Endogenous Amyloid-formed Ca2+-permeable Channels in Aged 3xTg AD Mice

Abstract Alzheimer’s disease (AD), the leading cause of dementia, is characterized by the accumulation of beta-amyloid peptides (Aβ). However, whether Aβ itself is a key toxic agent in AD pathogenesis and the precise mechanism of Aβ-elicited neurotoxicity are still debated. Emerging evidence demonstrates that the Aβ channel/pore hypothesis could explain Aβ toxicity, because Aβ oligomers are able to disrupt membranes and cause edge-conductivity pores that may disrupt cell Ca2+ homeostasis and drive neurotoxicity in AD. However, all available data to support this hypothesis have been collected from “in vitro” experiments using high concentrations of exogenous Aβ. It is still unknown whether Aβ channels can be formed by endogenous Aβ in AD animal models. Here, we report an unexpected finding of the spontaneous Ca2+ oscillations in aged 3xTg AD mice but not in age-matched wild-type mice. These spontaneous Ca2+ oscillations are sensitive to extracellular Ca2+, ZnCl2, and the Aβ channel blocker Anle138b, suggesting that these spontaneous Ca2+ oscillations in aged 3xTg AD mice are mediated by endogenous Aβ-formed channels.


Introduction
Alzheimer's disease (AD), the leading cause of dementia, is characterized by the accumulation of beta-amyloid peptides (A β) in senile plaques in the brain of affected patients. 1 Many cellular mechanisms are thought to play important roles in the development and pr ogr ession of AD pathogenesis, but A β depositinduced toxicity is still considered to be one of, if not the most, important factors in the pathogenesis of AD. 2 Several lines of evidence support the A β channel/pore hypothesis stating that A β is a b le to disrupt membranes by causing pore formation. [3][4][5][6] Ther efor e, A β to xicity can be e xplained at least in part on the basis of dysregulation of Ca 2 + homeostasis by direct lipid disruption. 3 , 7 , 8 This hypothesis was based on microscopic conductance changes induced by A β pore formation in artificial membranes that wer e highl y complex and showed cation selectivity. 9 , 10 Recent evidence shows that A β, similar to gramicidin, caused micro and macro perforation of cellular membranes, leading to neurotoxicity by a Ca 2 + -dependent mechanism in cultured neurons. 7 , 8 , 11-13 The A β channel/pore shar es sev eral pr operties: heter o-dispersity, irr ev ersibility, nonselectivity, long open times, blockade by zinc, inhibition by Anle138b, and enhancement by "aging" or acidic pH. 14 These properties would lead to cells gradually becoming "leaky," leading to loss of ionic gradients, dysregulation of calcium homeostasis, and high consumption of energy supplies. Although n umer ous studies have laid the foundation for this hypothesis, the predominant evidence is indirect and limited to in vitro system tests. Ideally, if one could record ionic currents under patch-clamp conditions on a cell, isolated from relevant regions in AD animal models, after all the nati v e ion channels on such a hypothetical cell wer e b locked without altering the physiology, one could expect empirical evidence that would be direct, widel y acce pted and most useful to the understanding of AD. 15 Based on this type of experimental setup, we provide direct evidence of an endogenous A β channel/pore in aged 3xTg AD mice.
The pancreatic acinar cell is a classical cell model, for studies of Ca 2 + signal tr ansduction mec hanisms, because it has been possib le to dir ectl y obtain considera b le insight into intracellular Ca 2 + handling under both normal and pathological conditions. 16 Unlike nerve and endocrine, as well as muscle cells, exocrine cells such as pancreatic acinar cells are non-excitable cells and do not possess v olta ge-gated Ca 2 + channels, and the cytosolic Ca 2 + signals gov erning pancr eatic acinar secr etion ar e primarily generated by release of Ca 2 + from intracellular stores, principally the endoplasmic reticulum (ER). [17][18][19] For example, acetylcholine (ACh) can cause physiological cytosolic Ca 2 + oscillatory signals (monitored by changes in the Ca 2 + -acti v ated Cl − curr ent) thr ough the G pr otein-IP 3 pathw ay. Mor eov er, additional factors work together to keep the intracellular Ca 2 + homeostasis in acinar cells, such as the plasma membrane Ca 2 + -acti v ated ATPase, sarco(endo) plasmic reticulum Ca 2 + activated ATPase and storeoper ated c hannels.
As a highly innervated organ, the pancreas shares a number of striking pathophysiologic similarities in A β deposition comparing to that in brain 20 . AD can cause the A β deposition in the pancreases of mice ov er expr essing human APP, suggesting that A β deposits may occur in organs other than the brain. 20 Here , w e show that A β de posits can be detected in pancr eatic acinar cells in aged 3xTg AD model mice . Furthermore , unexpected spontaneous Ca 2 + oscillations were found in such AD mouse acinar cells. Since mouse pancreatic acinar cells do not expr ess v olta ge-gated Ca 2 + c hannels, w e hypothesize that the spontaneous Ca 2 + oscillations mediated through endogenous A β-formed Ca 2 + -permea b le channels in these a ged AD mice.

Materials and Methods
C57BL/6 J mice w ere pur c hased from Beijing Vital Ri v er La boratory Animal Technology company. 3xTg AD mice, 12-24-moold, were a gift from Prof. Diling Chen (Guangdong Institute of Microbiology, China), who has published a paper using 3xTg AD mice. 21 This AD mouse carries transgenes encoding mutants of presenilin-1 (PS1; M146V), amyloid precursor protein (APP; swe), and tau (P301L). These mutant genes lead to rapid accumulation of amyloid in mice and pr ematur e dysfunction of synaptic transmission and long-term potentiation. 22 All mice were kept in groups under constant standard conditions of temperature and humidity, with ad libitum access to food and water, and on a 12-h light/dark cycle. All experimental pr ocedur es using mice conform to the Medical Animal Care and Welfare Committee and were approved by the Laboratory Animal Ethics Committee of Shantou Uni v ersity Medical Colle ge .
Single isolated mouse pancreatic cells were prepared as pr eviousl y described. [23][24][25][26] Briefly, wild-type or 3xTg AD mice were anesthetized with isoflurane. The pancreas was quickly r emov ed and a small amount of collagenase solution was injected into the pancreas for digestion (150-200 U/mL, 18 min, 37 • C; Wako Pure Chemicals, Osaka, Japan). At the end of the collag enase dig estion, the cell suspension was g ently triturated with a pipette to further dissociate the cells, and then the cells were washed several times with standard external solution (composition listed below). Thereafter, 200 μL aliquots of the suspension w ere tr ansferred to a 35-mm culture dish containing 2 mL of standard external solution. Dissociated cells were typically used within 3 h after dissociation. All experiments were performed at 22 ± 1 • C.
As pr eviousl y r e ported, [23][24][25][26] patch-clamp conv entional whole-cell recordings were used to record Ca 2 + -activated Cl − currents to monitor intracellular Ca 2 + signaling oscillations. When the recording pipette was filled with K + -containing pipette solution, the resistance was 3-4 M . After a G seal was formed between the cell membrane and the pipette, the A β perforated whole-cell conformation was formed as judged by gradual (5-30 min) reduction of the access resistance (to < 60 M ). Whole-cell recording mode w as achiev ed by brief suction. The cells were held at −30 mV and the series r esistance w as not compensated. Tr ansmembr ane currents wer e r ecorded with a patch clamp amplifier (Axopatch 200B; Molecular Devices; Sunnyvale, CA, USA).
The standard extracellular solution contained (in m m ): 140 NaCl, 4.7 KCl, 1.2 MgCl 2 , 1 CaCl 2 , 1.13 MgCl 2 , 10 glucose, and 10 HEPES, pH 7.3 adjusted using Tris-base. A Ca 2 + -free solution was prepared by replacing Ca 2 + with Na + (142 m m NaCl and 0 m m Ca 2 + ) and adding 1 m m EGTA. The pipette solution for whole-cell recordings contained (in m m ): 140 KCl, 0.24 EGTA, 1.13 MgCl 2 , 5 Na 2 ATP, 10 glucose, and 10 HEPES, pH 7.2. ACh, 2APB, and atropine sulfate used in this study were purchased from Sigma (St. Louis, MO, USA). Ruthenium red was pur c hased from Wako Chemical, and CdCl 2 and ZnCl 2 were purchased from Macklin (Shanghai Macklin Biochemical Co., Ltd). For external drug application, a "U-tube" rapid application system was employed. Data wer e filter ed at 2 kHz, acquired at 5 kHz, and digitized online (Clampex 10.6 softw ar e, Digidata 1550B, Axon Instruments, Union City, CA, USA). All data were displayed and stored on a computer.

Thioflavin S Staining
To identify the A β accumulation, the thioflavin S staining was used, samples were prepared as described. 27 Briefly, 12mo-old 3xTg mice or wild type (WT) mice of the same age were anesthetized, and the brain and pancr eas wer e fixed by 4% paraformaldehyde perfusion through the heart. Paraffinembedded bloc ks w ere prepared by sequential dehydration in graded ethanol before embedding, and tissues were serially sectioned to a thickness of 4 μm and mounted on pre-coated Poly-llysine glass slides. Sections were first pre-incubated with potassium permanganate solution and oxalic acid, then placed in 3% sodium borohydride solution for 5 min. Staining was performed using filtered 0.05% Thioflavin S (Sigma) in 50% ethanol for 30 min in the dark, and differentiated with two changes of 80% ethanol for 10 s. This was followed by 3 washes with large volumes of distilled water and an incubation step in 5 × PBS buffer at 4 • C for 30 min. Finally, slides were briefly rinsed in PBS and cov er ed with cov erslips using Vectashield Hard Set mounting media with DAPI (Vector). Slides were allowed to set in the dark at 4 • C and imaged immediately thereafter. Fluor escence ima ges wer e acquir ed on a Zeiss LSM800 confocal microscope.

Prepar a tion of oA β 1-42
For the oligomerization of A β 1-42 (oA β), a 1 m m A β solution was pr e par ed by dissolving as-synthesized A β 1-42 powder (DGpeptides Co.) in 1,1,1,3,3,3-hexafluor oisopr opanol, and b low dr ying the liquid with nitrogen to form a peptide film at the bottom of the centrifuge tube. The A β 1-42 peptide membrane was resuspended by adding DMSO and sonicated under sterile conditions for 10 min before use, then diluted to a 100 μm stock solution in PBS at 4 • C. For the application of monomers, the 100 μm stork solution was directly diluted to the r equir ed concentration with standard extracellular solution. For the pr e paration of oligomers, the 100 μm stork solution was vortexed (15 s), centrifuged, and transferred to a 4 • C freezer for 24 h. To avoid fibril formation, all samples were used within one day.

Sta tistical Anal ysis
The net charge of the curr ent w as obtained by dividing the curr ent ar ea by the cell membrane capacitance (Cm) over a certain time (usually for 2 min). Before values were measured from the baseline of spontaneous Ca 2 + oscillations from 3xTg AD mice for appr oximatel y 2 min (normalized to 1) and wer e compar ed to the changes induced by drugs exposure. Data are presented as mean ± SEM. One-w ay ANOVA w as employed to compare the time to first calcium spike generation at different concentrations of oA β. Differences with P < 0.05 were considered significant.

Unexpected Spontaneous Ca 2 + Oscillations in Aged 3xTg AD Mice
Usually, in mouse pancreatic acinar cells, Ca 2 + oscillations can be induced by different stimulations, such as ACh, CCK, InsP3, or Ca 2 + ion. 28 However, by using patch-clamp whole-cell recordings in fr eshl y dissociated pancr eatic acinar cells fr om a ged 3xTg AD mice, we found unexpected spontaneous Ca 2 + oscillations ( Figure 1 A). In 16 aged (24-mo-old) 3xTg AD mice, we tested 34 acinar cells, and 23 (23/34) showed spontaneous Ca 2 + oscillations, whereas in 8 age-matched WT mice, there were no detecta b le spontaneous Ca 2 + oscillations in the 15 acinar cells tested ( Figure 1 B). In 5 WT cells (5/15), after whole-cell recording for 30 min (during which no responses were observed), 10 n m ACh was able to evoke Ca 2 + oscillations ( Figure 1 B), suggesting that both cell function and recording quality are good. Compared to both frequency and individual spike duration of ACh (10 n m )induced Ca 2 + oscillation responses ( Figure 1 Ba, 0.29 ± 0.07 Hz, n = 5 cells, and Figure 1 Bb, 1.08 ± 0.15 s, n = 5), the spontaneous Ca 2 + oscillations exhibited much lower oscillational frequency but longer oscillatory spike duration ( Figure 1 Ba, 0.03 ± 0.02 Hz, n = 7 cells and Figure 1 Bb, 2.67 ± 0.27 s, n = 7). Student's ttest analysis showed that the difference in both oscillatory frequencies and duration between the spontaneous Ca 2 + oscillations in pancreatic acinar cells fr om a ged 3xTg AD mice and the ACh (10 n m )-induced Ca 2 + oscillations in pancreatic acinar cells from age-matched WT mice wer e highl y significant ( P < 0.001).

Spontaneous Ca 2 + Oscillations Are Dependent on Extracellular Ca 2 + , and Sensiti v e to Zn 2 + and Anle138b
To explore the nature of these spontaneous Ca 2 + oscillations, we performed 3 experiments. F irst, w e r emov ed extracellular Ca 2 + by r e placing standard e xtracellular solution with an e xtracellular Ca 2 + -free solution containing 1 m m EGTA. In 8 cells tested, we found that the r emov al of extracellular Ca 2 + r ev ersib l y a bolished these oscillations ( Figure 2 A), suggesting that the spontaneous Ca 2 + oscillations ar e trigger ed by extracellular Ca 2 + influx into cells. Then, we found that a high concentration of ZnCl 2 (3 m m ) eliminated spontaneous Ca 2 + oscillations ( Figure 2 B, n = 7). F inally, w e tested the effects of an amyloid (A β) c hannel bloc ker, Anle138b, and found that Anle138b (100 n m ) completely inhibited the spontaneous Ca 2 + oscillations ( Figure 2 C, n = 11), suggesting that the spontaneous Ca 2 + oscillations in aged AD acinar cells are elicited by extracellular Ca 2 + efflux into cell through the A β-formed channels. Statistical analysis ( Figure 2 D) demonstrates that all 3 treatments (removal of extracellular Ca 2 + , ZnCl 2 , and Anle138b) induced reduction of spontaneous Ca 2 + oscillations are highly significant ( P < 0.0001).

Spontaneous Ca 2 + Oscillations Are Mediated Through Ca 2 + -induced Ca 2 + Release (CICR), Rather Than InsP 3 -induced Ca 2 + Release (IICR)
In these experiments, we further identified which intracellular Ca 2 + -signal pathways (CICR and/or IICR) mediated the spontaneous Ca 2 + oscillations. As shown in Figure 3 A, in 6 cells tested, the CICR blocker ruthenium red (10 μm ) inhibited the spontaneous Ca 2 + oscillations. To test whether the IICR participates in the spontaneous Ca 2 + oscillations, we performed 2 experiments. First, we tested the effects of 2APB, both InsP 3 r ece ptor and store-operated Ca 2 + channel blocker, and found that 100 μm 2APB failed to block the spontaneous Ca 2 + oscillations ( Figure 3 B, n = 7). Second, we acti v ated the IICR pathway, by application of ACh (10 n m ), and showed that the ACh-induced Ca 2 + oscillations can be blocked by 10 μm atropine ( Figure 4 A, n = 6), but cannot be blocked by 100 n m Anle138b ( Figure 4 B, n = 5). These results suggest that the spontaneous Ca 2 + oscillations in acinar cells from aged 3xTg AD mice are mediated by extracellular Ca 2 + influx into cytosol through the A β-formed channels, subsequently triggering the CICR signal pathway.

A β-like Peptides Accumulate in the Pancreas of Aged 3xTg AD Mice
It has been r e ported that APP is expressed in the normal mouse pancreas, and in the APP/PS1 mouse model of AD, APP is ov er expr essed within pancr eatic islets. 29 Ther e ar e also extensi v e amyloid deposits in the pancreas of 8-mo-old APP/PS1 transgenic mice. 30 In addition, transgenic mice ov er expr essing both amyloid beta-protein and perlecan in pancreatic acinar cells. 31 , 32 We detected A β deposition, in the pancreas of 3xTg AD mice, using thioflavin staining. Brain tissue (hippocampal sections) of 3xTg AD mice was used as a positi v e contr ol. Hippocampal sections of 12-mo-old 3xTg AD mice showed large numbers of A β de posits ( Figur e 5 B), which were not found in age-matched hippocampal sections from WT mice ( Figure 5 A). In pancreatic tissue sections, similar A β de posits wer e observ ed in 12-moold 3xTg AD mice ( Figure 5 D), but were not seen in WT mice ( Figure 5 C).

Exogenous A β Induces Spontaneous Ca 2 + Oscillations in Pancreatic Acinar Cells From Adult Mice
Data presented thus far indicate that pancreatic acinar cells of aged 3xTg AD mice display spontaneous Ca 2 + oscillations induced by extracellular Ca 2 + influx into cells through the A β channels likely formed by endogenous A β. To further confirm this, w e mimic ked A β c hannel formation by addition of e xo genous oA β to pancreatic acinar cells, which have no spontaneous Ca 2 + oscillations, from 6-mo-old WT mice ( Figure 6 C).
F irst, w e tested the ability of oA β (500 n m ) to induce Ca + oscillations by using perforated patch r ecording ( Figur e 6 A). After esta b lishment of a "giga seal" for about 5-30 min, the membrane series resistance gradually decreased to about 40 M , indicating formation of a perforated whole-cell recording configur ation ( F igure 6 Ba, left column). Based on our previous r e port, ketone pr ev ents oA β cell entr y in cultur ed hippocampal neurons. 33 Therefore , w e examined the effects of 1 m m ketone (pr etr eatment for 2 h or added sim ultaneousl y with oA β) on the time r equir ed for oA β channel formation. Either ketone co-application (oA β and ketone were added into peptide solution together) or ketone pr e-tr eatment (tr eatment of cells with ketone first for 2 h, then do patch recording with electrode that oA β and ketone were added into peptide solution together) significantly prolonged the formation of oA β channels, based on perforated whole-cell configuration measurements ( Figure 6 Ba). Furthermore, we examined the effects of 10 μm congo red on the time of oA β channel formation in the perfor ated whole-cell configur ation, and found similar prolongation of oA β channel formation ( Figure 6 Ba). Neither ketone nor congo red affected the recording conditions, including the series r esistance ( Figur e 6 Bb) and electr ode tip r esistance ( Figur e 6 Bc). Then, we tested whether bath-perfusion of 500 n m oA β could induce Ca 2 + oscillations like the spontaneous oscillations that we observed in aged APP mice.

Pharmacological Properties of Exogenous A β-induced Ca 2 + Oscillations in Pancreatic Acinar Cells From Adult Mice
To evaluate the nature of the e xo genous A β-induced Ca 2 + oscillations, we performed similar pharmacological tests as for the spontaneous Ca 2 + oscillations found from the aged 3xTg AD mice ( Figure 2 ). As shown in Figure 7 , bath-perfusion of e xo genous oA β (500 n m ) induced Ca 2 + oscillations ( Figure 7 A, n = 6), and the r emov al of extracellular Ca 2 + (by Ca 2 + -free extracellular solution containing 1 m m EGTA) r ev ersib l y stopped the oscillations ( Figure 7 B, n = 8). Like spontaneous Ca 2 + oscillations found in the pancreatic acinar cells from the aged 3XTg AD mice, the A β-induced Ca 2 + oscillations were sensitive to 3 m m ZnCl 2 ( Figure 1 C, n = 6) and 100 μm ruthenium red ( Figure 7 D, n = 6), but were not sensitive to 100 μm 2APB ( Figure 7 E, n = 6). In 6 cells tested, we found that bath-applied atropine did not affect the oA β-induced Ca 2 + oscillations ( Figure 7 F, n = 4). These results suggest that the oA β-induced Ca 2 + oscillations are mediated through the A β channels.

Comparison of Exogenous A β-induced Ca 2 + Oscillations and ACh-induced Ca 2 + Oscillations
In these experiments, we compared the oA β-induced Ca 2 + oscillations (see Figure 7 ) and the ACh-induced oscillations. Figure 8 shows that in acinar cells, from 6-mo-old WT mice, there were no detected oscillations during the 30 min recording, but bath application of ACh (10 n m ) could induce typical Ca 2 + oscillations ( Figure 8 A, n = 6). The ACh-induced Ca 2 + oscillations were sensiti v e to the m uscarinic r ece ptor b locker, atr opine ( Figur e 8 B, n = 4), but were not sensitive to either removal of extracellular Ca 2 + ( Figure 8 B, n = 6) or treatment with the A β channel blocker Anle138b ( Figure 8 C, n = 5). These results suggest that the trigger and pharmacology of ACh-induced Ca 2 + oscillations are different from oA β-induced Ca 2 + oscillations.

Discussion
The major and novel finding in this study is the unexpected spontaneous Ca 2 + oscillations in aged 3xTg AD mice but not in  age-matc hed WT mice . These spontaneous Ca 2 + oscillations ar e sensiti v e to extracellular Ca 2 + and high concentrations of ZnCl 2 , and mor e importantl y can be eliminated by Anle138b, an A β-formed c hannel bloc ker, suggesting that these spontaneous Ca 2 + oscillations in aged 3xTg AD mice are mediated by Ca 2 + influx through endogenous A β-formed channels. To confirm this, we mimicked A β-formed channels in fr eshl y isolated pancreatic acinar cells from adult (6-mo-old) WT mice, and used patch-clamp recording to show that oA β in the recording electrode (500 n m ) can perforate the membrane and form a pore, and bath-perfusion of e xo genous 500 n m oA βcan induce Ca 2 + oscillations that exhibit properties similar to the spontaneous Ca 2 + oscillations observed in aged 3xTg AD mice. To our knowledge, this is the first evidence of the   A β-formed channels by endogenous A β in aged 3xTg AD mice.
AD is a dementing, neurode gener ative disorder c har acterized by increased accumulation of A β, selecti v e de gener ation of forebr ain c holinergic neurons, and progressive deficits in learning and memory. It has been postulated that A β triggers cytotoxicity that in turn causes AD, but the mechanisms inv olv ed r emain elusi v e . The A β c hannel hypothesis has been postulated to explain A β toxicity, and states that A β is a b le to allostericall y assemb le into an ion channel structur e embedded in a membr ane lipid bilay er, resulting in membr ane leakage and unbalanced Ca 2 + homeostasis, leading to neuronal damage and death. 9 In 1992, Har d y and Higgins r e ported that A β can eliminate neuronal calcium homeostasis, thus making neurons more vulnera b le to environmental damage. 34 In 1993, Arispe et al. used electrophysiological techniques to show that A β1-40 can combine with the planar lipid bilayer to further form a cationselecti v e channel, thereby demonstrating the ability of A β to form channels and in turn induce an imbalance of Ca 2 + homeostasis, which may underlie A β neurotoxicity. 35 Although this hypothesis sounds v er y inter esting and is likel y an important mechanism of A β toxicity, the evidence supporting this hypothesis was collected using artificial membranes or cell culture at a nascent stage. 3 , 9 , 10 , 36-40 There are 2 major concerns for this hypothesis: (1) A β concentrations used to form the channels are too high and the lack of the evidence demonstrating such channels can be formed by endogenous A β, 15 and (2) the existence of A β channels has not been shown in AD animal models. These knowledge gaps significantly diminish enthusiasm for this hypothesis. In this study, we unexpectedly found the spontaneous Ca 2 + oscillations in pancreatic acinar cells isolated from aged 3xTg AD AD model mice, but not from agematc hed WT mice . This had surprised us because ther e ar e no classical v olta ge-gated Ca 2 + channels expr essed in pancr eatic acinar cells. Then, the question was how do these Ca 2 + ions enter the cell? Considering that the pancreas shares a number of striking pathophysiologic similarities in A β deposition compared to that in brain, and AD also can cause A β deposition in the mouse pancreas upon ov er expr ession of human APP, suggesting that A β deposits may occur in other organs than the brain. 20 Ther efor e, we hypothesized that A β channels may mediate these spontaneous Ca 2 + oscillations in aged 3xTg AD mice. We designed 4 experiments to test this hypothesis. First, we excluded the possibility that the spontaneous Ca 2 + oscillations wer e mediated thr ough classical ACh r ece ptor-G-pr otein-IP 3 -Ca 2 + (IICR) pathwa y b y showing that atropine had no effect on the spontaneous oscillations ( Figure 1 C). Then, we identified that the spontaneous Ca 2 + oscillations are mediated by extracellular Ca 2 + ion influx into cytosol, because the oscillations w ere quic kly limited by removal of extracellular Ca 2 + , and are triggered by CICR because the ryanodine receptor antagonist ruthenium r ed w as a b le to eliminate the oscillations ( Figure 1 D). Mor e importantl y, the A β channel b locker Anle138b completel y blocked the oscillations ( Figure 1 E). These results suggest that in aged 3xTg AD mice, there are A β depositions in the pancr eas ( Figur e 5 D), [29][30][31][32] wher e A β forms Ca 2 + -permea b le A β channels and causes spontaneous Ca 2 + oscillations. To further confirm this, w e mimic ked A β-formed channels in pancreatic acinar cells, dissociated from 6-mo-old adult mice, by bath-perfusion of oA β. Under whole-cell path-clamp recording conditions, these adult acinar cells did not show any spontaneous Ca 2 + oscillations before perfusion of oA β ( Figure 2 A), but after a 30-min bath perfusion of 500 n m oA β, Ca 2 + oscillations occurred ( Figure 2 B). The time course of oA β-induced Ca 2 + oscillations depended on the oA β concentration ( Figure 2 C). Importantly, the oA β-induced Ca 2 + oscillations exhibited the similar features as the spontaneous Ca 2 + oscillations of acinar cells from aged 3xTg AD mice, including the dependence of extracellular Ca 2 + ( Figure 2 D), sensitivity to ruthenium red ( Figure 2 E), and inhibition by Anle138b ( Figure 2 F).
To our knowledge, this is the first evidence that the A β formed channels from endogenous oA β in an aged AD model. A β accum ulates in pancr eatic tissues and pancr eatic acinar cells are not expressed in classical voltage-gated Ca 2 + channels, which gi v es us the ability to find the special phenomenon of spontaneous Ca 2 + oscillations. Although pancreatic acinar cells possess v er y selecti v e Ca 2 + r elease acti v ated Ca 2 + (CRAC) channels, and these ar e vitall y important for continuous Ca 2 + signaling as the intracellular (ER) store is not infinite. Direct electr ophysiological r ecordings of the CRAC curr ent in mouse pancreatic acinar cells have been reported, 41 in which, it is shown dir ectl y that the CRAC current is acutely (and r ev ersib l y) b locked by 100 μm 2APB, the exact same concentration that is shown in Figure 4 B, in which, 2APB did not block the spontaneous Ca 2 + oscillations in the 3xTg AD mice, str ongl y suggesting that the spontaneous Ca 2 + oscillations are not mediated through the CRAC channels.
The pancreatic acinar cell is a classical model for studies of Ca 2 + signal transduction mechanisms because it ena b les one to dir ectl y obtain considera b le insight into intracellular Ca 2 + handling under both normal and pathological conditions. 42 Unlike nerve and endocrine cells, as well as muscle cells, exocrine cells such as pancreatic acinar cells are non-excitable and do not possess v olta ge-gated Ca 2 + channels, and the cytosolic Ca 2 + signals gov erning pancr eatic acinar secr etion ar e primaril y generated by the release of Ca 2 + from intracellular stores, principally the ER 17,19,42 ( Figure 1 ). For example, ACh can cause physiological cytosolic Ca 2 + oscillatory signals through a G pr otein-IP 3 pathw ay ( Figur e 1 ). The incr ease in intracellular Ca 2 + generates Ca 2 + -induced Ca 2 + release by affecting the open-state pr oba bility of the IP 3 R or RyR channels. 43 As a highly innervated organ, pancr eas shar es a n umber of pathophysiologic similarities to A β deposition in the brain . 20 Our results show that A β deposits can be detected in both pancreatic acinar cells and hippocampal neurons in aged APP AD model mice ( Figure 5 ). Based on this finding, we expect that A β channels are also expressed in brain neurons, such as hippocampal and cortical neurons. Ther efor e, A β toxicity can be explained in part on the basis of dysregulation of Ca 2 + homeostasis by A β channels. 3 , 7 , 8 Recent evidence shows that A β, similar to gramicidin, causes micro and macro perforation in cellular membranes to induce neurotoxicity by a Ca 2 + -dependent mechanism in cultured neurons. 7 , 8 , 11-13 A β channels/por es shar e common pr operties of heter odispersity, irr ev ersibility, non-selecti vity, long open times, blockade by zinc, inhibition by congo red, and enhancement by "aging" or acidic pH. 14 These properties lead to cell leakage, decreases in ionic gr adients, dysre gulation of calcium, and consumption of energy supplies.
We show that A β deposits in pancreatic acinar cells can form Ca 2 + permea b le channels/por es in a ged 3xTg AD mice. Nota b l y, w e demonstr ate the ability of A β to perforate intact animal cell membranes, supporting the A β channel/pore hypothesis. Our results also show that A β-mediated disruption of Ca 2 + homeostasis in acinar cells increases intracellular Ca 2 + concentrations, which may be inv olv ed in A β toxicity. Recently, the failures of clinical trials associated with reducing A β deposition in mild AD patients raises the question of whether A β is the critical targ et for AD pathog enesis and treatment. Our findings suggest that A β may play an important role in triggering cell pathogenesis in the middle or late stages of AD. However, our results also suggest that pharmacological blockade of A β channels is a likely novel potential therapeutic strategy to impr ov e AD pathogenesis and learning and memory deficits. Collecti v el y, we identify an endogenous A β channel in an aged AD mouse model, providing new insight into the understanding of AD pathogenesis, and the intervention of AD pathological processes, as well as providing a potential treatment to improve AD cognitive deficits by targeting the A β channels.

Limitations of the Study
In this study, we unexpectedly found spontaneous Ca 2 + oscillations in pancreatic acinar cells isolated from aged 3xTg AD mice but not in age-matched WT mice. Since pancreatic acinar cells have no classical voltage-gated Ca 2 + channels, which gi v es us a chance to find the endogenous A β-formed channels in aged 3xTg AD mice. On the other hand, this cell pr e paration limits our finding dir ectl y linking A β-formed channels and cogniti v e deficits in AD model mice. We believe that it is hard to find endogenous A β-formed channels in neurons because neurons expr ess v arious types of v olta ge-gated Ca 2 + channels, wher eas pancreatic acinar cells have no suc h Ca 2 + c hannels. To overcome this limitation, we need to extend our study to hippocampal neurons and evaluate the effects of A β-formed Ca 2 + channel blockers (eg, Anle138b) on A β-induced neuronal toxicity and neurode gener ation, and also on cognitive deficits and learning and memory behavioral impairment in AD models. These experiments are ongoing.