The Role of Intracellular Ca2+ and Mitochondrial ROS in Small Aβ1-42 Oligomer-Induced Microglial Death

Alzheimer’s disease (AD) is the most common form of dementia worldwide, and it contributes up to 70% of cases. AD pathology involves abnormal amyloid beta (Aβ) accumulation, and the link between the Aβ1-42 structure and toxicity is of major interest. NMDA receptors (NMDAR) are thought to be essential in Aβ-affected neurons, but the role of this receptor in glial impairment is still unclear. In addition, there is insufficient knowledge about the role of Aβ species regarding mitochondrial redox states in neurons and glial cells, which may be critical in developing Aβ-caused neurotoxicity. In this study, we investigated whether different Aβ1-42 species—small oligomers, large oligomers, insoluble fibrils, and monomers—were capable of producing neurotoxic effects via microglial NMDAR activation and changes in mitochondrial redox states in primary rat brain cell cultures. Small Aβ1-42 oligomers induced a concentration- and time-dependent increase in intracellular Ca2+ and necrotic microglial death. These changes were partially prevented by the NMDAR inhibitors MK801, memantine, and D-2-amino-5-phosphopentanoic acid (DAP5). Neither microglial intracellular Ca2+ nor viability was significantly affected by larger Aβ1-42 species or monomers. In addition, the small Aβ1-42 oligomers caused mitochondrial reactive oxygen species (mtROS)-mediated mitochondrial depolarization, glutamate release, and neuronal cell death. In microglia, the Aβ1-42-induced mtROS overproduction was mediated by intracellular calcium ions and Aβ-binding alcohol dehydrogenase (ABAD). The data suggest that the pharmacological targeting of microglial NMDAR and mtROS may be a promising strategy for AD therapy.


Introduction
Alzheimer's disease (AD) is the most common cause of dementia in older adults, with an estimated 55 million cases worldwide, which are predicted to reach 150 million by 2050 [1]. AD pathogenesis involves the abnormal accumulation of amyloid beta (Aβ) peptides, and the action mechanism of various Aβ 1-42 assemblies is of major interest. The neurotoxicity of Aβ 1-42 oligomers includes the neuronal N-methyl D-aspartate receptor (NMDAR)-dependent disruption of Ca 2+ homeostasis, the destabilization of the neuronal cell membrane [2,3], the promotion of inflammatory reactions [4,5], or mitochondrial injury [6,7]. Most studies refer to neuronal NMDAR as the primary target in the Aβ damage pathway, yet the role of NMDAR in the microglia-related events in AD remains unclear. We have previously reported that small Aβ 1-42 oligomers (but not larger aggregates or monomers) induce neuronal and microglial plasma membrane depolarization and neuronal death; however, only microglial depolarization was prevented by the NMDAR blocker MK801 [8]. Recently, it has been shown that NMDA stimulates microglial proliferation and morphological transformation toward an activated state that is characterized

The Role of NMDARs in Aβ 1-42 -Induced Microglial Cytoplasmic Calcium Increase and Death
First, we investigated whether Aβ 1-42 species of different aggregation states might cause changes in the cytosolic calcium ion concentration in primary microglial cultures ( Figure 1). After 15 min of incubation of the cultured microglial cells with small Aβ  oligomers, the level of intracellular Ca 2+ -dependent fluorescence increased by 18% compared with the control, and after 45 and 90 min, the fluorescence levels rose to 36% and 50%, respectively (Figure 1b,c,i). In contrast, the level of intracellular Ca 2+ in microglia remained unchanged during a 90-minute treatment with Aβ 1-42 monomers, large oligomers, and fibrils.
Next, we evaluated whether small Aβ 1-42 oligomers induced microglial death. As shown in Figure 2a, in pure microglial cultures, small Aβ 1-42 oligomers caused rapid necrosis in a concentration-and time-dependent manner: after 30 min of incubation with 0.5 µM and 1 µM small Aβ 1-42 oligomers, the levels of necrotic (PI-positive) cells were about 10% and increased gradually over 4.5 h of incubation to eventually reach 22% and 61% necrosis at 0.5 µM and 1 µM concentrations, respectively. After 24 h, about 90% of the cells were necrotic in the 1 µM small Aβ 1-42 oligomer-treated primary microglial cultures (Figure 2a,b); in the 0.5 µM small Aβ 1-42 oligomer-treated primary microglial cultures, about 60% necrosis was observed (Figure 2a). Large oligomers and fibrilar Aβ 1-42 did not cause necrosis during the 24 h treatment of microglial cultures (Figure 2a). The preparation of Aβ species involves the solvent hexafluoroisopropanol (HFIP), which is evaporated in further steps; however, trace amounts of HFIP in Aβ solutions might be present. Therefore, we performed additional control experiments in which cultures were treated with solvent solutions (HFIP) prepared in exactly the same way as Aβ oligomers but in the absence of peptides. HFIP applied at the same concentration as in the Aβ 1-42 preparations did not influence the microglial viability (Figure 2a 2+ was assessed by loading microglia with Fluo-3AM dye and measuring the fluorescence inten sity of microscope images. In (a-h), there are representative images of the cells; the scale bar (10 µ m) is the same for all images. In (i)-a quantitative evaluation of Ca 2+ -dependent fluorescence i microglial cells. HFIP-hexafluoroisopropanol is a solvent used for Aβ preparation. BAPTA-1, Bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetrakis(acetoxymethyl ester) was used as negative control of the assay. ATP was used as a positive control. The fluorescence intensity in eac experiment was expressed as a percentage of ATP-stimulated fluorescence, and BAPTA-treated ce fluorescence level was eliminated as a nonspecific baseline. The data are presented as averages wit a standard deviation of 5-9 experiments performed in triplicates (5 images each, i.e., 15 images pe sample in total). ***, **-statistically significant difference compared with the HFIP control at th same time point. ***-p < 0.001, **-p < 0.01.
Next, we evaluated whether small Aβ1-42 oligomers induced microglial death. A is the same for all images. In (i)-a quantitative evaluation of Ca 2+ -dependent fluorescence in microglial cells. HFIP-hexafluoroisopropanol is a solvent used for Aβ preparation. BAPTA-1,2-Bis(2-aminophenoxy)ethane-N,N,N ,N -tetraacetic acid tetrakis(acetoxymethyl ester) was used as a negative control of the assay. ATP was used as a positive control. The fluorescence intensity in each experiment was expressed as a percentage of ATP-stimulated fluorescence, and BAPTA-treated cell fluorescence level was eliminated as a nonspecific baseline. The data are presented as averages with a standard deviation of 5-9 experiments performed in triplicates (5 images each, i.e., 15 images per sample in total). ***, **-statistically significant difference compared with the HFIP control at the same time point. ***-p < 0.001, **-p < 0.01. further steps; however, trace amounts of HFIP in Aβ solutions might be present. The fore, we performed additional control experiments in which cultures were treated w solvent solutions (HFIP) prepared in exactly the same way as Aβ oligomers but in absence of peptides. HFIP applied at the same concentration as in the Aβ1-42 preparatio did not influence the microglial viability (Figure 2a,b). In there are representative images of microglial nuclei stained for viability detection w Hoechst33342 and PI after 24 h treatment with the small Aβ1-42 oligomers. Cell cultures were a lyzed at 20X magnification. ***, **-statistically significant difference compared with the contro the same time point; ***-p < 0.001; **-p < 0.01. The data described above show that small Aβ1-42 oligomers elevated intracellular C and induced cell death in pure microglial cultures. Next, we tested the effects of differ Aβ1-42 preparations on microglia in mixed neuronal-glial cell cultures. After 15 min of cubation, the average microglial calcium-dependent fluorescence in the small Aβ1-42 olig mer-treated cultures was 7% higher compared to the untreated control ( Figure 3a). Af 45 min, the difference from the control increased to 20% and remained similar after min. In contrast, there was no significant difference between the control and monome large oligomer-, and fibril-treated cultures.
The effect of small Aβ1-42 oligomers on microglial viability was also observed in mix neuronal-glial cell cultures. The percentage of necrotic microglia after a 30 min of tre ment with small Aβ1-42 oligomers was 17% and gradually increased in time to reach 3 after 4.5 hours and 53% after 24 h (Figure 3b,d,e). There were almost no necrotic cells the controls (Figure 3c) and other Aβ1-42 species-treated samples. Thus, the levels of ce killed by small oligomers were significantly higher after all investigated periods. . HFIP-a solvent used for Aβ preparation. In (b), there are representative images of microglial nuclei stained for viability detection with Hoechst33342 and PI after 24 h treatment with the small Aβ 1-42 oligomers. Cell cultures were analyzed at 20X magnification. ***, **-statistically significant difference compared with the control at the same time point; ***-p < 0.001; **-p < 0.01. The data described above show that small Aβ 1-42 oligomers elevated intracellular Ca 2+ and induced cell death in pure microglial cultures. Next, we tested the effects of different Aβ 1-42 preparations on microglia in mixed neuronal-glial cell cultures. After 15 min of incubation, the average microglial calcium-dependent fluorescence in the small Aβ  oligomer-treated cultures was 7% higher compared to the untreated control ( Figure 3a). After 45 min, the difference from the control increased to 20% and remained similar after 90 min. In contrast, there was no significant difference between the control and monomer-, large oligomer-, and fibril-treated cultures.
The effect of small Aβ 1-42 oligomers on microglial viability was also observed in mixed neuronal-glial cell cultures. The percentage of necrotic microglia after a 30 min of treatment with small Aβ 1-42 oligomers was 17% and gradually increased in time to reach 37% after 4.5 h and 53% after 24 h (Figure 3b,d,e). There were almost no necrotic cells in the controls ( Figure 3c) and other Aβ 1-42 species-treated samples. Thus, the levels of cells killed by small oligomers were significantly higher after all investigated periods.
In summary, small Aβ 1-42 oligomers increased Ca 2+ levels and induced necrosis in the microglia in pure and mixed neuronal-glial cell cultures; however, the effects were more prominent in pure cultures. No Ca 2+ or viability changes were observed after treatment with other Aβ 1-42 forms, such as monomers, large oligomers, or fibrils.
We have previously found that small Aβ 1-42 oligomers induced neuronal toxicity via NMDARs [8]. In this study, we examined whether NMDARs were involved in microglial Ca 2+ elevation and cell death triggered by small Aβ 1-42 oligomers by incubating primary microglial cell cultures with the peptide in the presence of the NMDAR inhibitors MK801, memantine, or D-2-Amino-5-phosphopentanoic acid (DAP5).  In summary, small Aβ1-42 oligomers increased Ca 2+ levels and induced necrosis in t microglia in pure and mixed neuronal-glial cell cultures; however, the effects were mo prominent in pure cultures. No Ca 2+ or viability changes were observed after treatme with other Aβ1-42 forms, such as monomers, large oligomers, or fibrils.
We have previously found that small Aβ1-42 oligomers induced neuronal toxicity v NMDARs [8]. In this study, we examined whether NMDARs were involved in microgl Ca 2+ elevation and cell death triggered by small Aβ1-42 oligomers by incubating prima microglial cell cultures with the peptide in the presence of the NMDAR inhibitors MK8 memantine, or D-2-Amino-5-phosphopentanoic acid (DAP5).
All the applied inhibitors at 10 μM concentrations completely prevented the Aβ1 induced increase in microglial intracellular Ca 2+ (Figure 4a). In the cells affected by sm Aβ1-42 oligomers without the inhibitors, the Ca 2+ level was comparable to the samp All the applied inhibitors at 10 µM concentrations completely prevented the Aβ 1-42induced increase in microglial intracellular Ca 2+ (Figure 4a). In the cells affected by small Aβ 1-42 oligomers without the inhibitors, the Ca 2+ level was comparable to the samples treated with a specific NMDAR agonist 1 mM NMDA and was lower than in the cells treated with glutamate, which is an activator of NMDA, AMPA, and kainate receptors. However, the NMDAR inhibitors were not as effective in protecting the microglia from death caused by small Aβ 1-42 oligomers (Figure 4b). Although the noncompetitive NMDAR inhibitors MK801 and memantine significantly prevented small Aβ 1-42 oligomer-induced increases in the necrotic cell number after 90, 180, and 270 min of treatment, the levels of dead cells remained substantially higher than in the control cultures (compare Figure 2a,b). A competitive NMDAR inhibitor DAP5 was even less protective and significantly prevented cell death only after 180 and 270 min, but not after 90 min. Neither of the inhibitors significantly affected the microglial viability in the presence of the small Aβ 1-42 oligomers after 30 min or 24 h. The data suggest that small Aβ 1-42 oligomers (but not monomers, large oligomers, or fibrils) elevate the intracellular Ca 2+ concentration in microglia by triggering the NMDARs; however, other mechanisms contribute to this outcome in causing microglial death. levels of dead cells remained substantially higher than in the control cultures (compare Figure 2a,b). A competitive NMDAR inhibitor DAP5 was even less protective and significantly prevented cell death only after 180 and 270 min, but not after 90 min. Neither of the inhibitors significantly affected the microglial viability in the presence of the small Aβ1-42 oligomers after 30 min or 24 h. The data suggest that small Aβ1-42 oligomers (but not monomers, large oligomers, or fibrils) elevate the intracellular Ca 2+ concentration in microglia by triggering the NMDARs; however, other mechanisms contribute to this outcome in causing microglial death.

The Effect of Aβ1-42 Species on Superoxide Production by Mitochondria
In this part of the study, we investigated whether various Aβ1-42 species could stimulate mitochondrial superoxide generation in microglia and neurons in mixed neuronalglial cell cultures. As can be seen in Figure 5a,b, after 30 min of treatment with small Aβ1-42 oligomers, superoxide-dependent MitoSOX Red fluorescence in the neuronal cells was 35% higher than in the control, and it remained at this level for 1 h and increased further by 153% after 2 h (Figure 5a,b). Large Aβ1-42 oligomers and Aβ1-42 fibrils did not cause any detectable changes in the mitochondrial superoxide production. Similarly to the neuronal data, small Aβ1-42 oligomers induced a substantial increase in MitoSOX fluorescence in the microglial cells, which reached 195%, 155%, and 167% of the control after 30 min, 1 h, and 2 h of incubation, respectively (Figure 5a,c). Again, other Aβ1-42 forms did not significantly influence the mitochondrial superoxide production. In addition, small and large Aβ1-42 oligomers or fibrils did not cause any superoxide-related fluorescence in the astrocytes after 0.5-2 h of incubation in neuronal-glial cell cultures. HBSS or 2 μM Antimycin A without MitoSOX did not affect the MitoSOX-related fluorescence in the neurons and mi-

The Effect of Aβ 1-42 Species on Superoxide Production by Mitochondria
In this part of the study, we investigated whether various Aβ 1 -42 species could stimulate mitochondrial superoxide generation in microglia and neurons in mixed neuronal-glial cell cultures. As can be seen in Figure 5a,b, after 30 min of treatment with small Aβ 1-42 oligomers, superoxide-dependent MitoSOX Red fluorescence in the neuronal cells was 35% higher than in the control, and it remained at this level for 1 h and increased further by 153% after 2 h (Figure 5a,b). Large Aβ 1-42 oligomers and Aβ 1-42 fibrils did not cause any detectable changes in the mitochondrial superoxide production. Similarly to the neuronal data, small Aβ 1-42 oligomers induced a substantial increase in MitoSOX fluorescence in the microglial cells, which reached 195%, 155%, and 167% of the control after 30 min, 1 h, and 2 h of incubation, respectively (Figure 5a,c). Again, other Aβ 1-42 forms did not significantly influence the mitochondrial superoxide production. In addition, small and large Aβ 1-42 oligomers or fibrils did not cause any superoxide-related fluorescence in the astrocytes after 0.5-2 h of incubation in neuronal-glial cell cultures. HBSS or 2 µM Antimycin A without MitoSOX did not affect the MitoSOX-related fluorescence in the neurons and microglia. To get more mechanistic insights into Aβ1-42-induced superoxide production by m tochondria, we applied a set of pharmacological modulators. The possible involvement NMDARs and elevated Ca 2+ was tested with MK801 and BAPTA, a selective cell-perme ble Ca 2+ chelator. Frentizole was used to identify whether the mitochondrial superoxi was induced by Aβ-ABADs. N-Acetyl L-Cysteine (NAC) has mitochondria-specific an oxidant effects [20] and was applied to examine the importance of the mitochondrial red state, and Apocynin, an inhibitor of NADPH oxidase, was appliedto verify the role of RO produced by this enzyme. None of the compounds (except the selective mitochondr superoxide scavenger MitoTEMPO) affected the small Aβ1-42 oligomer-induced neuron mitochondrial superoxide (Figure 5a,d). However, the microglial Aβ1-42-induced mit To get more mechanistic insights into Aβ 1-42 -induced superoxide production by mitochondria, we applied a set of pharmacological modulators. The possible involvement of NMDARs and elevated Ca 2+ was tested with MK801 and BAPTA, a selective cell-permeable Ca 2+ chelator. Frentizole was used to identify whether the mitochondrial superoxide was induced by Aβ-ABADs. N-Acetyl L-Cysteine (NAC) has mitochondria-specific antioxidant effects [20] and was applied to examine the importance of the mitochondrial redox state, and Apocynin, an inhibitor of NADPH oxidase, was appliedto verify the role of ROS produced by this enzyme. None of the compounds (except the selective mitochondrial superoxide scavenger MitoTEMPO) affected the small Aβ 1-42 oligomer-induced neuronal mitochondrial superoxide (Figure 5a,d). However, the microglial Aβ 1-42 -induced mitochondrial superoxide was significantly reduced by the Frentizole, NAC, and BAPTA (and MitoTEMPO), but not by the Apocynin and MK801 (Figure 5e). The results indicate that Aβ-mediated mtROS overproduction is differentially regulated in neurons and microglia in mixed neuronal-glial cultures and point to microglial intracellular Ca 2+ -and ABADmediated mitochondrial oxidative stress.
For evaluating whether small Aβ 1-42 oligomer-induced mtROS cause mitochondrial depolarization and whether it is related to ABAD activation, we measured the mitochondrial membrane potential using MitoTracker Orange CM-H2TMRos in the presence of MitoTEMPO or Frentizole. The exposure of the neuronal-glial cultures to small Aβ 1-42 oligomers for 1 h caused a 25% decrease in the membrane potential compared to the control (Figure 6a,b). The membrane potential was restored by MitoTEMPO (to 90%) but not with Frentizole (Figure 6a,b), thus suggesting no considerable role of the ABADs in the Aβ-mediated decrease in mitochondrial activity. In addition, MitoTEMPO was not effective in preventing Antimycin A-induced mitochondrial depolarization (Figure 6b), most likely because the latter was caused by respiratory chain complex III inhibition but not by the mtROS. Thus, the result was small Aβ 1-42 -induced mtROS-dependent mitochondrial depolarization in the neuronal-glial cell culture. For evaluating whether small Aβ1-42 oligomer-induced mtROS cause mitochond depolarization and whether it is related to ABAD activation, we measured the mitoch drial membrane potential using MitoTracker Orange CM-H2TMRos in the presence MitoTEMPO or Frentizole. The exposure of the neuronal-glial cultures to small Aβ1-42 gomers for 1 h caused a 25% decrease in the membrane potential compared to the cont (Figure 6a,b). The membrane potential was restored by MitoTEMPO (to 90%) but not w Frentizole (Figure 6a,b), thus suggesting no considerable role of the ABADs in the A mediated decrease in mitochondrial activity. In addition, MitoTEMPO was not effect in preventing Antimycin A-induced mitochondrial depolarization (Figure 6b), most lik because the latter was caused by respiratory chain complex III inhibition but not by mtROS. Thus, the result was small Aβ1-42-induced mtROS-dependent mitochondrial polarization in the neuronal-glial cell culture.

The Role of Mitochondrial ROS in Small Aβ1-42 Oligomer-Induced Neuronal and Microgli Death
We have previously shown that small Aβ1-42 oligomers induce the opening of mi chondrial permeability transition pore in neurons, thereby leading to mitochondrial d function and cell death [8]. Mitochondrial permeability transition pore sensitivity to C is modulated by mtROS [21], thus suggesting that Aβ1-42-affected neurons could be, principle, rescued by mitochondrially targeted antioxidants or by suppressing ABADs. test this hypothesis, neuronal-glial cultures were preincubated with MitoTEMPO or Fr tizole and then treated with 1 μM small Aβ1-42 oligomers for 24 h. Aβ1-42 reduced the n ronal viability to 43%, and both Frentizole and MitoTEMPO significantly prevented t

The Role of Mitochondrial ROS in Small Aβ 1-42 Oligomer-Induced Neuronal and Microglial Death
We have previously shown that small Aβ 1-42 oligomers induce the opening of mitochondrial permeability transition pore in neurons, thereby leading to mitochondrial dysfunction and cell death [8]. Mitochondrial permeability transition pore sensitivity to Ca 2+ is modulated by mtROS [21], thus suggesting that Aβ 1-42 -affected neurons could be, in principle, rescued by mitochondrially targeted antioxidants or by suppressing ABADs. To test this hypothesis, neuronal-glial cultures were preincubated with MitoTEMPO or Frentizole and then treated with 1 µM small Aβ 1-42 oligomers for 24 h. Aβ 1-42 reduced the neuronal viability to 43%, and both Frentizole and MitoTEMPO significantly prevented this loss of viability, thus resulting in 72% and 81%, respectively, of the cells remaining alive (Figure 7a). Twenty-four-hour treatment with small oligomers reduced the microglial viability to 47% of the control (Figure 3b), and the decrease was not MitoTEMPO-or Frentizole-sensitive. and Frentizole on extracellular glutamate levels in Aβ1-42 oligomer-treated neuronal-g cell culture medium. Indeed, 1 h treatment with small Aβ1-42 oligomer elevated the ex cellular glutamate from 2.1 µ M in the control to 3.9 µ M in the treated cultures ( Figure 7 The pretreatment with MitoTEMPO or Frentizole significantly prevented glutamate creases by causing the levels to drop to 2.6 and 2.2 µ M, respectively, thereby indicat glutamate release mediation by mtROS and ABAD. In death experiments (a), cells were exposed to 1 μM Aβ1-42 oligomers with or without MitoTEM (MitoT) or Frentizole (FR) for 24 h. Cell death was quantified by assessing nuclei morphology a staining with PI and Hoechst 33342. Cells were counted in at least 5 microscopic fields per well ( wells per treatment). Data are expressed as percentage of specific neuronal cells of the total num of neuronal cells per field. Neurons were recognized according to characteristic morphology us phase contrast microscopy. For extracellular glutamate measurements (b), cells were treated wi μM small Aβ1-42 oligomers for 1 h with or without MitoTEMPO or Frentizole. Glutamate in cult medium was evaluated by Amplex Red glutamic acid/glutamate oxidase assay kit. Glutamate c centrations were calculated with L-glutamic acid standard curve. All the quantitative data are p sented as averages with a standard deviation of 4-8 experiments. ***, statistically significant dif ence compared with the control; ###-with Aβ; #-p < 0.05; ***, ###-p < 0.001.

Discussion
In this study, we used Aβ1-42 species of different oligomerization states-small a large oligomers, fibrils, and monomers-to investigate Aβ neurotoxicity mechanisms rat brain primary cell cultures. The main finding of the study was that the microglial culture was extremely sensitive to toxic small Aβ oligomers. Small Aβ1-42 oligomers tr gered an NMDAR-mediated increase in the intracellular Ca 2+ in microglial cells, wh partially correlated with microglial necrosis. We have previously shown that small oli mers, but not other Aβ forms, induced neuronal necrosis in neuronal-glial cell cultu [22], and now we demonstrate that small oligomers cause the concentration-and tim dependent necrosis of microglial cells. Dystrophic microglia have been found in ag AD-and frontotemporal-lobar-degeneration-affected brains [23][24][25]. Amyloid plaquesociated microglia have impaired Ca 2+ signaling and increased apoptosis rates in the A mouse models [26,27]; however, there are no data about the link between microglial de and neuronal damage in Aβ pathology. MtROS might modulate glutamatergic signaling leading to calcium overload and excitotoxicity [15]; therefore, we further investigated the potential effects of MitoTEMPO and Frentizole on extracellular glutamate levels in Aβ 1-42 oligomer-treated neuronal-glial cell culture medium. Indeed, 1 h treatment with small Aβ 1-42 oligomer elevated the extracellular glutamate from 2.1 µM in the control to 3.9 µM in the treated cultures (Figure 7b). The pretreatment with MitoTEMPO or Frentizole significantly prevented glutamate increases by causing the levels to drop to 2.6 and 2.2 µM, respectively, thereby indicating glutamate release mediation by mtROS and ABAD.

Discussion
In this study, we used Aβ 1-42 species of different oligomerization states-small and large oligomers, fibrils, and monomers-to investigate Aβ neurotoxicity mechanisms in rat brain primary cell cultures. The main finding of the study was that the microglial cell culture was extremely sensitive to toxic small Aβ oligomers. Small Aβ 1-42 oligomers triggered an NMDAR-mediated increase in the intracellular Ca 2+ in microglial cells, which partially correlated with microglial necrosis. We have previously shown that small oligomers, but not other Aβ forms, induced neuronal necrosis in neuronal-glial cell cultures [22], and now we demonstrate that small oligomers cause the concentration-and time-dependent necrosis of microglial cells. Dystrophic microglia have been found in aged, AD-and frontotemporallobar-degeneration-affected brains [23][24][25]. Amyloid plaque-associated microglia have impaired Ca 2+ signaling and increased apoptosis rates in the AD mouse models [26,27]; however, there are no data about the link between microglial death and neuronal damage in Aβ pathology.
We determined that small Aβ 1-42 oligomers caused mtROS, mitochondrial depolarization, high extracellular glutamate levels, and neuronal death in neuronal-glial cell cultures; and mitochondrial antioxidants prevented all these events. Aβ-induced mtROS overproduction in microglia was mediated by intracellular Ca 2+ and ABADs. The interaction of Aβ with ABAD enhances mitochondrial oxidative stress, mitochondrial toxicity, and cognitive decline in AD patients and transgenic mice [16], whereas the inhibition of ABAD-Aβ interaction reduces oxidative stress, as well as improves mitochondrial functions and spatial memory in a mouse AD model [28]. Several studies indicate that the molecular pathways under redox control may be implicated in the Aβ pathogenesis targeting microglia [29]. In our study, Apocynin was not effective in the Aβ-mediated ROS production in microglia, and this result suggests that mtROS, not NADPH oxidase derived-ROS, might be responsible for mediating microglial damage in response to small Aβ oligomers.
Mitochondrial dysfunction in microglia plays a significant role in the pathogenesis of AD and other neurological disorders [30]. Previously, we found that small Aβ 1-42 oligomers decreased the capacity of isolated brain mitochondria to retain calcium, thus inducing mitochondrial permeability transition pore opening [8]. The overload of calcium in mitochondria leads to the opening of mitochondrial permeability transition pore, which is an initial step in activating necrotic and apoptotic cell death [31,32]. Moreover, ROS overproduction in neurons and glia can serve as an additional trigger for pore opening [33]. One of the possible causes of enhanced mtROS might be the direct impairment of complex I activity by Aβ [34].
Here, the Aβ 1-42 oligomers caused a rapid increase in the NMDAR-dependent cytosolic calcium in microglia and a calcium-dependent increase in the mtROS and extracellular glutamate, thereby inducing neuronal and microglial death. The overproduction of mtROS in neurons alters the NMDARs, thus leading to impaired glutamatergic signaling, calcium overload, and excitotoxicity [15]. Oxidative glutamate toxicity has been hypothesised as the component of excitotoxicity-initiated cell death when parts of the neurons are killed by glutamate exposure directly in an antioxidant-dependent manner, and another part is damaged due to the activation of the NMDARs [35]. Aβ 1-42 oligomers stimulated the excessive formation of mtROS through a mechanism requiring NMDAR activation in hippocampal neuronal cultures [36]. Ionotropic glutamate receptors are also expressed in microglial cells, and their stimulation leads to a Ca 2+ influx [37] and ROS production [38]. In addition, glutamate might cause an extracellular Ca 2+ influx via Na + /Ca 2+ exchanger [39].
One of the possible explanations for NMDAR-mediated mitochondrial response induced by Aβ oligomers could be the involvement of mitochondria-associated membranedependent NMDARs. It has been shown that the activation of NMDARs may trigger efficient Ca 2+ release from endoplasmic reticulum (ER) stores [40,41], and the NMDA receptor antagonist DAP5 may suppress ER Ca 2+ release and mitochondrial Ca 2+ import in mammalian neurons [42]. Earlier investigations on postmortem brains affected by AD have determined the increase of contact sites between the ER and mitochondria, as well as the upregulation of proteins of the mitochondria-associated membranes [43]. Additionally, Aβ-induced NMDAR activation can cause mitochondrial dysfunction via Ca 2+ flux from the ER in primary cortical neurons [44]. Thus, the activation of the NMDAR is required for Ca 2+ release from ER stores and the Ca 2+ import into neuronal mitochondria. However, it is not known whether NMDARs are important for ER/mitochondria Ca 2+ fluxes in microglial cells.
Our findings suggest that the downstream effects of small Aβ oligomers leading to microglial calcium overload and NMDAR activation, microglial and neuronal mitochondrial damage, and glutamate release may contribute to the mitochondrial oxidative stress hypothesis and may provide a basis for the development of therapeutic measures for delaying the progression of AD. The effects of small Aβ oligomer-induced toxicity on neurons and microglia, as well as their potential pharmacological modulation sites, are summarized in a scheme in Figure 8.
laying the progression of AD. The effects of small Aβ oligomer-induced toxicity on neurons and microglia, as well as their potential pharmacological modulation sites, are summarized in a scheme in Figure 8. Only small Aβ1-42 peptide oligomers are toxic to neurons and microglia; the toxicity mechanisms in these two cell types share some similarities but also have differences. Both microglia and neurons suffer from NMDA-dependent Ca 2+ increases followed by mitochondrial ROS increases, as well as the loss of mitochondrial inner membrane potential ΔΨ (most likely due to the Ca 2+ and ROS-induced mitochondrial permeability transition pore mPTP). The mitoROS feedback to NMDARs via glutamate (GLUT) increase can lead to neuronal death due to excitotoxicity and deenergization. In microglia, NMDAR is not the only pathway of Aβ1-42-induced Ca 2+ entry, and the importance of ABAD and the mitochondrial redox state is evident. Small Aβ1-42 oligomer-mediated damage in neurons and microglia can be pharmacologically controlled by NMDAR inhibitors MK801, memantine, and DAP5, as well asmitochondrial ROS scavenger MitoTEMPO. In microglia, this damage can be prevented by ABAD inhibitor Frentizole and mitochondrial redox state modulator N-Acetyl L-Cysteine (NAC). Only small Aβ 1-42 peptide oligomers are toxic to neurons and microglia; the toxicity mechanisms in these two cell types share some similarities but also have differences. Both microglia and neurons suffer from NMDA-dependent Ca 2+ increases followed by mitochondrial ROS increases, as well as the loss of mitochondrial inner membrane potential ∆Ψ (most likely due to the Ca 2+ and ROS-induced mitochondrial permeability transition pore mPTP). The mitoROS feedback to NMDARs via glutamate (GLUT) increase can lead to neuronal death due to excitotoxicity and deenergization. In microglia, NMDAR is not the only pathway of Aβ 1-42 -induced Ca 2+ entry, and the importance of ABAD and the mitochondrial redox state is evident. Small Aβ 1-42 oligomer-mediated damage in neurons and microglia can be pharmacologically controlled by NMDAR inhibitors MK801, memantine, and DAP5, as well asmitochondrial ROS scavenger MitoTEMPO. In microglia, this damage can be prevented by ABAD inhibitor Frentizole and mitochondrial redox state modulator N-Acetyl L-Cysteine (NAC).

Cell Cultures and Treatments
The procedures used in this study were approved by The State Food and Veterinary Service of the Republic of Lithuania in accordance with European Convention for the protection of vertebrate animals used for experimental and other purposes. The rats were bred and maintained at Lithuanian University of Health Sciences Animal House under controlled conditions. Wistar rats were anaesthetised by CO 2 , followed by cervical dislocation. Primary pure microglial cultures from rat cerebral cortices were prepared from 7-8-day-old Wistar rats of both genders as described in [45].
Primary neuronal-glial cultures from rat cerebellum were prepared from 7-8-day-old Wistar rats as described in [46]. Cells were grown for 7 days in vitro (DIV) before exposure to Aβ species. The neuronal-glial culture contained 80% neurons and 7% astrocytes as assessed by cellular morphology and 13% microglia indicated by staining with Isolectin GS-IB4 conjugated with Alexa Fluor488.

Measurement of Intracellular Calcium Concentration
Intracellular calcium concentration was assessed by fluorescence microscopy using the Fluo-3AM dye. Cultures were exposed to Aβ 1-42 compounds (monomers, small oligomers, large oligomers, and fibrils), at 1.0 µM, and Fluo-3AM at 2.0 µM for 15, 45, and 90 min; all were visualized by OLYMPUS IX71S1F-3 fluorescent microscope (at 495/521 nm wavelengths). To investigate the role of NMDARs, the cell cultures were preincubated with NMDAR blockers (memantine, MK801, and DAP5) all at 10 µM for 30 min. A total of 1 mM ATP-induced Fluo-3AM fluorescence was evaluated as maximal and was considered 100%. Fluo-3AM fluorescence after treatment with BAPTA (6 µM) was considered minimal (0%). For calcium evaluation in microglia in neuronal-glial cultures, the cells were treated and assessed the same way as in pure culture, and fluorescence intensity was measured in individual microglial cells that were traced according to their brightfield images, as demonstrated in Figure 9.

Cell Viability Assay
The viability of cells in cultures was assessed by propidium iodide (PI, 7 µM) and Hoechst 33342 (4 µg/mL) staining using a fluorescence microscope OLYMPUS IX71S1F-3, as described in [22]. Neurons were recognised according to characteristic morphology (round in shape, small somata with a few dendrites) using phase-contrast microscopy. PI-positive cells were classified as necrotic, and cells with condensed/fragmented nuclei (Hoechst 33342, bright blue) as apoptotic. Microglial cells were identified by Isolectin GS-IB4 conjugated with AlexaFluor488 (7 ng/mL) staining. Neuronal and microglial cell numbers in neuronal-glial cultures were assessed by counting cells in at least 5 microscopic images/wells by means of ImageJ software. Cell viability was expressed as the percentage of viable, necrotic, or apoptotic cells of the total number of specific cells in the microscopic image. The number of neurons/microglia in the control group was considered as 100%. Cultures were treated with Aβ 1-42 species (0.5 or 1 µM) in the presence/absence of NMDAR blockers memantine, MK801, or DAP5. The blockers were applied at 10 µM concentrations and added 30 min before the Aβ treatment. MitoTEMPO (10 µM) or Frentizole (5 µM), where applied, were added 15 min before the Aβ treatment. List of inhibitors used in the study is indicated in Table 1.

Cell Viability Assay
The viability of cells in cultures was assessed by propidium iodide (PI, 7 µ M) and Hoechst 33342 (4 µ g/mL) staining using a fluorescence microscope OLYMPUS IX71S1F-3, as described in [22]. Neurons were recognised according to characteristic morphology (round in shape, small somata with a few dendrites) using phase-contrast microscopy. PIpositive cells were classified as necrotic, and cells with condensed/fragmented nuclei (Hoechst 33342, bright blue) as apoptotic. Microglial cells were identified by Isolectin GS-IB4 conjugated with AlexaFluor488 (7 ng/mL) staining. Neuronal and microglial cell numbers in neuronal-glial cultures were assessed by counting cells in at least 5 microscopic images/wells by means of ImageJ software. Cell viability was expressed as the percentage of viable, necrotic, or apoptotic cells of the total number of specific cells in the microscopic image. The number of neurons/microglia in the control group was considered as 100%. Cultures were treated with Aβ1-42 species (0.5 or 1 µ M) in the presence/absence of NMDAR blockers memantine, MK801, or DAP5. The blockers were applied at 10 µ M concentrations and added 30 min before the Aβ treatment. MitoTEMPO (10 μM) or Frentizole (5 μM), where applied, were added 15 min before the Aβ treatment. List of inhibitors used in the study is indicated in Table 1.

Inhibitor
Mechanism of Action
BAPTA A selective cell-permeable Ca 2+ chelator (information provided by manufacturer).

MitoTEMPO
A mitochondrially targeted antioxidant and specific scavenger of mitochondrial superoxide (information provided by manufacturer).

Frentizole
An inhibitor of amyloid beta peptide binding alcohol dehydrogenase (ABAD)-Aβ interaction (information provided by manufacturer).
N-Acetyl L-Cysteine A disulfide reductant, a direct scavenger of oxidants, and a driver of glutathione synthesis [20,49].
Apocynin A potent and selective inhibitor of NADPH oxidase [50].
Antimycin A An inhibitor of mitochondrial cytochrome bc1 (complex III), often used as a positive control for superoxide generation.

Mitochondrial Membrane Potential Assessment
Mitochondrial membrane potential in neuronal-glial cultures was monitored using the fluorescent dye MitoTracker ® Orange CM-H2TMRos (200 nM) according to the manufacturer protocol. Cultures were treated with small Aβ 1-42 oligomers plus/minus MitoTEMPO or Frentizole, as described in Section 4.5. The ability of the dye to respond to plasma membrane potential changes was tested by a mitochondrial membrane potential destabilizer Antimycine A (2 µM) for 30 min. Pictures of at least 5 randomly selected fields per well were taken using fluorescence microscope OLYMPUS IX71S1F-3 with 554/576 nm bandpass filters. The changes in red fluorescence were analyzed by Image J 1.52 v software and calculated per total number of cells.

Measurement of Glutamate Concentration
Extracellular glutamate was measured using the commercially available Amplex Red glutamic acid/glutamate oxidase assay kit according to the manufacturer's protocol. Neuronal-glial cultures were pretreated with small Aβ 1-42 oligomers for 1 h plus/minus MitoTEMPO or Frentizole. After incubation, 50 µL aliquots of culture medium were taken and mixed with a working solution of 100 µM Amplex Red reagent containing 0.25 U/mL HRP, 0.08 U/mL L-glutamate oxidase, 0.5 U/mL L-glutamate-pyruvate transaminase, and 200 µM L-alanine; they were then incubated for 30 min at 37 • C and protected from light. The fluorescence was measured in a fluorescence microplate reader Fluoroskan Ascent (Thermo Scientific, Waltham, MA, USA) using excitation in the 530-560 nm range and emission at 590 nm. Glutamate concentrations of the test samples were calculated from the L-glutamic acid standard curve.

Statistical Analysis
Data were expressed as mean ± standard deviation SD of 4-9 experiments on separate cultures. Statistical comparison between experimental groups was performed using a one-way ANOVA followed by a Tukey's or LSD post hoc test using SPSS 20.0 software. p values < 0.05 were considered significant.