mGlu5 receptors and cellular prion protein mediate amyloid-β-facilitated synaptic long-term depression in vivo

NMDA-type glutamate receptors (NMDARs) are currently regarded as paramount in the potent and selective disruption of synaptic plasticity by Alzheimer’s disease amyloid β-protein (Aβ). Non-NMDAR mechanisms remain relatively unexplored. Here we describe how Aβ facilitates NMDAR-independent long-term depression of synaptic transmission in the hippocampus in vivo. Synthetic Aβ and Aβ in soluble extracts of Alzheimer’s disease brain usurp endogenous acetylcholine muscarinic receptor-dependent long-term depression, to enable long-term depression that required metabotropic glutamate-5 receptors (mGlu5Rs). We also find that mGlu5Rs are essential for Aβ-mediated inhibition of NMDAR-dependent long-term potentiation in vivo. Blocking Aβ binding to cellular prion protein with antibodies prevents the facilitation of long-term depression. Our findings uncover an overarching role for Aβ-PrPC-mGlu5R interplay in mediating both LTD facilitation and LTP inhibition, encompassing NMDAR-mediated processes that were previously considered primary.

I ncreasing our understanding of how amyloid-b protein (Ab) causes synaptic dysfunction should provide new means of therapeutically targeting early Alzheimer's disease (AD) 1 . It is now well established that Ab has rapid, profound and selective disruptive effects on synaptic plasticity of excitatory synaptic transmission in vulnerable brain regions, including the hippocampus 2 . In addition to causing strong inhibition of longterm potentiation (LTP), Ab has been reported to enhance longterm depression (LTD). Most research has focused on the actions of Ab on forms of LTP and LTD that require NMDA-type glutamate receptors (NMDARs) [3][4][5][6] . Indeed, as NMDARdependent LTP is likely to underlie synaptic memory mechanisms 7 , the inhibition of this form of LTP by Ab is highly congruent with the ability of Ab to impair learning and memory 8,9 . Somewhat similarly, excessive enhancement of LTD that requires NMDARs can cause memory retrieval deficits 10,11 . Remarkably, the disruption of NMDAR-dependent synaptic plasticity by Ab is itself mediated through NMDARs, in particular, those containing the GluN2B subunit [12][13][14][15] .
In contrast, little is known about how Ab affects forms of synaptic plasticity that do not require NMDARs. Whereas Ab potently inhibits acetylcholine-induced LTP 16 , NMDARindependent LTP induced by strong high-frequency conditioning stimulation (HFS) appears to be resistant to disturbance by Ab 17 in the hippocampus in vitro. Recently, Ab was reported to enable an NMDAR-independent LTD that was blocked by metabotropic glutamate-5 receptor (mGlu5R) antagonists in hippocampal slices 5,8 . Indeed, synaptically evoked activation of mGlu5R or other similar G-protein coupled receptors including M1 muscarinic acetylcholine receptors (mAChRs) can induce LTD that does not require NMDARs 11,[18][19][20] . Moreover, mAChRdependent LTD has been proposed to underlie visual recognition memory in the perirhinal cortex 21 and to provide a neurophysiological basis for preserved memory function in the ageing hippocampus 22 . Considering the early vulnerability of cholinergic pathways and related signalling in AD 23,24 , we hypothesize that Ab would inhibit mAChR-dependent LTD.
Remarkably, in vivo exposure to low-dose Ab facilitated an NMDAR-independent form of LTD but does not appear to affect mAChR-dependent LTD. This Ab-facilitated LTD is found to be mGlu5R-dependent. Moreover, Ab-mediated inhibition of LTP is also dependent on metabotropic glutamate-5 receptors (mGlu5Rs), indicating a key overarching role of this glutamate receptor subtype. We also discover that cellular PrP, a receptor for certain synaptotoxic Ab assemblies 25,26 , is necessary for Ab to facilitate LTD. These data are strongly congruent with recent molecular evidence that Ab and cellular prion protein (PrP C ) form a complex with mGlu5R at the postsynaptic density 27 and thereby disrupt synaptic plasticity.

Results
In vivo induction of mAChR-dependent LTD. In order to study the effects of Ab on mAChR-dependent LTD in vivo, we In four animals, two stimulation electrodes (S1, black and S2, purple) were implanted in different locations in the stratum radiatum to allow independent activation of the Schaffer collateral-commissural pathway. One hour after stable baseline recording from both S1 and S2 pathways, application of LFS-900 to S1 induced LTD in the S1 pathway but not in the S2 pathway. Conversely, 1 h after the first LFS, a second LFS was applied to S2 pathway that only induced LTD in pathway S2. As summarized in (d) One hour after application of LFS1, the EPSP in pathway S1 decreased to 67.4 ± 5.6% (Po0.05 compared with Pre; one-way ANOVA-Tukey) but did not change significantly in the S2 pathway (103.6±4.5%, P40.05 compared with Pre; Po0.05 compared with S1 pathway; t-test). developed a novel induction protocol that makes use of the reported requirement for high-intensity pulses to ensure robust synaptic ACh release during low-frequency conditioning stimulation (LFS) in the neocortex 28 . We found that application of strong LFS, consisting of 900 high-intensity pulses at 1 Hz (LFS-900), in the stratum radiatum of anaesthetized rats triggered synaptic LTD that (i) was stable for B3 h (Fig. 1a,b), (ii) was readily reversible by HFS (Fig. 1a,b) and (iii) was input specific (Fig. 1c,d). Consistent with the essential requirement for activation of cholinergic mechanisms in the induction of this form of LTD, LFS-900 failed to induce LTD of synaptic transmission after pretreatment with the mAChR antagonist scopolamine (Fig. 2a,b). In contrast, the LTD was not dependent on the activation of nicotinic AChRs, the magnitude of LTD being unaffected by injection of the nicotinic AChR antagonist mecamylamine before LFS-900 ( Fig. 2a,b). Consistent with a role for the M1 subtype of mAChR in LTD induction 19 , the M1-selective antagonist pirenzepine significantly reduced the magnitude of LTD (Fig. 2c,d). mAChR activation did not appear to be required for LTD maintenance/expression, as injection of scopolamine after the conditioning stimulation, using the same dose that completely prevented LTD induction, did not significantly affect the magnitude of LTD (Fig. 2e,f). Further evidence that physiological release of ACh is a key factor in LTD induction in vivo was the ability of an agent that enhances the effects of endogenously released ACh, the acetylcholinesterase inhibitor donepezil, to lower the threshold of LTD induction. Thus, we found that a relatively weak LFS conditioning protocol, consisting of 300 high-intensity pulses at 1 Hz (LFS-300) that was at or just below the threshold to induce significant LTD in vehicle-pretreated animals, triggered a large and robust LTD that was stable for at least 3 h in animals pretreated with donepezil ( Fig. 2g,h). Moreover, as described below, the induction of this in vivo synaptically evoked mAChR-dependent LTD did not require the activation of NMDA or mGlu5Rs. Because Ab can interfere with mAChR-related signalling 29 , we went on to examine the ability of Ab to disrupt this form of LTD.
Ab enhances an mAChR-independent form of LTD. We investigated the effects of Ab on synaptically evoked mAChRdependent LTD in vivo by the injection of Ab into the lateral cerebral ventricle via a cannula. Initially, we used a soluble synthetic Ab 1-42 preparation that had been centrifuged to remove any fibril aggregates. We chose a dose (160 pmol) of soluble Ab 1-42 that did not affect baseline synaptic transmission but strongly inhibited NMDAR-dependent LTP, as described below and previously 30 . To our surprise, in animals pre-injected with soluble Ab 1-42 the application of LFS-900 triggered an LTD that was more stable than the control LTD induced in the absence of Ab. Thus, LTD induced in the presence of Ab was stable during the 5-h recording period, whereas control LTD decayed significantly between 3 and 5 h post LFS (Fig. 3a,b). Although we had hypothesized that mAChR-dependent LTD would be inhibited by Ab, we wondered whether this Ab-facilitated LTD required mAChRs. We therefore pretreated the rats with scopolamine before Ab. In contrast to control LTD, which was completely abrogated by the mAChR antagonist (Fig. 2a,b), the time course and magnitude of LTD was only partly reduced by scopolamine in Ab-treated animals (Fig. 3a,b). These findings indicate that Ab had enabled an additional LTD that was more stable and independent of mAChRs while at the same time leaving a residual mAChR-dependent LTD relatively unscathed.
We wondered whether this Ab-facilitated additional, mAChRindependent, LTD was due to the ability of Ab to lower the threshold for LTD induction in vivo. We therefore used the weak LFS conditioning protocol (LFS-300). In addition to our standard soluble Ab 1-42 preparation we also tested a preparation of soluble Ab 1-42 enriched with protofibrils ( Fig. 4). We combined the results obtained with the two synthetic Ab 1-42 preparations because there was no quantitative difference in their effects on LTD. The application of weak LFS-300 induced a large and robust   LTD that was stable for at least 3 h in animals injected with Ab 1-42 (Fig. 5a,b), but not vehicle or a control, reverse sequence peptide Ab 42-1 (Fig. 5a,b). This dose (160 pmol) of Ab 1-42 did not affect baseline synaptic transmission (Fig. 5a,b) and consistent with a relatively selective action of Ab on the mechanisms underlying LTD induction, the same dose applied after the LFS-300 conditioning stimulation failed to facilitate LTD (Fig. 5c,d).
Moreover, the LTD induced by weak LFS-300 in the presence of Ab, like the additional LTD induced by the strong LFS-900 protocol, was also mAChR-independent, not being blocked by scopolamine pretreatment (Fig. 5e,f). Although synthetic Ab is most commonly used in studies of synaptic plasticity disruption, it is important to determine whether similar effects are caused by natural Ab. The presence of water-soluble SDS-stable Ab dimer in post-mortem brain extracts is highly correlated with ante-mortem dementia status 31 and such Ab can inhibit LTP and promote LTD in vitro 5,8 . It is therefore of great interest to assess whether AD brain-derived Ab can also facilitate LTD induction in vivo. Consequently, we tested the ability of Ab in water-soluble extracts of two different AD brains to mimic the ability of synthetic Ab 1-42 to lower the threshold for LTD induction in vivo. As can be seen from the western blot of one of the AD brain extracts (Fig. 6a), Ab runs on SDS gel predominantly as either monomer or dimer. These water-soluble SDS-stable species include a wide distribution of assemblies when analysed by size exclusion chromatography (SEC), ranging from monomer to Z70 kDa (ref. 8). Similar to synthetic Ab, the injection of Ab-containing AD brain soluble extract enabled the induction of robust and stable LTD by LFS-300 (Fig. 6b,c). Importantly, immunodepletion of Ab from the AD brain sample abrogated its ability to enable LTD induction. This finding indicates that soluble Ab is responsible for the lowering of the LTD induction threshold by the AD brain extract. Which SDS-stable Ab assembly is responsible for the facilitation of LTD by AD TBS brain extract remains to be determined.
Ab-facilitated LTD is NMDAR-independent. Because Ab has been reported to promote NMDAR-dependent LTD 5,6 , we The application of weak LFS (bar, LFS-300; 300 high-intensity pulses at 1 Hz) triggered a robust and stable LTD after acute i.c.v. injection (hash) of 160 pmol Ab 1-42 but not vehicle or the reverse peptide Ab 42-1 (Ab reverse). This dose of Ab 1-42 did not affect baseline synaptic transmission in the absence of LFS-300 (Ab no LFS). Data for soluble and protofibril Ab are combined and some animals had an additional separate i.c.v. injection of 5 ml vehicle 15 min before Ab. As summarized in (b) at 3 h the EPSP measured 93.3 ± 3.6% in controls (n ¼ 8, P40.05 compared with Pre; paired t), 69.9 ± 3.8% in Ab-injected rats (n ¼ 10, Po0.05 compared with Pre and vehicle group; paired t and one-way ANOVA-Tukey) and 97.3±4.3% in reverse peptide (n ¼ 4, P40.05 compared with Pre). Injection of Ab 1-42 (160 pmol, i.c.v.) did not affect baseline synaptic transmission (102.6 ± 1.6% at 3 h, n ¼ 7, P40.05 compared with Pre). ARTICLE postulated that activation of NMDARs in the presence of Ab may bypass the need for mAChRs in the induction of LTD in vivo. Contrary to our prediction, the NMDAR antagonist CPP, at a dose (10 mg kg À 1 , intraperitoneal (i.p.)) that completely blocks HFS-induced LTP 32 , did not affect the induction of LTD by LFS-300 in the presence of soluble Ab 1-42 (Fig. 7a,b). As CPP is a competitive antagonist and NMDARs containing GluN2B subunits are particularly implicated in Ab-mediated synaptic plasticity disruption [12][13][14][15] , we also tested the GluN2B-selective negative allosteric modulator Ro 25-6981 (ref. 33). Using a dose (6 mg kg À 1 , i.p.) that prevents Ab-mediated inhibition of LTP 12 , Ro 25-6981 had no effect on the facilitation of LTD by Ab (Fig. 7c,d). We concluded that like control LTD induced by LFS-900 (Fig. 7e,f), Ab-facilitated LTD induced by LFS-300 is NMDAR-independent.
HFS-induced de-depression of LTD is NMDAR-dependent. In the light of the contrasting findings regarding the involvement of NMDARs in the disruptive effects of Ab on LFS-induced LTD (present study) and HFS-induced LTP 12-15 , we also examined the effect of Ab on another form of synaptic plasticity, HFS-induced de-depression. De-depression is the persistent reversal of LTD by conditioning stimulation and is believed to be an essential component of bidirectional synaptic plasticity 34,35 . Although the induction of control LTD did not require activation of NMDARs, the reversal of this LTD by HFS in vivo was NMDAR-dependent. Thus, whereas the NMDAR antagonist CPP did not affect control LTD induced by LFS-900, it completely prevented the reversal of this mAChR-dependent LTD by HFS conditioning stimulation (Fig. 7e,f). To our surprise, HFS-induced de-depression was not prevented by Ab. Thus, HFS rapidly and persistently reversed Ab-facilitated LTD (Fig. 7a,b). Moreover, HFS-induced de-depression of Ab-facilitated LTD, like the persistent reversal of control LTD, was NMDAR-dependent, being abrogated in animals pretreated with CPP ( Fig. 7a,b). This indicates that HFSinduced NMDAR-dependent de-depression is resistant to Ab, unlike HFS-induced NMDAR-dependent LTP, as described previously 3,4 and below. This lack of effect of Ab on NMDARdependent de-depression, taken together with the inability of NMDAR antagonists to prevent the facilitation of LTD by Ab, underlines the potential importance of non-NMDAR mechanisms in mediating the synaptic plasticity disrupting effects of Ab in vivo.
Ab-facilitated LTD is mGlu5R-dependent. Apart from NMDARs, metabotropic glutamate receptors, in particular the mGlu5R subtype, have been implicated in the synaptic plasticity disrupting actions of Ab in vitro 5,8,36 . Bearing in mind the apparently differential roles of NMDARs in the effects of Ab on different forms of synaptic plasticity, next we assessed the involvement of mGlu5R in both Ab-mediated inhibition of LTP as well as Ab-facilitated LTD in vivo. Remarkably, systemic administration of the selective mGlu5R antagonist (negative allosteric modulator) MTEP prevented both of these disruptive actions of Ab without affecting either control LTP or control LTD. Thus, in animals administered with MTEP before intracerebroventricular (i.c.v.). injection of either synthetic or AD brain-derived Ab the application of LFS-300 failed to induce LTD ( Fig. 8a-d). Importantly, the same dose of MTEP had no effect on control LTD induced by LFS-900 (Fig. 8e,f), indicating that whereas Ab-facilitated LTD is mGlu5R-dependent, this was not the case for the control mAChR-dependent LTD. Somewhat similarly, whereas Ab 1-42 strongly inhibited LTP in vehiclepretreated animals, an identical HFS-triggered robust LTP in animals injected with MTEP followed by Ab (Fig. 8g,h). These Cross-reactive immunoglobulin-derived proteins that were detected when TBS buffer was immunoprecipitated are indicated (NS). (b,c) Similar to soluble synthetic Ab 1-42 , acute i.c.v. injection of unmanipulated TBS extract of AD brain (AD, 5 ml) also enabled the induction of robust and persistent LTD by the weak LFS-300 protocol. In contrast, the same extract of AD brain that had been immunodepleted of Ab using a polyclonal anti-Ab antibody (ID) did not enable the induction of LTD by LFS-300. findings strongly indicate that Ab enables LTD induction in vivo with an essential role of mGlu5, bypassing a requirement for activation of muscarinic ACh receptors. Moreover as MTEP prevented Ab's effects on both LTP and LTD, mGlu5Rs appear to be more pivotal to the synaptic plasticity disrupting actions of Ab than NMDARs.
Cellular prion protein mediates Ab-facilitated LTD. The question arises as to whether or not the facilitation of LTD by Ab shares other common mechanisms with LTP inhibition by Ab. Ab oligomers can bind very potently and selectively to cellular prion protein especially in a region that encompassed the aminoacid sequence 95-105, and thereby mediate inhibition of LTP by synthetic Ab 1-42 (ref. 25). The disease relevance of this finding is underscored by the PrP C -dependence of the inhibition of LTP by Ab oligomer-containing soluble extract of AD brain 37 . We examined the role of PrP C in mediating the facilitation of LTD by AD brain Ab and synthetic Ab 1-42 using monoclonal antibodies to PrP C . We started with the previously characterized anti-PrP C antibody 6D11, with an epitope that falls within the amino-acid 93-109 sequence, thereby preventing Ab 1-42 oligomer binding and inhibition of LTP 25 . Pretreatment with 6D11 completely prevented the facilitation of LTD by Ab-containing soluble AD brain extract (Fig. 9a,b). In order to further explore the role of PrP C , we compared the effect of two other high-affinity anti-PrP C antibodies (Fig. 9c,d). ICSM18, an antibody directed to helix-1 of PrP C , is known to inhibit Ab binding to PrP C and to prevent the LTP disrupting effect of AD brain extracts 37 . ICSM41 is an antibody to the structured region of PrP C with an undefined ARTICLE epitope that does not map to the Ab-binding region 38,39 . Although ICSM41 binds with similar high affinity to recombinant PrP C as ICSM18 (IC 50 : 0.41±0.04 and 0.3±0.1 nM, respectively, n ¼ 9, mean ± s.e.m.), unlike ICSM18, ICSM41 did not prevent Ab 1-42 protofibril binding to PrP C (Fig. 10b,c). Consistent with the differential ability of these two antibodies to prevent Ab 1-42 binding to PrP C , ICSM18 abrogated the facilitation of LTD by soluble AD brain extract, whereas the same dose of ICSM41 had no effect (Fig. 9c,d). These findings provide strong evidence that PrP C is required for the enablement of LTD by the most disease relevant form of soluble Ab, Ab from AD brain. We also tested the ability of ICSM18 to prevent the facilitation of LTD by synthetic Ab 1-42 . Ab from water-soluble extracts of AD brain contain a mixture of high-and lowmolecular weight components 8 , some of which bind to PrP C with high affinity 40,41 . In the case of synthetic Ab, protofibrillar We tested an eightfold lower dose of ICSM18 in this study because we found that ICSM18 bound to N2A cells, which express glycosylated mature PrP C , with an approximately eightfold higher affinity than ICSM41 (XC 50 4±1 and 33 ± 7 nM, respectively) (Fig. 10d). We found that this dose of ICSM18 completely abrogated the facilitation of LTD by protofibril Ab 1-42 (Fig. 9e,f). On the basis of the present and our previous 37 findings, PrP C appears to be a key site of binding and action for Ab-mediated disruption of both NMDARdependent and independent synaptic plasticity in vivo.

Discussion
We describe here for the first time the in vivo induction by synaptic stimulation of an mAChR-dependent homosynaptic LTD. The induction of mAChR-dependent LTD does not require NMDA or mGlu5R activation. Moreover, both chemically synthesized and human brain-derived Ab enhanced synaptically induced LTD in vivo. Remarkably, in Ab-treated animals the additional LTD does not require mAChRs, leaving mAChRdependent LTD relatively intact. However, like mAChR-dependent LTD, the Ab-facilitated LTD is NMDAR-independent. We found evidence that mGlu5R activation usurps the requirement for mAChRs to enable LTD induction via a process dependent on PrP C . Furthermore, Ab-mediated inhibition of LTP also requires mGlu5R and PrP C , placing Ab-PrP C -mGlu5R interactions central to the synaptic plasticity disrupting actions of Ab in vivo. LTD that requires mAChR activation has been proposed to be essential for certain forms of learning 21 , and the preservation of mAChR-dependent hippocampal LTD as animals age may be critical for maintaining cognitive performance 22 . The apparent ARTICLE dearth of studies of mAChR-dependent LTD in vivo may be owing to difficulties in optimizing suitable synaptic stimulation protocols. The present approach utilizes the insights gained from investigations of mAChR-dependent LTD in slices of cerebral cortex 28 . Currently used in vitro synaptic stimulation protocols to induce mAChR-dependent LTD at CA3-to-CA1 synapses have been reported to induce an LTD that is at least partly inhibited by mAChR antagonists 19 . The present finding that synaptic conditioning stimulation can induce LTD that is completely blocked by scopolamine provides strong evidence that mAChRdependent LTD that lasts for over 5 h can be induced by endogenously released ACh in vivo and therefore supports its proposed role in synaptic information storage.
Because previous reports had indicated that in vitro, Ab strongly impairs mAChR-mediated signalling 29 that may underlie mAChR-dependent LTD in the cerebral cortex 28 , we predicted that mAChR-dependent LTD in the hippocampus would be inhibited by Ab in vivo. To our surprise, Ab enabled additional LTD while at the same time leaving a scopolamine sensitive component of LTD relatively unscathed. It was apparent that Ab usurped mAChR-dependent LTD by lowering the synaptic stimulation threshold to induce another form of LTD that was mAChR-independent. The mechanisms of the additional LTD, however, appear to be at least partly shared with mAChRdependent LTD, as the initial phase of the control LTD was partly occluded by the Ab-enabled LTD.
Particularly surprising was the apparent lack of involvement of NMDARs in the facilitation of LTD by Ab, especially in view of the presumed essential role of NMDARs in the relatively selective binding of Ab oligomers to synapses 42 . Moreover, antagonists of GluN2B subunits prevent Ab-mediated facilitation of NMDAR-dependent LTD 5,6 and inhibition of NMDARdependent LTP [12][13][14][15] . These findings have led to the elucidation of a key role of GluN2B subunits in mediating the synaptic plasticity disrupting actions of Ab and have been extended to include many other deleterious effects of Ab 40,43 . However, based on the present results, targeting GluN2B is unlikely to prove to be an effective therapeutic strategy on its own and underlines the need to also explore non-NMDAR mechanisms.
Further undermining the putative primacy of NMDARs in the synaptic actions of Ab was the finding that Ab did not significantly affect NMDAR-dependent de-depression. This is all the more remarkable considering that Ab strongly inhibited NMDAR-dependent LTP at these same synapses using the same HFS induction protocol. Previous research has found that the persistent reversal of LTD by conditioning stimulation requires the recruitment of different signalling pathways to those usually necessary for LTP induction 44,45 . Thus the lack of inhibition of NMDAR-dependent de-depression at these synapses indicates that the inhibition of LTP by Ab is not due to the dependence of LTP on NMDARs. Furthermore, the present findings indicate that pharmacological inhibition of NMDARs may prevent potentially physiological reversal of LTD and leave any deleterious effects of Ab-facilitated NMDAR-independent LTD unopposed.
The present findings underscore a much more central role for the mGlu5R in mediating the synaptic plasticity disrupting effects of Ab and suggest that the lowering of the threshold for LTD and inhibition of LTP are two sides of one coin. Our finding that Abfacilitated LTD, like Ab-mediated inhibition of LTP, is blocked by antibodies that prevent Ab binding to PrP C provides an explanation for the pivotal role of mGlu5Rs. Previous research 46 has revealed that Ab acts as an extracellular scaffold to promote the inappropriate synaptic mobilization and activation of mGlu5R on cultured neurons. The membrane binding of Ab is prevented by both anti-mGlu5R and anti-PrP C antibodies in a non-additive manner 46 , consistent with the key role of PrP C in the binding of the Ab oligomer to plasmalemma 25 . The aberrant clustering of mGlu5R at synapses by Ab by binding to PrP C may trigger disruptive signalling activity that can enable LTD and inhibit LTP induction. Very recently direct evidence that PrP C mediates multiple effects of Ab oligomers, including dendritic spine loss in cultured neurons, by a direct physical linkage of PrP C with mGlu5Rs at or near the postsynaptic membrane was reported 27 . If the formation of Ab-PrP C -mGlu5R complexes is primary, then the requirement for NMDARs that contain GluN2B subunits in the inhibition of LTP by Ab is likely to be a downstream consequence. Indeed, mGlu5Rs provide a (d) FACS analysis revealed that ICSM18 bound to N2A cells, which express glycosylated mature PrP C , with an approximately eightfold higher affinity than ICSM41 (XC 50 4 ± 1 nM and 33 ± 7 nM, respectively, n ¼ 4, mean ± s.e.m.).
transduction link in the Ab-PrP C complex-mediated transmembrane coupling to NR2B subunits via activation of the tyrosine kinase Fyn 27,40,47 . In addition to Fyn, an Ab-PrP C -mGlu5R-mediated dysregulation of intracellular Ca 2 þ , eukaryotic elongation factor 2 and Arc 27 may contribute to synaptic plasticity disruption 11 by Ab in vivo.
Overall, the present research provides strong evidence that an Ab-PrP C -mGlu5R triad is critical for synaptic plasticity disruption, enabling an NMDAR-independent LTD to usurp mAChRdependent LTD and inhibit NMDAR-dependent LTP. Selectively targeting this Ab-PrP C -mGlu5R triad offers many possible means of preventing dysfunction of critical brain plasticity mechanisms in early AD.

Methods
Animals and surgery. Adult (250-350 g, 8-11 weeks old) male Wistar rats (BioResources Unit, Trinity College, Dublin) were used in all experiments. The animals were housed under a 12-h light-dark cycle at room temperature (19-22°C). Before the surgery, animals were anesthetized with urethane (1.5-1.6 g kg À 1 , i.p.). Lignocaine (10 mg, 1% adrenaline, subcutaneously) was injected over the area of the skull, where electrodes and screws were to be implanted. The body temperature of the rats was maintained at 37-38°C with a feedback-controlled heating blanket. The animal care and experimental protocol were approved by the Department of Health, Republic of Ireland.
Cannula implantation. In order to inject drugs or Ab into the brain, a stainlesssteel cannula (22 gauge, 0.7 mm outer diameter) was implanted above the right lateral ventricle (1 mm lateral to the midline and 4 mm below the surface of the dura). i.c.v. injection was made via an internal cannula (28 gauge, 0.36 mm outer diameter). The solutions were injected in a 5 ml volume over a 3-min period or 10 ml volume over a 6-min period. Verification of the placement of cannula was performed post mortem by checking the spread of ink dye after i.c.v. injection.
Electrode implantation. Monopolar recording electrodes were constructed from Teflon-coated tungsten wires (75 mm inner core diameter, 112 mm external diameter) and twisted bipolar stimulating electrodes were constructed from Tefloncoated tungsten wires (50 mm inner core diameter, 75 mm external diameter) separately 12 . Field excitatory postsynaptic potentials (EPSPs) were recorded from the stratum radiatum in the CA1 area of the right hippocampus in response to stimulation of the ipsilateral Schaffer collateral-commissural pathway. Electrode implantation sites were identified using stereotaxic coordinates relative to bregma, with the recording site located 3.4 mm posterior to bregma and 2.5 mm lateral to midline, and stimulating site 4.2 mm posterior to bregma and 3.8 mm lateral to midline. In some animals, another stimulating electrode was implanted at a site located 2.5 mm posterior to bregma and 2.2 mm lateral to the midline. The final placement of electrodes was optimized by using electrophysiological criteria and confirmed via post-mortem analysis.
Electrophysiology. Test EPSPs were evoked by a single square-wave pulse (0.2 ms duration) at a frequency of 0.033 Hz and an intensity that triggered a 50% maximum EPSP response. LTD was induced using 1 Hz LFS consisting of 900 pulses (0.2 ms duration). During the LFS the intensity was raised to trigger EPSPs of 95% maximum amplitude. A relatively weak LFS protocol, used to study the Ab-mediated facilitation of LTD, consisted of 300 pulses (0.2 ms duration) at 1 Hz, with an intensity that evoked 95% maximum amplitude. LTP was induced using 200 Hz HFS consisting of one set of 10 trains of 20 pulses (inter-train interval of 2 s). The stimulation intensity was raised to trigger EPSPs of 75% maximum during the HFS. None of the conditioning stimulation protocols elicited any detectible abnormal changes in background EEG, which was recorded from the hippocampus throughout the experiments.
Synthetic Ab. We made two main different preparations of synthetic Ab, soluble and protofibril Ab 1-42 . Our standard, soluble Ab 1-42 (Bachem or Biopolymer Laboratory, UCLA Medical School) was prepared as a stock solution of 64 mM in mild alkali (0.1% ammonium hydroxide) in milliQ water (Millipore Corporation, Ireland) to avoid isoelectric precipitation and then centrifuged at 100,000 g for 3 h to remove any fibril aggregates. An aliquot of the supernatant was taken to estimate peptide concentration using the micro BCA protein assay (Thermo-Fisher Scientific Life Science Research Products, Rockford, IL) and the remaining supernatant was stored at À 80°C until required. Whereas the test dose (160 pmol) of soluble Ab 1-42 did not affect baseline transmission in the absence of LFS (see Results), double this dose (320 pmol, i.c.v.) caused a small (B15%) decrease in baseline at 3 h.
Differentially aggregated protofibril Ab 1-42 and biotinylated Ab 1-42 were synthesized, and purified by Dr James I. Elliott at Yale University (New Haven, CT). Peptide (B10-20 mg) was weighed into a screw-cap 50-ml Sterilin tube, dissolved in anhydrous DMSO with gentle mixing for 2 min to produce a 5-mM solution and then diluted to 100 mM in phenol red-free Ham's F12 medium (Caisson Labs) and vortexed for 15 s. Samples were aggregated without shaking for 48 h, transferred to a 2-ml eppendorf tube, centrifuged at 16,100 g for 20 min to remove any large preformed aggregates and the upper 90% for each solution collected, aliquoted, snap frozen in liquid N 2 and stored at À 80°C. Samples were then tested for the presence of large protofibrillar assemblies, known to bind avidly to PrP and cause PrP-dependent toxicity 26 . Ab 1-42 protofibrils used for electrophysiology were further dialysed against 2 Â 5 l of PBS in an 8000 MWCO semi-permeable membrane to ensure all DMSO and cell media were exchanged before freezing and characterization.
Electron microscopy. Five microlitre of peptide solution was applied to glowdischarged carbon-coated copper grids and left to bind for 60 s. Excess solution was removed using grade 4 Whatman filter paper. Samples were negatively stained with 2% uranyl acetate for 30 s, blotted then allowed to air dry. Images were acquired on an FEI Tecnai T10 electron microscope operating at 100 kV and recorded on a 1k Â 1k charged couple device camera (Gatan) at a typical magnification of 34,000 with a pixel size of 5.03 Å.
SEC and multi-angle light scattering. Aliquots (0.33 ml) of Ab 1-42 protofibrils were injected onto a Superdex 200 10/30 column (GE Healthcare) and eluted with PBS at a flow rate of 0.5 ml min À 1 using an Agilent HPLC and peptide elution monitored by absorbance at 275 nm. Light scattering was performed using a Wyatt DAWN HELEOS II multi-angle light scattering module with Ab concentrations calculated using the refractive index.
TBS extract of human brain. AD brain 1 was obtained and used in accordance with the UCD Human Research Ethics Committee guidelines (under approval LS-E-10-10-Walsh). AD brain 2 was obtained and used in accordance with the Partner's Institutional Review Board (Walsh BWH 2011). In both cases informed consent was obtained from subjects. Samples of temporal cortex were obtained from 2 AD cases referred to as AD1 and AD2. AD1 was from an 85-year-old male with dementia and fulminant amyloid and tangle pathology (Braak stage ¼ 4) and was provided by Drs Dykoski and Cleary of Minneapolis VA Health Care System, and potently inhibits LTP 48 . AD2 was from an 81-year-old female who died with severe AD and was kindly provided by Dr Cindy Lemere of Brigham and Women's Hospital. Frozen cortex (0.9 g) was allowed to thaw on ice, chopped into small pieces and homogenized in 4.5 ml of ice-cold 20 mM Tris-HCl, pH 7.4, containing 150 mM NaCl with 25 strokes of a Dounce homogenizer (Fisher, Ottawa, Ontario, Canada) 31,48 . Water-soluble Ab was separated from membrane-bound and plaque Ab by centrifugation at 91,000 g and 4°C in a TLA 55 rotor (Beckman Coulter, Fullerton, CA, USA) for 78 min. To eliminate bioactive small molecules the supernatant was exchanged into ammonium acetate. Thereafter, extracts were divided into two parts: one aliquot was immunodepleted of Ab by three rounds of 12-h incubations with our anti-Ab antibody, AW8 (ref. 31), and protein A at 4°C. The second portion was not manipulated in any way and is simply referred to as AD. Aliquots of samples were stored at À 80°C or used to assess Ab content with a sensitive immunoprecipitation/western blot procedure. Our rabbit polyclonal antibody, AW8, was used (at a dilution of 1:80) for immunoprecipitation and a combination of the anti-Ab40 and Ab42 monoclonal antibodies, 2G3 and 21F12 (each at 1 mg ml À 1 ) for western blot. Ab concentration was estimated by reference to known quantities of synthetic Ab 1-42 . Antibodies 2G3 and 21F12 were kindly provided by Drs P. Seubert and D. Schenk (Elan Pharmaceuticals).
Data analysis. The magnitude of LTD is expressed as the percentage of pre-LFS baseline EPSP amplitude ( ± s.e.m.). The sample size was chosen based on our knowledge of what is appropriate for in vivo electrophysiology to determine whether synaptic plasticity is induced or affected by Ab or other interventions 3,6,10,12 . No data were excluded, and control experiments were interleaved randomly throughout. Two-tailed paired Student's t-tests (paired t) or one-way ANOVA with Tukey's multiple comparison test (one-way ANOVA-Tukey) were used to evaluate LTD within groups, and two-way ANOVAor unpaired Student's t-tests (unpaired t) were used to compare between groups. Kruskal-Wallis one-way ANOVA with Dunn's multiple comparison test was used to compare the effects of antibodies on Ab binding to recombinant PrP C . A Po0.05 was considered as statistically significant.