Proteomic analysis links alterations of bioenergetics, mitochondria-ER interactions and proteostasis in hippocampal astrocytes from 3xTg-AD mice

The pathogenesis of Alzheimer’s disease (AD), a slowly-developing age-related neurodegenerative disorder, is a result of the action of multiple factors including deregulation of Ca2+ homeostasis, mitochondrial dysfunction, and dysproteostasis. Interaction of these factors in astrocytes, principal homeostatic cells in the central nervous system, is still poorly understood. Here we report that in immortalized hippocampal astrocytes from 3xTg-AD mice (3Tg-iAstro cells) bioenergetics is impaired, including reduced glycolysis and mitochondrial oxygen consumption, and increased production of reactive oxygen species. Shotgun proteomics analysis of mitochondria-ER-enriched fraction showed no alterations in the expression of mitochondrial and OxPhos proteins, while those related to the ER functions and protein synthesis were deregulated. Using ER- and mitochondria-targeted aequorin-based Ca2+ probe we show that, in 3Tg-iAstro cells, ER was overloaded with Ca2+ while Ca2+ uptake by mitochondria upon ATP stimulation was reduced. This was accompanied by the increase in short distance (≈8–10 nm) contact area between mitochondria and ER, upregulation of ER-stress/unfolded protein response genes Atf4, Atf6 and Herp, and reduction of global protein synthesis rate. We suggest that familial AD mutations in 3Tg-iAstro cells induce mitochondria-ER interaction changes that deregulate astrocytic bioenergetics, Ca2+ homeostasis and proteostasis. These factors may interact, creating a pathogenic loop compromising homeostatic and defensive functions of astroglial cells predisposing neurons to dysfunction.


Introduction
Alzheimer's disease (AD) is a devastating age-related neurodegenerative disorder with a complex and slowlydeveloping pathogenesis 1 . Modern hypotheses, aiming at explanation the mechanisms of AD development, link ageing with destructive effects of Familial Alzheimer Disease (FAD) mutations and β-amyloid (Aβ) burden. Among such hypotheses are the calcium hypothesis of neurodegeneration 2 and the mitochondrial cascade hypothesis, the latter including also the oxidative stress and the energy dysbalance hypotheses 3,4 . The Ca 2+ hypothesis concerns the enhancement of neuronal Ca 2+ signaling leading to excitotoxicity and neuronal death 5 , and the over-activation of calcineurin (CaN) signaling 6 linked to altered neuronal plasticity and memory deficit. Mitochondrial dysfunction occurs early in AD 7 . The aging processes and FAD-associated mutations can promote mitochondrial dysfunctions in the central nervous system (CNS) [8][9][10] . Alterations of oxidative phosphorylation determine the decrease of adenosine triphosphate (ATP) and the increase of reactive oxygen species (ROS) production accompanied by fatty acid peroxidation. Both oxidative stress and decrease in ATP strongly impact neuronal function by impairing all energy-demanding processes, affecting protein folding, and causing DNA damage 11 . Recent evidence indicates that Ca 2+ and mitochondria can be tandem players in AD development. A significant reduction in expression of mitochondrial Ca 2+ efflux controlling proteins is observed in prefrontal cortex of both 3xTg-AD mice models and human individuals with AD 12 . Moreover, the study provides substantial experimental evidence that mitochondrial Ca 2 + overload contributes to AD progression by promoting mitochondrial dysfunction, superoxide generation, and neuronal cell death. Mitochondria-endoplasmic reticulum (mito-ER) interaction has recently been suggested to play an important role in cellular physio-pathology 13,14 . Studies suggest that mito-ER interaction is altered in fibroblasts from FAD patients 15 and in cellular AD models 16 . Forcing mito-ER tethering in a Drosophila model of AD increased lifespan of model animals 17 . These findings were focused and hypotheses considered principally for the neuronal dysfunction, while their application for astroglial cells has not been studied in details.
Astrocytes are fundamental homeostatic cells in the CNS. They provide structural, metabolic, and signaling support to neurons 18,19 . Growing body of evidence suggests that, during AD pathogenesis, astrocytic dysfunction may precede or be parallel to the neuronal dysfunction [20][21][22] . In this aspect, mitochondrial function in astrocytes is of Fig. 1 Mitochondrial and glycolytic energy metabolism is impaired in 3Tg-iAstro compared with WT-iAstro cells. Bioenergetics of iAstro cells were assessed by Seahorse Flux Analyzer using Cell Mito Stress Kit. In a, there are mitochondrial oxygen consumption curves presented as averages ± standard deviations of each measurement time point (n = 9). Initial three measurement points represent basal mitochondrial respiration, the next three points after blocking adenine nucleotide translocating by adding oligomycin represent proton leak-stimulated oxygen consumption, then follow three points representing maximal mitochondrial oxygen consumption capacity when mitochondrial inner membrane is uncoupled by FCCP, and the last three points are for non-mitochondrial oxygen consumption when mitochondrial respiratory chain is inhibited by rotenone and antimycin A (see more detailed explanation on manufacturer's website https://www.agilent.com/en/products/cell-analysis/seahorse-xf-consumables/ kits-reagents-media/seahorse-xf-cell-mito-stress-test-kit). The summary data calculated from the curves in a are shown in b. In c representing efficiency of glycolysis, there are curves of pH changes in the cell culture medium measured simultaneously with oxygen consumption rate. Section (d) shows mitochondrial and glycolytic ATP production in iAstro cells assessed by Seahorse Flux Analyzer using Real-Time ATP Rate Assay (n = 9). e Mitochondrial membrane potential was measured using JC-1 fluorescent ratiometric probe on n = 3 independent cultures assayed in quadriplicates (p = 0.15). special interest as it may specifically be involved in deregulation of synaptic transmission 23 . While the role of astrocytic mitochondria in AD is being now acknowledged 24,25 , it appears difficult to dissect the astrocytic versus neuronal mitochondrial dysfunction in AD brain, and the knowledge about astrocyte-specific mitochondrial alterations, their link to astrocytic Ca 2+ signaling and other cellular processes as endoplasmic reticulum (ER)stress proteostasis remains limited 26 .
Recently, we have generated and characterized immortalized astrocytic cell lines from hippocampi of 3xTg-AD mice, a popular and well-studied AD mouse model 27 . These lines, named WT-and 3xTg-AD immortalized astrocytes (WT-iAstro and 3Tg-iAstro) recapitulate the features of primary astrocytic cultures from 3xTg-AD mice in terms of gene profiling, protein expression and Ca 2+ signaling 27,28 . Here we used WT-iAstro and 3Tg-iAstro lines to study mitochondrial functions and their association with Ca 2+ signaling, mito-ER interaction and proteostasis. Our data suggest that the functional impairment of mitochondrial respiration in FAD astrocytes may be associated with deregulations of cellular Ca 2+ homeostasis and protein synthesis through altered mitochondria-ER interaction.

3Tg-iAstro astrocytes have impaired ATP synthesis and mitochondrial functions
Investigation of metabolic activity by Seahorse XF Cell Mito Stress Assay revealed significant decrease in basal mitochondrial oxygen consumption rate (OCR) of 3Tg-iAstro cells compared to the WT-iAstro line (Fig. 1a, b). The average OCR in 3Tg-iAstros was 19% lower than in astrocytes without AD mutations. The same tendency remained for the mitochondrial OCR sensitive to ATPase inhibitor oligomycin; ATP production-coupled OCR in 3Tg-iAstro cells was also by 19% lower than in WT-iAstro cells. Moreover, when mitochondria were stressed by permeabilizing inner membrane for H + with carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) to reveal maximal mitochondrial respiratory capacity, the responsive increase in OCR of 3Tg-iAstro cell mitochondria was five times lower compared to that in healthy WT-iAstro cells (see OCR after addition of FCCP in Fig.  1a and Spare respiratory capacity bars in Fig. 1b). There was no significant difference in proton leak-driven OCR observed between 3Tg and WT-iAstro cell lines.
Glycolysis activity of the cells was assessed as extracellular acidification rate (ECAR) simultaneously with the OCR in the same samples. There was no difference in ECAR between the two cell lines when the cells were not stressed by mitochondrial inhibitors (Fig. 1c). After blocking phosphorylation by oligomycin, the ECAR induced by WT-iAstro cells started to increase indicating glycolysis activation to restore energetic balance. However, in the probes containing 3Tg-iAstro cells, this increase was significantly lower. The impairment of both mitochondrial and glycolytic energy metabolism in 3Tg-iAstro cells was also confirmed by Seahorse XF Real-Time ATP Rate Assay (Fig. 1d). Compared to the WT-iAstro cells, the average decrease in calculated ATP amount produced by mitochondrial and glycolytic pathways was 41% and 49%, respectively.
Altered OCR and ECAR may be associated with impairment of mitochondrial membrane potential (Δψm). However, we found that, although there was a trend to a lower Δψm in 3Tg-iAstro cells, the difference was not significant (81.67 ± 3.87% of WT-iAstro, p = 0.15) (Fig. 1e).

Increased mitochondrial ROS production in 3Tg-iAstro cells
Mitochondrial dysfunction is usually related with ROS production, and both features are characteristic to neuronal pathology in AD. To investigate if mitochondrial failure is also accompanied by ROS generation in hippocampal astrocytes from AD mouse model, we have compared ROS levels in 3Tg-iAstro and WT-iAstro cells. Assessment of intracellular hydroxyl, peroxyl and other ROS by DCFDA assay revealed significant increase of the species inside 3Tg-iAstro compared to WT-iAstro cells (Fig. 2a, c, d). The difference in the level of 2,7-dichlorofluorescein (DCF) fluorescence between 3Tg-iAstro and WT-iAstro samples increased during the first 30 min of monitoring in a plate reader, and was by 76% higher in 3Tg-iAstro cells at this time point (Fig. 2c). The fluorescence intensity in 3Tg-iAstro cells remained higher by 78% and 85% after 60 and 90 min, respectively. Similarly, 73% higher DCF fluorescence intensity in 3Tg-iAstro cells compared to WT-iAstro samples was found by evaluation of cell fluorescence intensity in microscope images by ImageJ software (Fig. 2d).
The primary ROS produced within mitochondria is superoxide, most of which is converted to hydrogen peroxide by superoxide dismutase 29 . Hydrogen peroxide is membrane permeable and freely diffuses to cytosol where it might be further converted to other ROS such as hydroxyl radical by Fenton reaction. On the contrary, superoxide cannot pass membranes and remains at the production site. If the source of cytosolic ROS is mitochondrial hydrogen peroxide, the levels of mitochondrial superoxide should also be increased. Indeed, images of 3Tg-iASTRO cells loaded with mitochondrially targeted fluorescent superoxide indicator MitoSOX TM Red had more intensively stained mitochondrial network compared to WT-iAstro cells (Fig. 2b, Supplementary Fig. 1). Quantitative evaluation of the images revealed 63% higher MitoSOX fluorescence in 3Tg-iAstro cells (Fig. 2e). This was equal to the level induced by mitochondrial respiratory complex III inhibitor 100 μM Antimycin A, a well-known inducer of mitochondrial superoxide, which was applied as a positive control of the assay.
Proteomic analysis of MERE fraction from WT-and 3Tg-iAstro lines shows absence of protein expression alterations in mitochondria Next, we thought to investigate if the impairment of cellular bioenergetics was a result of altered expression of protein components including glycolysis and mitochondrial OXPHOS pathway as it was recurrently suggested for AD [30][31][32] . For this, we have isolated Mitochondria/ Endoplasmic Reticulum-Enriched (MERE) fractions and performed shotgun mass spectrometry (SG-MS) proteomics followed by bioinformatic analysis (Fig. 3a). 1089 and 928 proteins have been identified in WT-iAstro and 3Tg-iAstro lines, respectively, with 777 proteins common for WT and 3Tg samples (Supplementary Table  1a). Quantification of proteins resulted in 53 differentially expressed proteins (DEPs) in 3Tg-iAstro vs WT-iAstro cells ( Fig. 3b and Supplementary Table 1b). Comparison with our previously published proteome dataset obtained on whole-cell lysates from same WT-iAstro and 3Tg-iAstro cells 27 showed 489 common proteins between MERE fraction and whole-cell lysate, while only six DEPs were found to be common between MERE fractions and whole-cell preparations (Fig. 4a). Next, we analyzed the sub-cell compartment composition of identified and DEPs proteins in MERE fraction in comparison with whole-cell dataset 27 . For this, proteins were assigned to one of the following ten groups according to Genecards database (https://www.genecards.org/): cytosol, plasma membrane, mitochondria, ER, Golgi apparatus (GA), intracellular vesicles, ribosomes, cytoskeleton, nucleus, and extracellular proteins. In the identified proteins list we found near twofold enrichment of mitochondrial and ER proteins and three-fold enrichment of GA proteins in MERE fraction compared with whole-cell proteins with 30% decrease of PM and the absence of extracellular proteins (Fig. 4b). Surprisingly, in MERE fraction DEPs, there was only one mitochondrial protein, dimethylarginine dimethylaminohydrolase 2 (Uniprot_ID DDAH2_MOUSE, mouse gene Ddah2). Instead, there were 43 and 33% more ER and GA proteins, respectively (Fig. 4c). Cytosolic, intracellular vesicle, ribosomal, cytoskeletal, and nuclear proteins were less abundant (by 30%, 59%, 79%, 54%, and 77%, respectively). Of note, components of the glycolytic pathway were absent in DEPs of MERE fraction, while whole-cell lysate DEPs list contained two glycolytic enzymes, namely fructose-bisphosphate aldolase A (Aldoa, 1.47 fold change) and of pyruvate kinase isoform M1 (Pkm, −2.21 fold change).
Bioinformatic analysis of MERE fraction and whole-cell lysate DEPs from WT-and 3Tg-iAstro lines shows alterations in adhesion, ER and protein synthesis To investigate the functional significance of protein expression alterations we merged MERE fraction DEPs with those of whole-cell lysates. DAVID gene ontology analysis of 120 DEPs of the merged list resulted in 81 significantly overrepresented GO terms. Top overrepresented GO terms were as follows: Biological Processestranslation, cell-cell adhesion, and protein folding; among Cell Component termsmyelin sheath, focal adhesion, intracellular ribonucleoprotein complex, cell-cell adherens junction, and ribosome; among Molecular Function termspoly(A) RNA binding, cadherin binding involved in cell-cell adhesion, RNA binding, structural constituent of ribosome, protein binding, unfolded protein binding. The overrepresented KEGG pathway was Ribosome (Fig. 5 and Supplementary Table 2).

Poor overlap with MAMs and mitochondrial datasets from mouse AD models
Next, we compared MERE fraction and whole cells 3Tg-iAstro datasets with mitochondria-associated membranes (MAMs) and mitochondrial proteome datasets reported by Völgyi et al. 33 and Yu et al. 31 , respectively. Völgyi et al. reported deregulation of many ribosomal proteins in MAMs from 3-month-old APP/PS1 AD mouse. However, there were only two proteins common with our MERE fraction dataset: oppositely regulated ribosomal protein 21 (Rpl21) and co-downregulated Syntenin-1 (Sdcbp) (Supplementary Table 3a, b). Dataset by Yu et al. features mitochondrial proteome from hippocampus and cortex of symptomatic 15-month-old 3xTg-AD mice, the same from which 3Tg-iAstro originated. Also in this case two protein were common with our MERE fraction dataset: co-downregulated vacuolar proton pump subunit B2 (Atp6v1b2) and oppositely regulated vesicular-fusion ATPase NSF (Supplementary Table 3c, d).
Taken together, these results suggest that the expression of mitochondrial proteins, which have been identified in MERE fractions, was largely unaltered in WT-iAstro and 3Tg-iAstro lines. Instead, the protein expression alterations are mainly related to ER and protein synthesis and processing. We reasoned that, in terms of association with protein expression, OXPHOS deficiency in 3Tg-iAstro cells may be associated with extra-mitochondrial factors. Therefore, we decided to focus on investigation of the following aspects: (i) relationships between Ca 2+ signaling in the cytosol, ER and mitochondria; (ii) mitochondria-ER interaction; and (iii) assessment of the ER-stress/UPR and protein synthesis in 3Tg-iAstro vs WT-iAstro cells.

Shortened mitochondrial-ER distance in 3Tg-iAstro cells
The deficient mitochondrial Ca 2+ signaling, among other reasons, may be mediated by the alterations in the relationships between mitochondria and ER 35 . We have investigated this exploiting the recently generated SPLICS probe which allows visualizing mitochondrial-ER interaction point at two defined distances: ≈8-10 nm (SPLICS-Short) and ≈40-50 nm (SPLICS-Long) 15 . In WT-iAstro cells SPLICS-Short was visualized as fluorescent dots localized mainly in the perinuclear zone (Fig. 7a). In 3Tg-iAstro cells SPLICS-Short expression resulted in a significantly higher number of fluorescent dots some of them appeared to be fused. Quantification of fluorescence area normalized for cell surface showed near doubling of the short-distance mito-ER contacts in 3Tg-iAstro cells as compared with WT-iAstro cells (0.055 ± 0.0037 SPLICS/ cell ratio vs 0.031 ± 0.0032 SPLICS/cell ratio, respectively, p < 0.0001) (Fig. 7a). SPLICS-Long expression in both WT-iAstro and 3Tg-iAstro cells was visualized as structures with diverse morphology, from dots of different size to tubular structures reminiscent shape of mitochondria. Overall, the SPLICS-Long fluorescence area in WT-iAstros was significantly higher than that of SPLICS-Short (p < 0.0001). Nevertheless, the difference in SPLICS-Long/cell area ratio between 3Tg-iAstro and WT-iAstro cells was not significative (0.084 ± 0.008 SPLICS/cell ratio vs 0.069 ± 0.0062 SPLICS/cell ratio, respectively, p = 0.15) ( Fig. 7b) indicating that specifically short distance mito-ER interaction is augmented in 3Tg-iAstro cells.
Activation of unfolded-protein response and reduced protein synthesis in 3Tg-iAstro compare to WT-iAstro cells Proteomic analysis suggests that protein synthesis and processing may be impaired in 3Tg-iAstro cells. Therefore, we first measured mRNA levels of ER-stress and UPR-related gene. Real-time PCR analysis shows that Atf4, Atf6, and Herp mRNA were significantly upregulated in 3Tg-iAstro cells by 136% (p = 0.037), 73% (p = 0.020) and 166% (p = 0.031), respectively (Fig. 8a). Xbp1 showed a trend of upregulation (by 88%) although the difference failed to reach significance (p = 0.059). To investigate protein translation efficiency, we used a method of surface sensing of translation (SUnSET) 36,37 . After puromycin pulsing of WT and Tg-iAstro we observed a 20% (p = 0.033) decrease of puromycin incorporation in 3Tg-iAstro compare to WT-iAstro (Fig.  8b, c).

Discussion
In the present contribution we took advantage of the recently generated AD astrocytic cellular model, immortalized hippocampal astrocytes from 3xTg-AD mice (3Tg-iAstro and WT-iAstro cells) to investigate the possible association between astrocytic bioenergetic status, ER and mitochondrial Ca 2+ homeostasis and proteostasis, which includes protein synthesis and degradation machinery.
We found that in 3Tg-iAstros (i) glycolysis, OCR and ECAR are significantly lower compared to WT-iAstros, (ii) intracellular and intramitochondrial ROS production is elevated, however (iii) the changes are not due to alterations in expression of mitochondrial proteins, but rather due to extra-mitochondrial protein expression changes, resulting in (iv) ER Ca 2+ overload and reduced ATP-evoked mitochondrial Ca 2+ transients; (v) increase of area with a shorter (8-10 nm) mitochondria-ER interaction distance; (vi) increased expression of ER-stress/ UPR-associated genes; and (vii) reduced protein synthesis rate in 3Tg-iAstro compare with WT-iAstro cells.
Two glycolytic proteins were found to be changed in our dataset collected on whole-cell iAstro cells, namely fructose-bisphosphate aldolase A (Aldoa) and isoform M1 of pyruvate kinase PKM (Pkm). Aldoa catalyzes the reversible conversion of fructose-1,6-bisphosphate to glyceraldehyde 3-phosphate and dihydroxyacetone phosphate and is critically involved in glycogen storage: deficiency of this enzyme causes Glycogen Storage Disease XII leading to excessive liver accumulation of glycogen 38 . Therefore, increase of Aldoa expression in 3Tg-iAstro cells may potentially lead to the depletion of glycogen from astrocytes which would reduce bioenergetic reserve of the AD CNS. The second enzyme, Pkm, a member of pyruvate kinases (PK) family, catalyzes the last, and ratelimiting, step of glycolysis by the transfer of a phosphoryl group from phosphoenolpyruvate to ADP generating ATP and pyruvate 38 . Downregulated glycolytic response to mitochondrial stress was also confirmed in human iPSCderived astrocytes with PS1-ΔE9 mutations 39 . Astrocytes have significantly higher glycolytic metabolism than neurons through which they feed neurons, store glycogen and metabolize and recycle neurotransmitters 40,41 . Slowing down astrocytic glycolysis may have detrimental effects on neuronal activity and survival.
Ca 2+ is an important regulator of mitochondrial oxidative metabolism and is required for the activity of enzymes of the tricarboxylic acid cycle, the proteins of the electron transport chain, and the ATP synthase 42,43 . In these terms, the decrease of mitochondrial Ca 2+ may explain the reduced OCR and respiratory capacity of 3Tg-iAstro cells. Moreover, our results are in line with those obtained on APPswe/PS1A246E-expressing astroglioma cells in which reduced mitochondrial OCR was accompanied by insignificant change in mitochondrial Ca 2+ in front of strongly potentiated ATP-induced cytosolic Ca 2+ increase 44 . However, it has been reported that in iPSCderived astrocytes form patients bearing PS1-ΔE9 mutation basal mitochondrial respiration was increased 39 . Such discrepancies may arise from differences in generation of cell models as well as from different AD-related mutations.
A growing body of evidence suggests that the mitochondrial-ER interaction, in terms of both distance and protein composition, may be central for deregulation of Ca 2+ signaling, mitochondrial functions and ERstress 14,35,45,46 . It has been found that the distance between ER and mitochondria is kept in a range of ≈10-50 nm 13 . Increase as well as decrease of this distance affects multiple aspects of mitochondrial functions: Ca 2+ signaling, lipid transfer and biogenesis, protein synthesis and transport. We found that in 3Tg-iAstro cells the short distance (≈8-10 nm) sites area is increased by 77%. Although at the moment we cannot draw cause-effect relationships between the mito-ER contact sites shortening and the other findings of this work, we propose a hypothetical mechanistic sequence of events which may occur in astrocytes and has in its center FAD mutationsinduced increase in mito-ER tethering at a short distance. First of all it explains reduced mitochondrial Ca 2+ signals in concomitance with ER Ca 2+ overload. The mito-ER distance shorter than 10-12 nm creates steric encumbrance for IP 3 Rs which hampers Ca 2+ transfer from IP 3 Rs to mitochondria 13 . Reduced mitochondrial Ca 2+ signals may lead to reduced OXPHOS efficiency and ATP production 42 , while cellular Ca 2+ overload may be a cause for ER-stress/UPR 47 . Both mitochondrial and ER-related abnormalities due to increased agonist-induced Ca 2+ release and impaired supply of ATP may lead to global Fig. 8 Activation of ER-stress and reduction of protein synthesis in 3Tg-iAstro vs WT-iAstro cells. a Real-time PCR of ER-stress/UPR responserelated genes Atf4 (p = 0.037), Atf6 (p = 0.02), Herp (p = 0.031) and Xbp1 (p = 0.059) of 3Tg-iAstro vs WT-iAstro cells. Data are expressed as mean ± SEM ΔC(t), n = 14 for WT-iAstro, n = 12 biological replicates for 3Tg-iAstro cells. b WT and 3Tg-iAstro were pulsed with puromycin for 3 h. Where indicated, cells were also treated with cycloheximide (CHX, 10 μM) ten minutes before adding puromycin. Anti-puromycin antibody was used to detect neo-synthesized peptides. Ponceau staining was used for band normalization. c Quantification of anti-puromycin detected bands. Data are expressed as mean ± SEM, six and four independent cultures were used for WT and Tg-iAstro, respectively (p = 0.033; two-tailed unpaired Students's t test with Mann-Whitney correction). deregulation of cellular Ca 2+ homeostasis. At the same time ER-stress and energy deficit may slow down protein synthesis and degradation machinery as these processes are among the most energy consuming in the cell. While details of this hypothetic scenario, illustrated in Fig. 9, are still to be elucidated, recent publication corroborate our results. It has been show that mito-ER tethering in a short mito-ER distance range is augmented in both presenilin 2 N141I mutant human fibroblasts 15 and upon overexpression of AD-relevant caspase 3-cleaved WT Tau, which is localized specifically at mito-ER contact sites 48 .

Conclusions
In the CNS, neuronal performance, survival and resilience in stressful or diseased conditions strongly depends on homeostatic support of astroglial cells 19 . During AD pathogenesis, astrocytes, in their turn, undergo complex, brain region-specific and disease stage-dependent remodeling 21,49 . Our data suggest that these changes concern multiple aspects of astroglial biology, characteristic for AD as the deregulation of cellular bioenergetics and Ca 2+ signaling, mitochondria-ER interaction and proteostasis.
The effects of these changes, in turn, may have multiple consequences, e.g., we recently have shown that conditioned medium from 3xTg-AD astrocytes reduces synaptic proteins in cultured neurons 50 , while 3Tg-iAstros fail to support the integrity of the blood-brain barrier in vitro 51 . In light of this, our work exemplifies the importance of multidisciplinary approach in studying such complex phenomena as AD and illustrates the use of iAstro cells as versatile and cost-effective astrocytic cellular model of AD. To conclude, our results highlight the important role that astrocytes play in AD pathogenesis and suggest that the alterations in their homeostatic capabilities contribute to AD progression at early stages of the disease.

Evaluation of iAstro energetic metabolism and ROS production Mitochondrial respiration
Assessment of mitochondrial and glycolytic functionality was performed by Seahorse XFp Analyser (Agilent Technologies, Santa Clara, CA, USA) using Seahorse XFp Cell Mito Stress Test Kit (Agilent Technologies) and Seahorse XFp Real-Time ATP Rate Assay Kit (Agilent Technologies) according to manufacturer's instructions. Briefly, iAstro cells were seeded into Agilent Seahorse XFp well plates at a density of 100,000 cells/cm 2 and kept in complete culture medium (described in 4.1.) in incubator for 14-16 h. One hour before the measurement, the medium was replaced with Seahorse XF Assay Medium supplemented with 2 mM L-glutamine, 1 mM sodium pyruvate and 10 mM glucose and cells were placed to non-CO 2 incubator. Just before the measurement, medium was changed again to the fresh Assay Medium with the same supplements. Final inhibitor concentrations in the wells were 1.5 μM oligomycin, 1 μM carbonyl cyanide-4-phenylhydrazone (FCCP), 0.5 μM antimycin A, and 0.5 μM rotenone. Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were normalized to total cellular protein content determined directly in the plate immediately after each experimental run by Fig. 9 Scheme of possible relationships between FAD mutations, mitochondrial-ER interaction, mitochondrial and ER Ca 2+ signaling, bioenergetic status of mitochondria, ER-stress/UPR and protein synthesis. FAD mutations-provoked increase in short distance mitochondrial-ER contact sites may lead to (1) deficient mitochondrial Ca 2+ signaling, (2) decrease of OxPhos function and production of ATP and (3) to increase of ROS. From the ER side it may lead to (4) overload of the ER with Ca 2+ and deregulation of cellular Ca 2+ homeostasis, (5) induction of ER-stress/UPR response as well as (6) to inhibition of protein synthesis (red arrows). Secondary events may include negative feedbacks of the ATP deficit and the increased ROS on (7) the cellular Ca 2+ homeostasis and (8) protein synthesis (orange arrows).
Bradford assay. Optical density after reaction with the Bradford reagent (Cat. No. B6916, Merck) was assessed by a multifunctional plate reader Infinite 200 Pro M Nano Plex (Tecan, Männedorf, Switzerland). The data were analyzed and single run reports were generated by Wave software (Agilent Technologies), and graphical images of summary data were created by SigmaPlot vs.13 (Systat Software, Slough, UK).

Reactive oxygen species determination
For intracellular ROS measurement, cells were seeded in black (for plater reader) and transparent (for microscopy) 96-well plates at a density of 60,000 cells/cm 2

Mitochondrial superoxide determination
For evaluation of mitochondrial superoxide, the cells were grown in clear 96-well plates the same way as for intracellular ROS assessment. 48 h after plating, the cells were 3x washed with HBSS and incubated with 2 µM MitoSox TM Red (Cat. No. M36008, Thermo Fisher Scientific) in HBSS at 37°C in the dark for 15 min. The images were taken by fluorescent microscope Zeiss Axio Observer.Z1 and fluorescence intensity in the images was evaluated by ImageJ software. For higher magnification images for monitoring MitoSox staining colocalization with mitochondrial network, the cells were grown on glass-bottom Petri dishes (Catalog No. 0030740009, Eppendorf). For visualization of total mitochondrial network, 500 nM MitoTracker Green FM (Cat. No. M7514, Thermo Fisher Scientific) and for nuclear chromatin staining, 6 μg/ml Hoechst33342 (Sigma-Aldrich) was added to incubation together with MitoSox prior to microscope imaging.

Mitochondrial membrane potential determination
Variation in mitochondrial membrane potential (Δψm) is an important parameter of mitochondrial function and is an indicator of cell health. Mitochondrial membrane potential (Δψm) was measured by using 5,51,6,61-tetrachloro-1,11,3,31tetraethylbenzimidazolyl carbocyanine iodide (JC-1; Cayman, Ann Arbor, Michigan, USA) according to manufacturer's instructions as has been previously describes 52,53 . Briefly, WT-and 3Tg-iAstro cells were plated (10 000 cells/well in 96-well plate) 24 h before measurement. Then the cells were incubated with JC-1 diluted in Assay Buffer for 15 min at 37°C. After washing (twice with Assay Buffer), the mitochondrial membrane potential was determined by measuring the red (excitation 535 nm/emission 595 nm) and green (excitation 485 nm/emission 535 nm) fluorescence using a spectrophotometer (VICTOR™ X Multilabel Plate Reader, PerkinElmer). Δψm was calculated as ratio of red/green fluorescence intensities and the data are expressed as a fold change of WT-iAstro cells.

Preparation of mitochondria-ER enriched fractions
WT-iAstro and Tg-iAstro were plated at concentration of 0.5 × 10^6 cells/dish in 10 cm Petri dishes (50 dishes per line) and grown until 80-90% confluence. On the day of experiment, the cells were washed twice with PBS, detached with trypsin, counted and collected 20 × 10^6 cells/tube. Three instrumental replicates for each sample were subjected to the DIA analysis 54,55 .
The mass spectrometry files were searched using Protein Pilot (AB SCIEX, Concord, Canada) and Mascot (Matrix Science Inc., Boston, USA). Samples were input in the Protein Pilot software v. 4.2 (AB SCIEX, Concord, Canada), with the following parameters: cysteine alkylation, digestion by trypsin, no special factors and False Discovery Rate at 1%. The UniProt Swiss-Prot reviewed database containing mouse proteins (version 12/10/2018, containing 25137 sequence entries). The Mascot search was performed on Mascot v. 2.4, the digestion enzyme selected was trypsin, with two missed cleavages and a search tolerance of 50 ppm was specified for the peptide mass tolerance, and 0.1 Da for the MS/MS tolerance. The charges of the peptides to search for were set to 2+, 3+ and 4+, and the search was set on monoisotopic mass. The instrument was set to ESI-QUAD-TOF and the following modifications were specified for the search: carbamidomethyl cysteines as fixed modification and oxidized methionine as variable modification.
The quantification was performed by integrating the extracted ion chromatogram of all the unique ions for a given peptide. The quantification was carried out with PeakView 2.0 and MarkerView 1.2. (Sciex, Concord, ON, Canada). Six peptides per protein and six transitions per peptide were extracted from the SWATH files. Shared peptides were excluded as well as peptides with modifications. Peptides with FDR lower than 1.0% were exported in MarkerView for the t-test.

Generation of stable iAstro lines expressing MIT-AEQ and ER-AEQ
For measurement of Ca2+ inside the ER and mitochondria, either mitochondrially or ER-targeted photoprotein aequorin (MIT-AEQ and ER-AEQ probes, respectively) were used. The details of generation of producation of lentiviral vectors expressing aequorin Ca 2 + probes is described elswhere 34 . 24 h after plating (10E4 cell/well in 24-well plate), WT-and Tg-iAstros were infected with lentiviral particles experssing MIT-AEQ and ER-AEQ at MOI from 5 to 20. The dilution of virus that gave more than 50% of infected cells, as was detected by fluorescence of reporter proteins 34 , were further processed. Upon reaching confluence, the cells were expanded, and MIT-AEQ and ER-AEQ expressing WT-and 3Tg-iAstro cells were inriched using fluorescenceactivated cell sorting (S3e Cell Sorter, Bio-Rad, Segrate, Milano). Sorted cells were expanded and frozen until needed.

Aequorin Ca 2+ measurements
For Ca 2+ measurement in the ER lumen, WT-and 3Tg-iAstro cells expressing ER-AEQ Ca 2+ sensor were plated on 13 mm round coverslips placed in 24-well plates (5 × 10 4 cell/well). 24 h after plating che cells were reconstituted in modified Krebs-Ringer buffer (KRB, 135 mM NaCl, 5 mM KCl, 0.4 mM KH2PO4, 1 mM MgSO4, 5.5 mM glucose, 20 mM HEPES (pH 7.4) supplemented with 600 μM EGTA, 5 μM coelenterazine-n (low lightemitting coelenterazine variant) and 3 μM ionomycin (all reagents from Sigma) for 1 h at 4°C. After reconstitution the cells were washed three times with KRB containing 600 μM EGTA and 2% BSA, followed by three washes with KRB containing 100 μM EGTA, after which the coverslips were trans-ferred into perfusion chamber of a custom-built aequorinometer (CAIRN research, UK). The cells were perfused with KRB supplemented with 100 μM EGTA at 37°C. After 3 min of baseline recording the perfusion solution was switched to KRB supplemented with 2 mM Ca 2+ and recording continued until the [Ca 2+ ] in the ER did reach the steady-state level. At the end of each experiment, for quantification of the ER Ca 2+ levels, the cells were perfused with distilled water containing 100 mM Ca 2+ and 0.1% Triton X-100 to disrupt cells and discharge the remaining AEQ pool.
For assessment of ATP-induced Ca 2+ transients in the mitochondrial matrix, MIT-AEQ-expressing WT-and 3Tg-iAstro cells were reconstituted with coelenterazine wt (5 μM) in complete culture medium in 37°C and 5% CO2 for 1 h. After washing (3 times) with KRB supplemented with 2 mM Ca 2+ , cells were transferred into perfusion chamber of a custom-built aequorinometer and acquisition started in perfutio with KRB supplemented with 2 mM Ca 2+ . After baseline recording, perfusion was switched to KRB supplemented with 2 mM Ca 2+ and 20 μM ATP. At the end of each experiment, for quantification of the ER Ca 2+ levels, the cells were perfused with distilled water containing 100 mM Ca 2+ and 0.1% Triton X-100 to disrupt cells and discharge the remaining AEQ pool. The light signals were calibrated off-line into [Ca 2+ ] values using an algorithm described in 34 .

Visualization of mitochondria-ER contact sites
Mitochondria-ER contact sites were visualized using SPLICS probe as described in 15 . Briefly, WT-iAstro and 3Tg-iAstro were plated onto 13 mm glass coverslips in 24well plates (3x10E4 cells/well). 24 h after plating cells were transfected with pairs of plasmids SPLICS-Mit-Short + SPLICS-ER (hereafter referred as SPLICS-Short) and SPLICS-Mit-Long + SPLICS-ER (hereafter referred as SPLICS-Long) in 1:1 ratio using Lipofectamine 2000 (Thermofisher, Milan, Italy). 24 h after transfection cells were washed with PBS and fixed in 4% formaldehyde (Sigma, Milan, Italy). Fluorescent image stacks were taken using Zeiss 710 confocal laser scanning microscope equipped with Plan-Apochromat 63x/1.40 Oil M27 objective and Zen software. Image stacks were processed offline to major intensity projections. Thresholding and area calculations were done using Fiji package of ImageJ software v.1.52p. Data are expressed as SPLICS/cell area ratio.

RNA extractiona and real-time PCR
Total RNA was extract by using the TripleXtractor reagent (Bio-Cell) as indicated by the supplier, and the AMV Reverse Transcriptase kit (Promega) was used to produce cDNA following the manufacturer's recommendations, by using 2 μg of total RNA. Quantitative PCR reactions were performed by using the Bio-Rad CFX96 qPCR thermocycler. The primer pair sequences for selected amplicons were designed using the online IDT PrimerQuest Tool software (IDT; https://eu.idtdna.com/ Primerquest/Home/Index). Primer sequences are available in Supplementary Table 1. Gapdh mRNA level was used as an internal control and results were expressed as previously described 56 .
Puromycin incorporation method (Surface sensing of translation, SUnSET) iAstro-WT and TG were incubated with 4 µM puromycin dihydrochloride (Sigma Cat. P8833) supplemented in normal medium at 37°C with 5% CO2 for 3 h 36 . Subsequently, cells lysates were subjected to Western blot assay.
30 µg of proteins were mixed with the right amount of Laemmli Sample Buffer 4X (Bio-Rad), and boiled for 7'. Then samples were loaded on a 12% polyacrylamidesodium dodecyl sulphate gel for electrophoresis. Proteins were transferred onto nitrocellulose membrane, using Mini Transfer Packs or Midi Transfer Packs, with Trans-Blot® Turbo TM (Bio-Rad) according to manufacturer's instructions (Bio-Rad).

Statistical analysis
The data for mitochondrial and glycolytic efficiency and ROS production (Figs. 1 and 2 except Fig. 2e) are presented as mean ± SD; statistical analysis was performed by SigmaPlot vs.13 (Systat Software, Slough, UK). Other data (Fig. 2e; Figs. 6-8) are presented as mean ± SEM and were analyzed using GraphPad Prism v.7. No samples/results were excluded from the analysis. Unpaired two-tailed Student's t test was used in all experiments and differences were considered significant at p < 0.05.