Aβ peptide (Aβ_1-42) was from Bachem Inc. (Torrance, CA, USA). Fluo4-AM, MitoSOX^TM Red Mitochondrial Superoxide Indicator, MitoTracker^® Orange CMTMRos, anti-rabbit Alexa Fluor^® 488 and anti-mouse Alexa Fluor^® 635 were from Molecular Probes, Inc. (Eugene, OR, USA). Hexafluoro-2-propanol (HFIP) and CMC were from Merck (Darmstadt, Germany), Neurobasal and Dulbecco’s modified essential medium (DMEM), B27 supplement and lipofectamine 2000 were from Gibco (Carlsbad, CA, USA). DOTAP Liposomal Transfection Reagent was from Sigma–Aldrich (Oakville, ON, Canada). Phosphodiester oligonucleotides (ODNs) were from Integrated DNA Technologies (Coralville, IA, USA). The mito-Pericam plasmid was donated by Dr. V. Eisner. Bicinchoninic acid assay (BCA) kit and mHsp-70 antibody were from Pierce Biotechnology (Rockford, IL, USA). Ryanodine was from Alexis (Lausen, Switzerland). PDVF membranes were from Millipore (Bedford, MA, USA). RyR2 antibody and Rhod2-AM was from Thermo-Fisher (Waltham, MA, USA). Gp91 ds-tat was from AnaSpec (Fremont, CA, USA).

The Aβ_1-42 peptide was prepared as previously described, as a HFIP film (De Felice et al., 2007; Paula-Lima et al., 2011; SanMartín et al., 2012a,b; Lobos et al., 2016). This film is dissolved next in DMSO to obtain a 5 mM stock solution, which is subsequently diluted with cold phosphate buffered saline (PBS) to 100 μM and incubated overnight at 4°C. After 24 h, the Aβ solutions (100 μM) were centrifuged at 4°C, 14,000 × g for 10 min to remove protofibrils and fibrils (insoluble aggregates). Supernatants with soluble AβOs were transferred to sterile tubes and protein levels were determined with a BCA kit. Fresh preparations of AβOs were used in all experiments.

Eighteen-day-old embryos from Sprague-Dawley rats were used to obtain primary hippocampal cultures, as we previously described (Paula-Lima et al., 2005, 2011; SanMartín et al., 2012a,b, 2014, Lobos et al., 2016). Concisely, after meninges removal from brains, hippocampi were dissected and hippocampal cells were dissociated softly in HANKS-glucose solution. Cells were then centrifuged and resuspended in DMEM plus 10% horse serum and plated on polylysine-coated plates. After 1 h, DMEM was replaced by Neurobasal medium plus B-27. Cells were maintained for 15–21 days in vitro (DIV) in a humidified 5% CO₂ atmosphere at 37°C prior to experimental handlings. Mature hippocampal cultures were enriched in neurons with a glial content <24% (Paula-Lima et al., 2011). This study was carried out in accordance with the recommendations of The Guidelines on the recognition of pain, distress and discomfort in experimental animals. The protocol was approved by the Bioethics Committee on Animal Research, Faculty of Medicine, University of Chile.

Hippocampal cultures at 21 DIV were fixed by adding an equal volume of 4% formaldehyde and 4% sucrose (in PBS buffer) for 10 min, rinsed three times with PBS, incubated with 10% normal goat serum plus 0.1% Triton X-100 (blocking-permeant solution) for 1 h and then immunolabeled by overnight incubation at 4°C with mHsp-70 diluted in blocking solution (1/750). After this incubation period, cultures were rinsed three times with PBS and were incubated for 1 h at room temperature with Alexa Fluor^® 488 anti-rabbit as secondary antibody (1/400 in blocking solution). Cells were rinsed three times with PBS, and coverslips were mounted in DAKO mounting medium for morpho-topological analysis of the mitochondrial network. Quantification of the percentage of neurons with fragmented mitochondria was carried out as described previously (SanMartín et al., 2012a, 2014). To label mitochondria, cells were labeled for 20 min at 37°C with 50 nM MitoTracker Orange and observed on a Carl Zeiss LSM Pascal 5 confocal microscope system (Zeiss, Oberkochen, Germany) or on a Nikon C2+ confocal Microscope (Melville, NY, USA). Images were digitally acquired using LSM software (Zeiss) or NIS-Elements C software (Nikon). Image deconvolution and generation of zeta projections from 0.4 μm 7–15 stacks were performed using the ImageJ software program (National Institutes of Health, USA). Neurons were typed as exhibiting filamentous or fragmented mitochondrial network. Ten optical fields were observed for each condition, counting approximately 15 neurons. The percentage of neurons with fragmented mitochondria was determined respect to the total number of neurons counted.

Neurons (14–21 DIV) were treated with 500 nM AβOs at the microscope stage, or for different incubation periods in the culture plates, depending on the type of experiment performed.

To down-regulate RyR2 expression, we used phosphodiester oligonucleotides (ODNs) with the following sequences. ODN RyR2: 5′-T^∗T^∗C GCCCGCATCAGCC^∗A^∗T-3′; ODN Scrambled (ODN Scr), 5′-C^∗G^∗GCAGGAGTCTGTG C^∗G^∗C-3. The ODN Scr was used as control, as previously described (Galeotti et al., 2008). Liposomal Transfection Reagent DOTAP (13 μM) was used to introduce ODNs into neurons. As controls, we also transfected neurons with ODNs specifically designed for the RyR1 and RyR3 isoforms; these ODNs did not modify RyR2 expression (data not shown).

RyR2 mRNA levels were determined by RT-PCR performed in a MX3000P Stratagene amplification system (La Jolla, CA, USA) using the DNA binding dye SYBR green and the following previously described Primer sense/Primer antisense sequences: 5′-AATCSanMartínGTGGCGGAATTTCTTG-3′/5′-TCTCCCTCAGCCTTCTCCGGTTC-3′ (Paula-Lima et al., 2011; Lobos et al., 2016). Levels of RyR2 mRNA were normalized respect to levels of β-actin mRNA and calculated by the relative 2-ΔΔCt method. For determination of RyR2 protein content, we performed western blot analysis. Cells homogenates were separated by SDS-PAGE (3.5–8% gradient or 10% polyacrylamide gels) and transferred to PVDF membranes for subsequent incubation with specific antibodies against RyR2 (Adasme et al., 2011; Paula-Lima et al., 2011).

Cells were preloaded with 5 μM Fluo4-AM in Tyrode solution (in mM: 30 glucose, 129 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, 25 HEPES-Tris, pH 7.3) for 30 min at 37°C. After washing three times with Tyrode, 500 nM AβOs were added to the cultures at the microscope stage and fluorescence images of intracellular Ca²⁺ signals were obtained every 15 s in an inverted confocal microscope (Carl Zeiss LSM Pascal 5) or every 3 s in an inverted confocal microscope (Nikon C2+). Regions of interest (ROIs) were determined in cell bodies and neurites. Relative Ca²⁺ levels are presented as F/F₀ values, where F₀ corresponds to the basal fluorescence and F to the experimental fluorescence. Experiments were done at room temperature (20–22°C).

Neurons at 14 or 15 DIV were transiently transfected with the mito-Pericam plasmid using a ratio of 1:3 DNA:lipofectamine 2000 as previously described (SanMartín et al., 2014). Twenty-four hours after transfection, cultures were treated with 50 μM ryanodine for 1 h, with 10 mM NAC for 30 min, or with vehicle. Next, cultures were washed three times with Tyrode solution and were maintained in this solution at the microscope stage. The 500 nM AβOs or 0.5 mM 4-CMC were added to the cultures. Mitochondrial Ca²⁺ signals from neuronal cells (identified as such by morphology) were recorded every 3 s in an Olympus Disk Scanning Unit (DSU) IX 81 confocal microscope (Olympus, Hamburg, Germany) using 60× oil immersion objective, excitation 420 nm and Hg/Ar lamp. Changes in Ca²⁺ levels are presented as F/F₀ values, where F corresponds to the experimental fluorescence and F₀ to the basal fluorescence. Experiments were done at room temperature (20–22°C).

Neurons at 14–21 DIV were incubated with 2.5 μM Rhod2 for 30 min, washed three times with Tyrode solution and incubated for additional 30 min to allow mitochondrial loading with Rhod2. Next, cells were transferred to Tyrode solution containing 5 μM Fluo4 and incubated for an additional 30 min period. Cells were then rinsed three times with Tyrode and AβOs (500 nM) were added to the cultures at the microscope stage. Simultaneous fluorescence images of intracellular and mitochondrial Ca²⁺ signals were obtained every 3 s in an inverted confocal microscope (Nikon C2+). ROIs were determined in cell bodies and neurites. Relative Ca²⁺ levels are presented as F/F₀ values, where F₀ corresponds to the basal fluorescence and F to the experimental fluorescence. Experiments were done at room temperature (20–22°C).

Cultures were treated for 1 h with 50 μM ryanodine, for 30 min with 10 mM NAC, or for 2 h with 20 μM EUK-134, in Neurobasal medium supplemented with B-27. Cultures were then placed in modified Tyrode solution for subsequent loading with 1 μM MitoSOX for 20 min at 37°C. After washing three times with modified Tyrode solution, AβOs (500 nM) were added to the cultures at the microscope stage. The fluorescence images generated by the mitochondrial superoxide probe in primary hippocampal neurons (identified as such by morphology) were recorded every 5 s in a confocal microscope (Carl Zeiss LSM Pascal 5). Fluorescence signals are presented as F/F₀ values, where F₀ corresponds to the basal fluorescence levels and F to the experimental fluorescence. Experiments were performed at room temperature (20–22°C).

Cultures at 14 or 15 DIV were transfected transiently with the HyperMito plasmid (Evrogen, Moscow, Russia) at a ratio of 1:3 DNA:lipofectamine 2000. Twenty-four hours after transfection, cultures were treated for 1 h with 50 μM ryanodine, for 30 min with 10 mM NAC, for 2 h with 20 μM EUK-134, or with vehicle in Neurobasal plus B27 medium. After three rinses with Tyrode solution, 500 nM AβOs were added to the cultures at the microscope stage. The fluorescent signals generated by the mitochondrial hydrogen peroxide probe were recorded from neuronal cells (identified as such by morphology) every 3 s in a confocal microscope (Carl Zeiss LSM Pascal 5). Relative mitochondrial hydrogen peroxide levels are presented as F/F₀ values, where F₀ corresponds to the basal fluorescence and F to the experimental fluorescence. Experiments were performed at room temperature (20–22°C).

Mitochondria were identified by staining fixed cultures with mHsp-70, as we previously described (Paula-Lima et al., 2011; SanMartín et al., 2012a, 2014). The specificity of mHsp-70 as a mitochondrial stain was previously confirmed by staining mitochondria with MitoTracker Orange, which yielded the same labeling pattern as mHsp-70 (SanMartín et al., 2014). To determine the levels of the mitochondrial protein mHsp-70 in neurites and soma, segmentations were performed to define different ROIs, as described in detail elsewhere (SanMartín et al., 2014). Confocal image stacks were captured with a confocal microscope (Zeiss LSM-5, Pascal 5 Axiovert 200), using the LSM 5 3.2, and deconvoluted using Huygens Scripting (Scientific Volume Imaging, Hilversum, Netherlands).

3D models were reconstructed from successive xy-images along the z-axis. Based on their volumes, we defined four different clusters to characterize mitochondrial connectivity as previously described (SanMartín et al., 2014). First, we determined the mean volume of single mitochondria, yielding 0.15 ± 0.04 μm³ (mean ± SE, n = 834). The mean volume of single mitochondria was used to define connected clusters: (i) 1–3 mitochondria (0–0.45 μm³); (ii) 4–10 mitochondria (0.45–1.5 μm³); (iii) 11–50 mitochondria (1.5–7.5 μm³); (iv) over 50 mitochondria (>7.5 μm³). Values obtained with control neurons were compared to those obtained from neurons treated with 500 nM or 1 μM AβOs for 24 h.

The significance of differences in the experiments was determined using paired Student’s t-test or one-way ANOVA followed by Bonferroni’s post hoc test.

We have shown in previous work that AβOs generate Ca²⁺ entry signals via NMDA receptors, which promote RyR-mediated Ca²⁺-induced Ca²⁺ release (Paula-Lima et al., 2011). Stimulation of RyR channels by Ca²⁺ is redox sensitive and does not occur if RyR channel cysteine residues are highly reduced (Marengo et al., 1998). Accordingly, we tested if inhibition of the NADPH oxidase type-2 (NOX2), an important neuronal source of superoxide radical generation (Kishida and Klann, 2007; Ma et al., 2011; Riquelme et al., 2011), affected AβOs-induced cytoplasmic Ca²⁺ signals. As illustrated in Supplementary Figure S1, incubation of neurons with gp91-ds-tat, an inhibitory peptide of NOX2 activity that precludes its assembly (Rey et al., 2001), prevented the generation of Ca²⁺ signals in response to AβOs. In contrast, hippocampal cells incubated with a scrambled gp91-ds-tat peptide (scr), displayed similar Ca²⁺ signal generation in response to AβOs as controls. Based on these findings, we suggest that AβOs stimulate NOX2 activity, presumably via NMDA receptor stimulation (Brennan et al., 2009), and that the increased Ca²⁺ and ROS levels induced by AβOs jointly stimulate RyR-mediated Ca²⁺ release.

Mito-Pericam is a plasmid that expresses a Ca²⁺-sensing protein that decreases its fluorescence upon Ca²⁺ binding (Rizzuto et al., 1992), which is conjugated to a GFP derivative and a mitochondrial destination sequence. Hippocampal neurons transfected with mito-Pericam, represented by the false colored neuron illustrated in Figure 1A, were treated at the microscope stage with AβOs (500 nM) or with the RyR channel agonist 4-CMC (0.5 mM). Addition of AβOs (Figure 1A, bottom) produced a significant decrease in mito-Pericam fluorescence compared to that registered under basal levels (Figure 1A, top), indicating that AβOs induce mitochondrial Ca²⁺ entry. Quantification of fluorescence changes revealed that, within 1 min after AβOs or 4-CMC addition, neurons displayed significantly lower fluorescence relative to neurons treated with vehicle (Figure 1B); after 500 s, the decrease was significantly higher in 4-CMC-treated compared to AβOs-treated neurons.

To investigate whether RyR-mediated Ca²⁺ release from the ER underlies the AβOs-induced mitochondrial Ca²⁺ increase, we pre-incubated neurons for 1 h with 50 μM ryanodine (Rya), which in these conditions abolishes RyR-mediated Ca²⁺ release without causing Ca²⁺ depletion from the ER (Adasme et al., 2015). Interestingly, neurons pretreated with ryanodine did not display differences in mito-Pericam fluorescence after AβOs addition (Figure 1B). Previous work indicated that AβOs promote cytoplasmic ROS production (De Felice et al., 2007). Hence, we evaluated the participation of ROS in the mitochondrial Ca²⁺ increase induced by AβOs. To this aim, we pre-incubated neurons with 10 mM NAC for 30 min before the addition of AβOs. As illustrated in Figure 1B, NAC completely prevented the mitochondrial Ca²⁺ increase induced by AβOs. Quantification of the fluorescence recorded 500 s after AβOs addition, illustrated in Figure 1C, shows that both inhibitory ryanodine and NAC prevented AβOs-induced mitochondrial Ca²⁺ increase. Accordingly, we propose that AβOs induce Ca²⁺ uptake in mitochondria through RyR-mediated Ca²⁺ release, which requires in turn NMDA-receptor mediated Ca²⁺ entry and NOX2-mediated ROS generation.

To determine mitochondrial superoxide levels we used the MitoSOX^TM Red reagent (MitoSOX), comprised of a hydroethidine linked to a triphenylphosphonium cationic group that target this probe to the mitochondrial matrix in response to the negative membrane potential (Robinson et al., 2006). Oxidation of MitoSOX by superoxide produces red fluorescence signals. Stimulation of neurons with AβOs produced a rapid and sustained increase in MitoSOX fluorescence, indicating that AβOs promote mitochondrial superoxide generation; Figure 2A illustrates the time course of superoxide generation and Figure 2B, the fluorescence intensities obtained at the end of the experiment. Pre-incubation for 1 h with 50 uM ryanodine prevented the increase in probe fluorescence produced by AβOs, revealing that RyR-mediated Ca²⁺ release is essential to this process. In accord, neurons treated at the microscope stage with 4-CMC (0.5 mM), a RyR-channel agonist, exhibited an increase in MitoSOX fluorescence (Figures 2A,B), which was significantly higher than the increase produced by AβOs. The addition of vehicle did not change probe fluorescence.

To detect mitochondrial hydrogen peroxide generation, primary hippocampal cultures were transiently transfected with the Hyper^TM-Mito plasmid. This plasmid codes for the mitochondrial protein HyPer-mito that has a circularly permuted yellow fluorescent protein inserted into the regulatory domain of the prokaryotic hydrogen peroxide-sensing protein (OxyR) (Belousov et al., 2006), allowing selective detection of mitochondrial hydrogen peroxide production in living cells. Figure 2C shows that addition of 500 nM AβOs produced within minutes a fluorescence increase in primary hippocampal neurons, indicating that AβOs promoted mitochondrial hydrogen peroxide generation (Figure 2C). In contrast, neurons in cultures pre-incubated for 1 h with 50 uM ryanodine to prevent RyR-mediated Ca²⁺ release did not exhibit changes in probe fluorescence in response to AβOs (Figure 2C). The quantification of the fluorescence recorded 1200 s after AβOs addition is shown in Figure 2D. Altogether, the combined results illustrated in Figure 2 indicate that AβOs-induced RyR-mediated Ca²⁺ release has a key role in AβOs-induced mitochondrial superoxide and hydrogen peroxide generation.

We evaluated the effects of the general antioxidant NAC on the mitochondrial superoxide production induced by AβOs. For this purpose, neurons were pre-incubated for 1 h with NAC (10 mM), and then AβOs were added at the microscope stage. Figure 3A shows representative fluorescence images of mitochondrial superoxide generation recorded before (left) and 1500 s (right) after AβOs addition, in the presence or absence of NAC. As illustrated in these images, NAC prevented the mitochondrial superoxide increase elicited by AβOs. The quantification of the results from several experiments indicates that NAC completely prevented the superoxide increase in the mitochondria (Figures 3B,C). In agreement with these findings, cultures pre-incubated for 2 h with the mito-protector agent EUK-134 (20 μM) exhibited a significant decrease in neuronal superoxide levels following AβOs addition, as observed in the pseudo color images shown in Figure 3D. Figure 3E illustrates the quantification of the kinetics of the MitoSox fluorescence changes and Figure 3F, the endpoint fluorescence values.

The effects of NAC and EUK-134 on AβOs-induced mitochondrial hydrogen peroxide production were tested next. For this purpose, cultures were pre-incubated with NAC and EUK-134 as described above. Figure 4A shows representative fluorescence images of neuronal mitochondrial H₂O₂ generation before (left) and 1200 s (right) after AβOs addition, in the presence or absence of NAC (Figure 4A) or EUK-134 (Figure 4D). Both NAC and EUK-134 prevented the increase in mitochondrial H₂O₂ levels caused by AβOs addition. Quantification of the kinetics of hydrogen peroxide generation is shown in Figures 4B,E, while the fluorescence intensities measured at the endpoint of the experiments are shown in Figures 4C,F.

Based on these combined findings, we conclude that both antioxidants, NAC and EUK-134, prevent the increases in mitochondrial superoxide and hydrogen peroxide levels induced by AβOs.

We described previously that AβOs promote the fragmentation of the mitochondrial network, and that the antioxidant NAC prevents AβOs-induced mitochondrial fragmentation by preventing RyR-mediated Ca²⁺ release (SanMartín et al., 2012a). To quantify independently changes in mitochondrial network in the soma and neurites, we performed a detailed morpho-topological analysis of the mitochondrial network before and after the exposure to 500 nM AβOs. This analysis defined four mitochondrial clusters according to their volume. The mean volume of single mitochondria (0.15 ± 0.04 μm³) was used to define all clusters (see Materials and Methods). As previously described (SanMartín et al., 2014), the mitochondrial network of hippocampal neurons in control conditions is highly interconnected, with elongated mitochondria that extend across the cell body and neuronal projections. Figure 5A illustrates mature hippocampal neurons (18–21 DIV) displaying a characteristic organization of their mitochondrial network, which may reflect specific cellular demands in the neuronal soma (Figure 5A) and neurites (Figure 5B). Compared to a representative control neuron (Figures 5A,B), the continuity of the mitochondrial network of a neuron incubated for 24 h with 500 nM AβOs exhibited a loss, and the proportion of small mitochondria in soma and neurites increased (Figures 5C,D). The quantification of morpho-topological analysis of mitochondrial fragmentation revealed that AβOs decreased the fraction of the biggest clusters (>7.5 μm³) and increased the proportion of the intermediate clusters of mitochondria (1.5–7.5 μm³), in the soma (Figure 5E) as well as in the neurites (Figure 5F). This effect was dose-dependent, but we set the subsequent experiments with the lower concentration of AβOs, 500 nM, which we have reported to be sub-lethal (Paula-Lima et al., 2011).

We investigated next the effects of the mitochondrial antioxidant EUK-134 on the mitochondrial fragmentation induced by AβOs. Analysis of fixed control neurons stained with MitoTracker Orange revealed that only 5% of primary hippocampal neurons contained fragmented mitochondria, while most of the neurons exhibited filamentous mitochondria in neurites and soma (Figure 6A). In contrast, a significantly higher percentage of neurons (67%) treated with 500 nM AβOs for 24 h contained punctuate mitochondria, exposing noteworthy fragmentation of the mitochondrial network (Figure 6B). Incubation with 20 μM EUK-134 before AβOs treatment significantly decreased (from 67 to 16%, Figure 6C) the fraction of neurons exhibiting fragmented mitochondria; 20 μM EUK-134 by itself (Figure 6D) did not elicit significant changes in the content of fragmented mitochondrial (11%) when compared to the controls (Figure 6E). Theses results indicate that the EUK-134 mitochondrial antioxidant prevents mitochondrial fragmentation induced by AβOs.

Oligotransfection of primary hippocampal cultures with an oligodeoxynucleotide against RyR2 (ODN RyR2) reduced by 50% RyR2 mRNA (Supplementary Figure S2A) and protein contents (Supplementary Figure S2B), determined in homogenates of the whole primary culture. We studied next the impact of RyR2 knockdown on agonist-induced RyR-mediated cytoplasmic Ca²⁺ signals elicited by 4-CMC, and found that neurons in ODN RyR2 transfected cultures exhibited 40% lower Ca²⁺ signals when compared to neurons present in cultures transfected with the scrambled oligonucleotide (ODN Scr) (Supplementary Figure S2C).

To evaluate if RyR2 knockdown affected AβOs-induced cytoplasmic and mitochondrial Ca²⁺ signals, we used neuronal cultures transfected with ODN RyR2 or ODN Scr and loaded with Fluo4 and Rhod2 (for a representative experiment, see Supplementary Figure S3). As reported previously (Sanz-Blasco et al., 2008; Paula-Lima et al., 2011; SanMartín et al., 2012a; Hedskog et al., 2013), we confirmed that treatment with AβOs caused an increase in both cytoplasmic and mitochondrial Ca²⁺ signals. The fluorescence of both dyes increased in response to 50 mM KCl addition at the end of the experiment, evidencing that neurons were still active after all the experimental manipulations (Supplementary Figure S3). The fluorescence intensities observed in ODN RyR2 and ODN Scr-treated neurons revealed that ODN RyR2 transfection caused a significant reduction in both the cytoplasmic (Figure 7A) and the mitochondrial (Figure 7B) Ca²⁺ signals induced by AβOs. The quantification of the last fifteen seconds of the average of three experiments shows that these differences are statistically significant (Figures 7C,D).

We further investigated the effects of transfection with ODN RyR2 on AβOs-induced mitochondrial fragmentation. Analysis of control neurons transfected with ODN Scr and loaded with MitoTracker Orange, revealed that about 15% of primary hippocampal neurons contained fragmented mitochondria, whereas 24 h after 500 nM AβOs addition 53% exhibited punctuate mitochondria, revealing fragmentation of the mitochondrial network (Figures 8A,B). Transfection with ODN RyR2 markedly reduced almost to zero the percentage of neurons that exhibited a mitochondrial punctuate pattern, even after AβOs treatment (Figures 8A,B). Based on these results, we propose that Ca²⁺ release mediated by the RyR2 isoform plays a central role in AβOs-induced mitochondrial fragmentation.