|Home | About | Journals | Submit | Contact Us | Français|
In Alzheimer disease (AD), the perturbation of the endoplasmic reticulum (ER) calcium (Ca2+) homeostasis has been linked to presenilins (PS), the catalytic core in γ-secretase complexes cleaving the amyloid precursor protein (APP) thereby generating amyloid-β (Aβ) peptides. Here we investigate whether APP contributes to ER Ca2+ homeostasis and whether ER Ca2+ could in turn influence Aβ production. We show that overexpression of wild-type human APP (APP695), or APP harboring the Swedish double mutation (APPswe) triggers increased Ryanodine receptors (RyR) expression and enhances RyR-mediated ER Ca2+ release in SH-SY5Y neuroblastoma cells and in APPswe-expressing (Tg2576) mice. Interestingly, dantrolene-induced lowering of RyR-mediated Ca2+ release leads to the reduction of both intracellular and extracellular Aβ load in neuroblastoma cells as well as in primary cultured neurons derived from Tg2576 mice. This Aβ reduction can be accounted for by decreased Thr-668-dependent APP phosphorylation and β- and γ-secretases activities. Importantly, dantrolene diminishes Aβ load, reduces Aβ-related histological lesions and slows down learning and memory deficits in Tg2576 mice. Overall, our data document a key role of RyR in Aβ production and learning and memory performances, and delineate RyR-mediated control of Ca2+ homeostasis as a physiological paradigm that could be targeted for innovative therapeutic approaches.
Alzheimer’s disease (AD) is the most common neurodegenerative disorder leading to dementia. Extracellular senile plaques, intracellular neurofibrillary tangles, and neuronal loss represent the main histological hallmarks of AD. Amyloid-β peptides (Aβ), the main components of senile plaques, result from the sequential endoproteolytic cleavage of amyloid precursor protein (APP) by β-secretase (BACE-1) and presenilin (PS)-dependent γ-secretase complex (Checler, 1995). Increased level of Aβ is considered a key event contributing to AD etiology. As a support of the amyloid cascade hypothesis, most of the mutations in APP and PS-1/2 responsible for early-onset familial AD (FAD) modulate Aβ production (Bekris et al., 2010).
Calcium (Ca2+) is one of the most important and versatile second messengers in cell signaling. In the nervous system, Ca2+ ions play crucial roles in neurotransmitters synthesis and release, signal transmission, dendrite growth, spine formation, regulation of gene expression, as well as in synaptic plasticity (Berridge et al., 2003). The ability of neurons to regulate the influx, efflux and subcellular compartmentalization of Ca2+ appears compromised in AD (Bezprozvanny and Mattson, 2008). Importantly, one of the main changes observed in AD is a rise in the amount of Ca2+ being released from the endoplasmic reticulum (ER) stores. Aβ enhances Ca2+ release from the ER through both the inositol 1,4,5-triphosphate Receptor (IP3R) and the Ryanodine Receptors (RyR) (Ferreiro et al., 2004). FAD-linked PS1 and PS2 mutations trigger abnormal ER Ca2+ homeostasis by potentiating IP3-and RyR-evoked Ca2+ liberation, and decreasing ER Ca2+ uptake (Leissring et al., 1999; Stutzmann et al., 2004; Stutzmann et al., 2006; Cheung et al., 2008; Green et al., 2008; Brunello et al., 2009). However, the role of PS in ER Ca2+ leakage is debated (Tu et al., 2006; Shilling et al., 2012).
Conversely, it was also reported that Ca2+ homeostasis may influence APP pathophysiological processing. Therefore, Aβ production is enhanced by elevation of intracellular [Ca2+] (Buxbaum et al., 1994; Querfurth and Selkoe, 1994), or increased RyR-mediated Ca2+ release (Querfurth et al., 1997), and is reduced in IP3R-deficient lines (Cheung et al., 2008).
While perturbations of Ca2+ homeostasis have been largely described in PS models; fewer studies focused on the direct impact of APP on Ca2+ homeostasis (Leissring et al., 2002; Lopez et al., 2007; Rojas et al., 2008; Niu et al., 2009). Nevertheless, the characterization of subcellular Ca2+ signaling dysregulation in APP-expressing models, and the possible implication of RyR in APP-mediated Ca2+ alteration have not been reported before. In addition, the blockade of RyR as a mean to modulate APP metabolism and Aβ production has not been investigated.
We provide here evidence that enhanced RyR-mediated Ca2+ release, occurs in SH-SY5Y neuroblastoma cell line stably overexpressing either wild-type human APP (APP695), or APP harboring the Swedish double mutation (K670N/M671L) (APPswe) and in primary neurons from APPswe-expressing mice (Tg2576). Interestingly, blockade of RyR-mediated Ca2+ release by dantrolene reduces Aβ production in vitro in both SH-SY5Y model, and Tg2576 primary neurons. Moreover, dantrolene diminishes Aβ load, reduces Aβ-related histological lesions and slows down learning and memory deficits in Tg2576 mice. All together, our data demonstrate that ER Ca2+ dysregulation acts as an amplification pathway in the Aβ cascade and identify RyR as a target of Ca2+ pathology linked to AD.
Dantrolene, SB415286, Roscovitine, SP600125, Caffeine, and Carbamoylcholine chloride were purchased from Sigma Aldrich.
Aβ, C99 and total APP were detected using the 6E10 antibody (Covance) recognizing 1–16 residues of Aβ. Aβ was also detected using FCA18 antibody, recognizing free Asp1 residue of Aβ1-x peptides (Barelli et al., 1997). Total APP was detected using APP N-terminal antibody (22C11, Millipore) recognizing 66–81 residues of APP, or APP C-terminal antibody (Sigma Aldrich) recognizing 676–695 residues of APP. Phosphorylated APP was detected using P-APP antibody (Thr-668) (Cell Signaling). Other antibodies directed towards the following proteins were as follows: GADPH (Millipore); PSD-95, SERCA2B, Aph1 and RyR-1/2/3 (Thermo Scientific Pierce Products); β-Actin and IP3R-1/2/3 (Santa Cruz); BACE-1 and CytP450 (Abcam); SNAP-25 (Covance); Vamp-2 (Synaptic System); Synapsin-I/II and Synaptotagmin (developed by F.B.); Nicastrin (Sigma Aldrich); and PS1 (a generous gift from Gopal Thinakaran).
Human SH-SY5Y neuroblastoma cells (CRL-2266, ATCC) were cultured following manufacturer’s instructions. SH-SY5Y cells stably expressing pcDNA3.1, APPswe or APP695 constructs were generated following standard protocols and maintained in the presence of 400 µg geneticin (Gibco).
For subcellular Ca2+ analyses, 150,000 cells were spotted on 13-mm coverslips, and placed 24 h later in contact with the appropriate Adenoviral system expressing cytosolic-(AdCMVcytAEQ) or ER-(AdCMVerAEQ) targeted aequorin probes as already described (Chami et al., 2008).
APPswe- and APP695-expressing SH-SY5Y cells and primary cultured neurons were treated over night (20 h) with respectively 50 µM dantrolene or 1 µM dantrolene or with vehicle (DMSO). Protein extracts were prepared using lysis buffer (50 mM Tris pH 8, 10 % glycerol, 200 mM NaCl, 0.5 % Nonidet p-40, and 0.1 mM EDTA) supplemented with protease inhibitors (Complete, Roche diagnostics).
To detect Aβ peptide, 40 µg of the total proteins were incubated with 70 % formic acid (Sigma) and Speed Vac evaporated for 40 min. The pellets were dissolved in 1 M Tris pH 10.8, 25 mM Betaine and diluted in 2× Tris-Tricine loading buffer (125 mM Tris-HCl pH 8.45, 2 % SDS, 20 % Glycerol, 0.001 % Bromophenol blue, and 5 % β-mercaptoethanol). Proteins were resolved by 16.5 % Tris-Tricine SDS-PAGE, transferred onto nitrocellulose membranes, and incubated overnight with specific antibodies as specified in legends. All the other proteins were detected on total extracts resolved by SDS-PAGE following standard procedures.
Cells were harvested by trypsinization and centrifuged at 600×g for 10 min at 4 °C. The pellets were resuspended in 1 ml isolation buffer (250 mM D-Mannitol, 5 mM HEPES pH 7.4, 0.5 mM EGTA, and 0.1 % Bovine Serum Albumin (BSA)) supplemented with protease inhibitor mixture. After chilling on ice for 20 min with frequent tapping, cells were disrupted by at least 200 strokes of a glass Dounce homogenizer and the homogenate was centrifuged at 1,500×g at 4 °C to remove unbroken cells and nuclei. The supernatant was centrifuged at 100,000×g at 4 °C for 1 h. The pellet containing the microsomal fraction was suspended in 0.25 M Sucrose, and 10 mM Tris-HCl pH 7.4 supplemented with protease inhibitors.
Culture medium without serum was supplemented with protease inhibitors, 1 mM PMSF and 0.1 % BSA. After a brief centrifugation, supernatants were mixed with equal volumes of 20 % TCA (Trichloroacetic acid), incubated at 4 °C for 30 min and then centrifuged at 18,000×g for 15 min at 4 °C. Pellets were washed with ice-cold acetone, centrifuged at 10,000×g for 5 min at 4 °C, then dried and dissolved with RIPA buffer (50 mM Tris-HCl pH 8, 150 mM NaCl, 1 % NP-40, 0.5 % sodium deoxycholate, and 0.1 % SDS) supplemented with protease inhibitors. Aβ content was assessed by 10–20 % Tris-Tricine SDS-PAGE (Invitrogen).
The concentrations of Aβ40 and Aβ42 were measured in the culture medium by using the respective ELISA kits (Invitrogen) following the manufacturer’s instructions.
CytAEQ was reconstituted with 5 µM coelenterazine for 2 h in Krebs-Ringer modified buffer (KRB) (in mM: 125 NaCl, 5 KCl, 1 Na3PO4, 1 MgSO4, 5.5 glucose, 20 HEPES pH 7.4) supplemented with 1 mM CaCl2 at 37 °C. Cytosolic Ca2+ signals were obtained upon application of 500 µM carbamoylcholine chloride, 135 mM KCl, or 30 mM caffeine.
For reconstitution with high efficiency of the erAEQ, the luminal [Ca2+] of this compartment was first reduced by incubating cells for 1 h at 4 °C in KRB supplemented with 5 µM n-coelenterazine, 1 µM ionomycin, and 600 µM EGTA. After this incubation, cells were extensively washed with KRB supplemented with 2 % BSA before the luminescence measurement was initiated. The ER was refilled by exposing cells to 1 mM extracellular CaCl2. All aequorin measurements were carried out in a purpose built luminometer. The experiments were terminated by lysis of cells with 100 µM digitonin in a hypotonic Ca2+-rich solution (10 mM CaCl2, H2O) to discharge the remaining aequorin pool. The light signal was collected and calibrated into [Ca2+] values, as previously described (Chami et al., 2008). After reaching the steady state value, cells were perfused with (50 µM) tBuBHQ, thus blocking SERCA pump and activating passive ER Ca2+ leakage.
Fura2-AM Ca2+ measurements were performed as described previously (Bisaillon et al., 2010). Briefly, cells were loaded with 4 µM Fura2-AM (Molecular Probes). Cells were then washed and bathed in HEPES buffered Hank's Balanced Salt Solution (HBSS) (in mM: 140 NaCl, 1.13 MgCl2, 4.7 KCl, 2 CaCl2, 10 D-glucose, and 10 HEPES pH 7.4) for 10 minutes before Ca2+ was measured. Fluorescence images of several cells were recorded and analyzed with a digital fluorescence imaging system (InCyt Im2, Intracellular Imaging, Cincinnati, OH). Fura2 Fluorescence was collected at 510 nm upon alternate excitation at 340 nm and 380 nm and ratio of fluorescence in response to 340 nm excitation to that in response to 380nm excitation was obtained on a pixel-by-pixel basis and represented as raw data.
Standard whole-cell patch clamp recordings were performed using an Axopatch 200B and Digidata 1440A (Axon Instruments) as previsouly published (Zhang et al., 2011). Clampfit 10.1 software was used for data analysis. Pipettes were pulled from borosilicate glass capillaries (World Precision Instruments, Inc.) with a P-97 flaming/brown micropipette puller (Sutter Instrument Company) and polished with DMF1000 (World Precision Instruments, Inc.) to a resistance of 2–4 MΩ when filled with pipette solutions (in mM: 145 Cs-methanesulfonate, 20 Cs-1,2-bis-(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (Cs-BAPTA), 8 MgCl2, and HEPES (pH adjusted to 7.2 with CsOH). Immediately before the experiments, cells were washed with bath solution (in mM: 110 TEA-Cl, 10 CsCl, 10 HEPES, 10 CaCl2 (pH was adjusted to 7.4 with CsOH). Only cells with tight seals (>16 GΩ) were selected for break in. Whole-cell currents were recorded every 2 seconds with a standard voltage ramp from −140 mV to 40 mV (lasting 180 ms) from a holding potential −80 mV. Currents were low-pass filtered at 5 kHz and sampled at a rate of 10 kHz.
Primary cultured neurons were obtained from individual e17 mice, then plated separately and genotyped for the presence of APP695 transgene by PCR. Cortices and hippocampi were dissected and digested in 0.125 % trypsin-HEPES-buffered saline solution for 30 min. Cells were seeded onto poly-L-lysine-coated tissue culture plates or glass coverslips. Cultures were incubated at 37 °C, 5 % CO2 in Neurobasal medium (Gibco) supplemented with B-27. Medium was changed after 4 h with Neurobasal containing B-27 supplemented with 0.1 % glutamine, and 1 % penicillin-streptomycin.
β-secretase activity was monitored using the β-secretase activity assay kit (Biovision). Briefly, 5,000,000 cells were lysed in ice-cold extraction buffer and centrifuged at 10,000×g for 5 min. The supernatant (50 µg) was then incubated in the presence of BACE-1 substrate at 37 °C, in the presence or absence of 50 µg BACE-1 inhibitor JMV2764 (Buggia-Prevot et al., 2008). BACE-1 activity corresponds to the JMV2764-sensitive fluorescence recorded at 320 nm (excitation) and 420 nm (emission) wavelengths.
In vitro γ-secretase assay was assessed as already described (Sevalle et al., 2009). Intact cell pellets were suspended in 10 mM Tris pH 7.5 supplemented with protease inhibitor mixture, and subjected to repeated passages through a 25G needle. Homogenates were first centrifuged at 800×g for 10 min at 4 °C and the resulting supernatant was subjected to an additional 20,000×g centrifugation for 1 h at 4 °C. Membrane-containing pellets were then resuspended in solubilization buffer (150 mM sodium citrate pH 6.4 containing 3-[(3-cholamydopropyl) dimethylammonio]-2-hydroxy-1-propanesulfonate 1 % (v/v)) supplemented with protease inhibitor mixture. All steps were performed at 4 °C. Solubilized membranes (1 mg/ml) were diluted once with sodium citrate buffer (150 mM pH 6.4), and with reaction buffer (150 mM sodium citrate pH 6.4, 20 mM dithiothreitol, 0.2 mg/ml BSA, 1 mg/ml egg phosphatidyl choline and 50 µg/mL recombinant C100-FLAG). The resulting reaction mix were then either incubated over constant agitation for 16 h at 37 °C or stored at 4 °C (negative controls). Samples were then supplemented with 2× Tris-Tricine loading buffer, boiled for 5 min and subjected to western blot for Aβ analysis using 16.5 % Tris-Tricine SDS-PAGE.
Total RNA was isolated using NucleoSpin RNA II (Macherey-Nagel) according to the manufacturer’s protocol. Total RNA extraction form the cortex of WT and Tg2576 mice was isolated using RNAeasy lipid tissue (Qiagen) according to the manufacturer’s protocol. Complementary DNA (cDNA) was synthesized from 2 µg of total RNA and random primers using GoScript Reverse Transcription System kit (Promega). Target gene expression was analyzed by real time PCR using Corbett Rotor-Gene 6000 (Invitrogen) and SYBR Green (Roche Applied Sciences). Cycling parameters were as follows: 20 sec at 95 °C, 20 sec at 60 °C, and 20 sec at 72 °C for 55 cycles. Primer sequences for human RyR isoforms were as follows: RyR1 forward: 5’-GCATGGCTTCGAGACTCAC-3’; RyR1 reverse: 5’-CATCTTCCAGACATAAGACTCCTG-3’; RyR2 forward: 5-’TGCTGGCTTGGGGCTGGAG-3’; RyR2 reverse: 5’-ACCATGGGCAGCGTCCACAG-3’; RyR3 forward: 5’-GACATGCGAGTCGGCTGGGC-3’; RyR3 reverse: 5’-GATGCCAACGCTGGCCCCTG-3’. Human β-Actin was used as control gene.
Primer sequences for mouse RyR isoforms were as follows: RyR1 forward: 5′-TCTTCCCTGCTGGAGACTGT-3′; RyR1 reverse: 5’-GTGGAGAAGGCACTTGAGG-3’; RyR2 forward: 5′-TCAAACCACGAACACATTGAGG-3′; RyR2 reverse: 5′-AGGCGGTAAAACATGATGTCAG-3′; RyR3 forward: 5′-CTGGCCATCATTCAAGGTCT-3′; RyR3 reverse: 5′-GTCTCCATGTCTTCCCGTA-3′. Mouse GAPDH was used as control gene.
Experiments were carried out in accordance with the European Community (86/609/EEC) directives regulating animal research, the Italian Ministry of Health (DL 116/92; DL 111/94-B), and by the “Institut Fédératif de Recherche Necker-Enfants Malades” animal care and use committee. Tg2576 mice were developed by Hsiao et al (Hsiao et al., 1996) and carrying human APP695 cDNA with the Swedish double mutation at positions (K670M→N671L) under the control of the hamster Prion promoter. The genotype of the mice was confirmed by PCR using DNA from tail tissues. Tg2576 mice and wild-type (WT) littermates of either sex between 12–15 month-old were treated with 5 mg/kg dantrolene or with PBS (vehicle) by intraperitoneal injections twice a week during 3 months. All mice were weighted each month.
Dantrolene was freshly prepared in pre-warmed PBS solution and subjected to brief sonication before injection. Dantrolene solution was mixed thoroughly before each injection. After 3 months of treatment, mice were subjected to behavioral testing and then sacrificed. One half of the brains was immediately fixed in freshly depolymerized 4 % paraformaldehyde and used for immunostaining. The other half of the brains was dissected to isolate cortices and hippocampi. Snap frozen cortices were then homogenized to have a powder mixture which was used for protein and RNA isolation.
Brains were sectioned to 8 µm-thickness with a cryostat and stained for amyloid plaques. Slices were first washed with PBS, incubated in formic acid for 6 min, and then in H2O2 for 15 min. Non-specific sites were saturated in PBS, 0.05 % tween, 5 % BSA for 1 h. Slides were incubated overnight with primary antibodies (6E10, 1:1000 or FCA18, 1:500) prepared in PBS, tween 0.025 %. After washes, sections were incubated with secondary HRP-conjugated (1:1000, Jackson Labs), or Alexa 488-fluorescent antibodies (1:1000, Molecular Probes) at Room Temperature for 1 hour. Fluorescent slides were incubated for 5 minutes with DAPI (Roche, 1:20000). Slides with HRP-conjugated antibodies were incubated with DAB-impact (Vector), rinsed and counter-stained with Cresyl violet. Images were captured using DM108 microscope (Leica), or an epifluorescence microscope (Axioplan2, Zeiss) under 10× and 20× magnification. Counting of Aβ plaques was performed on 15 serial slices from each animal blindly by two different researchers.
The water maze test was performed in a 1.2 m-diameter pool. A 10 cm-diameter platform was placed in the southwestern quadrant in the hidden trials as already described (Morris, 1984). The procedure consisted of 5 days of hidden platform tests, plus a probe trial 24 h after the last hidden platform test. In the hidden platform tests, mice were trained for 4 trials, with an inter-trial interval of 10 min. After the probe test, mice were trained in a visible platform tests. In the visible platform test, mice were tested for 4 trials with an inter-trial interval of 10 min. Tracking of animal movement was evaluated using an ANYmaze system (Ugo Basile, Italy).
Novel object recognition was performed in a 44 × 44 cm open field chamber with opaque walls equipped with a digital video recording system as already described (Bevins and Besheer, 2006). The objects used during the task were Object Lego different for shape and color. Mice were first habituated to the chamber for 10 min during which Any-Maze (Ugo Basile, Italy) software quantifies various locomotor parameters and anxiety-related behavior, including total distance traveled, time spent moving ≥ 50 mm/sec, number of entries into and time spent in the central part of open field chamber.
Twenty-four hours after the habituation session, mice were subjected to training in a 10 min session of exposure to two different objects in the open field box. The time spent exploring each object was recorded using the video tracking. Exploration consisted of any investigative behavior (i.e., head orientation, sniffing occurring within < 1.0 cm) or deliberate contact that occurred with each object. After the training session, the animal was returned to its home cage. After 24 h retention interval, the animal was returned to the arena with one familiar object and a novel one. Objects were counterbalanced between sessions and animals, and were cleaned with 70 % ethanol after each trial. The time spent in exploring each object was then measured. A discrimination index was calculated as following:
Animals that spent less than 20 sec exploring the objects during the 10 min test session were omitted from analysis.
Results are reported from at least three different experiments. Statistical analyses were done using t-test or one-way or two-ways Anova. Bonferroni, Dunnet’s, or Tukey’s multiple comparison post-hoc analyses were subsequently performed on ANOVA results to determine significance.
We set up neuroblastoma SH-SY5Y cell lines stably overexpressing APP695 or APPswe (Fig. 1A). Both APP695- and APPswe-overexpressing cells yield increased levels of C99 (issued from cleavage by β-secretase; Fig. 1A), and of Aβ40 and Aβ42 peptides (issued from sequential cleavages by β- and γ-secretases) (Aβ42 (pg/µg of protein): 21.6± 2.9, n=10, and 20.5 ± 2.4, n=13 in APPswe- and APP695-expressing cells respectively, versus 1.6 ± 0.5, n=11 in mock-transfected cells; Aβ40 (pg/µg of protein): 229 ± 46, n=5, and 160 ± 27, n=9 in APPswe- and APP695-expressing cells respectively, versus 17 ± 3, n=6 in mock-transfected cells; Fig. 1B).
We analyzed Ca2+ release from the ER by using cytosolic Ca2+-based aequorin probe (Chami et al., 2008). We first examined RyR-mediated Ca2+ release upon cell stimulation with the RyR agonist caffeine (30 mM) (Riddoch et al., 2005). As shown in Fig. 1C, caffeine elicits a fast and large Ca2+ transient that was amplified in APPswe- and APP695-expressing cells as compared to control (peak (µM): 1.96 ± 0.05, n=24, and 2.09 ± 0.04, n=22 respectively, versus 1.19 ± 0.02, n=24 in control) (Fig. 1C). We next investigated cytosolic Ca2+ signal upon stimulation of Ca2+ release through the IP3R. It was already reported that the stimulation of SH-SY5Y cells with the muscarinic agonist carbamoylcholine chloride (carbachol) caused a cytosolic Ca2+ response mainly mediated by Ca2+ release from IP3-sensitive stores (van Acker et al., 2000). Accordingly, carbachol (500 µM) application triggers a transient increase in cytosolic Ca2+, the extent of which was significantly larger in APPswe- and APP695- expressing cells as compared to control (peak (µM): 4.63 ± 0.07, n=24, and 4.30 ± 0.05, n=24 respectively, versus 1.34 ± 0.03, n=24 in control) (Fig. 1D).
The influx of extracellular Ca2+ through the plasma membrane also participates to the increase of cytosolic [Ca2+]. We therefore investigated the contribution of Voltage-Gated Ca2+ Channels (VGCC)-mediated Ca2+ entry. The application of KCl (135 mM) triggers membrane depolarization leading to the opening of VGCC thereby, inducing Ca2+ entry into the cytosol. Figure 1E shows a significant increase in KCl-evoked Ca2+ entry in APPswe- and APP695-expressing cells as compared to control (peak (µM): 1.51 ± 0.05, n=23, and 1.57 ± 0.08, n=24 respectively, versus 0.98 ± 0.05, n=24 in control) (Fig. 1E). VGCC-mediated Ca2+ entry may trigger Ca2+ release from internal stores through a mechanism known as Ca2+-induced Ca2+ release (CICR). To investigate CICR, we used dantrolene, a well characterized antagonist of RyR channels (Muehlschlegel and Sims, 2009), and measured Ca2+ entry upon application of KCl (135 mM). Since APP695- and APPswe-expressing cells harbor the same alteration of Ca2+ signals (Fig. 1 C–E), we performed these analyses on APPswe-expressing cells only. The results show that VGCC-mediated Ca2+ entry is reduced upon dantrolene treatment in APPswe-expressing SH-SY5Y cells but not in pcDNA3.1-expressing cells (Fig. 1F). These data led us to conclude that RyR-mediated Ca2+ signals contribute through CICR mechanism to increased VGCC-mediated Ca2+ entry in APPswe-expressing SH-SY5Y cells.
We also investigated Ca2+ influx through voltage-independent plasma membrane Ca2+ channels. Control and APPswe-expressing cells were incubated in EGTA-rich solution to buffer extracellular Ca2+ followed by restoration of a Ca2+-rich solution to the extracellular milieu, thereby assessing basal Ca2+ entry across the plasma membrane. We noticed that APPswe-expressing cells harbor increased basal [Ca2+] illustrated by a higher basal plateau values before the application of EGTA solution ((F340/F380): 1.111 ± 0.009, n=20, versus 1.075 ± 0.008, n=20 in control) and larger Ca2+ entry revealed by an increased plateau value reached upon addition of Ca2+-rich solution ((F340/F380): 0.263 ± 0.027, n=20, versus 0.170 ± 0.020, n=20 in control) (Fig. 1G).
Together, these data reveal that APP overexpression determines an increase of cytosolic Ca2+ signals due to combined increased Ca2+ release from the ER through IP3R and RyR (Fig. 1C, and 1D), and enhanced Ca2+ entry through voltage-dependent and voltage-independent plasma membrane Ca2+ channels (Fig. 1E, and 1G). Nevertheless, we show that elevated VGCC- mediated Ca2+ signals in APPswe-expressing SH-SY5Y cells is a consequence of CICR through RyR (Fig. 1F).
Increased Ca2+ release from the ER through IP3R and RyR could be associated with altered ER Ca2+ loading capacity. We investigated ER Ca2+ content by using Ca2+-based aequorin probe targeted to the ER (Chami et al., 2008). We measured the ER Ca2+ load capacity upon application of 1 mM CaCl2-rich solution. As shown in Figure 2A, ER Ca2+ loading is reduced in APPswe-expressing cells as compared to control (plateau (µM): 166.0 ± 3.7, n=15, versus 245.0 ± 15.8, n=21 respectively) (Fig. 2A). The analysis of ER Ca2+ uptake capacity (ascending slope phase of the curve) did not reveal any difference between control and APPswe-expressing cells, thus ruling out a possible alteration of the activity of SERCA (Sarco-Endoplasmic Reticulum Ca2+ ATPase) (uptake (µM/sec): 11.5 ± 0.4, n=15 in APPswe-expressing cells, versus 9.3 ± 0.9, n=21 in control). We also analyzed the ER Ca2+ passive leak upon SERCA inhibition by tBuBHQ. As displayed in Figure 2B, APPswe-expressing cells show an increased Ca2+ leakage from the ER as compared to control cells (as revealed by increased slope (µM/sec): 0.85 ± 0.03 n=15 in APPswe-expressing cells, versus 0.51 ± 0.02, n=21 in control) (Fig. 2B). These data demonstrate that the reduced Ca2+ loading capacity in APPswe-expressing cells is due to increased Ca2+ release through IP3R and RyR, and to elevated ER Ca2+ passive leakage.
It is known that depletion of ER Ca2+ activates Ca2+ influx through the plasma membrane, a mechanism known as store-operated Ca2+ entry (SOCE) (Smyth et al., 2010).
We investigated SOCE in APPswe-expressing cells upon ER Ca2+ depletion by carbachol-mediated IP3R Ca2+ release, or by Thapsigargin (TG)-mediated SERCA blockade in the presence of EGTA (Fig. 2C and 2D respectively). Under these conditions, we confirm that Ca2+ release from intracellular stores is larger in APPswe-expressing cells than in controls (carbachol peak (F340/F380): 1.093 ± 0.049, n=87, versus 0.409 ± 0.096, n=83 respectively, and TG peak (F340/F380): 0.306 ± 0.017, n=96, versus 0.212 ± 0.029, n=97 respectively) (Fig. 2C and 2D). Upon carbachol and TG-induced ER Ca2+ depletion, we notice that Ca2+-mediated SOCE is larger in APPswe-expressing cells than in control. Application of low concentrations of Gd3+ (5 µM; inhibitor of Orai-mediated Ca2+ entry) abolishes completely Ca2+ entry in both control- and APPswe-expressing cells, suggesting that SOCE in these cells is likely mediated by STIM/Orai signaling complexes, independently of transient receptor potential canonical (TRPC) channels. Surprisingly, Gd3+-mediated Ca2+ entry inhibition occurred with similar kinetics in APPswe-expressing cells and control, suggesting that Ca2+ pumping mechanisms are similar in control and APPswe-expressing cells.
Since Fura2 measurements are prone to artifacts and a constitutive Ca2+ entry under certain conditions could be amplified by the Ca2+ off/Ca2+ on protocol routinely used to assess SOCE, we also measured ICRAC (Ca2+ release-activated Ca2+ current), the main non-voltage-gated SOCE current using standard electrophysiological recordings as already described (Potier et al., 2009). We show that passive store depletion by high concentrations (20mM) of the fast chelator BAPTA activates an ICRAC current (sampled at −100mV) with similar size and kinetics in APPswe-expressing cells and pcDNA3.1 control cells. Importantly, we notice that Gd3+-dependent ICRAC blockade occurs in a similar manner in pcDNA3.1 and APPswe-expressing cells (Fig. 2E). These data are further confirmed by representing the current-voltage (I–V) relationships from the ramp protocol wherein current density was evaluated at various membrane potentials (Fig. 2F).
These experiments demonstrate that APPswe-expressing cells manifest a larger SOCE upon store depletion and did not reveal any alteration of ICRAC. Therefore, we postulate that increased Ca2+ entry in APPswe-expressing cells is likely due to exaggerated unregulated basal Ca2+ entry; and that this increase is not due to enhanced SOCE and ICRAC.
The experiments in Figure 1 and and22 reveal that ER Ca2+ homeostasis is largely deregulated in APPswe-expressing cells. Consequently, we focused our study on the molecular mechanisms underlying ER Ca2+ store emptying. We analyzed the expression of Ca2+ mobilizing proteins in this compartment, namely RyR, IP3R, and SERCA2b. We noticed a higher expression of both IP3R and RyR in APPswe- and APP695-expressing cells (RyR: 2.2 ± 0.4 and 1.8 ± 0.5 respectively, versus 0.8 ± 0.1 in control; IP3R: 2.0 ± 0.2 and 1.9 ± 0.3 respectively, versus 1.0 ± 0.1 in control) (Fig. 3A). As expected from Ca2+ uptake experiments (Fig. 2A), no significant change in SERCA2b expression was observed in APPswe- and APP695-expressing cells (SERCA2b: 1.3 ± 0.1 and 1.2 ± 0.1 respectively, versus 1.1 ± 0.2 in control) (Fig. 3A). Since Tg2576 mice are characterized by a major accumulation of Aβ in the cortex, the same analyses were also performed on cortices isolated from 12–15 month-old Tg2576 mice and wild type mice (WT). Our data show that Tg2576 mice harbor an increased expression of RyR (RyR: 1.5 ± 0.1 in Tg2576 mice, n = 4, versus 1.0 ± 0.1, n = 4 in WT mice), while the expression of IP3R is not significantly affected (Fig. 3B).
Since, the induction of IP3R expression is observed only in SH-SY5Y model and that the dysregulation of RyR expression is reported in both in vitro and in vivo APP-overexpressing models, we then compared mRNA expression levels of the three RyR isoforms in both SH-SY5Y model, and Tg2576 mice. By using quantitative RT-PCR, we show an increased expression of RyR-1/2/3 mRNAs in APPswe- and APP695- expressing cells (RyR1: 1.5 ± 0.11 and 1.7 ± 0.03; RyR2: 1.6 ± 0.14 and 1.7 ± 0.13; and RyR3: 1.52 ± 0.12 and 1.64 ± 0.13 and in APPswe- and APP695-expressing cells respectively, versus control cells taken as 1) (Fig. 3C). The same analyses performed on cortices isolated from 12–15 month-old Tg2576 and WT mice show a significant increase of the expression of RyR2 isoform, while the expression of RyR1 and RyR3 isoforms remain unchanged (RyR1: 0.69 ± 0.26; RyR2: 1.45 ± 0.17; and RyR3: 0.72 ± 0.30 in Tg2576 mice, versus WT mice taken as 1) (Fig. 3D). To note, comparative analyses of the expression (cycle threshold value which is defined as the number of cycles required for the fluorescent signal to exceed background level) of the three RyR isoforms reveal that in SH-SY5Y cells, RyR3 is more abundant than RYR2, which is more expressed than RyR1, while RyR2 and RyR3 are the major isoforms expressed in the cortex of Tg2576 and WT mice (data not shown).
Therefore, these data revel that RyR upregulation may underlies ER Ca2+ homeostasis dysregulation in both SH-SY5Y model and Tg2576 mice.
We then explored the potential implication of RyR-mediated Ca2+ release in the modulation of APP processing. It was already reported that caffeine-mediated RyR Ca2+ release stimulates Aβ production (Querfurth et al., 1997). Accordingly, treatment of APPswe-expressing cells with caffeine (5 mM) increases the production of C99 fragment derived from APP processing by β-secretase (Fig. 4A).
We used dantrolene, to modulate RyR-mediated Ca2+ release (Muehlschlegel and Sims, 2009). The concentration and duration of treatment with dantrolene were determined in SH-SY5Y cells using cell viability test and Ca2+ measurements analyses. Cell viability is not altered in APPswe-expressing SH-SY5Y cells treated for 20 h with dantrolene (50 µM) (data not shown). Under these experimental conditions, we show that dantrolene significantly reduces RyR-dependent Ca2+ release in APPswe-expressing cells, but not in control cells (Fig. 4B).
We then assessed whether dantrolene could modify the proteolytic fragments derived from APP processing by β-secretase (C99), or β- and γ-secretases (Aβ40/42). Interestingly, dantrolene treatment reduces the production of C99 fragment in both APPswe- and APP695- expressing cells. Quantification revealed a reduction of C99 production of about 30 % in APPswe- and APP695-expressing cells (Fig. 4C). Dantrolene treatment also significantly decreases Aβ42 production (Aβ42 (AU): 0.6 ± 0.1, n=10, and 0.7 ± 0.1, n=19 in dantrolene-treated APP695- and APPswe-expressing cells respectively, versus vehicle-treated cells taken as 1) (Fig. 4D).
In order to rule out any artifactual effect due to the immortalization of cell lines, we investigate the effect of dantrolene in primary cultured neurons isolated from WT and Tg2576 mice. Neurons from Tg2576 mice yield enhanced levels of C99 at 7 days in vitro (DIV) that is maintained at 12 and 15 DIV (Fig. 5A). Tg2576 primary cultured neurons harbor an alteration of intracellular Ca2+ signaling as demonstrated by the increased Ca2+ release upon stimulation with caffeine (30 mM) (peak (µM): 5.57 ± 1.05, n=7 and 3.29 ± 0.32, n=8 in Tg2576 versus WT neurons respectively; Fig. 5B), and an increased VGCC-dependent Ca2+ entry upon stimulation with KCl (50 mM) (peak (µM): 2.70 ± 0.25, n=13 and 1.83 ± 0.17, n=13 in Tg2576 versus WT neurons respectively; Fig. 5C).
In primary neurons, treatment with dantrolene (1 µM, 20 h) does not alter cell viability (data not shown). Under these conditions, dantrolene reduces C99 peptide production (0.6 ± 0.3 in dantrolene-treated Tg2576 neurons, versus vehicle-treated Tg2576 neurons taken as 1; Fig. 5D), and total Aβ peptide present in culture medium (54 % ± 13 in dantrolene-treated Tg2576 neurons, versus vehicle-treated Tg2576 neurons taken as 100 %; Fig. 5E).
Both data obtained in SH-SY5Y expressing cells and primary cultured neurons clearly demonstrate that the inhibition of RyR-mediated Ca2+ release controls APP processing and the production of C99 fragment and Aβ peptide.
Amyloidogenic metabolism of APP implies its sequential cleavage by β- and γ-secretases (Checler, 1995). It was also reported that APP phosphorylation on Thr-668 (P-APP) plays a major role in APP metabolism and the production of Aβ (Pierrot et al., 2006). Therefore, the reduction of C99 and Aβ peptide production upon dantrolene treatment may be linked to decreased expression and/or activity of β- and γ-secretases or alteration of APP phosphorylation on Thr-668.
We show that SH-SY5Y stably overexpressing APPswe or empty vector display similar expression levels of BACE-1 (β-secretase) and PS1, Aph1 and Nicastrin (components of γ-secretase complex (Checler, 1995)), the expression levels of which remained unaffected by dantrolene (Fig. 6A).
We performed two sets of experiments in APPswe-expressing SH-SY5Y cells to investigate APP phosphorylation. First, we analyzed the extent of Thr-668 P-APP upon dantrolene treatment, and used, as controls, inhibitors of candidate kinases thought to be implicated in APP phosphorylation (CdK5, GSK3β, and JNK) (Muresan and Muresan, 2007). As shown in Figure 6B, the addition of dantrolene or CdK5, GSK3β, and JNK kinase inhibitors (Roscovitine, SB415286, and SP600125 respectively) significantly reduce the extent of Thr-668 P-APP (Fig. 6B). We then analyzed the time courses of APP phosphorylation and C99 production in dantrolene-treated APPswe cells and showed that dantrolene concomitantly and persistently reduces P-APP and C99 production level as soon as after one hour-treatment (Fig. 6C). These data establish that dantrolene like kinases inhibitors modulate APP phosphorylation on Thr-668 residue.
We then investigated the effect of dantrolene and of kinase inhibitors on in vitro β- and γ-secretases activities. Our data show that β-secretase activity is significantly decreased upon treatment with dantrolene and roscovitine (48.0 % ± 8.5, and 61.2 % ± 16.6 respectively, versus 100.0 % ± 10.8 in vehicle-treated cells, n=5; Fig. 6D), but not with SB415286 and SP600125 (92.0 % ± 23.0, and 79.5 % ± 23.3 respectively, n=5; Fig. 6D).
We also monitored in vitro γ-secretase activity in reconstituted membranes prepared from dantrolene- or kinase inhibitors-treated cells. We found that: i) recombinant C100 fragment is cleaved at 37 °C and to a much lesser extent at 4 °C (negative control), and ii) Aβ production by membranes prepared from mock- or non-treated APPswe-transfected cells is similar. Interestingly, a significant reduction of Aβ production was observed with membranes prepared from dantrolene-treated APPswe-expressing cells as compared to vehicle treated ones (Aβ versus C100 signal: 0.65 ± 0.07 versus 1.0 ± 0.1 respectively) (Fig. 6E). In order to rule out a putative direct effect of dantrolene on γ-secretase that would have interfered with the in vitro assay, we incubated the C100 fragment with membranes isolated from untreated APPswe cells in the absence or the presence of dantrolene. Dantrolene does not modify the C100 fragment cleavage (data not shown), thus demonstrating that the reduction of γ-secretase activity upon dantrolene treatment is not linked to a direct interaction of dantrolene with the γ-secretase complex. Our data also reveal that kinase inhibitors reduce γ-secretase activity, in a significant manner with SB415286 (Aβ versus C100 signal: 0.52 ± 0.09 versus 1.0 ± 0.1 in vehicle-treated cells), and to a lesser extent with Roscovitine and SP600125 (Aβ versus C100 signal: 0.73 ± 0.09 and 0.64 ± 0.08 respectively) (Fig. 6E).
These data demonstrate that dantrolene reduces both β- and γ-secretases activities and that under our experimental conditions, β-secretase activity is reduced upon Cdk5 inhibition, while γ-secretase activity is reduced upon GSK3β inhibition.
Our consistent data obtained in the SH-SY5Y cells and in primary neurons led us to explore the functional consequences of dantrolene in vivo. We used Tg2576 mice developed by Hsiao et al. (Hsiao et al., 1996). This model shows an impairment of learning and memory starting from 9–10 months of age accompanied by an increase in Aβ40 and Aβ42–43 peptides and the development of mature senile plaques (Hsiao et al., 1996). The chronic treatment (3 months) with dantrolene was administered to 12– 15 month-old mice, i.e when mice already displayed significant AD-related histological lesions and cognitive deficits. It is noteworthy that dantrolene has been already used in vivo (Chen et al., 2011), and recent evidence suggests that it readily crosses the blood brain barrier (reviewed in (Muehlschlegel and Sims, 2009)). Our data demonstrate that dantrolene treatment significantly reduces the density of Aβ plaques in Tg2576 mice ((Aβ plaques /section): 37 ± 8, n=6 in dantrolene-treated Tg2576 mice versus 89 ± 20, n=5 in vehicle-treated Tg2576 mice) as revealed using the 6E10 antibody recognizing 1–16 residues of Aβ peptides and C99 fragment (Fig. 7A). A similar result was obtained using the FCA18 antibody that recognizes Asp 1 residue of Aβ1-x peptides and C99 (Barelli et al., 1997) (Fig. 7A). No staining was detected with these antibodies in WT mice. Dantrolene-treated Tg2576 mice also exhibit a lower production of C99 and total Aβ peptide than vehicle-treated mice (C99: 0.6 ± 0.1, and Aβ: 0.4 ± 0.1, n=13–18 in dantrolene-treated mice as compared to vehicle-treated mice taken as 1, n=10–13) (Fig. 7B).
As dantrolene reduced Aβ burden in Tg2576 mice in vivo, we hypothesized that this may lead to prevention of AD-related phenotype in this model i.e. alteration of synaptic function and learning and memory decline.
We analyzed the expression of pre-synaptic proteins implicated in vesicles mobilization and docking (Synapsin-I, SNAP-25, VAMP-2, and Synaptotagmin (Stg)), as well as of post-synaptic scaffold protein (PSD-95). PSD-95 expression is significantly reduced in 15–18 month-old Tg2576 mice as compared to age-matched WT mice (0.6 ± 0.1, n=10, versus 1.0 ± 0.1, n=13 in Tg2576 and WT mice respectively), while there is no significant modification of the expression of pre-synaptic SNAP-25, VAMP-2, Stg and Synapsin-I proteins (Fig. 7C). Consequently, we examined the impact of dantrolene on the expression of PSD-95 in WT and Tg2576 mice. Dantrolene abolishes the reduction of PSD-95 expression observed in Tg2576 mice (1.1 ± 0.1, n=13, versus 0.5 ± 0.1 n=10 in dantrolene- versus vehicle-treated mice respectively) (Fig. 7D) and remain pharmacologically inert in WT mice (1.2 ± 0.1, n=9, versus 1.0 ± 0.1, n=13) (Fig. 6D). These data indicate that the restoration of normal PSD-95 levels by dantrolene parallels the reduction of Aβ burden observed in dantrolene-treated Tg2576 mice.
It was previously reported that Tg2576 mice harbor learning and memory deficits (Hsiao et al., 1996). We thus investigated the impact of dantrolene treatment on these two parameters by using two complementary tests: the Morris Water Maze (MWM) (Morris, 1984), which tests spatial learning memory, and the novel object recognition paradigm, which records recognition memory (Bevins and Besheer, 2006). In the MWM, WT and Tg2576 mice treated with vehicle or dantrolene have similar escape latencies to find visible platform (Fig. 8A), thus indicating that motility and vision are not affected in Tg2576 mice and that dantrolene treatment does not affect these parameters. Both WT and Tg2576 mice are also able to learn the MWM task, as the average escape latency for each group gradually decrease to reach a predetermined criterion (<25 sec average latency) during 5 days of hidden-platform training trials. However, vehicle-treated Tg2576 mice show significantly lower learning performance since they reach criterion on day 5, while vehicle-treated WT mice reach it on day 4 (escape latency on day 4 (sec) : 39.2 ± 2.7, n=11 for vehicle-treated Tg2576 mice, versus 24.6 ± 2.2, n=8 for vehicle-treated WT mice) (Fig. 8B). Analyses of the path length and of the path efficiency confirm these data, ((Path length on day 4 (m): 5 ± 0.5, n=11 for vehicle-treated Tg2576 mice, versus 2.5 ± 0.3, n=8 for vehicle-treated WT mice; Fig. 8C), and (path efficiency on day 4: 0.15 ± 0.03, n=11 for vehicle-treated Tg2576 mice, versus 0.29 ± 0.03, n=8 for vehicle--treated WT mice; Fig. 8D)). Importantly, dantrolene improves learning ability in Tg2576 mice as compared to vehicle-treated Tg2576 mice (escape latency on day 4 (sec): 21.5 ± 4.3, n=10, versus 39.2 ± 2.7, n=11 respectively) (Fig. 8B), (path length (m): 2.3 ± 0.4, n=10, versus 5.0 ± 0.5, n=11 respectively) (Fig. 8C), and (path efficiency: 0.33 ± 0.06, n=10, versus 0.15 ± 0.03, n=11 respectively) (Fig. 8D). Our data also reveal that dantrolene treatment does not affect learning ability in WT mice (Fig. 8B–D), and restores learning ability in Tg2576 mice to a statistically similar level to that observed in WT mice (p-value> 0.5). At the probe trial, no difference in the time spent in the target quadrant was found between dantrolene-treated Tg2576 mice and vehicle-treated Tg2576 mice (data not shown). Therefore, we also explored recognition memory using the novel object recognition (NOR) paradigm (Taglialatela et al., 2009). In this test, mice are less exposed to stress conditions as compared to the MWM test. During the set up of the NOR apparatus and training paradigm, we confirmed the absence of any artifactual preference for a specific object (color and form) between all groups of mice, and verified that Tg2576 mice were not anxious and did not harbor motility decline (data not shown). The total object exploration time during training session was not different in dantrolene- and vehicle-treated WT and Tg2576 mice (data not shown). After 24h retention, we performed a testing session where the sample objects were reintroduced, one being identical to the training object, i.e. the familiar object, the other being a novel object. Total exploration time during the testing session is not significantly different between dantrolene- and vehicle-treated WT and Tg2576 mice (data not shown). However, vehicle-treated Tg2576 mice show a clear reduction in the object discrimination ratio as compared to vehicle- and dantrolene treated WT mice (47.6 ± 5.7, n=11, versus 66.6 ± 4.9, n=11, and 55 ± 6.3, n=6 respectively) (Fig. 8E). Importantly, dantrolene treatment increased the object discrimination index as compared to vehicle-treated Tg2576 mice (72.7 ± 3.6, n= 10, versus 47.6 ± 5.7, n=11 respectively), thus reflecting an increase in the exploration time of the novel object versus the familiar object in dantrolene-treated Tg2576 mice (Fig. 8E). As for the MWM, in the NOR paradigm, we also reveal that dantrolene treatment restores the object discrimination index in Tg2576 mice to a statistically similar level to that observed in vehicle-treated WT mice (72.7 ± 3.9, n= 10, versus 66.6 ± 4.9, n=11 respectively, p-value > 0.5). These results demonstrate that dantrolene reduces both learning and memory decline in Tg2576 mice. All together, our data demonstrate that the blockade of RyR-mediated Ca2+ release by dantrolene simultaneously reduces Aβ load, prevents the loss of PSD-95 expression and prevents learning and memory deficits in vivo.
We report herein that WT or mutated APP overexpression triggers a large increase of cytosolic Ca2+ signals mainly linked to increased ER Ca2+ release and passive Ca2+ leakage (Fig. 1 and and2).2). Importantly, we reveal the implication of RyR in APP-associated Ca2+ alteration. Therefore, we show that RyR expression and RyR-mediated Ca2+ release are enhanced in both in vitro and in vivo APP-overexpressing models. We also reveal the participation of CICR through RyR in Ca2+ entry via VGCC in APPswe-expressing cells. Interestingly, the CICR-associated pathway was not observed in control cells, suggesting that the larger responses of VGCC in APPswe-expressing cells may arise principally from greater CICR through RyR. Exacerbated IP3R-evoked Ca2+ signals observed in APPswe-expressing cells may also be due to increased CICR through the RyR. This phenomenon was reported in two other AD mice models (PS1M146V and 3xTg-AD) (Stutzmann et al., 2006).
It is known that depletion of ER Ca2+ activates SOCE (Smyth et al., 2010). However, our data reveal that increased Ca2+ entry in APPswe-expressing cells cannot be accounted for by altered SOCE or ICRAC. We suggest that the elevated cytosolic [Ca2+] observed in APPswe-expressing cells is contributed by alternative mechanisms namely: 1) alteration of Ca2+ extrusion by PMCA and Na+/Ca2+ exchanger (NCX); 2) activation of NCX in Ca2+ influx/Na+ efflux mode; 3) reduction of the buffering capacity; or 4) as already reported exaggeration of Ca2+ entry through Aβ pore in the plasma membrane (Demuro et al., 2011).
The kinetics of Ca2+ slope after Ca2+ responses are not altered in APPswe-expressing cells as compared to control (Fig. 1 C–E); thus excluding an alteration of PMCA and NCX pumping function. We also did not see any evidence for a modification of Ca2+ entry upon addition of KB-R7943, an inhibitor of NCX reverse mode (Magi et al., 2005). This excludes the implication of NCX operating in Ca2+ influx/Na+ efflux mode in the observed increased Ca2+ entry (data not shown).
APPswe-expressing cells harbor an increased basal Ca2+ level as revealed in Figure 1G. Accordingly, elevated resting [Ca2+] was previously reported in APPswe-derived neurons (Lopez et al., 2007), and reduced expression of the calcium binding protein calbidin-D28K was also described in AD (Riascos et al., 2011). Moreover, it was recently demonstrated that Aβ oligomers aggregate into a Ca2+ permeable pore in the plasma membrane (Demuro et al., 2011). It is therefore tempting to speculate that elevated Ca2+ entry in APPswe-expressing cells may be a consequence of two mutually non-exclusive mechanisms: i) a constitutive Ca2+ entry through Aβ oligomers in the plasma membrane; and/or ii) reduced buffering capacity.
Alteration of ER Ca2+ homeostasis was reported in various AD models. Importantly, deviant RyR-mediated Ca2+ release and enhanced RyR expression were described in 3xTg-AD, and PS1M146V-expressing mice models (Chan et al., 2000; Smith et al., 2005; Stutzmann, 2007; Chakroborty et al., 2009). It was proposed that PS were the predominant Ca2+-deregulating factor in AD and that they may trigger RyR expression and activation in these models. Accordingly, it was recently demonstrated that PS1 and PS2 directly increase RyR single channel activity through protein-protein interaction (Hayrapetyan et al., 2008; Rybalchenko et al., 2008). We provide here evidence that enhanced RyR expression and RyR-mediated ER Ca2+ release occurred in an AD-related model independently of PS mutation or overexpression. Thus, our data reinforce the implication of ER Ca2+ homeostasis dysregulation in AD and pointed out RyR expression and/or function dysregulation as a common key player in AD “calciopathy”.
Elevated RyR levels have been described early in human AD cases and in mild cognitive impairment (Kelliher et al., 1999; Bruno et al., 2011). Accordingly, alterations of RyR expression and/or function were found to occur in 3xTg-AD mice model before Aβ formation, tau deposits, or memory deficits (Chakroborty et al., 2009). These data suggest that dysregulation of RyR may represent an etiological trigger that may contribute to the setting of histopathological lesions and synaptic deficits that are associated with the later disease stages. Our study reveals that alterations of RyR-dependent Ca2+ signals likely contribute to the progression of AD pathogenesis through the amplification of Aβ peptide production and memory decline. In these contexts, RyR emerges as a key factor that could be implicated in both initiation and progression of AD.
We show an induction of RyR1/2/3 isoforms in APP695- and APPswe-expressing SH-SY5Y cells. Upregulation of RyR2 isoform, but not of RyR1 and RyR3, was also observed in the cortex of Tg2576 mice. Similar results were reported in 3xTg-AD mice (Chakroborty et al., 2009). Therefore, we suggest that RyR2 upregulation may underlie the enhanced RyR-mediated Ca2+ release in Tg2576-derived neurons. Since the induction of different RyR isoforms at the mRNA and protein levels was described in distinct AD models and at different stages of AD pathology, it appears of most interest, to study the molecular pathway(s) underlying the control of RyR expression. In the context of APP-overexpressing models, we may speculate that RyR expression may be regulated by APP-intracellular domain fragment (AICD), a transcriptively active modulator (Pardossi-Piquard and Checler, 2012) which has already been described to modulate IP3-mediated Ca2+ signaling (Leissring et al., 2002).
Enhanced ER Ca2+ emptying in APPswe models may also be linked to pathophysiological post-translational modifications in the macromolecular complex containing RyR1 or RyR2 resulting in “leaky channel” (Marx et al., 2000; Bellinger et al., 2009; Gant et al., 2011). Specific experiments are needed to demonstrate if the increased ER passive Ca2+ leakage observed in APPswe-expressing cells (Fig. 2 B) is associated to a dysfunction in the RyR macromolecular complex. It was initially demonstrated that RyR1 and RyR3 are the targets of dantrolene (Zhao et al., 2001). Interference of dantrolene with cardiac and neuronal RyR2 isoform has been disputed, although it has recently been proposed to have effects on cardiac RyR2 (Jung et al., 2012; Maxwell et al., 2012). Our finding demonstrate the potential use of dantrolene as a tool to modulate RyR-mediated Ca2+ signals, however other approaches must be considered such as modulators of the RyR macromolecular complex (Bellinger et al., 2009).
It was previously reported that APP phosphorylation on Thr-668 residue is necessary for intra-neuronal accumulation of Aβ (Lee et al., 2003; Pierrot et al., 2006; Muresan and Muresan, 2007), and that the activity of Cdk5 and GSk3β kinases implicated in APP phosphorylation are calpain- and Ca2+-dependent (Nath et al., 2000; Lebel et al., 2009).
It is possible to envision the following scenario (Fig. 9): dantrolene, through the modulation of RyR-mediated Ca2+ release, reduces APP phosphorylation on Thr-668 residue likely through the control of Cdk5 and GSk3β kinases activities; in parallel, dantrolene also reduces β- and γ-secretases activities. This may occur directly since Ca2+ interacts with β- and γ-secretases to enhance their activities (Hayley et al., 2009; Ho et al., 2010), or indirectly through the control of Cdk5 and GSK3β activities. Therefore, dantrolene modulates in concert both APP phosphorylation on Thr-668 and β- and γ-secretases activities leading to the reduction of C99 and Aβ production likely preventing learning and memory decline (Fig. 9).
We show herein that Tg2576 mice harbor a reduced level of PSD-95 (a component of the post-synaptic density membrane associated guanylate kinase (PSD-MAGUK) scaffolding proteins). In line with these results, reduced PSD-MAGUKs expression, i.e. PSD-95 and SAP-102 were also reported in autopsied AD brains (Proctor et al., 2010). It is well established that PSD-MAGUK indirectly regulates synaptic plasticity and memory through the control of the number and compartmentalization of both (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and N-methyl-D-aspartate (NMDA) glutamate receptors around the PSD (Elias et al., 2008). In addition, Goussakov et al. showed a profound RyR-mediated Ca2+ increase within dendritic processes and spines and larger NMDA-evoked Ca2+ signals in the 3xTg-AD strain (Goussakov et al., 2010). We hypothesize that excessive post-synaptic RyR-mediated Ca2+ release and subsequent increased Aβ load may have contributed to PSD-95 expression decline in Tg2576 mice. This may have led directly or indirectly to learning and memory decline.
About 30 millions individuals are estimated to be affected with AD worldwide and to date no effective treatment exists to arrest disease progression. Therapeutic approaches targeting Ca2+ influx have demonstrated efficacy in animal AD models, very few have been successful in clinical trials, namely the L-type Ca2+ channel blocker nimodipine (Tollefson, 1990), and the NMDA open receptor blocker memantine (Bullock, 2006). Targeting of ER Ca2+ homeostasis as a therapeutic approach for AD was not investigated before. Dantrolene was originally used for the treatment of malignant hyperthermia (Harrison, 1975). However, recent in vitro and in vivo studies revealed the neuroprotective effect of dantrolene. Thus, dantrolene was shown to protect cells in vitro against the adverse consequences of the PS1 mutation (Guo et al., 1999), and to be neuroprotective in vivo in spinocerebellar ataxia type 2 and 3 and in Huntington's disease (Chen et al., 2008; Liu et al., 2009; Chen et al., 2011).
We provide here evidence that dantrolene treatment reduces Aβ burden in vitro and in vivo and prevents the reduction of PSD-95 expression and learning and memory decline in vivo. Our study reveal RyR as a potential target for the treatment of AD and paves the way for the development of therapeutic strategies for AD based on modulating ER-dependent Ca2+ release mechanisms.
This work was supported by INSERM, CNRS, AFM (11456 and 13291), «Fondation pour la Recherche Médicale» (DEQ20071210550) and, the Italian Institute of Technology. This work has been developed and supported through the LABEX (excellence laboratory, program investment for the future) DISTALZ (Development of Innovative Strategies for a Transdisciplinary approach to Alzheimer’s disease). We acknowledge grant support from the National Institutes of Health (5R01HL097111, to M.T.), “L’Ecole de l’INSERM Liliane Bettencourt” for supporting the MD-PhD curriculum of B.O. and the Italian Institute of Technology, Genova, Italy for supporting the PhD curriculum of D.D.P.