All experiments followed the policies and guidelines of the Canadian Council on Animal Care and the animal care regulations of McGill University. TgCRND8 mice bear Swedish KM670/671NL and Indiana V717F mutations in the hAβAPP-encoding gene and overexpress human Aβ by 3–4 months. Male TgCRND8 and non-transgenic (nTg) mice were maintained on an outbred C3H/C57BL6 background and kept on a 12 h light/dark cycle with food and water ad libitum.

βCTF content was quantified from hippocampal homogenates from Tg mice, as preliminary examination indicated that it is undetectable in nTg mice (data not shown). This was expected, as the antibody used in the enzyme-linked immunoabsorbent assay (ELISA) kit (IBL International, Japan) is directed against the human βCTF protein. Hippocampi from male TgCRND8 mice (n = 8 mice) were dissected on ice, snap frozen and stored at −80°C until proteins were extracted. Hippocampus samples were homogenized in radioimmunoprecipitation assay (RIPA) buffer (100 μl per hippocampus) and left on ice for 30 min. The samples were then centrifuged for 10 min at 1000 g at 4°C. The supernatants were collected and protein concentrations were determined using a BCA assay kit (Pierce, Rockford, IL, USA). The βCTF content was then assessed in triplicates using an ELISA kit (IBL International, Japan) as per manufacturer’s instructions. Briefly, samples were diluted in enzyme immunoassay (EIA) buffer (1:100) and standards were prepared as directed. 100 μl of each sample was loaded into the appropriate wells and incubated overnight at 4°C. The plate was then washed seven times using diluted wash buffer. One hundred microliter of labeled antibody solution was loaded into each well, and the plate was incubated for 1 h at 4°C. The plate was subsequently washed nine times and 100 μl of chromogen was added to each well. After 30 min of incubation in the dark at room temperature, 100 μl of stop solution was added to each well, and the plate was analyzed using a plate reader at 450 nm against the reagent blank.

Additional male mice (n = 5 mice per experimental group) were anesthetized by pentobarbital and transcardially perfused (PBS followed by 4% PFA). Brains were stored in fixative for 24 h at 4°C, then in a sucrose solution (30% in PBS) for 3 days at 4°C, frozen using dimethylbutane and stored at −80°C. Brains were sliced coronally using a freezing microtome at 40 μm for light microscopy, or using a vibrating microtome at 50 μm for fluorescence microscopy, and free-floating sections were stored in a cryopreservative solution (3:3:4 glycerol:ethylene glycol:PBS) at −20°C in preparation for IHC (IHC) staining. The serial sectioning fraction for IHC was 1/8). Washes in PBS preceded all steps except primary antibody addition. All steps were at room temperature unless otherwise specified.

For immunofluorescence, sections were incubated for 5 min with formic acid (90%). They were then incubated for 1 h with PBS containing 1% normal goat serum, 0.25% Triton X-100 and 0.45% gelatin. Sections were further incubated overnight at 4°C with either FCA3340 (Barelli et al., 1997) to detect Aβ (Millipore 171608) or CT20 (Millipore 171610) to detect βCTF (rabbit, 1:1000 dilution for both). Sections were incubated for 2 h with secondary antibody (Alexa 555-conjugated goat anti-rabbit IgG; ThermoFisher A-21428, 1:2000). Sections were incubated with DAPI (1:10,000 dilution) for 15 min at room temperature. Sections were then mounted on glass slides with Fluoromount G. Fluorescence was visualized using an epifluorescence (Axioplan2, Zeiss) microscope.

For the study of neuronal populations, sections were incubated in PBS with 0.2% Triton X-100 for 1.5 h, followed by 3% H₂O₂ prepared in PBS for 10 min. The sections were blocked with 2% normal serum in PBS + 0.2% Triton X-100 for 1 h. Primary antibodies (NeuN: mouse, 1:200, Millipore MAB377; NPY: rabbit, 1:2000, Abcam ab10980; PV: rabbit, 1:5000, Abcam ab11427) were applied overnight at 4°C, followed by biotinylated secondary antibodies (Vector BA-1000, 1:200; BA-2001, 1:500) for 1 h. The immunolabeled product was visualized using the avidin-biotin complex method (Vectastain elite ABC Kit, Vector Laboratories; 30 min), followed by DAB (for NeuN) or SG (for NPY and PV) chromogen development (Vector Laboratories). Stained sections were mounted on slides, dried overnight, dehydrated in ethanol, cleared in xylene and coverslipped. Parallel sections from the same animals were used across immunohistochemical experiments.

Immunoreactive (IR) cell somata were counted in each hippocampal region and sub-region (from Bregma −1.06 mm to Bregma −3.88 mm; Franklin and Paxinos, 2007), on a Nikon Eclipse E600 (Kanagawa, Japan) microscope with a 20X objective, by an experimenter blind to group identity. Anatomical regions were determined as per Franklin and Paxinos (2007). CA1 and CA2 were combined (CA1/2) due to the relatively amorphous boundary separating these two regions. Intra-subject counts by an additional experimenter (on a Leica DM 2500 microscope) to confirm count/total accuracy correlated significantly (p = 0.0062), with Pearson r and intraclass correlation values >0.5. The serial section fraction was 1/8. Cell quantifications are shown as numbers of cells per section to correct for number of sections and for consistency between experiments. Pairwise comparisons between nTg and TgCRND8 animals in hippocampal regions and sub-regions were performed using unpaired t-tests, with Welch’s correction applied when required.

For GABAergic markers, we were able to count each labeled cell in order to quantify. However, given the large number of NeuN-labeled cells, it was necessary to use stereological estimates to quantify this marker, as in previous studies (Pham et al., 2003). Although this method relies on estimates as opposed to absolute cell counts, and is not appropriate for quantification of all hippocampal cell types (Noori and Fornal, 2011), it is an optimal quantification method for this larger cellular population. NeuN-IR quantifications were performed by an experimenter blind to group identity on an Olympus BX51 microscope with a motorized stage, using StereoInvestigator software (MBF Bioscience). The stereological parameters (optimized for the NeuN-IR population) were: sampling grid area 28440 μm², counting frame 25 μm × 25 μm, dissector height 10 μm, guard zone 1 μm. Volume was assessed using a Cavalieri estimator (20 μm grid spacing) corrected for overprojection. Average coeffecients of error for Cavalieri probes were 0.025 for CA1/2 and CA3 and 0.033 for dentate gyrus (DG), and for optical fractionator (Gundersson m = 1) were 0.051 for CA1/2, 0.060 for CA3 and 0.054 for DG.

Tg and nTg mice (aged 30–35 days; n = 6/group) were decapitated, and the brain was rapidly removed and placed in ice-cold high sucrose artificial CSF (ACSF) solution (in mM: 252 sucrose, 3 KCl, 2 MgSO₄ 24 NaHCO₃, 1.25 NaH₂PO₄, 1.2 CaCl₂ and 10 glucose) and bubbled with carbogen (95% O₂ and 5% CO₂). The cerebellum and frontal cortex were removed with a razor blade, and the hemispheres were separated and allowed to recover for 2–3 min in the oxygenated sucrose solution. Complete septo-hippocampal isolate was then removed from the remaining hemisection as described previously (Goutagny et al., 2009). After dissection, the complete septo-hippocampal preparation was left at room temperature in ACSF bubbled with carbogen for 60 min. For recording, the preparation was transferred quickly to a custom-made submerged recording chamber. Recordings were performed at 30–32°C after an additional 30 min period in the chamber. The preparation was continuously perfused with ACSF (25 ml/min, in mM: 126 NaCl, 24 NaHCO₃, 10 glucose, 4.5 KCl, 2 MgSO₄, 1.25 NaH₂PO₄ and 2 CaCl₂, pH 7.4, with 95% O₂/5% CO₂) via a gravity-fed perfusion system and maintained at 30–32°C. Local field potentials were recorded using glass micropipettes (2–6 MΩ) filled with ACSF. Signals were recorded through a differential AC amplifier (A-M Systems), filtered online (0.1–500 Hz), and sampled at 5 KHz. All drugs came from aliquots of stock solutions (stored at −80°C) and were added to the perfusing artificial ACSF at the concentrations indicated. Base line recording lasted for 20 min followed by 100 s of pharmacological stimulation (4AP at 150 μM) and 20 min recovery after stimulation. Changes in theta power were measured in mV²/Hz. Pairwise comparisons were performed with t-tests. For all experiments, a p value of ≤0.05 considered statistically significant. Bar graphs show experimental mean, with error bars indicating standard error of the mean.

Immunohistochemical staining for three neuronal markers (NeuN, NPY and PV) revealed distributional patterns in CA1/2, CA3, DG and subiculum (Figure 1A) that have been described previously (Albuquerque et al., 2015). Assessment of Aβ and amyloid-plaques using the FCA3340 antibody, which specifically recognizes human Aβ but not APP (Barelli et al., 1997), indicated no immunolabeling in hippocampus sections obtained from 1 month-old TgCRND8 mice, in contrast to the abundance of plaques present in the hippocampus of 11 month-old TgCRND8 mice, used as a positive control (Chishti et al., 2001; Figure 1B). However, the CT20 antibody which recognizes both human and murine full length APP and its C-terminal fragments (CTF) α, β and amyloid precursor protein intracellular domain (AICD) but not Aβ, revealed IR cells, notably in the pyramidal cell layer of the CA1/CA2 region (Figure 1B). CT20-IR cells displayed punctate cytoplasmic labeling (Figure 1B). Combined, the labeling data obtained with FCA3340 and CT20 indicated that immunoreactivity likely arises from the presence of APP and its first cleavage products (CTFs) in the absence of Aβ. Since the observed putative CTF-IR could arise from different fragments (αCTF, βCTF or AICD) generated along either the amyloidogenic and non-amyloidogenic pathways, we performed a selective βCTF ELISA to assess whether the CT20-IR could be due specifically to the βCTF. The presence of βCTF was confirmed and quantified (1.47 pg βCTF/mg protein ± 0.22, range 0.80–2.43 pg βCTF/mg protein) by ELISA in the hippocampal tissue of TgCRND8 mice aged 1 month.

Phenotypic analysis along the rostro-caudal axis (−0.94 to −2.86 mm from Bregma) using the neuronal marker NeuN indicated that both total number (Figure 2A) and density (Figure 2C) of neurons were similar between genotypes in all studied regions of the hippocampus (CA1/2, CA3 and DG). Interestingly, structural volume was significantly decreased for TgCRND8 mice in CA3 (p = 0.041) and DG (p = 0.017) but not CA1/2, as compared to controls (Figure 2B).

However, when assessing specifically NPY (Figure 3A; p = 0.016) or PV (Figure 3B; p = 0.016) subpopulations, a significant decrease in numbers of IR-cells was found in TgCRND8 mice compared to age-matched nTg littermates.

A more detailed analysis indicated that the overall number of NPY neurons is decreased in hippocampal sub-regions CA1/2 (p = 0.021) and DG (p = 0.0033), but not CA3 or subiculum (Figures 4A–D) of TgCRND8 mice as compared to controls. The most affected sub-regions in CA1/2 were the stratum pyramidale (p = 0.021) and oriens (p = 0.026) layers (Figure 4E). No significant difference was seen between genotypes in the layers of CA3 (Figure 4F). By contrast, in the DG, the polymorphic layer (PO; p = 0.0031) and granule cell layer (GR; p = 0.0051) showed alterations in TgCRND8 mice, whereas the molecular layer (MO) was unaffected (Figure 4G).

An analogous analysis of PV-expressing neuronal sub-populations showed a significant decrease in the number of these neurons in CA1/2 (Figure 5A; p = 0.017) and subiculum (Figure 5D; p = 0.030) whereas CA3 (Figure 5B) and DG (Figure 5C) were not significantly affected. Among the layers analyzed in these sub-regions (Figures 5E–G), the number of PV-expressing neurons in TgCRND8 mice was significantly decreased only in the pyramidal cell layer of the CA1/2 region (Figure 5E; p = 0.0004).

A recent study reported that PV-expressing neurons at the interface between CA1/2 and subiculum are the primary regulators of hippocampal neuronal network oscillations at theta frequencies (Amilhon et al., 2015). Therefore, we decided to focus further on the PV-expressing sub-population of interneurons.

To explore the mechanisms of the observed decrease in the number of PV-expressing interneurons in 1 month-old TgCRND8 mice, we used a double-labeling approach to determine whether, in the absence of Aβ (Figure 1B), PV-expressing neurons may co-express βCTF. Using anti-PV and FCA3340 antibodies, we first confirmed the absence of Aβ in the PV-IR neurons located in the pyramidal layer (PY) of CA1 in 1 month-old TgCRND8 mice (Figure 6A). By contrast, in the anatomically matched hippocampal region from 11 month-old TgCRND8 mice, which was used as a positive control, cytoplasmic Aβ labeling was revealed in pyramidal neurons.

As our ELISA data demonstrated βCTF expression in the hippocampus in 1 month-old TgCRND8 mice, we next asked whether PV-expressing neurons may also express βCTF. Our double-labeling approach indicated that CTF-IR (as revealed by the CT20 antibody) does not co-localize with PV-IR (Figure 6B). Interestingly, in 1 month-old TgCRND8 mice, APP and CTFs were apparently localized in the upper portion of the pyramidal cell layer where PV-IR neurons were virtually never detected (Figure 6B). Indeed, PV-expressing neurons were localized in the lower portion of the PY (Figure 6B).

Transgenic mouse models of AD at 5 months of age are characterized by high levels of Aβ peptides potentially leading to network hyperexcitability (Palop et al., 2007). However, our recent data examining Tg mice at 1 month of age has not indicated the presence of Aβ peptide (Goutagny et al., 2013); instead, we observed elevated βCTF and decreased numbers of hippocampal PV-expressing neurons in the current study. Aiming to assess whether these alterations could account for network hyperexcitability (changes in theta amplitude, theta frequency and burst (seizure) events), we utilized 4-aminopyridine (4AP) in a complete septo-hippocampal preparation (Goutagny et al., 2009). The hyperexcitability network state is achieved by blocking primarily K_v1 channels, which consequently induces activity reflecting the firing of GABA-releasing cells and is sustained by GABA_A receptor signaling (Avoli and de Curtis, 2011). Our hypothesis was that 4AP would reveal a state of hyperexcitability in juvenile TgCRND8 mice.

Treatment with 4AP did not elicit a visible change in the amplitude of theta activity in nTg mice (Figures 7A,B, red square). Conversely, in Tg mice 4AP treatment affected amplitude of hippocampal theta activity (Figure 7D, red square). Magnification of raw activity traces confirmed an increase in theta amplitude of Tg mice during 4AP treatment (Figure 7E). As shown in the spectrogram analysis, theta oscillations frequency remains stable (4.10 ± 1.68 Hz) in nTg mice when treated with 4AP (Figure 7C, red square). In contrast, although not statistically significant (p = 0.29), Tg mice showed changes in frequency during 4AP stimulation (Figure 7F, red square). Statistical analysis further confirmed our findings, as no significant changes were detected in frequency either during (p = 0.29) or after 4AP stimulation (p = 0.53; Figures 7G,H). Concomitantly, statistical analysis further confirmed the 7.5 ± 2.44-fold increase in theta power during 4AP stimulation in Tg mice (Figure 7I, p = 0.02). The increase was also significant after 4AP stimulation (Figure 7I, 5.03 ± 1.29 p = 0.01). The changes in theta peak power were also significant when Tg mice were compared to nTg during 4AP (Figure 7J, p = 0.038) and after stimulation (Figure 7J, p = 0.023).

At this point, the data suggested that the network state in TgCRND8 mice is closer to a hyperexcitability-like state when compared to controls. To further explore the excitability network state, we analyzed the number of burst events present in the nTg (Figure 7K) and Tg mice (Figure 7L). Our results showed an elevated number of burst events in Tg mice when treated with 4AP (Figure 7L), further supporting a hyperexcitability-like network state. In contrast, control mice showed a more stable network state (Figure 7K). Further statistical analysis confirmed a two-fold increase (2.4 ± 0.19-fold, p = 0.012) in relative burst events in Tg mice when compared to nTg mice (Figures 7M,N).