP/PS1 +Prnp+/−Grm5+/− mice is fully dependent on targeting amyloid-β oligomer-induced alterations in neuronal signalling pathways related to Alzheimer’s disease in these mice. Genetic removal of Prnp or Grm5 , as well as the double heterozygote state, is sufficient to rescue disease progression in APP/PS1 + mice without altered amyloid-β accumulation or gliosis. Improvements occur in response to removal of the amyloid-β oligomer cell-surface anchor PrP C ( Lauren et al. , 2009 ; Gimbel et al. , 2010 ), the transmembrane signal transducer mGluR5 ( Um et al. , 2013 ), or a reduced interaction between both proteins in the current study.
Targeting the PrP C -mGluR5 interaction for Alzheimer’s disease therapy
The current work verified a significant interaction between Prnp and Grm5 in mediating amyloid-β oligomer-induced phenotypes and provides further evidence for the amyloid-β oligomer-PrP C -mGluR5 pathway in Alzheimer’s disease pathophysiology. The data show that improvements of Alzheimer’s disease-like phenotypes in APP/PS1 + mice occurred despite continued high levels of amyloid-β, amyloid-β oligomers, and amyloid-β plaques. We hypothesize that targeting the PrP C -mGluR5 interaction in Alzheimer’s disease will ameliorate disease progression by reducing the signal transduction from extracellular amyloid-β oligomers onto intracellular protein mediators, a mechanism distinct from amyloid-β lowering agents. We have shown previously that PrP C -directed antibodies and mGluR5-directed compounds can potently modulate the interaction between PrP C and mGluR5 ( Haas et al. , 2014 ). Additional studies will further clarify the efficacy of these molecules to ameliorate Alzheimer’s disease phenotypes without negatively affecting endogenous signalling.
- APP/PS1 +
amyloid precursor protein/presenilin 1
calcium/calmodulin-dependent protein kinase II
eukaryotic elongation factor 2
excitatory postsynaptic potential
human embryonic kidney-293T
homer scaffolding protein 1 splice variants a/b/c
metabotropic glutamate receptor 5
- PrP C
cellular prion protein
proline rich tyrosine kinase 2
We thank Jacqueline Heiss and Stefano Sodi for assistance with mouse husbandry.
This work was supported by grants from NIH, BrightFocus Foundation, Alzheimer’s Association and Falk Medical Research Trust to S.M.S., and the German National Academic Foundation supports L.T.H.
S.M.S. is a co-founder of Axerion Therapeutics seeking to develop PrP-based therapeutics for Alzheimer disease.
Supplementary material is available at Brain online.