Impairment of Spike-Timing-Dependent Plasticity at Schaffer Collateral-CA1 Synapses in Adult APP/PS1 Mice Depends on Proximity of Aβ Plaques

Alzheimer’s disease (AD) is a multifaceted neurodegenerative disorder characterized by progressive and irreversible cognitive decline, with no disease-modifying therapy until today. Spike timing-dependent plasticity (STDP) is a Hebbian form of synaptic plasticity, and a strong candidate to underlie learning and memory at the single neuron level. Although several studies reported impaired long-term potentiation (LTP) in the hippocampus in AD mouse models, the impact of amyloid-β (Aβ) pathology on STDP in the hippocampus is not known. Using whole cell patch clamp recordings in CA1 pyramidal neurons of acute transversal hippocampal slices, we investigated timing-dependent (t-) LTP induced by STDP paradigms at Schaffer collateral (SC)-CA1 synapses in slices of 6-month-old adult APP/PS1 AD model mice. Our results show that t-LTP can be induced even in fully developed adult mice with different and even low repeat STDP paradigms. Further, adult APP/PS1 mice displayed intact t-LTP induced by 1 presynaptic EPSP paired with 4 postsynaptic APs (6× 1:4) or 1 presynaptic EPSP paired with 1 postsynaptic AP (100× 1:1) STDP paradigms when the position of Aβ plaques relative to recorded CA1 neurons in the slice were not considered. However, when Aβ plaques were live stained with the fluorescent dye methoxy-X04, we observed that in CA1 neurons with their somata <200 µm away from the border of the nearest Aβ plaque, t-LTP induced by 6× 1:4 stimulation was significantly impaired, while t-LTP was unaltered in CA1 neurons >200 µm away from plaques. Treatment of APP/PS1 mice with the anti-inflammatory drug fingolimod that we previously showed to alleviate synaptic deficits in this AD mouse model did not rescue the impaired t-LTP. Our data reveal that overexpression of APP and PS1 mutations in AD model mice disrupts t-LTP in an Aβ plaque distance-dependent manner, but cannot be improved by fingolimod (FTY720) that has been shown to rescue conventional LTP in CA1 of APP/PS1 mice.


Introduction
Alzheimer's disease (AD) is an age-related, multifaceted neurodegenerative disorder characterized by a progressive and irreversible cognitive decline. It is the most common cause of dementia and currently there is no disease-modifying therapy [1]. The main pathological hallmarks in AD are the presence of extracellular amyloid beta (Aβ) plaques, consisting of Aβ protein oligomers aggregates (Aβo, residues 1-40/42) and the accumulation of neurofibrillary tangles within neurons, composed of abnormally hyperphosphorylated tau protein [2][3][4][5][6]. The Aβ plaques induce microglial activation, cytokine release, reactive astrocytosis, and subsequently an induction of chronic neuroinflammation, leading to increased levels of pro-inflammatory cytokines, neurotoxic tryptophan metabolites (kynurenines), and anti-inflammatory cytokines (compare review [7]). Importantly, Int. J. Mol. Sci. 2021, 22, 1378 3 of 20 in [27]), it has not been previously reported whether AD alters spike timing-dependent plasticity (STDP) in the hippocampal formation. We used whole-cell patch clamp recordings in single postsynaptic cornu ammonis (CA) 1 pyramidal neurons from 6-month-old male C57Bl6/J wild-type (WT), APP/PS1 mice [31] and their WT littermates to study STDP at fully developed Schaffer collateral (SC)-CA1 synapses in 6-month-old adult animals. We tested different STDP paradigms to induce synaptic plasticity by repeatedly pairing a single presynaptic stimulation with either 1 or 4 postsynaptic action potentials (APs) with 10 ms interval (spike timings, ∆t).

Low Repeat STDP Paradigms Induced Timing-Dependent (t-) LTP in Adult WT Mice
To the best of our knowledge, STDP has been studied previously exclusively in 4-6 weeks old juvenile or adolescent animals [25,32,33]. To explore STDP properties in adult AD model mice, we first had to establish STDP paradigms that successfully induce t-LTP in adult WT mice. To this aim we investigated in a first series of experiments a low repeat (6× at 0.5 Hz) STDP paradigm that we previously established in juvenile WT mice [34], to induce t-LTP in 6-month-old (adult) C57Bl6/J mice (Figures 1 and 2A).

Results
Although synaptic dysfunctions such as alterations in conventional long-term potentiation (LTP) and long-term depression (LTD) of hippocampal synaptic transmission have been described as an early event in Alzheimer disease (AD) mouse models (reviewed in [27]), it has not been previously reported whether AD alters spike timing-dependent plasticity (STDP) in the hippocampal formation. We used whole-cell patch clamp recordings in single postsynaptic cornu ammonis (CA) 1 pyramidal neurons from 6-month-old male C57Bl6/J wild-type (WT), APP/PS1 mice [31] and their WT littermates to study STDP at fully developed Schaffer collateral (SC)-CA1 synapses in 6-month-old adult animals. We tested different STDP paradigms to induce synaptic plasticity by repeatedly pairing a single presynaptic stimulation with either 1 or 4 postsynaptic action potentials (APs) with 10 ms interval (spike timings, Δt).

Low Repeat STDP Paradigms Induced Timing-Dependent (t-) LTP in Adult WT Mice
To the best of our knowledge, STDP has been studied previously exclusively in 4-6 weeks old juvenile or adolescent animals [25,32,33]. To explore STDP properties in adult AD model mice, we first had to establish STDP paradigms that successfully induce t-LTP in adult WT mice. To this aim we investigated in a first series of experiments a low repeat (6× at 0.5 Hz) STDP paradigm that we previously established in juvenile WT mice [34], to induce t-LTP in 6-month-old (adult) C57Bl6/J mice (Figures 1 and 2A).   (1) and after t-LTP induction (2). Scale bars are shown in the respective insets.

High Repeat STDP Stimulation Paradigm Induced t-LTP in Adult APP/PS1 Mice
To further explore STDP in adult 6-month-old WT and APP/PS1 mice, we tested the 1:1 and the 1:4 protocols with higher repeat numbers that were shown previously to elicit t-LTP in juvenile animals (compare [25]). Here, we argued that fully matured SC-CA1 synapses in adult animals might need stronger STDP paradigms for robust t-LTP, and that this t-LTP might be more sensitive to amyloidosis pathology in APP/PS1 than the low repeat protocols. First, we asked whether t-LTP induced with high repeat paradigms (i.e., 35× 1:4 and 100× 1:1) is different in magnitude or time-course compared to t-LTP induced with low repeat paradigms in WT mice (see Figure 3A).

T-LTP Induced with 6× 1:4 Stimulation at SC-CA1 Synapses in Adult APP/PS1 Mice Is Impaired Selectively in CA1 Neurons Located Near to Aβ Plaques
Although high frequency stimulation induced conventional LTP at SC-CA1 synapses was reported previously to be impaired in 5-6-month-old animals of the same APP/PS1 mouse strain used in the present study [10,35], our above-described results showed that t-LTP tested was not significantly altered at this age. Since we argued that the proximity of Aβ plaques to the recorded CA1 neuron might be decisive to observe significant t-LTP deficits in APP/PS1 animals, we next focused on t-LTP in CA1 neurons in the vicinity of Aβ plaques that we visualized by staining with the dye methoxy-X04 (for staining procedure, compare Methods). Stained Aβ plaques in APP/PS1 mice were visible in stratum radiatum, stratum oriens, and in the pyramidal cell layer of the CA1 area, but at this age (6-month), the density of Aβ plaques is relatively sparse (compare [10,36]). Thus, we decided to record t-LTP selectively in CA1 pyramidal neurons in the vicinity of Aβ plaques. To this aim we determined the distance between the recorded CA1 neuron soma and the nearest plaque. We observed that in CA1 neurons with their somata <200 µm away from the border of the nearest Aβ plaque (AD near), t-LTP induced by 6× 1:4 stimulation was significantly impaired (p = 0.0093) in comparison to t-LTP in methoxy-X04 treated WT littermate mice. Moreover, we observed comparable t-LTP magnitude in CA1 neurons >200 µm away from the border of the nearest Aβ plaque (AD distant) relative to methoxy-X04 treated WT littermates ( Figure 4A; WT: 135.8 ± 10.6% (n = 16/N = 9); AD near: 92.4 ± 9.1% (n = 12/N = 9); AD distant: 129.6 ± 9.6% (n = 8/N = 4)).
ANOVA analyses showed a significant main effect (F (2,33) = 5.338; p = 0.0098), and post hoc Tukey's test revealed p = 0.0093 for WT vs. AD near, p = 0.9153 for WT vs. AD distant, and p = 0.0762 for AD near vs. AD distant. Interestingly, when plotting t-LTP magnitude of recorded CA1 cells vs. distance to nearest plaque, we observed a moderate positive correlation between both parameters ( Figure 4B; Pearson correlation coefficient r(18) = 0.5105, p = 0.0215).

Chronic Fingolimod Treatment of APP/PS1 Mice Does Not Rescue Impaired t-LTP in CA1 Neurons Located in the Vicinity of Aβ Plaques
The FDA approved anti-multiple sclerosis drug fingolimod (FTY720, sphingosine-1 phosphate receptor modulator) has been shown previously to ameliorate Aβ pathology, as well as associated deficits in synaptic plasticity and memory formation in AD mice [10,37,38]. Based on these evidences, we tested the same regime of 1-month chronic fingolimod treatment (compare with Methods) that we reported previously to rescue AD deficits in APP/PS1 mice (compare [10]) for its potential to rescue 6× 1:4 t-LTP deficits in CA1 neurons near to Aβ plaques. However, this treatment did not rescue 6× 1:4 induced t-LTP at SC-synapses of CA1 neurons with their somata near to plaques (AD near) ( Figure 4C; WT + fingolimod: 169.9 ± 20.1% (n = 12/N = 7); AD near + fingolimod: 111.1 ± 10.2% (n = 16/N = 7), Mann-Whitney U-test: U = 42.0, p = 0.012). This suggests that chronic fingolimod treatment is not sufficient to rescue the 6× 1:4 induced t-LTP deficits in CA1 pyramidal cells of APP/PS1 mice.

CA1 Pyramidal Neurons in APP/PS1 Animals Showed Comparable Basal Electrical Properties as WT Littermates, Irrespective of Aβ Plaques Location
In the aforementioned results, we observed clear deficits in t-LTP at single neuron level in APP/PS1 mice only in CA1 neurons near to Aβ plaques (AD near). Therefore, we investigated whether the AD pathology also affected the basal electrical properties of hippocampal CA1 neurons. First, we focused on general differences between the genotypes irrespective of Aβ plaques location (Supplementary Figure S1). Since these results did not reveal any general differences between hippocampal CA1 neurons from APP/PS1 and WT littermates (Supplementary Figure S1), we investigated whether any such differences were evident when comparing CA1 neurons located near vs. distant to Aβ plaques of APP/PS1 animals. When testing basal electrical and basal synaptic properties of SC inputs to CA1 neurons, we did not observe any significant differences between properties of neurons near plaques (AD near) compared to CA1 cells with their somata distant from plaques (AD distant). Likewise, both groups showed no significant differences in all these parameters compared to WT littermate controls. Thus, AP frequency ( Figure 5A: AP frequency in response to a 180 pA depolarizing current step: WT: 21.6 ± 0.9 Hz (n = 16/N = 9); AD near: 20.3 ± 1.4 Hz (n = 12/N = 9); AD distant: 19.9 ± 1.0 Hz (n = 8/N = 4), ANOVA ( These results revealed that the observed impairment in t-LTP in APP/PS1 mice cannot be accounted for by differences of neuronal excitability or basic synaptic properties in postsynaptic CA1 pyramidal cells of this AD mouse model.

Discussion
High frequency stimulation (HFS) induced long-term potentiation (LTP) represents a traditional correlation-based type of synaptic plasticity. In contrast, spike timing-dependent plasticity (STDP) captures the importance of causality in determining the direction of synaptic modification [39]. Moreover, STDP has rapidly gained interest, because of its combination of simplicity, physiological plausibility, and computational power. Nonetheless, the current knowledge of molecular and cellular mechanisms of STDP and its role in pathophysiology is scarce. While numerous previous studies reported deficits in HFS induced LTP in the hippocampus as an early event in Alzheimer's disease (AD; [40][41][42][43][44]), the impact of AD on STDP is unknown. In the current study, we therefore assessed timing-dependent (t-) LTP at hippocampal Schaffer collateral (SC)-cornu am- However, APP/PS1 and WT mice expressed significantly different distribution of sEPSC inter-event intervals (IEIs; p = 0.005). (E) CA1 neurons from both WT littermate and APP/PS1 mice showed similar mean mEPSC amplitudes (p = 0.28), but significantly different cumulative mEPSC amplitude distributions (p < 0.0001). (F) CA1 neurons from APP/PS1 animals displayed similar mean mEPSC frequencies (p = 0.25) and mEPSC IEI distribution as WT littermates (p = 0.52). Digits in the bars show the number of recorded neurons per condition, at least from three different animals per group. Data shown as mean ± SEM. *: p < 0.05, statistical analyses were performed with two-tailed Student's t-test (parametric data) and Mann-Whitney U-test (non-parametric data). Cumulative frequency distributions were analyzed with Kolmogorov-Smirnov test.
Taken together, the spontaneous and miniature EPSC analyses in CA1 neurons revealed that in APP/PS1 mice, sEPSC amplitudes are increased whereas sEPSC frequencies are decreased, although this effect did not reach statistical significance, compared to WT littermates. On the other hand, APP/PS1 and WT littermate mice express comparable mEPSC amplitudes and frequencies. Furthermore, sIPSCs (inhibitory postsynaptic currents) are slightly decreased in amplitude and frequency in APP/PS1 mice compared to WT littermates, but also this effect did not reach statistical significance. Moreover, mIPSC properties are indistinguishable between genotypes (see Supplementary Figure S2 for IPSCs).

Discussion
High frequency stimulation (HFS) induced long-term potentiation (LTP) represents a traditional correlation-based type of synaptic plasticity. In contrast, spike timing-dependent plasticity (STDP) captures the importance of causality in determining the direction of synaptic modification [39]. Moreover, STDP has rapidly gained interest, because of its combination of simplicity, physiological plausibility, and computational power. Nonetheless, the current knowledge of molecular and cellular mechanisms of STDP and its role in pathophysiology is scarce. While numerous previous studies reported deficits in HFS induced LTP in the hippocampus as an early event in Alzheimer's disease (AD; [40][41][42][43][44]), the impact of AD on STDP is unknown. In the current study, we therefore assessed timing-dependent (t-) LTP at hippocampal Schaffer collateral (SC)-cornu ammonis (CA) 1 synapses in slices of 6-month-old APP/PS1 mice, which serve as a β-amyloidosis mouse model. Using patch clamp recordings, we were able to identify STDP paradigms that successfully induced t-LTP at SC-CA1 synapses in fully matured 6-month-old adult WT mice. The 6× 1:4 stimulation induced t-LTP was not significantly impaired in 6-month-old APP/PS1 mice, under conditions when proximity of amyloid beta (Aβ) plaque to the recorded neurons was not resolved. However, this 6× 1:4 t-LTP was strongly impaired if the nearest Aβ plaque was <200 µm away from the recorded neuron, and was not restored by chronic treatment of APP/PS1 mice with the anti-inflammatory drug Fingolimod (FTY720).

Timing-Dependent LTP in 6-Month-Old Adult WT Mice
While t-LTP is typically studied in juvenile animals, it was not previously reported to exist in 6-month-old WT mice (e.g., [25,32,33,45,46]). The results of the present study reveal that t-LTP can be induced at SC-CA1 synapses in acute hippocampal slices of 6-month-old mice with a 1:1 (canonical) and a 1:4 (burst) protocol that mimics postsynaptic theta bursts of action potentials (APs)-a firing pattern that is observed during learning processes in vivo [47]. Importantly, 6 pairings of pre-and postsynaptic APs at low frequency (0.5 Hz) were sufficient to induce t-LTP (Figure 2A). Interestingly, while also high repeat numbers of STDP stimulation (i.e., 100× 1:1 and 35× 1:4 protocols at 2 Hz) successfully induced t-LTP; these stronger protocols were neither more successful nor induced a higher magnitude of t-LTP (compare Figures 2A and 3A). Since this mild physiologically relevant stimulation with 6× 1:1 or 6× 1:4 paradigms induced t-LTP in adult mice (this study), but are equally effective also in 1 month old juvenile mice [34], our results indicate that physiological maturation does not affect STDP in the hippocampus of normal control mice.

Intact Timing-Dependent LTP in Adult APP/PS1 Mice When Aβ Plaque Location Was Not Resolved
In 6-month-old APP/PS1 mice t-LTP elicited by either 6× 1:4 or 100× 1:1 stimulation was intact (Figures 2B and 3B). However, as an early event in AD mouse models, synaptic dysfunctions such as alterations in conventional HFS induced LTP of hippocampal synaptic transmission have been observed by several earlier studies [40][41][42][43][44]. Importantly, these previous studies used extracellular field potential recordings that measure the summed synaptic responses of a larger population of neurons. In contrast, STDP is measured at the single neuron level. Furthermore, all studies investigating HFS induced LTP (field potential measurements) typically employ high frequency/theta burst stimulations for LTP induction under conditions of intact GABAergic inhibition, while pairing of presynaptic and postsynaptic APs used for t-LTP induction in the current study is typically recorded in the presence of GABA A antagonists (see Methods; [25,48,49]). These differences in LTP induction can have great influence on the recruitment of signaling cascades for expression of LTP compared to t-LTP, and this could account for intact t-LTP but impaired HFS induced LTP in 6-month-old APP/PS1 mice.

Impaired t-LTP in CA1 Neurons Located in the Vicinity of Aβ Plaques in Adult APP/PS1 Mice
Another explanation for the intact 6× 1:4 t-LTP in APP/PS1 mice could be that recorded single CA1 neurons might not be located near enough to Aβ plaques. Importantly, given the relatively low density of Aβ plaques in 6-month-old APP/PS1 hippocampus (see Figure 4B; compare [10]), it is likely that we recorded from CA1 pyramidal neurons farther away from a plaque if the location of the nearest plaque was not determined. Since we argued that the density or location of Aβ plaques might be decisive to observe clear t-LTP deficits in APP/PS1 mice, we next investigated t-LTP in CA1 pyramidal neurons in the proximity of Aβ plaques. APP/PS1 mice showed clearly impaired t-LTP in CA1 cells near to Aβ plaques (<200 µm, Figure 4A). This indicates that the proximity of Aβ plaques is indeed decisive to observe t-LTP deficits under our recording conditions. Although the Aβ plaques were identified as a principle component in AD histopathology a century ago, the pathogenic mechanisms of Aβ plaques are still unclear (see reviews [30,[50][51][52][53]. However, our results suggest that Aβ plaques and most likely associated soluble Aβ species, which are present at higher concentration around plaques, interfere with t-LTP, thereby contributing to cognitive decline. In this respect it is important to note that in CA1 pyramidal neurons farther away from Aβ plaques in APP/PS1 mice (>200 µm, Figure 4A), t-LTP was unaltered and comparable to t-LTP in WT littermates. Furthermore, we observed a dependence of t-LTP magnitude from plaque distance ( Figure 4B), which further supports an important role of plaque-associated toxicity for t-LTP. However, the detailed mechanism(s) by which Aβ plaques disrupt t-LTP at SC-CA1 synapses remains to be investigated.

Chronic Fingolimod Treatment Did Not Rescue Impaired t-LTP in APP/PS1 Mice
Treatment with fingolimod (for 1-month) alone was not sufficient to rescue t-LTP in adult APP/PS1 mice ( Figure 4C). It has been described that some neuroprotective effects of fingolimod are mediated through the elevation of brain-derived neurotrophic factor (BDNF) expression [38,54,55] and through reduced neuroinflammation (reviewed in [7,56]). Interestingly, microglia are involved in neuroinflammation and has been reported to affect synaptic remodeling during activity-dependent synaptic plasticity (LTP) in the healthy adult brain (see review [57,58]). Moreover, Nazari and colleagues described a fingolimod mediated rescue of impaired conventional LTP after stroke [59], while others reported fingolimod induced rescue of correlation based synaptic plasticity (LTP) in mouse models of Huntington's and Alzheimer's disease [10,60]. While the former study suggests an ameliorating effect of fingolimod at the neuronal network level, our current study reveals no rescue of 6× 1:4 induced impaired t-LTP in the proximity of Aβ plaques in adult APP/PS1 mice. These disparate findings might indicate differences in signaling cascades involved in conventional LTP, an associative activity among larger groups of cells, compared to t-LTP, which relies on plasticity at the single cell level.

Unaltered Basal Properties of CA1 Neurons in APP/PS1 Mice
Since we observed t-LTP deficits only in the proximity of Aβ plaques in APP/PS1 mice, we also studied basal electrical properties of CA1 pyramidal neurons in plaque distance-dependent manner. The studied basal electrical properties of the hippocampal CA1 neurons, e.g., intrinsic excitability, paired-pulse facilitation, resting membrane potential and action potential amplitude were intact in APP/PS1 mice (see Figure 5 and Supplementary Figure S1). These observations are consistent with findings from earlier studies [61,62]. Although APP/PS1 mice displayed significantly higher mean EPSC amplitudes and a different cumulative distribution of EPSC amplitudes and inter-event intervals (IEI) for spontaneous EPSCs, we observed largely unaltered properties of miniature EPSC, indicating no gross changes in quantal synaptic transmission at hippocampal SC-CA1 synapses in APP/PS1 mice (EPSCs, Figure 6). In addition, CA1 pyramidal neurons in APP/PS1 mice expressed comparable non-evoked inhibitory synaptic transmission as WT littermate mice (IPSCs, Supplementary Figure S2). Overall, these data indicate that overexpression of APP and PS1 mutations does not generally compromise CA1 pyramidal neurons' neuronal excitability and basal synaptic transmission in 6-month-old APP/PS1 mice.
While our study is the first to illustrate t-LTP deficits in APP/PS1 mice in an Aβ plaque distance-dependent manner, the molecular and cellular mechanisms underlying this deficit remain to be elucidated in future studies.

Conclusions
Our study reveals that in 6-month-old wild type mice t-LTP is elicited with similar efficiency by robust high repeat (i.e., 35× 1:4 and 100× 1:1) and more physiological low repeat (i.e., 6× 1:1 and 6× 1:4) STDP protocols. In 6-month-old APP/PS1 mice that show on average a low level of Aβ plaques in CA1 [10,31], these protocols show unaltered t-LTP, although high frequency stimulation (HFS)-induced SC-CA1 LTP measured with field potential recordings is already impaired at this age [10]. However, if patch clamp recorded CA1 neurons are specifically selected for their proximity to an Aβ plaque (Figure 4), they show impaired t-LTP, suggesting that plaque associated molecules block this potentiation. Together, these findings indicate that field potential HFS-LTP is more sensitive to Aβ pathology than t-LTP in single cell recordings, thereby suggesting that the underlying cellular signaling cascades of t-LTP and HFS-LTP differ. This might also explain why HFS-LTP in APP/PS1 mice can be rescued by Fingolimod [10], while the same treatment does not rescue t-LTP (current study). Future studies in aged APP/PS1 mice are required to resolve the full capacity of Fingolimod to rescue synaptic deficits in AD mice.

Animals
For spike timing-dependent plasticity (STDP) experiments, 6-month-old male C57Bl/6J mice (Charles River, Sulzfeld, Germany), amyloid precursor protein/presenilin 1 (APP/PS1) double transgenic mice [31] and their WT littermates, derived from our own breeding colony were used. APP/PS1 mice co-express KM670/671NL mutated amyloid precursor protein (APP) and L166P mutated presenilin (PS) 1 under the control of a neuron-specific Thy1 promoter element. These APP/PS1 animals are well suited for studying the pathomechanism of amyloidosis, a hallmark of Alzheimer's disease (AD) pathology. The animals were group housed in Makrolon cages at a temperature of 21 ± 2 • C and 12:12 h light/dark cycle, lights on at 07:00 am. They had free access to food and water. Genotypes of APP/PS1 and their WT littermates were determined by PCR from ear punches before the STDP experiments and verified again by tail biopsies after performing and analyzing the experiments. All experimental procedures were performed during the light period of the animals, were in accordance with the ethical guidelines for the use of animals in experiments of the European Committees Council Directive (2010/63/EU), and were approved by the local animal care committee (Landesverwaltungsamt Saxony-Anhalt, approval numbers: IPHY/G/01-1383/16 and IPHY/G/01-1492/18).

Hippocampal Slice Preparation and Maintenance
The animal was anesthetized using isoflurane (Isofluran CP, cp-pharma, Burgdorf, Germany). The loss of consciousness was confirmed by the absence of reflex activity following a toe pinch. The mouse was decapitated immediately, and the brain was separated from the skull. The brain was kept in an ice-cold artificial cerebrospinal fluid (aCSF) containing (in mM): 125 NaCl, 2.5 KCl, 26 NaHCO 3 , 0.8 NaH 2 PO 4 , 25 glucose, 6 MgCl 2 , 1 CaCl 2 , saturated with 95% O 2 and 5% CO 2 (pH 7.2-7.4; 303-306 mOsmol/L, Fiske Micro-osmometer Model 210, Fiske associates, Waterford, PA, USA). The cerebellum, brainstem and one 3rd of the frontal brain were removed before brain slicing. Moreover, the ventral part of the brain was cut transversely at an angle of 11 • in order to obtain transversal slices. The brain was cut with a vibratome (LEICA VT1200S VIBRATOME, Leica Biosystems, Nuβloch, Germany) and 350 µm thick acute hippocampal slices were collected starting from the first slice that allowed identifying clearly separated dentate gyrus and all cornu ammonis (CA) regions. These slices were used for STDP experiments and represent the intermediate and ventral region of the hippocampus. After slicing, the CA1 region was isolated from excessive CA3 input by a single cut between CA3 and CA2 to reduce spontaneous excitatory post-synaptic potentials (EPSPs). In all STDP recordings, synaptic inhibition was blocked with γ-aminobutyric acid (GABA) A receptor antagonist picrotoxin (100 µM). About 6 slices were transferred into the interface-style chamber and allowed for 25 min to incubate in continuously carboxygenated (5% CO 2 , 95% O 2 ) pre-warmed aCSF (200 mL, same composition as slice preparation medium mentioned above) at 34-35 • C to allow the slice surface to recover from blade trauma, followed by at least one hour of recovery at room temperature. All slices were maintained in this interface chamber at room temperature until being transferred to the recording chamber of an upright microscope for electrophysiological recording.

Aβ Plaque Staining with Methoxy-X04 in Acute Hippocampal Slices
The blue fluorescent dye 1,4-bis-(4 -hydroxystyryl)-2-methoxybenzene (methoxy-X04; Tocris, Wiesbaden, Germany) has been used previously as a live stain for Aβ plaques in AD mice [10,36,63,64]. To localize Aβ plaques in recorded hippocampal slices, APP/PS1 mice were injected intraperitoneally 24 h before slice preparation with methoxy-X04 (dissolved in dimethyl sulfoxide (DMSO; Sigma-Aldrich, Taufkirchen, Germany); stock solution: 10 mg/mL) at a concentration of 25 mg/kg body weight. After slice preparation on the day of t-LTP measurement and transfer of a recorded slice into the recording chamber of an upright fluorescence microscope (Zeiss Examiner A1, Carl Zeiss Microscopy GmbH, Jena, Germany), Aβ plaques were visualized using a 10× and 63× water immersion objectives. Fluorescence of methoxy-X04 was excited with a band pass filter (352-402 nm) and the emitted light was passed through a filter cube allowing to selectively detect blue fluorescence (dichroic mirror: 409 nm; bandpass filter: 417-477 nm). Live cell imaging of blue fluorescent Aβ plaques in CA1 area and associated stratum radiatum was performed with a digital CCD camera Photometrics CoolSNAP ES 2 (Visitron Systems GmbH, Puchheim, Germany). Using a digitized micromanipulator system (Luigs and Neumann SM-5), the distance (x,y,z coordinates) between the recorded CA1 neuron soma and the border of the nearest Aβ plaque was determined. The diagonal distance (compare Figure 4) of Aβ plaque from the CA1 neuron soma was calculated offline with the Pythagorean equation.

Chronic Fingolimod Treatment of APP/PS1 Mice
As a therapeutic strategy to rescue AD related deficits in t-LTP, we tested chronic fingolimod (Abcam, Cambridge, UK) treatment in adult APP/PS1 mice [38,54]. APP/PS1 animals and their WT littermates were injected intraperitoneally (i.p.) with a dose of 1 mg fingolimod/kg body weight [38], every second day for a month. Fingolimod stock solution (200 mM in DMSO) was diluted with 0.9% saline and was stored in aliquots at −20 • C until use. On the day of administration, fingolimod solution was thawed at room temperature, and warmed in 37 • C water bath for 3-4 min before intraperitoneal injection.

Electrophysiology in Acute Hippocampal Slices
For all experiments, 350 µm thick acute, transversal hippocampal slices were used. For whole-cell recordings, pyramidal neurons in CA1 region of hippocampus were visualized with differential interference contrast infrared video microscopy (VX45 Optronis camera, Optronis GmbH, Kehl, Germany; Zeiss Examiner A1 microscope, Carl Zeiss Microscopy, Jena, Germany). For all STDP recordings, 100 µM picrotoxin (Sigma Aldrich, Taufkirchen, Germany, dissolved in ethanol) was added to the extracellular aCSF solution to block GABA A receptors. ACSF was composed of (in mM): 125 NaCl, 2.5 KCl, 25 NaHCO 3 , 0.8 NaH 2 PO 4 , 25 glucose, 2 CaCl 2 , 1 MgCl 2 , saturated with 95% O 2 and 5% CO 2 (pH 7.2-7.4; 301-304 mOsmol/L). Slices were incubated for 5-10 min in the recording chamber before start of recording. Whole-cell recordings were performed at 28-31 • C (aCSF perfusion rate, 1.8 mL/min), with glass pipettes (pipette resistance, 4-6 MΩ) filled with intracellular solution containing (in mM): 140 potassium gluconate, 10 HEPES, 20 KCl, 4 Mg-ATP, 0.3 Na-GTP, 10 Na-phosphocreatine; pH was adjusted to 7.2-7.4 using 1 M KOH (280-290 mOsmol/L). The liquid junction potential of +10 mV that was observed before seal formation was corrected. Pipette capacitance, CA1 pyramidal cell capacitance and series resistance were compensated with EPC8 patch clamp amplifier in all voltage clamp and current clamp recordings. Cells were held in current or voltage clamp mode at -70 mV for different experimental approaches. For stimulation of presynaptic Schaffer collaterals (SC; stimulus duration 0.3 ms), a glass pipette (0.7-0.9 MΩ) filled with intracellular solution was placed in stratum radiatum where SC axon bundles are located in the CA1 region. During control and test periods, EPSPs were evoked at 0.05 Hz in current clamp mode. Presynaptic stimulation which caused postsynaptic firing of action potential (AP) or the maximum EPSP amplitude observed was regarded as maximum stimulus intensity. Stimulus strength (µA) was adjusted to evoke 30-50% of maximal stimulation as optimal stimulation intensity and kept constant during the whole experiment. Input resistance was determined by injecting a 20 pA hyperpolarizing current through the recording electrode for a duration of 250 ms; 200 ms prior to each SC stimulation. Cells were accepted for analysis only if the resting membrane potential was between −50 and −70 mV at the start of the recording. Data were discarded if input resistance changed ±30 % throughout the recording or in case of a clear run-up or run-down of EPSP slopes during the first 10 min of a recording. The paired-pulse facilitation (PPF) was tested at 50 ms inter-stimulus interval (ISI) repeated 3 times at 0.05 Hz in voltage clamp mode (holding potential: −70 mV).
Timing-dependent LTP was induced by repeated pairings of an EPSP induced by single presynaptic stimulation evoked by stimulation of SC input and one or four postsynaptic APs induced by one or four somatic current injections (2 ms, 1 nA) via the recording electrode. Different numbers of pairings were used depending on the stimulation protocol as mentioned in the results. T-LTP was induced by causal (pre-post) pairings at positive spike timings. Paradigms using either a 1 EPSP/1 AP (indicated as 1:1) or a 1 EPSP/4 AP (specified as 1:4) sequence were used. For each recorded cell, positive spike timing (i.e., ∆t in ms) was determined between the onset of the EPSP and the peak of the first AP. As a (negative) control, recordings with ongoing synaptic stimulation for 40 min at 0.05 Hz, but without pairing with postsynaptic APs, were performed (indicated as 0:0).

Data Acquisition and Data Analyses
Whole-cell recordings were performed using an EPC8 patch clamp amplifier connected to a LiH8+8 interface (HEKA Elektronik, Lambrecht, Germany) and acquired with Patchmaster software (HEKA, Germany). Data were filtered at 3 kHz and digitized at 10 kHz. Data analyses were performed using FitMaster (HEKA, Germany). Synaptic signals were recorded in current clamp mode as EPSP, except for PPF, which was recorded in voltage clamp as excitatory postsynaptic currents (EPSCs). With the help of FitMaster, initial EPSP slopes (i.e., first 2-3 ms after EPSP onset) were determined. In control and t-LTP experiments, EPSP slopes were normalized to the respective mean baseline recorded during the first 10 min prior to STDP stimulation, which was set to 100%. In all experiments, magnitude of synaptic changes (e.g., t-LTP) was determined as the normalized change in mean response size during the last 10 min of measurement (between 21-30 min after t-LTP induction). Further, the numbers of APs fired by a CA1 neuron, in response to different depolarizing somatic current injections (0-180 pA for 1000 ms, 20 pA increments) through the recording electrode, were determined to analyze intrinsic excitability. As another read-out of intrinsic excitability, the stimulation current (depolarization step (10 ms) of 0-400 pA, 40 pA increment) required to elicit one AP in the recorded CA1 cell [65,66] was determined as rheobase. The PPF was estimated by the paired pulse ratio (PPR), calculated as the normalized change in mean response size of the 2nd divided by the 1st EPSC. AP peak amplitude and after-depolarization were measured using FitMaster (demonstrated in supplementary Figure S1E). Similarly to the EPSP slopes, the input resistance was normalized to the respective mean baseline during the first 10 min of recording, which was then set to 100%. Data were discarded if input resistance changed ±30% throughout the STDP measurement. Further, EPSP rise time and decay time was calculated using the Trace Fit function (low level-high level: 10-90%) for the 10 min long baseline in FitMaster.

Whole Cell EPSC Recordings
Spontaneous and miniature EPSCs were studied in hippocampal CA1 pyramidal neurons in acute slices with no CA3-CA1 cut, prepared from APP/PS1 and WT littermate mice brains as described above. The EPSCs were recorded with the intracellular solution containing (in mM): 140 potassium gluconate, 10 HEPES, 20 KCl, 4 Mg-ATP, 0.3 Na-GTP, 10 Na-phosphocreatine; pH was adjusted to 7.2-7.4 using 1 M KOH (280-290 mOsmol/L; [25]). Spontaneous EPSCs were studied in the presence of bicuculine (10 µM, Sigma Aldrich) and miniature EPSCs were examined in the presence of bicuculine (10 µM) and tetrodotoxin (TTX, 1 µM, Tocris) in the extracellular aCSF [67]. EPSCs were recorded at −70 mV holding potential, for the period of 10 min; 250 events from each sweep were used for data analyses. The temperature of external solution in the recording chamber was maintained at 34-36 • C (aCSF perfusion rate-2.4 mL/min) for EPSC recordings. Data were analyzed manually using the 'Mini Analysis program' (version 6.0.7, Synaptosoft, Decatur, GA, USA).

Statistics
Data are specified as mean ± standard error of mean (SEM), and experiments were combined from at least three different animals per group. Furthermore, the number of experiments (n) and number of animals (N) are indicated in the respective parts of the Results. Group size of 10 was determined by power analysis (G*Power, Heinrich-Heine University Düsseldorf, Germany). The distribution of variables was determined with the Shapiro-Wilk test. Statistical analyses were performed with one-sample t-test, paired or unpaired two-tailed Student's t-test, as applicable. Non-parametric data were analyzed by Mann-Whitney U-test. Multiple comparisons were performed with ANOVA and post hoc Dunnett's or Tukey's test. Pearson correlation coefficient was used to determine correlation between distance of CA1 cell from Aβ plaques and t-LTP magnitude. The Kolmogorov-Smirnov test was used to determine significance for cumulative distributions data (E(I)PSCs). Significance levels are designated by *: p < 0.05. The statistical procedures used in each experiment are mentioned in the respective text passages. All data were analyzed using Origin 8.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy.