The following antibodies were purchased from Abcam (Cambridge, United Kingdom): mt-CO2 (ab79393), VDAC (ab14734), actin (ab8226), COX4 (ab14744), ATF5 (ab184923), SIRT3 (ab217319), COX15 (ab201082), NDUFS4 (ab137064), NAMPT (ab236874), ATF4 (ab184909), CHOP (ab11419), Nrf2 (ab62352), NDUFA9 (ab14713), SIRT1 (ab110304). COQ7 antibody (NBP1-98496) was purchased from Novus Biologicals (Colorado, CO, United States). NDUFAF6 antibody (DPABH-03128) was purchased from Creative Diagnostics (New York, United States). mtND1 (MBS9402205) and mtND3 (MBS8551680) antibodies were purchased from MyBioSource (San Diego, CA, United States). mtHSP70 antibody (MA3-028), HSP60 antibody (MA3-012), Tau antibody (MN1000), MitoTracker Deep Red FM (M22426), Donkey anti-Rabbit IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor 555 (A-31572) and Donkey anti-Mouse IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor 488 (A-21202) were purchased from Thermo Fisher (Waltham, MA, United States). eif2α (5324), P-eif2α (9721), mtND3 (82,933), anti-acetylated lysine 9441) antibodies were purchase from Cell Signaling (Danvers, MA, United States). Galactose (sc-202564), paraformaldehyde (sc-253236B), oligomycin (sc-203342), antimycin A (sc-202467A), FCCP (sc-203578), DAPI (sc-3598), nicotinamide (sc-208096), riboflavin (sc-205906), thiamine (sc-205859), L-carnitine (sc-205727) and HEPES (sc-29097) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, United States). Saponin (S7900-25G), valproic acid (P4543-10G), LDN-1931189 (SML0559-5MG), Db-cAMP (D0260-100MG), CHIR99021 (SML1046-5MG), Goat Anti-Rabbit IgG H&L (HRP) (401353-2ml), Goat Anti-Mouse IgG, H&L Chain Specific peroxidase Conjugate (401253-2ml) and donkey serum (D9663) were purchased from Merck (Darmstadt, Germany). Pterostilbene and 3-TYP was purchased from Cayman Chemical (Ann Harbor, MI, United States). SB431542 (1614/10), Noggin (6057-NG-100), LM-22A4 (4607/10), GDNF (212-GD-010) and NT3 (267-N3-025) were purchased from R&D systems (Minneapolis, MN, United States). PBS (Phosphate Buffer Saline, 102309) 10x was purchased from Intron Biotechnology (Seongnam, South Korea) and then diluted to 1× PBS pH 7.4. BSA (Bovine Serum Albumin, A6588.0100) was purchased from Applichem (Darmstadt, Germany).

Cultured fibroblasts were derived from a skin biopsy of patients (COQ7, COX15, NDUFAF6, NDUFS1, ND3 and NDUFS4) with the following mutations: COQ7: Compound heterozygous mutation in COQ7 gene; c.161_161deIG (p.Val55fs) and c.319C>T (p.Arg107Trp). COX15: Homozygous mutation in COX15 gene; c.C649T (p.Arg217Trp). NDUFAF6: Compound heterozygous mutation in NDUFAF6 gene; c.371T>C (p.Ile124Thr) and c.554_558delTTCTT (p.Tyr187AsnfsTer65). NDUFS1: Homozygous mutation in NDUFS1 gene; c.A755G (p.Asp252Gly). ND3: Mitochondrial heteroplasmic mutation in MT-ND3 gene with ∼80% heteroplasmy; m.10191T>C (p.Ser45Pro). NDUFS4: Homozygous mutation in NDUFS4 gene; c.350+5G>A.

Control fibroblasts were human skin primary fibroblasts from healthy volunteer donors. Samples from patients and controls were obtained according to the Helsinki Declarations of 1964, as revised in 2001. Fibroblasts from patients and controls were cultured at 37°C in DMEM (Dulbecco’s Modified Eagle Medium) containing 4.5 gL⁻¹ glucose, L-glutamine, and pyruvate supplemented with 1% antibiotic Pen-Strep solution (Thermo Fisher, Waltham, MA, United States) and 20% Fetal Bovine Serum (FBS) (Thermo Fisher, Waltham, MA, United States). All the experiments were performed with fibroblasts on a passage number lower than 8.

Drug screening was performed in restrictive culture medium with galactose as main carbon source. Our aim was to deprive cells from glycolysis as energy source (due to the use of galactose) and hence have them rely exclusively on the mitochondrial electron transport chain for ATP production (Coelho et al., 2015; Kamalian et al., 2019). In this condition, their inherent mutations would make mitochondrial patients’ fibroblasts unable to survive in such medium.

Galactose medium was prepared with DMEM without glucose and glutamine and supplemented with 10 mM galactose, 1% antibiotic solution and 10% FBS. Cells were seeded in 24-well plates in DMEM glucose. After 24 h cells were treated for 72 h with several compounds. Then the medium was removed, and cells were washed with PBS prior to the addition of the galactose medium(T0). Then, the treatments were re-applied in the same concentration.

Cell viability was assessed by live cell imaging counting and trypan blue 0.2% staining. Cell counting and representative images were acquired using the BioTek™ Cytation™ 1 Cell Imaging Multi-Mode Reader (Biotek, Winooski, VT, United States).

Western blotting was performed using standard methods. After transferring the proteins to nitrocellulose membranes (BIORAD, Hercules, CA, United States, #1620115), these were incubated with primary antibodies, which were diluted 1:1,000 in BSA 5% overnight. Then washed twice with TTBS and incubated with the corresponding secondary antibody for 1h at 4°C. Secondary antibodies were diluted 1:2,500 in BSA 5%. Multiple blots were run and several proteins of interest were serially detected. Every membrane was checked for protein loading using Ponceau staining and actin protein levels. Stripping was not used. If possible, membranes were re-probed with different antibodies. This is when the molecular weight of the new protein of interest did not interfere with that of the previous one. Moreover, if the proteins were sufficiently separated from one another during gel electrophoresis, membranes were cut and each respective piece was used to detect a different target protein.

Mitochondrial respiratory function of control and mutant fibroblasts was measured using a mito-stress test assay with an XF24 extracellular flux analyzer (Seahorse Bioscience, Billerica, MA, United States, 102340-100) according to the manufacturer’s instructions. Cells were seeded at a density of 1.5 × 10⁴ cells/well with 500 µl growth medium (DMEM medium containing 20% FBS) in XF24 cell culture plates and incubated for 24 h at 37°C, 5% CO₂. Subsequently, growth medium was removed from the wells, leaving on them only 50 µl medium. Then, cells were washed twice with 500 µl of pre-warmed assay XF base medium (102353-100) supplemented with 10 mM glucose (103577-100), 1 mM glutamine (103579-100) and 1 mM sodium pyruvate (103578-100); pH 7.4) and eventually 450 µl of assay medium (500 µl final) were added. Cells were incubated at 37°C without CO₂ for 1 h to allow pre-equilibrating with the assay medium. Mitochondrial functionality was evaluated by sequential injection of four compounds affecting bioenergetics. The final concentrations of the injected reagents were: 1 µM oligomycin, 2 µM FCCP, 1 and 2.5 µM rotenone/antimycin A. The best concentration of each inhibitor and uncoupler, as well as the optimal cells seeding density were determined in preliminary analyses. A minimum of five wells per treatment were used in any given experiment. This assay allowed for an estimation of basal respiration, maximal respiration and spare respiratory capacity. The studied parameters were the following: 1) Basal respiration: Oxygen consumption rate (OCR) used to meet cellular ATP demand resulting from mitochondrial proton leak. It shows energetic demand of the cell under baseline conditions. 2) ATP Production: This parameter is calculated by subtracting the OCR after oligomycin injection from the basal respiration. It shows ATP produced by the mitochondria that contributes to meeting the energetic needs of the cell. 3) Maximal respiration: The maximal OCR attained by adding the uncoupler FCCP. FCCP mimics a physiological “energy demand” by stimulating the respiratory chain to operate at maximum capacity to meet this metabolic challenge. It shows the maximum rate of respiration that the cell can achieve. 4) Spare respiratory capacity: This measurement indicates the capability of the cell to respond to an energetic demand as well as how closely the cell is to respire to its theoretical maximum.

Activity of complex I and complex IV was assessed according to the protocol of the Complex I (ab109720)/Complex IV (ab109876) Enzyme Activity Dipstick Assay Kit (Abcam, Cambridge, United Kingdom). In this technique the proteins from cellular lysates migrate through a nitrocellulose membrane. Then, Complex I is immunocaptured (i.e., immuno-precipitated in active form) on the dipstick. Then, the dipstick is immersed in Complex I activity buffer solution containing NADH as a substrate and nitrotetrazolium blue (NBT) as the electron acceptor. Immunocaptured complex I oxidizes NADH and the resulting H+ reduces NBT to form a blue-purple precipitate at the Complex I antibody line on the dipstick. The signal intensity of this precipitate corresponds to the level of Complex I enzyme activity in the sample. Dipsticks images were taken with a Molecular Imager ChemiDoc XRS+ System (BIORAD, Hercules, CA, United States) and quantified by the Image Lab software.

NAD⁺/NADH levels in cellular pellets were assessed by the NAD⁺/NADH Colorimetric Assay Kit (Abcam, Hercules, CA, United States, ab65348) protocol. Colour intensity was measured using a POLARstar Omega plate reader (BMG Labtech, Offenburg, Germany).

Mitochondrial extracts were obtained using the Mitochondrial Isolation Kit for Cultured Cells (Abcam, Hercules, CA, United States, ab110170). Then, SIRT3 activity was determined by the SIRT3 Fluorometric Activity Assay Kit (Abcam, Hercules, CA, United States, ab156067) protocol. Fluorescence was measured using a POLARstar Omega plate reader (BMG Labtech, Offenburg, Germany).

Neurons were generated from mutant NDUFAF6 and control fibroblasts by direct neuronal reprogramming as previously described by Drouin-Ouellet et al. [28, 29]. Controls and mutant NDUFAF6 patient-derived fibroblasts were plated on μ-Slide 4-Well Ibidi plates (Ibidi, Gräfelfing, Germany) and cultured in DMEM + Glutamax (61965059) with 1% Pen-Strep solution and 10% FBS.

The day after, dermal fibroblasts were transduced with one-single lentiviral vector containing neural lineage-specific transcription factors (ASCL1 and BRN2) and two shRNA against the REST complex, which were generated as previously described with a non-regulated ubiquitous phosphoglycerate kinase (PGK) promoter [30]. The plasmid was a gift from Dr. Malin Parmar (Developmental and Regenerative Neurobiology, Lund University, Sweden). Transduction was performed at a multiplicity of infection (MOI) of 30. On the following day cell culture medium was switched to fresh DMEM and after 48 h to neural differentiation medium (NDiff227; Takara-Clontech, Kusatsu, Japan, Y40002) supplemented with neural growth factors and small molecules at the following concentrations: LM-22A4 (2 μM), GDNF (2 ng/ml), NT3 (10 ng/ml), dibutyryl cyclic AMP (db-cAMP, 0.5 mM), CHIR99021 (2 μM), SB-431542 (10 μM), noggin (50 ng/ml), LDN-193189 (0.5 M) and valproic acid (VPA, 1 mM). Half of the neuronal differentiation medium was refreshed every 2–3 days. Eighteen days post-infection (DPI), the medium was replaced by neuronal medium supplemented with only growth factors until the end of the cellular conversion. At day 21, cells were treated with CoC3 and the medium was changed every 2–3 days for 10 more days. Neuronal cells were identified by the expression of Tau protein, using the anti-TAU clone HT7 antibody from Invitrogen/Molecular Probes (Eugene, OR, United States). Nuclei were stained with DAPI (Invitrogen/Molecular Probes, Eugene, OR, United States, D1306). DAPI+/Tau+ cells were considered iNs. Conversion efficiency was calculated as the number of Tau+ cells over the total number of fibroblasts seeded for conversion. Neuronal purity was calculated as the number of Tau+ cells over the total cells in the plate after reprogramming.

Treated and untreated fibroblasts were grown on 1 mm width glass coverslips for 72 h in normal growth medium with/without the addition of CoC3. Cells were stained with 100 nM MitoTracker DeepRed FM 45 min before fixation. Afterwards, they were washed twice with PBS, fixed in 3.8% paraformaldehyde for 15 min at room temperature, incubated in blocking buffer (BSA 1% in PBS) and permeabilized with 0.1% saponin in blocking buffer for 1 h. In the meantime, primary antibodies were diluted 1:100 in antibody buffer (BSA 0.5% plus saponin 0.1% in PBS). Fibroblasts were incubated overnight at 4°C with the antibodies and subsequently washed twice with PBS. Secondary antibodies were similarly diluted 1:400 antibody buffer, but their incubation time on cells was reduced to 2 h at room temperature. Coverslips were then washed twice with PBS, incubated for 5 min with PBS containing DAPI 1 μg/ml and washed again with PBS. Next, they were mounted on microscope slides using Vectashield Mounting Medium (Vector Laboratories, Burlingame, CA, United States, H1000).

Neurons were grown on 4 μ-SLIDE 4-well plates (Ibidi, Gräfelfing, Germany, 80,426) and stained with 100 nM MitoTracker Deep Red FM 45 min before fixation. Cells were washed with PBS, fixed in 4% paraformaldehyde for 10 min at room temperature, and permeabilized with 0.1% Triton X-100 for 10 min. Then, blocking buffer consisting of PBS 5% donkey serum was added for 1 h. Primary antibodies were diluted 1:100 in PBS 5% donkey serum and incubated on the cells overnight at 4°C. The following morning neurons were washed twice with PBS prior to the addition of the secondary antibodies. These were diluted 1:300 in PBS 5% donkey serum and incubated for 2 h at room temperature. Finally, cells were washed twice with PBS, incubated for 15 min with PBS containing DAPI dilution 1 μg/ml and washed with PBS.

Samples were analyzed using an upright fluorescence microscope (Leica DMRE, Leica Microsystems GmbH, Wetzlar, Germany). Images were taken using a DeltaVision system (Applied Precision; Issaquah, WA, United States) with an Olympus IX-71 microscope using a ×100 objective. Images were analysed using the softWoRx and ImageJ software.

We used non-parametric statistics, which do not consider distributional assumptions, given the low reliability of normality testing for small sample sizes like those used in this work (Le Boedec, 2016). To compare parameters between groups, variables were evaluated using the Wilcoxon match-paired signed rank test, the Friedman Test or a 2-way ANOVA Test. All results were expressed as mean ± standard deviation (SD) of three independent experiments and a p-value < 0.05 was considered as statistically significant.

Control cells and mutant mtND3 or NDUFAF6 cells were cultured for 3 days on glucose-rich DMEM medium. Then, the medium was replaced by galactose medium. As expected, almost no differences could be observed on the growth rate of control cells after the switch to galactose (Figures 1A,D). In contrast, mtND3 and NDUFAF6 mutant cells did not survive in galactose medium (Figures 1E,F). Interestingly, 1 μM pterostilbene treatment, a SIRT3 activator (Li YR. et al., 2018), enabled the survival of mtND3 and NDUFAF6 cells on galactose medium, even though their growth rate was slower than before (Figures 1H,I). The efficacy of pterostilbene was enhanced when it was co-administered with mitochondrial cofactors: nicotinamide 5 μM (Canto et al., 2012), riboflavin 1 μM (Henriques et al., 2016), thiamine 1 μM (Depeint et al., 2006), biotin 1 μM (Rodriguez-Fuentes et al., 2007), lipoic acid 5 μM (Solmonson and DeBerardinis, 2018) and L-carnitine 1 μM (Figures 1K,L). This cocktail plus pterostilbene 1 μM was named CoC3. In contrast, treatment with CoC3 without pterostilbene was insufficient to promote mutant fibroblasts survival and/or growth (Figures 1N,O), indicating that pterostilbene is necessary for CoC3 beneficial effects. Also, individual treatment of each CoC3 component was unable to promote cell survival by their own (Supplementary Figures S1, S2).

To assess the role of SIRT3 activation under CoC3 supplementation control and mutant cells were treated with 3-TYP, a SIRT3 inhibitor, in all experimental conditions. No differences were observed in control cells on glucose or galactose medium (Supplementary Figures S3A,D, S4A). However, mutant cells cultured in galactose medium did not survive even with CoC3 supplementation (Supplementary Figures S3E,F,H,I,K,L, S4B,C). This data suggests that SIRT3 activation may play a crucial role in mutant cell survival and the mechanism of action of pterostilbene and CoC3.

The positive effect of CoC3 treatment was evaluated in six patient cell lines bearing mutations in mtND3 (Figure 2A), NDUFAF6 (Figure 2B), COX15 (Figure 2C), NDUFS1 (Figure 2D), COQ7 (Figure 2E) and NDUFS4 (Figure 2F).

First, mitochondrial protein expression levels were assessed in the 6 cell lines via Western blotting. For each mutant cell line, the affected protein as well as several critical mitochondrial proteins were examined: subunits of complex I (mtND3 and NDUFA9), subunits of complex IV (mtCO2 and nCOX4), VDAC as mitochondrial mass marker and actin as loading control. In addition, the changes in protein expression levels upon treatment with CoC3 were evaluated. Given the broad diversity in genetic backgrounds and mutations, results varied among the mutant cell lines. Notwithstanding, most mitochondrial proteins presented lower expression levels in all mutant cell lines, though to different extents. Interestingly, treatment with CoC3 significantly enhanced mitochondrial protein expression levels in all mutant cell lines (Figures 2A,F; Supplementary Figures S5A–F).

The efficacy of CoC3 treatment in the mutant NDUFAF6 cell line was also checked by an immunofluorescence assay. NDUFAF6 protein signal was low compared to controls associated with mitochondrial network disruption. CoC3 treatment partially corrected both NDUFAF6 protein signal and the normal mitochondrial network morphology (Figure 3A). In addition, MitoTracker signal was markedly low in mutant cells compared to controls, suggesting mitochondrial depolarization. As expected, CoC3 treatment was also able to partially increase mitochondrial membrane potential (Figures 3B,C).

To test the efficacy of the CoC3 treatment in improving mitochondrial activity, complex I and complex IV enzyme activity were determined in control and mutant cells. Complex I activity was tested in mtND3 (Figure 4A), NDUFAF6 (Figure 4B), NDUFS1 (Figure 4D), COQ7 (Figure 4E) and NDUFS4 (Figure 4F) cells. The activity of complex IV was tested in mutant COX15 cells (Figure 4C). The activity of both mitochondrial complexes markedly increased after treatment with CoC3 in all mutant cell lines.

As mutant COQ7 cells showed coenzyme Q₁₀ deficiency (Supplementary Figure S6), cells were also treated with CoC3 as well as with CoC3 plus 4 μM coenzyme Q₁₀ (Figure 4E). However, coenzyme Q₁₀ supplementation did not significantly improve the positive effects of CoC3 alone in complex I activity.

Next, mitochondrial bioenergetics was assessed using the Mito-stress test assay in an XF24 extracellular flux analyzer (Seahorse Bioscience, Billerica, MA). Similarly, to other pathological alterations, mitochondrial bioenergetics was affected in mutant cell lines in different degrees but they all present a general decrease in mitochondrial activity. Confirming the positive effect previously observed, the supplementation with CoC3 significantly restored mitochondrial bioenergetics in all mutant cell lines (Figures 5A,F).

It is known that the mechanism of action of pterostilbene analogues, such as resveratrol (Houtkooper et al., 2013; Regitz et al., 2016) or piceatannol (Papandreou et al., 2015; Hosoda et al., 2021), is explained by their ability to regulate mitochondrial proteostasis. Thereafter, we decided to explore the possibility of UPR^mt activation by pterostilbene plus mitochondrial cofactors (Figure 6). For this, protein expression levels of UPR-related proteins was assessed in treated and untreated cell lines: eif2α and its active form p-eif2α as initiators of the Integrated Stress Response (ISR) (Pakos-Zebrucka et al., 2016; He et al., 2020), Nrf2 as mitochondrial function regulator upon stress conditions (He et al., 2020), CHOP, ATF4 and ATF5 as main UPR^mt effectors in humans (Melber and Haynes, 2018; Wang YT. et al., 2019; Jiang et al., 2020), and HSP60 and HSP70 as mitochondrial chaperones (Ruan et al., 2020). Additionally, we examined the expression levels of SIRT3 as a pterostilbene target (Bai et al., 2018; Wang T. et al., 2019) and NAMPT which catalyzes the rate-limiting step in NAD biosynthesis. Application of the CoC3 treatment led to a significant increase in protein expression levels of ISR, UPR^mt-related proteins and NAMPT in both mutant and control cells (Figures 6A,F; Supplementary Figures S7A–F).

SIRT3 is a canonical mitochondrial sirtuin (Zhang J. et al., 2020) able to regulate several mitochondrial proteins (Yang et al., 2016). Aiming to assess the activity of such sirtuin in a mitochondrial disease context we measured NADH levels and the NAD⁺/NADH ratio, given that these are sirtuins cofactors related with cellular fitness (Supplementary Figures S8A–D). We selected mutant mtND3 and NDUFAF6 cells as representatives cell lines and they were treated with CoC3. Our results showed a marked increase of NAD⁺ content and NAD⁺/NADH ratio after CoC3 treatment in both control and mutant cell lines (Supplementary Figures 8C,D) which is associated with increased NAMPT protein expression levels (Figure 6).

Furthermore, the deacetylase activity of SIRT3 in such patients was measured with two different assays. The first approach consisted in measuring SIRT3 activity through the addition of an acetylated peptide that acts as a fluorescent reporter once deacetylated. Accordingly, SIRT3 activity increased in treated controls and mutant cells (Figure 7A). Second, cells were fractionated into mitochondria, cytoplasm and nucleus and measured acetylation levels in each compartment via Western blot (Figures 7B,C; Supplementary Figures S9A,B). Both in control and mutant cells a significant reduction in mitochondrial proteins’ acetylation was observed upon CoC3 treatment, being such decrease less noticeable for nuclear proteins.

The fibroblast model provided useful information on the pathophysiology of this disease, however, the most affected cell types in the majority of mitochondrial pathologies are muscle cells and/or neurons (Gorman et al., 2016). Therefore, direct reprogramming of mitochondrial diseases patient-derived fibroblasts into induced neurons (iNs) is a valuable tool to understand the pathogenesis of these disorders. For this reason, control and mutant NDUFAF6 fibroblasts were direct-reprogrammed to iNs. Reprogrammed cells presented a typical neuron-like morphology and positive immunoreactivity against Tau, a microtubule-associated protein primarily found in neuronal axons of vertebrates’ brain. In contrast, unprogrammed cells did not show Tau staining.

Tau+ cells were used to calculate neuronal conversion efficiency (Tau+ cells over the total number of fibroblasts seeded for conversion), which was around 10% in control (9 ± 1.8%) and 9% in mutant NDUFAF6 cells (8 ± 1.9%) cells. Neuronal purity (Tau+ cells over the total cells in the plate after reprogramming) was around 13% (10 ± 2.6%) in control cells and up to 12% (10 ± 2.1%) in mutant NDUFAF6 cells.

The efficacy of CoC3 treatment in mutant NDUFAF6 iNs was then evaluated. NDUFAF6 protein levels were examined by immunofluorescence microscopy (Supplementary Figure S10A). Additionally, mitochondrial network integrity and potential was assessed by MitoTracker Deep Red FM staining. In mutant NDUFAF6 iNs, NDUFAF6 protein and MitoTracker signals were markedly low compared to controls (Supplementary Figures S10B,C). Interestingly, CoC3 treatment partially reverted the disease phenotype by increasing both NDUFAF6 protein and MitoTracker signals on mutant NDUFAF6 iNs as previously seen in fibroblasts.