<scp>FBXL</scp> 4 deficiency increases mitochondrial removal by autophagy

Abstract Pathogenic variants in FBXL 4 cause a severe encephalopathic syndrome associated with mt DNA depletion and deficient oxidative phosphorylation. To gain further insight into the enigmatic pathophysiology caused by FBXL 4 deficiency, we generated homozygous Fbxl4 knockout mice and found that they display a predominant perinatal lethality. Surprisingly, the few surviving animals are apparently normal until the age of 8–12 months when they gradually develop signs of mitochondrial dysfunction and weight loss. One‐year‐old Fbxl4 knockouts show a global reduction in a variety of mitochondrial proteins and mt DNA depletion, whereas lysosomal proteins are upregulated. Fibroblasts from patients with FBXL 4 deficiency and human FBXL 4 knockout cells also have reduced steady‐state levels of mitochondrial proteins that can be attributed to increased mitochondrial turnover. Inhibition of lysosomal function in these cells reverses the mitochondrial phenotype, whereas proteasomal inhibition has no effect. Taken together, the results we present here show that FBXL 4 prevents mitochondrial removal via autophagy and that loss of FBXL 4 leads to decreased mitochondrial content and mitochondrial disease.

Here, we combined three different models, i.e., knockout mice, patient‐derived fibroblasts, and a CRISPR/Cas9 knockout human cell line, to study the physiological role of Fbxl4. We report that Fbxl4 is involved in mitochondrial quality control and that its absence causes an increased lysosomal turnover of mitochondria leading to a decreased cellular mitochondrial content. Although the remaining mitochondria are fully functional, their numbers are insufficient to prevent disease.

Mitochondrial diseases are a heterogeneous group of inherited metabolic disorders characterized by deficient oxidative phosphorylation (OXPHOS) that leads to a variety of secondary metabolic defects and pleiotropic clinical phenotypes (Gorman et al , 2016 ). Mutations that cause mitochondrial disorders can be localized to either nuclear DNA or mtDNA and may affect a variety of mitochondrial processes such as mtDNA expression, biogenesis of the OXPHOS system, mitochondrial protein import, or mitochondrial dynamics. Although a given pathogenic mutation often is rare, the collective burden of such mutations is substantial, making mitochondrial diseases the most common form of inherited metabolic disorders. In order to keep the mitochondrial population of the cell in shape, several quality control mechanisms act on mitochondria. These mechanisms include the cytosolic ubiquitin–proteasome system, several intramitochondrial proteolytic systems, autophagic clearance of damaged mitochondria, and the mitochondria‐derived vesicle pathway (Bragoszewski et al , 2017 ; Zimmermann &amp; Reichert, 2017 ; Pickles et al , 2018 ). Mutations in Parkin and PINK1 cause early‐onset forms of Parkinson's disease, and the corresponding proteins have been reported to be involved in mitochondrial protein control (Hauser &amp; Hastings, 2013 ; Moon &amp; Paek, 2015 ; Hernandez et al , 2016 ), but their in vivo roles remain controversial (Lee et al , 2018 ; Sliter et al , 2018 ). Apart from PARKIN, other ubiquitin ligases, such as FBXO7, MARCH5, and HUWE1, have been related to mitochondrial quality control by directing proteins for proteasomal degradation or by triggering mitophagy (Chen et al , 2017 ; Di Rita et al , 2018 ; Zhou et al , 2018 ). Recently, ubiquitination has also been reported to occur in the inner mitochondrial membrane and this process may possibly contribute to fine‐tune metabolism according to the energetic demand of the cell (Lavie et al , 2018 ). FBXL4 is a nuclear gene that is implicated in control of mitochondrial function as biallelic mutations recently have been linked to encephalopathy associated with an mtDNA maintenance defect syndrome (Bonnen et al , 2013 ; Gai et al , 2013 ). The disease onset varies from the neonatal period to a few years after birth, and affected patients present a wide range of manifestations, including developmental delay, dysmorphology, lactic acidosis, generalized hypotonia, and decreased mtDNA levels (Huemer et al , 2015 ; El‐Hattab et al , 2017 ). FBXL4 mutations are found in ~ 0.7% of all mitochondrial patients and in ~ 14% of children with congenital lactic acidosis, making it one of the most common causes of mitochondrial disease (Dai et al , 2017 ). Despite the high prevalence and severe consequences of pathogenic FBXL4 mutations, the molecular function of the FBXL4 protein has remained poorly understood (Antoun et al , 2015 ; Huemer et al , 2015 ; Wortmann et al , 2015 ; Dai et al , 2017 ; El‐Hattab et al , 2017 ). FBXL4 belongs to a family of F‐box proteins, which typically serve as substrate adaptors in so‐called Skp1‐cullin1‐F‐box (SCF) E3 ubiquitin ligase protein complexes (Skaar et al , 2013 ; Zheng et al , 2016 ). However, F‐box proteins may also have SCF‐independent functions, e.g., as regulators of cellular proliferation, intracellular trafficking, mitochondrial dynamics, and other processes (Nelson et al , 2013 ; Zhou et al , 2018 ). Consequently, SCF‐independent functions of FBXL4 cannot be excluded, although it is suggested to be part of an SCF complex (Van Rechem et al , 2011 ). Previous studies have focused on using patient‐derived fibroblasts to characterize the molecular phenotype of FBXL4 mutations. These studies have shown that loss of FBXL4 is associated with decreased mtDNA levels, decreased levels of OXPHOS protein components, reduced OXPHOS activity, and low oxygen consumption (Bonnen et al , 2013 ; Gai et al , 2013 ), but the precise role of FBXL4 in mitochondria is still unknown. Moreover, cytosolic roles for FBXL4 acting as a component of a ubiquitin ligase complex have been described both in mammalian cell lines and in fruit flies (Van Rechem et al , 2011 ; Li et al , 2017 ). Here, we describe an Fbxl4 knockout mouse and show that it recapitulates important phenotypes present in patients with mitochondrial disease caused by FBXL4 mutations. Using proteomic approaches, we demonstrate that there is a general decrease in mitochondrial proteins accompanied by an increase in lysosomal proteins in Fbxl4 mouse knockout tissues as well as in fibroblasts derived from patients with loss‐of‐function FBXL4 mutations. Surprisingly, expression of nuclear genes encoding mitochondrial proteins and mitochondrial translation remained unaffected in the absence of FBXL4. We present data showing that the molecular phenotype instead is explained by increased autophagic removal of mitochondria, leading to a global decrease in cellular mitochondrial content. Treatment with t

Levels of mtDNA were significantly decreased in liver, kidney, and brain but not in heart and skeletal muscle of Fbxl4 −/− mice at the age of 1 year (Fig 2 A). Also, the mtDNA‐encoded transcripts were significantly lower in all analyzed tissues, whereas the levels of nuclear transcripts remained mostly unaffected (Fig 2 B). Figure 2 Fbxl4 knockout mice have reduced amounts of mt DNA and mitochondrial proteins Relative levels of mtDNA in tissues of 1‐year‐old Fbxl4 knockout mice determined by quantitative real‐time PCR (normalized to controls). Data are presented as mean ± SEM, n = 6 for controls and n = 8 for knockout animals. Student's t ‐test; ** P &lt; 0.01; *** P &lt; 0.001. Transcript levels in indicated tissues of 1‐year‐old Fbxl4 knockout mice determined by quantitative real‐time PCR. Relative mRNA levels in control animals were set as 1 (dashed line). Data are presented as mean ± SEM, n = 6 for controls and n = 8 for knockout animals. Student's t ‐test; * P &lt; 0.05; ** P &lt; 0.01; *** P &lt; 0.001. Western blot analysis of protein steady‐state levels in liver protein extracts from wild‐type ( Fbxl4 + / + ) and Fbxl4 knockout ( Fbxl4 − / − ) animals. Quantification of the Western blot in (C). Signal normalized to the average of controls and presented as mean ± SEM, n = 4 for both controls and knockout animals. Student's t ‐test; ** P &lt; 0.01; *** P &lt; 0.001. Quantitative TMT‐based proteomic analysis of wild‐type and Fbxl4 knockout liver tissue protein extracts. Proteins from three control and three Fbxl4 knockout liver protein extracts were TMT‐labeled, quantified, and analyzed separately as described in Materials and Methods ; the data from control and knockout animals were averaged after the analysis and presented as −log 10 of the adjusted P ‐value versus log 2 fold change (logFC). Mitochondrial proteins were selected according to mouse MitoCarta 2.0 (Calvo et al , 2016 ) and highlighted in red; lysosomal proteins were selected according to hLGDB database (Brozzi et al , 2013 ) and highlighted in yellow. Enrichment analysis of lysosomal and mitochondrial proteins according to the logFC median of each group of proteins. Data are presented as 25–75 percentile box with the indicated median, and whiskers representing the ± 1.5× inter‐quartile range. Proteins included in each group: total = 1,523, lysosome = 50, mitochondria = 324. The decrease in mtDNA and mtDNA‐encoded transcripts can be due to a selective depletion of mtDNA or a general decrease in mitochondrial content in the tissue. To investigate this question, we determined the cellular content of mitochondrial proteins and the protein content of isolated mitochondria by using Western blot and mass spectrometry analyses (Figs 2 C–F and EV2 A–F). The steady‐state levels of selected mitochondrial proteins, as determined by Western blotting, were strongly decreased in whole liver homogenates of Fbxl4 − / − animals, irrespective of if they were encoded by mtDNA or nuclear DNA (Fig 2 C). Tandem mass tag (TMT)‐based mass spectrometry analysis of liver tissue homogenate confirmed a global, pathway‐independent reduction in mitochondrial proteins, whereas there was an enrichment of lysosomal proteins, such as cathepsins (Ctsb, Ctsc, Ctsd, Ctsl) and other hydrolases (Fig 2 E and F, and Dataset EV1 ). A total of 83 proteins showed significant changes, 57 of which were downregulated and 26 upregulated in homogenate from whole Fbxl4 − / − liver (Fig 2 E and Dataset EV1 ). In contrast, label‐free, global proteomic analyses of isolated mitochondria from liver, kidney, and heart showed only mild changes in the proteome affecting, e.g., amino acid degradation pathways and propanoate metabolism, consistent with mild metabolic derangement in Fbxl4 − / − mitochondria. Importantly, we found no changes in the expression of nuclear‐encoded proteins controlling mtDNA maintenance. Analysis of selected proteins by Western blots confirmed the results from proteomics (Fig EV2 B, D and F). The proteomic changes displayed a strong tissue specificity with only six proteins (Pgam5, Grhpr, Acsm5, Spr, Nmat3, and Nit2) consistently changed in mitochondria from all three analyzed tissues (Fig EV2 G and Dataset EV2 ). Kidney and liver, the most affected organs, had 52 commonly changed proteins, mostly belonging to mitochondrial metabolic enzymes and not related to mtDNA maintenance and expression ( Dataset EV2 ). Figure EV2 Protein steady‐state levels in mitochondria of Fbxl4 knockout animals Label‐free proteomic analysis of isolated liver mitochondria in wild‐type versus Fbxl4 knockout animals. Data are represented as log 10 of the P ‐value on y ‐axis against log 2 of knockout‐to‐control ratio (log 2 fold change, logFC) on the x ‐axis. Western blot analysis of protein steady‐state levels in isolated liver mitochondria from wild‐type ( Fbxl4 + / + ) and Fbxl4 knockout ( Fbxl4 − / − ) animals. Label‐free proteomic analysis of isolated kidney mitochondria in wild‐type versus

We generated an FBXL4 knockout RKO cell line by CRISPR/Cas9‐mediated gene disruption to get further insights into the role of FBXL4 in a defined cell culture system. The resulting clone had a 20‐bp‐long deletion in exon 4 of the FBXL4 gene, leading to a frameshift mutation and a stop codon in the open reading frame (Fig 4 A). Figure 4 Lysosomal degradation of mitochondrial proteins is increased in FBXL 4 knockout cells A Genomic DNA sequence from FBXL4 knockout (KO) and wild‐type (WT) RKO cells was compared to the NCBI reference FBXL4 coding sequence (CDS) NM_001278716.2 and showed a frameshift‐causing deletion in the knockout cells. B Steady‐state protein levels in wild‐type (WT) and FBXL4 knockout (KO) RKO cells. C (Upper panel) Pulse‐chase labeling of mitochondrial translation products in wild‐type (WT) and FBXL4 knockout (KO) cells. The cells were incubated for 1 h with a 35 S methionine–cysteine labeling mix in the presence of anisomycin to inhibit cytosolic translation (pulse). After removal of the medium, the cells were incubated in complete growth medium without inhibitors and labeled amino acids for 16 or 24 h (chase). A representative experiment of four biological replicates is shown. A Coomassie‐stained part of the gel is shown as loading control. (Lower panel) Quantification of the MTCO1‐ND4 signal presented as mean ± SEM; n = 4 biological replicates; Student's t ‐test; * P &lt; 0.05. D (Upper panel) Mitochondrial translation products were labeled in wild‐type (WT) and FBXL4 knockout (KO) RKO cells for 60 min as in (C), followed by 24‐h chase in the presence of 5 μM epoxomicin (epoxo) or 20 mM ammonium chloride (NH 4 Cl). A representative experiment of n ≥ 3 biological replicates is shown. A fragment of the Coomassie‐stained gel is shown as loading control. (Lower panel) Quantification of the ND4‐MTCO1 signal. n ≥ 3 biological replicates; data are presented as mean ± SEM. Student's t ‐test; * P &lt; 0.05. E, F Steady‐state levels of indicated proteins in wild‐type and FBXL4 knockout RKO cells treated with NH 4 Cl (E), or epoxomicin (Epoxo, F). The levels of mitochondrial proteins were reduced in the FBXL4 knockout cell line, similar to the findings in FBXL4 mutant patient fibroblasts (Fig 4 B). To distinguish whether the altered levels of mitochondrial proteins were due to alterations in their biogenesis or degradation, we performed 35 S labeling to study protein synthesis followed by a 16‐ to 24‐h chase to monitor protein stability (Fig 4 C). Mitochondrial translation was not affected in the knockout cell line, whereas the amounts of labeled products were significantly reduced after a 24‐h chase, indicating an increased turnover of the newly synthesized proteins in the absence of FBXL4. Given that FBXL4 is an F‐box protein suggested to play a role in proteasomal degradation of proteins, we investigated whether inhibition of proteasomal activity could rescue the observed phenotype. To this end, we performed 35 S labeling of cellular translation products and added the proteasomal inhibitor epoxomicin during the chase phase of the experiment. Epoxomicin treatment increased the differences in protein stability between wild‐type and knockout cells and thus promoted increased mitochondrial protein turnover (Fig 4 D). In contrast, treatment with the lysosomal inhibitor ammonium chloride rescued the phenotype seen during the chase part of the cellular in vitro translation experiment (Fig 4 D). Further analyses showed that LC3‐II, p62/SQSTM1, and ubiquitin conjugates accumulated to the same extent in both wild‐type and FBXL4 KO cells after treatment with the lysosomal inhibitor NH 4 Cl, as well as after treatment with the proteasomal inhibitor epoxomicin (Fig 4 E and F).

The pathophysiology of human FBXL4 deficiency causing an encephalopathic syndrome associated with mtDNA depletion has remained enigmatic. Here, we present evidence that FBXL4 acts as a regulator preventing excessive turnover of mitochondria via lysosomes. We have characterized Fbxl4 knockout mice, FBXL4‐deficient patient fibroblast, and a human FBXL4 knockout cell line and report several lines of evidence showing that loss of FBXL4 leads to increased mitochondrial turnover via lysosomes. Quantitative proteomics showed an overall decrease in mitochondrial content and an increase in lysosomal proteins both in mouse tissues and in human fibroblasts lacking FBXL4. These results led us to hypothesize that mitochondrial content was decreased due to increased lysosomal degradation. This assumption was supported by 35 S labeling experiments, which showed that mitochondrial protein stability in FBXL4 knockout cells was reduced but could be rescued by inhibiting lysosomal hydrolases with ammonium chloride. In further support of the hypothesis, the mito‐QC expression in patient fibroblasts showed an increased number of mitolysosomes consistent with increased mitochondrial turnover. However, the levels of LC3‐II and also p62, a global autophagy marker (Calvo‐Garrido et al , 2019 ), were similar in wild‐type and FBXL4 knockout cells, also there was no difference in levels of these proteins in control and FBXL4 mutant human fibroblast lines. This finding can be explained by an alternative autophagy pathway that is independent of LC3. Such a pathway has been reported in Atg5 knockout mice and is dependent on the Rab GTPase, Rab9a (Kuma et al , 2004 ). This pathway involves the participation of vesicles from the trans‐Golgi that engulf the cargos for subsequent fusion with lysosomes (Kuma et al , 2004 ). There is increasing evidence that Rab GTPases can play an important role during autophagy and several members of this protein family have been related to different steps in the autophagy process (Ao et al , 2014 ). Interestingly, a slight increase in different Rab GTPases was observed in both mouse and human fibroblasts lacking FBXL4, including Rab9a in the case of human fibroblasts, which lends support to the hypothesis of the existence of an alternate autophagy pathway driven by this group of proteins. Another option that cannot be excluded is a role for the MDV pathway which is also independent of LC3 conversion (Sugiura et al , 2014 ). As we did not observe changes in the conversion of LC3‐I to LC3‐II, we assume that the classical pathway for autophagic flux in FBXL4‐deficient cells is not affected. Instead, an alternate autophagy pathway may be upregulated leading to increased mitochondrial removal. This model seemingly contradicts the proposed role of FBXL4 as an F‐box protein in the SCF complex. One possible explanation for this apparent discrepancy implies an important role for FBXL4 in mitochondrial quality control as a substrate‐recognizing subunit of the ubiquitin ligase complex. Consistent with this model, the absence of FBXL4 would lead to accumulation of aberrant mitochondrial proteins that could affect mitochondrial function and thus increase the degradation of mitochondria by the lysosome. However, there is no evidence showing the presence of the SCF complex in mitochondria and it is therefore unlikely that ubiquitination could happen in the mitochondrial intermembrane space (IMS) via FBXL4. Recently, it has been reported that ubiquitination could occur in the inner mitochondrial membrane, but the ubiquitin ligases or complexes catalyzing this reaction have not been identified (Lavie et al , 2018 ). Another possibility, although perhaps also unlikely, is that FBXL4 could recognize unfolded or damaged proteins in the IMS and promote their export to the cytosol for degradation. A somewhat similar mechanism has been shown in yeast, in which unfolded proteins can be retro‐translocated from the IMS to the cytosol through the TOM complex and further degraded by the ubiquitin–proteasomal system in the cytosol (Bragoszewski et al , 2015 ). The phenotypes of Fbxl4 − / − mice agree with the alternate autophagy pathway model. The finding that Fbxl4 − / − embryos are present at Mendelian ratios and develop normally until E13.5 despite having somewhat lower mtDNA levels argues that the mitochondrial phenotype is not the leading cause of the perinatal death. It should be pointed out that neonatal lethality has been reported in many autophagy‐deficient animals, e.g., Atg3 , Atg5 , Atg7, and Atg12 knockouts (Kuma et al , 2004 , 2017 ; Mizushima &amp; Levine, 2010 ). Interestingly, Atg4b knockout mice were reported to be born at non‐Mendelian ratios, but developed almost normally after birth, demonstrating a striking similarity to the Fbxl4 − / − phenotype (Read et al , 2011 ). Although Fbxl4 knockout animals are not autophagy‐deficient, we assume that increased mitophagy flux at birth could imbalance the overall autophagic

For heart, kidney, and liver microscopy analyses, tissues were fixed in 4% PFA, dehydrated, and paraffin‐embedded. Sections of 5 μm in thickness were mounted on glass slides, de‐paraffinized with xylene, and stained with hematoxylin and eosin.

COX/SDH histochemistry was performed as previously described (Filograna et al , 2019 ). Briefly, quadriceps from wild‐type and FBXL4 knockout animals were washed in PBS and frozen in liquid nitrogen‐cooled isopentane. Sections, 14 μm, were mounted in poly‐lysine‐coated glass slides and left air‐dry briefly. Sections were incubated for 45 min at 37°C in COX medium (100 μM cytochrome c, 4 mM diaminobenzidine tetrahydrochloride, catalase (20 μg/mL), and 0.2 M phosphate buffer [pH 7]). Next, slides were washed in PBS and incubated for 30 min at 37°C in SDH medium (130 mM sodium succinate, 200 μM phenazinemethosulphate, 1 mM sodium azide, 1.5 mM nitroblue tetrazolium, and 0.2 M phosphate buffer [pH 7]). Finally, slides were washed in PBS, dehydrated in increasing ethanol concentrations, and mounted for bright‐field microscopy.

Total DNA was isolated with standard phenol/chloroform and proteinase K digestion protocol, and the concentration was measured with Qubit DNA assay (Thermo Fisher Scientific). Quantitative PCR was performed to quantify mtDNA, using 5 ng of total DNA in each reaction. Specific probes for ND1 (mouse) or ND6 (human) were used to detect mtDNA, whereas the Actb probe (for mouse) or the 18S ribosomal RNA probe (for human DNA) was used to detect nuclear DNA. The relative mtDNA levels were calculated as the ratio of mtDNA/nDNA based on the obtained Ct values.

For Lamp2 immunofluorescence, cells were grown on glass coverslips. Coverslips were washed twice with PBS and fixed with 4% PFA in PBS for 15 min. After, cells were incubated for 1 h in 1% BSA, and 0.1% saponin in PBS. Cells were then incubated with Lamp2 antibody (Abcam) in 1% BSA, and 0.1% saponin in PBS at 4°C overnight. Coverslips were washed two times with PBS and then incubated with Alexa 568 anti‐mouse secondary antibody for 1 h at room temperature. Finally, coverslips were stained with DAPI, washed five times with PBS, and mounted using Prolong Diamond Antifade Mountant (Life Technologies). The slides were imaged in a Zeiss LSM700 confocal microscope.

Mito‐QC pBabe vector was kindly provided by Dr. Ian Ganley. Transfection of fibroblast lines was done as previously described (Allen et al , 2013 ). Cells stably expressing mito‐QC were plated on glass coverslips. Coverslips were washed with PBS and fixed with 4% PFA, and 200 mM HEPES‐KOH pH = 7 for 15 min and washed twice with PBS. Coverslips were mounted using Prolong Diamond Antifade Mountant (Life Technologies) and imaged in a Zeiss LSM700 confocal microscope. Obtained images were analyzed using the ImageJ mito‐QC counter semi‐automated tool (Montava‐Garriga et al , 2020 ). Ratio threshold was set to 0.6, radius for smoothing = 1, and standard deviation over mCherry mean intensity = 3. At least three technical replicates were done for each line and &gt; 15 cells analyzed in each replicate.

The protein content of samples was measured by Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific), and equal protein amounts were loaded on a gel. For SDS–PAGE, precast Bolt Bis‐Tris Plus Gels (Thermo Fisher Scientific) were used according to manufacturer's instructions. For Western blot analysis, the gels were blotted and membranes probed with antibodies of interest. Signal detection was done using HRP‐coupled secondary antibodies with enhanced chemiluminescence substrate (Bio‐Rad). A list of the used primary antibodies is provided in Appendix Table S2 . BN–PAGE was performed using the NativePAGE system (Thermo Fisher Scientific). In brief, 75 μg of mitochondria was solubilized in solubilization buffer: 1% (w/v) digitonin, 20 mM Tris–HCl pH 7.4, 0.1 mM EDTA, 50 mM NaCl, and 10% glycerol (v/v). After 10 min of incubation on ice, samples were centrifuged to eliminate non‐solubilized material and supernatant was mixed with 5% w/v Coomassie Brilliant Blue G‐250. Samples were separated in NativePAGE 3–12% Bis‐Tris gels (Thermo Fisher Scientific) followed by staining with Imperial Staining (Thermo Fisher Scientific) or in‐gel complex activity staining as described previously (Matic et al , 2018 ). For complex I staining, gels were soaked in Tris–HCl pH 7.4 buffer containing 0.1 mg/ml NADH and 2.5 mg/ml iodotetrazolium chloride. For complex IV staining, gels were soaked in 50 mM phosphate buffer pH 7.4 with 0.5 M sucrose, 0.5 mg/ml 3,3′‐diaminobenzidine tetrahydrochloride, 1 mg/ml cytochrome c, and 20 μg/ml catalase.

Labeling of mitochondrial translation products with 35 S‐methionine was performed as described previously (Chomyn, 1996 ). Cells were grown in 24‐well plates, extensively washed with PBS, and pre‐incubated in Cys‐Met‐free DMEM medium (Gibco) for 20 min. Next, the cells were treated for 20 min with 100 μg/ml anisomycin (Sigma) to block cytosolic translation, and 3.7 MBq of 35 S‐ l ‐methionine and cysteine mix (PerkinElmer) was added to the cells for 60 min to label newly synthesized mitochondrial translation products. For the chase experiment, cells were extensively washed with PBS and complete DMEM + FBS medium to remove radioactive substances and anisomycin, and then grown for 16–24 h. The radiolabelled translation products were resolved by SDS–PAGE, and the gels were dried and exposed to phosphor screens for radioactive signal detection.

Samples, 2 μg aliquots, were injected into a 50‐cm‐long C18 EASY spray column (Thermo Fisher Scientific) installed in a nano‐LC‐1000 system online coupled to a Orbitrap Fusion mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). The chromatographic separation of the peptides was achieved with the organic gradient as follows: 2–26% AcN in 110 min, 26–35% AcN in 10 min, 35–95% AcN in 5 min, and 95% AcN for 15 min at a flow rate of 300 nl/min. Full mass scans were acquired in m / z 375–1,500 at resolution of R = 120,000 (at m / z 200), followed by data‐dependent HCD fragmentations from maximum 15 most intense precursor ions with a charge state 2+ to 7+. The tandem mass scans were acquired with a resolution of R = 60,000, targeting 5 × 10 4 ions, setting isolation width to m / z 1.4, and normalized collision energy to 35%. For label‐free proteomic analysis, peptides were separated on a 25‐cm, 75‐μm internal diameter PicoFrit analytical column (New Objective) packed with 1.9 μm ReproSil‐Pur 120 C18‐AQ media (Dr. Maisch,) using an EASY‐nLC 1200 (Thermo Fisher Scientific). The column was maintained at 50°C. Buffer A and buffer B were 0.1% formic acid in water and 0.1% formic acid in 80% acetonitrile. Peptides were separated on a segmented gradient from 6 to 25% buffer B for 135 min, from 25 to 31% for 20 min, and from 31 to 50% buffer B for 20 min at 200 nl/min. Eluting peptides were analyzed on a QExactive HF (Thermo Fisher Scientific). Peptide precursor m / z measurements were carried out at 60,000 resolution in the 300–1,800 m / z range. The ten most intense precursors with charge state from 2 to 7 only were selected for HCD fragmentation using 25% normalized collision energy. The m / z values of the peptide fragments were measured in the Orbitrap at 30,000 resolution, using a minimum AGC target of 8e3 and 55‐ms maximum injection time. Upon fragmentation, precursors were put on a dynamic exclusion list for 45 s.

Data were statistically analyzed using GraphPad Prism 8 software (except proteomics data as detailed above). Neither randomization nor blinding was implemented for animal studies and data analysis. All data are presented by means ± SEM. When two groups were compared, a two‐sided unpaired Student's t ‐test was used, while a one‐way ANOVA and Tukey's post hoc analysis were used for multiple group comparison. Mito‐QC data are presented as 5–95% percentile boxplot, and non‐parametric Mann–Whitney test was used for statistical analysis. Genotype distribution in mice and embryos was analyzed using chi‐square test by comparing the obtained percentages against the expected Mendelian percentages. P ‐values &lt; 0.05 were considered statistically significant. The exact number of replicates and P ‐values for each figure are summarized in Appendix Table S3 .

DA, OL, and N‐GL conceived the project, designed the experiments, and wrote the manuscript. DA and OL performed and interpreted the majority of the experiments. DA, OL, CK, and N‐GL advised on methodology and interpreted the data. AS, IA, and MJ performed experiments and analyzed the data. FS contributed to proteomic data analysis. AWe, RWT, and AWr provided critical input for the project. N‐GL supervised the project. All the authors commented on the manuscript.

Appendix Click here for additional data file. Expanded View Figures PDF Click here for additional data file. Dataset EV1 Click here for additional data file. Dataset EV2 Click here for additional data file. Dataset EV3 Click here for additional data file. Source Data for Expanded View Click here for additional data file. Review Process File Click here for additional data file.