Mitochondrial respiratory dysfunction disturbs neuronal and cardiac lineage commitment of human iPSCs

Mitochondrial diseases are genetically heterogeneous and present a broad clinical spectrum among patients; in most cases, genetic determinants of mitochondrial diseases are heteroplasmic mitochondrial DNA (mtDNA) mutations. However, it is uncertain whether and how heteroplasmic mtDNA mutations affect particular cellular fate-determination processes, which are closely associated with the cell-type-specific pathophysiology of mitochondrial diseases. In this study, we established two isogenic induced pluripotent stem cell (iPSC) lines each carrying different proportions of a heteroplasmic m.3243A>G mutation from the same patient; one exhibited apparently normal and the other showed most likely impaired mitochondrial respiratory function. Low proportions of m.3243A>G exhibited no apparent molecular pathogenic influence on directed differentiation into neurons and cardiomyocytes, whereas high proportions of m.3243A>G showed both induced neuronal cell death and inhibited cardiac lineage commitment. Such neuronal and cardiac maturation defects were also confirmed using another patient-derived iPSC line carrying quite high proportion of m.3243A>G. In conclusion, mitochondrial respiratory dysfunction strongly inhibits maturation and survival of iPSC-derived neurons and cardiomyocytes; our presenting data also suggest that appropriate mitochondrial maturation actually contributes to cellular fate-determination processes during development.

Mitochondria possess multiple copies of their own genome (mitochondrial DNA; mtDNA) and play some crucial roles in cellular energy metabolism. From the viewpoint of developmental biology, several recent studies have clearly indicated that mitochondria are functionally and morphologically reorganized for adaptation to an embryonic stem cell (ESC)-like intracellular environment during induced pluripotent stem cell (iPSC) generation. [1][2][3][4][5][6] Moreover, mtDNA haplogroups (i.e., genetic population groups that share a common ancestor), which are known to be associated with various phenotypes (e.g., disease susceptibility, environmental adaptation or aging), also affect their intrinsic gene expression signatures involved in pluripotency, differentiation, DNA methylation and mitochondrial energy metabolism. 7 Thus, appropriate mitochondrial rejuvenation or maturation may be one important step for bona fide cellular reprogramming or differentiation, as well as for epigenetic modification or resetting in nuclear DNA.
Most parts of pathogenic mutations in mtDNA-specific tRNA genes responsible for various types of mitochondrial diseases have been reported as heteroplasmy (i.e., wild-type mtDNA and mutant mtDNA coexist within a single cell), and induced mitochondrial dysfunction emerges only when mutation ratios of mtDNA exceed their intrinsic pathogenic thresholds at a cellular level. 8 Mitochondrial diseases caused by heteroplasmic mtDNA mutations present a wide variety of affected tissues and organs (e.g., central nervous system or cardiovascular system) among patients, 9,10 probably due to variations in mutant mtDNA proportions at each tissue and organ level. Therefore, disease-relevant iPSCs carrying heteroplasmic mtDNA mutations will greatly help us to open new avenues for studying the patient-specific definitive genotype-phenotype relationship of affected tissues and organs in mitochondrial diseases. 11 In fact, several groups and we have reported the generation and the application of patient-derived iPSCs carrying various heteroplasmic mtDNA mutations toward in vitro human mitochondrial disease modeling; [12][13][14][15][16][17][18] however, it remains uncertain whether and how such heteroplasmic mtDNA mutations affect particular cellular fate-determination processes during development. Recently, we also demonstrated that mitochondrial respiratory dysfunction caused by a heteroplasmic m.3243 A4G mutation in MT-TL1 gene, 19 which is the most representative mutant mtDNA, strongly inhibits cellular reprogramming but does not affect maintenance of the pluripotent state. 20 Our findings may indicate that the degree of the molecular pathogenic influence of heteroplasmic mtDNA mutations actually changes during cellular lineage-commitment processes along with the degree of functional maturation in mitochondria.
In this study, we established two isogenic iPSC lines carrying different proportions of m.3243 A4G from the same patient; one exhibited apparently normal and the other showed most likely impaired mitochondrial respiratory function. Using these isogenic iPSC lines, we demonstrated that induced mitochondrial respiratory dysfunction triggered by high proportions of m.3243 A4G strongly inhibits maturation and survival of iPSC-derived neurons and cardiomyocytes. Such in vitro neuronal and cardiac maturation defects were also confirmed by using another patient-derived iPSC line carrying quite high proportion of m.3243 A4G. Our presenting data therefore demonstrate that isogenic iPSC lines with different proportions of m.3243 A4G would make enormous contributions as in vitro human cellular disease models to greatly facilitate iPSC-based drug discovery and regenerative therapeutics in mitochondrial diseases.

Results
Generation of patient-derived isogenic iPSC lines carrying different proportions of m.3243A4G. First, we generated two isogenic iPSC lines from the same patient, each of which possessed different proportions of m.3243A4G (approximately 40 and 90% proportions of mutant mtDNA; denoted as P1-3243[40] and P1-3243[90], respectively). We also established two additional iPSC lines, each of which were derived from healthy control subject (denoted as Control) and from another patient carrying over 90% proportion of m.3243A4G (denoted as P2-3243[490]), respectively. We have previously reported that the molecular pathogenic threshold level of m.3243A4G with regard to mitochondrial respiratory function is~90% in patient-derived clonal fibroblasts. 20 We confirmed that no marked difference was observed between all iPSC lines with regard to ESC-like pluripotent characteristics such as pluripotency markers expression and embryoid body (EB)-mediated in vitro spontaneous differentiation into three germ layers (Figures 1a and  b), in addition to pluripotency genes expression and silenced transgenes expression (Supplementary Figures S1A and B). Genetic identity of these isogenic iPSC lines was also verified by analysis of short tandem repeat variations ( Figure 1c). We measured the overall mitochondrial respiration profile of all iPSC lines by a flux analyzer. Although no statistical significance was observed between two isogenic iPSC lines (P1-3243[40] and P1-3243[90]), mitochondrial energy metabolic potentials (e.g., basal respiration and ATP production) of P1-3243[90] iPSC line were both lower than those of P1-3243 [40] iPSC line (Figure 1d). We further analyzed enzymatic activities of mitochondrial respiratory chain complexes in all iPSC lines. In fact, mitochondrial respiratory chain complex I activity was significantly suppressed by over 90% proportion of m.3243A4G (P2-3243[490] vs Control), whereas mitochondrial respiratory chain complex IV activity was apparently unaffected in all iPSC lines (Figure 1e). Although no statistical significance was observed between two isogenic iPSC lines (P1-3243[40] and P1-3243[90]), mitochondrial respiratory chain complex I activity of P1-3243[90] iPSC line was actually lower than that of P1-3243[40] iPSC line. We also randomly selected several iPSC colonies from each patient-derived iPSC line to determine m.3243A4G proportions at each single-iPSC-colony level and found no significant segregation in m.3243A4G proportions during self-renewal of iPSCs throughout this study (i.e., at least 5-10 passages in culture of each iPSC line) (Figure 1f). We therefore concluded that two isogenic iPSC lines with different proportions of m.3243 A4G from the same patient (P1-3243[40] and P1-3243[90]) were successfully established; one exhibited apparently normal and the other showed most likely impaired mitochondrial respiratory function.
Inhibited cardiac maturation triggered by exceeding the pathogenic threshold level of m.3243A4G. Next, we asked whether and how heteroplasmy levels of m.3243A4G affect cardiac maturation (Figure 2a) Movies S1-S3). Of note, no marked difference was observed during the time course of cardiac induction between these iPSC lines (see also Figure 2b). Interestingly, however, all cardiomyocytes derived from P1-3243[90] iPSC line possessed less than 90% proportions of m.3243 A4G (Figure 2e). To confirm the molecular pathogenic influence of m.3243A4G on cardiac lineage commitment, we measured mitochondrial respiratory function of iPSC-derived cardiomyocytes by a flux analyzer; as expected, cardiomyocytes derived from P1-3243[90] iPSC line, all of which exhibited below the molecular pathogenic threshold level of m.3243A4G, showed apparently normal mitochondrial respiration profile and mitochondrial energy metabolic potentials (e.g., basal respiration and ATP production) similarly to those derived from P1-3243[40] iPSC line ( Figure 2f). In addition, we found that one out of five iPSCderived cardiomyocytes showed a significant decrease in m.3243A4G heteroplasmy level (approximately 60% proportion of mutant mtDNA) when compared with the distributions of m.3243A4G heteroplasmy levels in the parental P1-3243 [90] iPSC line (77-92% proportions of mutant mtDNA) (see also Figure 2e). We also added mtDNA copy number analysis for iPSC-derived cardiomyocytes and their parental iPSCs. Cardiomyocytes derived from two isogenic iPSC lines (P1-3243[40] and P1-3243[90]) had more mtDNA copies per cell than those in the parental iPSCs; however, no significant difference in mtDNA copy number was observed between two isogenic iPSC lines (P1-3243[40] and P1-3243[90]), or among their iPSC-derived cardiomyocytes (Supplementary Figure S2). We therefore concluded that mitochondrial respiratory dysfunction caused by exceeding the pathogenic threshold level of m.3243A4G induced cardiac maturation defects.
Induced neuronal cell death triggered by exceeding the pathogenic threshold level of m.3243A4G. We then differentiated these iPSC lines into neurons using our stepwise induction method to clarify whether and how    Figure S3). We therefore concluded that mitochondrial respiratory dysfunction caused by exceeding the pathogenic threshold level of m.3243A4G also induced neuronal cell death; this phenomenon is similar to, but more pronounced than, that in cardiac lineage.
Neuronal maturation defect was also recapitulated by using mtDNA-depleted neuroblastoma cells. We further addressed whether severe mitochondrial respiratory dysfunction, which is triggered by mtDNA depletion, is also able to recapitulate neuronal maturation defect and even neuronal cell death during neuronal lineage commitment. We used SH-SY5Y neuroblastoma cell line (SH-SY5Y WT) to prepare its mtDNA-depleted cell line (SH-SY5Y ρ 0 ) and to differentiate both neuroblastoma cell lines into neurons (Figure 4a). We confirmed, in advance, that SH-SY5Y ρ 0 line showed severe mitochondrial respiratory dysfunction triggered by mtDNA depletion (Figure 4b). In fact, cytochemical staining of cytochrome c oxidase (COX), an indicator of mitochondrial respiration activity, also indicated that SH-SY5Y WT line showed strongly COX-positive, whereas SH-SY5Y ρ 0 line showed COX-negative ( Figure 4d). As we expected, neuronal maturation was markedly suppressed by severe mitochondrial respiratory dysfunction triggered by mtDNA depletion, and in some cells, induced neuronal cell death was also observed in SH-SY5Y ρ 0 line during neuronal lineage commitment (Figure 4c and Supplementary Movies S4 and S5) with similar trend to iPSC-derived neurons carrying high proportions of m.3243 A4G. In this case, the remaining 'survival' neurons in SH-SY5Y ρ 0 line were COX-negative (Figure 4d). Although neuroblastoma cells have several distinct genetic, epigenetic and energy metabolic properties from iPSCs, we concluded that mitochondrial respiratory dysfunction caused by defective mtDNA with various mutation types actually induced neuronal maturation defect and even neuronal cell death in vitro.

Discussion
In this study, we generated two isogenic iPSC lines from the same patient; one exhibited apparently normal and the other showed most likely impaired mitochondrial respiratory function. Using these isogenic iPSC lines, our lineage-specific directed differentiation methods demonstrated that induced mitochondrial respiratory dysfunction triggered by high proportions of m.3243 A4G strongly inhibits maturation and survival of iPSC-derived neurons and cardiomyocytes. Such in vitro maturation defects in both neuronal and cardiac lineages were also confirmed using another patient-derived iPSC line carrying over 90% proportion of m.3243 A4G. In addition to our results, Hämäläinen et al. 15 reported the pathogenic influences of a heteroplasmic m.3243 A4G mutation on neuronal differentiation; briefly, their established patient-origin iPSC-derived neurons carrying approximately 80-85% proportions of m.3243 A4G exhibited specific downregulation of mitochondrial respiratory chain complex I at both transcript and protein levels and showed accelerated mitophagy via the PARKIN-PINK1 pathway, probably due to clearance of damaged mitochondria for further neuronal differentiation and maturation. Taking these previous findings with our presenting data, we propose that appropriate mitochondrial rejuvenation or maturation must be required for bona fide cellular reprogramming or differentiation, and the degree of molecular pathogenic influence of mutant mtDNA actually determines the severity of the cell-type-specific disease phenotypes in vitro, including the differentiation efficiency into particular cell types ( Figure 5).
As we noted above, severe mitochondrial respiratory dysfunction strongly induces neuronal cell death in vitro; however, most parts of mitochondrial disease patients carrying mutant mtDNA undergo normal brain development in vivo before symptomatic appearance. What is the crucial difference between iPSC-based in vitro cellular disease phenotypes and in vivo clinical symptoms? Some previous molecular neuropathological studies using postmortem brain of mitochondrial disease patients found that neuronal cells carrying higher proportions of mutant mtDNA frequently remained in some patients' cerebellar lesions (e.g., dentate nucleus neurons, olivary neurons and Purkinje cells). 21,22 These findings suggest that neuronal cell death does not always correlate with mutant mtDNA proportions, leading to the discrepancy between our iPSC-based in vitro recapitulation of neuronal development and in vivo brain pathology of mitochondrial disease patients. With regard to this discrepancy, it is hypothesized that physiological and physical interaction with other non-neuronal cell types in the brain (e.g., astrocytes) may strongly enhance maturation and long-term survival of neurons having damaged mitochondria. Astrocytes are known to play a role as an energy supplier to neurons through the release of lactate; 23 for example, a co-culture system with astrocytes is generally used for accelerated functional maturation and long-term survival of neurons in vitro. Moreover, the predominant energy metabolic system in astrocytes is glycolysis, while that in neurons is mitochondrial respiration, 23 suggesting no apparent influence of mitochondrial respiration defects on physiological function in astrocytes to support neurons. In fact, we displayed the successive observation of TUJ1-positive neurons derived from P2-3243[490] iPSC line through EB-mediated in vitro spontaneous differentiation, and even this iPSC line did not produce neurons using the directed neuronal differentiation method. On the other hand, aberrant early embryogenesis was reported using fertilized eggs derived from a female mito-mouse carrying 70% proportion of 4696-bp mtDNA NF-H (red). Marked neuronal cell death in SH-SY5Y ρ 0 was also observed at day 8. Representative movies of differentiating neurons derived from SH-SY5Y WT and SH-SY5Y ρ 0 were also shown in Supplementary Movies S4 and S5, respectively. (d) Representative images of cytochemical COX staining for undifferentiated and differentiated SH-SY5Y WT and SH-SY5Y ρ 0 ; COX (brown). Cell nuclei were co-stained with hematoxylin (purple). Both samples of SH-SY5Y WT and SH-SY5Y ρ 0 were stained simultaneously for the same period deletion; 24 this mouse mutant mtDNA exhibited the intrinsic pathogenic threshold level (60-80% proportions of mutant mtDNA) in relation to mitochondrial respiratory function, which was confirmed by in vitro biochemical analysis using transmitochondrial cellular systems and by in vivo phenotypic analysis using several lines of mito-mice. 25 Although there are some experimental differences between their findings and our presenting data (iPSC-based human model vs mouse model, mtDNA point mutation vs mtDNA partial deletion, etc.), the defective in vitro differentiation into particular cell types triggered by severely impaired mitochondrial respiration may be suggestive of such in vivo embryonic lethality.
Mitochondrial diseases present a broad clinical spectrum even among patients carrying the same heteroplasmic mtDNA mutations (e.g., variations in age of onset, in affected tissues and organs, or in disease progression and phenotypic severity), and vice versa, different mtDNA mutations share similar clinical features in mitochondrial diseases. Such clinical phenotypic diversity frequently makes us complicated to understand the overall pathology of mitochondrial diseases; therefore, curable treatments have yet to be established. Thus, our established isogenic iPSC lines from the same mitochondrial disease patient exhibiting either apparently normal or impaired mitochondrial respiratory function must be promising tools not only to recapitulate tissue-and organ-specific disease phenotypes but also to efficiently explore candidate chemical compounds (i) that ameliorate mitochondrial respiratory dysfunction or (ii) that induce reduced mutant mtDNA proportions. We believe that our presenting data display new insights not only into understanding how mitochondrial respiratory dysfunction triggered by heteroplasmic mtDNA mutations influences cellular fate-determining processes but also into facilitating the applications in future iPSC-based drug discovery and regenerative therapeutics in mitochondrial diseases.

Materials and Methods
Patients. This study was approved by NCNP Institutional Review Board and was stringently conducted in accordance with the ethical principles of the 'Declaration of Helsinki'. Patient biopsy was performed for diagnostic purposes only after we received written informed consent with permission to study patient-derived iPSCs.
To evaluate the distributions of m.3243 A4G proportions in each patient-derived iPSC line, we randomly picked up several iPSC colonies from each patient-derived iPSC line to extract DNA for determination of m.3243 A4G proportions at each single-iPSC-colony level.
Characterization of patient-derived iPSCs. Characterization of patientderived iPSCs via detection of pluripotency markers was performed according to our previous report: 17 Briefly, cultured and harvested patient-derived iPSCs were transferred onto MEF-seeded multi-well culture plates and were maintained in primate ESC culture medium at 37°C under humidified atmosphere of 5% CO 2 . Culture medium was changed daily. After 3 days in culture, patient-derived iPSCs were characterized by standard immunocytochemical protocol. Fluorophoreconjugated primary antibodies used were as follows: Cy3-conjugated anti-OCT4 in vitro spontaneous differentiation of patient-derived iPSCs into EB-mediated three germ layers was also performed according to our previous report: 17 Briefly, cultured and harvested patient-derived iPSCs were transferred onto ultra-lowadherent culture dishes (HydroCell; CellSeed, Tokyo, Japan) and were maintained in primate ESC culture medium without bFGF at 37°C under humidified atmosphere of 5% CO 2 . Culture medium was changed every other day. After 7 days in floating culture, emergent EBs were transferred onto Geltrex (Gibco)-coated multi-well culture plates and were maintained in primate ESC culture medium without bFGF at 37°C under humidified atmosphere of 5% CO 2 . Culture medium was changed every other day. After 14 additional days in adherent culture, spontaneously differentiated cells were characterized by standard immunocytochemical protocol. Primary antibodies used were as follows: anti-TUJ1 (1:200 dilution; Abcam, Cambridge, UK), anti-αSMA Figure 5 Graphical summary showing the relationship between mtDNA pathogenic threshold and inhibited neuronal differentiation. iPSC lines carrying over the pathogenic threshold level of m.3243A4G showed neuronal maturation defect and even neuronal cell death during neuronal lineage commitment; however, no cell death was observed during NSC specification and expansion in these iPSC lines, suggesting that m.3243A4G has minimal molecular pathogenic influence on NSCs Short tandem repeat (STR) analysis was performed to confirm the genetic identity of the established isogenic iPSC lines: Briefly, extracted DNA as template (0.5 ng) was amplified using a thermal cycler (GeneAmp PCR System 9700; Applied Biosystems, Waltham, MA, USA) with a PowerPlex 16 HS System kit (Promega, Fitchburg, WI, USA) according to the manufacturer's instructions. The amplified DNA fragments were electrophoresed using a DNA analyzer (ABI PRISM 3130xl; Applied Biosystems). The obtained data were analyzed using GeneMapper Software (Ver. 5.0; Applied Biosystems).
Analysis of mtDNA mutation. Long PCR-based whole-mtDNA sequencing for the patient was performed as described elsewhere 27 with modifications to eliminate any adverse results arising from pseudo-sequences in nuclear DNA: Briefly, extracted DNA as a template (10 ng) was amplified via mtDNA-specific longrange PCR and the following mtDNA-specific nested PCR using a thermal cycler (GeneAmp PCR System 9700; Applied Biosystems). The amplified mtDNA fragments were sequenced using a DNA analyzer (ABI PRISM 3130xl; Applied Biosystems).
Pyrosequencing was performed to determine m.3243 A4G proportions: Briefly, extracted DNA as a template (10-20 ng) was amplified using a thermal cycler (GeneAmp PCR System 9700; Applied Biosystems). The amplified mtDNA fragments were sequenced using a pyrosequencing instrument (PyroMark Q24 Advanced; Qiagen, Venlo, Netherlands) with a PyroMark Q24 Advanced Reagents kit (Qiagen) according to the manufacturer's instructions. The obtained data were analyzed using PyroMark Q24 Advanced Software (Ver. 3.0.0; Qiagen). Primers used are listed in Supplementary Table S1.
Analysis of mtDNA copy number. mtDNA copy number analysis was performed according to our previous report: 20 Briefly, extracted DNA as a template (1 ng) was used for quantitative PCR with a SYBR Green I PCR Master Mix kit (Roche, Basel, Switzerland) according to the manufacturer's instructions. A realtime PCR system (LightCycler 480II; Roche) was used to measure mtDNA copy number per cell. Measurement for each sample was performed in triplicate.
ΔΔC T -based relative quantification method was adopted for data analysis. Primers used are listed in Supplementary Table S1.
Analyses of pluripotency genes expression and transgenes silencing. Reverse transcription was performed with PrimeScript RT Master Mix kit (TaKaRa Bio, Shiga, Japan) according to the manufacturer's instructions. After reverse transcription of extracted total RNA, total cDNA as a template (10 ng) was used for quantitative PCR with a SYBR Green I PCR Master Mix kit (Roche) according to the manufacturer's instructions. A real-time PCR system (LightCycler 480II; Roche) was used to measure pluripotency genes expression and transgenes silencing. Measurement for each sample was performed in triplicate. ΔΔC T -based relative quantification method was adopted for data analysis. Primers used are listed in Supplementary Table S1.
Emergent cardiomyocytes were maintained in cardiac maturation medium and were characterized according to standard immunocytochemical protocol. Primary antibody used was as follows: anti-cTNT (1:200 dilution; ThermoFisher Scientific, Waltham, MA, USA). Secondary antibody used was as follows: AlexaFluor 568 (1:800 dilution; Molecular Probes). Stained samples were observed under a fluorescent microscope (IX71 System; Olympus).
Analysis of mitochondrial respiration. Analysis of mitochondrial respiratory potential was performed using a flux analyzer (Seahorse XF e 24 Extracellular Flux Analyzer; Seahorse Bioscience, North Billerica, MA, USA) with a Seahorse XF Cell Mito Stress Test Kit (Seahorse Bioscience) according to the manufacturer's instructions. Basal respiration and ATP production were calculated to evaluate mitochondrial respiratory function according to the manufacturer's instructions. After the measurement, cells were harvested to count the cell number, and each plotted value was normalized relative to the number of cells used. The detailed procedures are as follows: For iPSCs, several pieces of iPSC colonies were transferred onto each well of Geltrex (Gibco)-coated XF e 24 cell culture plates (Seahorse Bioscience) and were maintained in primate ESC culture medium. After 3 days in culture, iPSCs were equilibrated in unbuffered XF e assay medium (Seahorse Bioscience) supplemented with 10 mM glucose, 1 mM sodium pyruvate and transferred to a non-CO 2 incubator for 1 h before measurement. Oxygen consumption rate (OCR) was measured with sequential injections of 2 μM oligomycin, 1 μM FCCP and each 2 μM of rotenone/ antimycin A.
For iPSC-cardiomyocytes, each beating cardiomyocyte-aggregate at 23 days of differentiation in total was transferred onto each well of Geltrex (Gibco)-coated XF e 24 cell culture plates (Seahorse Bioscience) and was maintained in Cardiac maturation medium. At 26 or 27 days of differentiation in total, beating cardiomyocytes were equilibrated in unbuffered XF e assay medium (Seahorse Bioscience) supplemented with 10 mM glucose and 1 mM sodium pyruvate, and transferred to a non-CO 2 incubator for 1 h before measurement. OCR was measured with sequential injections of 1 μM oligomycin, 0.5 μM FCCP and each 2 μM of rotenone/ antimycin A.
For iPSC-neurons, each neurosphere at 8 days of differentiation in total was transferred onto each well of Geltrex (Gibco)-coated XF e 24 cell culture plates (Seahorse Bioscience) and was maintained in NSC expansion medium for next 6 days, followed by terminal differentiation in Neuron induction medium. At 26 or 27 days of differentiation in total, neurons were equilibrated in unbuffered XF e assay medium (Seahorse Bioscience) supplemented with 10 mM glucose and 1 mM sodium pyruvate, and transferred to a non-CO 2 incubator for 1 h before measurement. OCR was measured with sequential injections of 2 μM oligomycin, 0.5 μM FCCP and each 2 μM of rotenone/antimycin A.
Analysis of mitochondrial respiratory chain complex activity. Analysis of mitochondrial respiratory chain complex activity was performed according to our previous report: 20 Briefly, mitochondrial respiratory complex activity was measured with Complex I Human Enzyme Activity Microplate Assay kit (Abcam) and with Complex IV Human Enzyme Activity Microplate Assay kit (Abcam) according to the manufacturer's instructions, respectively. Cell extracts (150 μg for complex I, 50 μg for complex IV) were used to measure time-dependent absorbance alterations on a multi-well plate reader (SPECTROstar Nano; BMG Labtech, Ortenberg, Germany).