Intracellular oligomeric amyloid-beta rapidly regulates GluA1 subunit of AMPA receptor in the hippocampus

The acute neurotoxicity of oligomeric forms of amyloid-β 1-42 (Aβ) is implicated in the pathogenesis of Alzheimer’s disease (AD). However, how these oligomers might first impair neuronal function at the onset of pathology is poorly understood. Here we have examined the underlying toxic effects caused by an increase in levels of intracellular Aβ, an event that could be important during the early stages of the disease. We show that oligomerised Aβ induces a rapid enhancement of AMPA receptor-mediated synaptic transmission (EPSCA) when applied intracellularly. This effect is dependent on postsynaptic Ca2+ and PKA. Knockdown of GluA1, but not GluA2, prevents the effect, as does expression of a S845-phosphomutant of GluA1. Significantly, an inhibitor of Ca2+-permeable AMPARs (CP-AMPARs), IEM 1460, reverses the increase in the amplitude of EPSCA. These results suggest that a primary neuronal response to intracellular Aβ oligomers is the rapid synaptic insertion of CP-AMPARs.


Intracellular infusion of oligomerised Aβ1-42 (Aβ) causes a rapid increase in the AMPAR-mediated EPSC (EPSC A ) in CA1 pyramidal neurons.
Since Aβ oligomers are toxic 3-5 , we were interested in determining the intracellular effects of Aβ oligomers on synaptic function. Neurons were injected with oligomerised Aβ via passive diffusion from the patch pipette, whilst basal synaptic transmission was measured. Aβ oligomers caused a rapid increase in the amplitude of the AMPAR-mediated excitatory postsynaptic current (EPSC A ) (181 ± 15%, n = 7, Fig. 2a). In contrast, neither the infusion of non-aggregated, monomeric Aβ nor Aβ oligomers that had been pre-incubated with clusterin, a chaperone that sequesters oligomers 28 , had any significant effect upon EPSC A (81 ± 8%, n = 6, Fig. 2b and 103 ± 9%, n = 7, Fig. 2c, respectively). The effect of Aβ oligomers was independent of the need to evoke EPSC A , since stopping stimulation for 15 min, shortly after obtaining whole-cell configuration, did not prevent the increase in synaptic transmission (closed circle: 192 ± 26%, n = 6, Fig. 2d). In addition, the effect of Aβ oligomers did not require the activation of NMDA receptors (NMDAR), since EPSC A was enhanced in the presence of the NMDAR antagonist D-AP-5 (147 ± 14%, n = 6, Fig. 2e). This effect was also specific for EPSC A since a pharmacologically-isolated NMDAR-mediated EPSC (EPSC N : holding voltage -40 mV, 10 μ M NBQX perfusion) was unaffected by Aβ oligomer infusion (99 ± 14%, n = 6, Fig. 2f). 2+ and PKA. We next investigated the signalling cascades that underlie the rapid action of Aβ oligomers on AMPAR-mediated synaptic transmission (Fig. 3). Changes in postsynaptic Ca 2+ levels initiate signal cascades involved in the modulation of synaptic transmission [29][30][31] . Therefore we tested whether blockade of postsynaptic Ca 2+ mobilisation affects Aβ -mediated EPSC A regulation. The Aβ oligomer-induced increase was dependent on postsynaptic Ca 2+ , since it was prevented by postsynaptic infusion of the Ca 2+ chelator BAPTA (95 ± 12%, n = 7, Fig. 3a), and relied on Ca 2+ release from intracellular stores, since bath applied ryanodine also prevents the Aβ -induced EPSC A increase (108 ± 18%, n = 7, Fig. 3b). We were interested in examining the Ca 2+ -dependent mechanism responsible for these effects, and possible downstream effectors. Ca 2+ -induced changes in synaptic transmission are known to involve, among other kinases, protein kinase A (PKA) 32,33 . Accordingly, we tested the involvement of PKA in the observed Aβ -induced EPSC A increase. We found that the effect required the activation of PKA, since it was prevented by either Rp-cAMPS, a cyclic AMP analogue that acts as a competitive antagonist of cAMP-induced activation of PKA (97 ± 8%, n = 6, Fig. 3c) or H89, a PKA inhibitor (94 ± 9%, n = 6, Fig. 3d), but not PKC since it was unaffected by both the PKC inhibitor Ro 32-0432 (171 ± 7%, n = 6, Fig. 3e) or PKC19-31, a pseudosubstrate of PKC which functions to inhibit the kinase (166 ± 13%, n = 6, Fig. 3f).

Aβ oligomer-induced increase in EPSC A is dependent on postsynaptic Ca
Calcium-calmodulin kinase II (CaMKII) is a Ca 2+ -sensitive kinase that has also been implicated in the regulation of AMPAR expression 34,35 . We therefore tested the involvement of CaMKII in Aβ -induced EPSC A increase. When cells were infused with Aβ and the CaMKII inhibitor KN-62, we observed an initial increase in EPSC A that rapidly declined (97 ± 16%, n = 7, Fig. 3g).    Aβ oligomer-induced enhancement of EPSC A is mediated by the GluA1 subunit of AMPARs. Since homomeric forms of GluA1-AMPARs characteristically display greater conductance than GluA2 containing AMPARs 36,37 , we hypothesized that infusion of Aβ facilitates EPSC A through an increase in synaptic homomeric GluA1 AMPARs. Indeed, the activation of PKA can lead to the insertion of GluA1-containing, GluA2-lacking AMPARs, known as Ca 2+ -permeable AMPARs (CP-AMPARs) 38,39 .
Collectively, these data suggest that the rapid Aβ oligomer-induced changes in EPSC A may be due to a PKA-dependent synaptic insertion of CP-AMPARs. To test this directly, we bath-applied IEM 1460 (IEM), a compound that selectively blocks CP-AMPARs 41 . IEM had no effect on EPSC A when using control pipette solution (99 ± 8%: 30 min after IEM treatment, n = 6, Fig. 4d), which is consistent with a negligible contribution by CP-AMPARs to basal AMPAR-mediated transmission. However, IEM dramatically reduced the EPSC A following infusion with Aβ oligomers (159 ± 11%: 10 min after infusion of Aβ oligomers; 77 ± 10%: 30 min after the start of IEM treatment, n = 8, Fig. 4e). This suggests that Aβ oligomer infusion causes a rapid increase in the synaptic expression of CP-AMPARs, resulting in the observed facilitation of EPSC A amplitude.
Using biotinylation assays from hippocampal slices, we found that the surface expression of GluA1 was significantly increased with exogenous Aβ treatment but that there was no change in GluA2/3 expression (Fig. 5A). This suggests that exogenously applied Aβ also induces the insertion of CP-AMPARs. To support these findings, we measured EPSC A during the extracellular perfusion of Aβ . We found that there was an increase of EPSC A on application of Aβ (145 ± 7%, n = 6, Fig. 5B), which was prevented when slices were continually perfused with IEM (89 ± 6%, n = 6, Fig. 5C).

Discussion
Here we have revealed a rapid synaptic response to intracellular accumulation of Aβ oligomers. Several lines of evidence suggest that extracellular Aβ oligomers are taken up into neurons where they impair synaptic function 22 . By studying the effects of intracellularly applied Aβ oligomers we have found a rapid action: the insertion of CP-AMPARs via a PKA-dependent phosphorylation of s845 of GluA1. These effects, occurring as a primary response to the emergence of cytosolic Aβ oligomers, could contribute to a key catalyzing mechanism of subsequent aberrant synaptic transmission. This finding therefore highlights a surprising discrepancy in our current understanding of the effects of Aβ on synaptic receptors. Whereas previous studies have shown that Aβ can drive the downregulation of synaptic transmission, in some cases mediated by the internalization of AMPARs and NMDARs 42-45 , our findings and those of others seem to indicate, in contrast, that Aβ can actually facilitate synaptic transmission, possibly through inducing the expression of receptors [46][47][48][49] . Presumably this is due to different time courses of Aβ -mediated toxic effects (the above studies, for example, range in treatment times from minutes to hours) and/or CP-AMPARs mediated secondary toxic insults 50,51 .
The mechanisms regulating the trafficking of GluA1-containing AMPARs have previously been characterised, and generally converge on C-terminus phosphorylation events [52][53][54] . One canonical mechanism is the phosphorylation of the s845 residue on GluA1, priming its expression at the synapse 52 . Our finding that the expression of S845A, a mutant form of GluA1 which cannot be phosphorylated at s845, blocks the Aβ -induced enhancement of EPSC A , suggests that Aβ operates this rapid effect via a regulated physiological mechanism; the PKA-mediated phosphorylation of GluA1-s845. CaMKII has previously been implicated in AMPAR regulation 34,35 . Consistent with this role, we found that inhibiting CaMKII blocks the Aβ -induced enhancement of EPSC A . Interestingly, we observed a delayed effect under these conditions; whilst there was an initial increase in EPSC A , this rapidly declined. This might be explained by previously reported roles for CaMKII in the synaptic stabilization of AMPARs 55 . Therefore, PKA and CaMKII may act in concert in this mechanism, promoting the expression and then stabilization of synaptic AMPARs, respectively. Together, these data raise the interesting question as to whether Aβ might actually operate physiologically to regulate synaptic glutamate receptor expression, and whether its aberrant cytosolic presence leads to a dysregulated physiological process. Clearly, more work is required to further understand a possible non-pathological role of Aβ .
CP-AMPARs are expressed at an early postnatal age and are replaced with GluA2-containing Ca 2+ -impermeable AMPARs during development [56][57][58] . CP-AMPARs are critically involved in physiological 34,59,60 and pathological plasticity in the matured synapse [61][62][63] . Furthermore, growing evidence suggests CP-AMPARs prime neurodegenerative diseases including stroke, ischaemia and amyotrophic lateral sclerosis 62,64 . We found that blocking CP-AMPARs prior to exposure to exogenous Aβ prevented the facilitation of synaptic transmission. Therefore, our findings support the hypothesis that progressive Aβ -mediated CP-AMPAR expression is a pivotal catalyst for the onset of pathology.
The early accumulation of intracellular Aβ has been shown to be neurotoxic 65 and transgenic models have shown it to be sufficient for cognitive impairments prior to the increase in extracellular Aβ 21,66 . Indeed, the accumulation of intracellular Aβ has previously been shown to be prevalent in the brains of AD patients [67][68][69] , and this is thought to be one of the earliest events in the pathology, preceding Aβ plaques and neurofibrillary tangles 67,70 . A recent report has shown that the infusion and accumulation of Aβ into neurons can have significant impairing effects on synaptic function 71 . Accounting for these findings and our data, the accumulation of intracellular Aβ will likely prove to be a catalyzing event in the pathogenesis of the disease. Given that the primary response to an increase in intracellular Aβ  appears to be the expression of CP-AMPARs at the synapse, targeting CP-AMPARs may provide a means of restoring synaptic function in AD.

Methods
Amyloid-β preparation. The amyloid-β 1-42 peptide (Aβ ; Millipore, UK) was first dissolved at a concentration of 1 mg / ml in 100% HFIP (1,1,1,3,3,3-hexafluoro-2-propanol [Sigma-Aldrich]). This solution was incubated at room temperature for 1 h with occasional vortexing. Next, the solution was sonicated for 10 min in a water bath sonicator. The solution was then dried under a gentle stream of nitrogen gas. 100% DMSO was then used to resuspend the peptide, which was then incubated at room temperature for 12 min with occasional vortexing. This solution was finally aliquoted into smaller volumes and stored at -80 °C. For a working solution, D-PBS (Invitrogen, UK) was added to the peptide stock solution and incubated for 2 h at room temperature to allow for peptide aggregation. To prepare monomeric Aβ , the same proceedure outlined above was followed, with the exception of the 2 h room temperature aggregation step.

Electrophysiology. All animal experiments were carried out in accordance with the UK Scientific
Procedures Act, 1986 and associated guidelines. The methods were carried out in accordance with the approved guidelines. All experimental protocols were approved by the University of Bristol Animal Welfare & Ethical Review Body. Acute hippocampal slices were prepared from 26 -to 32 -day-old male Wistar rats. Animals were sacrificed by dislocation of the neck and then decapitated. The brain was rapidly removed and placed in ice-cold artificial CSF (aCSF) containing (in mM): 124 NaCl, 3 KCl, 26 Only cells with series resistance < 25 MΩ with a change in series resistance < 10% from the start were included in this study. The amplitude of the excitatory postsynaptic currents (EPSCs) was measured and expressed relative the normalized baseline (first 5 min of recording).
Slice biotinylation and NeutraAvidin pull-down. Surface biotinylation of acute slices was performed as described previously with some modifications 72 . Briefly, slices were initially washed twice in aCSF and subsequently incubated in aCSF containing 1 mg / ml Sulfo-NHS-SS-Biotin (Thermo Scientific, Rockford, USA) for 45 min at 4 °C to allow for labelling of all surface membrane proteins. Excess biotin was removed by washes in aCSF containing NH 4 Cl. Tissue was then homogenised in lysis buffer containing 25 mM Tris (pH 7.5), 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 10m M NaF and a cocktail of protease inhibitors (Sigma, St Louis, USA) and incubated for 30 min prior to centrifugation at 1,000 g to remove cellular debris. The total protein concentration was determined using the Pierce BCA kit. Subsequently, 100 μ l of StreptaAvidin beads (Upstate, USA) were added to 500 μ g of protein lysate and placed on a rotator at 4 °C for 2 hr. Samples were then washed five times in lysis buffer; beads were pulled-down after each wash by gentle centrifugation. Bound proteins were eluted by adding 2 X SDS reducing buffer and moderate heating at 60 °C for 30 min. The resulting supernatant was transferred to new tubes and heated at 90 °C for 5 min prior to gel loading. Statistical Analyses. Data were analyzed from one slice per rat (i.e., n = number of slices = number of rats). Data pooled across slices are expressed as the mean ± s.e.m. Significance (p < 0.05) was tested using two-tailed t-tests. For electrophysiology experiments, mean ± s.e.m. data from the 40 min timepoint are described.