Abstract Mitochondrial diseases are associated with neuronal death and mtDNA depletion. Astrocytes respond to injury or stimuli and damage to the central nervous system. Neurodegeneration can cause astrocytes to activate and acquire toxic functions that induce neuronal death. However, astrocyte activation and its impact on neuronal homeostasis in mitochondrial disease remain to be explored. Using patient cells carrying POLG mutations, we generated iPSCs and then differentiated these into astrocytes. POLG astrocytes exhibited mitochondrial dysfunction including loss of mitochondrial membrane potential, energy failure, loss of complex I and IV, disturbed NAD^+/NADH metabolism, and mtDNA depletion. Further, POLG derived astrocytes presented an A1-like reactive phenotype with increased proliferation, invasion, upregulation of pathways involved in response to stimulus, immune system process, cell proliferation and cell killing. Under direct and indirect co-culture with neurons, POLG astrocytes manifested a toxic effect leading to the death of neurons. We demonstrate that mitochondrial dysfunction caused by POLG mutations leads not only to intrinsic defects in energy metabolism affecting both neurons and astrocytes, but also to neurotoxic damage driven by astrocytes. These findings reveal a novel role for dysfunctional astrocytes that contribute to the pathogenesis of POLG diseases. Background In an increasingly aging global population, neurodegeneration is now a leading threat to human health. Understanding the process of neurodegeneration is changing from a purely neuron-centric view to a more comprehensive perspective in which several cell types interact to drive this process, and where astrocytes, previously considered only supportive, have moved center stage [49]^1^-[50]^5. Despite advances, mechanistic understanding remains constrained within paradigms such as oxidative stress, proteasome impairment, accumulation of abnormal protein aggregates and mitochondrial dysfunction [51]^6^, [52]^7. Of these, mitochondrial dysfunction is a recurring theme, particularly in common diseases [53]^8^-[54]^10; such as Parkinson's (PD) and Alzheimer's (AD) diseases that together affect ~10% of population, and diabetic retinopathy, the most common form of blindness in working aged people [55]^11^, [56]^12. How a mitochondrial defect leads to tissue destruction and subsequent loss of functional neurons remains, however, unclear. Of interest in the context of mitochondrial dysfunction are proteins involved in mitochondrial DNA replication (mtDNA), particularly polymerase gamma (pol γ) as the enzyme that replicates and repairs mtDNA [57]^13. Mutations in the POLG gene, which encodes the catalytic subunit of polymerase γ, are the primary cause of inherited mitochondrial diseases. These mutations, along with variations in mtDNA, significantly affect mitochondrial function. Clinically, these mutations cause a continuum of overlapping phenotypes from infantile to adult disorders [58]^14. At the molecular level, mutations in POLG lead to mtDNA maintenance defects and mitochondrial dysfunction. Interestingly, the molecular mtDNA defect differs depending on the tissue; multiple deletions occur in skeletal muscle while neurons and hepatocytes show mtDNA depletion [59]^15. The underlying mechanisms behind decreased mtDNA quality and mitochondrial dysfunction in POLG related disease remain, however, elusive. In post-mortem studies, we have shown that the loss of neurons in POLG related disease is driven by severe mtDNA depletion [60]^15. The contribution of glial cell dysfunction to neurodegenerative disease has come more into focus [61]^3^, [62]^16^-[63]^18. Astrocytes are both highly abundant [64]^19 and play a crucial role in the support and modulation of neuronal function including regulating glutamate turnover, ion and water homeostasis, synapse formation/modulation, tissue repair, energy storage, and defense against oxidative stress [65]^19^, [66]^20. These cells are also critical for neuronal metabolism [67]^21^, [68]^22 and are a major component of neurovascular coupling [69]^23. Greater understanding of the importance of astrocytes has shifted focus to the role of these cells in neurological diseases such as PD, AD [70]^24^, [71]^25, Huntington's disease (HD) [72]^26 and amyotrophic lateral sclerosis (ALS) [73]^27. However, the role of astrocytes in mitochondrial diseases such as POLG has yet to be explored. Both disease and cerebral injury can induce astrocytes to enter a 'reactive' state in which gene expression changes markedly [74]^3. These reactive astrocytes are mainly divided into A1 and A2 types according to their function: LPS-induced neuroinflammation A1 astrocytes lose their supportive role and acquire toxic functions mediated by secreted neurotoxins, thereby inducing neurons and oligodendrocytes die rapidly; Ischemia-induced A2-reactive astrocytes promote neuronal survival and tissue repair. Recent studies suggested that activated microglia induced astrocyte conversion to the A1 reactive phenotype by releasing interleukin 1 alpha (IL1α), tumor necrosis factor alpha (TNFα), and the complement component subunit 1q (C1q) [75]^3. In neurodegenerative states such as AD [76]^28, PD [77]^29, multiple sclerosis (MS) [78]^28 and ALS [79]^17. Reactive astrocytes can display both neuroprotective and neurodegenerative functions. The role of astrocyte reactivation and the consequences this has for neuronal homeostasis in mitochondrial disease and mitochondrial dysfunction has not been explored. Stem cell technologies, including induced pluripotent stem cells (iPSCs) and multicellular organoid models are revolutionizing our ability to investigate complex systems and human disease [80]^17^, [81]^30^, [82]^31. Our aim in this study was to use a human stem cell-based culture system to examine the astrocytic contribution to POLG related disease of the brain and investigate whether mitochondrial dysfunction would stimulate astrocytes to become toxic for neurons. To achieve this, we generated iPSC-derived astrocytes from two patients harboring POLG mutations. Using cellular, metabolic, and transcriptomic approaches, we found that POLG mutations not only cause intrinsic defects in energy metabolism affecting neurons and astrocytes, but also astrocyte-driven neurotoxic damage. Here, we describe the mitochondrial profile of iPSC-derived astrocytes from POLG related diseases, validating this model as a powerful tool for studying disease mechanisms and for non-invasive drug-targeting assays in vitro. Our findings reveal novel roles for dysfunctional astrocytes that contribute to the pathogenesis of mitochondrial diseases, which provide a novel disease target. Materials and methods Derivation of iPSCs and neural stem cells (NSCs) generation Human iPSCs were reprogrammed from fibroblasts as described in our previous publication [83]^31^, [84]^32. As shown in [85]Supplementary Table 1, two POLG patients derived iPSCs were used in this study, including one homozygous c.2243G>C, p.W748S/W748S (WS5A) and one compound heterozygous c.1399G>A/c.2243G>C, p.A467T/W748S (CP2A). A panel of control were used in this study, Detroit 551 (ATCC^® CCL 110^™), AG05836B fibroblasts (RRID: CVCL_2B58) and two human embryonic stem cell (ESC) lines: HS429 and HS360. All iPSC and ESC lines were maintained in E8 medium (Invitrogen, A1517001) on Geltrex (Invitrogen, A1413302) coated 6-well plate (Thermo Scientific, 140675). NSCs were generated from the POLG and control iPSC lines by lifting cells into a Chemically Defined Medium (CDM) with epidermal growth factor (EGF) and fibroblast growth factor-2 (FGF-2) both at 100 ng/ml, as described previously [86]^32. Astrocyte differentiation iPSC-derived NSCs were placed on poly-D-lysine (PDL) coated coverslips (Neuvitro, GG-12-15-PDL). The following day, the cells were changed into astrocyte differentiation medium, as described in [87]Supplementary Table 1. The medium was changed every other day for the first week, every two days for the second week and every three days for the third and fourth week. After 28 days of differentiation, the cells were cultured in maturation medium AGM^TM Astrocyte Growth Medium BulletKit^TM (Lonza, CC-3186) as described in [88]Supplementary Table 2, for one more month. Dopaminergic neuron (DA neuron) differentiation iPSC-derived neurospheres which were generated from 5 days' neural induction, were maintained in CDM supplemented with 100 ng/ml FGF8b (R&D systems, 423-F8) over a period of 7 days to initiate DA neuron progenitor induction. The following 7 days, the medium was changed to CDM supplemented with 1 µM purmorphamine (PM) (EMD Millipore, 540220-5MG) and 100 ng/ml FGF8b. Termination of the suspension cultures was performed by dissociating the spheres into single cells by incubation with TrypLE^™ Express followed by trituration and subsequent plating into monolayers. The DA neurons were matured in DA medium: CDM supplemented with 10 ng/ml BDNF (PeproTech, 450-02) and 10 ng/ml GDNF (PeproTech, 450-10) on Poly-L-Ornithine (Sigma-Aldrich, P4957) and laminin (Sigma-Aldrich, L2020) coated plates. Immunocytochemistry and immunofluorescence (ICC/IF) staining Cells were fixed with 4% (v/v) paraformaldehyde (PFA, VWR, 100503-917) and blocked using blocking buffer containing 10% (v/v) normal goat serum (Sigma-Aldrich, G9023) with 0.3% (v/v) Triton^™ X-100 (Sigma-Aldrich, X100-100ML). The cells were then incubated with primary antibody solution overnight at 4°C and further stained with secondary antibody solution (1:800 in blocking buffer) for 1 hour (h) at room temperature (RT). NSCs were stained with rabbit anti-PAX6, anti-NESTIN, anti-SOX2.Mitochondrial respiratory chain complex I subunit were stained with anti-NDUFB10. Astrocytes and oligodendrocytes were stained with anti-GFAP, anti-S100β, anti-EAAT-1, anti-GS and anti- DCX, respectively. The antibodies used for DA neuron staining were anti-TH, anti-TuJ1 and anti-MAP2. The secondary antibodies used were Alexa Fluor^® goat anti-rabbit 488 Alexa Fluor^® goat anti-mouse 594 and Alexa Fluor^® goat anti-chicken 594. After incubation with secondary antibodies, the coverslips were mounted onto cover slides using prolong diamond antifade mounting medium with DAPI (Invitrogen, [89]P36962). Details of the antibodies used were listed in Key Resources Table. For staining of neurospheres, spheres were spread directly onto cover slides and left at RT until completely dry and then fixed with 4% (v/v) PFA. After two washes with PBS, the spheres were covered in PBS with 20% sucrose, sealed with parafilm, and incubated overnight at 4°C. The spheres were blocked with blocking buffer for 2 hours at RT and the primary antibodies were added to the samples overnight at 4°C. After washing the samples for 3 hours in PBS with a few changes of buffer, incubation with secondary antibodies (as described above) was conducted overnight at 4°C in a humid and dark chamber. Coverslips were mounted using Fluoromount G^® (Southern Biotech, 0100-01) before imaging was performed using the Leica TCS SP5 or SP8 STED confocal microscope (Leica Microsystems, Germany). Mitochondrial volume and mitochondrial membrane potential (MMP) measurement To measure mitochondrial volume and MMP, cells were double stained with 150 nM MitoTracker Green (MTG) and 100 nM Tetramethylrhodamine Ethyl Ester (TMRE) for 45 min at 37°C. Cells treated with 100 µM Carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP) (Abcam, ab120081) were used as negative control. Stained cells were washed with PBS, detached with TrypLE^™ Express and neutralized with media containing 10% FBS. The cells were immediately analyzed on a FACS BD Accuri^™ C6 flow cytometer (BD Biosciences, San Jose, CA, USA). The data analysis was performed using Accuri™ C6 software. L-lactate production measurement L-lactate generation was analyzed by colorimetric L-lactate assay kit (Abcam, ab65331) according to the manufacturer's instructions. Endpoint lactate concentration was determined in a 96-well plate by measuring the initial velocity (2 min) of the balance between NAD^+ and NADH by lactate dehydrogenase. Immediately following the extracellular flux assay, the plate was measured at OD 450 nm in a microplate reader (VICTOR^™ XLight, PerkinElmer). Intercellular and mitochondrial reactive oxygen species (ROS) production Intracellular ROS production was measured by flow cytometry using dual staining of 30 µM 2′,7′-Dichlorodihydrofluorescein diacetate (DCFDA) (Abcam, ab11385) and 150 nM MitoTracker Deep Red (MTDR) (Invitrogen, [90]M22426), which enabled us to assess ROS level related to mitochondrial volume. Mitochondrial ROS production was quantified using co-staining of 10 µM MitoSOX^™ Red Mitochondrial Superoxide Indicator (Invitrogen, [91]M36008) and 150 nM MTG to evaluate ROS level in relation to mitochondrial volume. Stained cells were detached with TrypLE^™ Express and neutralized with media containing 10% FBS. The cells were immediately analyzed on a FACS BD Accuri^™ C6 flow cytometer. NADH metabolism and ATP measurement using Liquid Chromatography Mass Spectrometry (LC-MS) analysis Cells were washed with PBS and extracted by addition of ice-cold 80% methanol followed by incubation at 4°C for 20 min. Thereafter, the samples were stored at -80°C overnight. The following day, samples were thawed on a rotating wheel at 4°C and subsequently centrifuged at 16000 g at 4°C for 20 min. The supernatant was added to 1 volume of acetonitrile and the samples were stored at -80°C until analysis. The pellet was dried and subsequently reconstituted in a lysis buffer (20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 2% SDS, 1 mM EDTA) to allow for protein determination with BCA protein assay (Thermo Fisher Scientific, 23227). Separation of the metabolites was achieved with a ZIC-pHILC column (150 x 4.6 mm, 5 μm; Merck) in combination with the Dionex UltiMate 3000 (Thermo Scientific) liquid chromatography system. The column was kept at 30°C. The mobile phase consisted of 10 mM ammonium acetate pH 6.8 (Buffer A) and acetonitrile (Buffer B). The flow rate was kept at 400 µL/min and the gradient was set as follows: 0 min 20% Buffer B, 15 min to 20 min 60% Buffer B, 35 min 20% Buffer B. Ionization was subsequently achieved by heated electrospray ionization facilitated by the HESI-II probe (Thermo Scientific) using the positive ion polarity mode, and a spray voltage of 3.5 kV. The sheath gas flow rate was 48 units with an auxiliary gas flow rate of 11 units, and a sweep gas flow rate of 2 units. The capillary temperature was 256°C and the auxiliary gas heater temperature was 413°C. The stacked-ring ion guide (S-lens) radio frequency (RF) level was at 90 units. Mass spectra were recorded with the Q Exactive mass spectrometer (Thermo Scientific) and data analysis was performed with the Thermo Xcalibur Qual Browser. Standard curves generated for NAD^+ and NADH were used as references for