Fancd2 in vivo interaction network reveals a non-canonical role in mitochondrial function

Fancd2 is a component of the Fanconi anemia (FA) DNA repair pathway, which is frequently found defective in human cancers. The full repertoire of Fancd2 functions in normal development and tumorigenesis remains to be determined. Here we developed a Flag- and hemagglutinin-tagged Fancd2 knock-in mouse strain that allowed a high throughput mass spectrometry approach to search for Fancd2-binding proteins in different mouse organs. In addition to DNA repair partners, we observed that many Fancd2-interacting proteins are mitochondrion-specific. Fancd2 localizes in the mitochondrion and associates with the nucleoid complex components Atad3 and Tufm. The Atad3-Tufm complex is disrupted in Fancd2−/− mice and those deficient for the FA core component Fanca. Fancd2 mitochondrial localization requires Atad3. Collectively, these findings provide evidence for Fancd2 as a crucial regulator of mitochondrion biosynthesis, and of a molecular link between FA and mitochondrial homeostasis.

Fanconi anemia (FA) is a genetic disorder associated with congenital developmental defects, bone marrow failure and predisposition to cancers, particularly acute myelogenous leukemia [1][2][3] . In recent decades, the function of FA proteins has been extensively studied and very well documented. As a genetically heterogeneous disease, more than a dozen FA and FA-associated proteins (FANCA, − B, − C, − E, − F, − G, − L, − M, FAAP100, FAAP24, FAAP20, HES1, MHF1, and MHF2) form a core complex 4,5 . In response to DNA damage or DNA replication stress, this multimeric FA core complex monoubiquitinates two downstream FA proteins, FANCD2 and FANCI, which recruit other components of the FA/BRCA pathway to damaged DNA loci and consequently affect DNA replication, cell cycle control and DNA damage repair processes [4][5][6][7][8] .
While many studies point to an essential role for the FA pathway in DNA damage repair and genome maintenance, emerging evidence suggests that FA deficiency causes mitochondrion dysfunction, which may play roles in the pathogenesis of bone marrow failure and leukemia progression in FA. For example, several studies show that FA cells exhibit abnormalities in mitochondrial metabolism, Ca 2 + homeostasis, and gene expression [9][10][11] . Additional studies revealed mitochondrion defects in FA patients and mice [12][13][14][15] . At the molecular level, it has been shown that FA deficiency causes altered mitochondrial morphology and mitochondrial complex defect, and decreased mitochondrial membrane potential and ATP production 11,16,17 . More recently, it has been reported that some of the FA proteins are required for removal of damaged mitochondria and reduction of mitochondrial reactive oxygen species (ROS) 18 . Furthermore, the utility of antioxidants shows in vivo protective effects against the onset of malignancies and bone marrow failure in FA knockout mice 19,20 . Although these studies have indicated a correlation between mitochondrion defects and FA deficiency, direct evidence for a mechanistic link is missing.
In this study, we developed a Fancd2 knock-in mouse model that enabled the identification of a non-canonical function of the FA pathway in mitochondrion biosynthesis. We further demonstrated that Fancd2 localizes in the mitochondrion and associates with the nucleoid complex components Atad3 and Tufm, thus providing a molecular link between FA and mitochondrial homeostasis. tion of the FA pathway, we generated a Fancd2 knock-in mouse model, in which a duel tandem (3XFLAG and HA) tag was inserted at the C-terminus of the endogenous Fancd2 locus ( Fig. 1A; Supplemental Fig. 1A,B). We confirmed that the tagged Fancd2 protein was expressed and could be pulled down by FLAG and HA antibodies using 2-step immunoprecipitation in Fancd2 KI ES cells (Supplemental Fig. 1C). In addition, immunoblotting of mouse embryonic fibroblast (MEF) cells from wild-type (WT) and Fancd2 KI/KI mice with anti-FLAG and anti-Fancd2 antibodies demonstrated that the tagged Fancd2 and the WT endogenous Fancd2 proteins were expressed at roughly the same levels (Supplemental Fig. 1D), indicating that the FLAG-HA tag does not affect the expression of the Fancd2 protein. Furthermore, treatment of Fancd2 KI ES cells showed that both mitomycin C (MMC) and hydrourea (HU) induced monoubiquitination of the tagged Fancd2 protein (Supplemental Fig. 1E). Immunofluorescence microscopy for Fancd2 KI/KI MEF cells with an anti-HA antibody showed that tagged Fancd2 retained the ability to form DNA damage foci in nuclei after MMC treatment (Supplemental Fig. 1F). These results indicate that the tagged Fancd2 protein retains its functionality in the context of the DNA damage response at the cellular level. Fancd2 genomic locus, Fancd2 locus following homologous recombination (targeted DNA) and the final Fancd2 3XFLAG/HA locus following excision of the loxPflanked Neomycin resistance gene (Neo) is illustrated. Fancd2 exons are shown as gray boxes and are numbered. FLAG/HA (red box) tags are also indicated. (B) Tissues isolated from Fancd2-KI mice were analyzed by immunoblotting with the M2 anti-FLAG antibody. Fancd2 is highly expressed in ES cells, testes, spleen, lymph nodes and ovaries. (C) Bone marrow and spleen mononuclear cells isolated from Fancd2-KI mice were sorted and subjected to western blot. Fancd2 is highly expressed in B, erythroid, Lin − , MPP populations and relatively low in T, myeloid, Lin + cells. (D) Fancd2 expression in bone marrow, spleen and testis is dramatically reduced after a single dose (3 mg/kg) of MMC IP injection in mice. (E) Fancd2 expression was analyzed in sorted population cells in bone marrow after MMC treatment. Full-length blot is presented in Supplementary Fig. 4. The data represent a summary of more than three mice of each genotype from two independent experiments. To examine the function of the tagged Fancd2 alleles in vivo, we subjected Fancd2 KI/KI animals to MMC injection. Unlike Fancd2-KO (Fancd2 −/− ) mice, which showed hypersensitivity to MMC and died within 10 days after injection, homozygous Fancd2 KI/KI mice were resistant to MMC (Supplemental Fig. 2A). Histopathological analysis showed that the Fancd2 KI/KI mice had normal testes and ovaries compared to Fancd2 −/− mice, characteristic of severe gonadal defects (Supplemental Fig. 2B) 2 . In the context of hematopoiesis, the colony-forming activity of Fancd2 KI/KI bone marrow progenitor cells was not affected by MMC at a dose that significantly inhibited the proliferation of Fancd2 −/− progenitor cells (Supplemental Fig. 2C). We also examined primary MEF cells derived from the homozygous Fancd2 KI/KI mice for other two FA cellular hallmarks: MMC-induced cell death and G 2 /M arrest 21,22 . Not surprisingly, unlike the Fanca −/− or Fancd2 −/− cells, the Fancd2 KI/KI cells were completely resistant to MMC-induced cell death (Supplemental Fig. 2D) and G 2 /M arrest (Supplemental Fig. 2E). Taken together, the Fancd2 KI/KI mice display normal development and the tagged Fancd2 protein retains full functionality in the DNA damage response.
Tissue-specific expression of Fancd2 in mice. We next determined the expression of the Fancd2 protein in different tissues of Fancd2 KI/KI mice. We found that Fancd2 was highly expressed in the testes, which is consistent with previously reported data 23 . We also observed high levels of Fancd2 expression in ES cells, lymph node, spleen and ovary tissue (Fig. 1B). Since bone marrow failure and later myeloid malignancies are the major symptoms in FA patients 1-3 , we analyzed Fancd2 expression in different hematopoietic lineages. Fancd2 was found highly expressed in B (B220 + ), erythroid (CD71 + Ter119 + ), lineage-negative (Lin − ), committed progenitors and Lin − Sca1 + c-Kit + (LSK) cell populations, and was relatively low in T (CD3e + ), myeloid (Gr-1 + CD11b + ), Lin + cell populations (Fig. 1C). Interestingly, little ubiquitinated (Ub) form of Fancd2 was detected in the fresh tissues or cells, unlike the obvious Ub-Fancd2 in ES cell lines (Fig. 1B).
Since hypersensitivity to MMC-induced DNA damage is the cellular hallmark of FA cells, we examined whether MMC treatment would alter the level of Fancd2 protein in vivo. We found that the levels of the Fancd2 protein were dramatically reduced in bone marrow, spleen, testis, committed progenitors and LSK cells after a single dose (3 mg/kg body weight) of MMC injection (Fig. 1D,E). Unexpectedly, MMC treatment did not induce robust Fancd2 monoubiquitination in the testis (Fig. 1D).

Proteomic analysis of Fancd2-associated proteins. To identify Fancd2 interactors in vivo, we isolated
Fancd2-containing complexes from ES cells, E11.5 mouse embryos, testes and spleen mononuclear cells ( Fig. 2A), as these compartments express high levels of Fancd2 (Fig. 1B). The identities of Fancd2-associated proteins were determined by liquid chromatography and tandem mass spectrometry (LC-MS/MS; Supplementary Table 1). We validated the association of Fancd2 with some of the interactors using anti-Flag pull-down. Among them, we detected known Fancd2-binding partners, including Fanci, Msh2 and DDB1 (Supplemental Fig. 3). Screening the interactors for ontology categories showed enrichment for cell cycle (P < 0.001) and DNA replication (P < 0.001) proteins. Unexpectedly, proteins involved in mitochondrion homeostasis were enriched the most (P < 0.001) (Fig. 2B). The Fancd2 interaction network reveals different patterns of Fancd2-associated polypeptides in these four compartments (Fig. 2C), suggesting the presence of developmental and tissue-specific differences in the repertoire of Fancd2 interactors. Significantly, we observed three proteins, Fanci, Atad3, and Tufm that interacted with Fancd2 in all four tissues (Fig. 2C). We confirmed the interaction of Fancd2 with Atad3 and Tufm in the spleen (Fig. 2D). Fanci is a well-known Fancd2-binding partner 7 ; whereas Atad3 (component of the mitochondrial nucleoid complex) and Tufm (component of mitochondrial translation machinery) are essential for mitochondrion biosynthesis 24-27 . Loss of Fancd2 or Fanca causes dysregulation of mitochondrion genes. Since FA proteins have been shown to regulate oxidative stress and mitochondrion respiration complex enzymes 9,12,13,[16][17][18] , and since mitochondrial Fancd2-interactors show the highest enrichment score in our proteomic analysis ( Fig. 2 above), we decided to study the link between FA and the mitochondrion further. We first examined whether FA deficiency affected mitochondrion morphology. Transmission electron microscopy analysis revealed swollen mitochondria with disorganized cristae in spleen MNCs and MEFs of Fanca− /− and Fancd2− /− mice (Fig. 3A).
Since the FA pathway is essential for the response to replicative stress and repair of DSBs generated during DNA replication, we suspected that the FA proteins were directly involved in mtDNA replication and repair. To our surprise, the mtDNA copy number in the bone marrow and spleen MNCs showed no significant difference between the one year old WT, Fanca− /− and Fancd2− /− mice (Fig. 3B). To determine if mtDNA integrity was impaired in FA-deficient mice, we performed mtDNA sequencing and found no significant difference in mutation rates between WT and Fanca− /− or Fancd2− /− mtDNA of spleen MNCs (Fig. 3C). These results indicate that loss of Fanca or Fancd2 in adult mice has no effect on mtDNA copy number or the mutation frequency, suggesting that mtDNA variance was likely not the cause of the mitochondrion defect in FA.
To search for clues about Fancd2 function in mitochondria, we first analyzed nuclear-and mtDNA-encoded gene expression at the mRNA and protein levels. Quantitative PCR analysis of nuclear genes functioning in the mitochondrion showed specific up-regulation of genes associated with mitochondrial biogenesis (Atad3, Tufm, Tfam) and down-regulation of genes involved in oxidative phosphorylation (Atp1a1, Atp2a2, Scl25a5) in Fancd2 is required for the stability of the mitochondrion nucleoid complex. We then turned our attention to the two mitochondrial Fancd2-interactors, Atad3 and Tufm, which interacted with Fancd2 in all four tissues (Fig. 2C,D). Both Atad3 and Tufm are among the most frequently identified components of the mitochondrion nucleoid complex [24][25][26][27] , which is essential for mitochondrion biosynthesis. In addition, Atad3 and Tufm have been reported to be required for the mitochondrial transcription and translation process [24][25][26][27][28] . Since Atad3 and Tufm are localized exclusively in mitochondria and are present in the mitochondrial nucleoprotein complex 25 , we postulated that Fancd2 might play a role in the mtDNA transcription and translation process, probably by stabilizing the Atad3/Tufm/Tfam nucleoid complex. We first determined whether Fancd2 was localized to the mitochondria. Immunofluorescence staining of tagged-Fancd2 in Fancd2 KI/KI MEFs with an anti-HA antibody showed that Fancd2 was predominantly localized in the nucleus with a small fraction found in the cytosol (Fig. 4A). Cytoplasmic Fancd2 exhibited co-localization with the mitochondrial maker Mito tracker RED (Fig. 4A), suggesting that Fancd2 is also localized to the mitochondria. To confirm this finding, we performed immunoblotting on fractionated lysates of Fancd2 KI/KI MEFs cells. The mitochondrial, cytosol, and nuclear fractions were marked by Atp5a1, GAPDH and Histone2B, respectively. Consistent with the immunostaining result, the Fancd2 protein was detected in both the nuclear and cytosol fractions (Fig. 4B). Importantly, a significant  We next determined whether Fancd2 was an integral component of the mitochondrion nucleoid complex in vivo. To this end, we first performed immunoprecipitation experiments using antibodies specific for Atad3, Tufm, and Aifm (Apoptosis-Inducing Factor, Mitochondrion-Associated), which is a Fancd2 interactor found in testes, spleen and embryos (Fig. 2C,D). Fancd2 co-immunoprecipitated with Atad3, Tufm and Aifm in the lysates of Fancd2 KI/KI spleen MNCs (Fig. 4C). In the meantime, Tufm associated with Atad3, which co-immunoprecipitated with Tfam (Transcription Factor A, Mitochondrial) (Fig. 4C), another component of the mitochondrion nucleoid complex 25 . Collectively, these results suggest that Fancd2 is part of the mitochondrion nucleoid complex, probably through interacting with Atad3 and Tufm (Fig. 4D).
We then asked whether loss of FA proteins would affect the Fancd2/Atad3/Tufm complex by carrying out co-immunoprecipitation experiments using Fancd2 KI/KI , Fanca− /− Fancd2 KI/KI and Fancd2− /− spleen MNCs cells. We observed that the Atad3/Tufm interaction was reduced in the absence of Fanca and Fancd2 (Fig. 4E). Altogether, loss of Fanca or Fancd2 disrupted Fancd2/Atad3/Tufm mitochondrion nucleoid complex, which likely contributes to the mtDNA transcription and translation defect in FA cells.
Atad3 mediates mitochondrial import of Fancd2. Since the expression of numerous genes encoding mitochondrial proteins, including Atad3, is induced by mitochondrial dysfunction leading to a protective response 29,30 , we examined next whether the expression of Fancd2, as an Atad3-interacting partner, also responded to mitochondrial stress. We treated Fancd2 KI/KI MEFs with different mitochondrion respiration complex inhibitors oligomycin (Complex V inhibitor), FCCP (Uncoupling agents), or antimycin A (Complex III inhibitor) combined with rotenone (complex I inhibitor). We observed that the level of Fancd2 was significantly increased 16hr after oligomycin treatment, in the mitochondrion and cytosol fractions but not in the nuclear fraction (Fig. 4F). In addition, more Atad3 was co-immunoprecipitated with Fancd2 in oligomycin treated MEFs (Fig. 4G). These results indicate that Fancd2 is involved in the response of cells to mitochondrial stress.
Atad3 has been found anchored in the inner mitochondrial membrane and to deliver substrate into mitochondria 31 . This prompted us to ask whether Atad3 mediated Fancd2 import to mitochondria. Fancd2 KI/KI MEFs were transfected with siRNA targeting Atad3, Tufm or Tfam, and then analyzed for mitochondrial Fancd2. We obtained significant knockdown (> 80%) for these proteins in the mitochondrion (Fig. 4H). Knocking down Atad3 but not Tufm or Tfam ablated Fancd2 in both resting (untreated) and stressed (oligomycin-treated) mitochondria (Fig. 4I). Thus, Atad3 is indispensable for the localization and importation of Fancd2 to the mitochondrion with or without stress.

Discussion
Most studies on FA cellular and molecular mechanisms have been conducted in cultured cell lines in vitro. For in vivo studies, several FA knockout mice have been reported, to date, with only a few phenotypic features recapitulating FA patients, including hypersensitivity to DNA-crosslinking agents, abnormal gonads and stressed hematopoiesis 32,33 . Studies on FA mice or human patient-derived primary cells identified additional phenotypes including apoptotic response deregulation, oxidative stress hypersensitivity, and mitochondrion deficiency, which all appear to fall outside of DNA damage repair 9-20 . Using an innovative Fancd2 knock-in mouse model, the recent study shows that most Fancd2-binding proteins do not have a recognized function in DNA damage response and repair. Particularly, many Fancd2-interactors are mitochondrion-specific. We also show that Fancd2 is localized in the mitochondrion and associates with the nucleoid complex components Atad3, Tufm and Tfam, all of which are required for mitochondrial biosynthesis. Moreover, the Atad3/Tufm/Tfam complex is disrupted in Fancd2− /− mice and those deficient for the FA core component Fanca. These findings add another facet to the non-repair function of FA proteins.
A major limitation for studying the molecular mechanisms in FA murine models is the lack of antibodies with high-affinity for mouse FA proteins. We generated a 3 × FLAG-HA tagged Fancd2 knock-in mouse model that overcomes this limitation and this has enabled us to identify Fancd2-binding proteins in different organs, thus gaining new insights into FA protein function in vivo. We have shown that the 3 × FLAG/HA tagged Fancd2 is essentially wild-type and that the homozygous Fancd2 KI/KI mice are indistinguishable from their wild-type counterparts. The 3 tandem FLAG is ideal for detecting Fancd2 in a small amount or for immunoprecipitation. Indeed, month old mice spleen MNCs was amplified. The PCR product was subjected to small genome sequencing. The  this duel tag has enabled us to perform 2 step purification of the Fancd2-containing complex with limited amount of materials from different tissues.
A thorough examination of Fancd2 KI/KI mice reveals tissue-and cell type-specific expression patterns that likely correspond to differential Fancd2 function, especially in the reproductive and hematopoietic organs. The general absence of Fancd2 monoubiquitination in the organs of the Fancd2 KI/KI mice are particularly striking and suggest that a low level of DNA repair activity of Fancd2 may be sufficient for the maintenance of tissue homeostasis in vivo. Conversely, we detected a substantial amount of monoubiquitinated Fancd2 in ES cells and MEFs derived from Fancd2 KI/KI embryos, suggesting that environmental exposure during in vitro cell culture may have induced Fancd2 monoubiquitination, likely by triggering cellular stress including DNA damage. These findings demonstrate a very different in vivo operation of the FA pathway with a majority of the Fancd2 protein exhibiting as non-ubiquitination form, which leads us to propose the existence of ubiquitination-independent functions outside of DNA repair. However, it is also possible that the status of Fancd2 monoubiquitination may reflect cell proliferation or tissue turnover.
Our proteomic study reveals that the most predominant Fancd2 interactors in all four tested tissues belong to those involved in mitochondrial homeostasis, suggesting FA deficiency as a potential contributing factor to mitochondrion deficiency. This intriguingly universal interaction between Fancd2 and mitochondrial proteins also lends support to previous observations that FA proteins regulate oxidative stress response and mitochondrion functions 9,12,13,[16][17][18] . Remarkably, we identified two important mitochondrial proteins, Atad3 and Tufm, with which Fancd2 interacts in all four tissues. Atad3 and Tufm are among the most frequently identified components in the mitochondrion nucleoid complex 27 . Since the mitochondrion nucleoid complex participates in mitochondrial protein synthesis characteristic of uniquely coupled transcription and translation processes 34,35 , both Atad3 and Tufm proteins have been demonstrated to be required for mitochondrial transcription and translation processes 24,27 . In contrast to the assumption that Fancd2 impacts mtDNA replication and integrity, we found that Fancd2 interferes with mtDNA-encoded gene transcription and translation, by interacting with, and stabilizing, the Atad3/Tufm/Tfam mitochondrial nucleoid complex. Indeed, our results indicate that Atad3 mediates Fancd2 localization and importation to mitochondria. Additionally, the deregulation of nuclear-encoded mitochondrial proteins in the Fanca-and Fancd2-deficient cells appear to be part of the feedback loop ultimately leading to mitochondrial dysfunction. However, the exact mechanism underling the functional interaction between Fancd2, the Atad3, Tufm, Tfam mitochondrial nucleoid complex and nuclear-coded mitochondrial proteins remains to be further investigated. Also, how Fanci or Fanca regulates Fancd2 in mitochondrial function remains unknown. Nevertheless, these findings shed new light into the potential link between FA and mitochondrial defects.
Further investigation is needed to determine whether the Fancd2-Atad3-Tufm complex formation requires Fancd2 monoubiquitination or an intact FA core complex. Since we do not observe monoubiquitinated Fancd2 in the cytoplasm or mitochondria, we envision another intriguing scenario, where both a non-monoubiquintinated Fancd2 or the Fancd2 protein in cells lacking the key FA core component Fanca could still form complexes with Atad3-Tufm. However, given that mice deficient for several components of the FA core complex (e.g. Fanca, Fancb, Fancc, Fancg) show exactly the same phenotypes as those deleted for Fancd2 [19][20][21][22][23]32,33 , we also speculate that both Fancd2/Fanci monoubiquitination and FA core complex have a crucial role in Fancd2-Atad3-Tufm complex formation.
In summary, the present study employs an innovative Fancd2 knock-in mouse model to demonstrate that, in addition to its well-established role in DNA repair, Fancd2 influences gene expression and post-translational function of mitochondrial proteins Atadf3, Tufm, and Tfam.

Methods
Generation of 3XFLAG/HA-tagged Fancd2 knock-in mice. The C-terminally tagged allele (Fancd2 3XFlag/HA ) was generated by inserting sequences encoding 3XFLAG/HA tags immediately upstream of the STOP codon (TGA) within the forty-forth exon of the Fancd2 gene (Fig. 1A). Fancd2 targeting vectors were constructed by recombineering as previously described 36 . The linearized targeting plasmid was electroporated into 129/B6 F1 hybrid V6.5 embryonic stem cells (Novus biologicals) and DNA from G418-resistant ES cell clones were analyzed. The floxed Neomycin selection marker was then excised by transient expression of Cre recombinase. Positive ES cell lines were then injected into mouse blastocysts. Chimeric animals were obtained and crossed to C57BL/6J mice, and homozygous Fancd2 KI/KI mice on a mixed 129S4, C57BL/6J background were generated. Founder animals were also backcrossed to pure C57BL/6J mice for six generations to generate a population of homozygous Fancd2 KI/KI mice on the C57BL/6J background.
Purification of Fancd2-containing complexes, mass spectrometry and proteomic analysis. For one purification round of Fancd2-containing complexes, we pooled around 40 10 cm dishes of cultured ES cells, or 20 11.5D embryos, or 8 adult testes, or 24 adult spleens. The same number of cells or organs from WT mice was used for mock purifications. Organs were lysed using a Homogenizer, on ice, in 3 volumes of ice-chilled Lysis buffer with Roche Complete proteinase inhibitor. Lysates were cleared by centrifugation (40,000 g, 20 min, at 4 °C), and the supernatant was incubated with 100 ul of agarose beads conjugated to anti-FLAG antibody (M2, Sigma) at 4 °C. After a few washing steps with ice-cold lysis buffer, Fancd2, together with the associated proteins was eluted in 400 ul of the lysis buffer containing 200ug/ml FLAG peptide (Sigma). The eluate was incubated with 20 ul of anti-HA beads (Roche) for 4 hrs at 4 °C. After several washes with the lysis buffer, Fancd2 together with the associated proteins was boiled and run on a NuPAGE Bis-Tris gel (Invitrogen). In parallel, we performed identical "mock" purifications from the same number of organs or tumors dissected from wild-type animals (which do not express tagged Fancd2 All peptides and corresponding spectral count results are shown in Supplementary Table 1. Each protein was evaluated for mock or Fancd2 immunoprecipitated samples. All proteins that were present in mock samples were excluded from further analysis for each tissue. We observed about 200-300 proteins for each tissue. Proteins with more than 1 or 2 peptides were subjected to further study. Toppgene was used for Gene Ontology (GO) categories enrichment analysis. STRING and Cytoscape were used for protein interaction study.
Markus Grompe (Oregon Health & Sciences University) 33,37 , respectively. All animals were sacrificed 2-3 weeks after the last injection for experiments. Mice were maintained on a C57BL/6J background in the animal barrier facility at Cincinnati Children's Hospital Medical Center. Animals were kept in accordance with the protocol approved by the CCHMC Institutional Animal Care and Use Committees. All animal experiments were performed in accordance with the institutional guidelines and were approved by the Institutional Animal Care and Use Committee of Cincinnati Children's Hospital Medical Center (IACUC2013-0159).
MtDNA copy number. Total DNA was extracted from cell samples using the DNeasy Blood and Tissue kit (Qiagen, USA). To quantify the mtDNA copy number, real-time PCR was performed using iTaq Universal SYBR Green Supermix (Bio-rad, USA) with a Bio-Rad CFX system. For determination of the amount of nuclear DNA, the apoB gene and B2M gene were used as a reference. The quantification using the COXII or ND4 region in mtDNA showed similar results. These real-time qPCRs were carried out in quadruplicate for all measurements. All the primers are listed in Supplementary Table S2.
MtDNA sequencing. The integrity of mtDNA was analyzed by long range PCR, using TaKaRa PrimeSTAR GXL DNA polymerase (Qiaqen) to generate an approximately 16 k bp molecule from primers spanning the entire 16 kb mouse mitochondrial genome (forward 5′ -CCCAGCTACTACCATCATTCAAGT-3′ and reverse 5′ -GAGAGATTTTATGGGTGTAATGCGG-3′ ). The PCR product was gel purified and submitted to small genome sequencing at the DNA sequencing facility of CCHMC. The NGS result was analyzed by Nextgene software. siRNA transfection. siRNA for Tufm (Qiagen, S101458919), Atad3 (Qiagen, S100905723), Tfam (Qiagen, S100183974), and Ubr5 (Sigma, EMU08461) was transfected to the MEFs using Lipofectamine ™ RNAiMAX reagent. The effect of targeted siRNA sequences was compared with negative control siRNA (Qiagen).