*2.4. Cell Culture*

Human-derived primary skin fibroblast cells (patients and controls) were obtained from the Department of Pediatrics, Salzburger Landeskliniken, Salzburg, Austria. Fibroblasts were obtained by a superficial punch skin biopsy, collected from the patient under local anesthesia. Two individuals, from whom the skin fibroblasts were taken as controls, had no genetic defects and were hospitalized for sepsis. In order to meet the requirement of individual experiments, cells were grown in slightly di fferent media. For proteome and metabolome profiling, as well as blue native-polyacrylamide gel electrophoresis (BN-PAGE), cell lines were cultivated in high glucose Dulbecco's modified eagle medium (DMEM, Thermo Scientific, Waltham, MA, USA, # 31966) containing 4.5 g/<sup>L</sup> glucose, 1 mM pyruvate, and GlutaMAX, and supplemented with 10% fetal bovine serum (FBS, Merck, Darmstadt, Germany, # F7524), 1% penicillin-streptomycin-neomycin (PSN) antibiotic mixture (Thermo, # 15640055) at 37 ◦C in a normoxia incubator with a humidified atmosphere of 5% CO2. Cells were grown to 90% confluency in one T75 or one T300 polystyrene flask (TPP, Trasadingen, Switzerland) in biological triplicates for proteome and metabolome experiments, respectively. For the live-cell respiration assay with the Seahorse XFe96 Analyzer (Agilent, Santa Clara, CA, USA), fibroblasts were cultured in basic DMEM (Thermo, # A14430), supplemented with 1 mM sodium pyruvate (Merck, # P2256), 2 mM L-glutamine (Merck, # G3126), 1% PSN antibiotic mixture (Thermo, # 15640055), 10% dialyzed FBS (Silantes, # 281000900), and 25 mM glucose (Merck, # G7021) in three biological replicates in T150 flasks.

#### *2.5. Metabolite Extraction and Profiling by Targeted LC-MS*

Metabolite extraction was done as reported previously with a minor modification for cell culture samples [32]. In brief, T300 flasks with fibroblasts between the passages of 11–15 at 90% confluence were harvested in triplicates for each experiment. Twenty-four hours before harvest, the cell culture was replenished with fresh medium. In order to keep the original metabolic state of the cell and minimize metabolite degradation, cells were harvested within 2 minutes. Culture medium was aspirated and the cells were rinsed quickly twice with ice-chilled 1× phosphate-bu ffered saline, pH 7.4. Then, 1 mL water was added into the flask, which was immediately shock-frozen in liquid nitrogen. The flask was kept on ice, and cell lysates were collected with a cell scraper (TPP) and transferred into a 15 mL tube for three thaw-and-freeze cycles to extract the metabolites. Metabolites were extracted with methyl tert-butyl ether (MTBE), methanol, and water [32]. The remaining protein pellet was used in the bicinchoninic acid (BCA) protein assay for normalization among samples. Extracts were aliquoted equally into three tubes for later reconstitution in water, acetonitrile, and 50% methanol in acetonitrile, respectively. Additionally, an internal standard mixture, containing chloramphenicol and C13-labeled L-glutamine, L-arginine, L-proline, L-valine, and uracil was added to each sample (10 μM final concentration). A SpeedVac was used to dry the aliquots. Dry residuals were dissolved in three di fferent solvents (1) 100 μL 50% acetonitrile in MeOH with 0.1% formic acid, (2) 100 μL MeOH with 0.1% formic acid for analysis by hydrophilic interaction liquid chromatography (HILIC) column, or (3) 100 μL water with 0.1% formic acid for C18 column mode. The supernatants were transferred to micro-volume inserts. Then, 20 μL per run was injected for subsequent LC-MS analysis.

Over 400 metabolites were selected to cover most of the important metabolic pathways in mammals. Metabolites are very diverse in their chemical properties. Therefore, two di fferent LC columns have been used for metabolite separation: a Reprosil-PUR C18-AQ (1.9 μm, 120 Å, 150 × 2 mm ID; Dr. Maisch, Ammerbuch, Germany) column, and a zicHILIC (3.5 μm, 100 Å, 150 × 2.1 mm ID; Merck). The settings of the LC-MS instrument, 1290 series ultra high pressure liquid chromatography (UHPLC) (Agilent) online coupled to a QTRAP 6500 (Sciex, Foster City, CA) were reported previously [33]. The bu ffer conditions were A1—10 mM ammonium acetate, pH 3.5 (adjusted with acetic acid); B1—99.9% acetonitrile with 0.1% formic acid; A2—10 mM ammonium acetate, pH 7.5 (adjusted with ammonia solution); and B2—99.9% methanol with 0.1% formic acid. All bu ffers were prepared in LC-MS grade water and organic solvents.

A list of all metabolites, including multiple reaction monitoring (MRM) ion ratios, retention times, and Kyoto Encyclopedia of Genes and Genomes (KEGG) or Human Metabolome Database (HMDB) metabolite identifiers can be found in Table S6. Peak integration was performed using MultiQuant software v.2.1.1 (Sciex, Foster City, CA) without any smoothing and reviewed manually. Peak intensities were normalized, first against the internal standards, and subsequently against protein abundances obtained from the BCA assay. The first transition of each metabolite was used for relative quantification between samples and controls. All original LC-MS-generated QTrap wi ff- files, as well as MultiQuant processed peak integration q.session files can be downloaded via http://www.peptideatlas.org/PASS/PASS01195.

#### *2.6. Proteomics Sample Preparation with Label-Free Quantification (LFQ)*

Proteomics sample preparation was done according to a published protocol with minor modifications [34]. Three biological replicates of each patient and control fibroblast cell lines between the passages of 8–11 were harvested from T75 flasks and lysed under denaturing conditions in a bu ffer containing 6 M guanidinium chloride (GdmCl), 5 mM tris(2-carboxyethyl)phosphine, 20 mM chloroacetamide, and 50 mM Tris-HCl pH 8.5. Lysates were denatured at 95 ◦C for 15 min shaking at 800 rpm in a thermal shaker and sonicated in a water bath for 15 min. A small aliquot of cell lysate was used for the BCA assay to quantify the protein concentration. Lysates (100 μg proteins) were diluted with a dilution bu ffer containing 10% acetonitrile and 25 mM Tris-HCl, pH 8.5, to reach a 1 M GdmCl concentration. Then, proteins were digested with 2 μg LysC (MS-grade, Roche, enzyme

to protein ratio 1:50) shaking at 800 rpm at 25 ◦C for 2 h. The digestion mixture was diluted again with the same dilution bu ffer to reach 0.5 M GdmCl. Then, 2 μg trypsin (MS-grade, Roche, enzyme to protein ratio 1:50) was added and the digestion mixture was incubated at 37 ◦C overnight in a thermal shaker at 800 rpm for 14 h. Solid phase extraction (SPE) disc cartridges (C18-SD, Waters, Milford, MA) were used for peptide desalting according to the manufacturer's instructions. Desalted peptides were reconstituted in 0.1% formic acid in water and further separated into four fractions by strong cation exchange chromatography (SCX, 3M Purification, Meriden, CT). Eluates were first dried in a SpeedVac, then dissolved in 20 μL 5% acetonitrile and 2% formic acid in water, briefly vortexed, and sonicated in a water bath for 30 seconds prior injection to nano-LC-MS.

#### *2.7. LC-MS Instrument Settings for Shotgun Proteome Profiling and Data Analysis*

LC-MS/MS was carried out by nanoflow reverse-phase liquid chromatography (Dionex Ultimate 3000, Thermo) coupled online to a Q-Exactive Plus Orbitrap mass spectrometer (Thermo). Briefly, the LC separation was performed using a PicoFrit analytical column (75 μm ID × 55 cm long, 15 μm Tip ID; New Objectives, Woburn, MA) in-house packed with 2.1 μm C18 resin (Reprosil-AQ Pur, Dr. Maisch, Ammerbuch, Germany). Peptides were eluted using a non-linear gradient from 2% to 40% solvent B over 101 min at a flow rate of 266 nL/min (solvent A: 99.9% water, 0.1% formic acid; solvent B: 79.9% acetonitrile, 20% water, 0.1% formic acid). 3.5 kilovolts were applied for nanoelectrospray ionization. A cycle of one full fourier transformation scan mass spectrum (300−1750 *<sup>m</sup>*/*<sup>z</sup>*, resolution of 60,000 at *m*/*z* 200, automatic gain control (AGC) target 1 × 106) was followed by 12 data-dependent MS/MS scans (200–2000 *<sup>m</sup>*/*<sup>z</sup>*, resolution of 30,000, AGC target 5 × 105, isolation window 2 *m*/*z*) with normalized collision energy of 25 eV. Target ions already selected for MS/MS were dynamically excluded for 15 s. In addition, only peptide charge states between two to eight were allowed.

Raw MS data were processed with MaxQuant software (v1.6.0.1) and searched against the human proteome database UniProtKB with 21,074 entries, released in December 2018. Parameters of MaxQuant database searching were a false discovery rate (FDR) of 0.01 for proteins and peptides, a minimum peptide length of seven amino acids, a first search mass tolerance for peptides of 20 ppm and a main search tolerance of 4.5 ppm, and using the function "match between runs". A maximum of two missed cleavages was allowed for the tryptic digest. Cysteine carbamidomethylation was set as fixed modification, while N-terminal acetylation and methionine oxidation were set as variable modifications. Contaminants, as well as proteins identified by site modification and proteins derived from the reversed part of the decoy database, were strictly excluded from further analysis. The MaxQuant processed output files can be found in Table S2, showing peptide and protein identification, accession numbers, percentage of sequence coverage of the protein, q-values, and LFQ intensities. The mass spectrometry data have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository [35] with the dataset identifier PXD009743.

#### *2.8. Experimental Design, Statistical Rationale, Pathway, and Data Analyses*

The correlation analysis of biological replicates and the calculation of significantly di fferent metabolites and proteins were done with Perseus (v1.6.0.2). LFQ intensities, originating from at least two di fferent peptides per protein group, were transformed by log2. Only protein groups with valid values within compared experiments were used for further data evaluation. Statistical analysis was done by a two-sample *t*-tests with Benjamini–Hochberg (BH, FDR of 0.05) correction for multiple testing. Significantly regulated metabolites and proteins between patients and controls were indicated by a plus sign in Tables S1 and S3.

One-way ANOVA with Tukey's multiple comparison test (significance level, alpha = 0.05) was performed using GraphPad Prism 5 to compare concentration ratios of reduced and oxidized glutathione (GSH/GSSG) in patients and controls, as well as data from live cell respiration assay.

For comprehensive proteome data analyses, we applied gene set enrichment analysis (GSEA, v2.2.3) [36] in order to see if a priori defined sets of proteins showed statistically significant, concordant differences between mutations and controls. All proteins with ratios calculated by Perseus were used for GSEA analysis. The Galaxy online tool (https://usegalaxy.org/) was used to calculate the average of the ratios of the few duplicate gene names. We used GSEA standard settings, except that the minimum size exclusion was set to 5 and Reactome v5.2 and KEGG v5.2 were used as gene set databases. The cutoff for significantly regulated pathways was set to be *p*-value ≤ 0.05 and FDR ≤ 0.05.

#### *2.9. Simulations of Electron Transfer between the Iron–Sulfur Clusters of NDUFS1 and Prediction of Protein Stability of p.Phe124Leu in ND5*

We have applied tunneling calculations of electron transfer between N4 and N5 iron–sulfur clusters of the NDUFS1 subunit and studied the consequences of mutations of the key residues involved in the process using a method described previously [37].

The change of the Gibbs free-energy gap, ΔΔG, which measures the gain or loss of protein stability upon mutations, was calculated by the online tool STRUM for the p.Phe124Leu substitution in ND5 [38].

#### *2.10. Measurement of Respiratory Chain Enzyme Activities*

Sample preparation of muscle homogenates for the spectrophotometric assay of enzyme activities was done as reported previously [39–41].

#### *2.11. Live Cell Respiration Assay by Seahorse XFe96*

The operation and the calibration of sensor cartridges of the Seahorse XFe96 instrument were done according to the manufacturer's instructions. A pilot experiment was performed to optimize the cell number at seeding and the concentration of carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP), an uncoupling agen<sup>t</sup> of the mitochondrial electron transfer chain and the ATP synthase. In the final assay, fibroblasts were seeded at 40,000 cells per well and cultured under normoxic condition (5% CO2) for 6 h in order to allow them to attach to the bottom of the culture plate. Then, the culture medium was replaced with the Seahorse assay medium and the plate was transferred to a non-CO2 incubator for 45 min right before the start of the assay. The Mito Stress Test Kit (Agilent, #103015-100) and the Glycolytic Rate Assay Kit (Agilent, #103344-100) were used according to the user manuals. Inhibitor concentrations were used as followed: oligomycin (2 μM), FCCP (1.5 μM), rotenone (0.5 μM), antimycin (0.5 μM), and 2-deoxy-glucose (500 mM).

#### *2.12. Blue Native PAGE, Western Blot, and In-Gel Activity Assay of CI*

Fibroblasts were trypsinized, collected by centrifugation, and lysed in mitochondria isolation and storage buffer (83 mM sucrose, 3.3 mM Tris-HCl, pH 7.0, 0.3 mM ethylenediaminetetraacetic acid (EDTA), 1.7 mM 6-aminohexanoic acid, and protease inhibitor cocktails) by passing them through a needle (Ø 0.45 × 25 mm, 26 G × 1") 30 times on ice, on the basis of published protocols with minor modifications [42,43]. In brief, crude mitochondria fractions were solubilized with digitonin in a ratio of 10 μL 20% digitonin per 20 mg cell pellet. Next, 8 μg solubilized mitochondria per lane were loaded onto the precast NativePAGE 3%–12% gradient Bis-Tris protein gels (Thermo, # BN1001BOX) and run at 4 ◦C with pre-chilled buffers. The blue native PAGE was first run with the dark blue cathode buffer (0.02 % Coomassie Blue G-250) until the running front reached one-third of the gel length. Then, the light blue cathode buffer (0.002% Coomassie Blue G-250) was used to finish the gel running. The gel was further processed for western blot or in-gel activity assay of CI.

Immunodetection of OXPHOS enzymes on blue native gels was performed following an established protocol [43]. The antibodies used in western blot were anti-NDUFS1 (Proteintech, Rosemont, IL, USA, # 12444-1-AP) and all others were purchased from Merck: anti-NDUFS2 (# SAB2702088), anti-NDUFB8 (# HPA003886), anti-UQCRC2 (# HPA007998), and anti-SDHB (# HPA002868).

In-gel activity assays of CI were conducted as described previously [43]. Briefly, blue native gels were incubated in the assay bu ffer directly after electrophoresis. One half of the gel piece was cut out and incubated in the assay bu ffer for 1 hour and documented. The other half, as an identical replicate, was stained with Coomassie G-250 following a published protocol [44].

#### *2.13. Protein Sequence Alignment*

Protein sequences of NDUFS1 and ND5 were aligned using T-Co ffee [45] and visualized in Espript 3.0 [46].
