Alzheimer's disease-associated peptide Aβ42 mobilizes ER Ca2+ via InsP3R-dependent and -independent mechanisms

Dysregulation of Ca2+ homeostasis is considered to contribute to the toxic action of the Alzheimer's disease (AD)-associated amyloid-β-peptide (Aβ). Ca2+ fluxes across the plasma membrane and release from intracellular stores have both been reported to underlie the Ca2+ fluxes induced by Aβ42. Here, we investigated the contribution of Ca2+ release from the endoplasmic reticulum (ER) to the effects of Aβ42 upon Ca2+ homeostasis and the mechanism by which Aβ42 elicited these effects. Consistent with previous reports, application of soluble oligomeric forms of Aβ42 induced an elevation in intracellular Ca2+. The Aβ42-stimulated Ca2+ signals persisted in the absence of extracellular Ca2+ indicating a significant contribution of Ca2+ release from the ER Ca2+ store to the generation of these signals. Moreover, inositol 1,4,5-trisphosphate (InsP3) signaling contributed to Aβ42-stimulated Ca2+ release. The Ca2+ mobilizing effect of Aβ42 was also observed when applied to permeabilized cells deficient in InsP3 receptors, revealing an additional direct effect of Aβ42 upon the ER, and a mechanism for induction of toxicity by intracellular Aβ42.


INTRODUCTION
Alzheimer's disease (AD) is a progressive and irreversible brain disorder, which results in severe memory loss, behavioral as well as personality changes and a decline in cognitive abilities. While the most common type of AD remains idiopathic in origin, with age the most significant risk factor for disease onset (sporadic AD, sAD), ∼5% of cases show a Mendelian pattern of inheritance (familial AD, fAD). The amyloid β-peptide (Aβ) is hypothesized to be central to the pathogenesis of both sporadic and familial AD (Hardy and Selkoe, 2002). Aβ is a small, hydrophobic polypeptide, consisting of 39-42 amino acid residues, which occurs principally as a 40 or 42 amino acid peptide, Aβ 40 and Aβ 42 , respectively (Zhang et al., 2011). An imbalance between the production and clearance of Aβ, as occurs in fAD and sAD, respectively, leads to the accumulation of Aβ and, in turn, to its aggregation. This aggregation process represents a critical step in the pathogenesis of AD because the neurotoxic properties of Aβ are associated only with aggregated forms of the peptide (Kuperstein et al., 2010). Protein aggregation is highly dynamic and involves a wide range of intermediate structures such as oligomers, comprising dimers, trimers, dodecamers, and higher-molecular weight complexes, before aggregating into protofibrils and finally into mature amyloid fibrils (Dobson, 2003).
In this study, we investigated (1) the contribution of Ca 2+ mobilization from the ER to the increase in intracellular Ca 2+ induced by oligomeric Aβ 42 , (2) the mechanism (s) by which Aβ 42 elicited this effect, (3) the capacity for Aβ 42 to mobilize Ca 2+ directly from the ER. To allow isolation of effects on the ER from other plasma membrane targets of Aβ 42 , model cells systems were used that allowed fundamental aspects of ER Ca 2+ regulation to be studied. We determined that Ca 2+ release from the intracellular ER substantially contributed to the increase in intracellular Ca 2+ concentration induced by oligomeric Aβ 42 . The Aβ 42 -induced Ca 2+ elevation comprised InsP 3 dependent and independent components. Using DT40 cells deficient in the three InsP 3 R isoforms that were permeabilized to allow direct access of Aβ 42 to the ER, we also demonstrated that it had the capacity to release Ca 2+ from the ER independent of InsP 3 Rs. Together, these data place the ER and Ca 2+ released from it as central to the actions of both extracellular Aβ and Aβ that has reached an intracellular location.

MATERIALS
Peptides were purchased from The American Peptide Company and rPeptide. Cell culture reagents and chemicals were from Invitrogen or Sigma, unless otherwise stated.

PREPARATION OF Aβ 42 OLIGOMERS
Wild type and scrambled Aβ 42 were obtained at a purity of >95%. Peptide mass was verified by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry and peptides from the same batch were used throughout. Samples of synthetic Aβ 42 oligomers were prepared as previously described (Demuro et al., 2005) and remained stable for at least 3 weeks. Samples of Aβ 1-42 scrambled peptide (KVKGLIDGAHIGDLVYEFMDSN SAIFREGVGAGHVHVAQVEF) were prepared in the same way as Aβ 42 oligomers. All Aβ samples were stored at 4 • C and were used within 10-15 days of preparation. Toxicity of Aβ 42 preparations was confirmed by MTT assay before use in Ca 2+ imaging experiments ( Figure S1A). The oligomeric nature of the Aβ 42 preparation was established by surface plasmon resonance (SPR) spectroscopy using an antibody specific to oligomeric Aβ 42 ( Figure S1B). All Aβ 42 concentrations stated are based on the molar mass of the peptide.

LIVE CELL Ca 2+ IMAGING
Methods for single cell analysis of intracellular Ca 2+ concentration were as previously described (Peppiatt et al., 2003). Cells were loaded at 37 • C with 2 μM of the acetoxymethyl (AM) ester form of fura-2 for 30 min followed by an equivalent period in dye free media to allow de-esterification of the indicator. Imaging experiments were performed using either Ca 2+containing (121 mM NaCl, 5.4 mM KCl, 0.8 mM MgCl 2 , 1.8 mM CaCl 2 , 6 mM NaHCO 3 , 25 mM HEPES, 5.5 mM glucose, pH 7.3) or Ca 2+ free (as for Ca 2+ containing with 1.8 mM CaCl 2 replaced with 1 mM EGTA) buffer as indicated. Fura-2 imaging was carried out using an imaging system configured around a Nikon TE300 inverted epi-fluorescence microscope equipped with a 20 × 0.75 NA multi-immersion objective. Samples were illuminated by alternate excitation at 340 and 380 nm using a Sutter filter changer (340HT15 and 380HT15; Sutter Industries) and emitted light was filtered at >460 nm (1 ratio pair per 2 s). Images were captured using a Hamamatsu ORCA ER CCD camera. The imaging system was controlled with Ultraview software (PerkinElmer Life Sciences Ltd., UK). Acquired images were processed with Ultraview software and analyzed in MATLAB. Background subtracted fura-2 ratios were calibrated according to standard procedures (Grynkiewicz et al., 1985), using the maximum and minimum ratio values obtained through exposing cells sequentially to Ca 2+ free and Ca 2+ containing imaging buffer to which 2 μM ionomycin had been added. Parameters analyzed from the Ca 2+ responses included the peak amplitude, the time to peak and the integral of the response (the area under the curve) and the percentage of responding cells.

MTT REDUCTION ASSAY
The Cell Titer 96 Non-Radioactive Cell Proliferation Assay (Promega) was used to validate the cytotoxic effect of Aβ 42 on SH-SY5Y cells and was performed according to manufacturer's instructions. Briefly, cells were incubated with Aβ 42 (n = 4) for 24 h prior to the addition of the MTT dye solution and a further 4 h incubation at 37 • C, 5% CO 2 . Thereafter, the solubilization/stop solution was added and incubated overnight at room temperature. Absorbances were read at 570 nm with a reference wavelength of 650 nm using a fluorescence plate reader (Synergy HT, BIO-TEK). The data is expressed as the percentage of MTT reduction relative to both live-and dead-cell controls and thus represents the percentage of viable cells. Aβ 42 samples were considered to be toxic if 25-40% of cells remained metabolically healthy at an Aβ 42 concentration of 1 μM and if more than 50% remained metabolically healthy at a concentration of 100 nM.

STATISTICAL ANALYSIS
Data is presented as the mean value of the combined datasets ± SEM. Statistical significance was determined by Student's t-test (two-tailed). Data was accepted as significant when p < 0.05 and is denoted by * p < 0.05, * * p < 0.01, or * * * p < 0.001.

INTRACELLULAR Ca 2+ IS ELEVATED IN CELLS EXPOSED TO OLIGOMERIC Aβ 42
Experiments were first performed to validate the Ca 2+ mobilizing properties of oligomeric Aβ 42 over the concentration range of its toxicity. Application of Aβ 42 spanning its cytotoxic range (1, 5 and 10 μM) caused an elevation in intracellular Ca 2+ ( Figure 1A). The increase in cytosolic Ca 2+ concentration immediately followed the addition of Aβ 42 , developed to a peak within minutes of application and subsequently returned to baseline, despite the continued presence of the peptide. No Ca 2+ responses were detected when Aβ 42 below 1 μM was applied (data not shown). Between 1 μM and 10 μM Aβ, the number of responding cells, the peak amplitude and the integral of the Ca 2+ responses increased in a concentration-dependent manner. The number of responding cells reached 100% at 5 μM Aβ 42 (Figures 1Bi,iii,v). To test cell viability as well as to determine whether metabotropic Ca 2+ signaling was affected by Aβ, carbachol (CCH) was applied subsequent to Aβ. CCH elicited Ca 2+ responses in 100% of cells pre-exposed to 1 or 5 μM oligomeric Aβ 42 or to a vehicle control (10%) (Figures 1Bii,iv,vi). At 10 μM Aβ, however, the number of cells responding to CCH was significantly reduced (Figure 1Bii). The peak amplitude and integral of the Ca 2+ responses to CCH subsequently applied were inversely related to the magnitude of the Ca 2+ responses elicited by oligomeric Aβ 42 (Figures 1Biv,vi). This observation suggested that exposure to Aβ 42 oligomers was depleting the intracellular CCH-sensitive ER Ca 2+ store. These Ca 2+ mobilizing effects of oligomeric Aβ 42 were significantly greater than observed in cells exposed to Aβ 42 that had been prepared in a manner to yield a monomeric form of the peptide (Figures S2, S1B). From these results, due to its potency in mobilizing Ca 2+ whilst preserving agonist responses, a concentration of 5 μM oligomeric Aβ 42 was selected for use in subsequent experiments.
As a control for the application of peptide, experiments were also performed using a scrambled Aβ sequence, which had been prepared in the same manner as the wild type Aβ 42 . Although significantly less toxic than the wild type sequence (Figure S1A), scrambled Aβ peptide also evoked Ca 2+ responses in all cells (Figure 2Ai). However, consistent with its lower toxicity, both the amplitude and the integral of the Ca 2+ transients elicited by scrambled Aβ were significantly lower than those induced by oligomeric Aβ 42 and, in addition, they required a significantly longer time to reach peak (Figures 2Bi,Ci,Di). Furthermore, concordant with the less potent effect of scrambled Aβ in mobilizing intracellular Ca 2+ , the amplitude and integral of CCH-induced Ca 2+ transients elicited following prior exposure to scrambled Aβ were significantly greater than those stimulated following prior exposure to oligomeric Aβ 42 (Figures 2Bii,Cii,Dii).
Taken together, the comparison of the effects of Aβ scramble and oligomeric Aβ 42 demonstrates that the amino acid sequence of Aβ 42 has potent Ca 2+ mobilizing properties, which are distinct from the action of Aβ scramble.

Aβ 42 OLIGOMERS MOBILIZE Ca 2+ FROM INTRACELLULAR STORES
The reduced magnitude of CCH-induced Ca 2+ signals observed in cells previously exposed to oligomeric Aβ 42 suggested that this form of Aβ 42 was exerting an effect on intracellular Ca 2+ stores. Therefore, we tested the relative contributions of Ca 2+ influx from the extracellular space and its release from intracellular stores to Aβ 42 -induced Ca 2+ transients.
To determine the contribution of extracellular Ca 2+ and Ca 2+ influx to Aβ 42 oligomer-induced Ca 2+ transients, we performed experiments using Ca 2+ -free imaging buffer. Under these conditions, Aβ 42 oligomers retained their ability to induce Ca 2+ responses, with 100% of cells responding (Figure 3Ai). While no significant difference was observed in the peak amplitude (Figure 3Aiii) of Aβ 42 oligomer-induced Ca 2+ transients, the integral of the response was significantly decreased in the absence of extracellular Ca 2+ (Figure 3Av).
In contrast to the Aβ 42 oligomer-induced Ca 2+ response, the peak amplitude and the integral of the Ca 2+ responses to CCH applied following Aβ 42 oligomer exposure were significantly decreased by removal of extracellular Ca 2+ from the imaging buffer (CCH, after Aβ 42 ; Figures 3Aiv,vi). This effect on the CCH-induced Ca 2+ responses is likely due to lack of storeoperated Ca 2+ entry, which would replenish the Ca 2+ released from stores by Aβ 42 . Indeed, the peak amplitude and the integral of CCH-induced Ca 2+ responses elicited in Ca 2+ free buffer were significantly greater in naïve cells (CCH, no Aβ 42 ) than when Aβ 42 oligomers were previously applied (Figures 3Aiv,vi). Since Aβ 42 oligomer-induced Ca 2+ transients were not significantly affected by removal of extracellular Ca 2+ , these results suggest CCH-induced Ca 2+ responses were normalized to control experiments conducted on the same experimental day. Bar graphs are mean ± SEM from at least three independent experiments. * p < 0.05; * * p < 0.01; * * * p < 0.001. that oligomeric Aβ 42 and CCH mobilize Ca 2+ from a common intracellular Ca 2+ pool. The requirement of Ca 2+ release from the ER Ca 2+ store for the Ca 2+ transients elicited by Aβ-induced was next investigated. To this end, ER Ca 2+ stores were depleted by exposure of cells the SERCA pump inhibitor thapsigargin (Tg; 2 μM, 15 min) prior to the application of Aβ 42 . In the absence of replete ER Ca 2+ stores, Aβ 42 -induced Ca 2+ transients were completely abrogated (Figures 3Bi,iii,v). Similarly, CCH-induced Ca 2+ responses were eliminated in Tg-treated cells (Figures 3Bii,Biv,Bvi), confirming the effect of Tg. Taken together, these experiments establish that Aβ 42 oligomers mobilize Ca 2+ from the ER.

Aβ 42 -INDUCED Ca 2+ RELEASE OCCURS IN PART THROUGH INSP 3 Rs
Having determined that Aβ 42 oligomers mobilize Ca 2+ from the intracellular ER Ca 2+ store, we aimed to identify the mechanism by which Ca 2+ release occurs. We therefore tested whether Aβ 42 was causing Ca 2+ release from the ER through activation of InsP 3 R or ryanodine receptor (RyR) Ca 2+ release channels localized to this organelle.
Although SH-SY5Y cells have been reported to express functional RyRs, application of caffeine (10 mM), an agonist of the three RyR isoforms (10 mM) did not elicit a Ca 2+ response in the SH-SY5Y cells used in this study ( Figure S2A). Furthermore, the neuronally-expressed type 2 RyR could not be detected by immunoblot analysis (Figure S2B). Based on these observations, a role for RyR2 in Aβ 42 oligomer-mediated Ca 2+ release was ruled out.
Caffeine application did not affect the number of cells exhibiting Ca 2+ responses following Aβ 42 oligomer application, with 100% of cells responding (Figure 4B). However, caffeine significantly decreased the peak amplitude and the integral of the Aβ 42 oligomer-induced Ca 2+ transients ( Figure 4B). In contrast, Aβ scramble-induced Ca 2+ transients were unaffected by caffeine application (Figure 4C). Ca 2+ responses to 0.5 μM CCH were abolished by caffeine, demonstrating its inhibitory effect upon IICR (Figures 4A-C). Data is presented as percentage of responding cells, peak amplitude and integral of the response. Bar graphs are mean ± SEM from at least three independent experiments. * * * P < 0.001.
Although caffeine inhibits InsP 3 Rs (Bezprozvanny et al., 1994), it also acts on targets other than the InsP 3 R such as cyclic nucleotide phosphodiesterases and phospholipase C (PLC) (Toescu et al., 1992;Taylor and Broad, 1998). Therefore, to investigate further the role of InsP 3 signaling in the generation of Aβ 42 oligomer-induced Ca 2+ transients, InsP 3 signaling was inhibited by GFP-5 P overexpression. Using this strategy, InsP 3 -mediated Ca 2+ signals induced by CCH were prevented, validating this approach for suppression of InsP 3 signaling ( Figure 5A). As observed for caffeine, however, GFP-5 P overexpression did not prevent Aβ 42 oligomer-induced Ca 2+ transients, with 100% of cells responding (Figure 5B). However, the peak amplitude and the integral of Aβ 42 oligomer-induced Ca 2+ transients were significantly decreased by overexpression of GFP-5 P ( Figure 5B) when compared to the magnitude of Ca 2+ transients in control cells, expressing GFP alone. Significantly, Aβ scramble-induced Ca 2+ transients were not affected by GFP-5 P overexpression with no significant impact of its expression upon the peak amplitude or the integral of Aβ scramble-induced Ca 2+ transients ( Figure 5C). Taken together, these results demonstrate that Ca 2+ transients elicited by Aβ 42 oligomers arise as a result of release from the ER intracellular Ca 2+ store and that activation of InsP 3 Rs contributes to this effect.

Aβ 42 OLIGOMER-INDUCED Ca 2+ LEAK FROM THE ER
The data presented above indicates that externally applied Aβ 42 rapidly induces an increase on cytosolic Ca 2+ that involves InsP 3 -dependent and -independent Ca 2+ release from the ER. Since Aβ 42 has also been shown to elicit some of its cytotoxic effects as a result of intracellular accumulation (Wirths et al., 2004), we investigated whether it mobilized Ca 2+ from the ER when directly applied. We also tested whether InsP 3 Rs were required for its intracellular action.
To this end, an established permeabilized cell high-throughput functional assay of ER Ca 2+ release was used (Tovey et al., 2006). This model uses as substrate for specific analysis of ER Ca 2+ release, a plasma membrane-permeabilized preparation of the DT40 chicken B-lymphocyte cell line. A derivative of this cell line in which the 3 InsP 3 R isoforms have been deleted by homologous recombination (DT40 TKO), allows the requirement for InsP 3 Rs for Ca 2+ release to be tested (Sugawara et al., 1997). Cell permeabilization and substantial dilution in intracellular buffer rules out the contribution of endogenously generated InsP 3 to signaling in this assay. Using this assay, a significantly greater InsP 3 independent ER Ca 2+ leak was observed in both wild-type (p = 0.002) and DT40 TKO cells (p = 0.0195) exposed to Aβ 42 oligomers compared to the passive Ca 2+ leak detected in each cell type (Figures 6A,B). The maximal Ca 2+ leak rate induced by Aβ 42 oligomers was not significantly different between wildtype and DT40 TKO cells (p = 0.2606, Figure 6C), suggesting that InsP 3 Rs were not required for Aβ 42 oligomers to trigger Ca 2+ release.

FIGURE 5 | Aβ 42 oligomer-induced Ca 2+ release occurs is reduced by InsP 3 5 P expression. (A)
Imaging protocol employed to investigate the involvement of InsP 3 Rs in Aβ 42 oligomer-mediated Ca 2+ release from the ER. to the passive Ca 2+ leak observed in each cell type (Figure 6B), and thus there was no significant difference in the maximal Ca 2+ leak rate following Aβ scramble application between these two cell types (p = 0.2522, Figure 6C). Importantly, a significant difference between the Ca 2+ leak rates triggered by exposure to Aβ 42 oligomers and Aβ scramble in wild-type DT40 cells (p = 0.0056) and DT40 TKO cells (p = 0.0045) was observed, indicating that Aβ-induced Ca 2+ leak from the ER is dependent and specific to the amino acid sequence of Aβ 42 . Taken together, these results suggest that Aβ 42 oligomers trigger a Ca 2+ leak from the ER, which does not depend upon a direct interaction with InsP 3 Rs.

DISCUSSION
Here we show that the oligomeric form of the AD-associated peptide Aβ 42 has potent Ca 2+ mobilizing properties and we identify mechanisms responsible for its action. Using both intact and permeabilized cell assays to investigate the effects of extracellular and internalized Aβ 42 , respectively, we establish that Ca 2+ release from the ER makes the greatest contribution to the Ca 2+ mobilizing effects of Aβ 42 . The InsP 3 signaling pathway also contributes to the Ca 2+ mobilizing properties of oligomeric Aβ 42 in intact cells. InsP 3 Rs were not required for Aβ 42 -stimulated Ca 2+ flux in permeabilized cells ruling out a direct regulation of InsP 3 Rs by Aβ 42 . Central to the Ca 2+ hypothesis of amyloid toxicity is the property of Aβ to induce Ca 2+ elevations in its target cells. This sets in motion a cascade of events, which culminates in neuronal death. Ever since this hypothesis was put forward more than 20 years ago (Khachaturian, 1989(Khachaturian, , 1994, numerous reports have described Aβ-induced changes in intracellular Ca 2+ in a number of cell types including primary neurons and astrocytes as well as neuroblastoma cell lines (Abramov et al., 2004b;Demuro et al., 2005). While there is general consensus that Aβ affects Ca 2+ homeostasis, the mechanisms underlying this action of Aβ are many. Contributing to this diversity are the different experimental models used, the peptide sequence applied, the conformational state of the peptide and the method used for peptide preparation. Indeed, a number of shorter Aβ sequences have been employed in in vitro studies and depletion of ER Ca 2+ store content reported (Ferreiro et al., 2004. Since Aβ 42 is considered to be more relevant to the pathology of AD, we focused on its effects on intracellular Ca 2+ homeostasis. Not only is an accumulation of Aβ 42 observed in AD, this longer and more hydrophobic peptide is also more prone to self-assemble than Aβ 40 , the other principle length at which Aβ occurs. As a result, Aβ 42 exerts a greater degree of neurotoxicity (Jarrett and Lansbury, 1993). Consistent with the growing body of evidence that soluble oligomeric forms of Aβ constitute the primary neurotoxic species (Walsh et al., 2002;Gong et al., 2003;Cleary et al., 2005;Klyubin et al., 2005), this species of Aβ 42 potently induced Ca 2+ fluxes and cytotoxicity in this study (Figures 1, 2 and Figure S2). Highlighting the requirement for appropriate peptide controls when studying Aβ 42 , Ca 2+ release and cytotoxicity was also induced by a scrambled peptide sequence of Aβ 42 , although the magnitude of these responses was significantly lower than that induced by the wild type sequence. From these results, we concluded that the peptide sequence of Aβ 42 was the major contributor to the toxicity and Ca 2+ mobilizing properties. The temporal properties of the Ca 2+ transients we observed were reminiscent of those reported elsewhere, being relatively slow in reaching peak and returning to baseline levels after a few minutes (Demuro et al., 2005;Simakova and Arispe, 2006). The return of these Ca 2+ signals to baseline does, however, suggest that the Ca 2+ elevations induced by Aβ 42 were not immediately toxic. The Ca 2+ mobilizing properties of the scrambled peptide, however, may reflect the previously described intrinsic properties of an oligomeric/amyloid peptide (Bucciantini et al., 2002;Yoshiike et al., 2008). For example, oligomeric forms of polyQ and insulin have been shown to induce Ca 2+ transients (Demuro et al., 2005). The solvent HFIP used for preparation of the peptide has also previously been shown to exhibit cytotoxicity and to affect ion conductance of membranes (Capone et al., 2009). Both Ca 2+ influx from the extracellular space and release from ER-localized intracellular stores have been reported to be induced by Aβ and involved in its toxic action (Blanchard et al., 2004;Ferreiro et al., 2004Ferreiro et al., , 2006Kayed et al., 2004;Demuro et al., 2005Demuro et al., , 2011Kelly and Ferreira, 2006;Simakova and Arispe, 2006;Arispe et al., 2007;De Felice et al., 2007;Resende et al., 2008;Demuro and Parker, 2013). Although Ca 2+ entry from the extracellular space was a component of the Ca 2+ elevation induced by Aβ 42 in this study, the greatest contribution was due to release from the ER. Moreover the lack of an effect of removal of extracellular Ca 2+ upon the initial peak of the Ca 2+ response or the number of responding cells suggested that Ca 2+ entry across the plasma membrane was secondary to Ca 2+ release from the ER. Since Aβ 42 was acting to deplete the ER stores, the Ca 2+ influx could arise via a store-operated Ca 2+ entry pathway. These observations are not, however, incompatible with an additional mechanism for Ca 2+ entry via plasma membrane pores formed by Aβ 42, which have been shown to require a longer period to develop (Demuro et al., 2011). Whether the Ca 2+ fluxes associated with the formation of membrane pores, which were generally local to the pore and were of a relatively small magnitude, contribute to the global Ca 2+ transient is not clear (Demuro et al., 2011).
Analysis of the mechanisms underlying Ca 2+ release from the ER revealed that while InsP 3 Rs contributed to Aβ 42 -induced Ca 2+ release from the ER in intact cells, the greater part of the Ca 2+ elevation induced by Aβ 42 was due to an alternative mechanism. However, IICR did not contribute to the Ca 2+ responses induced by scrambled peptide. From these results, we concluded that Aβ 42 -induced Ca 2+ release from the ER comprises an Aβ 42 sequence-specific component, which is InsP 3 -dependent, and a second component, which is peptide sequence-and InsP 3independent. Comparison of these Aβ 42 and Aβ 42 scrambled datasets reveals that although the InsP 3 -dependent component of the total Aβ 42 signal is relatively minor, when considered as a fraction of the Aβ 42 -specific Ca 2+ signal (i.e., Aβ 42 -Aβ 42 scrambled Ca 2+ transient), its importance is increased.
Our demonstration of the participation of InsP 3 signaling in Aβ 42 -induced Ca 2+ responses provides robust evidence in support of this pathway in Aβ 42 -mediated Ca 2+ signals thus far. In particular, the use of InsP 3 5 phosphatase overexpression to suppress InsP 3 signaling is a highly selective strategy, overcoming issues regarding incomplete knockdown of InsP 3 Rs and contribution of the isoforms not targeted when using siRNA approaches. The inhibition of Ca 2+ signals by caffeine is also consistent with a role for the InsP 3 signaling pathway in the Ca 2+ mobilizing effects of Aβ (Parker and Ivorra, 1991;Bezprozvanny et al., 1994). Not only does caffeine inhibit InsP 3 Rs directly (Bezprozvanny et al., 1994), by also inhibiting PLC, caffeine is a potent inhibitor of InsP 3 generation (Taylor and Broad, 1998). These findings are consistent with the reduction in the Aβ 42 -induced Ca 2+ transient observed following application of the PLC inhibitor U73122 (Resende et al., 2008) although U73122 has numerous non-specific effects. The mechanism by which InsP 3 signaling is engaged by Aβ 42 in this study remains to be established. Since the effects of inhibition of InsP 3 signaling persist in the absence of extracellular Ca 2+ , activation of PLC and InsP 3 generation by Aβ 42 -stimulated Ca 2+ influx can be excluded. Thus, a more likely scenario would involve Aβ 42 engagement of a PLC-coupled G protein coupledreceptor (GPCR). Indeed, a number of different GPCRs, including metabotropic glutamate receptors, are activated by Aβ 42 , contributing to modulation of LTP, Aβ 42 synthesis and processing and cytotoxicity (Wang et al., 2004;Thathiah and De Strooper, 2011).
The internalization of Aβ from the extracellular space (Bucciantini et al., 2004;Pierrot et al., 2004;Wirths et al., 2004;Kaminski Schierle et al., 2011) raises a further possibility that Aβ acts to either directly activate/sensitize InsP 3 Rs or to alter InsP 3 generation/metabolism. Since significant intracellular Aβ 42 accumulation would require up to 1 h (Bucciantini et al., 2004;Kaminski Schierle et al., 2011), it is unlikely that this endocytosed Aβ 42 contributes to the acute modulation of Ca 2+ fluxes observed in this study and elsewhere in intact cells. Endocytosis of Aβ 42 may, however, contribute to the more chronic effects on Ca 2+ homeostasis as well as cytotoxicity previously reported (Ferreiro et al., 2004(Ferreiro et al., , 2006Resende et al., 2008). The possibility that Aβ 42 could directly affect ER Ca 2+ homeostasis from an intracellular location was therefore also considered. Using a permeabilized cell assay to allow control of cytosolic conditions and access of Aβ 42 to the ER, an Aβ 42 -stimulated Ca 2+ efflux from the ER was observed. Unlike that observed for intact cells, the difference between Aβ 42 and Aβ 42 scrambled was dramatic, revealing a highly specific effect of Aβ 42 upon ER Ca 2+ mobilization. These effects were observed in the absence of exogenous InsP 3 suggesting that the effects were InsP 3 R-independent. The extensive dilution of cytosol following permeabilization of the DT40 cells would also likely preclude a contribution of Aβ 42stimulated InsP 3 generation. More significantly, InsP 3 Rs were not required for the Ca 2+ mobilizing properties of Aβ 42 , since deficiency in all three InsP 3 R isoforms did not affect the Ca 2+ mobilizing properties of Aβ 42 . The absence of a requirement for InsP 3 Rs for Aβ 42 -stimulated Ca 2+ flux in the permeabilized cell system does not rule out the possibility that IICR contributes to Ca 2+ fluxes and toxicity mediated by intracellular Aβ 42. Indeed, by activating Ca 2+ -sensitive PLC and generation of InsP 3 , Ca 2+ mobilized by Aβ 42 could promote IICR. Consistent with this notion, microinjected Aβ 42 was recently shown to promote Ca 2+ signals in Xenopus oocytes in a manner that involved InsP 3 generation (Demuro and Parker, 2013).
The depletion of the ER Ca 2+ store by Aβ 42 has important implications for the mechanisms of its toxicity. Depletion of ER Ca 2+ stores results in the accumulation of unfolded proteins and activation of the ER stress response, which via caspase 12 activation and Bap31 cleavage can subsequently induce mitochondrial apoptotic cascades (Verkhratsky, 2005;Xu et al., 2005;Mekahli et al., 2011). The engagement of InsP 3 Rs during Aβ 42stimulated depletion of ER Ca 2+ may be of greater consequence. Specifically, InsP 3 R-induced Ca 2+ release from the ER and its subsequent sequestration by neighboring mitochondria could lead to mitochondrial Ca 2+ overload, permeability transition and death (Csordas et al., 2006). These pathways also lead to increased reactive oxygen species generation, which is commonly observed in AD (Ferreiro et al., 2004Arduino et al., 2009;Clark et al., 2010).
The use of SH-SY5Y neuroblastoma cell line and permeabilized DT40 B-lymphocytes in this study, rather than primary neurons allowed careful dissection of the role of ER Ca 2+ signaling to Aβ-induced Ca 2+ signals independent from Ca 2+ fluxes that may arise in neurons as a result of electrical or synaptic activity. Moreover, using this cell line, contributions from other Aβ targets described in neurons such as NMDA receptors are excluded. Analogous to a number of other studies in electrically non-excitable primary and cultured cells including Xenopus oocytes (Demuro and Parker, 2013) astrocytes and PC12 cells (Abramov et al., 2003(Abramov et al., , 2004aSimakova and Arispe, 2006), our data indicates that certain of the Ca 2+ mobilizing properties of Aβ 42 are neuron-independent and do not require the expression of any other of its reported targets. Fundamental aspects of the Ca 2+ mobilizing properties of Aβ 42 were further revealed and exemplified by the Aβ 42 -stimulated Ca 2+ flux from the InsP 3 Rdeficient ER of permeabilized DT40 B-lymphocytes. These latter data demonstrate for the first time that Aβ 42 has the capacity to directly induce Ca 2+ flux from the ER. Given the importance of the ER and InsP 3 Rs in neuronal functions, future studies will be required to test whether InsP 3 Rs contribute to Aβ-mediated neuronal pathology.

AUTHOR CONTRIBUTIONS
Laura E. Jensen: substantial contributions to conception and design, acquisition, analysis and interpretation of data as well as writing of manuscript. H. Llewelyn Roderick: substantial contributions to conception and design, interpretation of data as well as writing of manuscript. Geert Bultynck and Tomas Luyten: designed, acquired, analysed and interpreted data of Figure 6. Hozeefa Amijee: designed, acquired and interpreted data of Figure S2. Martin D. Bootman: proof-read manuscript.
Antibodies (Bii) A11 and (Biii) 12F4 were injected over the immobilized Aβ 42 of each flow cell at a concentration of 50 μg/ml and 10 μg/ml, respectively. The injection of the anti-oligomer antibody, A11, was followed by a regeneration step prior to injection of 12F4. The binding of injected antibodies, present in the flow phase, to the immobilized Aβ 42 was measured by response units (RU) elicited. All values were corrected for the RU obtained from the reference cell, flow cell 1, which was saturated with biotinylated Aβ 42 only. Soluble Aβ monomers and Aβ oligomers were prepared as previously described (Demuro et al., 2005). This method of Aβ preparation reportedly results in homogeneous populations of Aβ monomers and oligomers (also