Secreted Amyloid Precursor Protein-Alpha Enhances LTP Through the Synthesis and Trafficking of Ca2+-Permeable AMPA Receptors

Regulation of AMPA receptor expression by neuronal activity and neuromodulators is critical to the expression of both long-term potentiation (LTP) and memory. In particular, Ca2+-permeable AMPARs (CP-AMPAR) play a unique role in these processes due to their transient, activity-regulated expression at synapses. Secreted amyloid precursor protein-alpha (sAPPα), a metabolite of the parent amyloid precursor protein (APP) has been previously shown to enhance hippocampal LTP as well as memory formation in both normal animals and in Alzheimer’s disease models. In earlier work we showed that sAPPα promotes trafficking of GluA1-containing AMPARs to the cell surface and specifically enhances synthesis of GluA1. To date it is not known whether de novo synthesized GluA1 form CP-AMPARs or how they contribute to sAPPα-mediated plasticity. Here, using fluorescent non-canonical amino acid tagging–proximity ligation assay (FUNCAT-PLA), we show that brief treatment of primary rat hippocampal neurons with sAPPα (1 nM, 30 min) rapidly enhanced the cell-surface expression of de novo GluA1 homomers and reduced levels of de novo GluA2, as well as extant GluA2/3-AMPARs. The de novo GluA1-containing AMPARs were localized to extrasynaptic sites and later internalized by sAPPα-driven expression of the activity-regulated cytoskeletal-associated protein, Arc. Interestingly, longer exposure to sAPPα increased synaptic levels of GluA1/2 AMPARs. Moreover, the sAPPα-mediated enhancement of LTP in area CA1 of acute hippocampal slices was dependent on CP-AMPARs. Together, these findings show that sAPPα engages mechanisms which specifically enhance the synthesis and cell-surface expression of GluA1 homomers, underpinning the sAPPα-driven enhancement of synaptic plasticity in the hippocampus.

Our previous work has provided evidence suggesting that the neuroprotective and neurotrophic metabolite of the transmembrane glycoprotein amyloid precursor protein (APP), secreted APP-alpha (sAPPα), may regulate similar processes. sAPPα, as well as peptide derivatives, enhances hippocampal LTP (Ishida et al., 1997;Richter et al., 2018), and rescues memory impairments (Meziane et al., 1998;Mileusnic et al., 2004;Fol et al., 2016;Xiong et al., 2017;Tan et al., 2018). We have shown that sAPPα not only promotes synaptodendritic protein synthesis (Claasen et al., 2009) and new gene transcription (Ryan et al., 2013), but brings about a protein synthesisdependent trafficking of GluA1-AMPARs to the cell surface (Mockett et al., 2019) and enhances Arc protein expression (Livingstone et al., 2019), yet how these changes contribute to the documented sAPPα-driven enhancement of synaptic plasticity in the hippocampus is unresolved. Here, we show that sAPPα rapidly and specifically induces the synthesis of CP-AMPARs which are transported to the cell surface. Further, we show that sAPPα-mediated enhancement of LTP in area CA1 of acute hippocampal slices is dependent on CP-AMPARs. Consistent with the slot-hypothesis of AMPAR expression, the increase in cell surface GluA1 occurs alongside a reduction in cell surface GluA2/3-containing AMPARs and de novo GluA1-homomeric receptors are ultimately replaced by sodium channel forming-GluA1/2 receptors derived from pre-existing pools. Thus sAPPα acts as a neuromodulator by harnessing paradigmatic synaptic plasticity mechanisms, to prime the response of synapses to co-existent activity.

Animals
All experimental protocols were approved by the University of Otago Animals Ethics Committee and conducted in accordance with New Zealand Animal Welfare Legislation under the ethics approval ET18/15 and AUP-18-136 for cell culture work and DET19/16 for all acute slice work. All experiments conducted in primary hippocampal cultures were prepared from postnatal Sprague-Dawley rat pups (male or female, P0-P1) sourced from a breeding colony maintained at the Hercus Taieri Resource Unit by the University of Otago (Dunedin, New Zealand). The preparation of primary hippocampal cultures followed a modified protocol based on Banker and Goslin (1998) and Kaech and Banker (2006). Hippocampi were dissociated using papain (Sigma) and plated at a low density on glass-bottomed culture dishes (40,000 cells/cm 2 ; Mattek), or 96-well assay plates (67,500 cells/cm 2 ; Corning, #3603) for immunolabelling. Cells were cultured in Neurobasal A medium (Life Technologies #10888-022), supplemented with B27 (Life Technologies, #17504-001) and Glutamax (Life Technologies, #35050-061) at 37 • C/5% CO 2 for 21-27 days in vitro (DIV). Control treatments were undertaken at the same time in matched sets of culture dishes (FUNCAT-PLA, immunocytochemistry).

Drugs and Reagents
For experiments examining the role of Ca 2+ -permeable receptors the antagonist IEM-1460 was used (100 µM, Abcam, #AB141507). sAPPα production and purification was carried out according to Turner et al. (2007).

Immunocytochemistry
For experiments examining the effect of siRNA treatment on protein expression, DIV21-27 primary hippocampal neurons were treated with sAPPα (1 nM; 2 h), or culture media only. Following pre-treatment with siRNA (1 µM, 60 min), primary hippocampal cultures were co-treated with either Arc or NT siRNA (1 µM, 2 h), in the presence or absence of sAPPα (1 nM).

PLA
The labeling of cell-surface receptor subunit dimers was conducted as per (Lin et al., 2015) and adapted for the detection of AMPARs present at the cell surface. In brief, cells were treated with sAPPα (1 nM) or existing media. Following incubation, cells were washed with PBS-MC (pH 7.4), and fixed in PFA in PBS-MCS (pH 7.4). For experiments investigating GluA1/2containing AMPARs, using both C-terminal and N-terminal antibodies, cells were incubated in blocking buffer (4% normal goat serum in PBS, 60 min at RT), probed with the N-terminal antibody (90 min, RT), washed (PBS pH 7.4, 3 × 5 min), and subsequently permeabilized with 0.5% Triton X-100 in PBS (15 min, RT). Cells were then probed with the C-terminal antibody in addition to anti-MAP2 for visualization of neuronal structure (90 min, RT) and washed (PBS pH 7.4, 3 × 5 min). For experiments utilizing two N-terminal antibodies, such as those investigating GluA2/3-containing AMPARs, permeabilization was omitted and cells were incubated in blocking buffer (4% normal goat serum in PBS, 60 min at RT) followed by incubation of primary antibodies (90 min, RT). Following this, cells were washed (PBS pH 7.4, 3 × 5 min), permeabilized with 0.5% Triton X-100 in PBS (15 min, RT) and incubated in blocking buffer again (4% normal goat serum in PBS, 60 min at RT) before addition of anti-MAP2 and DAPI for visualization of neuronal structure (90 min, RT). Following addition of antibodies, proximity ligation assay was performed as described above.

Field Potential Electrophysiology
All experiments conducted on acute tissue were prepared from young adult male Sprague-Dawley rats (42-56 d), as described previously (Mockett et al., 2019). Rats were deeply anesthetized with ketamine (100 mg/kg, i.p.) and decapitated via guillotine. The brains were removed and chilled in ice-cold and oxygenated modified artificial cerebrospinal fluid (aCSF) for which sucrose was substituted for NaCl (composition in mM: sucrose 210, glucose 20, KCl 2.5, NaH 2 PO 4 1.25, NaHCO 3 26, CaCl 2 0.5, MgCl 2 3, pH 7.4 when gassed with 95% O 2 -5% CO 2 ). Hippocampi were dissected and slices (400 µm) cut using a vibroslicer (Leica, VT1000). Slices were transferred to a porous, transparent membrane in an incubation chamber, and maintained at the interface between air and standard aCSF (in mM: NaCl 124, KCl 3.2, NaH 2 PO 4 1.25, NaHCO 3 26, CaCl 2 2.5, MgCl 2 1.3, D-glucose 10, equilibrated with carbogen 95% O 2 -5% CO 2 ; 32 • C) for 30 min followed by RT for an additional 90 min. Slices were then transferred to the recording chamber containing recirculating aCSF (95% O 2 , 5% CO 2 ; 32.5 • C), superfused continuously at a rate of 2 mL/min. Baseline field excitatory postsynaptic potentials (fEPSPs) were elicited in area CA1 by stimulation of the Schaffer collateral-commissural pathway at 0.017 Hz (diphasic pulses, 0.1 ms half-wave duration) using a Teflon-coated 50 µm tungsten wire monopolar electrode (A-M Systems Inc., Carlsborg, WA) connected to Grass P511 AC amplifiers via high impedance probes (Grass Instruments Company, West Warwick, United States). Signals were amplified (x1000) with half-amplitude filter cut-offs of 0.3 Hz and 3 kHz. Evoked responses were recorded with a glass microelectrode (A-M systems, 1.0 mm x 0.58 mm, 4"; Catalog No. #601000) filled with aCSF (1.9-2.9 M ) and placed in stratum radiatum of area CA1 (approximately 300 µm from the stimulating electrode). During periods of baseline recording the stimulation intensity was adjusted to elicit a fEPSP with an initial slope value of 40% of the maximum elicited when delivering 200 µA of current. Non-saturated LTP was induced by applying a mild theta burststimulation protocol (TBS; 5 bursts of 5 pulses at 100 Hz delivered at 200 ms intervals) at baseline stimulus intensity as per Mockett et al. (2019). sAPPα (1 nM) and IEM-1460 (100 µM) were bathapplied by switching to an identical preheated and oxygenated aCSF solution that contained the compound of interest. sAPPα was delivered 30 min prior to TBS and IEM-1460 was delivered 20 min prior to and during the sAPPα administration and continued 10 min post-TBS. Control experiments were routinely interleaved randomly between experimental treatments.

Image Analysis
To quantify the PLA signal a custom-made ImageJ script created by Maximilian Heumüller (Max Planck Institute for Brain Research, Frankfurt) was used (tom Dieck et al., 2015). Following thresholding of the FUNCAT-PLA signal, a 'MAP2 mask' was generated capturing the area of the neuron, and the size and intensity of FUNCAT-PLA puncta within the mask were measured and recorded. To analyze the signal within the somatic and dendritic compartments, somata were isolated from each cell as above, and the proximal 50 µm segment all dendrites were straightened using the 'straighten' plugin in ImageJ, and analyzed for PLA signal. For the detection of somatic receptor subunit dimers using PLA, analysis included a step to exclude non-specific signal detected within the nucleus using a mask generated by the DAPI channel. For image representation, ImageJ was used to adjust the brightness and contrast equally for all treatment groups. To quantify immunofluorescence, neurons were outlined using ImageJ. An 'integrated intensity/neuronal area' value was generated for each cell and somatic compartment, including all dendrites up until intersection with neighboring dendrites. This value was corrected for average background fluorescence by subtracting mean gray values of background fluorescence.

Colocalization
To analyze the proximity of PLA signal to the synaptic marker synapsin-1, the ImageJ plugin Just Another Colocalization Plugin (JACoP) was used. In brief, images were separated into single channel images and the proximal 50 µm of MAP2 and PLA signal was isolated. Average PLA and synapsin-1 signal was determined for all treatment groups and applied within the JACoP plugin. In order to determine colocalization, the Mander's overlap coefficient was used to generate a value proportional to the level of overlap of PLA on synapsin-1 signal, of which a value of -1 indicates negative correlation, 0 indicates no correlation and + 1 indicates a positive correlation. In addition, the proximity of PLA puncta to the synapsin-1 signal was determined by manual measurements of the center of each PLA puncta (determined by the brightest pixel) to the center of mass of the nearest synapsin-1 puncta. Distances defining synaptic (0-2 µm), extrasynaptic (2-4 µm), and non-synaptic (4 + µm) puncta were determined based on the size and observed overlap of the PLA and synapsin-1 puncta (approximately 1 µm; Jarvius et al., 2006;Fantuzzo et al., 2017) at the resolution used.

FUNCAT-PLA & PLA
Prior to data amalgamation, outliers were removed from each experiment using Grubb's tests, and normality was assessed (D'Agostino & Pearson omnibus normality tests). To examine the expression of newly synthesized proteins and AMPAR complexes at the cell surface using FUNCAT-PLA and PLA, respectively, significance was calculated on amalgamated data per experiment expressed relative to each experimental control by use of one sample t-tests (having met the assumptions of independence of observation, homogeneity of variances and approximation to a normal distribution) and F tests to compare variances were applied to datasets when applicable. To examine the effect of sAPPα treatment (1 nM, 30 min) relative to sAPPα (1nM, 2 h) the data were assessed by Mann-Whitney twotailed U-test. The effect of siRNA was assessed by expressing the normalized sAPPα values + siRNA relative to siRNA alone. To determine whether there was a significant difference between treatment groups and the mean of the controls, data were assessed by use of one sample t-tests. To determine whether there was a significant difference between the mean effect of sAPPα alone (fold change relative to control) and the siRNA (sAPPα + siRNA/siRNA), the data were assessed by Mann-Whitney two-tailed U-test. For experiments examining the expression of proteins relative to synaptic puncta, normality was determined by Shapiro-Wilk normality test and significance of the Mander's overlap coefficient was determined by unpaired t-test. For the comparison of synaptic, extrasynaptic, and nonsynaptic puncta significance was assessed by two-way ANOVA and Šidák's multiple comparisons.

Immunocytochemistry
Statistics for all immunocytochemistry experiments were performed on raw data using Kruskal-Wallis one-way ANOVA followed by Dunn's multiple comparisons test. Data were not normally distributed (D'Agostino & Pearson normality test). Outliers within the raw data sets were detected using the Grubb's test.

Electrophysiology
The initial slopes of the fEPSPs were analyzed offline as a measure of synaptic efficacy and expressed as a percentage change from the baseline level. To account for a small but non-significant effect of IEM-1640 on baseline synaptic transmission (−8.3 ± 6.71%, p = 0.1127, n = 6, one-way ANOVA with Tukey's multiple comparisons test), the induction of LTP (0-10 min post-TBS, in the presence of IEM-1460) was calculated as the average of the last 10 min of the baseline recording period prior to the application of TBS. Following washout of IEM-1460, responses measuring the persistence of LTP (50-60 min post-TBS) were normalized to the baseline value prior to the addition of IEM-1460. Group means were expressed as the percentage change ± SD. Data acquisition and analysis were performed by custom-built software developed in the Department of Psychology, University of Otago. Normal distribution of data was determined by D'Agostino and Pearson omnibus normality test. Statistical differences between experimental groups was determined by oneway ANOVA followed by Tukey's multiple comparisons tests at the p < 0.05 significance level.

RESULTS
sAPPα Promotes the Rapid and Transient Trafficking of de novo GluA1-Containing AMPARs to the Somatic and Dendritic Cell Surface It is not known whether sAPPα induces the rapid synthesis of GluA1-containing AMPARs or whether these de novo molecules are trafficked to the cell surface and thus contribute to plasticity. Thus, to examine specifically the cell-surface population of de novo GluA1, we utilized FUNCAT-PLA under detergent-free conditions. We found that sAPPα treatment (1 nM; 30 min) significantly increased the cell-surface expression of de novo GluA1 in both the soma (2.01 ± 0.33 mean ± SEM, p = 0.0039; Figure 1A) and dendrites (3.97 ± 0.32, p = 0.0009; Figure 1B) of cultured hippocampal neurons. Interestingly, following 2 h of sAPPα treatment, there was no detectable change in de novo somatic (0.83 ± 0.15, p = 0.265; Figure 1A) or dendritic (0.86 ± 0.29, p = 0.673; Figure 1B) GluA1. Indeed, cell-surface expression of GluA1 at 2 h was significantly decreased relative to 30 min of sAPPα treatment on both the soma (p = 0.0058) and dendrites (p = 0.0005). Interestingly, in parallel experiments using immunocytochemistry with non-permeabilized conditions to allow detection of both extant and newly made cell surface GluA1, we were able to detect increases in GluA1 in the dendrites. We observed a modest increase by 30 min (2.37 ± 0.37, p = 0.0787), which reached statistical significance by 2 h (3.00 ± 0.44, p = 0.0002) and remained significantly elevated at 4 h (2.50 ± 0.39, p = 0.0204). We found no changes in the soma at any time point (30 min: 1.09 ± 0.12, p = 0.8998; 2 h: 1.066 ± 0.82, p = 0.2694, 4 h: 1.043 ± 0.15, p = 0.3191; see Supplementary Figure 1). Together, these results suggest that sAPPα induces the rapid de novo synthesis of GluA1 subunits, which are trafficked to the cell surface by 30 min. These newly made subunits are later internalized within 2 h of treatment, yet levels of extant GluA1-containing receptors remain elevated for up to 4 h.

Rapid Decrease in Dendritic de novo Cell Surface GluA2 Following sAPPα Treatment
As the GluA1 subunit is a constituent of either GluA1/2 or GluA1-homomeric receptors, to determine the likely composition of de novo AMPARs formed following sAPPα treatment, we next examined the temporal expression of de novo GluA2 at the cell surface, in both the somatic and dendritic compartments. Here, we found that de novo cell-surface GluA2 remained unaffected at the soma by either 30 min (0.86 ± 0.28, p = 0.654) or 2 h (1.32 ± 0.21, p = 0.148; Figure 2A) sAPPα treatments. Conversely, there was an early decrease of de novo cell-surface GluA2 at the dendritic cell surface (30 min: 0.54 ± 0.17, p = 0.009), which returned to control levels within 2 h (1.92 ± 1.08, p = 0.395; Figure 2B). From this, we can infer that while sAPPα upregulates the expression of de novo GluA1, it inhibits the synthesis or otherwise restricts expression of de novo GluA2 at the cell surface. Together, these results Outliers were removed from each experiment prior to amalgamation using Grubb's tests, and normality was detected by D'Agostino and Pearson omnibus normality tests. All data are expressed relative to control, as mean ± SEM from 3 experiments. Significance relative to control was assessed by student's t-test, significance between sAPPα-treated groups was assessed by Mann-Whitney two-tailed U-test. **p ≤ 0.01, ***p ≤ 0.001. Representative images show neuronal soma (upper panels) and dendrites (lower panels; MAP2, ; GluA1, ; DAPI, ). Scale bars = 10 µm.
indicate that sAPPα acts both to upregulate cell surface de novo GluA1-containing homomeric AMPARs and downregulate de novo GluA2-containing AMPARs.
the extrasynaptic pool of de novo GluA1-containing AMPARs within the dendrites.
As de novo GluA1 levels at the cell surface did not persist following prolonged (2 h) treatment of sAPPα, we next employed siRNA knockdown of de novo Arc protein to assess whether inhibition of Arc protein synthesis affects de novo GluA1 cell surface levels. As expected, sAPPα treatment for 2 h resulted in no change in either somatic (0.73 ± 0.16, p = 0.118) or FIGURE 3 | Secreted amyloid precursor protein-alpha enhances de novo GluA1 at the extrasynaptic membrane. Representative images showing de novo cell-surface GluA1 in (A) control and (B) sAPPα-treated conditions. Representative images show dendrites (50 µm; synapsin-1, ; GluA1, ). Scale bar = 10 µm. (C) No significant difference was observed in the Mander's overlap coefficient following sAPPα-treatment (1 nM, 30 min; n = 20 cells, 58-65 dendrites). Significance was assessed by Mann-Whitney two-tailed U-test. (D) Frequency histogram of the distribution of de novo GluA1 puncta in relevance to synapsin-1 immunofluorescence. Distances were calculated for individual PLA puncta with respect to the closest synapsin-1 center of mass. Synaptic PLA showed overlapping centers within 0-2 µm, puncta within 2-4 µm were considered extrasynaptic, and puncta beyond 4 µm were considered non-synaptic (n = 58-65 dendrites, 130-205 puncta). (E) Quantification of GluA1 puncta abundance at the synaptic, extrasynaptic, and non-synaptic membrane (n = 58-65 dendrites, 130-205 puncta). Normality was determined by Shapiro-Wilk normality test. All data are expressed relative to control, as mean ± SEM from 3 experiments. Significance was assessed by two-way ANOVA and Šidák's multiple comparisons, ***p ≤ 0.0005. dendritic (0.85 ± 0.85, p = 0.374) expression of de novo cell surface GluA1, relative to control conditions (Figures 4A,B). However, in the presence of Arc siRNA (preincubation: 1 µM, 60 min, followed by co-incubation with sAPPα: 1 nM, 2 h), we observed a significant increase in the expression of de novo GluA1 at both the somatic (1.55 ± 0.36, p = 0.0043; Figure 5A) and dendritic (4.14 ± 0.65, p ≤ 0.0001; Figure 5B) cell surface, relative to sAPPα-only conditions. Finally, treatment of cultures with the NT siRNA alone had no observable effect on the levels of de novo cell surface GluA1 at the soma (0.94 ± 0.15, p = 0.7247, n = 23; Supplementary Figure 2A Figure 2B), indicating that the observed effects are due to the specific to the addition of Arc siRNA and subsequent knockdown of Arc protein expression and not as a result of non-specific siRNA-mediated effects. Therefore, these results show that following the rapid expression of de novo GluA1-containing AMPARs at the cell surface, these AMPARs are subsequently endocytosed in an Arc-dependent manner.
While previous research examining acute sAPPα treatments (Mockett et al., 2019) found no evidence for an increase in de novo GluA2 synthesis in response to sAPPα, this does not exclude the possibility that sAPPα additionally promotes the trafficking of pre-existing GluA2-containing AMPARs to the cell surface. Therefore, we extended our analysis to examine cell surface populations of GluA1-and GluA2-containing (GluA1/2) AMPARs within somatic and dendritic compartments. Here, we utilized PLA to detect the coincident proximity of GluA1 and GluA2 AMPAR complexes at the cell surface. We found no significant increase in GluA1/2-containing AMPARs at the cell surface following a 30 min sAPPα treatment, in either the soma (1.04 ± 0.11, p = 0.709; Figure 6A) or dendrites (1.05 ± 0.09, p = 0.54; Figure 6B) of cultured primary hippocampal neurons. However, the cell surface expression of GluA1/2 AMPARs significantly increased in both the soma (2.15 ± 0.30, p = 0.0013; Figure 6A), and dendrites (2.28 ± 0.14, p ≤ 0.0001; Figure 6B) by 2 h. These results indicate that GluA1/2 AMPARs are expressed at the cell surface following prolonged sAPPα treatment. Outliers were removed from each experiment prior to amalgamation using Grubb's tests, and normality was detected by D'Agostino and Pearson omnibus normality tests. All data are expressed relative to control, as mean ± SEM from 3 experiments. Data assessing the effect of siRNA are expressed as mean ratio of sAPPα + Arc siRNA/Arc siRNA alone ± SEM. Significance between control and treatment was assessed by student's t-test, significance between sAPPα-treated and sAPPα + siRNA-treated was assessed by Mann-Whitney two-tailed U-test, **p ≤ 0.001, ****p ≤ 0.0001. Representative images show neuronal soma (upper panels) and dendrites (lower panels; MAP2, ; GluA1, ; DAPI, ). Scale bars = 10 µm.
previous research has shown that GluA3-containing AMPARs are regulated by in vivo LTP (Williams et al., 2007), and in vitro LTD (Holman et al., 2007), and growth factor treatment (Narisawa-Saito et al., 1999). Curiously, GluA3-containing AMPARs do not appear to directly regulate the expression of LTP or context fear memory formation (Meng et al., 2003;Humeau et al., 2007), but their removal from the synapse may be an essential step required for LTD (Holman et al., 2007). Regardless, their synaptic expression is considered essential for basal synaptic transmission (Meng et al., 2003).
We next examined whether the enhancement of LTP induction following sAPPα treatment (1 nM, 30 min) was dependent on the function of CP-AMPARs. Here, slices were perfused with IEM-1460 (100 µM, 50 min) before TBS, and 10 min post-TBS stimulation. Slices were also perfused with sAPPα 30 min before TBS in the presence or absence of IEM-1460. As shown in Figure 9, relative to control slices, pre-treatment with sAPPα (1 nM, 30 min) enhanced both the induction (126.2 ± 13.55% of baseline, p ≤ 0.0001; Figures 9C,D) and persistence (77.41 ± 10.66% of baseline, p = 0.0002; Figures 9C,E) of LTP, confirming previously observed enhancements described by Mockett et al. (2019). Co-application of IEM-1460 significantly inhibited the sAPPα-enhanced early potentiation (84.19 ± 11.74% of baseline, p = 0.0334; Figures 9C,D), while wash-out of IEM-1460 resulted in the recovery of potentiation, which was no longer different from sAPPα treatment alone (70.78 ± 15.44% of baseline, p = 0.3518; Figures 9C,E), but was significantly different from control slices (p = 0.0149). These results indicate that at least the initial period of the sAPPα-mediated enhancement of LTP is due to the rapid incorporation of functional CP-AMPARs into the synapse.

DISCUSSION
Secreted amyloid precursor protein-alpha plays a crucial role in synaptic plasticity and memory. Knockdown of APP or inhibition of sAPPα significantly reduces LTP in the dentate gyrus and impairs spatial memory and inhibitory avoidance in rats and passive avoidance in chicks (Mileusnic et al., 2000;Taylor et al., 2008). Similarly, inhibition of α-secretase activity impairs hippocampal LTP and spatial memory (Taylor et al., 2008). Importantly, these deficits are restored by either acute administration of sAPPα or peptide derivatives (Taylor et al., 2008;Morrissey et al., 2019) or by genetic overexpression (Ring et al., 2007;Tan et al., 2018). Understanding how sAPPα orchestrates these effects is a critical step if the administration of sAPPα or derivatives is to be considered as a therapeutic strategy in neurodegenerative diseases. Mounting evidence suggests that sAPPα regulates LTP and memory through the regulation of new gene transcription, protein synthesis, and glutamate receptor trafficking (Richter et al., 2018;Mockett et al., 2019). In the current experiments, we have shown that sAPPα differentially  (2), and 60 min post-TBS (3). Scale bar = 1 mV, 5 ms. (C) Average traces of control slices, receiving only a mild TBS (5 bursts at 5 Hz, 5 pulses per 100 Hz/burst; G, n = 8 slices), IEM-1460 (100 µM; G, n = 6 slices), sAPPα (1 nM; G, n = 6 slices) and sAPPα + IEM-1460-treated (G, n = 6 slices) conditions. All conditions are normalized to the average of baseline 10 min before TBS, and all data are presented as mean% change ± SD. Data shows summary histograms for (D) early and (E) late potentiation following TBS. To account for the minor effect of IEM-1460 on baseline synaptic transmission, data examining early potentiation in the presence of IEM-1460 are averaged from the first (D) 10 min post-TBS for each experimental group relative to baseline, averaged 10 min before TBS. Following washout late potentiation was measure from the final (E) 10 min post-TBS relative to baseline, averaged 10 min before application of IEM-1460. *p = 0.0334, # p = 0.0149, ### p = 0.0002, #### p ≤ 0.0001. Normality was detected by Shapiro-Wilk normality test. Significance was assessed by one-way ANOVA followed by Tukey's multiple comparisons tests. Hashes (#) indicate significance between control and sAPPα-treated; asterisks (*) indicate significance between sAPPα-and sAPPα + IEM-1460-treated. SC = Schaffer collaterals, s. radiatum = Stratum radiatum.
regulates the synthesis and cell surface expression of AMPARs in hippocampal neurons. Importantly, this includes the rapid synthesis and trafficking of de novo GluA1 homomeric AMPARs, crucial to the sAPPα-driven enhancement of LTP.

Regulated Synaptic Expression of CP-AMPARs
Secreted amyloid precursor protein-alpha enhances the synthesis of GluA1 but not GluA2 subunits (Mockett et al., 2019). Although sAPPα alters transcription profiles (Stein et al., 2004;Ryan et al., 2013), it has not been shown to alter levels of AMPAR subunit mRNA within a 2 h treatment window. sAPPα does, however, enhance protein synthesis within isolated hippocampal synapses (Claasen et al., 2009), and it is well established that AMPAR subunits, including GluA1, undergo synthesis within dendrites (Torre and Steward, 1992;Kacharmina et al., 2000;Grooms et al., 2006; for direct exocytosis at extrasynaptic sites (Choquet, 2018). Thus, it is likely that the rapid de novo synthesis of GluA1 results from translation of pre-existing mRNA. It is yet to be determined whether sAPPα-induced GluA1 synthesis occurs within dendrites or is delivered to dendrites via lateral diffusion from somatic sites (Borgdorff and Choquet, 2002;Adesnik et al., 2005) or active transport throughout the dendrites (Passafaro et al., 2001;Hangen et al., 2018).

Role of de novo CP-AMPARs in sAPPα-Enhanced LTP
We have shown that de novo GluA1 is trafficked to extrasynaptic sites and is later internalized, but not degraded, as intracellular levels remain elevated (Mockett et al., 2019). As extrasynaptic AMPARs are necessary for the induction of LTP (Oh et al., 2006;Penn et al., 2017) and increasing the proportion of extrasynaptic AMPARs enhances the expression of LTP (Oh et al., 2006), the enhanced GluA1 expression likely provides reservoirs of CP-AMPARs which could underpin the sAPPα-mediated priming of LTP. Indeed, we confirmed that sAPPα does not affect responses to baseline stimulation (Richter et al., 2018;Mockett et al., 2019), nor did IEM-1460 affect baseline responses in sAPPα-treated slices any more than control slices, all supporting the view that the enhancement of cell-surface GluA1 expression by sAPPα is strictly extrasynaptic. However, pre-treatment with sAPPα did enhance LTP induction in a manner that was dependent on CP-AMPARs, likely due to lateral diffusion of the recently inserted extrasynaptic CP-AMPARs.

Regulation of GluA2-Containing AMPARs
The observed elevation in extrasynaptic de novo GluA1 occurs alongside a reduction in cell-surface de novo GluA2 and extant dendritic GluA2/3 while no cell-wide reduction in levels was evident. As GluA2 mRNA is also expressed in dendrites, local downregulation of GluA2 synthesis may be controlled locally via the RNA interacting protein, CPEB3 (Savtchouk et al., 2016) or FMRP (Muddashetty et al., 2007), or via microRNA (miRNA). Indeed, we have previously shown that sAPPα transiently upregulates miR-30 (Ryan et al., 2013), a miRNA known to regulate the expression of GluA2 (Song et al., 2019). Alternatively, association with Protein Interacting with C Kinase-1 (PICK1) may restrict GluA2 trafficking (Terashima et al., 2004;Jaafari et al., 2012) or AP2-GluA2 binding may also promote internalization (Hanley, 2018). Furthermore, the NO-cGMP-PKG and MAPK pathways have been found to be involved in de-clustering of GluA2/3-containing AMPARs (Endo and Launey, 2003), an event which is typically associated with subsequent internalization of receptors (Matsuda et al., 2000). As synaptic protein synthesis, expression of Arc protein, and GluA1 trafficking are dependent on the activity of PKG and MAPK (Claasen et al., 2009;Livingstone et al., 2019;Mockett et al., 2019), sAPPα's induction of these pathways may also contribute to GluA2/3 receptor internalization. Further, rapid downregulation of GluA2-containing calcium impermeable AMPARs (CI-AMPARs) may facilitate GluA1 cellsurface expression via increasing the available "slots" within the PSD (MacDougall and Fine, 2013) to be filled by CP-AMPARs Lüscher, 2005, 2006;Hong et al., 2013). Indeed, in vitro live-cell imaging shows simultaneous downregulation of cell-surface GluA2/3 and upregulation of GluA1-containing AMPARs following LTP induction (Tanaka and Hirano, 2012).
Interestingly, we also observed a delayed increase in the proportion of GluA1/2 AMPARs following sAPPα treatment, at both the somatic and dendritic cell-surface primarily at synaptic sites. These receptors were likely derived from preexisting pools as they were not tagged in the FUNCAT-PLA assays. Previous work has shown that chronic overexpression of APP in vitro upregulates cell-surface GluA2 expression (Lee et al., 2010), indicating that APP, and likely sAPPα, may regulate the persistence of synaptic plasticity through the long-term expression of GluA2-containing AMPARs. As their increase occurs at a timepoint at which both de novo GluA1 and extant GluA2/3 are internalized, this would result in a switch to synapses expressing GluA1/2 CI-AMPARs (Shi et al., 2001;Bellone and Lüscher, 2005;Mameli et al., 2007), as observed during reconsolidation of memories (MacDougall and Fine, 2013). As we have shown that sAPPα enhances Arc expression and that internalization of de novo GluA1 is Arc dependent, Arc expression likely facilitates the addition of synaptic GluA1/2containing AMPARs (McCormack et al., 2006;Hong et al., 2013). Indeed, previously knockdown of Arc has been shown to increase cell-surface GluA1-but not GluA2-containing AMPARs (Shepherd et al., 2006), while overexpression of Arc protein significantly decreases CP-AMPAR-associated rectification in neurons (DaSilva et al., 2016), a mechanism which may permit addition of CI-AMPARs. Recently, increased Arc expression has been shown to protect neurons against CP-AMPARmediated oxyhemoglobin excitotoxicity (Chen et al., 2020), via endocytosis of these AMPARs. While reduced phosphorylation of GluA1 S845 may also contribute to internalization (Man et al., 2007;Lussier et al., 2015), our observations support a role for Arc-dependent endocytosis in the regulation of de novo CP-AMPARs.
Of note, sAPPα has been previously shown to enhance LTP and plasticity-related protein synthesis via a mechanism involving α7nAChRs (Richter et al., 2018;Livingstone et al., 2019) in that antagonism of α7nAChRs during the priming phase, but not during the induction of LTP, impairs the expression of sAPPα-enhanced LTP (Richter et al., 2018). Our current work suggests that this occurs by promoting the synthesis and trafficking of CP-AMPARs. Indeed, activation of postsynaptic α7nAChRs by acute nicotine treatment promotes the rapid trafficking of CP-AMPARs (Lozada et al., 2012;Halff et al., 2014;Tang et al., 2015), while chronic α7nAChR activation enhances the cell-surface expression of GluA1/2 AMPARs (Duan et al., 2015).

CONCLUSION
Many pathological changes in Alzheimer's disease are associated with changes in the expression of AMPA-subtype glutamate receptors, the regulation of which is critical to memory formation. We have shown one mechanism by which sAPPα enhances LTP is by de novo synthesis of GluA1-containing CP-AMPARs which are trafficked to extrasynaptic sites. Simultaneously, sAPPα promotes the removal of GluA2/3containing AMPARs, increasing the number of available synaptic slots, and permitting the trafficking CP-AMPARs from extrasynaptic domains in response to stimulation events which enhance synaptic transmission. The synthesis of Arc protein facilitates removal of synaptic AMPARs which are replaced by GluA1/2-containing CI-AMPARs. Future research should aim to investigate the contribution of Ca 2+ -permeable and GluA2-containing AMPAR to sAPPα-enhanced LTP, and the role of excitatory and inhibitory neurons in this mediating these effects. This work is crucial to understanding the mechanisms harnessed by sAPPα to promote memory formation and further inform the development of alternative treatment strategies for Alzheimer's disease.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.

ETHICS STATEMENT
The animal study was reviewed and approved by University of Otago Animals Ethics Committee and conducted in accordance with New Zealand Animal Welfare Legislation under the ethics approval ET18/15 and AUP-18-136 for cell culture work and DET19/16 for all acute slice work.

AUTHOR CONTRIBUTIONS
RL: was the major contributor to the experimental aspects of the study, contributed to the design of the study, prepared the primary neuronal cultures, carried out immunocytochemistry, FUNCAT-PLA, PLA, and electrophysiology experiments, and analyzed and interpreted the corresponding data and drafted the manuscript. ME: participated in the design and co-ordination of the study and critically assessed the manuscript. AS: participated in the design and co-ordination of the study and critically revised the manuscript. CW: carried out immunocytochemistry and corresponding data analysis. WT: co-ordinated sAPPα and critically revised the manuscript. WA: participated in the design and co-ordination of the study, undertook data analysis and interpretation, and critically revised the manuscript. JW: conceived and participated in the design and co-ordination of the study, undertook data analysis and interpretation, and critically assessed the manuscript. All authors contributed to the article and approved the submitted version.

FUNDING
This research was supported by grants from the Health Research Council of New Zealand to JW, WA, and AS, and by University of Otago postgraduate scholarships to RL and ME.