Intraneuronal Aβ accumulation induces hippocampal neuron hyperexcitability through A-type K⁺ current inhibition mediated by activation of caspases and GSK-3

Abstract

Amyloid β-protein (Aβ) pathologies have been linked to dysfunction of excitability in neurons of the hippocampal circuit, but the molecular mechanisms underlying this process are still poorly understood. Here, we applied whole-cell patch-clamp electrophysiology to primary hippocampal neurons and show that intracellular Aβ₄₂ delivery leads to increased spike discharge and action potential broadening through downregulation of A-type K⁺ currents. Pharmacologic studies showed that caspases and glycogen synthase kinase 3 (GSK-3) activation are required for these Aβ₄₂-induced effects. Extracellular perfusion and subsequent internalization of Aβ₄₂ increase spike discharge and promote GSK-3-dependent phosphorylation of the Kv4.2 α-subunit, a molecular determinant of A-type K⁺ currents, at Ser-616. In acute hippocampal slices derived from an adult triple-transgenic Alzheimer’s mouse model, characterized by endogenous intracellular accumulation of Aβ₄₂, CA1 pyramidal neurons exhibit hyperexcitability accompanied by increased phosphorylation of Kv4.2 at Ser-616. Collectively, these data suggest that intraneuronal Aβ₄₂ accumulation leads to an intracellular cascade culminating into caspases activation and GSK-3-dependent phosphorylation of Kv4.2 channels. These findings provide new insights into the toxic mechanisms triggered by intracellular Aβ₄₂ and offer potentially new therapeutic targets for Alzheimer’s disease treatment.

1. Introduction

Alzheimer’s disease (AD) is the most common form of dementia with a high prevalence among the aging population (Citron, 2010). Extracellular fibrillar amyloid β-protein (Aβ) accumulation and neurofibrillary tangles composed of hyperphosphorylated tau are considered the neuropathologic hallmarks of AD (Selkoe, 2001). Aβ originates from the amyloid β-precursor protein (APP) through sequential cleavage by the β- and the γ-secretase enzyme complex (Thinakaran and Koo, 2008). Most of the full-length Aβ peptide consists of the 40-amino-acid long Aβ (Aβ₄₀), whereas a small proportion (10%) is the 42-residues variant (Aβ₄₂), a species that because of its high hydrophobicity has been considered more prone to fibril formation and neurotoxicity than the Aβ₄₀ (Jarrett et al., 1993).

A plethora of neurophysiological studies have reported toxic activity of Aβ₄₂ oligomers on synaptic function and activity-dependent plasticity in the hippocampus and cerebral cortex building the notion that AD is a synaptopathy (Crimins et al., 2013; Klyubin et al., 2012; Selkoe, 2002; Shankar et al., 2008; Sheng et al., 2012). It is believed, indeed, that disruption in synaptic structure and function induced by Aβ₄₂ is one of the underlying mechanisms leading to the aberrant neuronal processing and network dysfunction that characterize AD cognitive impairment and memory loss (Mucke and Selkoe, 2012). Surprisingly, although, much less knowledge has been gained on other Aβ toxic effects on neuronal activity that might precede synaptic dysfunction and as such could be targeted for early therapeutic interventions against AD.

Emerging evidence indicates that changes in neuronal excitability could play a fundamental role in initiating early Aβ₄₂ pathology predisposing to and/or triggering subsequent synaptic dysfunction. In vitro and animal model studies have demonstrated that Aβ pathology associates with increased excitability of hippocampal neurons, leading to hypersynchronous network activity and higher risk for seizures (Born et al., 2014; Brown et al., 2011; Busche et al., 2012; Davis et al., 2014; Del Vecchio et al., 2004; Minkeviciene et al., 2009; Palop et al., 2007; Putcha et al., 2011). Epidemiologic studies have confirmed comorbidity between patients with AD and epilepsy, showing increased risk of seizures in AD patients (Amatniek et al., 2006; Hommet et al., 2008). This repertoire of cellular and clinical studies indicating potential commonalities in early AD and epilepsy opens new horizons in the understanding of early AD triggers and therapies. However, despite this evidence, the molecular mechanisms linking Aβ pathology to aberrant neuronal firing is still poorly understood. A missing link in associating Aβ effects to neuronal excitability is the identification of the source of toxic oligomers.

Early studies based on the observation of senile plaques in the extracellular space (Hardy and Selkoe, 2002) have pointed for extracellular Aβ₄₂ (eAβ₄₂) as the primary cause of AD. However, in studies on postmortem AD and mild cognitively impaired patients immunoreactivity of intracellular Aβ₄₂ (iAβ₄₂) has been identified in neurons of the hippocampus and entorhinal cortex (Gouras et al., 2000; Mori et al., 2002), the 2 main brain regions affected in early AD pathology. These findings have been corroborated by results from transgenic mouse models (Guzman et al., 2014; Youmans et al., 2012; for review see; LaFerla et al., 2007), leading to the overall hypothesis of iAβ accumulation as an early triggering event in the AD progression preceding senile plaques (Kuo et al., 2001; LaFerla et al., 2007; Oakley et al., 2006). Intracellular Aβ-driven effects on neuronal firing might therefore be one of the earliest detectable triggers of the AD pathology preceding and predisposing to synaptic deficits.

Building on the growing interest in identifying early Aβ targets, we studied the effect of iAβ₄₂ on neuronal excitability in primary hippocampal neurons and in CA1 hippocampal neurons of triple-transgenic AD (3xTg-AD) mouse model that harbor the mutant genes for amyloid precursor protein (APPSwe), presenilin 1PS1M146V and for tauP301L (Oddo et al., 2003) and overexpresses intracellular Aβ₄₂.

We found that Aβ₄₂ internalization is required for inducing increase in intrinsic excitability in hippocampal pyramidal neurons, an effect that is mediated by inhibition of A-type K⁺ channels (Kv), and requires caspases activation and glycogen synthase kinase 3-(GSK-3) dependent phosphorylation of Kv4.2. Parallel changes in intrinsic excitability, A-type K⁺ currents, and Kv4.2 phosphorylation are found in CA1 hippocampal neurons derived from brain slices of 3xTg-AD mice.

2. Methods

2.1. Animals

A colony of 3xTg-AD (harboring the PS1 M146V, APPSwe KM670/671NL, and tau P301L transgenes) and non-transgenic (B6129SF2/J, named non-Tg) mice were used for electrophysiological experiments in brain slices and for Western blot analysis. The colonies were established in-house from breeding pairs purchased by the Jackson Laboratory. All animal procedures were approved by the Ethics Committee of the Università Cattolica “S. Cuore” and were fully compliant with the Italian and European Union legislation on animal research.

2.2. Preparation of Aβ₄₂ solutions

Freeze-dried purified Aβ₄₂, Aβ_42-1, and Aβ₄₂ variant harboring oxidized methionine in position 35 ($\mathrm{A}{\beta }_{42}^{\mathrm{MO}}$) were purchased from rPeptide (Bogard, GA) and AnaSpec (Fremont, CA). Protein solutions were prepared as previously described (Ripoli et al., 2013), according to standard procedures. Briefly, peptides were diluted to 1 mM in 1,1,1,3,3,3,-hexafluoro-2-propanol to disassemble preformed aggregates and stored as dry films at −20 °C before use. The films were dissolved at 1 mM in dimethyl sulfoxide (DMSO), sonicated for 10 minutes, diluted to 100 μM in cold phosphate-buffered saline (PBS), and incubated for 12 hours at 4 °C to promote protein oligomerization. The final working concentration of 200 nM was obtained by diluting the 100 μM Aβ₄₂ in extracellular or intracellular solutions (for salt concentrations see the following sections). The same amount of DMSO to PBS contained in Aβ₄₂ solutions was used as vehicle.

2.3. Primary hippocampal neuron cultures

Hippocampal neurons from postnatal day 0–2, C57Bl/6 mice brains were prepared according to standard procedure (Piacentini et al., 2008) with some modifications. Briefly, hippocampi were incubated for 10 minutes at 37 °C in PBS containing trypsin-ethylenediaminetetraacetic acid (0.025%/0.01% wt/vol; Biochrom AG, Berlin, Germany), and the tissue was mechanically dissociated at room temperature (RT) with a fire-polished Pasteur pipette. The cell suspension was harvested and centrifuged at 235g for 8 minutes. The pellet was suspended in 88.8% Minimum Essential Medium (Biochrom), 5% fetal bovine serum, 5% horse serum, 1% glutamine (2 mM), 1% penicillin-streptomycin-neomycin antibiotic mixture (Invitrogen, Carlsbad, CA), and glucose (25 mM). Cells were plated at a density of approximately 100,000 cells on 20-mm coverslips precoated with poly-L-lysine (0.1 mg/mL; Sigma, St. Louis, MO). Twenty-four hours later, the culture medium was replaced with a mixture of 96.5% Neurobasal medium (Invitrogen), 2% B-27 (Invitrogen), 0.5% glutamine (2 mM), and 1% penicillin-streptomycin-neomycin antibiotic mixture. After 72 hours, this medium was replaced with a glutamine-free version of the same medium, and the cells were grown for 10 more days before experiments. The removal of L-glutamine from the culture medium was used to limit the grown of “mitotic” cells, such as astrocytes, and to reduce the possible excitotoxicity induced by glutamate obtained from the enzymatic conversion of L-glutamine.

2.4. Study of Aβ internalization

To study internalization of Aβ₄₂, both wt and $\mathrm{A}{\beta }_{42}^{\mathrm{MO}}$ were labeled with the IRIS 5-NHS active ester dye (IRIS 5; λex: 633 nm; λem: 650–700 nm; Cyanine Technology, Turin, Italy) according as previously described (Ripoli et al., 2013). IRIS-5 dye is suitable for conjugation of any biomolecules carrying free primary amines, such as proteins and peptides. Briefly, Aβ solutions (100 μM in PBS) were mixed with 6 mM IRIS 5 in DMSO for 4 hours in the dark under mild shaking conditions. After this time, labeled Aβs were purified with Vivacon 500 ultrafiltration spin columns (2 KDa cutoff; Sartorius Stedim Biotech GmbH, Goettingen, Germany) and then resuspended in PBS at a concentration of 100 μM before final dilution in the culture medium. Time-dependent internalization IRIS-5-labeled Aβ₄₂ (either wt or MO) was then studied by immunocytochemistry in hippocampal neurons derived from C57 mice.

Hippocampal neurons cultured for 15 days and treated with IRIS-5-labeled Aβ42 analogs were fixed with 4% paraformaldehyde (Sigma) in PBS for 15 minutes at RT. After being permeabilized (15 minutes incubation with 0.3% Triton X-100 [Sigma] in PBS), cells were incubated for 20 minutes with 0.3% BSA in PBS to block nonspecific-binding sites and then overnight at 4 °C with mouse anti-microtubule–associated protein 2 (MAP2, 1:300; Sigma). The subsequent day, cells were washed twice in PBS and then reincubated for 90 minutes at RT with Alexa-fluor 488-labeled donkey anti-mouse secondary antibody. Images (512 × 512 pixels) were acquired at 63× magnification with a confocal laser scanning system (TCS-SP2, Leica) and an oil-immersion objective (N.A. 1.4). Alexa-fluor 488 and IRIS-5 were excited at 488 and 633 nm, respectively with Ar/Kr and He/Ne lasers. The fluorescent signals emitted from fluorophores were revealed in spectral windows ranging from 500 to 530 nm for Alexa 488 and from 650 to 700 nm for IRIS-5. DAPI staining was imaged after 2-photon excitation (760 nm) with an ultrafast, tunable, mode-locked titanium:sapphire laser (Chamaleon, Coherent Inc).

2.5. Electrophysiology on primary hippocampal neurons

Recordings were obtained with an Axopatch 200B amplifier (Molecular Devices), and stimulation and data acquisition were performed with the Digidata 1200 series interface and pCLAMP 10 software (Molecular Devices). Patch electrodes, fabricated from borosilicate glass capillaries with the aid of a micropipette puller (P-97, Sutter Instruments, Novato, CA) had resistances of 3–5 MΩ when filled with the internal solution that contained (in mM): 146 mM K-gluconate, 18 mM HEPES, 1 mM EGTA, 4.6 mM MgCl₂, 4 mM NaATP, 0.3 mM Na₂GTP, and 15 mM creatine phosphate, (pH 7.4). For recordings, cells were constantly perfused with an external Tyrode solution containing the following (in mM): 140NaCl; 2 KCl; 10 HEPES; 10 glucose; 4 MgCl₂; and 4 CaCl₂ (pH 7.4; 312 mOsm). Primary neuronal cultures used in our study contained a mixed population of neurons and glial cells. According to previous studies (Bonin et al., 2007), pyramidal neurons were selected based on their pyramidal-like shape soma (approximately 15–20 μm in diameter) with prominent apical dendrites and physiological properties. A few cells (approximately 2%) exhibited large, fast after hyperpolarizations and higher discharge rates than those of most cells—features that are typical of fast-spiking GABAergic interneurons— and they were excluded from further investigation. We monitored the access resistance and membrane capacity before and at the end of the experiments to ensure recording stability and the health of studied cells (Grassi et al., 2004). Recordings were considered stable when the series and input resistances, resting-membrane potential, and stimulus artifact duration did not change >20%. To measure spike firing, current pulses (800 ms duration, from −50 to 250 pA) were applied in current-clamp mode. Depolarizing current pulses were applied every 5 minutes. Membrane potentials were held at approximately −70 mV during interpulse interval by passage of direct current with the patch amplifier. Changes in firing after 20 minutes of exposure to intracellular or extracellular Aβ₄₂ given compound were determined as the percent change in number of action potentials (APs) generated relative to baseline (T = 0). For eliciting single spikes, brief depolarization currents (1 nA; 2 ms) were injected. Spike halfwidth was measured at 50% of the peak. The transient A-type K⁺ current involtage–clamp mode was recorded by applying 300 ms voltage step from −90 to +50 mV. The transient channel was inactivated for a voltage step from −30 to +50 mV, leaving the sustained current of the total outward current. The transient current was then isolated by a digital subtraction of the sustained current from the total outward current.

2.6. Slices preparation and electrophysiology

Coronal slices (300- to 400-μm thick) containing the hippocampi were prepared using standard procedures (Podda et al., 2004, 2008; Curcio et al., 2013). Briefly, the animals were anesthetized with halothane (Sigma) and decapitated. The brains were rapidly removed and placed in ice-cold cutting solution containing in mM: 124 NaCl, 3.2 KCl,1 NaH₂PO₄, 26 NaHCO₃, 2 MgCl₂, 1 CaCl₂, 10 glucose, 2 Na-pyruvate, and 0.6 ascorbic acid (pH 7.4, 95%O₂/5% CO₂). Slices were cut with a vibratome (VT1000 S, Leica Microsystems, Wetzlar, Germany) and incubated in the cutting solution at 32 °C for at least 1 hour and then at RT until use. Slices were then transferred to a submerged recording chamber and continuously perfused with artificial cerebrospinal fluid bubbled with 95% O₂ and 5% CO₂ (pH 7.4). The artificial cerebrospinal fluid contained (in mM): 124 NaCl, 3.2 KCl, 1 NaH₂PO₄, 26 NaHCO₃, 1 MgCl₂, 2 CaCl₂, 10 glucose. The flow rate was kept at 1.5 mL/min with a peristaltic pump (Minipuls 3, Gilson), and bath temperature was maintained at 30 °C–32 °C by an inline solution heater and temperature controller (TC-344B, Warner Instruments, Hamden, CT, USA). Hippocampal pyramidal neurons were identified with a ×40 water-immersion objective on an upright microscope equipped with differential interface contrast optics under infrared illumination (BX5IWI, Olympus) and video observation. Current-clamp recordings were made with a MultiClamp 700B amplifier (Molecular Devices). Whole-cell patch-clamp recordings were performed with pipettes (resistance of 3–5 MΩ) filled with internal solution containing (in mM): 145 K-gluconate, 2 MgCl₂, 0.1 EGTA, 2 Na₂ATP, and 10 HEPES (pH 7.2 with KOH; 290 mOsm). Access resistance was monitored throughout the recording and was typically <15 MΩ. Data acquisition and stimulation were performed with a Digidata 1440A Series interface and pClamp 10 software (Molecular Device). Data were filtered at 1 kHz, digitized at 10 kHz, and were analyzed off-line with pClamp 10 software. To measure firing, current pulses were applied using a patch amplifier in current-clamp mode. A series of 5–6 depolarizing current pulses (800 ms duration, 50 pA apart) were applied (D’Ascenzo et al., 2009). Membrane potentials were held at approximately −70 mV during inter-pulse intervals by injecting Direct current with the patch-clamp amplifier. For eliciting single spikes, brief depolarization currents (1 nA; 2 ms) were injected. Spike halfwidth was measured at 50% of the peak. In some experiments membrane input resistance was evaluated in current-clamp mode using series of 600 ms hyperpolarizing steps (from −60 to 0 pA in 20 pA increments). Input resistance was calculated from the peak voltage achieved at each step. The input resistance values were derived from the linear portion of the intensity-to-voltage (I–V) relationship as the slope of the linear regression fitting line calculated by using the Clampfit program (pCLAMP9 software, Molecular Devices).

2.7. Immunoblot analysis

For protein expression and phosphorylation, cells or tissues were lysed in ice-cold lysis buffer (NaCl 150 mM, Tris-HCl 50 mM pH 8, EDTA 2 mM) containing 1% Triton X-100, 0.1% SDS, ×1 protease inhibitor mixture (Sigma), 1 mM sodium orthovanadate, and 1 mM sodium fluoride (Podda et al., 2012). After 15 minutes on ice with occasional vortexing, cells were spun down at 22,000g, 4 °C to remove debris, and supernatant quantified for protein content (DC Protein Assay; Bio-Rad). Equal amounts of cell protein lysates were diluted in 6× Laemmli buffer, boiled and resolved by SDS-PAGE.

Western blotting was performed using primary antibodies 1:1000 diluted overnight and revealed with horseradish peroxidase-conjugated secondary antibodies diluted to 1:2000 (Cell Signaling). The following primary antibodies were used: anti-total Kv4.2 rabbit polyclonal antibody from Abcam, anti-pKv4.2 phosho-Ser-616 rabbit polyclonal antibody from MyBioSource, anti-total GSK-3β rabbit monoclonal antibody from Cell Signaling, anti-GSK-3β phosho-Ser-9 rabbit polyclonal antibody from Immunological Sciences, and anti-tubulin mouse monoclonal antibody from Sigma.

For analysis of Aβ₄₂ peptide protein, preparations were analyzed by Western blotting as previously described (Attar et al., 2012; Ripoli et al., 2013). Briefly, the Aβ samples were mixed with NuPAGE LDS sample buffer 4× (final concentration of 200 nM) and separated on 10%–20% gradient Novex Tricine precast gels (Invitrogen) according to the manufacturer’s protocol. After electrophoresis the proteins were transferred to 0.2-μm nitrocellulose membranes (Amersham Biosciences, Buckinghamshire, UK). Membranes were blocked for 1 hour, at RT, in a suspension of 5% non-fat milk in Tris-buffered saline containing 0.1% Tween-20 before incubation overnight at 4 °C with mAb 6E10 (Signet, Dedham, MA, USA; 1:1000). Membranes were washed 3 times with Tris-buffered saline containing 0.1% Tween-20 and then incubated with horseradish peroxidase-conjugated anti-mouse secondary antibody (Cell Signaling; 1:2000) at RT for 1 hour. Development was done after 5 minutes of incubation with enhanced chemiluminescence reagents (Super-Signal west Femto Pierce) and exposed to Hyperfilm (Amersham Biosciences). Molecular sizes for immunoblot analysis were determined using color marker ultra-low range (Sigma).

2.8. Statistical analysis

Data are expressed as means ± standard errors of the means. Statistical significance was assessed with either Student t test or paired t test and 1-way analysis of variance for multiple groups comparison (with Tukey post hoc test). Statistical analysis was performed with SYSTAT 10.2 software (Statcom, Inc, Richmond, CA). The level of significance was set at 0.05.

3. Results

3.1. Intracellular Aβ₄₂ increases neuronal firing in primary mouse hippocampal neurons

To investigate whether iAβ₄₂ affects intrinsic excitability, whole-cell patch-clamp recordings were performed on primary mouse hippocampal neurons in current-clamp mode. Action potential firing was examined by exposing cells to a series of 800 ms current pulses (1 every 5 minutes), whereas recombinant Aβ₄₂ (200 nM) was intracellularly perfused through the patch pipette. The amplitude of current pulses ranged from −50 pA (hyperpolarizing, sub-threshold) to 200 pA (depolarizing, suprathreshold) in 50 pA increments. After 20 minutes (T₂₀) of Aβ₄₂ perfusion, the number of spikes significantly increased compared with the beginning of the experiment, right after achieving whole-cell configuration (T₀) (Fig. 1A, C, and D; number of APs at T₀ was 8.7 ± 0.6 versus 12.4 ± 0.8 at T₂₀; n = 22; p < 0.001; paired t test). The Aβ₄₂-induced firing enhancement developed quite slowly and peaked approximately 10 minutes after intracellular application of Aβ₄₂ (Fig. 1C, black circles). Similar effects were also found at a more physiological temperature (36 °C; Supplementary Fig. 1). To rule out nonspecific effects of Aβ₄₂ either the vehicle or 200 nM of the reverse protein Aβ_42-1 were injected into cells and firing frequency examined. As shown in Fig. 1B–D, neither vehicle nor Aβ_42-1 significantly affected spike number (vehicle, number of APs at T₀ 8.8 ± 1.0; number of APs at T₂₀ = 9.1 ± 1.1, n = 10, p > 0.05, paired t test; Aβ_42-1, number of APs at T₀ = 8.3 ± 1.6; number of APs at T₂₀ = 8.4 ± 1.3, n = 7, p > 0.05, paired t test). The oligomer distribution profile of recombinant Aβ₄₂ used in the previously mentioned experiments was verified by Western blot analysis, which showed that the vast majority of the injected Aβ₄₂ was in the form of monomers, dimers, trimers, and tetramers (Fig. 1F), that are considered the most synaptotoxic species in AD (Mucke and Selkoe, 2012; Shankar et al., 2008).

Fig. 1.

Intracellular perfusion of Aβ₄₂ increases primary hippocampal neuron excitability. (A) Whole-cell recordings from a hippocampal neuron in current-clamp configuration showing the increase in APs number after 20 minutes of intracellular perfusion of Aβ₄₂ via the patch pipette. (B) Intracellular vehicle perfusion had no effect on neuronal excitability. (C) Time course of the APs number recorded during intracellular perfusion of Aβ₄₂ (n = 22), vehicle (n = 10), and the reverse protein Aβ_42-1 (n = 7). (D) Bar graph summarizing data from experiments shown in C. Multiple comparisons were performed with ANOVA test with Tukey post hoc (F_(3,40) = 5.47; *p < 0.05). Data are expressed as percentage of APs number recorded after 20 minutes of intracellular perfusion of either vehicle or Aβ_s (T₂₀) versus that recorded soon after establishing the whole-cell configuration (T₀). (E) Bar graph showing first spike latency in different experimental conditions (n = 22). (F) Representative Western blot of recombinant Aβ₄₂ (200 nM) incubated for 18 hours at 4 °C used in this study. (G) Representative traces showing the AP width in either an Aβ₄₂-injected (left) or a vehicle-injected neuron (right). Recordings, at T₀ and T₂₀, are superimposed to clarify the spike broadening in Aβ₄₂-injected neurons. (H) Bar graph showing mean values of spike half-width after 20 minutes of intracellular perfusion of Aβ₄₂ (n = 10), vehicle (n = 10), and iAβ_42-1 (n = 7). Error bars indicate SEM, *p < 0.05; ** p < 0.001 in this and in the following figures. Abbreviations: Aβ, amyloid β-protein; ANOVA, analysis of variance; APs, action potentials; iAβ, intracellular Aβ; SEM, standard errors of the means.

A set of dedicated experiments was performed to determine whether iAβ₄₂ affected membrane properties. We found that iAβ₄₂ had no effects on the resting membrane potential (T₀ = −70.0 ± 0.3 mV; T₂₀ = −72.4 ± 1.2 mV; n = 5; p > 0.05) or cell input resistance (T₀ = 265.5 ± 17.8 MΩ; T₂₀ = 272.7 ± 14.9 MΩ; n = 22; p > 0.05; paired t test).

Importantly iAβ₄₂, but not the vehicle and the reverse peptide Aβ_42-1, induced a modest but significant decrease in the first spike latency (Fig. 1E; iAβ₄₂: T₀ = 49.8 ± 13.2 ms; T₂₀ = 31.8 ± 6.7 ms; n = 22; p < 0.05; paired t test; vehicle: T₀ = 48.4 ±10.3 ms; T₂₀ = 46.9 ± 8.7 ms; n = 10; p > 0.05; paired t test; Aβ_42-1: T0 = 49.1 ± 8.8 ms; T₂₀ = 51.5 ± 9.5 ms; n = 7; p > 0.05; paired t test). A concomitant change in spike width was found in an additional set of experiments in which the AP shape was analyzed with a series of single depolarizing steps (1 nA; 2 ms, Fig. 1G). As shown in Fig. 1H, 20 minutes of iAβ₄₂ perfusion, but not vehicle and iAβ_42-1, significantly broadened the AP (iAβ₄₂: T₀ = 1.9 ± 0.2 ms; T₂₀ = 2.5 ± 0.3 ms; n = 10; p < 0.001; paired t test; iAβ_42-1: T₀ = 1.8 ± 0.2 ms; T₂₀ = 1.7 ± 0.1 ms; n = 7; p > 0.05; paired t test; vehicle: T₀ = 2.1 ± 0.2 ms; T₂₀ = 2.2 ± 0.1 ms; n = 7; p > 0.05; paired t test). Broadening of APs was also observed in the first AP in a train of spikes evoked by 800 ms depolarizing pulse (T₀ = 1.7 ± 01 ms; T₂₀ = 2.3 ± 0.2 ms; n = 22; p < 0.001). Reduced first spike latency and broadening of the spike width are consistent with a decrease in A-type K⁺ currents (Kim et al., 2005; McDermott and Schrader, 2011). The A-type K⁺ current is a voltage-dependent, transient, and outward current that has been shown to control action potential repolarization in hippocampal neurons (Carrasquillo et al., 2012; Kim et al., 2005). Taken together, these results suggest that intracellular accumulation of Aβ₄₂ increases intrinsic neuronal excitability, likely through an A-type K⁺ current-mediated mechanism.

We therefore hypothesized that the increased excitability induced by iAβ perfusion could be due, at least in part, to a downregulation of A-type K⁺ current. To test this hypothesis, we evaluated the magnitude of A-type K⁺ currents during intracellular perfusion of Aβ₄₂ or vehicle in voltage-clamp mode. Representative traces of the total outward current recorded in response to voltage steps from −90 to +50 mV are shown in Fig. 2A (upper trace). A prepulse of −30 mV (50 ms) was applied to inactivate A-type K⁺ transient currents and to isolate the sustained current component (Fig. 2A, middle trace; Hoffman and Johnston, 1998). The transient current was subsequently obtained by digital subtraction (total sustained, Fig. 2A, lower trace). When K⁺ currents were recorded in the presence of 5 mM 4-AP, a commonly used A-type K⁺ current blocker, the currents were strongly inhibited thus confirming the A-Type identity of the K⁺ currents (Fig. 2B). As shown in Fig. 2C, intracellular Aβ₄₂ infusion, but not vehicle and Aβ_42-1, significantly decreased A-type K⁺ currents with a time course similar to iAβ₄₂-induced increase in spike number (Fig. 2C–E; Aβ₄₂: T₀ 1916 ± 292 pA; T₂₀ = 1141 ± 196 pA; n = 11; p < 0.001; paired t test; vehicle: T₀ = 1866 ± 256 pA; T₂₀ = 1881 ± 232 pA; n = 10; p > 0.05; Aβ_42-1: T₀ = 1884 ± 297 pA; T₂₀ = 1836 ± 145 pA; n = 6; p > 0.05). These results demonstrate that intracellular accumulation of Aβ₄₂ increases intrinsic excitability in hippocampal neurons and suggest that this effect may be because of reduction of A-type K⁺ current.

Fig. 2.

Intracellular Aβ₄₂ reduces transient A-type K+ currents in primary hippocampal neurons. (A) Isolation of the transient A-type K⁺ current from a hippocampal neuron in the voltage-clamp configuration. The total outward current (upper trace; 1) was recorded during 300 ms voltage steps from −90 to +50 mV. To inactivate the transient K⁺ channel a prepulse to −30 mV (50 ms) was applied before the voltage step to +50 mV (middle trace; 2). The transient K⁺ current (lower trace; 3) was isolated by digital subtraction of the sustained current from the total outward current. (B) K⁺ current recorded in the presence of 5 mM of 4-APs, an A-type K⁺ current blocker (from the same neuron shown in A). (C) Representative traces showing the effect of 20 minutes intracellular application of Aβ₄₂ (left) or vehicle (right). (D) Time course of the transient A-type K⁺ current amplitudes during intracellular perfusion of Aβ₄₂ (black circles; n = 10) and vehicle (open circles; n = 10). (E) Bar graph summarizing data from experiments shown in C (average transient A-type K⁺ current at T₂₀ is expressed as percentage of transient A-type K⁺ current recorded at T0; iAβ42 n = 11; i-vehicle n = 10; iAβ42-1 n = 6; ** p < 0.001). Abbreviations: Aβ, amyloid β-protein; APs, action potentials.

To provide additional evidence for the involvement of A-type K⁺ current in the iAβ₄₂-induced neuronal hyperexcitability, we performed a set of experiments in which we added 60 μM of 4-AP to the bath solution before achievement of the whole-cell configuration. At this concentration, 4-AP reduces the A-type K⁺ current (approximately 30%) to a level that still allows the cell to fire (Mitterdorfer and Bean, 2002; Ye et al., 2003). Under these conditions, iAβ₄₂ had no effect on neuronal excitability (Fig. 1D; number of APs at T₀ = 12.0 ± 1.1; number of APs at T₂₀ = 11.4 ± 0.7; n = 7; p > 0.05; paired t test), thus supporting our hypothesis that Aβ₄₂-induced reduction of A-type K⁺ current leads to increased intrinsic excitability.

3.2. iAβ₄₂-induced increased excitability and A-type K⁺ current reduction require caspase and GSK-3 activation

Our next objective was to identify intracellular pathways that mediate the observed iAβ₄₂-induced hyperexcitability.

Caspases, key molecular proteases involved in apoptosis (Nicholson, 1999) have been implicated in hippocampal synaptic plasticity (Jo et al., 2011; Li and Sheng, 2012), have been associated with AD and can be activated on Aβ exposure (Allen et al., 2001; Cho and Johnson, 2004; Chong et al., 2006; Heneka et al., 2013; Jo et al., 2011; Mattson, 2004).

To determine whether caspases might be involved in the iAβ₄₂-induced increase in intrinsic excitability, we tested whether Z-VAD-FMK, a pan-caspase inhibitor (Li and Sheng, 2012) could prevent iAβ₄₂-induced hyperexcitability. When Aβ₄₂ was intracellularly perfused in hippocampal neurons that were preincubated (20 minutes) with 50 μM of Z-VAD-FMK, no changes in excitability were detected (Fig. 3A and B; number of APs at T₀ = 8.0 ± 1.2; number of APs at T₂₀ = 7.5 ± 0.8; n = 10; p > 0.05; paired t test). Likewise, A-type K⁺ currents and AP width were unaffected by the Aβ₄₂ treatment (for A-type K⁺ currents Fig. 3C–E, T₀ = 1932 ± 164 pA; T₂₀ = 1783 ± 174 pA; n = 8; p > 0.05; for AP width, Fig. 3F–G; halfwidth at T₀ = 1.94 ± 0.08; halfwidth at T₂₀ = 2.01 ± 0.05; n = 8; p > 0.05; paired t test). Of note, exposure of neurons, treated with vehicle, to Z-VAD-FMK had no effects on spike number (Fig. 3B; number of APs at T₀ = 9.3 ± 0.9; number of APs at T2₀ = 8.7 ± 0.7; n = 10; p > 0.05; paired t test).

Fig. 3.

Preincubation of the neurons with the pan-caspase inhibitor Z-VAD-FMK prevents iAβ₄₂-induced increase in membrane excitability. (A) Representative traces showing no effect of iAβ₄₂ on membrane excitability when neurons were preincubated (20 minutes) with the pan-caspase inhibitor Z-VAD-FMK (50 μM). (B) Bar graph summarizing data from experiments shown in A (ANOVA test with Tukey post hoc; F_(3,42) = 5.50; **p < 0.001). Note that Z-VAD-FMK itself had no effect on membrane excitability (n = 10). For comparison the effect of vehicle and iAβ₄₂ are shown (same data of Fig. 1). (C) Representative traces showing that iAβ₄₂ had no effect on the transient A-type K⁺ current when neurons were preincubated with Z-VAD-FMK. (D) Bar graph showing the mean current amplitudes in the 2 different conditions. (E) Bar graph summarizing data for the transient A-type K⁺ current experiments (ANOVA test with Tukey post hoc; F_(3,33) = 12.8; **p < 0.001). (F) Preincubation of neurons with Z-VAD-FMK also prevented the iAβ₄₂-induced increase in spike broadening (n = 8). (G) Bar graph summarizing data from experiments shown in F (ANOVA test with Tukey post hoc; F_(2,24) = 14.5; **p < 0.001). Abbreviations: ANOVA, analysis of variance; iAβ, intracellular Aβ.

The results demonstrate that inhibition of caspases occludes the effect of iAβ₄₂ on excitability and strongly suggest that caspases are a downstream target of iAβ₄₂.

Another protein that has been widely implicated in AD is GSK-3β (Medina and Avila, 2013; Phiel et al., 2003; Takashima, 2006). GSK-3β activity is normally suppressed by its N-terminal phosphorylation at Ser-9 induced by protein kinase type 1 (Akt-1) (Emamian et al., 2004; Jo et al., 2011). Caspase-3 activation cleaves Akt1 resulting in derepression and increased activity of GSK-3β (Deng et al., 2014; Jo et al., 2011). Upregulation of GSK-3β interferes with synaptic plasticity mechanisms (Bradley et al., 2012) and voltage-gated Na⁺, Ca²⁺, and K⁺ channel function (Wildburger and Laezza. 2012).

We, therefore, speculated that Aβ₄₂-induced neuronal hyperexcitability could be mediated by GSK-3 activation, as a downstream of caspase-3. In support of this hypothesis, we found that pre-incubation (20 minutes) of the hippocampal neurons with the specific GSK-3 inhibitor CT-99021 (1 μM) prevented iAβ₄₂-induced spike discharge increase (Fig. 4A and B; APs at T₀ = 6.4 ± 1.9; APs at T₂₀ = 7.4 ± 2.6; n = 7; p > 0.05; paired t test), AP width increase (Fig. 4F and G; halfwidth at T₀ = 1.97 ± 0.24; halfwidth at T₂₀ = 2.16 ± 0.28; n = 11; p > 0.05; paired t test) and A-type K⁺ current reduction (Fig. 4C–E; T₀ = 1900 ± 379 pA; T₂₀ = 1719 ± 344 pA; n = 7; p > 0.05). These results strongly suggest that the iAβ₄₂-induced increase in intrinsic excitability requires the activity of the GSK-3 pathway.

Fig. 4.

Pharmacologic inhibition of GSK-3 prevents iAβ₄₂-induced neuronal hyperexcitability. (A) Representative traces showing that when neurons were preincubated with the GSK-3 inhibitor CT-99021 (1 μM; 20 minutes) intraneuronal application of Aβ₄₂ had no effect on spike firing. (B) Bar graph showing percentage of APs number recorded after 20 minutes intracellular perfusion of Aβ₄₂ in the presence of CT-99021 (ANOVA test with Tukey post hoc; F_(3,37) = 4.4; *p < 0.05). For comparison, the effect of vehicle and iAβ₄₂ are shown (same data of Fig. 1). (C) A-type K⁺ currents were unaffected by iAβ₄₂ when neurons were preincubated with CT-99021. (D) Bar graph showing the mean current amplitudes in the 2 different conditions (n = 7). (E) Bar graph summarizing data for transient A-type K⁺ current experiments (ANOVA test with Tukey post hoc; F_(3,32) = 14.2; **p < 0.001). (F) Preincubation of neurons with CT-99021 prevented the iAβ₄₂-induced increase in the spike broadening (n = 11). (G) Bar graph summarizing data from experiments shown in F (ANOVA test with Tukey post hoc; F_(2,29) = 11.5; **p < 0.001). Abbreviations: Aβ, amyloid β-protein; ANOVA, analysis of variance; APs, action potentials; iAβ, intracellular Aβ.

3.3. Intracellular Aβ₄₂ accumulation is required for the extracellular Aβ-induced increase in intrinsic excitability

We next asked whether eAβ could mimic the effects of iAβ₄₂ and act through similar mechanisms. As shown in Fig. 5, eAβ₄₂ oligomers (200 nM) significantly increased the number of spikes evoked in response to current steps (Fig. 5A–C; APs at T₀ = 8.5 ± 0.9; APs at T₂₀ = 12.7 ± 1.5; n = 7; p < 0.05; paired t test). The magnitude of the phenotype was comparable with iAβ₄₂ although the time course of the effect was slightly slower (Fig. 5B, solid circles). Similarly to iAβ₄₂, eAβ₄₂ affected spike kinetics (halfwidth: T₀ = 1.8 ± 0.1; T₂₀ = 2.1 ± 0.1; n = 7; p < 0.05; first spike latency: T₀ = 48.9 ± 12.0; T₂₀ = 27.8 ± 11.2; n = 7; p < 0.05) and reduced the amplitude of A-type K⁺ currents (Supplementary Fig. 1; T₀ = 1988 ± 375 pA; T₂₀ = 1232 ± 295 pA; n = 6; p < 0.05; paired t test), a phenotype that was prevented by preincubation of neurons with either Z-VAD-FMK (Fig. 5C; APs at T₀ = 9.6 ± 0.8; APs at T₂₀ = 8.5 ±1.1; n = 8; p > 0.05; paired t test) or CT-99021 (Fig. 5C; APs T₀ = 9.2 ± 1.8; APs T₂₀ = 8.5 ± 2.4; n = 7; p > 0.05; paired t test). These results suggest that eAβ₄₂ and iAβ₄₂ share common mechanisms of action and signal transduction.

Fig. 5.

Extracellular applied Aβ₄₂ increases neuronal excitability to the same extent of iAβ₄₂. (A) Representative traces showing the increase in spike numbers after 20 minutes of extracellular application of Aβ₄₂ (200 nM). (B) Time course of increased excitability during extracellular application of Aβ₄₂ (n = 7). For comparison, the time course of iAβ₄₂ is also shown (same data of Fig. 1). (C) Bar graph summarizing the effect of eAβ₄₂ on spike discharge. Note that preincubation of neurons with either Z-VAD-FMK (1 μM) or CT-99021 (1 μM) prevented the eAβ₄₂-induced increase in excitability (ANOVA test with Tukey post hoc; F_(2,20) = 17.2; **p < 0.001). (D) Representative Western blots of hippocampal proteins showing decreased immunoreactivity for phosphorylated GSK-3 at Ser-9 following 20–40 minutes of Aβ₄₂ treatment. (E) Densitometry for the blots probed with Ser-9 phosphorylated GSK-3 normalized to total GSK-3 is shown (n = 3; *p < 0.05; statistics by Mann-Whitney test). Abbreviations: Aβ, amyloid β-protein; ANOVA, analysis of variance; eAβ, extracellular Aβ; iAβ, intracellular Aβ.

To gain further mechanistic insights into the eAβ₄₂ activity, total cell lysates obtained from hippocampal neurons treated with eAβ₄₂ were processed for Western blot analysis and probed for a GSK-3β Ser-9 antibody, commonly used to estimate the pool of active GSK-3 (Fig. 5D). Band quantification revealed a time-dependent decrease of active GSK-3 induced by eAβ₄₂, whereas the total GSK-3 pool remained constant (Fig. 5E). Furthermore, preincubation of hippocampal neurons with the caspase inhibitor ZVAD-FMK (50 μM; 20 minutes) prevented the eAβ₄₂-induced decrease in GSK-3 phosphorylation, suggesting that GSK-3 is a downstream target of Aβ₄₂ via caspases 3 activation (Supplementary Fig. 3). Thus, by decreasing the level of tonic GSK-3 inhibition eAβ₄₂ might markedly increase the GSK-3 active pool with consequences for neuronal excitability (Kapfhamer et al., 2010; Wildburger and Laezza, 2012).

Overall, these results demonstrate that exposure of neurons to either eAβ₄₂ or iAβ₄₂ (by intracellular perfusion) leads to a common pathophysiological outcome on excitability raising the question of whether eAβ₄₂ requires internalization for its toxic activity. Two different approaches were used to answer this question.

First, we sought to determine whether extracellular perfusion of a cell membrane impermeable form of Aβ₄₂ harboring the methionine 35 sulfoxide chemical modification ($\mathrm{A}{\beta }_{42}^{\mathrm{MO}}$) (Ripoli et al., 2013), retained Aβ₄₂ activity. As shown in Fig. 6A, $\mathrm{A}{\beta }_{42}^{\mathrm{MO}}$ had no effect on neuronal excitability if bath applied (APs at T₀ = 9.4 ± 0.8; APs at T₂₀ = 9.5 ± 1.3; n = 7; p > 0.05; paired t test) but was significantly active if injected intracellularly (Fig. 6A; APs T₀ = 9.0 ± 1.3; APs T₂₀ = 12.7 ±1.8; n = 7; p < 0.05; paired t test). These results indicate that the lack of effect of bath applied $\mathrm{A}{\beta }_{42}^{\mathrm{MO}}$ can be attributed to impaired internalization rather than disrupted intracellular activity. Furthermore, to provide evidence for lack of $\mathrm{A}{\beta }_{42}^{\mathrm{MO}}$ internalization, we conjugated Aβ42 wild type ($\mathrm{A}{\beta }_{42}^{\mathrm{WT}}$) and $\mathrm{A}{\beta }_{42}^{\mathrm{MO}}$ with the fluorescent dye IRIS-5-NHS and examined the cellular distribution of the 2 Aβ₄₂ variants with confocal microscopy. Results of this set of experiment showed that more than 90% of neurons treated for 30 minutes with IRIS-5-labeled $\mathrm{A}{\beta }_{42}^{\mathrm{WT}}$ exhibited intracellular accumulation of the peptide (mean number of fluorescent spots for single neuron was 13 ± 3), although no internalization was observed on treatment with IRIS-5-labeled $\mathrm{A}{\beta }_{42}^{\mathrm{MO}}$ analog (Fig. 6C). Taken together these results suggest that the Aβ₄₂-induced increase of intrinsic excitability likely requires its intracellular accumulation.

Fig. 6.

Extracellular Aβ₄₂-induced hyperexcitability requires neuronal internalization. (A) Summary graphs showing that $\mathrm{A}{\beta }_{42}^{\mathrm{MO}}$ increases neuronal excitability only if applied intracellularly (ANOVA test with Tukey post hoc; F_(2,20) = 9.8; **p < 0.001). (B) Bar graph showing that intracellular perfusion of 6E10 by patch pipette significantly reduced the effect of extracellular Aβ₄₂ perfusion on neuronal excitability (n = 10; * p < 0.05). (C) Representative examples of neurons following 20 minutes extracellular application of 200 nM IRIS-labeled Aβ₄₂ (green; left) and $\mathrm{A}{\beta }_{42}^{\mathrm{MO}}$ (green; right). Red staining indicates MAP2 immunoreactivity. Scale bar 10 μm. Bottom boxes represent XZ cross-sections from confocal Z-stacks showing the different neuronal accumulation of Aβ₄₂ analogues after 20-minute treatments. Abbreviations: Aβ, amyloid β-protein; ANOVA, analysis of variance; MAP2, microtubule-associated protein 2. (For interpretation of the references to color in this Figure, the reader is referred to the Web version of this article.)

Second, we conducted a set of experiments aimed at preventing the activity of iAβ₄₂ with the 6E10 antibody that recognizes human Aβ₄₂, but has no reported effects on endogenous Aβ₄₂ or APP fragments (Tampellini et al., 2007). As shown in Fig. 6B, intracellular perfusion of the 6E10 antibodyalone (1:300) had no effect on intrinsic excitability (APs at T₀=8.8±1.3; APs at T₂₀=8.4±1.3; n=7; p>0.05; paired t test), but significantly lowered the increase in firing induced by eAβ₄₂ (+16.9%±7.9% n=10 compared with +47.0%±9.1%; n=7; p < 0.005; unpaired t test). Thus, the increase in intrinsic excitability induced by eAβ42 requires its intracellular action.

3.4. Aβ₄₂ increases α-subunit Kv4.2 phosphorylation

Our data indicate that intracellular accumulation of Aβ₄₂ increases intrinsic excitability in hippocampal neurons through inhibition of A-type K⁺ currents via caspases and GSK-3 activation. The molecular determinants of 4-AP-sensitive A-type K⁺ currents vary depending on brain regions. The Kv4.2 subunit is prominently expressed in CA1 hippocampal neurons, where it has been proposed to mediate sub-threshold and fast inactivating A-type K⁺ currents (Coetzee et al., 1999; Kim and Hoffman, 2012; Lin et al., 2011; Varga et al., 2000). Previous studies showed that Kv4.2 can be phosphorylated by extracellular signal-regulated kinases at Ser-616 resulting in suppression of the channel activity (Hu et al., 2006), indicating that phosphorylation of Kv4.2 can lead to functional outcomes on K⁺ currents.

Inspection of the primary sequence of Kv4.2 indicates that Ser-616 lies in a putative GSK-3 phosphorylation motif (S/TXXXS/T;Fig. 7A), where the target site is flanked by a proline residue (P617) and is 3 residues upstream of another S residue (S-620) that usually acts as a phosphorylation priming site (Wildburger and Laezza, 2012). Thus, we hypothesized that Aβ₄₂ treatment could result in a GSK-3-dependent phosphorylation of the Kv4.2 subunit at Ser-616.

Fig. 7.

Aβ₄₂ accumulation leads to GSK-3-dependent phosphorylation of Kv4.2 channel. (A) Primary sequence of the C-terminal domain of Kv4.2 channels around the putative GSK-3 phosphorylation site Ser-616. (B) Representative blots of total cell lysate from primary hippocampal neurons probed with either anti-phospho-Ser-616 (upper panel) or total Kv4.2 (lower panel) antibody following 20 minutes exposure of 200 nM of Aβ₄₂ in the presence or absence of CT-99021 (1 μM; 30 minutes). (C) Densitometry analysis for the blots probed with the anti-phospho-Ser-616 and normalized to total Kv 4.2 channel is shown (n = 3; **p < 0.001; statistics by Mann-Whitney test). Abbreviation: Aβ, amyloid β-protein.

To test this hypothesis, we preincubated hippocampal neurons with Aβ₄₂ (200 nM; 20 minutes) in the presence or absence of CT-99021 (1 μM; 30 minutes) and used Western blot analysis of total cell lysates to examine the level of phosphorylation of Ser-616. As illustrated in Fig. 7B and C, the Aβ₄₂ treatment led to a significant increase (p < 0.05; n = 3) in Ser-616 immunoreactivity that was prevented by CT-99021 (p > 0.05; n = 3). These results suggest that phosphorylation of Kv4.2 by GSK-3 might be a downstream target of Aβ₄₂-mediated increase in intrinsic excitability in hippocampal neurons.

3.5. Intracellular Aβ accumulation in an AD mouse model recapitulates in cell studies

The 3xTg-AD mouse model harbors the mutant genes for APPSwe, presenilin 1PS1M146V and tauP301L (Oddo et al., 2003) and exhibits temporal- and region-specific Aβ and tau pathologies that closely resemble the human AD pathology. At early AD stages (3–5 months), 3xTg-AD mice show intracellular accumulation of Aβ that precedes extracellular plaques (Oddo et al., 2003). As such, 3xTg-AD mice are an ideal model to examine early toxic effects of intracellular Aβ₄₂. To determine whether intracellular Aβ accumulation in 3xTg-AD mice recapitulated ours in cell studies, we performed whole-cell patch-clamp recordings of CA1 pyramidal neurons in hippocampal slices obtained from 5-month-old 3xTg-AD and non-Tg mice. Visually identified CA1 pyramidal neurons were exposed to a series of 800 ms current pulses ranging from −50 pA to 200 pA in 50 pA increments. To control for possible effects of membrane depolarization on neuronal firing, the membrane potential was kept at approximately −70 mV by injecting hyperpolarizing current during step intervals. Under these experimental conditions, we observed a significant increase in firing in 3xTg-AD neurons compared with control in almost all ranges of stimulation (Fig. 8A and B). The average AP number at 150 pA current injection was significantly higher in 3xTg-AD mice (16.4 ± 2.1, n = 10; p < 0.05 unpaired t test) compared with non-Tg (9.1 ± 1.9; n = 10). We also found a significant increase in the AP halfwidth in 3xTg-AD neurons (1.15 ± 0.09, n = 10; p < 0.05, unpaired t test) compared with non-Tg neurons (0.83 ± 0.02, n = 10). To determine whether these phenotypes were GSK-3-dependent, we performed a set of experiments in which hippocampal slices from 3xTg-AD mice were preincubated with CT-99021 (30±40 minutes; 1 μM). Under these experimental conditions, the averages of AP number and AP halfwidth were similar to those observed in slices from non-Tg mice (Fig. 8B and D; AP number = 10.9 ± 1.3 AP halfwidth = 0.85 ± 0.04; n = 10; p > 0.05). Of note, the increased excitability found in CA1 pyramidal neurons of 3x-Tg-AD mice was not associated with changes in input resistance (3xTg-AD: 189.1 ±13.6MΩ, n = 10; non-Tg: 165.2 ± 8.8 MΩ, n = 10; p > 0.05; unpaired t test) and membrane resting potential (3xTg-AD: −65.5 ±1.62 mV, n = 10; non-Tg: 60.2 ± 2.9 mV, n = 10, p > 0.05, unpaired t test). Furthermore, the same changes in intrinsic excitability induced by intracellular Aβ₄₂ accumulation were observed in control non-Tg mice following Aβ₄₂ intracellular injection through the patch pipette (Supplementary Fig. 4).

Fig. 8.

CA1 hippocampal pyramidal neurons from 3xTg-AD mouse show increased intrinsic excitability and Kv4.2 channel phosphorylation. (A) Representative current-clamp traces from CA1 hippocampal pyramidal neurons of either 3xTg-AD or non-Tg mice showing AP trains in response to current injection of 150 pA. (B) Input-output curves of APs number versus injected current intensity recorded in slices from 3xTg-AD and non-Tg mice. Note that in 3xTg-AD slices preincubated with CT-99021, the APs number was similar to what was observed in non-Tg mice (p < 0.05, unpaired t test; n = 10). (C) Representative AP traces to illustrate AP width in response to a single step current injection (1 nA; 1 ms) in either 3xTg-AD or non-Tg neurons. (D) Bar graph summarizing the mean AP half width in the different experimental groups (n = 10). (E) Representative Western blots of total cell lysate showing increased immunoreactivity for Ser-616 phosphorylation of Kv4.2 in the 3xTg-AD mice compared with non-Tg mice. Note that the total Kv4.2 pool was unchanged in the 2 experimental groups. Abbreviations: 3xTg-AD, triple-transgenic mouse model of Alzheimer’s disease; AP, action potential; non Tg, non transgenic.

Notably, Western blot analysis of hippocampal total cell lysates revealed that the levels of Ser-616 phosphorylation of the Kv4.2 subunit were dramatically higher in 3xTg-AD mice compared with control (n = 8; p < 0.001), whereas no changes in the total pool of Kv4.2 were observed (Fig. 8E). Altogether, these results confirm our primary culture studies and indicate that increased intrinsic excitability in CA1 hippocampal neurons is an early functional outcome of AD pathology induced by intracellular Aβ.

4. Discussion

The present study revealed that intracellularly injected Aβ₄₂ downregulates A-type K⁺ currents and increases spike width of hippocampal pyramidal neurons, thereby increasing excitability. These effects require caspases and GSK-3 activation and correlate with an increase in phosphorylation of a molecular determinant of A-type K⁺ currents, the Kv4.2 subunit, at Ser-616, a predicted GSK-3 site. At an early AD stage (5 months old), the 3xTg-AD mouse model exhibits intracellular accumulation of Aβ that precedes the appearance of extracellular plaques. In agreement with our findings in cell studies, we found that excitability of CA1 hippocampal neurons is increased in 3xTg-AD mice compared with controls along with phosphorylation of the Kv4.2 subunit at Ser-616. Converging evidence highlights a critical role of iAβ in AD pathology (Gouras et al., 2012; Ripoli et al., 2014; Takahashi et al., 2014; Zepa et al., 2011) and in its symptoms (Steinerman et al., 2008). Our findings provide new important insights into the mechanism of action of iAβ in AD and argue for a toxic effect of iAβ in early stages of the disease.

Previous studies have shown that extracellular Aβ can affect A-type K⁺ currents in hippocampal pyramidal neurons (Chen, 2005; Good and Murphy, 1996). However, the mechanism underlying these changes is poorly understood and whether these effects on A-type K⁺ currents require Aβ internalization is not known.

In this study, we show that extracellular Aβ application increases intrinsic excitability of hippocampal neurons and downregulates A-type K⁺ currents to the same extent of intracellular Aβ. Importantly, we show that this effect required intracellular Aβ buildup. Our electrophysiological data demonstrate a lack of effect of bath applied $\mathrm{A}{\beta }_{42}^{\mathrm{MO}}$, an Aβ variant that exhibited reduced internalization in hippocampal neurons (Fig. 6; Ripoli et al., 2013, 2014). Importantly, $\mathrm{eA}{\beta }_{42}^{\mathrm{MO}}$ does not activate caspase-3, preventing the sequel of caspase-induced effects on apoptosis (Clementi et al., 2006) and lacks toxic effects on synaptic function (Ripoli et al., 2013). Our immunofluorescence studies demonstrated the inability of $\mathrm{A}{\beta }_{42}^{\mathrm{MO}}$ to accumulate intracellularly and showed that $\mathrm{A}{\beta }_{42}^{\mathrm{MO}}$ perfused intracellularly retained the activity of the $\mathrm{iA}{\beta }_{42}^{\mathrm{WT}}$ (Fig. 5). Furthermore, chelation of eAβ₄₂, achieved by patch-pipette injection of the 6E10 antibody, reduced the effect on excitability. The monoclonal antibody 6E10 recognizes the N-terminus of human Aβ₄₂, and efficiently immunoprecipitates oligomeric and monomeric Aβ₄₂ present in conditioned medium (Tampellini et al., 2007). In our experimental conditions, 6E10 antibody did not completely abolished the eAβ₄₂,-induced hyperexcitability (Fig. 6C). This may reflect the inability of 6E10 antibody to neutralize all internalized Aβ₄₂ in agreement with previous studies (Klyubin et al., 2005; Tampellini et al., 2007).

Extracellular Aβ can bind to several cell surface receptors including LPR, RAGE, FPRL1, NMDA, and α-7AChRs (LaFerla et al., 2007; Penke et al., 2012), and the resulting receptor-Aβ complex can be then internalized into early endosomes (LaFerla et al., 2007). Furthermore, the soluble Aβ₄₂ complex associated with apolipoprotein E promotes Aβ internalization (Gylys et al., 2003; Kim et al., 2009; Kuszczyk et al., 2013). Thus, even at later stages of the disease when the predominant form of Aβ is extracellular, the intracellular pool of Aβ might be replenished and kept at a constant level by eAβ₄₂ reuptake. Based on these observations, it is tempting to speculate that in the human AD pathology Aβ₄₂ may exert complex effects on neuronal excitability through a mixed pool of extracellular and intracellular Aβ₄₂.

Intracellular Aβ₄₂ has been shown to suppress BK channels in pyramidal neurons located in the layer II/III of the rat visual cortex (Yamamoto et al., 2011). In hippocampal neurons, however, Aβ₄₂ reduces A-type K⁺ currents (Chen, 2005; Good and Murphy, 1996, Fig. 2), suggesting brain area specific effects of Aβ₄₂ on excitability. Our observation that 4-AP occludes the effect of iAβ₄₂ on A-type K⁺ currents (Fig. 2) rules out a significant role of BK channels and supports the notion that Aβ₄₂-induced A-type K⁺ current hypofunction represents the primary mechanism underlying increased neuronal excitability in the hippocampus in the Aβ pathology. Other mechanisms, however, should also be considered. Noteworthy are the results from a recent study demonstrating a role of α-7n-AChR in neuronal hyperexcitability following long-term exposure (7 days) to fibrillar Aβ₄₂ (Liu et al., 2013). The precise molecular mechanisms of these effects, however, are still unknown.

We show that Aβ₄₂-induced increase in excitability depends on caspases and GSK-3 activation. Caspases are a class of proteases instrumental for many cellular functions including cell differentiation, remodeling, and death. Recently, the role caspase-3, specifically, has been identified in nonapoptotic processes in the central nervous system, such as synaptic plasticity, spine structure, and memory formation (D’Amelio and Rossini, 2012; Snigdha et al., 2012).

Emerging evidence associates caspases with AD. Increased immunoreactivity for caspase-3 has been observed in postsynaptic compartments of postmortem brains from early AD patients (Louneva et al., 2008; Sultana et al., 2010). Caspase-3 can be activated by Aβ₄₂ exposure (Allen et al., 2001) and can exacerbate Aβ pathology (for review see Rohn, 2010). In a seminal work from Jo et al., 2011 the pan anti-caspases inhibitor, Z-VAD-FMK, was shown to prevent inhibition of LTP induced by Aβ₄₂ in CA1 hippocampal neurons, a finding supported by the lack of the Aβ₄₂ effect on CA1 LTP in caspase-3 knockout mice. A conclusion from these studies was that caspase-3 mediates Aβ inhibition of LTP by inducing cleavage of Akt1, a negative suppressor of GSK-3, and subsequent increase in the pool of active GSK-3β, detected as a decrease in the levels of p-GSK-3 (Ser-9). Pharmacologic inhibition of GSK-3 prevented the Aβ-induced inhibition of LTP, suggesting an Aβ-induced mechanism in which activation of casapse-3 leads to cleavage of Akt1 and loss of tonic inhibition of GSK-3 (Jo et al., 2011). Our findings showing that pharmacologic inhibition of either caspases (by Z-VAD-FMK) or GSK-3 (by CT-99021) prevented Aβ₄₂-induced hyperexcitability (Figs. 3–4) are consistent with the model described by Jo et al. (2011) and further emphasize the role of caspases signaling in nonapoptotic pathways in AD.

GSK-3 is a multifunctional kinase that modulates many fundamental cell processes including cell proliferation and differentiation (Frame and Cohen, 2001), cell survival (Takashima et al., 1993), cell motility (Lucas et al., 1998), and memory formation (Bradley et al., 2012). Hyperphosphorylation of tau and Aβ production in AD have been linked to GSK-3 dysfunction (DaRocha-Souto et al., 2012; Noble et al., 2005; Phiel et al., 2003; Pooler et al., 2012; White et al., 2006; Kimura et al., 2013). In double transgenic APP/tau mice, GSK-3 inhibition ameliorates plaque-related neuritic changes, suggesting a role of this kinase in Aβ-induced pathology (DaRocha-Souto et al., 2012). However, the molecular mechanisms linking Aβ oligomers toxicity to GSK-3 are still poorly understood. Our Western blot experiments demonstrating that Aβ₄₂ treatment markedly decreased the level of Ser-9 phosphorylated (inactive) GSK-3 (Fig. 5) are consistent with the emerging concept of a pathophysiological link between AD and altered GSK-3 signaling (Deng et al., 2014). Importantly, we discovered that treatment of hippocampal neurons with Aβ₄₂ results in GSK-3-dependent Kv4.2 phosphorylation at Ser-616 (Fig. 5). The Kv channel family includes the most heterogenous and abundant group of ion channels in excitable cells, comprising more than 40 subunit genes divided in separate families based on structural and functional properties (Gutman et al., 2003). So far, the only Kv channel subtype that has been validated as GSK-3 substrate is Kv7.2, which is encoded by the gene KCNQ2 (Jentsch, 2000). Here, we provide novel evidence for another Kv channel, Kv4.2, as a putative GSK-3 target.

A number of studies have demonstrated complex phosphorylation-dependent regulation of Kv4.2 with functional consequences on transient A-type K⁺ currents. In vitro phosphorylation of recombinant fragments of Kv4.2 revealed PKA- (Anderson et al., 2000), ERK- (Adams et al., 2000), and CaMKII- (Varga et al., 2004) mediated phosphorylation of Kv4.2. It is noteworthy that ERK-mediated phosphorylation of Kv4.2 leads to decreased Kv4.2 and that ERK directly phosphorylates Kv4.2 at T602, T607, and Ser-616 (Schrader et al., 2006). The consensus sequence for GSK-3 targets typically consists of S/T-X-X-X-S/T in which the first S/T residue is the GSK-3 phosphorylation site flanked by a proline residue and followed by an S/T site (3 residues downstream), acting as a phosphorylation priming site (Jope and Johnson, 2004; Wildburger and Laezza, 2012). Ser-616 flanked by P617 and followed by S620 matches with a GSK-3 consensus motif (Fig. 7A). In support, we show that Aβ₄₂ treatment leads to increase Ser-616 phosphorylation of Kv4.2, an effect that is strongly attenuated by pharmacologic inhibition of GSK-3 (Fig. 7B) and is reproduced in 5-month-old 3xTg-AD brains (Fig. 8E). Altogether, these results support the notion that Aβ₄₂ induces a decrease of A-type K⁺ currents and an increase in excitability through GSK-3-dependent phosphorylation of Kv4.2 at Ser-616.

The 5-month-old 3xTg-AD mouse model used in this study recapitulates early AD pathology, where endogenously produced Aβ accumulates inside neurons before the appearance of extracellular plaques (LaFerla et al., 2007; Oddo et al., 2003). Therefore, this model represents an elective tool for investigating the effects of intraneuronal Aβ accumulation. In support of a key role of iAβ in triggering early AD pathology, we found that hyperphosphorylation of Kv4.2 channels at Ser-616 was paralleled by increased firing and broadening of the AP of CA1 pyramidal neurons in 3xTg-AD mice (Fig. 8) similarly to what was observed in primary hippocampal neurons on iAβ exposure.

In animal models that overexpress Aβ, spontaneous network hyperexcitability has been documented (Born et al., 2014; Davis et al., 2014; Marcantoni et al., 2014; Minkeviciene et al., 2009; Palop et al., 2007). Data from the epilepsy literature inform us that similar hyperexcitability can arise from synaptic deficits, usually from unbalance between excitatory and inhibitory transmission (Bateup et al., 2013) or changes in intrinsic excitability (McNamara et al., 2006). However, the deficits in synaptic function and plasticity described in Aβ overproducing mice (Randall et al., 2010) are unlikely the primary cause of hyperexcitability (Davis et al., 2014). Changes in neuronal membrane properties and intrinsic hyperexcitability in AD mouse models have been reported. Minkeviciene et al. (2009) identified an approximately 10 mV depolarizing shift in the membrane resting potential in layer II/III cortical pyramidal neurons of mice carrying human APPsve and PS1dE9 genes (APdE9 mice), whereas Brown et al., (2011) found enhanced excitability in CA1 pyramidal neurons of Aβ overproducing PSAPP double transgenic mice at 8 months of age. This effect was paralleled by an enhanced after-depolarizing potential and changes in the AP waveform, including reduced half-width. In a different model of Aβ overproducing mouse (CRDN8 mice), Wykes et al. (2012) found spikes in CA1 pyramidal neurons to be narrower, a finding in line with previous studies (Brown et al., 2011). On the contrary, in our experimental conditions, we found that the AP halfwidth was increased in 3xTg mice (Fig. 8). The reasons for this apparent discrepancy remains unclear, however, it is tempting to speculate that this may result from different age-dependent accumulation of intracellular versus extracellular Aβ₄₂ in various Aβ overproducing mouse models.

Our finding showing increased excitability of CA1 hippocampal pyramidal neurons in 3xTg-AD mice adds new insights into the mechanisms underlying Aβ₄₂-dependent network hyperexcitability further supporting a link between increased risk of seizures and AD. To this regard, it should be considered that although seizures are increased in the AD patients, overall epilepsy is not a common feature in AD (Amatniek et al., 2006; Palop and Mucke, 2009). Results from imaging scans, such as FDG-PET and/or SPECT brain scans, on AD patients, indicate reduced glucose utilization or blood flow in many brain regions, including hippocampus, supporting reduced resting brain activity (Ishii, 2013). According to these data, the expected increase in energy demand induced by neuronal hyperexcitability may be masked by the overall decrease in energy requirements caused by synaptic failures and neuronal loss.

Collectively, our results provide novel evidence for the role of intracellular Aβ accumulation in hyperexcitability observed in early AD and identify the Kv4.2 channel as a new endpoint of a caspase-dependent GSK-3 pathway and a potential therapeutic target for early AD treatment and prevention.