The Neuroprotective Beta Amyloid Hexapeptide Core Reverses Deficits in Synaptic Plasticity in the 5×FAD APP/PS1 Mouse Model

Alzheimer’s disease (AD) is the most common cause of dementia in the aging population. Evidence implicates elevated soluble oligomeric Aβ as one of the primary triggers during the prodromic phase leading to AD, effected largely via hyperphosphorylation of the microtubule-associated protein tau. At low, physiological levels (pM-nM), however, oligomeric Aβ has been found to regulate synaptic plasticity as a neuromodulator. Through mutational analysis, we found a core hexapeptide sequence within the N-terminal domain of Aβ (N-Aβcore) accounting for its physiological activity, and subsequently found that the N-Aβcore peptide is neuroprotective. Here, we characterized the neuroprotective potential of the N-Aβcore against dysfunction of synaptic plasticity assessed in ex vivo hippocampal slices from 5×FAD APP/PS1 mice, specifically hippocampal long-term potentiation (LTP) and long-term depression (LTD). The N-Aβcore was shown to reverse impairment in synaptic plasticity in hippocampal slices from 5×FAD APP/PS1 model mice, both for LTP and LTD. The reversal by the N-Aβcore correlated with alleviation of downregulation of hippocampal AMPA-type glutamate receptors in preparations from 5×FAD mice. The action of the N-Aβcore depended upon a critical di-histidine sequence and involved the PI3 kinase pathway via mTOR. Together, the present findings indicate that the non-toxic N-Aβcore hexapeptide is not only neuroprotective at the cellular level but is able to reverse synaptic dysfunction in AD-like models, specifically alterations in synaptic plasticity.


Introduction
Alzheimer's disease (AD) is clinically characterized by impairments in cognitive memory and function. Loss of critical pre-and post-synaptic markers have been reported for postmortem AD brain tissue [1,2], suggesting that AD-related cognitive impairments are based, in large part, on synaptic dysfunction and loss. Additionally, accumulating evidence shows a strong link between excess soluble oligomeric amyloid-β (Aβ) and synaptic dysfunction in AD [3][4][5]. Cognitive decline and synaptic plasticity deficits are reported to occur prior to the accumulation of Aβ plaques and tau neurofibrillary tangles in prodromic phases leading to AD [6], supporting the idea that synaptic dysfunction and mild cognitive impairment are early events driven by soluble oligomeric Aβ rising to abnormally high levels years prior to AD diagnosis.
Synaptic dysfunction and eventual degeneration lead to abnormal synaptic transmission and impaired long-term potentiation (LTP) and/or long-term depression (LTD), which are important in synaptic plasticity and learning and memory. Pathological levels (high nM to μM) of Aβ have been shown to inhibit LTP-induction [3,7,8] and enhance LTD [9,10] in the hippocampus. On the other hand, low physiological levels (pM) of Aβ was found to enhance LTP and memory, indicating a hormetic effect of Aβ on synaptic plasticity [11][12][13].
Dysregulation of synaptic plasticity in AD pathogenesis involves altered regulation of NMDA-type and AMPA-type glutamate receptors. In addition to mediating Aβ-induced excitotoxicity, NMDA receptors can be depressed by Aβ at high concentrations [14], inducing LTD [15,16] as a consequence of subsequent downstream AMPA receptor internalization [15,16] and dendritic spine loss [16].
We have shown that at low concentration (pM-nM) the N-terminal Aβ fragment comprising amino acids 1-15/16 of the Aβ sequence, an endogenous peptide cleaved from Aβ via -secretase [17], is more effective as a neuromodulator than full-length Aβ1-42, stimulating receptor-linked increases in neuronal Ca 2+ , enhancing synaptic plasticity and enhancing contextual fear memory [13]. The Aβ1-16 peptide sequence corresponds to the C-terminal 16 amino acid sequence in soluble amyloid precursor protein- (sAPP-), referred to as the CT16, which has also been shown to enhance synaptic plasticity [18]. An essential core sequence comprising amino acids 10-15 of Aβ (N-Aβcore) was identified as the active region of the N-terminal Aβ fragment and was further shown to protect against Aβ-induced neuronal toxicity [19]. Here, we aimed to better understand the neuroprotective mechanism of the N-Aβcore on synaptic plasticity. We investigated whether the N-Aβcore could rescue LTP and LTD dysfunction resulting from prolonged, elevated levels of Aβ in an APP/PS1 transgenic mouse model harboring mutations found in familial Alzheimer's disease (FAD), while assessing the impact on AMPA-type glutamate receptor expression in reference to the neuroprotective action of the N-Aβcore in Aβ-synaptotoxicity.

Transgenic mice and cannulation surgery
All animal handling, surgery, use and euthanasia were performed under an approved IACUC protocol (11- For bilateral cannulation and injection, the following protocol was employed (as per ref 19). Cannulation into the dorsal aspects of both hippocampi of the 5FAD mice at 7to 8-months of age of both sexes was performed under full anesthesia (general: 1.2% Avertin; local at site: lidocaine) using stereotaxis (coordinates: -1.5mm anteroposterior; ±1mm lateral; -2mm depth). After the brief surgical procedure and recovery (full righting reflex), mice were subsequently housed in sound-isolated, ventilated hotels prior to peptide injection one week later. On the day of microinjection (morning), sterile saline or 500 nM N-Aβcore peptide was administered bilaterally through the cannulae in the 5FAD mice via microinjectors over 30s (0.5μL/side) and the mice were returned to their cages in the mouse hotel. Hippocampi were collected from euthanized mice 24 h after the bilateral microinjection of the saline or peptide, and lysates extracted from the hippocampi were prepared for immunoblot analysis (30 µg each). Euthanasia was performed under an approved IACUC protocol (11-1219-6 / 16-2282-2). This study was not preregistered and followed ARRIVE guidelines.

Preparation of Aβ peptides
Solutions of Aβ 1-42 (American Peptide; Anaspec) were prepared from aqueous stock solutions, followed by bath sonication. This preparation of full-length Aβ was previously shown to exist predominantly in the oligomeric state [see 13]. The N-Aβcore peptide (synthesized and purified to Peptide 2.0) was prepared from aqueous stock solutions.

Extracellular field potential recordings in hippocampal slices
Hippocampal slices were prepared from 7-to 8-month-old 5FAD (Tg6799) or B6.SJL (control mice) (as per ref 13). Cervical dislocation and decapitation were performed under an approved IACUC protocol (11-1219-6 / 16-2282-2), compliant with NIH and Society for Neuroscience guidelines for use of vertebrate animals in neuroscience research.
Brains were removed into ice-cold artificial cerebral spinal fluid (aCSF) consisting of 130mM NaCl, 3.5 mM KCl, 10mM glucose, 1.25mM NaH 2 PO 4 , 2.0mM CaCl 2 , 1.5mM MgSO 4 , and 24mM NaHCO 3, bubbled in 95% O 2 /5% CO 2 . Transverse brain slices of 400μm were obtained using a Leica vibrating microtome (Leica, VT1200S) and quickly transferred to fresh ice-cold aCSF for hippocampi isolation. Extracted hippocampi slices were incubated in bubbled aCSF in a holding chamber for 30 mins at room temperature (23°C) after which the holding chamber was transferred to a 32°C water bath for 1 h. The chamber was then removed from the water bath and placed at room temperature for another 1 h prior to recording. The slices were subsequently transferred to a recording chamber and perfused at 3mL/min with aCSF (bubbled with 95% O 2 /5% CO 2 ) at 32°C.
The Schaffer collateral fibers were stimulated at a frequency 0.1 Hz using a bipolar stimulating electrode and CA1 field excitatory postsynaptic potentials (fEPSPs) were recorded with a glass electrode filled with 3M NaCl (resistance 1-1.

Immunoblot analysis
Hippocampi injected with N-Acore. Hippocampi  were added to diluted SDS-solubilized protein samples for a final protein concentration of 2 g/L. The samples were boiled at 95°C for 10 min, immediately cooled on ice and then centrifuged. Equal amounts of protein were subjected to electrophoretic separation on a 4-20% Tris-Glycine polyacrylamide gel (ThermoFisher Scientific, #XP04200), transferred to Nitrocellulose membrane (LI-COR, # 92631092) via the iBlot2 semidry system (ThermoFisher). Recovered blots were incubated in primary antibody (see below) overnight at 4°C. The transfer blots were washed 3x (10 min each wash) in 0.1% Tween-20 in Tris-buffered saline (TBS) and incubated in the appropriate IR-dye-conjugated secondary antibody (LI-COR Biosciences) for 1h. An Odyssey IR imaging system (LI-COR Biosciences, Lincoln, NE) was used for signal detection. Analysis was performed via Image Studio v5.2.5 software (LI-COR Biosciences).

Cultured hippocampal neurons
Hippocampal neuron cultures were prepared as described [21] from neonatal mouse pups The cells were dissociated using sequential trituration (polished Pasteur pipettes of decreasing diameter) and collected by low-speed centrifugation. The dissociated cells were pre-plated in standard tissue culture dishes to remove adherent non-neuronal cells (glia; fibroblasts) for 10-15 mins. The neuron-enriched preparation was diluted to 1x10 5 cells/mL and then plated into poly-D-lysine-coated 24-well dishes in NB plus serum and Gentamicin. The cultures were washed with Neurobasal A medium containing B27 and Gentamicin to remove the serum and then cultured in this media for 7-10 days prior to treatment with Aβ, N-Aβcore or the combination for an additional 7 days. qPCR RNA was extracted from treated cultured hippocampal neurons using the PureLink® RNA Mini Kit (Ambion, Life Technologies, #12183025) as per the manufacturer's protocol.
Genomic DNA contamination was eliminated the RNA preparation by digesting with RNase-free DNase (Qiagen, #79254). The iScript™ cDNA Synthesis Kit (Bio-Rad) was used to synthesize cDNA. The expression levels of various genes were then determined using SYBR green via qPCR (Bio-Rad iCycler iQ™ Multicolor Real-Time PCR Detection System) using the primers listed in the accompanying table (Table 1). Cycling conditions were as follows: 95 °C for 15 min, followed by 40 cycles of 94 °C for 15 s and 60 °C for 60 s and the fold-changes in the variously treated samples compared to untreated (vehicle control) samples were calculated after normalizing to GAPDH gene expression. Table 1 Gene (Accession number) Sequence (5'-3') Forward GATGTTATGGAAGCAAGT

Data and statistical analysis
Treatment and units were randomized as to order for all assays and experiments.
Biological replicates were based on independent samples (n). All experiments were repeated at least three times unless otherwise noted. After testing for normality, multiple comparisons of the data were made using one-way ANOVA with Bonferroni or Tukey's post hoc tests, as indicated. Paired comparison was made using Student's t-tests. Pvalues <0.05 were considered the minimum for significance (as rejection of the null hypothesis). Unless otherwise noted, data were analyzed and graphed with GraphPad Prism 5 (GraphPad v5.0b; RRID:SCR_002798) using the appropriate statistical tests.

Results
The N-Aβcore reversed LTP deficits induced by pathological levels of full-length Aβ We have previously shown that the N-terminal Aβ fragment (Aβ 1-15/16 ) enhances synaptic plasticity and contextual fear memory while protecting against Aβ-linked synaptic impairment [13]. Considering the evidence that the N-Aβcore accounts for the neuromodulatory activity of the N-terminal Aβ fragment, we assessed whether the N-Aβcore is capable of reversing Aβ-linked synaptic dysfunction in an ex vivo model. We represented by input-output curves shows that the fEPSP slopes versus stimulus strength for both the 5FAD and B6.SJL mice were comparable (Fig 1A), ruling out any issues with regard to impact of the Aβ fragment peptides on baseline synaptic strength.
Interestingly, treatment with the N-Aβcore during baseline recordings induced increases in baseline synaptic transmissions for both 5FAD and B6.SJL but was only significant in the B6.SJL slices (Fig 1B; average increase relative to untreated controls: 110% ± 2% s.d.).
To assess sustained changes in synaptic plasticity, we used a 3x-TBS stimulation protocol at the Schaffer collaterals to measure LTP in the CA1 region (see cross-sectional diagram of the hippocampus in Fig 1A,

The N-Aβcore reversed full-length Aβ-linked downregulation of hippocampal AMPA-type glutamate receptors
Regulation of synaptic expression of AMPA-type glutamate receptors (AMPARs) has been shown to underlie LTP [23,24]. As downregulation of AMPARs is linked to the impairment in hippocampal LTP in APP/PS1 mice [15], the impact in vitro and in vivo of the N-Aβcore on the regulation of hippocampal AMPARs was assessed.
Utilizing hippocampal neuronal cultures derived from the background wild-type mice, B6.SJL, and treated with exogenous full-length Aβ (Aβ 42 ) for 7 days, the N-Aβcore was shown to alleviate the downregulation of AMPAR1 (GluA1) transcript expression assessed via qPCR in this in vitro Aβ toxicity model (Fig 2A; 22% ± 0.8% s.d. of Aβ 42 condition). There was no significant impact of the N-Aβcore on the modest downregulation of AMPAR2 (GluA2) transcript expression.
Proteins solubilized from hippocampi isolated from 5FAD mice bilaterally injected with the N-Aβcore or saline vehicle were assayed for changes in in vivo expression of hippocampal AMPAR1 (GluA1) via immunoblot analysis. Total GluA1 levels in the hippocampi were increased with exposure to the N-Aβcore as compared to saline-injected controls (Fig 2B; 248% ± 67% s.d.). The increase in relative pAMPAR1 (pS831) in the hippocampi exposed to N-Aβcore was accounted by the increase in total GluA1 (Fig 2B; 127% ± 35% s.d.). Together, these findings indicate that reversal of the LTP impairment in the 5FAD APP/PS1 by the N-Aβcore involves regulation of AMPAR expression.

Structure-function and concentration-dependence of the N-Aβcore in reversing LTP impairment in hippocampal slices from 5FAD APP/PS1 mice
We also tested for basic concentration-dependence of the N-Aβcore in reversing LTP impairment in the 5FAD mouse hippocampal slices. Low concentration (<pM) of the N-Aβcore showed no difference on LTP compared to control 5FAD slices (Fig 1C & 1E; 107% ± 7% s.d. compared to 5FAD).
Through Aβ-interacting receptor-linked Ca 2+ and neurotoxicity assays, we had previously shown that mutating the tyrosine residue in the N-Aβcore to a serine [Y10S] or mutating the two histidine residues to two alanines [H13A,H14A] reduces activity, indicating these amino acid residues in the N-Aβcore sequence are essential for activity [13,19]. To confirm the specificity of the N-Aβcore in reversing LTP impairment in 5FAD compared to 5FAD). It is important to note that there was no change in basal synaptic transmission or a trend toward increasing LTP in the wild-type slices treated with the reverse N-Aβcore (not shown), as seen for the N-Aβcore (Fig 1). Taken together, these results confirm the contribution of the two essential histidine residues to the positive neuromodulatory activity of the N-Aβcore.

N-Aβcore rescue of Aβ-induced LTP deficits involves PI3 kinase and mTOR
LTP has shown to involve multiple protein kinase and phosphatase pathways [25,26]. As prior evidence implicates the PI3 kinase and mTOR pathways in the regulation of Aβ neurotoxicity [27,28] and in the regulation of LTP [29], we evaluated the roles of PI3 kinase and mTOR in the action of the N-Aβcore in reversing impaired LTP via treatment of the hippocampal slice preparations with selective inhibitors. As shown in Figure 3, application of PI3 kinase inhibitor LY294002 had no impact on LTP in hippocampal slices from control (background B6.SJL) mice or 5FAD mice. In contrast, application of LY294002 prior to treatment with the N-Aβcore prevented the rescue by the N-Aβcore of LTP in the slices from 5FAD mice (Fig 3; 94.3% ± 15% s.d. compared to 5FAD).
Prior inhibition of the PI3 kinase pathway effector mTOR by rapamycin also prevented the rescue by the N-Aβcore of LTP in the slices from 5FAD mice (Fig 4; 101% ± 20% s.d. compared to 5FAD). However, rapamycin did reduce LTP in the control B6.SJL slices, to a level similar to that seen for the LTP in 5FAD slices. The inhibitor had no significant effect on the reduced level of LTP in the 5FAD slices.
To further probe the mechanism by which the N-Aβcore regulates the PI3 kinase pathway, the impact of the core peptide on Aβ-linked regulation of the PI3 kinase and its downstream effectors Akt and mTOR was investigated using mouse hippocampal neuron cultures and mouse hippocampal slices. Treatment of neuron cultures with full-length Aβ (1-42) was shown to downregulate expression of various PI3 kinase, Akt and mTOR transcripts (Fig 5). Co-treatment with the N-Aβcore alleviated the Aβ-induced downregulation of Akt1 and mTOR transcripts (Fig 5).

Elevated levels of Aβ enhances long-term depression and the N-Aβcore reverses Aβ-linked LTD enhancement in hippocampal slices from 5FAD APP/PS1 mice
LTD is an essential component of synaptic plasticity underlying memory processing in the hippocampus, as synapses cycle between LTP and LTD, a process known as synaptic scaling [30]. . The N-Aβcore had no effect on metabotropic glutamate receptor-induced LTD (Fig 7). Taken together, these data suggest that Aβ plays a role in facilitating LTD and the N-Aβcore may protect against Aβ-induced LTD enhancement. The role of the NMDA receptor in Aβ facilitation of LTD warrants further investigation.

Discussion
Previous studies have established a strong link between the progression of AD and the extent of synaptic dysfunction occurring in the early stages of the disease, prior to the formation of Aβ plaques and tau neurofibrillary tangles [1,5,34,35]. In AD-like models it has been widely demonstrated that elevated levels of soluble oligomeric Aβ drive LTP inhibition [3,7,8,36], coupled to downregulation of synaptic AMPARs. By contrast, the link between pathological levels of Aβ and LTD are less well understood, and as previously noted, investigations of the impact of Aβ on LTD have had conflicting results.
In accordance with previous findings [36], we found that LTP was near absent in the hippocampal slice model from APP/PS1 5FAD transgenic mice, previously shown to be accounted by elevated Aβ in the brains of the transgenic model mice [20,22].
Treatment here with the N-Aβcore reversed this deficit back to the LTP observed in slices from the background B6.SJL mice. Treatment of 5FAD slices with the N-Aβcore trended towards LTP enhancement, suggesting that the reversal of the LTP deficits in the 5FAD slices by the N-Aβcore was not solely due to competitive binding for target receptors and may possibly involve activation of a neuroprotective pathway that enhances synaptic plasticity [see 37]. Here, we identified the PI3 kinase/Akt/mTOR in the reversal of LTP deficits in 5FAD slices by the N-Aβcore as a primary pathway which has been shown to be a key link to long-term memory [38]. Other downstream pathways engaged by the N-Aβcore are not yet definitely identified, but we suspect that key players involved in synaptic modulation are affected, such as regulation of CREB, PKA, and/or CAMKII or downregulation of calcineurin and/or PP1, subsequently altering AMPA receptor trafficking to the synapses [23,24,39,40], consistent with the observed reversal of downregulation of hippocampal AMPARs by the N-Aβcore in 5FAD mouse hippocampus. Additionally, the enhancement of the basal synaptic transmission with the treatment of the N-Aβcore suggests a receptor-linked influx of Ca 2+ , which further supports the idea that the N-Aβcore activates an alternative neuroprotective pathway that enhances synaptic plasticity, consistent with results for neuroprotection by the N-Aβcore in Aβ-triggered neurotoxicity [19].
Previously, it has been shown that BDNF enhances basal synaptic transmission [41], therefore, we suspect that the N-Aβcore may be upregulating BDNF expression, possibly through a Ca 2+ -dependent increase in CREB activation and/or expression [42,43]. Another possibility is that the N-Aβcore-induced Ca 2+ influx could also regulate BDNF release at the synapses, thus, enhancing baseline synaptic transmission and ultimately LTP. It would be interesting to examine the effect of the basal synaptic transmission by the N-Aβcore long-term, and whether the enhancement of LTP observed required changes in baseline transmission prior tetanic stimulation.
In the context of synaptic plasticity, LTD is necessary for neural homeostasis.
NMDA receptor-dependent LTD often involves internalization of AMPA receptors via a caspase-dependent pathway [16,44]. To date, however, there is limited understanding in regard to the effects of pathological Aβ on LTD, where some groups show that synthetic Aβ enhances LTD [31,32,45] and others show no effect [33,46], though the differential action of Aβ may have resulted from recording in different subregions of the hippocampus (eg. CA1 vs. dentate gyrus). Here, we found that high concentrations of endogenous soluble Aβ shown to be present in the brains of 5FAD mice resulted in enhanced LTD in hippocampal slices, and treatment with the N-Aβcore prior to and during the LFS induction of LTD reverses this enhancement. Interestingly, Hu et. al. found that applying synthetic soluble Aβ prior to LFS did not affect the early phase of LFS-induced LTD (<2h post LFS), but facilitated the late phase (3-5h post LFS) [45], thus, possibly accounting for different findings. It is important to note that late phase LTP and LTD require new protein synthesis, and mTOR is linked to the regulation of protein synthesis [47]. Indeed, the reversal by the N-Aβcore of LTP deficit in the 5FAD slices was dependent upon mTOR. As LTD and LTP work in concert to allow for reversible synaptic plasticity and synaptic scaling, the LFS-induced enhancement of LTD in the 5FAD slices could affect subsequent LTP, and this may be another reason why an LTP deficit was observed in the 5FAD slices compared to wild-type preparations.
Although NMDA and AMPA receptors are involved in different aspects of LTP and LTD, and their expression is affected by elevated Aβ, metabotropic glutamate receptors (mGluRs) have also been implicated in Aβ-induced synaptic dysfunction [9,45,48,49].
Interestingly, NMDA-independent, mGluR-induced LTD was found to be unaffected by the N-Aβcore. While mGluRs have been linked via cellular prion to Aβ-induced cellular toxicity [49], our findings support a divergence in the Aβ-linked signaling pathways affected by the N-Aβcore in synaptic plasticity. Further work is needed to elucidate the detailed molecular mechanisms involved in N-Aβcore protection or reversal of Aβ-linked synaptic dysfunction, including caspase-dependent intracellular pathways leading to the regulation of AMPA receptors, and eventual synaptic loss in AD.

Conclusions
The essential core hexapeptide sequence, YEVHHQ, or N-Aβcore, within the     Bonferroni post hoc tests showed that rescue of the LTP deficit in the 500nM N-Aβcore-