Linear mtDNA fragments and unusual mtDNA rearrangements associated with pathological deficiency of MGME1 exonuclease

MGME1, also known as Ddk1 or C20orf72, is a mitochondrial exonuclease found to be involved in the processing of mitochondrial DNA (mtDNA) during replication. Here, we present detailed insights on the role of MGME1 in mtDNA maintenance. Upon loss of MGME1, elongated 7S DNA species accumulate owing to incomplete processing of 5′ ends. Moreover, an 11-kb linear mtDNA fragment spanning the entire major arc of the mitochondrial genome is generated. In contrast to control cells, where linear mtDNA molecules are detectable only after nuclease S1 treatment, the 11-kb fragment persists in MGME1-deficient cells. In parallel, we observed characteristic mtDNA duplications in the absence of MGME1. The fact that the breakpoints of these mtDNA rearrangements do not correspond to either classical deletions or the ends of the linear 11-kb fragment points to a role of MGME1 in processing mtDNA ends, possibly enabling their repair by homologous recombination. In agreement with its functional involvement in mtDNA maintenance, we show that MGME1 interacts with the mitochondrial replicase PolgA, suggesting that it is a constituent of the mitochondrial replisome, to which it provides an additional exonuclease activity. Thus, our results support the viewpoint that MGME1-mediated mtDNA processing is essential for faithful mitochondrial genome replication and might be required for intramolecular recombination of mtDNA.

Primer extension mapping of RNA-DNA 5' ends in human control and MGME1deficient patient fibroblasts (A) Reverse transcription activity of Taq polymerase. PEx was performed on ssRNA-DNA chimeric substrates (RNA8-DNA25 and RNA15-DNA18) or ssDNA (DNA33) using a radioactively labelled, 17-nt-long primer as indicated (top). Reactions were analysed by 15 % urea PAGE and subjected to autoradiography. (B) PEx was performed using total DNA preparations extracted from human control and patient fibroblasts (FB1976), which were not initially treated with RNase A, using a radioactively labelled primer annealing at nt position 16,569-18 of the human mtDNA. In the last two lanes the same analysis was performed using samples pretreated with RNaseH1. (C) siRNA downregulation of MGME1 in human cells. Western blot of total HeLa cell lysate transfected with two different siRNAs to MGME1 for 3 and 6 days as indicated. siRNA to GFP was used as a transfection control. β-actin serves as a loading control.

Supplementary Figure 4
Overexpression of MGME1 alters steady-state levels and length of 7S DNA (A) Western blot (short and long exposures) of HEK293T cells expressing WT MGME1 under a doxycycline inducible promoter (uninduced and induced cells), control fibroblasts, and fibroblasts transduced with WT MGME1 using two different lentiviral titers (1x and 10x) as previously described (1). Note: Flag.Strep2 tag causes an increase in molecular weight of MGME1 in HEK293T cells (B) Total DNA from cells expressing the wild-type version of MGME1 (left) or the K253A catalytic mutant (right) in a 10-day time course was restricted using EcoRI and BamHI and analysed by 1D Southern blotting using a probe specific for the NCR of human mtDNA (mtDNA region: 16,341-151; detecting both restriction fragment-length mtDNA and 7S DNA). (C) Mapping of DNA 5' ends in the human mtDNA NCR of HEK293T cells expressing the wild-type version of MGME1 (left) or the K253A catalytic mutant (right) in a 10-day time course using primer extension (PEx). A radioactively-labelled primer annealing at nt position 16,569-18 of the human mtDNA was extended using a thermal cycler. Samples were analysed on a 4 % denaturing polyacrylamide gel and subjected to autoradiography. A sequencing reaction from the same primer was run alongside in order to identify the PEx bands. H, non-transfected control. CSB, conserved sequence block. LSP, light-strand promoter.  The 11-kb sub-genomic fragment lacks the minor arc of the mtDNA (A) Restriction mapping for the 11-kb sub-genomic fragment of human mtDNA as per Figure 3B of the main text, but using a DNA probe located in the minor arc of the mtDNA (nt 607-1,204). The black bar indicates the probe binding site. Note the absence of shorter bands, as opposed to Figure 3B. (B) Digested FB1976 mtDNA retains a proportion of 7S DNA unless heat treated. 6 µg of total FB1976 DNA was restricted using AflII/XhoI or AflII/BamHI, then half of the digested product was heated at 95 o C for 5 minutes. Products were separated on a 0.6 % agarose gel, blotted and probed for the D-loop region. The restriction product present immediately above the expected 6.4 kb restriction fragment (indicated by an asterisk) is eliminated by heat treatment. This species is believed to represent restricted mtDNA that still contains a stably-bound 7S DNA molecule, based upon its gel mobility and elimination by heat treatment.

Supplementary Figure 7 mtDNA rearrangements in a patient with pathogenic POLG mutations
Circos representation of detected mtDNA breakpoints in a patient compound heterozygous for pathogenic POLG mutations W748S and L752P. Arrowheads indicate the orientation of the deletions. The part of the genome that is deleted spans counter clockwise from the root to the head of the arrow (as indicated by dashed line in the panel "del" of the scheme on the right). A detailed list of the detected breakpoints is available in Suppl. Table 3. Breakpoints that conform to regular major-arc deletions are marked in blue, grey arcs indicate breakpoints that remove O L . wt, wild-type; del, deleted mtDNA molecules.
Supplementary Table 1 Half-lives of mtDNA and sub-genomic fragments during ddC treatment of

MGME1-deficient and control fibroblasts
Half-life times (in days) were determined by nonlinear regression analysis of qPCR data assuming simple exponential decay kinetics. 7S DNA, the 11-kb sub-genomic fragment, and complete mtDNA amounts were calculated as described in Methods. Two controls and two patient samples were investigated, each determined in three independent reactions. a,b , significant difference, p < 0.01 (t test). Breakpoints were determined by sequencing single-molecule PCR products that were amplified using the indicated primers. 'Start' and 'Stop' indicate the first and last deleted nucleotide positions, respectively. Count values are indicated if breakpoints were detected more than once. a , direct repeat lengths of at least 5 nucleotides are indicated; PrimerF, forward primer used for single-molecule PCR amplification; PrimerR, reverse primer.

Single-molecule PCR
To identify specific breakpoints in the mitochondrial genome, a singlemolecule PCR approach was used as described previously (2). Single mtDNA breakpoints were amplified in 42 cycles of PCR using TaKaRa LA Taq Hot Start polymerase and primers 3137F26 and 45R22, primers 10F and 16469R, or primers 15974F23 and 15623R20. Total template DNA was diluted to degrees, where only a part of multiple identical reactions resulted in amplification products (ideally less than 50 %). It is reasonable to assume that, under these conditions, each positive reaction originated from a single mtDNA molecule. Deletion breakpoints were mapped by reamplifying single-deletion amplicons using diverse primer pairs located within the amplified region and direct sequencing of re-amplified products.

Ligation-mediated PCR
To determine free ends of linear mtDNA molecules, we used a modified version of the ligation-mediated PCR method (3). An asymmetric one-side blunt-end doublestranded DNA adaptor was created by annealing two non-phosphorylated oligonucleotides of different sizes, as described by Kang et al. [23]. 100 pmole of this linker was ligated with 0.2 µg of sample DNA using T4 ligase (New England Biolabs) in a volume of 60 µl. Amplification of the ligated strand was done using an mtDNA-specific and an adaptor-specific primer, the latter at 1/8 concentration of the former. To map the exact ends of mtDNA fragments, we performed the amplification under single-molecule conditions. Single amplicons were then sequenced, and end points were defined as the last mtDNA position before the first nucleotide of the adaptor sequence (Suppl. Figure 2). Since not all linear molecules of interest were blunt-ended in vivo, we optionally pre-treated DNA samples to create ligatable blunt ends. To investigate 5' ends of the 7S DNA we pre-treated samples with T4 DNA polymerase (New England Biolabs) that filled in the 3' end of the opposite strand if it was shorter, and removed if it was a 3' overhang (Suppl. Figure 2A). 3' ends of the 7S DNA were investigated on DNA samples that were pre-treated with mung bean nuclease (New England Biolabs) that removes overhangs (Suppl. Figure 2B).
Depending on the region to be investigated, mitochondrial primers varied. min at 68°C; finally 10 min followed at 72°C. The extended elongation time enabled the polymerase to amplify fragments of several thousand base pairs in size. Such long molecules can be amplified in addition to the short fragments if more than one copy of the primer-binding region are present on some of the mtDNA molecules ( Figure   5B, main text). PCR products were analysed on an ethidium-bromide stained agarose gel.
of intensities for all peptides for the given protein. For the purpose of analysis, if a given protein was not detected in a sample then its intensity was arbitrarily set to 1.
MaxQuant software was used to match peptides between MS runs to deal with the possible problem of insufficient sequencing during LC-MS runs and for accurate ranking of the identified proteins according to MS signal intensity. Matching between runs is based on comparing peptide identification from any of the LC-MS runs in which the peptide has been identified by MS/MS to another LC-MS run where no MS/MS spectrum has been acquired for that MS peptide feature or the peptide could not be identified based on the MS/MS spectrum.

Protein purification and in vitro activity assays
Clonal HEK293T cells were induced to overexpress MGME1.Flag.Strep2 using 50 ng/ml doxycycline for 48 hours. Mitochondria were isolated from cells using a hypotonic lysis and differential centrifugation procedure as described previously (9), and MGME1 protein isolated from mitochondria using a Strep-tactin gravity flow column (IBA) as described previously (1). 1 pmol of labelled DNA substrate was incubated with incremental concentrations of MGME1 protein for 30 min at 37 o C in a reaction buffer containing 10 mM tris pH 7.6, 20 mM MgCl 2 , 1 mM DTT and 0.1 mg/ml BSA. Reactions were then frozen on dry ice and separated on 10 % polyacrylamide, 7 M urea gels, dried and imaged using a phosphorimager.