Mice lacking the mitochondrial exonuclease MGME1 accumulate mtDNA deletions without developing progeria

Replication of mammalian mitochondrial DNA (mtDNA) is an essential process that requires high fidelity and control at multiple levels to ensure proper mitochondrial function. Mutations in the mitochondrial genome maintenance exonuclease 1 (MGME1) gene were recently reported in mitochondrial disease patients. Here, to study disease pathophysiology, we generated Mgme1 knockout mice and report that homozygous knockouts develop depletion and multiple deletions of mtDNA. The mtDNA replication stalling phenotypes vary dramatically in different tissues of Mgme1 knockout mice. Mice with MGME1 deficiency accumulate a long linear subgenomic mtDNA species, similar to the one found in mtDNA mutator mice, but do not develop progeria. This finding resolves a long-standing debate by showing that point mutations of mtDNA are the main cause of progeria in mtDNA mutator mice. We also propose a role for MGME1 in the regulation of replication and transcription termination at the end of the control region of mtDNA.

M itochondrial diseases represent the most common group of inherited metabolic diseases in humans with a prevalence of about 1 in 5000 1 . These genetically heterogeneous disorders can be caused by mutations in mitochondrial DNA (mtDNA) or in nuclear genes that encode proteins with mitochondrial function 2 . Mutations in nuclear genes cause mtDNA instability resulting in mtDNA depletion or accumulation of deletions and/or point mutations, ultimately leading to impaired oxidative phosphorylation (OXPHOS). The vast majority of mutations causing human mtDNA instability map to genes encoding proteins involved in mtDNA replication, e.g., the catalytic subunit of mtDNA polymerase (POLGA) 3 , the accessory subunit of mtDNA polymerase (POLGB) 4 , the replicative helicase (TWNK) 5 , DNA replication helicase/nuclease 2 (DNA2) 6 , mitochondrial genome maintenance exonuclease 1 (MGME1) 7 and ribonuclease H1 (RNASEH1) 8 , or nucleotide pool regulation, e.g., thymidine phosphorylase (TP) 9 , mitochondrial thymidine kinase (TK2) 10 , deoxyguanosine kinase (DGUOK) 11 , ATP-dependent succinate-CoA ligase (SUCLA2) 12 , and GTP-dependent succinate-CoA ligase (SUCLG1) 1,13,14 . Mitochondrial disorders are highly variable in disease severity and clinical presentation, and display tissue-specific manifestations. The basis for the clinical heterogeneity and tissue specificity of mutations in these ubiquitously expressed genes is unknown. The fact that mtDNA maintenance defects primarily involve high energy demanding tissues, such as brain and muscle, can only partially explain the tissue specificity seen in mitochondrial disorders. Other organs are also frequently affected causing, e.g., cardiomyopathy, diabetes mellitus, liver dysfunction, and optic neuropathy 1 . Extensive in vitro work has led to significant progress in our understanding of the biochemical processes underlying mtDNA maintenance disorders 4,15,16 , but animal models are nevertheless essential to understand the wide range of phenotypes and secondary metabolic consequences of mtDNA instability in different tissues 17 I   II  III  IV  V   II  I  III  IV  V   I  II  III  IV  V   I  I I  I  c Western blot analysis of MGME1 levels in heart, liver, kidney, and spleen mitochondrial extracts of Mgme1 knockout (−/−) and wild-type (+/+) mice and heart mitochondrial extracts of control (L/L) and tissue-specific knockout mice (L/L, cre). Mitochondrial HSP60 was used as a loading control Various animal models have increased our understanding of progression and disease mechanisms in mtDNA maintenance disorders and this will eventually open new avenues for therapeutic intervention 3,10,[18][19][20][21] .
To gain further insight into diseases of defective mtDNA replication, we created a knockout mouse model for the recently described disease gene encoding MGME1 (also known as Ddk1) 7,22 . Loss-of-function mutations in MGME1 were reported to cause a severe multisystem mitochondrial disorder in humans with depletion and rearrangements of mtDNA 7,23 . Patients with lossof-function MGME1 mutations have progressive external ophthalmoplegia, skeletal muscle wasting/weakness, emaciation, respiratory distress, severe dilated cardiomyopathy, microcephalus, mental retardation, and severe gastrointestinal symptoms. Biochemically, MGME1 is a single-stranded DNA nuclease involved in processing of 5′ mtDNA ends generated during replication 7,22,24 . Loss of MGME1 expression, either in siRNA treated cells or in patient fibroblasts, leads to an accumulation of 7S DNA 7,23 , which is the single-stranded DNA species formed by premature replication termination at the end of the control region of mtDNA 23 , thus suggesting a role for MGME1 in repressing formation or increasing turnover of these molecules.
We have studied the in vivo mtDNA replication phenotypes associated with MGME1 deficiency in various mouse tissues of knockout mice. Although MGME1 is not essential for embryonic development, its loss leads to accumulation of multiple deletions and depletion of mtDNA in a range of different mouse tissues. Furthermore, our data show that MGME1 is involved in regulation of heavy-strand replication and transcription termination. Remarkably, we report that MGME1 knockout mice display tissue-specific replication stalling patterns, with different tissues accumulating distinct replication intermediates, showing that this mouse model is a valuable tool to investigate tissuespecific pathology caused by the absence of MGME1.
Widespread distribution of deleted mtDNA in Mgme1 −/− mice. We performed long-extension PCR on total heart DNA to assess mtDNA integrity in Mgme1 −/− mice (Fig. 2a). When we used a primer pair ( Supplementary Fig. 2a) designed to amplify the major arc of mtDNA, which is the region most commonly affected by deletions, we found multiple shorter molecules on long-extension PCR, consistent with the presence of mtDNA rearrangements in Mgme1 −/− mice (Fig. 2a). Moreover, Southern blot analyses of heart DNA from Mgme1 loxP/loxP ; +/Ckmm-cre mice and various tissues of Mgme1 −/− mice showed both mtDNA depletion and a prominent deletion of mtDNA ( Fig. 2b-d). With restriction enzyme mapping, we established that the deleted species was a linear mtDNA molecule of~11 kb that lacks sequences corresponding to the minor arc of mtDNA, which is the shorter region extending between the origin of replication of the leading (heavy, O H ) and lagging (light, O L ) strand of mtDNA. Consistent with the location of the deletion, both SacI (Fig. 2b, d and Supplementary Fig. 2d) and XhoI ( Supplementary Fig. 2b) digested the linear mtDNA species into two fragments. In contrast, EagI, which only digests mtDNA in the minor arc region, did not cut the linear mtDNA molecule ( Supplementary Fig. 2c). Similar to the findings in Mgme1 −/− mice, patients with pathogenic MGME1 mutations display multiple mtDNA deletions in skeletal muscle, blood and urine on longextension PCR analyses 7 , as well as mtDNA depletion and high levels of a linear subgenomic mtDNA molecule on Southern blot analyses of fibroblast mtDNA 7,23 . We have previously reported that the prematurely ageing mtDNA mutator (PolgA mut/mut ) mouse has high levels of a linear subgenomic mtDNA molecule in addition to high levels of point mutations in mtDNA 25,26 . The linear subgenomic mtDNA molecules are very similar in size (Fig. 2b, Supplementary Fig. 2b, d) in Mgme1 −/− and PolgA mut/ mut mice. However, Mgme1 −/− mice accumulate more than double the amount of the 11 kb subgenomic fragment in comparison with PolgA mut/mut mice ( Supplementary Fig. 2d, e). The finding of linear subgenomic mtDNA molecules of a similar size and extension in patients and mice lacking functional MGME1 and in mtDNA mutator mice expressing mutant POLGA suggests that both enzymes function in the same pathway and that there is a common mechanism for the formation of those subgenomic fragments 27 .
Mgme1 knockout mice do not age prematurely. We assessed the mtDNA point mutation load in Mgme1 −/− mice by using a highfidelity polymerase to amplify mtDNA fragments followed by cloning and sequencing of individual clones. We detected wildtype levels of point mutations in liver of young (11 weeks of age)  Supplementary Fig. 3b). The increase in the point mutation load in mtDNA mutator mice is associated with premature onset of ageing-related phenotypes, such as decreased life span, male infertility and anemia 25,28,29 . In contrast, Mgme1 −/− male mice are fertile with normal testis weight, normal sperm count, normal sperm motility and unaltered testis morphology ( Fig. 3a-e).
At 20 weeks of age, Mgme1 −/− mice have normal blood hemoglobin concentration, whereas older animals show a mild decrease in hemoglobin at the age of 70 weeks ( Supplementary  Fig. 4a, b). Importantly, the anemia of Mgme1 −/− mice is much milder than the anemia in mtDNA mutator mice despite the observation that Mgme1 −/− mice have much higher levels of subgenomic deleted mtDNA than mtDNA mutator mice 30 . At the age of 70 weeks Mgme1 −/− mice have moderate anemia without reticulocytosis accompanied by moderate splenomegaly (Supplementary Fig. 4b-d), whereas mtDNA mutator mice at 40 weeks of age have profound anemia, marked reticulocytosis and massive splenomegaly with extramedullary hematopoiesis 25,28 .
Finally, we performed combined cytochrome c oxidase/ succinate dehydrogenase (COX/SDH) enzyme histochemistry of heart, colon and skeletal muscle of Mgme1 −/− mice and controls at the ages of 45 and 70 weeks. We found COX deficient cells in heart and colon from the age of 45 weeks in Mgme1 −/− mice, whereas there were no changes in skeletal muscle (Fig. 4).
Increased stability of 7S DNA in the absence of MGME1. Given the observed effects on mtDNA quantity and integrity in both patients and mice lacking MGME1, we investigated regulation of replication in the control region. Typically, most mtDNA replication events initiated at the O H region are abortive due to premature termination at the end of the control region, which results in the formation of a~650 nt long nascent DNA species (7S DNA) that creates a characteristic triple-stranded DNA structure, the displacement loop (D loop) 31 . Although the smear around 7S DNA in Mgme1 −/− mice samples prevented us from accurate quantification of the 7S DNA band, Southern blot analyses suggested an increase of steady-state levels of 7S DNA in Mgme1 −/− mice ( Fig. 5a, b). To further investigate the mechanism behind the altered levels of 7S DNA levels, we performed in organello mtDNA replication experiments ( Fig. 5c) 32,33 . Freshly isolated mitochondria from heart tissue were pulse labeled for 2 h with 32 P-dATP followed by a one-hour chase to follow the synthesis and stability of de novo synthesized mtDNA (Fig. 5c). Surprisingly, no de novo synthesis of a distinct 7S DNA species was found in heart mitochondria from Mgme1 −/− mice (Fig. 5c, lanes 2 and 6). However, we observed robust ongoing mtDNA synthesis as there was labeling of full-length mtDNA and a smear of shorter mtDNA species in Mgme1 −/− mitochondria, similar to the findings in wild-type mitochondria (Fig. 5c). The replicative DNA helicase TWINKLE is essential for mtDNA replication 34 , and we, therefore, also analyzed Twnk knockout (Twnk loxP/loxP ; +/Ckmm-cre) heart mitochondria to ensure that the observed labeling is due to de novo mtDNA synthesis. As expected, no de novo mtDNA replication products were detected in Twnk knockout mitochondria (Fig. 5c, lanes 10 and 11).
Moreover, we performed chase experiments and found that 7S DNA is less stable than full-length mtDNA in wild-type mitochondria (Fig. 5c, lanes 1, 5, and 9 vs lanes 4, 8, and 12), consistent with previous observations in cultured cells 35 . To investigate the puzzling discrepancy between the low de novo synthesis and the high steadystate levels of 7S DNA in Mgme1 −/− mice, we followed the stability of mtDNA with Southern blot analyses after inhibiting mtDNA replication with the chain-terminating nucleotide analog 2′,3′dideoxycytidine (ddC) in mouse embryonic fibroblasts (MEFs). In the presence of ddC no 7S DNA is observed after three days of treatment. In contrast, 7S DNA is stabilized in the absence of MGME1 and can still be observed after 3 days of ddC treatment   Fig. 5a, b). Ligation-mediated (LM)-PCR analysis showed a tendency of shift towards longer 5′ DNA ends in the absence of MGME1 ( Supplementary Fig. 5a, b). In contrast, LM-PCR analyses as well as 3′ polyadenylation-mediated PCR amplification revealed only modest changes in the 3′ ends of 7S DNA of Mgme1 −/− mice ( Supplementary Fig. 5a). Extended 5′ ends of 7S DNA have also been reported in fibroblasts from patients with MGME1 mutations and it has been speculated that this may explain the increased stability of 7S DNA 23 .
MGME1 interacts with mtDNA replication factors. To further establish a role for MGME1 in mtDNA replication we performed a search for interacting partners by using a proximity-biotinylation assay (BioID) 36 followed by affinity purification and mass spectrometry. The expression of human MGME1-BirA* in cultured human cells leads to an increase of protein biotinylation (Supplementary Fig. 6a) and the BirA tag does not interfere with the mitochondrial localization of the fusion protein ( Supplementary  Fig. 6b). Our results confirm that MGME1 interacts with the catalytic subunit of the mtDNA polymerase (POLGA) (Fig. 6) 23,37 . Furthermore, we identified a number of other mitochondrial replication-related proteins that could be interactors of MGME1, e.g., mitochondrial single-stranded DNA-binding protein (SSBP1), mitochondrial RNA polymerase (POLRMT) and TWINKLE (Fig. 6, Supplementary Data 1). As expected, several naturally biotinylated mitochondrial proteins, such as pyruvate carboxylase (PC), propionyl-CoA carboxylase (PCC), and methylcrotonyl-CoA carboxylase (MCCC1) were also identified in the affinity purification experiments. In addition, we identified very abundant mitochondrial proteins, such as proteases and respiratory chain subunits, which likely represent contaminants as we often find these proteins in various types of protein interaction studies.
Lack of MGME1 influences mitochondrial transcription. Next, we investigated the steady-state levels of mitochondrial rRNAs, tRNAs, and mRNAs in heart tissue from Mgme1 −/− mice. Unexpectedly, there was a decrease in the abundance of

Mgme1 Twnk
Pulse Chase  (Fig. 7a, b, Supplementary Fig. 7a). Promoter-distal LSP transcripts, such as tRNA Asn and tRNA Cys , did not change significantly in Mgme1 −/− mice. Normally, there is frequent termination of LSP transcription at CSB1 to generate the abundant polyadenylated 7S RNA (Fig. 7c) with a poorly understood function 38 . The steady-state levels of 7S RNA were markedly decreased in Mgme1 −/− mice (Fig. 7c, Supplementary Fig. 7b). In contrast, we found markedly increased levels of an antisense Hstrand transcript that spans over the control region, the so called anti control region transcript (ACR) (Fig. 7d). This transcript, which has an opposite sense to LSP transcripts, has previously been reported to increase in abundance in response to decreased mtDNA transcription initiation, decreased mtDNA replication or inhibition of mitochondrial translation [38][39][40]41 . This finding is in nice agreement with previous results showing that in the absence of premature termination of mtDNA replication (leading to decreased de novo formation of 7S DNA) the levels of the ACR transcript are increased 38,41 .
Loss of MGME1 does not affect mitochondrial protein levels.
As loss of MGME1 leads to decreased levels of full-length mtDNA and the formation of high levels of a linear subgenomic mtDNA molecule (Fig. 2), we decided to investigate the steadystate levels of mitochondrial proteins encoded by mtDNA and nuclear genes (Fig. 8). The steady-state levels of (OXPHOS) subunits were normal in mitochondrial protein extracts (Fig. 8a, b) and the organization of the respiratory complexes was unaltered on blue native polyacrylamide gel electrophoresis (BN-PAGE) of mitochondrial extracts from Mgme1 −/− mice (Fig. 8c).
We also found normal levels of proteins involved in mtDNA replication, transcription, and RNA maturation, such as mitochondrial transcription termination factor 1 (MTERF1) 42 (Fig. 8a, b). Despite accumulation of linear subgenomic mtDNA, depletion of mtDNA and replication stalling in young animals, there were no clear OXPHOS defects. This is probably due to compensatory mechanisms acting on various levels of the mtDNA expression axis thus resulting in high threshold levels, i.e., the overall mtDNA expression must be substantially reduced before respiratory chain deficiency occurs.
Tissue-specific replication stalling in Mgme1 knockout mice.
To gain a more detailed insight into the integrity of mtDNA we performed next generation sequencing of mtDNA from liver ( Fig. 9a), heart (Fig. 9d) and brain ( Supplementary Fig. 8) of Mgme1 −/− mice. The relative sequence coverage pattern of mtDNA from the liver of Mgme1 −/− mice (Fig. 9a) was very similar to the previously published pattern in liver from PolgA mut/ mut mice 25,26 and is consistent with the presence of a linear subgenomic mtDNA fragment extending from O H to O L in both Mgme1 −/− (Fig. 2d, Supplementary Fig. 2d) and PolgA mut/mut mice 25 . Further analysis of mtDNA by neutral-neutral twodimensional agarose gel electrophoresis (2DNAGE) showed a prominent and site-specific stalling of mtDNA replication in the region of O L (Fig. 9c) consistent with the sharp decline in sequence coverage after this point. In striking contrast, the sequence coverage pattern of mtDNA from heart (Fig. 9d) and brain ( Supplementary Fig. 8a) of Mgme1 −/− mice showed more of a gradual decrease with the most abundant sequences being present close to O H . 2DNAGE analysis of heart and brain mtDNA from Mgme1 −/− mice showed a generalized increase in the abundance of replication intermediates at all points along the replication fork arc (Fig. 9e, f; Supplementary Fig. 8b, c; Supplementary Fig. 9). This is consistent with a non-specific replication stalling phenotype as documented by the decreased sequence coverage as the distance from the origin of replication increases. In this study, we report that the loss of MGME1 has marked effects on mtDNA replication in vivo. The Mgme1 −/− mice are viable but exhibit an mtDNA replication stalling phenotype that surprisingly shows a tissue-specific pattern and is associated with decreased mtDNA copy number and an accumulation of mtDNA deletions. Furthermore, we report an unexpected role for MGME1 in the regulation of mtDNA replication and transcription termination at the end of the D-loop region. Proximity labeling experiments with the Bio-ID method showed that MGME1 interacts with a number of components of the replication machinery. MGME1, thus, has an important role in the regulation of mtDNA replication.
Loss of MGME1 in mice causes an increase in the steady-state levels of 7S DNA, consistent with the previously suggested role of MGME1 in degrading this mtDNA species 7,23 . Importantly, de novo replication of mtDNA occurs in the absence of MGME1, but the formation of 7S DNA is severely decreased. This striking discrepancy between impaired de novo synthesis and increased steady-state levels of 7S DNA is likely explained by increased stability of these molecules in the absence of MGME1 activity. It has been estimated that the 7S DNA in patient fibroblasts has a four-fold longer half-life than in control cells 23 , and here we report increased stability of the 7S DNA species in Mgme1 −/− MEFs after mtDNA replication inhibition with the chainterminating nucleotide analog ddC. Interestingly, the severely decreased de novo formation of 7S DNA argues that MGME1 may also modulate replication termination at the end of the Dloop region. Support for a regulatory role for MGME1 at the end of the D-loop region comes from the observation that transcription from HSP, which is normally terminated at this region, is aberrantly regulated in the absence of MGME1. Lack of HSP transcription termination results in accumulation of an ACR transcript over the D-loop region. This ACR transcript has also been described to form in response to thiamphenicol or ddC treatment of tissue culture cells 38,40 . We have previously reported in vivo evidence for ACR transcript accumulation in transgenic mice with impaired mtDNA transcription initiation 41 . In this mouse strain, knockout of Tfam in heart, which typically results in severe cardiomyopathy, was rescued by expression of the human TFAM gene. Interestingly, in addition to ACR transcript accumulation in those rescue mice the levels of 7S DNA were significantly reduced. We propose that MGME1 might be part of regulatory switch acting at the end of the D-loop region that controls mtDNA replication and H-strand transcription termination. Similarly, we have recently shown that the mitochondrial transcription termination factor 1 (MTERF1) blocks transcription initiated from LSP, just downstream of the ribosomal transcription unit, preventing formation of an antisense transcript over the D-loop region and interference with the activity of the LSP promoter in the mouse 42 . Moreover, MTERF1 was recently suggested to display contrahelicase activity and would therefore be able to counteract DNA unwinding by the TWINKLE helicase, which is a termination mechanism that may prevent the Interestingly, ChIP sequencing experiments recently demonstrated that TWINKLE may also be important for the regulation of replication at the 3′ end of 7S DNA in the control region 38 . Importantly, the Bio-ID experiments we present here show an interaction between MGME1 and TWINKLE. The prevention of interference between the mitochondrial replication and transcription machineries, as well as prevention of promoter interference seems to require complex regulation on multiple levels.
In line with the above-discussed results, Mgme1 −/− mice display reduced steady-state levels of the transcript originating from the HSP promoter, whereas promoter-proximal transcripts from LSP are increased and promoter-distal transcripts unaltered. It is likely that the ACR transcript, if abundant, interferes with the HSP promoter leading to a decrease of HSP transcription initiation. As a consequence, the transcription apparatus would shift to the LSP promoter thus resulting in an increased transcription initiation from LSP and higher levels of LSP promoter-proximal transcripts. Interestingly, despite boosted proximal transcription from LSP, the 7S RNA levels were decreased in Mgme1 −/− mice. The function of this promoter proximal transcript is unknown, but it is often discussed in the context of replication primer formation 38,47 . We propose that most LSP initiation events are used for DNA synthesis in our In agreement with this model, we report increased sequence coverage reads consistent with a replication initiation boost at O H and an increase of promoter-proximal LSP transcripts in Mgme1 −/− mice. Lack of MGME1 causes mtDNA depletion and accumulation of mtDNA deletions in a range of different mouse tissues. A hallmark of MGME1 deficiency in patient fibroblasts and mice is an 11 kb linear mtDNA fragment spanning the entire major arc of the mtDNA, which has been previously described in mtDNA mutator mice 25 and flies 48 . According to a recently suggested model, the 11 kb linear abnormal replication products arise from failed ligation of the nascent H-strand at O H and are a consequence of non-ligatable flaps created at the origin of replication 27,49 . Numerous studies suggest that mtDNA mutations and deletions contribute to the ageing phenotypes in experimental animals and in humans 13,[50][51][52] . Indeed, mtDNA mutator mice develop progressive premature ageing syndrome phenotypes 25,53 . In addition to the presence of the above mentioned 11 kb subgenomic mtDNA species, the mtDNA mutator mice also accumulate an increased number of point mutations, that most likely drive the ageing phenotype 54 . Consistent with this hypothesis, Mgme1 −/− mice do not accumulate point mutations and do not display a progeroid phenotype. In line with this finding, mtDNA subgenomic fragments have not been detected in tissues from aging mammals further indicating that this lesion on its own does not induce ageing 51 .
Both 2DNAGE and sequence analysis of mtDNA isolated from various Mgme1 −/− tissues revealed remarkable tissue-specific molecular phenotypes. The 2DNAGE analysis of Mgme1 −/− heart and brain showed a general and unspecific accumulation of replication intermediates along the replication arc, suggesting that replication is stalling along the entire restriction fragment. Similarly, downregulation of MGME1 in tissue culture cells induces a comparable non-specific stalling effect 7 . This result was different to what was seen in liver tissue of the Mgme1 −/− mice, where a very prominent stalling site was observed in the vicinity of O L . Furthermore, deep sequencing of liver mtDNA from Mgme1 −/− mice showed a sequence coverage plot that reflects the presence of linear fragments with a large deletion. A similar sequence coverage has been previously reported from liver samples of mtDNA mutator mice 26 . The coverage profiles of Mgme1 −/− samples from brain and heart displayed a different pattern including a pronounced peak of 7S DNA sequence and loss of sequence coverage in the direction of replication. These differences between the liver on the one hand and heart tissue on the other hand are in agreement with the distinct replication stalling profiles revealed by 2DNAGE data from these tissues. The increased reads along the major arc in the liver indicate that replication has trouble passing O L , predominantly replicating the major arc, explaining the specific stalling in the 2DNAGE. The unusual slope in sequence coverage from the heart suggests that replication is initiated faster than it can be completed, resulting in more reads closer to the origin and generalized stalling as seen on the 2D gel experiments. The observed tissue specificity of our Mgme1 −/− mouse model is in accord with the well-known tissue heterogeneity of mitochondrial disorders 55 . The tissue-specific phenotypes may, at least partly, be explained by the observation that diverse tissues have different energy demands and biosynthetic capacities. The synthesis of mtDNA relies tightly on a balanced dNTP pool, which depends on the mitochondrial dNTP salvage pathways and assisted transport of cytosolic dNTPs to mitochondrial matrix 56 . These processes are likely differentially regulated in various tissues, thereby contributing to the tissue-specific manifestations of mitochondrial mtDNA maintenance diseases as exemplified by patients carrying mutations in enzymes important for the mitochondrial nucleotide salvage pathways, e.g., TK2 10 and MPV17 57 .
In conclusion, we show here that MGME1 is a replicationrelated nuclease, necessary for 5′ 7S DNA processing and faithful mtDNA replication in mice. Absence of MGME1 results in tissuespecific replication stalling and accumulation of deleted mtDNA molecules but does not create a progeroid phenotype in affected animals. We propose that MGME1 is part of a termination complex acting at the end of the D-loop region where it modulates mtDNA replication and H-strand transcription termination. The tissue-specific molecular phenotypes associated with MGME1 deficiency suggest that the Mgme1 −/− mice will be a valuable model for investigation of the pathophysiology of mtDNA maintenance disorders. Generation of Mgme1 knockout mice. The targeting vector for disruption of Mgme1 in embryonic stem cells has been generated using BAC clones from the C57BL/6J RPCIB-731 BAC library and has been transfected into the Taconic Artemis C57BL/6N Tac ES cell line. To generate conditional knockout Mgme1 mice, exon III was flanked by loxP sites. The puromycin resistance cassette was introduced as a selection marker and removed by mating of Mgme1 +/loxP-pur mice with transgenic mice ubiquitously expressing Flp-recombinase. Mgme1 +/loxP mice were mated with mice ubiquitously expressing cre-recombinase (β-actin-cre) to generate heterozygous knockout Mgme1 +/− mice. Mgme1 +/− mice were further intercrossed to generate homozygous knockout Mgme1 +/− mice. To obtain tissuespecific (heart and skeletal muscle) knockout mice 34 Mgme1 loxP/loxP mice were crossed with transgenic mice expressing cre-recombinase under the control of the muscle creatinine kinase promoter (Ckmm-cre).
Isolation of mitochondria from mouse tissue. Mitochondria were isolated from mouse tissues using differential centrifugation as previously described 34 . Briefly, fresh tissues were cut, washed with ice cold PBS and homogenized in mitochondrial isolation buffer (MIB) containing 310 mM sucrose, 10 mM Tris-HCl, and 1 mM EDTA by using a Potter S pestle (Sartorius). The homogenate was centrifuged at 1000× g for 10 min at 4°C and the supernatant was subsequently spun at 10,000× g for 15 min at 4°C to isolate mitochondria. Crude mitochondrial pellets were suspended in MIB supplemented with 1 × Complete protease inhibitor cocktail (Roche). Protein concentration was determined by the Bradford method using BSA as a standard. Mitochondria from brain tissue were isolated following protocol from Miltenyi Biotec mitochondrial extraction tissue kit using TOM22 MicroBeads.
DNA extraction and Southern blot analysis. Genomic DNA was isolated by Gentra Puregene Tissue Kit (Qiagen) according to kit instuctions. DNA quantification was performed with the Qubit 1.0 fluorometer (Thermofisher). A volume of 2 μg of DNA were digested with SacI, XhoI, SphI, or EagI and DNA fragments were separated by agarose gel electrophoresis, transferred to nitrocellulose membranes (Hybond-N + membranes, GE Healthcare) and hybridization was carried out with α 32 P-dCTP-labeled probes. For 7S DNA Southern samples were heated for 3 min at 93°C prior to loading. List of probes can be find in Supplementary Data 2.
Uncropped scans of important Southern blots with marker lanes are presented in Supplementary Fig. 10.
In organello replication. A volume of 1 mg of freshly isolated heart mitochondria was resuspended in 0.5 ml of incubation buffer (25 mM sucrose, 75 mM sorbitol, 100 mM KCl, 10 mM K 2 HPO 4 , 0.05 mM EDTA, 5 mM MgCl 2 , 1 mM ADP, 10 mM glutamate, 2.5 mM malate, 10 mM Tris-HCl, pH 7.4) supplemented with 1 mg/ml fatty acid-free bovine serum albumin, 50 µM each of dTTP, dCTP, and dGTP and 20 µCi α-32 P-dATP (3000 Ci/mmol). Incubation was carried out at 37°C for 2 h on a rotating wheel. Chase experiments were followed by incubation with nonradiolabeled dATP (5 mM) for additional hour. After incubation, mitochondria were pelleted at 9000 rpm for 4 min and washed twice with 10% glycerol, 10 mM Tris-HCl, pH 6.8, 0.15 mM MgCl 2 . In the following step DNA isolation and Southern blot analysis were performed as described above.
Northern blot analysis. Northern blot transcript analysis was performed as previously described 34 . Briefly, RNA, 2 μg, was isolated using the ToTALLY RNA Total RNA isolation kit (Ambion), separated in formaldehyde agarose gels and transferred to Hybond-N + membranes (GE Healthcare) by northern blotting. DNA probes, α-32 P-dCTP-labeled, were used for visualization of mRNA and rRNA levels. Different tRNAs and 7S RNA were detected using specific oligonucleotides labeled with γ-32 P-ATP. For the detection of the ACR transcript riboprobe was synthetized using Riboprobe System T7 Kit (Promega). List of probes can be find in Supplementary Data 2.
Library preparation and Illumina pair-end DNA sequencing. Pair-end sequencing was performed by the Max Planck-Genome-centre Cologne, Germany (http:// mpgc.mpipz.mpg.de/home/). mtDNA Quality control was done using the Agilent TapeStation Genomic DNA ScreenTape and the Qubit BR kit. The library preparation protocol/kit was the NEBNext Ultra™ DNA Library Prep Kit for Illumina (NEB) after fragmentation of 150 ng gDNA with the Covaris sonicator to the requested insert size of >500 bp. The sequencing run conditions were 2 × 250 bp on the Illumina HiSeq2500, using HiSeq Rapid v2 Kits from Illumina. Each library was indexed individually with the provided barcodes and sequenced to 6,000,000 reads (tolerance range −30%) The reads were aligned to the C57Bl/6 J mouse mtDNA reference sequence (NC_005089.1), using the corona lite mapping algorithm (Applied Biosystems) with default settings. The first 49 bases of the mtDNA sequence were appended to the end of the reference to avoid that reads fail to align due to the circularity of the mitochondrial genome. This alignment procedure attempts to map each read at full-length to the reference sequence, allowing for at most 6 mismatches for each 50 bp read.
2DNAGE. For two-dimensional gels, mtDNA was isolated from fresh sucrose gradient purified mitochondria from liver and heart tissues and the by sequential phenol-chloroform extraction. The resulting DNA (3 μg per panel) was digested with BclI, precipitated and loaded onto 0.4% agarose without ethidium bromide. First dimension gels were separated at 27 V for 18 h at room temperature, then DNA-containing lanes were excised with a scalpel, rotated 90°counterclockwise, and molten 1% agarose containing 500 ng/ml ethidium bromide was cast around the gel slices. Second-dimension gels were run at a constant 260 mA for 6 h at 4°C. Gels were Southern blotted onto nylon membranes and hybridized with probes detecting either the O H -containing fragment or the O L -containing fragment. Primer sequences used for probe synthesis (5′-3′) were: O H forward, ATCAATGGTTCAGGTCATAAAATAATCATCAAC; O H reverse, GCCTTAGGTGATTGGGTTTTGC; O L forward, TGACTTGTCCCACTAA-TAATCGGAG; and O L reverse, CCCAAAGAATCAGAACAGATGCTG.
ddC treatment of mouse embryonic fibroblasts. 2,3′ Dideoxycytidine (Sigma) was used at a final concentration of 20 µM. Cells were treated for 3 days and afterwards resuspended in fresh medium at a concentration of 1.5-2 × 10 5 cells/ml. Total DNA was isolated by Gentra Puregene Tissue Kit and extraction was performed as instructed in the kit.
COX/SDH double-labeling enzyme histochemistry. COX/SDH double staining was performed as previously described 29 . Briefly, fresh heart and skeletal muscle tissues were dissected and immediately frozen in isopentane chilled with liquid nitrogen. Colon tissue was upon dissection frozen with liquid nitrogen. Tissues were further cryosectioned into sections (10 μm for skeletal muscle and colon; 7 μm for heart), mounted on slides and left to air dry briefly. Freshly prepared buffer A (0.8 ml of 5 mM 3,3′-diaminobenzidine tetrahydrochloride, 0.2 ml of 500 μM cytochrome c and 10 μl of catalase) was added to the slides. After incubation for 60 minutes at 37°C, slides were washed three times by 0.1 M phosphate buffered saline, pH 7.0. Then freshly prepared buffer B (0.8 ml 1.875 mM of nitroblue tetrazolium, 0.1 ml 1.3 M of sodium succinate, 0.1 ml 2.0 mM phenazine methosulphate, and 10 μl of 100 mM sodium azide) was applied and incubated for 30 minutes at 37°C. Slides were washed three times with 0.1 M phosphate buffered saline, pH 7.0, dehydrated and mounted for bright-field microscopy.
Sperm motility analysis. Sperm motility analysis was performed using Computerassisted semen analysis (CASA) as previously described 29 . Briefly, the cauda region of the right epididymis of each mouse was clamped proximally and distally, excised and rinsed briefly in pre-warmed PBS, and placed in a 1.5 ml Eppendorf tube containing fresh, pre-warmed M2 medium (SIGMA). The cauda epididymis was then unclamped and pierced with the point of a scalpel blade to allow sperm to diffuse into the medium. Sperm were allowed to disperse for 10 min at 37°C and were then appropriately diluted with fresh medium to permit sperm motility analysis of 10 μL sperm suspension by CASA detection (Hamilton Thorne Research Inc. Beverly, MA, USA).
Data availability. All the data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Information. The Illumina sequencing data is data freely available in the NCBI Sequence Read Archive under the Study ID SRP132950 under Accession Numbers SRX3712521-SRX3712536 (https://www.ncbi.nlm.nih.gov/Traces/study/?acc=SRP132950)[http://www.ncbi. nlm.nih.gov/bioproject/434306]. The additional data related to this paper may be requested from the authors.