Skip to main content

The two faces of synaptic failure in AppNL-G-F knock-in mice



Intensive basic and preclinical research into Alzheimer’s disease (AD) has yielded important new findings, but they could not yet been translated into effective therapies. One of the reasons is the lack of animal models that sufficiently reproduce the complexity of human AD and the response of human brain circuits to novel treatment approaches. As a step in overcoming these limitations, new App knock-in models have been developed that avoid transgenic APP overexpression and its associated side effects. These mice are proposed to serve as valuable models to examine Aß-related pathology in “preclinical AD.”


Since AD as the most common form of dementia progresses into synaptic failure as a major cause of cognitive deficits, the detailed characterization of synaptic dysfunction in these new models is essential. Here, we addressed this by extracellular and whole-cell patch-clamp recordings in AppNL-G-F mice compared to AppNL animals which served as controls.


We found a beginning synaptic impairment (LTP deficit) at 3–4 months in the prefrontal cortex of AppNL-G-F mice that is further aggravated and extended to the hippocampus at 6–8 months. Measurements of miniature EPSCs and IPSCs point to a marked increase in excitatory and inhibitory presynaptic activity, the latter accompanied by a moderate increase in postsynaptic inhibitory function.


Our data reveal a marked impairment of primarily postsynaptic processes at the level of synaptic plasticity but the dominance of a presumably compensatory presynaptic upregulation at the level of elementary miniature synaptic function.


There is a general consensus on two major histopathological characteristics of AD, extracellular amyloid plaques, consisting of fibrils and non-fibrillar forms of the polypeptide amyloid ß (Aß) and neurofibrillary tangles (NFTs), and intracellular aggregates that are composed of hyperphosphorylated forms of the tau protein [1]. Aß originates from the sequential endoproteolytic cleavage of amyloid precursor protein (APP) resulting in several forms of Aß of which the 42-residue Aß42 has the strongest propensity to form aggregates and the highest cellular toxicity [2,3,4].

The discovery of genetically inherited, early-onset, familial forms of AD (FAD) in the 1990s of the last century made Aß the primary focus of AD research for more than two decades. FAD patients carry missense mutations in the genes encoding amyloid precursor protein (APP), presenilin 1 (PSEN1) and presenilin 2 (PSEN2), respectively (see [5] for references). These mutations result invariably in the generation of longer or more aggregating forms of Aβ. As this seems sufficient for the full development of similar clinical features as the much more common spontaneous, late-onset AD (LOAD) [6,7,8], it is important to investigate the effect of Aß on the cellular manifestation of the disease [9].

Despite the fact that FAD contributes to less than 0.1% of AD cases, its discovery has boosted the generation of a variety of transgenic mouse models that carry a combination of these human mutations and replicate key aspects of human AD, like amyloid deposition and progressing cognitive decline [10, 11]. These “first-generation” transgenic AD mouse models have been invaluable for delineating multifarious molecular mechanisms of disease onset and progression, but they share the limitation that the proteolytic processing of overexpressed APP results not only in the overproduction of Aß, but also of some other APP fragments. Thus, in these mice, pathological changes per se cannot be clearly attributed to an increased Aβ production since they could also be due to the (patho) physiological effects of one or several of the other APP fragments [12]. To overcome these drawbacks of APP overexpression, a new generation of mouse models of sporadic AD has been developed by the Saido laboratory [13] using a knock-in strategy to introduce the Swedish mutation, which increases all Aß species, into the APP gene, together with either the Beyreuther/Iberian mutation or the Beyreuther/Iberian plus the Arctic mutation [13, 14]. These new models, denominated as AppNL-F and AppNL-G-F, express normal APP levels but develop robust Aß pathology resulting in synaptic degeneration and memory impairments [14]. More specifically, AppNL-F mice develop high levels of Aß42 and a high Aß42/40 ratio without changes in the expression of APP and other fragments (except a shifted ratio of CTF-α/CTF-ß). Addition of the Arctic mutation to AppNL-F resulted in mice (AppNL-G-F) that progress threefold faster to a more severe AD pathology and cognitive deficits compared to AppNL-F mice [12, 13]. Comparative studies of these mice with first-generation transgenic APP models confirmed the hypothesis that some findings with the latter are likely to be due to “side effects” of overexpressing APP and non-Aß fragments rather than the increased levels of Aß or Aß42/40 [12, 13, 15]. With regard to the ongoing characterization of cognitive performance of AppNL-F and AppNL-G-F mice, memory deficits were reported at 6 months in AppNL-G-F and 18 months in AppNL-F mice [13, 16,17,18]. However, some of the memory deficits observed at 6 months of age in AppNL-G-F mice have been subtle [17] or could not be reproduced by others [19]. Together, these data led to the conclusion that “App knock-in mice should be considered models of preclinical AD” [12].

Research on preclinical models of AD and their characterization requires sensitive tools to detect subtle indications of incipient pathology. Given that AD as the most common form of dementia [1, 20] is the integrative result of a complex interplay of multiple multicellular pathophysiological processes [9] with synaptic failure as a major downstream pathological deterioration [21, 22], early signs of pathological changes are likely to be discernible at the level of synapses. Electrophysiological measures of synaptic transmission and plasticity are sensitive to even minor changes in pre- and postsynaptic functions [23,24,25,26,27] and therefore meet these requirements optimally.

However, an evaluation of AppNL-G-F mice at the synaptic level, i.e., the primary locus of the pathological deterioration of cognition, is still lacking. To address this, we used in the current study long-term extracellular recordings in acute slices of the prefrontal cortex (PFC) and the hippocampus (HC) to evaluate activity-dependent synaptic changes at two different stages of AD pathology, 3–4 and 6–8 months. Whole-cell patch-clamp recordings of mEPSCs and mIPSCs at the age of 6–8 months, i.e., when an almost saturated amyloidosis is present in these mice [13], complemented these experiments. Measurements of soluble and insoluble Aß40 and Aß42 tissue levels served as an indicator of the progression of Aß pathology.

We found first signs of synaptic impairment already at 3–4 months of age in AppNL-G-F mice, becoming overt as faster decay of LTP in PFC. With further progression of pathology at 6–8 months, PFC LTP was severely impaired, paralleled by a marked reduction in basal synaptic transmission. In contrast, in the hippocampal CA1 region, basal synaptic transmission, short-term plasticity, LTP, and LTD were inconspicuous at 3–4 months, but at 6–8 months, LTP was clearly impaired and short-term plasticity (paired-pulse ratio at 10 ms interpulse interval) reduced. No changes were found in basal synaptic transmission and LTD. Whole-cell patch-clamp recordings at 6–8 months, the age of pronounced synaptic pathology, revealed increased mEPSC and mIPSC frequency pointing to an enhanced presynaptic activity. The increase in mIPSC amplitude suggested that the increase in GABAergic transmission included also postsynaptic mechanisms.

To our knowledge, this is the first electrophysiological characterization of hippocampal and prefrontal synaptic functioning of these second-generation AD models, which are expected to become a standard for identifying mechanisms and pathways upstream and downstream of Aβ amyloidosis [13].

Materials and methods


The housing conditions and procedures to prepare acute brain slices were approved by the KU Leuven Ethical Committee and in accordance with European Directive 2010/63/EU. Homozygous female AppNL and AppNL-G-F mice were derived from the breeding colony established in the laboratory of Bart De Strooper. The AppNL-G-F mice co-express the Swedish (KM670/671NL), the Beyreuther/Iberian (I716F), and the Arctic (E693G) mutations. AppNL mice that only express the Swedish mutation and do not develop any significant pathology [13, 14] served as controls. Saito et al. and Mehla et al. reported an age-dependent Aβ amyloidosis in homozygous AppNL-G-F mice [13, 18]. Notably, in the study of Saito et al., the cortical deposition began by 2 months and was almost saturated by 7 months, while Mehla et al. detected significant deposition in the cortex and hippocampus at 6 months, which peaked at 9–12 months of age. Sacher et al. came to similar results as Mehla et al. using a longitudinal PET imaging of amyloid load with 18F-florbetaben as tracer [28].

Electrophysiological recordings

Extracellular long-term recordings in the hippocampal CA1 region

Electrophysiological recordings were performed in hippocampal slices as previously described [29]. Briefly, 3–4 and 6–8 months old female AppNL-G-F and AppNL were tested. Animals were killed by cervical dislocation, and the hippocampus was rapidly dissected out into ice-cold (4 °C) artificial cerebrospinal fluid (ACSF), saturated with carbogen (95% O2/5% CO2). ACSF consisted of (in mM) the following: 124 NaCl, 4.9 KCl, 24.6 NaHCO3, 1.20, KH2PO4, 2.0 CaCl2, 2.0 MgSO4, 10.0 glucose, pH 7.4. Transverse slices (400 μm thick) were prepared from the dorsal area of the right hippocampus with a tissue chopper and placed into a submerged-type chamber, where they were maintained at 32 °C and continuously perfused with carbogen-saturated ACSF at a flow rate of 2.5 ml/min. After 90 min of incubation, one slice was arbitrarily selected and a tungsten electrode was placed in CA1 stratum radiatum. For recording of field excitatory postsynaptic potentials (fEPSPs), a glass electrode (filled with ACSF, 3–7MΩ) was placed in the stratum radiatum, opposite the stimulating electrode. The time course of the field EPSP was measured as the descending slope function for all sets of experiments. After input/output (I/O) curves had been established, the stimulation strength was adjusted to elicit a fEPSP slope of 35% of the maximum and kept constant throughout the experiment. For paired-pulse ratios, responses to two impulses given at an interval of 10, 20, 50,100, 200, or 500 ms were recorded as described in [30, 31]. During baseline recording, 3 single stimuli (0.1 ms pulse width; 10 s interval) were measured every 5 min and averaged for the 60-min fEPSP values. To induce L-LTP, three theta burst stimuli (TBS, separated by 10 min, 0.2 ms pulse width) were applied. L-LTD was generated by three trains of low-frequency stimulation (LFS) at 2 Hz for 10 min (0.2 ms pulse width) [29, 30, 32].

Extracellular long-term recordings in medial PFC

Electrophysiological recordings were performed in coronal PFC slices, cut at 1.5–2.5 mm rostral from bregma as described in [33]. Animals were killed by cervical dislocation, and the whole brain was rapidly dissected into ice-cold preoxygenated artificial cerebrospinal fluid (ACSF) consisting of (in mM) 124 NaCl, 4.9 KCl, 2.5 CaCl2, 1.3 MgSO4, 1.2 NaH2PO4, 25.6 NaHCO3, and 16.6 glucose, gassed with 95% O2/5% CO2, at pH 7.4. Usually, two slices (400 μm thick) were prepared per mouse using a custom-made tissue chopper, and incubated for 1 h at room temperature before being placed in a submerged-type four-chamber recording system (Campden Instruments LTD, Loughborough, Leics., UK), and maintained there at 32 °C and a flow rate of 1.8 to 2 ml/min/chamber. In all experiments, custom-made monopolar tungsten electrodes were used for stimulation and ACSF-filled glass electrodes (5–7 MΩ resistance) for recording of field excitatory postsynaptic potentials (fEPSPs). The initial slope of the fEPSPs served as a measure of this potential. To assess basic properties of synaptic responses, I/O curves were established by stimulation with 30 to 90 μA constant currents. The stimulation strength was adjusted to evoke a fEPSP slope of 40% of the maximum and kept constant throughout the experiment. During baseline recording, three single stimuli (0.1 ms pulse width; 10 s interval) were measured every 5 min. Once a stable baseline was established, LTP was induced by 4 episodes of high-frequency stimulation (HFS) at 100 Hz for 1 s, with 5 min interval between consecutive episodes. The sample sizes mentioned for recordings always reflect the number of animals and not slices used.

Whole-cell patch-clamp recordings of CA1 pyramidal neurons

Postsynaptic currents from single CA1 pyramidal cells were recorded in transverse hippocampal slices (400 μm thick) as described elsewhere [34]. Slices were prepared using a vibratome (Microm HM 650 V, Thermo Scientific, Waltham, MA, USA) and were placed after the cutting for about 90 min in an incubation chamber containing ACSF (in mM: 124 NaCl, 4.9 KCl, 1.2 NaH2PO4, 25.6 NaHCO3, 2.0 CaCl2, 2.0 MgSO4, 10.0 glucose, pH 7.3–7.4) continuously perfused with 95%O2/5%CO2 at 32 °C.

Whole-cell voltage clamp recordings were performed at room temperature using a MultiClamp 700B patch-clamp amplifier, and data were collected using pClamp software (Axon Instruments, Union City, CA, USA). Recording electrodes pulled from borosilicate glass (World Precision Instruments) were filled with a solution containing the following (in mM): 135.0 CsMeSO4, 4.0 NaCl, 4.0 Mg-ATP, 0.5 EGTA-Na, 0.3 Na-GTP, 10.0 K-HEPES, 5.0 QX-314; pH 7.3 (pipette resistance 3–5 MΩ). Access resistance was 10–20 MΩ and was then compensated to 75%. Only neurons with the input resistance changing < 25% during the recordings were included in the study.

Based on reversal potential, miniature excitatory and inhibitory postsynaptic currents (mEPSCs and mIPSCs) were mostly measured consecutively from the same neurons (e.g., [35,36,37]). That is, mEPSCs were first recorded at the reversal potential for GABAA receptor-mediated events (− 60 mV), and mIPSCs were recorded at the reversal potential for glutamatergic currents (+ 10 mV) with tetrodotoxin (1 μM) present in the bath medium. To verify that mEPSCs were indeed glutamatergic, they were blocked by 20 μM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and 10 μM d-aminophosphonovalerate (D-APV) at the end of the experiments. Similarly, mIPSCs could be blocked by 100 μM picrotoxin, a GABAA receptor antagonist.

Data were low-pass filtered at 2 kHz and acquired at 10 kHz using Digidata 1440 and pClamp 10 software. Off-line analysis of mEPSCs and mIPSCs was performed using MiniAnalysis software (v.6.0.7, Synaptosoft, Decatur, GA, USA).

40 and Aβ42 level quantification

Different brain regions (hippocampus, neocortex, and cerebellum) of AppNL and AppNL-G-F mice were dissected after transcardial perfusion with ice-cold phosphate-buffered saline (PBS). Tissue was homogenized in tissue protein extraction reagent (Pierce) supplemented with complete protease inhibitors (Roche). The homogenates were centrifuged at 4 °C for 1 h at 100,000×g (Beckman TLA 100.4 rotor), and the supernatants were used for ELISA. To assess the GuHCl-soluble Aβ fraction of the tissue, we used a guanidine-HCl extraction protocol. Aβ40 and Aβ42 levels were quantified on Meso Scale Discovery (MSD) 96-well plates by ELISA using end-specific antibodies provided by Dr. Marc Mercken (Janssen Pharmaceutica, Belgium). Monoclonal antibodies JRFcAβ40/28 and JRFcAβ42/26, which recognize the C terminus of Aβ species terminating at amino acid 40 or 42, respectively, were used as capture antibodies. JRF/AβN/25 labeled with sulfo-TAG was used as the detection antibody, and the plate was read in MSD Sector Imager 6000.


All data are shown as mean ± SEM. Differences between mean values were determined using 1-way or 2-way analysis of variance (ANOVA), or unpaired t test with Welch correction (Welch test). Time series were compared with 2-way repeated measures ANOVA (RM-ANOVA) procedures with the Holm-Sidak test for post hoc comparison. For data obtained from whole-cell patch-clamp recordings, statistical significance was determined using the Welch test or Kolmogorov-Smirnov test.


Determination of Aβ40 and Aß42 in brains of AppNL-G-F and AppNL mice

To get an indication of the age-dependent increase in the levels of soluble and insoluble Aß1-40 and Aß1-42 and to confirm previous data on the same genotypes [13], we performed ELISA measurements in a small sample of mice (n = 2 per genotype and age). In AppNL mice, no age-related changes in soluble Aβ40 and Aß42 were found in the hippocampus and cortex between 1 and 6 months (Fig. 1a, b). The levels of Aβ40 and Aß42, respectively, in both brain regions were similar. In AppNL-G-F animals, Aß40 is generated at low levels (less than a tenth compared to AppNL mice) without any noticeable age-dependent changes. The levels of Aß42 are higher than in AppNL mice and show similar values between 1 and 3 months but a clear upregulation in the hippocampus of 6-month-old mice. The upregulation is absent in the cortex. Insoluble Aß40 and Aß42, respectively, remain at similar levels in the hippocampus and cortex of AppNL mice across the whole age range, but show peptide-specific changes in the same regions in AppNL-G-F mice (Fig. 1c, d). Here, Aß40 is at plateau levels between 1 and 3 months and increases at 6 months, but levels of Aß42 accumulate almost linearly during aging and are more than two hundred-fold higher at 6 months than at 1 month of age. Between the ages of 2 and 3 months, the rise of Aß42 levels seems to be faster in the cortex than in the hippocampus. These results are consistent with previous reports [13].

Fig. 1

Quantification of Aβ40 and Aβ42 levels by sandwich ELISAs in the hippocampus and prefrontal cortex of AppNL-G-F and AppNL mice. a, b AppNL mice do not display changes of soluble Aß40 (a) and Aß42 (b) between 1 and 6 months of age. AppNL-G-F mice present low levels of Aß40 with no age-dependent changes, both in the hippocampus and cortex (a). While the Aβ42 levels in these mice are similar between 1 and 3 months, they increase in the hippocampus but not in the cortex of 6-month-old mice (b). c, d Insoluble Aß40 (c) and Aß42 (d) remain at similar levels in the hippocampus and cortex of AppNL mice across the whole age range. Interestingly, in AppNL-G-F mice, Aß40 is at plateau levels between 1 and 3 months and increases at 6 months, but levels of Aß42 accumulate almost linearly during aging. Of note, the increase in insoluble Aβ42 appears to start earlier in the cortex than in the hippocampus (d). Data represent mean ± SEM (n = 2 mice per indicated time point)

Intact synaptic function and synaptic plasticity in the hippocampus of AppNL-G-F mice at 3–4 months of age

The hippocampus belongs to the regions that are early affected by Aß pathology in humans, subsequently to the initial infiltration of the neocortex [38]. In particular, the CA1 region of the hippocampus is very vulnerable to any synaptotoxic effects. In addition, this region is the best investigated brain area for the processes of synaptic plasticity [long-term potentiation (LTP), long-term depression (LTD)], which react very sensitively to alterations in synaptic function during AD progression [10, 11, 39] and are established models for memory formation and storage at the cellular level [40, 41]. Here, we used long-term recordings of field excitatory postsynaptic potentials (fEPSPs) in the CA1 stratum radiatum to examine whether basal synaptic transmission, short-term plasticity, LTP, and LTD, evoked in acute slices of 3–4-month-old AppNL-G-F mice, show any discernible differences to AppNL control mice. As depicted in Fig. 2a, we did neither detect any genotype differences in basal synaptic transmission (input/output curves) nor in paired-pulse responses, a measure of presynaptically mediated short-term plasticity [42] (Fig. 2b).

Fig. 2

Unchanged basal synaptic transmission, short-term plasticity, and long-term plasticity in the hippocampal CA1 region of AppNL and AppNL-G-F mice at 3–4 months of age. a I/O curves of AppNL and AppNL-G-F mice were almost identical. b Paired-pulse ratio was intact at every interval tested in both groups. c AppNL and AppNL-G-F animals developed a robust LTP for 4 h. d Triple repetition of LFS stimulation induced robust LTD that did not differ between AppNL and AppNL-G-F mice. Data are shown as mean ± SEM

Next, we examined activity-dependent long-term synaptic changes. The stimulation protocol we employed to induce LTP [theta burst stimulation (TBS)] is based on the “hippocampal theta rhythm,” a large-amplitude oscillation seen in electroencephalographic recordings in the range of 4–8 Hz [30, 43, 44], and as such may be considered as better approximating the physiological conditions in vivo than high-frequency stimulation (HFS) consisting of long bursts at 100 Hz. Since we found in the past that weak tetanization protocols (single TBS) are more sensitive to detect synaptic deficits than strong protocols (e.g., 3× TBS [45];), we used a single TBS stimulation to induce LTP and obtained similar potentiation in both groups (Fig. 2c; AppNL—10 min after TBS application 137 ± 15%, 240 min 117 ± 12%, n = 6; App NL-G-F—10 min 150 ± 14, 240 min 127 ± 15, n = 6). Furthermore, triple LFS stimulation resulted in similar LTD in both groups that lasted 4 h (Fig. 2d; AppNL—10 min post-LFS application 67 ± 6%, 240 min 49 ± 8%, n = 6; App NL-G-F—10 min 78 ± 7%, 240 min 50 ± 8%, n = 5). These data clearly indicated that Aß-mediated synaptotoxic effects failed to cause any significant functional deficits in the hippocampal CA1 region of AppNL-G-F mice at 3–4 months of age.

AppNL-G-F mice show deficits in hippocampal long-term potentiation at 6–8 months

The absence of any synaptic deficit in the CA1 region at 3–4 months of age raised the question as to whether synaptic functions are affected by increased Aβ levels at the age of 6–8 months, in particular by the marked increase in the aggregation-prone Aß42 detected by ELISA, which should promote the formation of synaptotoxic oligomers and aggregation. We performed, therefore, the same electrophysiological tests at 6–8 months. Although we did not find any genotype differences in basal synaptic function (Fig. 3a), there were clear deficits in AppNL-G-F compared to AppNL in other measures. First, the paired-pulse ratio was significantly reduced at 10 ms (t = 2.102, p = 0.0484 Welch test; AppNL-G-F n = 12, AppNL n = 18; Fig. 3b), suggesting short-term plasticity has been affected. More obviously, when long-term potentiation was examined, we found it significantly impaired in slices from AppNL-G-F (n = 7) animals compared to AppNL (n = 7) across 4 h of recording (Fig. 3c; main effect of genotype for 4 h recording post-induction: F1, 12 = 12.17, p = 0.0045, RM-ANOVA). The difference was already present immediately after the LTP induction during the first 10 min (F1, 12 = 6.362, p = 0.0268, RM-ANOVA). Thereafter, LTP of AppNL mice returned to baseline, but the potentiation of AppNL-G-F decayed even to values below.

Fig. 3

Deficits in hippocampal synaptic plasticity in AppNL-G-F mice at 6–8 months of age. a Basal synaptic transmission as measured by input/output curves did not differ significantly between genotypes. b Paired-pulse ratio was decreased at the 10-ms interpulse interval in AppNL-G-F mice compared to AppNL controls (t = 2.102, p = 0.0484 Welch test). c Slices from AppNL-G-F developed a smaller initial amplitude than APP NL controls and displayed impaired maintenance of potentiation resulting in a significant difference to AppNL littermates (F1, 12 = 12.17, p = 0.0045, RM-ANOVA). d LFS-induced LTD was not affected by Aß-mediated pathology in AppNL-G-F mice. Data are shown as mean ± SEM. Asterisks indicate difference in fEPSPs between AppNL and AppNL-G-F: *p < 0.05, **p < 0.01. Insets depict representative analogue traces, taken during baseline recording (solid line), after TBS (broken line), and at the end of recording (240 min, dotted line)

When the LTP curves of 3–4 and 6–8 months old animals are compared by eye inspection, then it is already apparent that the magnitude of potentiation is lower in the older mice. Statistical analysis with RM-ANOVA confirmed a significant main effect of age in AppNL (F1, 11 = 4.990, p = 0.0472) and AppNL-G-F (F1, 11 = 10.82, p = 0.0072) animals.

In contrast to the deficits found in LTP at 6–8 months, LTD experiments did not reveal any significant difference between the two genotypes (AppNL—51 min (= 1 min after third LFS train) 70.11 ± 3.97%, 240 min 78.22 ± 7.48%, n = 6; AppNL-G-F—51 min 72.32 ± 11.30%, 240 min 67.31 ± 5.75%, n = 6; Fig. 3d).

Impairment of synaptic plasticity starts at 3–4 months and of basal synaptic transmission at 6–8 months in PFC

In the second set of experiments, medial PFC field potentials were evoked in cortical brain slices from 3–4- (Fig. 4a, b) and 6–8-month-old (Fig. 4c, d) AppNL and AppNL-G-F mice. The prefrontal cortex (PFC) is one of the areas that are very susceptible to amyloid pathology and first affected by it [38, 46]. Reduced synaptic density in PFC is considered as one of the strongest pathological correlates of cognitive decline and the severity of dementia [46,47,48]. A recent study reported significant microstructural changes in the PFC of 9–10-month-old AppNL-G-F mice, which were examined by diffusion tensor magnetic resonance imaging [49].

Fig. 4

Extracellular recordings in the medial PFC of cortical slices from AppNL and AppNL-G-F mice at 3–4 (a, b) and 6–8 months of age (c, d). a Basal synaptic transmission is unaffected at 3–4 months as shown by the input-output curves. b Long-term potentiation induced by 4× HFS in the prelimbic and infralimbic region of the PFC from AppNL-G-F mice is of similar magnitude as from the AppNL littermates, but the potentiation of AppNL-G-F decays faster resulting in a temporary difference to AppNL from 30 min post-induction until 120 min (F(1, 7) = 9.052, p = 0.0197, RM-ANOVA). c Input-output curves indicate a severe reduction in basal synaptic transmission in AppNL-G-F mice at 6–8 months of age (two-way RM-ANOVA F(1, 15) = 16.82, p = 0.0009). d Although LTP of a similar initial magnitude could be induced in AppNL-G-F mice (n = 8), the potentiation was not maintained and declined to baseline. In contrast, AppNL mice (n = 8) expressed a robust LTP that remained at about the same level until the end of recordings resulting in a highly significant genotype difference (RM-ANOVA F(1, 16) = 12.04, p = 0.0032). Data are shown as mean ± SEM. Asterisks indicate difference in fEPSPs between AppNL and AppNL-G-F: *p < 0.05, **p < 0.01, ***p < 0.001 (RM-ANOVA). Insets depict representative analogue traces, taken during baseline recording (solid line), after HFS (long-broken line), 90 min after LTP induction (broken line), and at the end of recording (225 min, dotted line). Note that the analogue traces of AppNL-G-F mice at 90 min and 225 min in b are exactly on top of each other

At the age of 3–4 months, we did not find significant differences in the input-output curves (Fig. 4a). However, the LTP recordings of AppNL-G-F mice (n = 4) decayed faster after induction as compared with AppNL (n = 5) animals resulting in a temporary significant difference from 30 min post-induction until 120 min (F (1, 7) = 9.052, p = 0.0197, RM-ANOVA; Fig. 4b). Thereafter, the LTP values of both groups converged to similar values.

At 6–8 months, AppNL-G-F (n = 8) mice were even more severely impaired, because they showed a pronounced reduction in basic synaptic transmission compared to AppNL (RM-ANOVA F(1, 15) = 16.82, p = 0.0009; Fig. 4c) and LTP decayed to baseline (Fig. 4d). Since LTP of AppNL (n = 10) was robustly maintained, this led to a significant genotype difference from 120 min after LTP induction until the end of recording (RM-ANOVA F (1, 16) = 12.16, p = 0.0041).

In addition to the “in-between” genotype differences in basal synaptic transmission and LTP at the two ages investigated, we tested also whether there was also an age-dependent “within” genotype change in basal synaptic transmission. Statistical comparison with RM-ANOVA confirmed a significant age difference for AppNL-G-F mice (F1, 12 = 14.26, p = 0.0026) while not reaching statistical significance in AppNL littermates (F1, 13 = 4.196, p = 0.0613).

Increased frequency of excitatory and inhibitory miniature postsynaptic currents in AppNL-G-F mice

The deficits in synaptic plasticity in AppNL-G-F mice indicated alterations in the complex machinery of the trans-synaptic signaling regulating crucial mechanisms like neurotransmitter release and recycling, postsynaptic receptor binding, and the balance between glutamatergic and GABAergic activity. In order to investigate such changes in more detail, i.e., at the level of elementary synaptic function, we performed patch-clamp whole-cell recordings of miniature excitatory postsynaptic currents (mEPSCs) and miniature inhibitory postsynaptic currents (mIPSCs), consecutively from the same CA1 pyramidal neurons [35,36,37] of 6–8-month-old mice.

Our measurements revealed a marked increase in the mean frequency of mEPSCs in AppNL-G-F mice as compared to AppNL control animals (inset in Fig. 5b, AppNL-G-F—0.519 ± 0.044 Hz, n = 8; AppNL—0.344 ± 0.025 Hz, n = 6, p = 0.0012 Welch test). In contrast, we did not find any significant change in the mean amplitude or half-width of mEPSCs (insets in Fig. 5a, c).

Fig. 5

Analysis of action potential-independent mEPSCs (ac) and mIPSCs (df) in 6–8 months old AppNL and AppNL-G-F mice. a, b The mean frequency but not the amplitude of mEPSCs is markedly increased in AppNL-G-F mice as compared to AppNL control animals (see insets in a and b, p = 0.0012 Welch test). c Representative traces from mEPSC recordings illustrate the higher mEPSC frequency of CA1 pyramidal neurons from AppNL-G-F mice compared to AppNL controls. d mEPSC half-width, a measure that characterizes the kinetics of inactivation, is unchanged in AppNL-G-F mice. Analyses of the probability distributions of the data (a, b, d) revealed significant genotype differences for all three parameters. e, f Both mean amplitude and mean frequency of mIPSCs show a pronounced increase in AppNL-G-F mice [amplitude AppNL-G-F (inset in e, p = 0.0003 Welch test), frequency (inset in f, p = 0.0013 Welch test)]. g Representative traces from mIPSC recordings illustrate the marked increase in the mIPSC frequency of CA1 pyramidal neurons from AppNL-G-F mice compared to AppNL controls. h The kinetics of inactivation as measured by the half-width of mIPSCs was the same in both genotypes. Kolmogorov-Smirnov tests yielded significant genotype differences for the amplitude, the frequency, and the half-width of mIPSCs. Asterisks indicate difference between AppNL and AppNL-G-F: **p < 0.01, ***p < 0.001

Thereafter, we examined the same parameters for mIPSCs and found here not only a pronounced increase in the mean frequency of mIPSCs in AppNL-G-F (inset in Fig. 5e, AppNL-G-F—3.321 ± 0.371 Hz, n = 8; AppNL—2.225 ± 0.311 Hz, n = 6, p = 0.0430 Welch test) but also significantly higher values of the mean IPSC amplitude in AppNL-G-F (inset in Fig. 5d, AppNL-G-F—17.65 ± 0.9069 pA, n = 8; AppNL—15.31 ± 0.3647 pA, n = 6, p = 0.0403 Welch test). As with the mEPSCs, there was no difference between genotypes in mIPSC half-width.

In-depth analyses of the probability distributions of the data (Fig. 5a–f) with the Kolmogorov-Smirnov test confirmed genotype differences for those parameters that had significantly different mean values as mentioned above. However, it detected also significant genotype differences in the cumulative probability distribution of several other parameters (mEPSC amplitude, mEPSC inter-event interval, mIPSC amplitude, mIPSC inter-event interval, mIPSC half-width: all p < 0.0001; mEPSC half-width p = 0.0033.


The study of genetically modified animals, mainly transgenic and knock-out mouse models, has been invaluable in the past to gain significant insights into the pathological mechanisms of AD. Although none of these models fully reproduces the complete spectrum of AD, they recapitulate major aspects of the disease [10] and their detailed characterization at the molecular, cellular, network, and behavioral levels is important to understand the utility and limitations of each model [50]. The AppNL-G-F mouse model, investigated in our study, belongs to the “second-generation” mouse models produced with the hope that they would display greater concordance between preclinical animal studies and human clinical trials. Aß-mediated pathology (as becoming overt by plaque deposition) starts already at the age of 2 months, and the cortex is almost saturated at 7 months. Likewise, neuroinflammation and changes in synaptic proteins, which are two of the main AD hallmarks, are also observed in these mice at early stages [13]. AppNL mice were used as control in our experiments, because they are generated by the same knock-in strategy as AppNL-G-F mice and show the same increase in the level of the C-terminal fragment β (CTF-β) as the latter, due to the Swedish mutation.

Here, we used an electrophysiological approach to gain insights into putative functional deficits at different levels of synaptic function, and examined basal synaptic transmission, synaptic plasticity, and miniature synaptic currents. Although there is an increasing awareness of AD as a multidimensional and multicellular process [9, 12], many of the pathological processes involved in AD that operate in parallel and/or interactively in neurons and other cells in the brain inevitably converge on the cellular mechanism underlying memory and compromise them in a complex manner [1, 20]. Thus, synapses and synaptic plasticity are major downstream sites and mechanisms of pathophysiological convergence [21, 22]. We recorded from the hippocampus and prefrontal cortex since these two brain regions are closely linked anatomically and their bidirectional functional interaction is essential for higher-order cognitive functions, including the encoding and retrieval of working, emotional, and episodic types of memory [51,52,53]. The synchronization of hippocampal and prefrontal neural activity allows the coordinated hippocampal-prefrontal replay which is considered a key mechanism for memory consolidation [54]. Importantly, the hippocampus (in particular, its CA1 region) and the PFC are highly susceptible to amyloid pathology, in particular to the synaptotoxic effects of soluble low-n Aß oligomers ([55,56,57]; see [3] for critical discussion).

In the present study, synaptic functioning was inspected at 3–4 and 6–8 months of age. Evidence for a beginning impairment was detected for PFC LTP in 3–4 months old AppNL-G-F mice, which was further aggravated at 6–8 months of age. The only other study of PFC LTP in an AD mouse model by Battaglia et al. reported an LTP deficit in the PFC of transgenic APP/PS1 mice, but did not specify the age [58]. In the hippocampal CA1 region, a deficit in LTP was only recognizable in 6–8 months old AppNL-G-F animals. The deficits in HC and PFC LTP were most apparent in the maintenance of potentiation and less in the initial magnitude, because LTP in both regions decayed to baseline values or even below, as observed in HC. Given the length of our LTP recordings of more than 3 h, we found late-LTP being impaired (≥ 3 h after induction), whose mechanisms have been implicated in long-term memory [59, 60]. Late-LTP was demonstrated to be highly susceptible to disruption by AD pathology but was rarely recorded in AD animal models in vitro in the past [61,62,63,64]. The earlier impairment of LTP in PFC compared to HC in AppNL-G-F mice is likely caused by a higher Aß42 burden and/or its earlier onset in this region at an early age, as indicated by the results of the ELISA measurements and further supported by significantly higher immunoreactive plaque areas in the PFC vs. HC of AppNL-G-F mice at the two ages investigated here [13]. According to Benilova et al. and Yang et al., amyloid plaques likely sequester soluble Aß oligomers and release them under certain conditions [3, 57]. Thus, a dynamic equilibrium between toxic oligomers and inert fibrils might exist around the plaques, resulting in the local release of neurotoxic oligomers into the surrounding tissue. According to this scenario, the higher plaque load in the cortex compared to the hippocampus of AppNL-G-F mice is likely to result in a higher tissue concentration of toxic oligomers in the cortex which would explain the earlier onset of synaptic deficits, as detected by a faster decay of LTP at the age of 3–4 months. Interestingly, in HC but not PFC, we found a significant reduction in the magnitude of potentiation with age. This difference was bigger in AppNL-G-F mice suggesting a major contribution of Aß-driven pathology to the age-dependent decline in potentiation.

In contrast to the pronounced impairment of LTP, NMDAR-dependent LTD induced by LFS at 2 Hz [32, 65] was not affected in AppNL-G-F mice. While LTP has been extensively studied in transgenic and knock-out animal models of AD [10, 11, 27], LTD as the physiological counterpart of LTP was largely neglected. Most of these AD-related LTD studies focused on the effects of amyloid ß-protein and its fragments. While some studies reported the resistance of LTD to the application of Aß peptides or to the pathology-driven increase of their endogenous levels [66], most studies found an increase in LTD or a facilitation of its induction [26, 56, 67,68,69,70,71]. The reasons for the missing deficits in LTD are not clear. It is unlikely that 2 Hz LTD is completely insensitive to Aß-mediated pathology because we obtained a clear impairment in 12-month-old APP/PS1 mice (unpublished laboratory data). Impaired LTD was also described for the cerebellum of APP/PS1 mice [72]. Thus, LTD is sensitive to Aß-mediated pathology, but compared to LTP, LTD may require higher levels of soluble Aß oligomers to get impaired [66, 73].

It is well accepted that (several forms of) LTP and LTD require the activation of NMDA receptors and the increase of intracellular calcium, but LTP and LTD are predominantly expressed and sustained by different signaling pathways. The lack of effects of Aß-mediated pathology on LTD in AppNL-G-F mice therefore indicates that Aß oligomers exert differential effects on signaling cascades that sustain LTP and LTD, respectively. This supposition is supported by published evidence. For example, Aβ oligomers block the accumulation of CaMKIIα, a central kinase in bidirectional synaptic plasticity, at excitatory synapses during LTP induction but not at inhibitory synapses during LTD [74]. Furthermore, Kootar et al. revealed by ex vivo experiments in hippocampal slices that the LTP impairment by Aß oligomers requires the functional crosstalk with glucocorticoid receptors, but the Aß oligomer-mediated LTD induction does not [75].

In addition to the marked deficit in LTP in HC and PFC at 6–8 months of age, basal synaptic transmission in PFC was significantly impaired in AppNL-G-F mice, as indicated by the lower input/output curve. While Battaglia et al. did not find a difference in basal synaptic transmission in the PFC of the transgenic APP/PS1 model [58], Roder et al. reported reduced basal synaptic transmission in the PFC of 6-month-old APP23 mice [76]. Such a severe disturbance of synaptic functioning was reported also for some transgenic AD mouse models like APP23, APPLd2, APPind H6, hAPPJ9, hAPPJ20, PS1M146V KI, and 3xTg ([76], see [27] for detailed references) in the hippocampal CA1 region, but in AppNL-G-F mice, we did not detect any significant reduction of basal synaptic transmission in this region.

The absence of any overt deficits in basal synaptic transmission and synaptic plasticity in AppNL-G-F at 3–4 months might be due to a composition and/or levels of Aß42 and Aß40 oligomers at this age that were not yet sufficient to exert discernible synaptotoxic effects. One of the reasons could be that the plaques that already exist at this age [13] sequester soluble high-n oligomers, a putative source for toxic low-n oligomers [57], to a higher degree at early than later stages of pathology.

Further of note, when we checked short-term plasticity by paired-pulse stimulation [77], we found a significantly lower value for AppNL-G-F mice at an interpulse interval of 10 ms. In previous studies of transgenic AD mouse models, significant changes in the paired-pulse ratio (PPR) were very rarely detected [27]. Since the interval of 10 ms is an integral part of the 100-Hz bursts in the TBS protocol that was used in our study to induce LTP, the decay of responses to repeated stimuli resulted presumably in a lower efficacy of the TBS protocol, leading to the observed LTP deficit in AppNL-G-F animals. A similar relationship between decreased paired-pulse response at 10 ms and defective LTP was previously described in neurogranin knock-out mice, provided a weak stimulation protocol was employed [78, 79]. The found reduction of the paired-pulse ratio at 10 ms is most likely due to increased GABAAergic inhibition, because it could be rescued in previous experiments by application of GABAA receptor antagonists such as picrotoxin [33].

In order to better understand at an elementary level to which extent pre- vs. postsynaptic changes and a shift in the balance between glutamatergic excitatory and GABAergic inhibitory activity contribute to the disturbances in synaptic function, we measured action potential-independent miniature EPSCs (mEPSCs) and miniature IPSCs (mIPSCs). The obtained significant changes in the mean frequency of mEPSCs in AppNL-G-F mice, together with an unchanged mean amplitude and half-width, suggest an increased presynaptic activity of glutamatergic synapses in response to the progressing Aß-mediated pathology. In contrast to our finding, Chang et al. observed in 2xKI mice that carry the Swedish mutation and the P264L presenilin 1 mutation but do not overexpress APP an age-dependent reduction in mEPSC amplitude [80]. Likewise, D’Amelio et al. found in 3-month-old Tg2576 transgenic mice a downregulation of mEPSC frequency [26].

The overall effect of synaptic pathology on mIPSCs in AppNL-G-F mice seems to be even stronger because we did not only find an increased mean frequency of mIPSCs in AppNL-G-F but also a significantly higher mean IPSC amplitude, i.e., synergistically acting pre- and postsynaptic mechanisms. The apparent increase in elementary inhibitory activity could be involved in the deficits in LTP induction at this age and is reminiscent of reported compensatory inhibitory mechanisms that developed in 4–7 months old hAPP-J20 mice in response to neuronal overexcitation [25]. These mice, carrying the Swedish and Indiana FAD mutations, show an enhanced frequency of mIPSCs similar to our findings in AppNL-G-F mice. In addition, the amplitude of large-amplitude mIPSCs was found to be increased in this study.


Taken together, the characterization of elementary synaptic functions and long-term synaptic plasticity in AppNL-G-F mice seems to point to an apparent synaptic enigma in these mice. The marked impairment of primarily postsynaptic processes at the level of synaptic plasticity contrasts with the found upregulation of presynaptic processes in elementary (miniature) synaptic function. Major changes in presynaptic function and markers are not unique to APP-KI mice; they were also found in AD patients [81,82,83]. Thus, the found upregulation is most likely the result of a compensatory “homeostatic” response. Such a mechanism could serve to counteract impairments of synaptic plasticity caused by postsynaptic deficits in order to confine the overall magnitude of AD-driven imbalances in synaptic function. The described changes in miniature pre- and postsynaptic mechanisms in concert with defects in synaptic plasticity are expected to stimulate further research, paving the way for new therapeutic strategies that target the vulnerable synaptic machinery that is central to AD-mediated cognitive decline.


The current study was designed to perform a first characterization of synaptic plasticity and other synaptic functional readouts of two new second-generation AD mouse models (AppNL-G-F and AppNL mice), which were produced by a knock-in strategy with the hope that they would display greater concordance between preclinical animal studies and human clinical trials. Indeed, recent studies support the supposition that they can serve as valuable models to examine Aß-related pathology in “preclinical AD.” The AppNL-G-F mice harbor a humanized APP construct that contains three mutations associated with familial Alzheimer’s disease: the Swedish, the Beyreuther/Iberian, and the Arctic mutation in the APP gene. These three mutations increase total Aß production, enhance the Aβ42/Aβ40 ratio, and promote Aß aggregation, respectively. AppNL mice carrying only the Swedish mutation serve as control. Of note, in AppNL-G-F mice, Aβ deposition starts already at 2 months and is nearly saturated by 7 months.

Against this background, we decided to investigate the two genotypes at two different stages of pathology, at an age of 3–4 months with mild to moderate Aß expression and an expected absence of major functional phenotypes, and at an age of 6–8 months with aggravated pathology, for which we anticipated discernible functional deficits in synaptic plasticity and other functional synaptic readouts. At the latter age, AppNL-G-F are still devoid of marked behavioral phenotypes according to published studies. Because our synaptic measurements are done in vitro, we have to use different batches of mice for this work. Given the attractiveness of these new mouse AD models, there was a high demand by several groups at KU Leuven to use them for a variety of methodological studies. This sometimes led to a bottleneck in breeding in the animal facility, making it difficult to receive batches of mice with the same sample size for all experiments.

For the statistical design of the study, we focused on pathophysiologically meaningful deficits, i.e., a rather large effect size and used homozygous females. Because we are conducting electrophysiological recordings of long-term potentiation (LTP) and long-term depression (LTD) with a sampling interval of 5 min, we are dealing with time series and use, therefore, repeated measures ANOVA (RM-ANOVA) as the standard of our analysis, with group/treatment as between-subject factor and time as within-subject factor. Due to the sensitivity of synaptic plasticity measures (e.g., LTP) towards pathological deteriorations, we observe usually rather big effect sizes. Therefore, the study has been powered to detect between-group differences, assuming an effect size F = 0.95, using an alpha err probe = 0.05, power (1-ß err probe) = 0.80, and a correlation between repeated measures of 0.85. Our common recording time of LTP or LTD is 3 to 4 h with a sampling interval of 5 min, resulting in a high number of repetitions in RM-ANOVA. In the event that we look at particular phases (periods) of LTP or LTD (e.g., early LTP, late-LTP), we work at least with periods of 90 min, which is equal to 19 repetitions in RM-ANOVA. Using these values in G-Power, we came to a minimal required total sample size of the two groups of 10 (5 per group), which was fulfilled throughout our measurements, except the 3-month-old AppNL-G-F group in the PFC recordings, where we lost the data of one animal due to technical problems during recording.

In terms of the reproducibility of the data, repeating the experiments was not an option as the staff who performed most of the measurements (a doctoral student, a postdoc) had to leave the laboratory because their contracts were running out and could not be renewed and new staff could not be hired because of the difficult local grant situation. However, according to our experience, it seems to be more important that new findings are replicated independently by other laboratories under (usually) slightly different methodical conditions to get a more robust picture of the particular phenotype. Thus, we noticed often in the past that results could be precisely replicated under the same conditions in one laboratory while other laboratories failed consistently to confirm them. Prominent examples for the latter are the controversies about the “molecular switch” in the field of mGluR receptors, the role of NR2A and NR2B NMDA receptor subunits in synaptic plasticity and learning, and the function of PKMzeta within the same functional circuits. Here, the laboratories that first described the function could consistently reproduce the initial finding, while other laboratories continued to fail. In relation to the present study, it would therefore be best if the same or similar experiments were carried out in other laboratories in order to check/validate the reproducibility of the presented results under different laboratory conditions.

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.



Amyloid ß


Artificial cerebrospinal fluid


Alzheimer’s disease


Analysis of variance


Amyloid precursor protein

App NL-F :

Knock-in mouse model of sporadic AD co-expressing the Swedish (KM670/671NL) mutation and the Beyreuther/Iberian (I716F) mutation

App NL-G-F :

Knock-in mouse model of sporadic AD carrying the Swedish mutation, the Beyreuther/Iberian mutation, and the Arctic (E693G) mutation






C-terminal fragment β




Familial forms of AD


Field excitatory postsynaptic potentials


High-frequency stimulation


Low-frequency stimulation


Long-term depression


Long-term potentiation


Late-onset AD


Miniature excitatory postsynaptic currents


Miniature inhibitory postsynaptic currents


Meso Scale Discovery


Neurofibrillary tangles


Phosphate-buffered saline


Prefrontal cortex


Paired-pulse ratio


Presenilin 1


Presenilin 2


Two-way repeated measures ANOVA


Theta burst stimulation

Welch test:

Unpaired t test with Welch correction


  1. 1.

    Querfurth HW, LaFerla FM. Alzheimer’s disease. N Engl J Med. 2010;362:329–44.

    CAS  PubMed  Google Scholar 

  2. 2.

    De Strooper B, Annaert W. Proteolytic processing and cell biological functions of the amyloid precursor protein. J Cell Sci. 2000;113:1857–70.

    PubMed  Google Scholar 

  3. 3.

    Benilova I, Karran E, De Strooper B. The toxic Abeta oligomer and Alzheimer’s disease: an emperor in need of clothes. Nat Neurosci. 2012;15:349–57.

    CAS  PubMed  Google Scholar 

  4. 4.

    Sheng M, Sabatini BL, Sudhof TC. Synapses and Alzheimer’s disease. Cold Spring Harb Perspect Biol. 2012;4:a005777.

    PubMed  PubMed Central  Google Scholar 

  5. 5.

    Tanzi RE, Bertram L. New frontiers in Alzheimer’s disease genetics. Neuron. 2001;32:181–4.

    CAS  PubMed  Google Scholar 

  6. 6.

    Selkoe DJ. The cell biology of beta-amyloid precursor protein and presenilin in Alzheimer’s disease. Trends Cell Biol. 1998;8:447–53.

    CAS  PubMed  Google Scholar 

  7. 7.

    LaFerla FM, Oddo S. Alzheimer’s disease: Abeta, tau and synaptic dysfunction. Trends Mol Med. 2005;11:170–6.

    CAS  PubMed  Google Scholar 

  8. 8.

    Selkoe DJ, Hardy J. The amyloid hypothesis of Alzheimer’s disease at 25 years. EMBO Mol Med. 2016;8:595–608.

    CAS  PubMed  PubMed Central  Google Scholar 

  9. 9.

    De Strooper B, Karran E. The cellular phase of Alzheimer’s disease. Cell. 2016;164:603–15.

    PubMed  Google Scholar 

  10. 10.

    LaFerla FM, Green KN. Animal models of Alzheimer disease. Cold Spring Harb Perspect Med. 2012;2:a006320.

    PubMed  PubMed Central  Google Scholar 

  11. 11.

    Puzzo D, Gulisano W, Palmeri A, Arancio O. Rodent models for Alzheimer’s disease drug discovery. Expert Opin Drug Discov. 2015;10:703–11.

    CAS  PubMed  PubMed Central  Google Scholar 

  12. 12.

    Sasaguri H, Nilsson P, Hashimoto S, Nagata K, Saito T, De Strooper B, Hardy J, Vassar R, Winblad B, Saido TC. APP mouse models for Alzheimer’s disease preclinical studies. EMBO J. 2017;36:2473–87.

    CAS  PubMed  PubMed Central  Google Scholar 

  13. 13.

    Saito T, Matsuba Y, Mihira N, Takano J, Nilsson P, Itohara S, Iwata N, Saido TC. Single App knock-in mouse models of Alzheimer’s disease. Nat Neurosci. 2014;17:661–3.

    CAS  PubMed  Google Scholar 

  14. 14.

    Nilsson P, Saito T, Saido TC. New mouse model of Alzheimer’s. ACS Chem Neurosci. 2014;5:499–502.

    CAS  PubMed  PubMed Central  Google Scholar 

  15. 15.

    Saito T, Matsuba Y, Yamazaki N, Hashimoto S, Saido TC. Calpain activation in Alzheimer’s model mice is an artifact of APP and presenilin overexpression. J Neurosci. 2016;36:9933–6.

    CAS  PubMed  PubMed Central  Google Scholar 

  16. 16.

    Shah D, Latif-Hernandez A, De SB, Saito T, Saido T, Verhoye M, D'Hooge R, Van der Linden A. Spatial reversal learning defect coincides with hypersynchronous telencephalic BOLD functional connectivity in APP (NL-F/NL-F) knock-in mice. Sci Rep. 2018;8:6264.

    PubMed  PubMed Central  Google Scholar 

  17. 17.

    Latif-Hernandez A, Shah D, Craessaerts K, Saido T, Saito T, De SB, Van der Linden A, D'Hooge R. Subtle behavioral changes and increased prefrontal-hippocampal network synchronicity in APP (NL-G-F) mice before prominent plaque deposition. Behav Brain Res. 2019;364:431–41. Epub;2017 Nov;20.:431–41.

    CAS  Article  PubMed  Google Scholar 

  18. 18.

    Mehla J, Lacoursiere SG, Lapointe V, McNaughton BL, Sutherland RJ, McDonald RJ, Mohajerani MH. Age-dependent behavioral and biochemical characterization of single APP knock-in mouse (APP (NL-G-F/NL-G-F)) model of Alzheimer’s disease. Neurobiol Aging. 2019;75:25–37. Epub;2018 Nov 5.:25–37.

    CAS  Article  PubMed  Google Scholar 

  19. 19.

    Whyte LS, Hemsley KM, Lau AA, Hassiotis S, Saito T, Saido TC, Hopwood JJ, Sargeant TJ. Reduction in open field activity in the absence of memory deficits in the App (NL-G-F) knock-in mouse model of Alzheimer’s disease. Behav Brain Res. 2018;336:177–81. Epub;2017 Sep 5.:177–81.

    Article  PubMed  Google Scholar 

  20. 20.

    Scheltens P, Blennow K, Breteler MM, De SB, Frisoni GB, Salloway S, Van der Flier WM. Alzheimer's disease. Lancet. 2016:10–6736.

  21. 21.

    Selkoe DJ. Alzheimer’s disease is a synaptic failure. Science. 2002;298:789–91.

    CAS  PubMed  Google Scholar 

  22. 22.

    Forner S, Baglietto-Vargas D, Martini AC, Trujillo-Estrada L, LaFerla FM. Synaptic impairment in Alzheimer’s disease: a dysregulated symphony. Trends Neurosci. 2017;40:347–57.

    CAS  PubMed  Google Scholar 

  23. 23.

    Choi S, Lovinger DM. Decreased probability of neurotransmitter release underlies striatal long-term depression and postnatal development of corticostriatal synapses. Proc Natl Acad Sci U S A. 1997;94:2665–70.

    CAS  PubMed  PubMed Central  Google Scholar 

  24. 24.

    Hsia AY, Masliah E, McConlogue L, Yu GQ, Tatsuno G, Hu K, Kholodenko D, Malenka RC, Nicoll RA, Mucke L. Plaque-independent disruption of neural circuits in Alzheimer’s disease mouse models. Proc Natl Acad Sci U S A. 1999;96:3228–33.

    CAS  PubMed  PubMed Central  Google Scholar 

  25. 25.

    Palop JJ, Chin J, Roberson ED, Wang J, Thwin MT, Bien-Ly N, Yoo J, Ho KO, Yu GQ, Kreitzer A, et al. Aberrant excitatory neuronal activity and compensatory remodeling of inhibitory hippocampal circuits in mouse models of Alzheimer’s disease. Neuron. 2007;55:697–711.

    CAS  PubMed  Google Scholar 

  26. 26.

    D'Amelio M, Cavallucci V, Middei S, Marchetti C, Pacioni S, Ferri A, Diamantini A, De ZD, Carrara P, Battistini L, et al. Caspase-3 triggers early synaptic dysfunction in a mouse model of Alzheimer’s disease. Nat Neurosci. 2011;14:69–76.

    CAS  PubMed  Google Scholar 

  27. 27.

    Marchetti C, Marie H. Hippocampal synaptic plasticity in Alzheimer’s disease: what have we learned so far from transgenic models? Rev Neurosci. 2011;22:373–402.

    CAS  PubMed  Google Scholar 

  28. 28.

    Sacher C, Blume T, Beyer L, Peters F, Eckenweber F, Sgobio C, Deussing M, Albert NL, Unterrainer M, Lindner S, et al. Longitudinal PET monitoring of amyloidosis and microglial activation in a second-generation amyloid-beta mouse model. J Nucl Med. 2019;60:1787–93.

    CAS  PubMed  PubMed Central  Google Scholar 

  29. 29.

    Ahmed T, Blum D, Burnouf S, Demeyer D, Buee-Scherrer V, D'Hooge R, Buee L, Balschun D. Rescue of impaired late-phase long-term depression in a tau transgenic mouse model. Neurobiol Aging. 2015;36:730–9.

    CAS  PubMed  Google Scholar 

  30. 30.

    Denayer E, Ahmed T, Brems H, Van Woerden G, Borgesius NZ, Callaerts-Vegh Z, Yoshimura A, Hartmann D, Elgersma Y, D'Hooge R, et al. Spred1 is required for synaptic plasticity and hippocampus-dependent learning. J Neurosci. 2008;28:14443–9.

    CAS  PubMed  PubMed Central  Google Scholar 

  31. 31.

    Shah D, Praet J, Latif HA, Hofling C, Anckaerts C, Bard F, Morawski M, Detrez JR, Prinsen E, Villa A, et al. Early pathologic amyloid induces hypersynchrony of BOLD resting-state networks in transgenic mice and provides an early therapeutic window before amyloid plaque deposition. Alzheimers Dement. 2016;12:964–76.

    PubMed  Google Scholar 

  32. 32.

    Balschun D, Wolfer DP, Gass P, Mantamadiotis T, Welzl H, Schutz G, Frey JU, Lipp HP. Does cAMP response element-binding protein have a pivotal role in hippocampal synaptic plasticity and hippocampus-dependent memory? J Neurosci. 2003;23:6304–14.

    CAS  PubMed  PubMed Central  Google Scholar 

  33. 33.

    Iscru E, Goddyn H, Ahmed T, Callaerts-Vegh Z, D'Hooge R, Balschun D. Improved spatial learning is associated with increased hippocampal but not prefrontal long-term potentiation in mGluR4 knockout mice. Genes Brain Behav. 2013;12:615–25.

    CAS  PubMed  Google Scholar 

  34. 34.

    Bhattacharya S, Herrera-Molina R, Sabanov V, Ahmed T, Iscru E, Stober F, Richter K, Fischer KD, Angenstein F, Goldschmidt J, et al. Genetically induced retrograde amnesia of associative memories after neuroplastin ablation. Biol Psychiatry. 2017;81:124–35.

    CAS  PubMed  Google Scholar 

  35. 35.

    El-Hassar L, Milh M, Wendling F, Ferrand N, Esclapez M, Bernard C. Cell domain-dependent changes in the glutamatergic and GABAergic drives during epileptogenesis in the rat CA1 region. J Physiol. 2007;578:193–211.

    CAS  PubMed  Google Scholar 

  36. 36.

    Torborg CL, Nakashiba T, Tonegawa S, McBain CJ. Control of CA3 output by feedforward inhibition despite developmental changes in the excitation-inhibition balance. J Neurosci. 2010;30:15628–37.

    CAS  PubMed  PubMed Central  Google Scholar 

  37. 37.

    Shao M, Hirsch JC, Peusner KD. Plasticity of spontaneous excitatory and inhibitory synaptic activity in morphologically defined vestibular nuclei neurons during early vestibular compensation. J Neurophysiol. 2012;107:29–41.

    PubMed  Google Scholar 

  38. 38.

    Thal DR, Rub U, Orantes M, Braak H. Phases of A beta-deposition in the human brain and its relevance for the development of AD. Neurology. 2002;58:1791–800.

    PubMed  Google Scholar 

  39. 39.

    Rowan MJ, Klyubin I, Cullen WK, Anwyl R. Synaptic plasticity in animal models of early Alzheimer’s disease. Philos Trans R Soc Lond Ser B Biol Sci. 2003;358:821–8.

    CAS  Google Scholar 

  40. 40.

    Citri A, Malenka RC. Synaptic plasticity: multiple forms, functions, and mechanisms. Neuropsychopharmacology. 2008;33:18–41.

    PubMed  Google Scholar 

  41. 41.

    Collingridge GL, Peineau S, Howland JG, Wang YT. Long-term depression in the CNS. Nat Rev Neurosci. 2010;11:459–73.

    CAS  PubMed  Google Scholar 

  42. 42.

    Zucker RS, Regehr WG. Short-term synaptic plasticity. Annu Rev Physiol. 2002;64:355–405 355–405.

    CAS  PubMed  Google Scholar 

  43. 43.

    Larson J, Lynch G. Theta pattern stimulation and the induction of LTP: the sequence in which synapses are stimulated determines the degree to which they potentiate. Brain Res. 1989;489:49–58.

    CAS  PubMed  Google Scholar 

  44. 44.

    Buzsaki G, Moser EI. Memory, navigation and theta rhythm in the hippocampal-entorhinal system. Nat Neurosci. 2013;16:130–8.

    CAS  PubMed  PubMed Central  Google Scholar 

  45. 45.

    Wilsch VW, Behnisch T, Jager T, Reymann KG, Balschun D. When are class I metabotropic glutamate receptors necessary for long- term potentiation? J Neurosci. 1998;18:6071–80.

    CAS  PubMed  PubMed Central  Google Scholar 

  46. 46.

    Serrano-Pozo A, Frosch MP, Masliah E, Hyman BT. Neuropathological alterations in Alzheimer disease. Cold Spring Harb Perspect Med. 2011;1:a006189.

    PubMed  PubMed Central  Google Scholar 

  47. 47.

    Scheff SW, Price DA. Alzheimer’s disease-related alterations in synaptic density: neocortex and hippocampus. J Alzheimers Dis. 2006;9:101–15.

    PubMed  Google Scholar 

  48. 48.

    Walsh DM, Selkoe DJ. Deciphering the molecular basis of memory failure in Alzheimer’s disease. Neuron. 2004;44:181–93.

    CAS  PubMed  Google Scholar 

  49. 49.

    Pervolaraki E, Hall SP, Foresteire D, Saito T, Saido TC, Whittington MA, Lever C, Dachtler J. Insoluble Abeta overexpression in an App knock-in mouse model alters microstructure and gamma oscillations in the prefrontal cortex, and impacts on anxiety-related behaviours. Dis Model Mech. 2019;dmm.

  50. 50.

    Webster SJ, Bachstetter AD, Nelson PT, Schmitt FA, Van Eldik LJ. Using mice to model Alzheimer’s dementia: an overview of the clinical disease and the preclinical behavioral changes in 10 mouse models. Front Genet. 2014;5:–88. eCollection;%2014.:88.

  51. 51.

    Varela C, Kumar S, Yang JY, Wilson MA. Anatomical substrates for direct interactions between hippocampus, medial prefrontal cortex, and the thalamic nucleus reuniens. Brain Struct Funct. 2014;219:911–29.

    CAS  PubMed  Google Scholar 

  52. 52.

    Jin J, Maren S. Prefrontal-hippocampal interactions in memory and emotion. Front Syst Neurosci. 2015;9:170. eCollection;2015.:170.

    Article  PubMed  PubMed Central  Google Scholar 

  53. 53.

    Eichenbaum H. Prefrontal-hippocampal interactions in episodic memory. Nat Rev Neurosci. 2017;10..

  54. 54.

    Preston AR, Eichenbaum H. Interplay of hippocampus and prefrontal cortex in memory. Curr Biol. 2013;23:R764–73.

    CAS  PubMed  PubMed Central  Google Scholar 

  55. 55.

    Cullen WK, Suh YH, Anwyl R, Rowan MJ. Block of LTP in rat hippocampus in vivo by beta-amyloid precursor protein fragments. Neuroreport. 1997;8:3213–7.

    CAS  PubMed  Google Scholar 

  56. 56.

    Shankar GM, Li S, Mehta TH, Garcia-Munoz A, Shepardson NE, Smith I, Brett FM, Farrell MA, Rowan MJ, Lemere CA, et al. Amyloid-beta protein dimers isolated directly from Alzheimer’s brains impair synaptic plasticity and memory. Nat Med. 2008;14:837–42.

    CAS  PubMed  PubMed Central  Google Scholar 

  57. 57.

    Yang T, Li S, Xu H, Walsh DM, Selkoe DJ. Large soluble oligomers of amyloid beta-protein from Alzheimer brain are far less neuroactive than the smaller oligomers to which they dissociate. J Neurosci. 2017;37:152–63.

    CAS  PubMed  PubMed Central  Google Scholar 

  58. 58.

    Battaglia F, Wang HY, Ghilardi MF, Gashi E, Quartarone A, Friedman E, Nixon RA. Cortical plasticity in Alzheimer’s disease in humans and rodents. Biol Psychiatry. 2007;62:1405–12.

    CAS  PubMed  Google Scholar 

  59. 59.

    Huang YY, Nguyen PV, Abel T, Kandel ER. Long-lasting forms of synaptic potentiation in the mammalian hippocampus. Learn Mem. 1996;3:74–85.

    CAS  PubMed  Google Scholar 

  60. 60.

    Kelleher RJ III, Govindarajan A, Tonegawa S. Translational regulatory mechanisms in persistent forms of synaptic plasticity. Neuron. 2004;44:59–73.

    CAS  PubMed  Google Scholar 

  61. 61.

    Seabrook GR, Rosahl TW. Transgenic animals relevant to Alzheimer’s disease. Neuropharmacology. 1999;38:1–17.

    CAS  PubMed  Google Scholar 

  62. 62.

    Li Q, Navakkode S, Rothkegel M, Soong TW, Sajikumar S, Korte M. Metaplasticity mechanisms restore plasticity and associativity in an animal model of Alzheimer’s disease. Proc Natl Acad Sci U S A. 2017;114:5527–32.

    CAS  PubMed  PubMed Central  Google Scholar 

  63. 63.

    Ronicke R, Mikhaylova M, Ronicke S, Meinhardt J, Schroder UH, Fandrich M, Reiser G, Kreutz MR, Reymann KG. Early neuronal dysfunction by amyloid beta oligomers depends on activation of NR2B-containing NMDA receptors. Neurobiol Aging. 2011;32:2219–28.

    PubMed  Google Scholar 

  64. 64.

    Nussbaum JM, Schilling S, Cynis H, Silva A, Swanson E, Wangsanut T, Tayler K, Wiltgen B, Hatami A, Ronicke R, et al. Prion-like behaviour and tau-dependent cytotoxicity of pyroglutamylated amyloid-beta. Nature. 2012;485:651–5.

    CAS  PubMed  PubMed Central  Google Scholar 

  65. 65.

    Ahmed T, Sabanov V, D'Hooge R, Balschun D. An N-methyl-D-aspartate-receptor dependent, late-phase long-term depression in middle-aged mice identifies no GluN2-subunit bias. Neuroscience. 2011;185:27–38.

    CAS  PubMed  Google Scholar 

  66. 66.

    Wang HW, Pasternak JF, Kuo H, Ristic H, Lambert MP, Chromy B, Viola KL, Klein WL, Stine WB, Krafft GA, et al. Soluble oligomers of beta amyloid (1-42) inhibit long-term potentiation but not long-term depression in rat dentate gyrus. Brain Res. 2002;924:133–40.

    CAS  PubMed  Google Scholar 

  67. 67.

    Kim JH, Anwyl R, Suh YH, Djamgoz MB, Rowan MJ. Use-dependent effects of amyloidogenic fragments of (beta)-amyloid precursor protein on synaptic plasticity in rat hippocampus in vivo. J Neurosci. 2001;21:1327–33.

    CAS  PubMed  PubMed Central  Google Scholar 

  68. 68.

    Hsieh H, Boehm J, Sato C, Iwatsubo T, Tomita T, Sisodia S, Malinow R. AMPAR removal underlies Abeta-induced synaptic depression and dendritic spine loss. Neuron. 2006;52:831–43.

    CAS  PubMed  PubMed Central  Google Scholar 

  69. 69.

    Cheng L, Yin WJ, Zhang JF, Qi JS. Amyloid beta-protein fragments 25-35 and 31-35 potentiate long-term depression in hippocampal CA1 region of rats in vivo. Synapse. 2009;63:206–14.

    CAS  PubMed  Google Scholar 

  70. 70.

    Li S, Hong S, Shepardson NE, Walsh DM, Shankar GM, Selkoe D. Soluble oligomers of amyloid Beta protein facilitate hippocampal long-term depression by disrupting neuronal glutamate uptake. Neuron. 2009;62:788–801.

    CAS  PubMed  PubMed Central  Google Scholar 

  71. 71.

    Hu NW, Nicoll AJ, Zhang D, Mably AJ, O'Malley T, Purro SA, Terry C, Collinge J, Walsh DM, Rowan MJ. mGlu5 receptors and cellular prion protein mediate amyloid-beta-facilitated synaptic long-term depression in vivo. Nat Commun. 2014;5:3374.

    Article  PubMed  PubMed Central  Google Scholar 

  72. 72.

    Kuwabara Y, Ishizeki M, Watamura N, Toba J, Yoshii A, Inoue T, Ohshima T. Impairments of long-term depression induction and motor coordination precede Abeta accumulation in the cerebellum of APPswe/PS1dE9 double transgenic mice. J Neurochem. 2014;130:432–43.

    CAS  PubMed  Google Scholar 

  73. 73.

    Raymond CR, Ireland DR, Abraham WC. NMDA receptor regulation by amyloid-beta does not account for its inhibition of LTP in rat hippocampus. Brain Res. 2003;968:263–72.

    CAS  PubMed  Google Scholar 

  74. 74.

    Cook SG, Goodell DJ, Restrepo S, Arnold DB, Bayer KU. Simultaneous live imaging of multiple endogenous proteins reveals a mechanism for Alzheimer’s-related plasticity impairment. Cell Rep. 2019;27:658–65.

    CAS  PubMed  PubMed Central  Google Scholar 

  75. 75.

    Kootar S, Frandemiche ML, Dhib G, Mouska X, Lorivel T, Poupon-Silvestre G, Hunt H, Tronche F, Bethus I, Barik J, et al. Identification of an acute functional cross-talk between amyloid-beta and glucocorticoid receptors at hippocampal excitatory synapses. Neurobiol Dis. 2018;118:117–28. Epub;2018 Jul 9.:117–28.

    CAS  Article  PubMed  Google Scholar 

  76. 76.

    Roder S, Danober L, Pozza MF, Lingenhoehl K, Wiederhold KH, Olpe HR. Electrophysiological studies on the hippocampus and prefrontal cortex assessing the effects of amyloidosis in amyloid precursor protein 23 transgenic mice. Neuroscience. 2003;120:705–20.

    CAS  PubMed  Google Scholar 

  77. 77.

    Curtis DR, Eccles JC. Synaptic action during and after repetitive stimulation. J Physiol. 1960;150:374–98 374–98.

    CAS  PubMed  PubMed Central  Google Scholar 

  78. 78.

    Pak JH, Huang FL, Li J, Balschun D, Reymann KG, Chiang C, Westphal H, Huang KP. Involvement of neurogranin in the modulation of calcium/calmodulin-dependent protein kinase II, synaptic plasticity, and spatial learning: a study with knockout mice. Proc Natl Acad Sci U S A. 2000;97:11232–7.

    CAS  PubMed  PubMed Central  Google Scholar 

  79. 79.

    Huang KP, Huang FL, Jager T, Li J, Reymann KG, Balschun D. Neurogranin/RC3 enhances long-term potentiation and learning by promoting calcium-mediated signaling. J Neurosci. 2004;24:10660–9.

    CAS  PubMed  PubMed Central  Google Scholar 

  80. 80.

    Chang EH, Savage MJ, Flood DG, Thomas JM, Levy RB, Mahadomrongkul V, Shirao T, Aoki C, Huerta PT. AMPA receptor downscaling at the onset of Alzheimer’s disease pathology in double knockin mice. Proc Natl Acad Sci U S A. 2006;103:3410–5.

    CAS  PubMed  PubMed Central  Google Scholar 

  81. 81.

    Masliah E, Mallory M, Alford M, DeTeresa R, Hansen LA, McKeel DW Jr, Morris JC. Altered expression of synaptic proteins occurs early during progression of Alzheimer’s disease. Neurology. 2001;56:127–9.

    CAS  PubMed  Google Scholar 

  82. 82.

    Yao PJ, Zhu M, Pyun EI, Brooks AI, Therianos S, Meyers VE, Coleman PD. Defects in expression of genes related to synaptic vesicle trafficking in frontal cortex of Alzheimer’s disease. Neurobiol Dis. 2003;12:97–109.

    CAS  PubMed  Google Scholar 

  83. 83.

    de Wilde MC, Overk CR, Sijben JW, Masliah E. Meta-analysis of synaptic pathology in Alzheimer’s disease reveals selective molecular vesicular machinery vulnerability. Alzheimers Dement. 2016;12:633–44.

    PubMed  PubMed Central  Google Scholar 

Download references


The authors wish to thank Bart De Strooper (VIB Centre for Brain Disease Research & KU Leuven, Leuven, Belgium; UK Dementia Research Institute, University College London, London, UK) for providing the AppNL-G-F and AppNL mice and for critical comments to the manuscript. We further thank Keimpe Wierda (VIB Centre for Brain Disease Research & KU Leuven, Leuven, Belgium) for a very detailed discussion of the manuscript. Marc Mercken (Janssen Pharmaceutica, Belgium) kindly provided end-specific antibodies.


This work was supported by the Research Foundation-Flanders (FWO grants G.0327.08 and G.0D76.14).

Author information




ALH, TA, and DB designed the study. ALH, VS, TA, and KC performed the experiments. ALH, VS, TA, KC, and DB analyzed the data. ALH, VS, TA, KC, TS, TS, and DB interpreted the data. ALH and DB wrote the manuscript with input from all co-authors. All authors read and approved the final version of the manuscript.

Corresponding author

Correspondence to Detlef Balschun.

Ethics declarations

Ethics approval and consent to participate

The housing conditions and procedures to prepare acute brain slices were approved by the KU Leuven Ethical Committee and in accordance with European Directive 2010/63/EU.

Consent for publication

Not applicable.

Competing interests

All authors declare that they have no competing interests concerning the present study.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Latif-Hernandez, A., Sabanov, V., Ahmed, T. et al. The two faces of synaptic failure in AppNL-G-F knock-in mice. Alz Res Therapy 12, 100 (2020).

Download citation


  • App knock-in mice
  • Long-term potentiation
  • Long-term depression
  • Miniature synaptic currents
  • Presynaptic glutamatergic and GABAergic upregulation
  • Electrophysiological phenotyping