Enhancing calmodulin binding to ryanodine receptor is crucial to limit neuronal cell loss in Alzheimer disease

Alzheimer’s disease (AD) is a neurodegenerative disorder characterized by progressive neuronal cell loss. Recently, dysregulation of intracellular Ca2+ homeostasis has been suggested as a common proximal cause of neural dysfunction in AD. Here, we investigated (1) the pathogenic role of destabilization of ryanodine receptor (RyR2) in endoplasmic reticulum (ER) upon development of AD phenotypes in AppNL-G-F mice, which harbor three familial AD mutations (Swedish, Beyreuther/Iberian, and Arctic), and (2) the therapeutic effect of enhanced calmodulin (CaM) binding to RyR2. In the neuronal cells from AppNL-G-F mice, CaM dissociation from RyR2 was associated with AD-related phenotypes, i.e. Aβ accumulation, TAU phosphorylation, ER stress, neuronal cell loss, and cognitive dysfunction. Surprisingly, either genetic (by V3599K substitution in RyR2) or pharmacological (by dantrolene) enhancement of CaM binding to RyR2 reversed almost completely the aforementioned AD-related phenotypes, except for Aβ accumulation. Thus, destabilization of RyR2 due to CaM dissociation is most likely an early and fundamental pathogenic mechanism involved in the development of AD. The discovery that neuronal cell loss can be fully prevented simply by stabilizing RyR2 sheds new light on the treatment of AD.


Results
App NL-G-F/NL-G-F mice (App NL-G-F mice), which harbor three familial AD mutations (Swedish: KM670/671NL, Beyreuther/Iberian: I716F, and Arctic: E693G) in the amyloid-β precursor protein, show cognitive dysfunction and accumulation of Aβ within several months 22 . We investigated whether enhancement of CaM binding to RyR2 influences development of AD, using App NL-G-F mice and RyR2 V3599K/V3599K mice (RyR2 V3599K ). RyR2 V3599K mice showed no apparent abnormalities. To examine whether stabilization of RyR2 through enhanced CaM binding is critical for AD pathogenesis, we generated App NL-G-F /RyR2 V3599K double homozygous mice (App NL-G-F /RyR2 V3599K mice), and evaluated how the AD phenotype is modified ( Supplementary Fig. 1). Consistent with a previous report 22 , spatial working memory was severely inhibited (Fig. 1A), with significant accumulation of Aβ in App NL-G-F mice (Fig. 1B,C, Supplementary Fig. 2). Interestingly, cognitive dysfunction was restored at 20 and 40 weeks in App NL-G-F /RyR2 V3599K mice, and at 40 weeks in dantrolene-treated App NL-G-F mice (Fig. 1A). There was hardly any accumulation of Aβ in the neural tissue until 40 weeks in WT and RyR2 V3599K mice, whereas Aβ markedly accumulated in App NL-G-F , dantrolene-treated App NL-G-F , and App NL-G-F /Ry R2 V3599K mice. There was no significant difference in Aβ levels at 20 and 40 weeks between App NL-G-F , dantrolenetreated App NL-G-F , and App NL-G-F /RYR2 V3599K mice (Fig. 1B,C). Of particular interest, significant neuronal cell loss occurred in App NL-G-F mice, whereas it did not occur in dantrolene-treated App NL-G-F and App NL-G-F /RYR2 V3599K mice (Fig. 1D). The neuronal cell loss was also confirmed by a decrease in NeuN in the isolated hippocampus of App NL-G-F , which was rescued by genetic gene substitution of RyR2 V3599K. Dantrolene also tended to rescue the neuronal cell loss in 40-week-old App NL-G-F mice (Fig. 1E). Long-term time course of spatial working memory, Aβ, and number of neuronal cells in WT, App NL-G-F , App NL-G-F + dantrolene (DAN), RYR2 V3599K , and App NL-G-F /RYR2 V3599K mice. (A) Time course of spatial working memory as aging. The Y-maze test was used for the evaluation of spatial working memory. Values for individual mice are plotted as mean ± SEM. N = 7-12 mice. Dotted line indicates standard deviation of 8-week-old WT mice. *p < 0.05 versus 8-week-old App NL-G-F mice (two-way ANOVA with post-hoc Dunnett test). + p < 0.05 vs. 24-week-old WT mice, † p < 0.05 versus 40 week-WT and -RYR2 V3599K mice, ##p < 0.01 vs. 40 week-old-App NL-G-F /RYR2 V3599K and -App NL-G-F + dantrolene (DAN) mice (one-way ANOVA with posthoc Tukey's multiple comparison test). (B) Amyloid beta (Aβ) in a cross section of hippocampus and the summarized data. Aβ was evaluated by immunocytochemistry using anti-Aβ. Aβ was expressed as an area (%) normalized by hippocampus area. Scale bars: 500 μm. N = 4-6 mice. Parentheses indicate the number of mice. The cross sections of whole brains are presented in Supplementary Fig. 2. ***p < 0.001 (one-way ANOVA with post-hoc Tukey's multiple comparison test). (C) The summarized data of Aβ evaluated by ELISA. each datum point reveals the average value of Aβ obtained form 20-week and 40-week mouse brains. N = 4-7 mice. Parentheses indicate the number of mice. ***p < 0.001 (one-way ANOVA with post-hoc Tukey's multiple comparison test). (D) NeuN stained cells and the summarized data. Neuronal cell density in a slice of Cornu Ammonis (CA) 1 and the dentate gyrus (DG) of the hippocampus (3 μm thickness) were measured by counting all NeuN stained cells devided CA 1 or DG area, and values for individual mice are plotted together with mean ± SEM. Scale bars: 100 μm. Parentheses indicate the number of mice. *p < 0.05, **p < 0.01, ***p < 0.001. (one-way ANOVA with post-hoc Tukey's multiple comparison test). E) Representative Western blots for NeuN in the isolated hippocampus, and the summarized data. Each datum point reveals the average value of NeuN obtained from 20-week and 40-week mouse brains. N = 4 mice. Parentheses indicate the number of mice. Fulllength blots are presented in Supplementary Fig. 10. *p < 0.05, ***p < 0.001 (one-way ANOVA with post-hoc Tukey's multiple comparison test). www.nature.com/scientificreports/ Next, we examined the protein expression of Glucose-Regulated Protein 78kD (GRP78), TAU, and the phosphorylation level of TAU (p-TAU) in whole brain homogenate. As shown in Supplementary Fig. 3, there was no significant difference in GRP78, TAU, p-TAU between all groups. These data are consistent with previously reported findings that no endoplasmic reticulum stress response was observed when using the same App NL-G-F mice 23 . So, we investigated the possibility that changes of ER stress markers and pTAU/TAU are limited to the neuronal cells. We examined the localization and protein expression of ER stress markers {GRP78, Activating Transcription Factor 6 (ATF6), C/EBP-homologous protein (CHOP)} and p-TAU/TAU in cultured neuronal cells at 20 weeks of age. Expressions of GRP78, ATF6 and CHOP increased in App NL-G-F neuronal cells compared to WT and RYR2 V3599K neuronal cells, but these expressions were restored to normal in App NL-G-F / RYR2 V3599K neuronal cells ( Fig. 2A-C). In App NL-G-F neuronal cells, GRP78 was expressed in the cytosol, but ATF6 and CHOP were expressed in the nucleus ( Fig. 2A-C). There was no difference in the expression level of TAU between WT, App NL-G-F , RYR2 V3599K , and App NL-G-F /RYR2 V3599K neuronal cells. However, the phosphorylation level of TAU (p-TAU) increased dramatically in App NL-G-F neuronal cells compared to neuronal cells from the other mouse strains (Fig. 2D).
At 20 weeks of age when cognitive dysfunction occurred, which is associated with Aβ accumulation and TAU phosphorylation in App NL-G-F mice, we measured the localization and expression level of RyR2 and CaM in isolated neuronal cells. CaM was largely co-localized with RyR2 especially around the nucleus where ER is known to be enriched (Fig. 3A). In App NL-G-F neuronal cells, CaM around the nucleus was dissociated from RyR2, whereas co-localization was restored in App NL-G-F /RYR2 V3599K neuronal cells (Fig. 3A). Direct association of exogenous CaM with RyR2 was also evaluated by binding of exogenous CaM cross-linked with sulfosuccinimidyl-6-[4′azido-2′-nitrophenylamino]hexanoate (Sulfo-SANPAH) to RyR2 11 . The apparent binding affinity of exogenous CaM to RyR2 decreased in App NL-G-F neural tissue, whereas it was restored in App NL-G-F /RYR2 V3599K neural tissue (Fig. 3B). The finding that cognitive dysfunction was ameliorated along with attenuation of TAU phosphorylation, but without decrease in Aβ in APP NL-G-F /RYR2 V3599K mice, strongly suggests that CaM dissociation from RyR2 play a critical role in cognitive dysfunction during AD development. On the other hand, at 8,20 and 40 W, there was no significant difference in RyR2 expression in brain homogenates between all groups ( Supplementary  Fig. 4). We also measured the changes in intracellular Ca 2+ before and after addition of ionomycin (10 μM) to neuronal cells isolated from WT, RyR2 V3599K , App NL-G-F , and App NL-G-F/RYR2V3599K mice brain tissue at 20 weeks of age ( Supplementary Fig.5). The time to peak was longer in App NL-G-F neuronal cells, suggesting a decrease in ER Ca 2+ content in App NL-G-F neuronal cells.
To further investigate how destabilization of RyR2 is linked with TAU phosphorylation and ER stress in neuronal cells, we examined the effect of tunicamycin (TM), a well-known ER stress inducer 24,25 , on baseline cytosolic [Ca 2+ ], response to ionomycin (10 μM), TAU phosphorylation, and ER stress in isolated neuronal cells. After addition of TM (3 μM) for 50 min, baseline cytosolic [Ca 2+ ] increased in WT neuronal cells, but not in the presence of dantrolene. Of interest, baseline cytosolic [Ca 2+ ] was not elevated after adding TM in RYR2 V3599K neuronal cells (Fig. 4A,B). Furthermore, in the presence of TM, ionomycin reduced the peak cytosolic [Ca 2+ ] and slowed the rate of increase (slope) and the amount of increase in cytosolic Ca 2+ (Δ[Ca 2+ ]), suggesting a decrease in ER Ca 2+ content in WT neuronal cells. However, this was not the case with WT neuronal cells treated with dantrolene or RYR2 V3599K neuronal cells.
The amount of CaM co-localized with RyR2 decreased after addition of TM in WT neuronal cells, whereas co-localization was restored in dantrolene-treated neuronal cells and in RYR2 V3599K neuronal cells (Fig. 4C). The apparent binding affinity of exogenous CaM to RyR2 also decreased after addition of TM in WT neuronal tissue, but not in dantrolene-treated neuronal tissue and RYR2 V3599K neuronal tissue (Fig. 4D).
Since decrease in ER Ca 2+ content has been reported to increase ER stress, which inhibits clearance of misfolded proteins 26 , we examined the effect of TM on localization and protein expression of GRP78, ATF6, and CHOP in isolated neuronal cells. GRP78, ATF6 and CHOP all increased 24 h after addition of TM in WT neuronal cells, whereas these changes were not observed in dantrolene-treated WT neuronal cells and RYR2 V3599K neuronal cells ( Fig. 5A-C). GRP78 was expressed in the cytosol after addition of TM in WT neuronal cells, while both ATF6 and CHOP were expressed in the nucleus. p-TAU increased after addition of TM in WT neuronal cells, but not in dantrolene-treated WT neuronal cells and in RYR2 V3599K neuronal cells (Fig. 5D). These findings are consistent with our results in App NL-G-F neuronal cells. Since TM has been previously shown to diminish the antioxidant capacity via inhibition of reduced glutathione (GSH) synthesis in liver cells 25 , we assessed whether TM may increase oxidative stress level in isolated neuronal cells ( Supplementary Fig. 6). Reactive oxygen species (ROS), evaluated by DCFH-DA fluorescence, increased similarly after addition of TM in both WT and RYR2 V3599K neuronal cells, and dantrolene did not reduce ROS, suggesting that oxidative stress is upstream from CaM dissociation from RyR2 ( Supplementary Fig. 6). In support of this notion, we previously reported that oxidative stress directly induces Ca 2+ leakage from cardiac RyR2 owing to domain unzipping between the N-terminal and central domains and subsequent CaM dissociation from RyR2 27,28 .
TM inhibits the DPAGT1 enzyme to inhibit one of the first steps of glycoprotein biosynthesis in the ER which results in the accumulation of misfolded proteins to cause subsequent ER stress 29 . Therefore, a mechanism that increases ER stress via DPAGT1 is also conceivable, although it is not known whether DPAGT1 dysfunction is involved in abnormal Ca 2+ processing in neurons.
We then evaluated whether chronic accumulation of Aβ increases ROS in isolated neuronal cells. Compared to WT or RyR2 V3599K mice, ROS markedly increased in App NL-G-F mice ( Supplementary Fig. 7). Of interest, ROS increased similarly in App NL-G-F /RYR2 V3599K mice as well, again suggesting that the increase in ROS is more upstream than Ca 2+ leakage during AD development ( Supplementary Fig. 7).
Collectively, we showed that accumulation of Aβ in genetic mouse models leads first in increased oxidative stress, then in dissociation of CaM from RyR2, in Ca 2+ leakage from ER, and in decreased ER Ca 2+ content, which promotes ER stress response, along with TAU phosphorylation. Pharmacological (by dantrolene) or

Discussion
In this study, we found that RyR2 destabilization due to CaM dissociation plays an essential role in the decrease in ER Ca 2+ content, increase in ER stress and p-TAU, subsequently leading to neuronal cell loss and cognitive dysfunction in App NL-G-F mice, and that either genetic (by V3599K substitution in RyR2) or pharmacological (by dantrolene) inhibition of CaM dissociation from RyR2 protected from AD ( Supplementary Fig. 8).
The amyloid cascade hypothesis has been widely recognized as a possible pathogenic mechanism of AD 2 . Currently, much effort has been made to reduce amyloid production or eliminate amyloid from the neural tissue. Unfortunately, most promising drugs failed to show any potential benefit for patients in clinical trials 5 , due to multiple reasons. First, accumulation of Aβ in neuronal cells may be only one of the possible factors that affect the pathogenesis of AD 2 . Second, inhibition of secretase activity to restrict toxic amyloid production may be harmful due to the many substrates that these enzymes target thus causing pleiotropic effects. For instance, γ-secretase modulates Notch signaling that plays a critical role in cell differentiation 30    www.nature.com/scientificreports/ and cognitive dysfunction were ameliorated without decreasing Aβ in chronically dantrolene-treated App NL-G-F mice and in RYR2 V3599K mice clearly indicates that accumulation of Aβ triggers initiation of an abnormal process leading to AD, but it is not necessary and sufficient for the pathogenesis of AD. Hence obviously other mechanisms need to be considered, particularly regarding AD treatment. Defective Ca 2+ homeostasis is another plausible mechanism involved in the pathogenesis of AD 5,6 . In support of this hypothesis, it has been reported that amyloid accumulation is linked with elevation of cytosolic Ca 2+ in neurons 31 , and that dysregulated ER Ca 2+ release alters synaptic transmission and plasticity mechanisms in an AD-type mouse model before the onset of histopathology and cognitive deficits 32 . Our study strongly suggests that displacement of CaM from RyR2 and subsequent ER Ca 2+ leakage increases ER stress in neuronal cells, thereby causing neuronal cell loss and cognitive dysfunction. Of interest, leaked Ca 2+ and displaced CaM induce activation of both Ca 2+ /calmodulin-dependent protein kinase II (CaMKII) and calcineurin (CaN) signals, both of which play important roles in the regulation of cognitive function 33,34 . CaN activates additional phosphatases, such as protein phosphatase 1, which further induces long-term depression that erases memories 35 . CaMKII has also been suggested to be a TAU kinase 36 . Hyper-activation of CaMKII may increase TAU phosphorylation; therefore, CaMKII is implicated in AD pathogenesis. Activation of CaN by CaM also disrupts the phosphatases interaction with TAU, possibly leading to TAU hyperphosphorylation 37 . Moreover, it has been recently shown that dantrolene improved cognitive dysfunction in association with decreased accumulation of amyloid [38][39][40][41] , thus linking defective Ca 2+ homeostasis with AD, although the precise mechanism has not been clarified yet.
In the cardiac muscle tissue, we previously reported that dantrolene specifically binds to amino acids 601-620 of RyR2 (corresponding with amino acids 590-609 in RyR1 to which dantrolene has been reported to bind 42 ), and restores defective N-terminal and central domain-domain interactions (domain unzipping), prevents dissociation of CaM from RyR2, suppresses aberrant Ca 2+ release, and thus inhibits lethal arrhythmia 43 and progression of heart failure 15 . Since RyR2 is abundantly expressed in the neural tissue, we anticipated that dantrolene also contributes to stabilization of RyR2 in AD-type diseased neural tissue. As expected, we observed that dantrolene inhibited ER Ca 2+ depletion by enhancing CaM binding affinity to RyR2 in TM-induced AD-type cells, thereby reducing ER stress and TAU phosphorylation. In contrast to the aforementioned reports, including ours, that showed a beneficial effect of dantrolene on AD prevention, one report showed that dantrolene rather promoted cognitive dysfunction, associated with further accumulation of amyloid 44 . There is no clear reason to account for the contradictory findings between these studies, although the experimental models used, the dosage, route, period, or timing of dantrolene administration can be speculated as influential factors. In our study, we did not observe any abnormal accumulation of Aβ or harmful effect after administration of dantrolene, although mice were fed with more than 20 times higher dose (100 mg/Kg/day) than the mice in Zhang, et al. ' s study (5 mg/ kg/day p.o., twice a week).
Although the effect of dantrolene on RyR2 channel function is specific 15 , we cannot entirely exclude possible off-target effects of dantrolene. In order to determine the critical role of RyR2 dysregulation in the pathogenesis of AD, a genetic approach is essential. In this study, we clearly demonstrated that when binding of CaM to RyR2 was stabilized pharmacologically in TM-induced AD-type cells or genetically in neural cells from App NL-G-F mice, most AD-related phenotypes (i.e. ER stress, neuronal cell loss, and cognitive dysfunction), except for Aβ accumulation were rescued. Recently TAU phosphorylation, rather than amyloid accumulation, has been recognized as an important determinant in the pathogenesis of developing AD 45 . Our finding that TAU phosphorylation, but not Aβ accumulation, was associated with improvement of cognitive function supports this idea. However, the fact that stabilization of RyR2 through increased binding affinity to CaM almost completely inhibited TAU phosphorylation in either TM-induced AD-type cells or neural cells from App NL-G-F mice strongly suggests that destabilization of RyR2 is an upstream of TAU phosphorylation in the pathogenesis of AD ( Supplementary  Fig. 8).
The mechanism by which the V3599K mutation prevents the decrease in CaM binding affinity for RyR2 in App NL-G-F mice remains unclear. Since the RyR2 V3599K mutation enhances the binding affinity of CaM to RyR2 in App NL-G-F mice or only when WT mice is subjected to TM (Figs. 3B, 4D), the accessibility of CaM near the V3599K mutation site of RyR2 can be allosterically enhanced only if there is a defective inter-subunit interaction in the tetrameric structure of RyR2. Recent higher-order structural analysis of RyR2 supports this idea. Namely, since CaM interacts closely with three domains of RyR2 (helix α − 1, helix 2b, helix α − 9) 46 , the accessibility of in situ CaM to its binding site may not be easily enhanced by the V3599K mutation. In contrast, the defective inter-subunit interaction due to ER stress may allosterically change the conformational state of the CaM binding region, facilitating CaM binding to the V3599K mutant (not WT) domain in helix α − 1. Of interest, dantrolene binds to a specific site (601-620 a.a.) near both the zipping interface and CaM, supporting the idea that dantrolene prevents CaM dissociation by stabilizing inter-subunit interaction (Supplementary Fig. 8).
There are several limitations in this study. First, the isolation process can affect the function of neuronal cells differently depending on the disease of each mouse model. Second, it is not clear whether the dissociation of CaM from RyR2 induced by endoplasmic reticulum stress preferentially targets neurons or also affects astrocyte / glial / non-neuronal cell types.
In conclusion, CaM dissociation from RyR2 plays a crucial role in the pathogenesis of AD, and enhancing CaM binding to RyR2 may be a novel, potent therapeutic strategy against AD. Especially, the discovery that neuronal cell loss can be fully prevented simply by stabilizing RyR2 would shed new light on the treatment of AD.

Methods
Chemicals. Dantrolene    were made by UNITECH Co., Ltd (Chiba, Japan). App NL-G-F mice were from RIKEN institute. All in vivo experiments were performed in contemporary random order. All strains were maintained on a C57BL/6 background. Measurement and analysis of the acquired data were performed by the investigators without knowing any genetic information of the mice.
Dantrolene administration. Dantrolene-treated App NL-G-F mice were fed, using a feeding apparatus (Roden CAFÉ; Oriental Yeast Co., Ltd., Tokyo, Japan), with 100 mg/kg/day dantrolene (Fuji Film Wako Chemicals, Tokyo, Japan) for 36 weeks (from 4-week-old to 40-week-old). The oral dose of dantrolene for chronic administration was determined as the dose by which inducible ventricular tachycardia was almost completely inhibited in CPVT-type RyR2R2474S+/− knock-in mice 9,47 . To avoid excessive moisture of dantrolene, food including dantrolene was changed at least every three days. Body weight was monitored during chronic administration of dantrolene; in all mice, body weight increased similarly regardless of dantrolene administration. The weight of the food consumed by every mouse was monitored and the concentration of dantrolene mixed in the food was adjusted at 100 mg/kg/day. In most cases, concentration of dantrolene in the food was 0.25%. No animals exhibited signs of toxicity and no mortality was detected in any group.
Y-maze test. The Y-maze test was used for the evaluation of spatial working memory, at 8-, 16 Polyornithine coating of culture dishes. Plastic culture dishes were coated with l-polyornithine prior to use as described by Eide et al. 48 . Each 35-mm plate was incubated overnight in 2.0 mL (or 0.5 mL for 24-dish plate) of 0.5 mg/mL borate buffer (10 mM Na2B4O7, pH 8.4). After incubation, plates were washes twice with water.
Isolation of neurons. Primary neuronal cultures were obtained from 20-week-old mice as described by Eide et al. 48 . Mice were anesthetized with overdose pentobarbital. The whole brain was isolated under sterile conditions, rinsed in HBS, and minced into small pieces. The minced tissue (approximately 0.5 mL) was then treated with 1 mL papain for 15 min at 37 °C. After gently leaving the solution, the papain/HBS supernatant was removed and replaced with 12 mL of prewarmed plating medium (2.0 mL/35-mm dishes), after which the tissue was immediately triturated. Approximately 2 mL of this cell solution was added to each 35-mm L-polyornithinecoated dish. Cells were allowed to settle for 15 min, washed once, and replaced with prewarmed plating medium (2.0 mL/35-mm dish). After plating, cells were incubated at 37 °C in 5% CO2 and incubated for 7 days without media replacement. On day 7, the plating media was replaced with the same volume of feeding medium and used for experiments.   Fig. 9). The absolute concentration of Ca 2+ was calculated by the following formula.
Immunocytochemistry analysis. Antibodies against the following proteins were used for immunohis- Preparation of brain homogenate. Whole brain homogenate prepared from 20-week-old mice was obtained according to Baghirova et al. 50 . Briefly, frozen mouse tissue was thawed and then minced into 2-4 mm pieces and washed with 1 mL ice-cold PBS. Then, 40-60 mg tissue was added into 500 μL ice-cold lysis buffer A. The tissue was disrupted in a tissue homogenizer and was centrifuged at 500×g for 10 min at 4 °C. The pellet was resuspended in lysis buffer A, followed by centrifugation at 4000×g for 10 min at 4 °C. The pellet was resuspended in lysis buffer B, followed by centrifugation at 6000×g for 10 min at 4 °C. Analysis of the binding characteristics of CaM to RyR2 with the CaM-SANPAH crosslinking method. Binding of CaM to RyR2 was evaluated using the photoreactive cross-linker, sulfosuccinimidyl-6-[4′-azido-2′-nitrophenylamino]hexanoate (Sulfo-SANPAH, Thermo Fisher Scientific, Waltham, USA), as described previously 11 . First, we made a CaM-SANPAH conjugate by mixing 50 μM recombinant CaM in conjugation buffer (150 mM KCl and 20 mM MOPS at pH 7.2) with 100 μM sulfo-SANPAH in the dark for 30 min. Conjugation was quenched by adding excess amount of lysine. CaM-SANPAH conjugate was purified using Amicon Ultra (MWCO 10 k). Mouse brain homogenates were diluted in binding buffer (150 mM KCl, 10 μM CaCl 2 , and 20 mM MES at pH 6.8) to 1 mg/mL and mixed with 100 nM CaM-SANPAH conjugate in the dark in a glass tube with or without CaMBPs. After a 10-min binding time, 30 s UV crosslinking was performed. Then, sample buffer was added to the crosslinked SR membrane followed by western blotting with anti-CaM (Merck, Millipore, Darmstadt, Germany). CaM-SANPAH crosslinked to RyR2 was detected as a 550 kDa band.
Tissue fixation and preparation of paraffin sections. Brain hemispheres prepared from 20-and 40-week-old mice were fixed by immersion in 4% paraformaldehyde in phosphate buffer solution (Nacalai tesque, Kyoto, Japan). Fixed brains were cut every 1.5 mm and embedded in paraffin, and 4 µm thick sections were mounted onto glass slides. Sections were stained with hematoxylin and eosin (H&E) or were used for immunofluorescence.
Immunohistofluorescence of Aβ and NeuN. As described by Wu et al. 41 , each brain of 20-and 40-weekold mice was fixed in 4% paraformaldehyde overnight at 4 °C prior to paraffin embedding. Coronal brain sec- www.nature.com/scientificreports/ tions of 10 μm thickness were deparaffinized and hydrated through a series of graded alcohol steps and washed in PBS. Antigen retrieval was performed by heating in Antigen Unmasking Solution (Vector Labs, Burlingame, USA) in a decloaking chamber (Biocare Medical, Concord, USA) for 2 min. Sections were incubated in 5% hydrogen peroxide to block endogenous peroxidase activity. Sections were incubated overnight at 4 °C with the anti-β-amyloid 6E10 (1/600, Covance, Dedham, USA) and anti-NeuN EPR12763 (1/1500, Abcam, Cambridge, UK) in 1% BSA and 0.5% Triton X-100, followed by labeling with an Alexa 488-conjugated secondary antibody (1/300). The sections were washed three times with PBS.
Neuronal cell density. NeuN immunostained, 3 μm thick brain slices were observed with a fluorescent digital microscope (BZ9000, Keyence, Japan, Plan 20 × Objective, Nikon, Japan), and all NeuN positive cells in the DG and CA areas of the entire hippocampus were counted and divided by the area of each region to calculate cell density.
Western blot for NeuN. Hippocampal homogenates prepared from 20 and 40 week old mice were obtained, as described by Baghirova et al. 50 . Briefly, frozen mouse tissue was thawed and then minced into 2-4 mm slices and washed with 1 mL ice-cold PBS. Then, hippocampal tissue was obtained added into 500 μL ice-cold lysis buffer A supplemented with digitonin and protease inhibitor cocktail. The tissue was disrupted in a tissue homogenizer and was centrifuged at 500×g for 10 min at 4 °C. The pellet was resuspended in lysis buffer A followed by centrifugation at 4000×g for 10 min at 4 °C. The pellet was resuspended in lysis buffer B containing igepal followed by centrifugation at 6000×g for 10 min at 4 °C. The pellet was denatured in SDSPAGE sample buffer. SDS-PAGE, blotting, and antibody detections were performed with anti-NeuN EPR12763 (1/1500, Abcam, Cambridge, UK).
Quantification of Aβ1-42 in mouse brain. Total proteins from whole brain of mice were extracted with lysis buffer. Then, Aβ1-42 was measured by an ELISA kit (WAKO-Fujifilm, Japan). In brief, protein solution diluted in PBS was placed into a 96-well micro plate coated with anti-Aβ1-16 antibody and incubated at 4 C overnight. After washing, the sample was incubated for 1 h in 100 μl of horseradish peroxide-conjugated antihuman Aβ1-42 antibody. The sandwiched Aβ1-42 was visualized with a TMB solution.
Statistics. Student t tests were used for statistical comparisons between two different conditions, whereas one-way or two-way ANOVA with a post-hoc Tukey's or Dunnett test was used for statistical comparison of more than two groups. All data were expressed as means ± SE. A probability value of less than 0.05 was considered statistically significant.