• mitochondrial DNA;
  • deletions;
  • duplications;
  • breakage hotspots;
  • mitochondrial disease;
  • tumors


  1. Top of page
  3. Introduction
  4. Data Source and Analysis
  5. Major Findings
  6. Discussion
  7. Conclusions
  8. Acknowledgment
  9. References
  10. Supporting Information

Mitochondrial DNA (mtDNA) rearrangements cause a wide variety of highly debilitating and often fatal disorders and have been implicated in aging and age-associated disease. Here, we present a meta-analytical study of mtDNA deletions (n = 730) and partial duplications (n = 37) using information from more than 300 studies published over the last 30 years. We show that both classes of mtDNA rearrangements are unequally distributed among disorders and their breakpoints have different genomic locations. We also demonstrate that 100% of cases with sporadic mtDNA deletions and 97.3% with duplications have no breakpoints in the 16,071 breakage hotspot site, in contrast with deletions from healthy and aged tissues. Notably, most deletions removing a section of the D-loop are found in tumors. Deleted mtDNA molecules lacking the origin of L-strand replication (OL) represent only 9.5% of all reported cases, whereas extra origins of replication occur in all duplications. As previously shown for deletions, imperfect stretches of homology are common in duplication breakpoints. Finally, we provide a dedicated Website with detailed information on deleted/duplicated mtDNA regions to facilitate the design of efficient methods for identification and screening of rearranged mitochondrial genomes (available at


  1. Top of page
  3. Introduction
  4. Data Source and Analysis
  5. Major Findings
  6. Discussion
  7. Conclusions
  8. Acknowledgment
  9. References
  10. Supporting Information

The human mitochondrial DNA (mtDNA) consists of a circular double-stranded DNA molecule of approximately 16,569 nucleotides (nt) encoding 13 essential proteins of the oxidative phosphorylation (OXPHOS) machinery. The occurrence of point mutations and large-scale rearrangements in the mtDNA is an important cause of mitochondrial disease, a clinically heterogeneous group of disorders related with OXPHOS dysfunction. Mitochondrial disorders typically affect postmitotic and energy-demanding tissues, such as nerve or muscle, can appear sporadically and currently have no cure [Park and Larsson, 2011; Schon et al., 2012; Tuppen et al., 2010]. In most cases, therapy is limited to relief of symptoms. Despite the ever-increasing number of reported cases, our understanding of the mechanisms involved in the formation and clonal expansion of mtDNA rearrangements remains limited.

Large-scale mtDNA rearrangements generally occur as partial deletions [Holt et al., 1988] or tandem direct duplications [Poulton et al., 1989a]. A circular deleted mtDNA (denoted simply as “deletion” from here on) is smaller than the normal 16,569-nt mtDNA by missing a section of the genome but remaining in a closed format. A circular partially duplicated mtDNA (denoted simply as “duplication” from here on) is larger than the normal mtDNA by having one section duplicated in tandem (Fig. 1). Deletions and duplications can coexist in the same patient; that is, the nondeleted section of a circular deleted mtDNA is found inserted into a full-length mtDNA to form the duplicated mtDNA, with both related molecules having the same breakpoints [Kajander et al., 2000; Manfredi et al., 1997; Poulton et al., 1989b]. The relationship between duplications and deletions is still not clear, but they could be originated in a common recombination event or one may be derived from the other [Holt et al., 1997; Manfredi et al., 1997; Poulton et al., 1993; Schon et al., 1997; Sembongi et al., 2007; Tang et al., 2000; Tengan and Moraes, 1998].


Figure 1. Schematic representation of a normal, partial duplicated and deleted mtDNA. A circular deleted mtDNA is smaller than the normal mtDNA by missing a section of the genome but remaining in a closed format. A circular partially duplicated mtDNA is larger than the normal mtDNA by having one section duplicated in tandem (light gray section). The two types can coexist in the same patient, with the nondeleted mtDNA region being the tandem duplicated region in partial duplications. The upstream region of the 5′ breakpoint (5fl) and the downstream region of the 3′ breakpoint (3fl) are found in all mtDNA types represented here (occurring twice in duplicated mtDNA), whereas the downstream region of the 5′ breakpoint (5del) and the upstream region of the 3′ breakpoint (3del) are only found in normal and duplicated mtDNAs, being removed (dark gray section) in mtDNA deletions.

Download figure to PowerPoint

Large-scale duplications were first described in patients with Kearns–Sayre syndrome (KSS), but they have also been found associated with other diseases and normal aging [Ballinger et al., 1994; Bodyak et al., 2001; Brockington et al., 1993; Dunbar et al., 1993; Fromenty et al., 1997; Manfredi et al., 1997; Muraki et al., 2001; Odoardi et al., 2003; Poulton et al., 1989a, 1989b; 1993]. The pathogenic nature of duplications remains uncertain and it has been suggested that they could act indirectly as a source of pathogenic mtDNA deletions, form aberrant translation products by the fusion of genes at the duplication breakpoint or cause an imbalance of tRNAs or proteins in mitochondria [Poulton et al., 1989b, 1993; Poulton, 1992; Schon et al., 1997]. Unlike mtDNA deletions, which are generally sporadic, duplications are frequently transmitted maternally [Ballinger et al., 1994; Dunbar et al., 1993].

The harmful effects of mtDNA deletions are unquestionable and probably occur due to the depletion of indispensable tRNAs required for proper translation in mitochondria [Nakase et al., 1990]. The clinical consequences of deletions are highly variable and depend on the ratio of deleted to intact mtDNAs in a cell (heteroplasmy), their tissue distribution and the vulnerability of each tissue to impaired oxidative metabolism (the threshold effect) [DiMauro and Hirano, 2003]. Some patients harbor a single form of deleted mtDNA (i.e., the same deletion length) in affected tissues, which typically arises spontaneously. These patients usually present one of three main syndromes: progressive external ophthalmoplegia (PEO) [Moraes et al., 1989; van Goethem et al., 2001], KSS [Kearns and Sayre, 1958; Zeviani et al., 1988], and Pearson syndrome (PS) [Pearson et al., 1979; Rotig et al., 1989]. Other patients present multiple mtDNA deletions (i.e., with different lengths) in affected tissues. These multiple deletions are typically caused by defects in nuclear genes related to mtDNA maintenance (e.g., POLG, SLC25A4) [Nishino et al., 1999; Spelbrink et al., 2001; van Goethem et al., 2001; Zeviani et al., 1989] and therefore follow a Mendelian pattern of inheritance. This category of disorders includes an inherited form of PEO known as autosomal dominant or recessive PEO (ad/arPEO) and mitochondrial neurogastrointestinal encephalomyopathy (MNGIE) [Nishino et al., 1999]. In addition, the accumulation of deletions might contribute to cell death in other diseases, such as Parkinson's disease [Bender et al., 2006; Ikebe et al., 1990; Kraytsberg et al., 2006] or inclusion body myositis [Bank et al., 2000; Moslemi et al., 1997]. Over the last years, mtDNA deletions have been found in healthy tissues at low levels, perhaps generated de novo on a continuous basis, with extensive variation between different tissues in the same individual [Chen et al., 1995; Cortopassi et al., 1992; Kajander et al., 2000; Linnane et al., 1990]. Their role in the aging of postmitotic tissues is still a matter of debate [Khrapko and Vijg, 2009; Kujoth et al., 2007; Park and Larsson, 2011].

Previous works have collected data from multiple sources to uncover important features of mtDNA rearrangements, such as the presence of homology [Guo et al., 2010; Lakshmanan et al., 2012; Mita et al., 1990; Samuels et al., 2004; Schon et al., 1989] and non-B DNA conformations at breakpoints [Damas et al., 2012; Hou and Wei, 1996; Oliveira et al., 2013; Pereira et al., 2008; Zeviani et al., 1989] and molecular differences or similarities among clinical groups [Guo et al., 2010; Kajander et al., 2000; Krishnan et al., 2008; Lopez-Gallardo et al., 2009; Poulton et al., 1994; Reeve et al., 2008; Sadikovic et al., 2010; Wanrooij et al., 2004]. Nevertheless, most of these studies were restricted to a few clinical backgrounds or to the analysis of a limited number of cases. Most importantly, deletions and duplications have never been analyzed together in a systematic manner. In order to overcome these limitations, we have performed a comprehensive survey of mtDNA rearrangements drawn from the published literature of the last 30 years. Overall, we identified unique properties of the two classes of rearranged mtDNA molecules, uncovered differences in breakpoint locations and provided useful information for the efficient identification of rearranged mitochondrial genomes.

Data Source and Analysis

  1. Top of page
  3. Introduction
  4. Data Source and Analysis
  5. Major Findings
  6. Discussion
  7. Conclusions
  8. Acknowledgment
  9. References
  10. Supporting Information

Data Collection

We collected data pertaining to human mtDNA rearrangements through a systematic review of the literature. In the case of deletions, 302 peer-reviewed publications from 1983 to 2011, one PhD thesis and the MITOMAP Web-based resource were analyzed [MITOMAP, 2011]. The information on duplications was obtained from 32 publications from 1988 to 2006 and from MITOMAP. The comprehensive collection of annotated rearrangements and the complete list of references are accessible at In total, we were able to recover sufficient clinical information for 730 different deletions and 37 different duplications. Each rearrangement was defined by a unique combination of two breakpoints (5′ and 3′), numbered according to the conventional light-strand (L-strand) positions of the revised Cambridge reference mtDNA sequence (rCRS; NC_012920.1). The 5′ breakpoints are upstream of the 5′ break and 3′ breakpoints are downstream of the 3′ break using the L-strand numbering. Each rearrangement was only considered once in most analyses, although several of them were reported in different studies. Nevertheless, the distribution of each rearrangement in the published literature can be obtained in the online database columns “Number of publications” and “References.” When a deletion and duplication was found in the same individual, they were included in both datasets.

We assigned syndromes, diseases and relevant clinical or nonpathological features to each mtDNA rearrangement according to the information provided in the associated peer-reviewed publication. The grouping into seven categories was based on clinical features and the abundance of reported rearrangements from the same group: single mtDNA deletions, multiple mtDNA deletions, healthy tissues, Parkinson's disease, inclusion body myositis, tumor, and “other clinical features” (Table 1). We also assigned mtDNA duplications to the groups named “single mtDNA deletions” and “multiple mtDNA deletions” when they were reported in patients with disorders typically defined by the presence of deletions (KSS, PEO, PS, etc.). The use of the same terms facilitates the comparisons between both classes of mtDNA rearrangements.

Table 1. Classification of mtDNA Rearrangements into Seven Groups
1Single mtDNA deletionsSingle cases of deletions found in PEO, KSS, or PS patients; usually sporadic
2Multiple mtDNA deletionsMultiple deletions detected in patients with defects in nuclear genes related to mtDNA; the related disease is inherited
3Healthy tissuesmtDNA rearrangements found in tissues with no signs of mitochondrial disease; including aged tissues, sublimons and control cases, among others
4Parkinson's diseasemtDNA rearrangements found in neurons from individuals with Parkinson disease
5Inclusion Body MyositismtDNA rearrangements found in patients with Inclusion Body Myositis and related disorders
6TumormtDNA rearrangements found in tumors
7Other clinical featuresAll mtDNA rearrangements that do not belong to any of the preceding categories

We have only included data from mtDNA duplications validated by Southern blot or long-range PCR. However, several studies reporting mtDNA deletions relied only on the PCR amplification of the rearrangement junctions. In such cases, it is possible that mtDNA duplications are being erroneously reported as deletions, considering that both rearrangements have a similar type of junction site detected by PCR. The use of restriction enzymes does not always distinguish deletions from duplications, such as when those with a recognition site within the duplicated segment (e.g., PvuII) are used [Poulton et al., 1993]. These enzymes cut out an mtDNA section with the length of the duplication that would run ahead of the full-length mtDNA band and suggest a deletion. A restriction enzyme (e.g., SnaBI) that cuts once within the nonduplicated segment (but not the deleted mtDNA) must be used to distinguish both classes of mtDNA rearrangements. Because such enzymes have been rarely used, duplications may have been overlooked. For instance, it has been shown that mtDNA duplications are common in KSS patients [Poulton et al., 1994], although several studies only searched for deletions in such patients. In order to allow readers to assess the possibility of likely missed duplications in the literature, we provide in the online database the information on the technique and restriction enzymes used to detect each mtDNA rearrangement.

Because we have only considered rearrangements with available sequence junctions, linear fragments of mtDNA would not be mislabeled as deletions or duplications since they lack a breakpoint junction site.

Measuring the Homology Length at Rearrangement Breakpoints

The breakpoints of each deletion and duplication are often located within or immediately adjacent to homologous sequences, usually in the form of perfect or imperfect direct repeats [Johns et al., 1989; Lakshmanan et al., 2012; Mita et al., 1990; Samuels et al., 2004; Schon et al., 1989]. In such cases, there are three possible outcomes of the rearrangement: the deletion or duplication junction may retain 0, 1, or 2 copies of the direct repeat. When a single copy of the homology is retained in the rearrangement junction, it is impossible to identify the exact sites of mtDNA breakage (i.e., 5′ and 3′ breakpoints). There is no way to know for sure which of the two homologous sequences are retained in the surrounding bases at the junction site (Supp. Fig. S1). As a consequence, the same rearrangement is sometimes described in the literature by arbitrary breakpoint positions inside the homologous region or by an interval of values (e.g., 8016–8019:15516–15519). In order to standardize the procedure, remove redundant data and facilitate the descriptive analyses, we have assigned the breakpoints to the 3′ nucleotide of the direct repeats (in the previous example, 8019:15519). In other words, breakpoints were always placed downstream (on the right of) the homology, according to the L-strand numbering (Supp. Fig. S1).

The deleted mtDNA molecule retains the upstream region of the 5′ breakpoint (5fl) and the downstream region of the 3′ breakpoint (3fl) next to each other, whereas the removed segment is delimitated by the downstream region of the 5′ breakpoint (5del) and the upstream region of the 3′ breakpoint (3del). 5del and 3del also denote the flanking region of the nonduplicated region in mtDNA duplications. In addition to these four regions, duplications also have a junction site connecting 5fl and 3fl (Fig. 1). In order to measure the extent of homology at breakpoints, we retrieved the length of the perfect repeat observed in all pairwise combinations of breakpoints’ flanking regions. For instance, the extent of homology between 5fl and 3fl in deletion 10370:15570 (TCTGGCCTAT ∥ GAGTGACTAC … TGATATTTCC ∥ TATTCGCCTA) was 3 nt.

In order to obtain an expected distribution of homology considering the mtDNA base content, we generated 20 sets of shuffled sequences with the same length and nucleotide composition of the rCRS mtDNA. We then measured the length of the direct repeats flanking 730 deletions and 37 duplications breakpoints located in the shuffled mtDNA sequences. The distribution of deletions and duplications from the shuffled mtDNAs (expected) was compared with the real (observed) distributions for each direct repeat length category (deletion and duplication in separate sets). For this purpose, we plotted the difference between the observed and expected frequencies (observed–expected). A positive value for the difference in a particular direct repeat length category indicates that real deletions or duplications have a higher frequency of cases within that length category than expected by chance. Conversely, a negative value for the difference indicates that real rearrangements have a lower frequency of cases with direct repeats of a certain length than expected.

We have also determined the extent of homology at breakpoints in the form of imperfect direct repeats by counting the number of paired bases (matches) between the aligned 15-nt flanking regions of 5′ and 3′ breakpoints (5fl vs. 3fl, 5fl vs. 3del, 5del vs. 3del, and 5del vs. 3fl). This procedure was used in real and 20 sets of shuffled mtDNA sequences as previously described. In order to consider the upstream and downstream regions of breakpoints simultaneously, we aligned and counted the number of matches between the aligned 30-nt sequences with 5′ and 3′ breaking sites as midpoints.

mtDNA Regulatory Sequences

Unless stated otherwise, we always refer to the origins of H-strand (OH) and L-strand (OL) replication using the strand-displacement model of mtDNA replication [Berk and Clayton, 1974; Robberson et al., 1972]. Although each replication origin is a range of nucleotides, we used here the nucleotide positions 407 for OH and 5747 for OL [Chang and Clayton, 1985; Hixson et al., 1986], which included the mtDNA regions encoding the RNA fragment used as primers to initiate DNA synthesis. Therefore, the minor arc (shortest mtDNA section between OL and OH) is located between positions 408 and 5,746 and the major arc (largest mtDNA section between OL and OH) is between positions 5,747 and 407. The D-loop or displacement loop refers to the triple-stranded DNA structure formed in the control region due to the premature arrest of H-strand synthesis near the control region 5′ end. The short H-strand (also known as 7S DNA) is hybridized to the parental L-strand, displacing the original H-strand. The trinucleotide stop point for the premature arrest of H-strand synthesis (D-loop 3′ end) is located between positions 16,104 and 16,106 [Doda et al., 1981]. We used here the largest possible D-loop region defined by the major OH site, extending from position 16,106 to 407 (OH). Nevertheless, shorter D-loop regions may occur due to the existence of other OH sites [e.g., Fish et al., 2004].

Statistical Analyses

The descriptive statistics, Student's t-test (independent samples with separate variance estimates), Pearson Chi-square test, and Fisher's exact test for contingency tables were obtained with the STATISTICA v7 software (StatSoft, Inc., Tulsa, OK). All reported P values are two-sided and a significance level of 0.05 was used.

Major Findings

  1. Top of page
  3. Introduction
  4. Data Source and Analysis
  5. Major Findings
  6. Discussion
  7. Conclusions
  8. Acknowledgment
  9. References
  10. Supporting Information

Distribution of mtDNA Rearrangements According to Associated Diseases

We started by determining the clinical distribution of mtDNA rearrangements using our comprehensive datasets. The rearrangements were classified into seven groups (Table 1). The percentage of rearrangements shared by two or more of these groups was 19.2% for deletions and 27.0% for duplications, suggesting that most rearrangements were exclusive to each group (Supp. Table S1). More than one third (36.6%) of all reported mtDNA deletions was found in patients with mitochondrial disease caused by the accumulation of a single (group 1 with 14.1%) or multiple (group 2 with 22.5%) deletions (Table 2). A total of 279 deletions (30.9% of the total) were observed in cases with no signs of mitochondrial disease, including aged tissues, sublimons, and control cases (group 3). The frequency of deletions was much lower in the remaining clinical groups (lower than 5.2% per group), indicating that deletions are less frequent in these clinical contexts or that the number of studies searching for them is much lower. Probably for the same reasons, mtDNA duplications have never been reported in Parkinson's disease, inclusion body myositis or tumors (Table 3). Nonetheless, the distribution of duplications is different from that of deletions, with most duplications (53.1%) reported in patients with single mtDNA deletions (group 1). In this group, all duplications were found in the context of KSS and PEO, with the exception of two studies where duplications were found in patients with PS [Muraki et al., 2001; Odoardi et al., 2003].

Table 2. General Features of 730 Unique Mitochondrial (mtDNA) Deletions Distributed by Seven Groups
     Location of 5′Location of 3′ 
   Length of deleted region (bp)Location of the deleted regionbreakpointsbreakpointsDeletion of replication origins
 Number of deletions per groupDeletions per group (%)MeanStandard deviationMin–MaxInside the minor arcInside the major arcRemoving part of minor and major arcsMinor arcMajor arcMinor arcMajor arcOLOHOL and OHNone
  1. a

    Total number of unique mtDNA deletions in the entire dataset.

  2. The deletions found in different groups are considered more than once (thus the total number of 902). A similar table describing only exclusive rearrangements (i.e., those only found in one group) is presented in the supplementary material. The number of deletions in each group is indicated according to their location in the minor and major arcs and the removal of replication origins (OL and OH). The mean length of the deleted region is also shown for each group.

1. Single mtDNA deletions12714.15,3921,8989–9,7682125021252125000127
2. Multiple mtDNA deletions20322.57,4631,590490–10,2731200222012201110201
3. Healthy tissues27930.97,4713,2153–15,6286212616721262736100218
4. Parkinson disease252.85,8822,1284–9,127025002502500025
5. Inclusion body myositis475.28,0881,7334,067–12,760042554204750042
6. Tumor323.54,5072,87213–8,162329032932900032
7. Other clinical features18921.06,1172,1934–10,9872185241852187200187
Total902 6,7762,5953–15,628148187083819158876910832
Total (nonredundant)a730 6,8932,6673–15,628106507079651117196910660
Table 3. General Features of 37 Unique Mitochondrial (mtDNA) Duplications Distributed by Seven Groups
   Length of duplicated region (bp) Location of the duplicated regionLocation of 5′ breakpointsLocation of 5′ breakpointsDuplication of replication origins
 Number of duplications per groupDuplications per group (%)MeanStandard deviationMin–MaxLength of interconvertible deletion (bp)Inside the minor arcInside the major arcDuplicating part of minor and major arcsMinor arcMajor arcMinor arcMajor arcOLOHOL and OHNone
  1. a

    Total number of unique mtDNA duplications in the entire dataset.

  2. The duplications found in different groups are considered more than once (thus the total number of 50). A similar table describing only exclusive rearrangements (i.e., those only found in one group) is presented in the supplementary material. The number of duplications in each group is indicated according to their location in the minor and major arcs and the duplication of replication origins (OL and OH). The mean lengths of the interconvertible deletions and of the duplicated regions are shown for each group.

1. Single mtDNA deletions2653.17,9594,249156–13,4918,610002652102605210
2. Multiple mtDNA deletions36.12,8234,493193–8,01113,74600321030210
3. Healthy tissues816.3568819156–2,55816,00100880080800
4. Parkinson disease
5. Inclusion body myositis
6. Tumor
7. Other clinical features1326.05,5593,335204–10,07411,0100013760130760
Total50 5,8444,512156–13,49110,72500502228050022280
Total (nonredundant)a37 6,8774,238156–13,4919,69200371324037013240

The distribution of breakpoints in the two families of rearranged molecules is plotted in Figure 2A and the length of deleted and duplicated regions in Figure 2B. When grouped in 100-nt bins, the frequency distribution of 5′ breakpoints was more scattered in duplications than in deletions. While there was a hotspot of 5′ deletion breakpoint near the OL (mtDNA window from position 5,701 to 5,800), duplications breakpoints were dispersed through that region. Additionally, 5′ deletion breakpoints were found across the COX2 gene, whereas duplications had a major peak at window 8,401 to 8,500 (ATP8 gene). Although more similar in both types of rearrangements, the distributions of 3′ breakpoints also presented differences, such as the absence in duplications of a hotspot around the mtDNA position 16,071 (Fig. 2A). There were only two out of 37 duplications with a 3′ breakpoint in the mtDNA window 16,001 to 16,100, significantly lower than the observed in deletions with 200 out of 730 cases (P value < 1 × 10−4; Pearson Chi-square test). In addition, seven short duplications (156 to 267 nt) had both breakpoints between positions 301 and 573 in the control region section near the tRNA-Phe (TRNF) gene (Supp. Table S2), where deletion breakpoints were rare.


Figure 2. Breakpoints of mtDNA deletions and duplications. A: Distribution of 5′ and 3′ deletion (blue and red bars) and duplication (yellow and green bars) breakpoints in the human mtDNA. The values were grouped in 100-nt windows and the locations of the mitochondrial genes are shown below the x-axis. B: Length distribution of the deleted region in 730 mtDNA deletions (gray bars) and the duplicated (black bars) and nonduplicated (orange bars) regions in 37 mtDNA duplications. The cases were grouped in 100-nt windows.

Download figure to PowerPoint

Although the missing region in deletions roughly corresponds to the nonduplicated section of duplications (black/white gradients of Fig. 3), the mean length of the deleted region was lower (6,893 nt, standard deviation of 2,667) than the mean length of the nonduplicated region (9,692 nt, with a standard deviation of 4,238 nt), as illustrated in Figure 2B. Moreover, only seven deletions and duplications had equal breakpoints. With the exception of a single case (2,059:16,071) that was only found in aged tissues [Bodyak et al.,

2001], all other duplication–deletion pairs were reported in at least one diseased individual, usually with KSS (Supp. Table S3).


Figure 3. Circular plot of the mitochondrial genome specifying the location of breakpoints and deleted or duplicated sites. The outer track depicts the mitochondrial genome with annotated tRNA (black), rRNA (brown) and protein-coding (green) genes. The control region is highlighted in gray. The black/white gradient indicates the percentage of mtDNA sites deleted in the 730 deletions (second outer track) or duplicated in the 37 duplications (intermediate track between deletions and duplications). The white/red gradient indicates the percentage of breakpoints (5′ + 3′) in each group (numbered according to the legend), measured in 100-nt windows. The shaded areas highlight the most conserved (light blue and light purple) and most deleted/duplicated regions (dark blue and purple) in our datasets. The major origins of replication (OH and OL) and the 16071 site are also indicated. The plot was produced using the Circos software, version 0.62 (

Download figure to PowerPoint

Different Diseases Reveal Distinct Patterns of Breakpoint Distributions in the Control Region

The strongest deletion breakpoint hotspot occurred inside the control region at position 16,071, where 41 from the 730 3′ breakpoints (5.62%) were observed (Figs. 2, 3, and 4A and B; Supp. Fig. S2). Its surrounding region (16,001 to 16,100) harbors 27.4% of all 3′ deletion breakpoints (n = 200), supporting previous observations [Damas et al., 2012; Guo et al., 2010; Krishnan et al., 2008; Samuels et al., 2004; Zeviani et al., 1989]. Only a single duplication (2,059:16,071) had a 3′ breakpoint at position 16,071, which was precisely the only deletion–duplication pair observed in aged tissues in both databases (Supp. Table S3) [Bodyak et al., 2001]. Despite being a deletion hotspot, we found marked differences in the distribution of breakpoints within the 16,001–16,100 region among the different groups (Fig. 4B and Supp. Fig. S3). This region was almost devoid of breakpoints in PEO, KSS, and PS (group 1), with only one case (8,648:16,085) observed in PS patients [Kleinle et al., 1997; Rotig et al., 1995], that is, 0.79% of all deletions (n = 127) from this group. Nevertheless, the deletion detected in PS had a 3′ breakpoint near but not on the 16,071 site (16,085) and was observed in all other groups, with the exception of Parkinson's disease. In contrast, the 16,071 surrounding region was a breakpoint hotspot in multiple mtDNA deletions (50.2%) and inclusion body myositis (63.8%), as shown in Figure 4B. It was also relatively common in healthy tissues (26.5%), as previously reported [Bodyak et al., 2001; Kajander et al., 2000; Khrapko et al., 1999]. Overall, there were significant differences in the proportion of 3′ breakpoints inside and outside the 16,001 to 16,100 region among the different groups (Fig. 4B). These observations suggest that the D-loop 3′ end is an obstacle to the distribution of mtDNA deletion breakpoints (Figs. 3 and 4A). Of the 730 deletions, only 17 (2.33%) had a 3′ breakpoint between position 16,086 and the OH. There is a significant difference (P value < 1 × 10−4; Pearson Chi-square test) between the number of sites with 3′ breakpoints in the 891-nt region between 16,086 and the OH (n = 17) and an upstream segment with the same length (from positions 15,185 to 16,085; n = 145). Only 18 mtDNA deletions had a breakpoint inside the D-loop region (14 with a 3′ breakpoint, 1 with a 5′ breakpoint, and 3 with both). Notably, there was a high prevalence of mtDNA deletions from tumors (group 6) inside the D-loop (7 out of 18), particularly from radiation-associated thyroid tumors [Rogounovitch et al., 2002]. In fact, the D-loop was partially removed in 21.9% of all deletions from tumors, a value that is significantly higher than in any of the remaining groups, with a P value < 0.05; Fisher's exact test (Fig. 4C). All results remain statistically significant when considering shorter D-loop regions, that is, when considering other OH sites (data not shown).


Figure 4. The spectrum of deletions in the mtDNA control region. A: Location of mtDNA deletions (black lines) within the control region. The major origin of H-strand replication (OH) and the D-loop are shown. B: Percentage of mtDNA deletions in each group with 3′ breakpoints between mtDNA positions 16,001 to 16,100. C: Percentage of mtDNA deletions in each group removing part of the D-loop region. The significance of the difference between groups was obtained with a Fisher's exact test.

Download figure to PowerPoint

The Presence and Absence of Replication Origins in Rearranged mtDNA Molecules

The only case of a deleted mtDNA (16,266:576) without OH was found in the skeletal muscle of a patient with MNGIE [Nishigaki et al., 2004]. In fact, most sites in the control region (including all sites for OH) were only found to be missing once out of 730 deletions (Fig. 4A and Table 2). Although most deletions (90.4%) in our dataset do not remove any origin of replication, 69 out of 730 deletions (9.45%) removed the OL (Table 2 and Supp. Fig. S4). Sixty-one out of 69 deletions lacking OL were observed in healthy or aged tissues (representing 21.9% of all deletions observed in this group), whereas no case was observed in PEO, KSS, or PS patients (group 1), Parkinson's disease patients (group 4) and tumors (group 6) (Fig. 5). Only one deletion lacking OL (4,469:13,923) was found in patients with multiple mtDNA deletions (group 2), associated with MNGIE [Nishigaki et al., 2004]. We also noticed that deleted mtDNAs without OL had a higher proportion of 3′ breakpoints in the 16,071 hotspot region in healthy tissues (38 out of 61) than in pathological situations (one out of eight), with a P value = 0.018 (Fisher's exact test).


Figure 5. Percentage of mtDNA deletions in each group lacking the origin of L-strand replication (OL). The significance of the difference between groups was obtained with a Fisher's exact test.

Download figure to PowerPoint

All mtDNA duplications observed thus far duplicate at least one of the replication origins (Table 3 and Supp. Fig. S4). Twenty-four of them duplicate both replication origins, 13 only the OH, while none duplicate only the OL. Although the small sample size in each group prevents significant comparisons, healthy tissues (group 3) stood out by lacking duplications of both replication origins, which were found in all other groups (Table 3). All exclusive duplications (n = 17) found in patients with KSS, PEO, or PS (group 1) duplicated OH and OL, in agreement with the large size of their duplicated region (mean of 10,033 nt, with a standard deviation of 2,225). Since origins of replication are always duplicated, there was no case of a duplication located exclusively inside the minor or major arcs (Table 3 and Supp. Fig. S4).

The breakpoints that define the 730 unique mtDNA deletions are highly concentrated (1,370 out of 1,460; 93.8%) in the major arc of the mtDNA (Fig. 3, Table 2, and Supp. Fig. S4), as previously recognized [Mita et al., 1990; Kajander et al., 2000; Samuels et al., 2004; Schon et al., 1989]. As a consequence, the deleted region of the mtDNA is often located inside the major arc (89.0% of the cases), whereas the minor arc only contains 10 deletions (1.4%), a significant difference with a P value < 1 × 10−4 (Pearson Chi-square test). The distribution of deletion lengths shows that there is a constraint at approximately 9,000 nt, with only 13.4% (98 deletions) of all deletions going beyond this value (Fig. 2B). The preferential location of the deletions inside the major arc explains the observed length constraint: the length of the major arc (11,230 nt) creates an upper limit for the size of the deletions. The central region of the major arc is often deleted, with positions 7,946 to 15,158 missing in more than 50% of the cases and positions 10,936 to 12,893, including tRNA-His (TRNH) and tRNA-Ser (TRNS2) genes, missing in 90% of the cases (Fig. 3 and Supp. Fig. S5).

Nonrandom Distribution of Homology at mtDNA Rearrangement Breakpoints

We started by determining the length of perfect direct repeats adjacent to breakpoints. There were 29 deletions (3.97% of the total) with a perfect repeat equal or larger than 10 nt (Supp. Table S4). The longer stretch of perfect homology had 13 nt and was found in cases 547:4,443 (CCCATACCCCGAA) and 8,482:13,460 (ACCTCCCTCACCA), the latter being known as the “common” deletion and the only case found in all seven groups (Supp. Table S5). We also noted that the only deletions observed in more than four different groups had direct repeats larger than 12 nt (Supp. Table S5). Overall, the mean length of the largest direct repeat observed in deletions present in three or more groups was significantly higher than in deletions found in two or less groups (6.6 nt vs. 2.9 nt, respectively, P value = 0.001, Student's t-test). There were five duplications with direct repeats equal or larger than 10 nt (13.51% of the total), being the largest direct repeat (13 nt) observed in the duplication that is related to the common deletion (8,482:13,460).

In order to test whether the observed homology could be produced by chance alone, we derived an expected distribution of direct repeat lengths using shuffled sequences of mtDNA (Fig. 6A). In all pairwise comparisons of flanking regions, the number of rearrangements with no homology was higher in random (expected distribution) than in the real cases (observed distribution). This result is represented as a negative difference (y-value) in the plots of Figure 6A. On the contrary, there was an enrichment of homology (positive values in the plots of Fig. 6A) in the upstream regions of both breakpoints (5fl vs. 3del) and, to a lesser extent, in other pairwise comparisons (5fl vs. 3fl and 5del vs. 3del). No perfect stretch of homology was recorded between the downstream regions of both breakpoints (5del vs. 3fl) due to the assignment of breakpoints to the most 3′ nucleotide of the direct repeats (Fig. 6A and Supp. Fig. S1). Overall, 13.2% of all mtDNA deletions and 16.2% of all duplications had no direct repeat at breakpoints and 70.3% of deletions and 67.6% of duplications had only direct repeats < 4 nt (Supp. Tables S4 and S6).


Figure 6. The presence of perfect and imperfect direct repeats at mtDNA deletion and duplication breakpoints. A: The histograms depict the difference between the distribution of rearrangements in real (observed) datasets and in shuffled mtDNA sequences (expected) according to the length of perfect direct repeats flanking breakpoints (deletions as black bars and duplications as white bars). When in the presence of homology, both breakpoints were placed downstream (on the right) of the direct repeats. For this reason, there are no perfect direct repeats between 5del and 3fl and no differences to show in a graph. B: The line charts depict the difference between the distribution of rearrangements in real and shuffled datasets according to the number of matches (base pairs of homology) in the alignment of 15-nt windows adjacent to breakpoints. This measures the relevance of imperfect repeats at breakpoints. C: Similar to (B) but with 30-nt windows centered at breakpoints. A positive value in the difference between observed and expected values indicates that real deletions or duplications have a higher frequency of cases within that length category than expected by chance. A negative value for the difference indicates that real rearrangements have a lower frequency of cases with a certain length or number of matches than expected by chance.

Download figure to PowerPoint

A similar result was obtained when imperfect stretches of homology around breakpoints were considered, with a higher than expected homology (number of equal aligned bases) in the 15-nt flanking regions of the reported rearrangements (Fig. 6B). In parallel with what was observed in 5fl versus 3del comparisons, the downstream regions of both breakpoints (5del vs. 3fl) also presented a higher homology than expected by chance (red line in Fig. 6B), which was not detected in the form of perfect direct repeats. The presence of homology was also detected when both flanking regions of each breakpoint were considered at the same time, measured in 30-nt windows centered at 5′ and 3′ breakpoints (Fig. 6C).

An Online Catalogue of Under- and Overrepresented mtDNA Regions in Rearrangements

In order to help researchers designing accurate methods for identification and screening of abnormal mtDNAs in different clinical contexts, we provide an online catalogue of breakpoints, junction sequences and deleted/duplicated sites from our collection of rearrangements (available at The Website includes the list of 500-nt mtDNA regions that alone or in combination encompass a higher number of 5′ and 3′ breakpoints and junction sites (a summary is presented in Supp. Fig. S6). All rearrangement junctions with breakpoints in two mtDNA regions of 500 nt can be detected using primers flanking both windows, yielding a PCR product with a maximum length of 1,000 nt (depending on the location of breakpoints inside windows) plus the distance between windows and primers.

Several screening methods for detection of mtDNA rearrangements use components that target specific regions of the genome believed to be underrepresented (deleted), overrepresented (duplicated), not altered by the rearrangement (used as control), or de novo generated by the rearrangement (e.g., junction site). Depending on the method, such regions can be used for binding of PCR primers, hybridization of probes, serve as restriction sites or be inspected for variations in the number of sequence reads (coverage) in next-generation sequencing of mtDNA [Calvo et al., 2012; DiMauro and Hirano, 2003; Krishnan et al., 2010; Tang et al., 2013]. Using our databases, we were able to find that the region more often removed in the deleted mtDNA molecules spans positions 12,204 to 12,214, being absent in 92.6% of reported cases (Fig. 3). This region is part of a large cluster of three regions in ND4, TRNH, TRNS2, and TRNL2 genes (positions 11,241 to 11,311, 11,369 to 11,422, and 12,114 to 12,282), which were found deleted in more than 92.3% of the cases. The mtDNA segment between positions 8,528 (ATP8/ATP6) and 13,950 (ND5) are absent in more than 70.0% of reported cases. At the opposite extreme, the 16,280–708 region encompassing part of the control region, TRNF and part of the RNR1 gene, was conserved in 98.8% of deleted mtDNAs (Fig. 3). The minor arc was conserved in 90.5% of the deleted mtDNAs. In duplications, the region between positions 301 and 965 was duplicated in more than 81.1% of the cases, including the short tandem duplications in the 3′ half section of the control region (Supp. Table S2). On the contrary, the region 12,206–13,451 was found duplicated only once (Fig. 3). These genomic regions are the ideal targets for a comprehensive detection of deleted or duplicated mtDNA segments.


  1. Top of page
  3. Introduction
  4. Data Source and Analysis
  5. Major Findings
  6. Discussion
  7. Conclusions
  8. Acknowledgment
  9. References
  10. Supporting Information

mtDNA Breakage Hotspots

Our analyses indicate that several breakage hotspots exist throughout the mitochondrial genome, either in mtDNA deletions or duplications (Figs. 2A and 3, Supp. Figs. S2 and S5). The breakpoints are often concentrated in narrow regions of the mitochondrial genome, such as the 16,071 site for deletion 3′ breakpoints or the WANCY cluster of tRNA genes for deletion and duplication 5′ breakpoints. Our data are therefore compatible with the existence of short stretches of DNA prone to breakage, known as “fragile sites” [Cohen et al., 1996; Pevzner and Tesler, 2003]. The fragile breakage model postulates that specific genomic sequences or structures increase the propensity of a DNA region to either produce or misrepair a break, leading to the formation of a rearrangement. In fact, mtDNA breakage hotspots often colocalize with specific characteristics of the local DNA sequence environment, such as direct repeats [Johns et al., 1989; Lakshmanan et al., 2012; Mita et al., 1990; Samuels et al., 2004; Schon et al., 1989], non-B DNA conformations [Damas et al., 2012; Hou and Wei, 1996; Oliveira et al., 2013; Pereira et al., 2008; Zeviani et al., 1989], or long imperfect homologous regions [Guo et al., 2010] that make them more predisposed to instability. In addition, our data also confirm that deletion breakpoints are located near replication origins (Figs. 2A and 3, Supp. Figs. S2 and S5), a correlation already noticed in other organisms (Di Rienzi et al., 2009]. Together with the observation that mutations of the replicative helicase Twinkle or the DNA polymerase γ (POLG) genes cause multiple mtDNA deletions (e.g., ad/arPEO), these results suggest that the distribution of deletions is influenced by the replication process [Spelbrink et al., 2001; van Goethem et al., 2001; Wanrooij et al., 2004]. Nevertheless, a multistep model including repair [Krishnan et al., 2008; Srivastava and Moraes, 2005] or recombination [Johns et al., 1989; Mita et al., 1990; Poulton et al., 1993] mechanisms have some credibility at present.

Despite the existence of breakage hotspots, the distribution of breakpoints from deletions and duplications had some noticeable differences (Figs. 2A and 3). For instance, several duplication breakpoints were found inside the control region in contrast with the observed in deletions (Supp. Table S2). Moreover, a major peak in the distribution of 5′ breakpoints at the ATP8 gene is characteristic of duplications, whereas the 5′ breakpoints of deletions are found dispersed across the COX2, ATP8, and surrounding regions (Fig. 2A). The irregular distribution of breakpoints in duplications may simply reflect the small sample size and additional cases may fulfill the distribution around COX2, as observed in deletions.

The Breakage Hotspot/Coldspot Dual Nature of Position 16,071

Another remarkable difference between the two families of rearranged molecules was the absence in duplications of a major breakage hotspot at position 16,071 (Figs. 2A and 3, Supp. Fig. S2), which is a common feature of deletions from healthy tissues and some disorders [Kajander et al., 2000; Moslemi et al., 1997; Zeviani et al., 1989]. The absence of duplications with breakpoints at 16,071 was even more striking given that they would correspond to small duplicated mtDNAs molecules, which are possibly more stable in mitochondria and are expected to be easier to detect in the laboratory. However, this result may simply reflect the fact that 53.1% of reported duplications were found associated with KSS, PEO, PS (group 1), whereas only three cases (6.1%) were found associated with multiple mtDNA deletion syndromes (group 2), as indicated in Table 3. Indeed, the 16,071 region is a coldspot of 3′ breakpoint in deletions from patients with KSS, PEO, PS (group 1) and Parkinson's disease (group 4) (Figs. 3 and 4B), whereas it behaves as a true hotspot in patients with multiple mtDNA deletions (group 2), as previously recognized [Zeviani et al., 1989]. The observed similarity of both types of rearranged molecules in avoiding the 16,071 breakage site might suggest a common origin in KSS, PEO, and PS patients.

Conversely, the marked difference between single and multiple deletions at the 16 071 breakage site is a clear sign that different mechanisms are acting in the formation or proliferation of rearranged mtDNAs. The reason why DNA breakage events, or any other process acting near the D-loop 3′ end, occur so often in certain situations but are rare in others remains unknown. The data suggest that either different mechanisms are causing mtDNA deletions in different syndromes and disorders or that differently deleted mtDNAs are positively or negatively selected in different situations, as previously suggested [Guo et al., 2010; Kajander et al., 2000; Lopez-Gallardo et al., 2009; Sadikovic et al., 2010; Wanrooij et al., 2004]. The different induction of break/recombination events in the 16,071 region might be related to the functions of the nearby D-loop structure and its regulation [Antes et al., 2010; He et al., 2007] or be influenced by local deformations to the canonical B-form of DNA in response to different trans-acting elements [Bacolla et al., 2004; Cooper et al., 2011; Damas et al., 2012; Pereira et al., 2008; Zeviani et al., 1989]. The differential expression of nuclear genes regulating mtDNA in tissues or individuals in response to distinct physiological demands might also explain this puzzling pattern.

Removal of mtDNA Replication Origins

Previous studies have noted that the removal of replication origins has a major effect in the spectrum of observed mtDNA deletions [Holt et al., 1997; Tang et al., 2000]. We found that deleted mtDNA molecules lacking an origin of replication represented only 9.6% of reported cases (Tables 2 and 3), which might be explained by the fact that circular deleted mtDNAs without OL or OH are unable to replicate or propagate efficiently in the cell, making difficult their detection using routine laboratory techniques. In fact, these rare deleted mtDNA molecules are usually recovered only by PCR in postmitotic tissues, such as muscle [Kajander et al., 2000; Khrapko et al., 1999; Moslemi et al., 1997; Nishigaki et al., 2004]. Nonetheless, the number of deleted species of mtDNA without OH (n = 1; 0.1%) and OL (n = 69; 9.5%) was significantly different (P value < 1.00 × 10−4; Pearson Chi-square test), as indicated in Table 2 and Supp. Figure S4. This result suggests that the removal of OL is less critical than the removal of OH for the proliferation and detection of mtDNA deletions, which might be an indirect indicator that alternate lagging strand initiation sites exist in mtDNA [Bowmaker et al., 2003; Brown et al., 2005].

How exactly deleted mtDNA molecules lacking OL replicate (if they really replicate) remains unknown. It is conceivable that mtDNA duplications may be responsible for clonal propagation of deletions lacking OL [Bodyak et al., 2001; Poulton et al., 1993]. The duplicated forms appear to be transient in cells under normal conditions; this is likely because they are constantly being resolved into a full-length (wild-type) molecule and a deleted mtDNA. In this manner, partially duplicated mtDNAs might provide a continuous source of deletions, which could slowly accumulate in postmitotic tissues without a need for replication. However, only a single deletion (2,059:16,071) without the OL had breakpoints equal to a reported duplication (Supp. Table S3), which was the only duplication-deletion pair in our database that was exclusively found in aged tissues [Bodyak et al., 2001]. Further studies are necessary to ascertain the role of duplicated mtDNA molecules in the propagation of deleted mtDNAs lacking the OL in pathological conditions.

A significant difference among the seven defined groups is the rare presence of deleted mtDNAs without OL in cases of mitochondrial pathology (groups 1, 2, 4, and 6) and their abundance in healthy or aged tissues (group 3), as shown in Figures 3 and 5, Table 2, and Supp. Table S1. Although the underlying cause of this pattern is not clear, the striking difference may reside in the differential formation, proliferation, or detection of deleted mtDNAs without OL in pathological and nonpathological cases. These different hypotheses are nonmutually exclusive and may act concomitantly to cause the observed result. First, OL-deficient mtDNAs might be generated preferentially in certain healthy or aged tissues by an unknown mechanism that somehow removes the OL region. In this respect, it is intriguing that a high proportion of 3′ breakpoints from deletions lacking OL are clustered around position 16,071 [Bodyak et al., 2001; Kajander et al., 2000; Khrapko et al., 1999], raising the possibility that the two factors might be somehow related. Second, the putative less efficient replication capacity of OL-deficient mtDNAs might also contribute to the observed difference. If pathological deletions result from a preexisting pool of rearranged mtDNAs present at very low levels [Kajander et al., 2000], from which some are amplified in abnormal situations, one might expect that those with a relative replicative advantage (two replication origins) would be more often found associated with disease. Third, the methods (such as PCR) used to detect deletions in healthy and aged tissues may be more likely to find deletions that exist in smaller quantities (i.e., sublimons), including those with a possible limited replicative capacity such as those without OL. In diseased tissues, mtDNA deletions are usually found at high frequencies for having exceeded the threshold level necessary to cause the mitochondrial dysfunction [Park and Larsson, 2011; Schon et al., 2012; Tuppen et al., 2010].

The Spectrum of Breakpoints in the mtDNA Control Region

The largest noncoding region of the human mtDNA, known as control region or D-loop, contains important regulatory elements for mtDNA replication and expression. The maintenance of conserved segments is believed to result from selective constraints acting on putative regulatory domains, such as those known as conserved sequences blocks proposed to control the H-strand replication initiation [Sbisa et al., 1997; Walberg and Clayton, 1981]. It is therefore expected that deletions and duplications with breakpoints on the control region are under strong negative selection since any sequence alteration may possibly disrupt the binding sites of nuclear-encoded proteins that regulate mtDNA. In fact, the rare presence of deletion breakpoints beyond the 16,071 hotspot region approaching the classical OH site is clearly in agreement with this hypothesis (Figs. 2, 3, and 4A). The 16,071 position is 35 to 37 nt away from the 3′ ends (16,104 to 16,106) of the 7S DNA. Indeed, there are only 18 deletions (2.47%) with one or both breakpoints inside the D-loop region (Fig. 4A). The 7S DNA is presumed to be a stalled or aborted replication intermediate with a 5′ end at OH [Berk and Clayton, 1974; Robberson et al., 1972]. This DNA has putative regulatory roles, either by serving as a primer for the synthesis of full-length mtDNA [Eichler et al., 1977] or by influencing the mitochondrial nucleotide pool via controlled 7S DNA synthesis/degradation cycles [Antes et al., 2010]. Other works suggest that the D-loop plays an important role as a protein recruitment center for nucleoid organization [He et al., 2007]. Whatever the case, the low frequency of breakpoints in the D-loop region might be explained by some type of protection offered by the triplex D-loop conformation and associated proteins that prevent DNA breakage. It is also possible that the reduced replicative capacity of deleted mtDNA molecules that lack parts of the functionally relevant D-loop region could prevent their proliferation in tissues and subsequent detection. This hypothesis is also compatible with the idea that the control region section near the tRNA-Pro (TRNP) gene harbors a cluster of potential start sites for mtDNA replication (named cluster II or Ori-b), according to a strand-coupled bidirectional replication model [Yasukawa et al., 2005]. Removal of these sites may hamper the proper replication of mtDNA. The observation that most deletions removing a section of the D-loop are found in radiation-associated thyroid tumors [Rogounovitch et al., 2002] has no parallel in other diseases and requires further investigation (Fig. 4C).

A large duplication encompassing the entire control region is expected to have less negative effects for mtDNA biogenesis than a short duplication with breakpoints disrupting a regulatory element. We detected seven short duplications of length 156 to 267 nt in the upstream region of the major OH site (Supp. Table S2), an important region for the regulation of the initiation of mtDNA replication [Bouzidi et al., 1998; Brockington et al., 1993; Tengan et al., 2002; Torroni et al., 1994; Wei et al., 1996]. It has been suggested that the tandem duplication of this region may contribute to the formation of pathological deletions [Brockington et al., 1993]. Nevertheless, six out of seven of these duplications were detected in healthy subjects (group 3), the same number observed in diseased individuals (Supp. Table S2), confirming that they are not always associated with large-scale deletions [Torroni et al., 1994; Wei et al., 1996] and that they are not the cause of disease [Hao et al., 1997]. The pathogenic nature of duplications disrupting the control region is also questioned by the fact that all duplications in healthy subjects (group 3) had at least one breakpoint inside this regulatory region.

Homology at Breakpoints

It is widely believed that significant regions of homology facilitate genomic rearrangements [Cooper et al., 2011; Zhang et al., 2009]. The only deletion found in all groups (Supp. Table S5) was the so-called “common deletion” (8,482:13,460), which also presented the longest stretch of perfect homology (13 nt) around breakpoints [Schon et al., 1989]. Although this deletion may be overrepresented in the published literature by being searched more often than other deletions [Krishnan et al., 2010], for instance by the use of a deletion-specific three primer PCR approach, we found that deletions detected in more clinical contexts also had the largest perfect direct repeats at breakpoints (Supp. Table S5). Our data are therefore compatible with the idea that large repeats correlate with mtDNA deletion breakpoints in many diseases, as well as in normal ageing [Holt et al., 1988; Lakshmanan et al., 2012; Mita et al., 1990; Nardi et al., 2012; Sadikovic et al., 2010; Samuels et al., 2004; Schon et al., 1989]. It is however possible that these perfect repeats are part of large regions of partial or interrupted homology, which are in turn responsible for bringing together distant segments of the mtDNA during the deletion formation [Guo et al., 2010]. Indeed, less than one third of deletions and duplications had perfect direct repeats larger than 4 nt flanking the breakpoints (Supp. Table S6). The higher than expected number of equal bases in the alignment of 5′ and 3′ breakpoint flanking regions (Fig. 6C), which takes into account mismatches interrupting perfect repeats, supports the claim that long imperfect repeated sequences correlate with deletion breakpoints [Guo et al., 2010]. The presence of homology at the breakpoints of duplications suggests that it may also contribute to the formation of this type of rearrangement. Because segments of overlapping homology can mediate the formation of a junction site either during DNA replication, recombination or repair, it is conceivable that different mechanisms might be associated with the formation of mtDNA deletions and duplications.


  1. Top of page
  3. Introduction
  4. Data Source and Analysis
  5. Major Findings
  6. Discussion
  7. Conclusions
  8. Acknowledgment
  9. References
  10. Supporting Information

We have characterized circular deleted and partially duplicated mtDNA molecules in diverse clinical contexts through an extensive review of the literature from the last 30 years. Our meta-analytical study reveals new features and confirms previous observations on the nature of these important classes of mtDNA abnormalities. Although several proprieties are common to deletion and duplication breakpoints, our study highlights important differences that require the attention of future investigations. The differences detected among clinical groups are a prerequisite to better understand the mutational mechanisms underlying these pathological lesions, which are still not fully comprehended. We also provide a detailed description of the mtDNA regions that should be targeted for an efficient detection of rearrangements. Although it is currently impossible to directly observe the sequence of events involved in the formation of mtDNA rearrangements, our systematic survey of reported cases in different scenarios can guide future investigations into their pathophysiological significance, which is a prerequisite for the development of effective therapeutic strategies.


  1. Top of page
  3. Introduction
  4. Data Source and Analysis
  5. Major Findings
  6. Discussion
  7. Conclusions
  8. Acknowledgment
  9. References
  10. Supporting Information
  • Antes A, Tappin I, Chung S, Lim R, Lu B, Parrott AM, Hill HZ, Suzuki CK, Lee CG. 2010. Differential regulation of full-length genome and a single-stranded 7S DNA along the cell cycle in human mitochondria. Nucleic Acids Res 38:64666476.
  • Bacolla A, Jaworski A, Larson JE, Jakupciak JP, Chuzhanova N, Abeysinghe SS, O'Connell CD, Cooper DN, Wells RD. 2004. Breakpoints of gross deletions coincide with non-B DNA conformations. Proc Natl Acad Sci USA 101:1416214167.
  • Ballinger SW, Shoffner JM, Gebhart S, Koontz DA, Wallace DC. 1994. Mitochondrial diabetes revisited. Nat Genet 7:458459.
  • Bank C, Soulimane T, Schroder JM, Buse G, Zanssen S. 2000. Multiple deletions of mtDNA remove the light strand origin of replication. Biochem Biophys Res Commun 279:595601.
  • Bender A, Krishnan KJ, Morris CM, Taylor GA, Reeve AK, Perry RH, Jaros E, Hersheson JS, Betts J, Klopstock T, Taylor RW, Turnbull DM. 2006. High levels of mitochondrial DNA deletions in substantia nigra neurons in aging and Parkinson disease. Nat Genet 38:515517.
  • Berk AJ, Clayton DA. 1974. Mechanism of mitochondrial DNA replication in mouse L-cells: asynchronous replication of strands, segregation of circular daughter molecules, aspects of topology and turnover of an initiation sequence. J Mol Biol 86:801824.
  • Bodyak ND, Nekhaeva E, Wei JY, Khrapko K. 2001. Quantification and sequencing of somatic deleted mtDNA in single cells: evidence for partially duplicated mtDNA in aged human tissues. Hum Mol Genet 10:1724.
  • Bouzidi MF, Poyau A, Godinot C. 1998. Co-existence of high levels of a cytochrome b mutation and of a tandem 200 bp duplication in the D-loop of muscle human mitochondrial DNA. Hum Mol Genet 7:385391.
  • Bowmaker M, Yang MY, Yasukawa T, Reyes A, Jacobs HT, Huberman JA, Holt IJ. 2003. Mammalian mitochondrial DNA replicates bidirectionally from an initiation zone. J Biol Chem 278:5096150969.
  • Brockington M, Sweeney MG, Hammans SR, Morgan-Hughes JA, Harding AE. 1993. A tandem duplication in the D-loop of human mitochondrial DNA is associated with deletions in mitochondrial myopathies. Nat Genet 4:6771.
  • Brown TA, Cecconi C, Tkachuk AN, Bustamante C, Clayton DA. 2005. Replication of mitochondrial DNA occurs by strand displacement with alternative light-strand origins, not via a strand-coupled mechanism. Genes Dev 19:24662476.
  • Calvo SE, Compton AG, Hershman SG, Lim SC, Lieber DS, Tucker EJ, Laskowski A, Garone C, Liu S, Jaffe DB. 2012. Molecular diagnosis of infantile mitochondrial disease with targeted next-generation sequencing. Sci Transl Med 4:118ra10.
  • Chang DD, Clayton DA. 1985. Priming of human mitochondrial DNA replication occurs at the light-strand promoter. Proc Natl Acad Sci USA 82:351355.
  • Chen X, Prosser R, Simonetti S, Sadlock J, Jagiello G, Schon EA. 1995. Rearranged mitochondrial genomes are present in human oocytes. Am J Hum Genet 57:239247.
  • Cohen O, Cans C, Gilardi JL, Roth H, Mermet MA, Jalbert P, Demongeot J, Cuillel M. 1996. Cartographic study: breakpoints in 1574 families carrying human reciprocal translocations. Hum Genet 97:659667.
  • Cooper DN, Bacolla A, Ferec C, Vasquez KM, Kehrer-Sawatzki H, Chen JM. 2011. On the sequence-directed nature of human gene mutation: the role of genomic architecture and the local DNA sequence environment in mediating gene mutations underlying human inherited disease. Hum Mutat 32:10751099.
  • Cortopassi GA, Shibata D, Soong NW, Arnheim N. 1992. A pattern of accumulation of a somatic deletion of mitochondrial DNA in aging human tissues. Proc Natl Acad Sci USA 89:73707374.
  • Damas J, Carneiro J, Goncalves J, Stewart JB, Samuels DC, Amorim A, Pereira F. 2012. Mitochondrial DNA deletions are associated with non-B DNA conformations. Nucleic Acids Res 40:76067621.
  • Di Rienzi SC, Collingwood D, Raghuraman MK, Brewer BJ. 2009. Fragile genomic sites are associated with origins of replication. Genome Biol Evol 1:350.
  • DiMauro S, Hirano M. 2003. Mitochondrial DNA deletion syndromes. In: Pagon R, Bird T, Dolan C, Stephens K, editors. GeneReviews. Seattle: University of Washington.
  • Doda JN, Wright CT, Clayton DA. 1981. Elongation of displacement-loop strands in human and mouse mitochondrial DNA is arrested near specific template sequences. Proc Natl Acad Sci USA 78:61166120.
  • Dunbar DR, Moonie PA, Swingler RJ, Davidson D, Roberts R, Holt IJ. 1993. Maternally transmitted partial direct tandem duplication of mitochondrial DNA associated with diabetes mellitus. Hum Mol Genet 2:16191624.
  • Eichler DC, Wang TS, Clayton DA, Korn D. 1977. In vitro replication of mitochondrial DNA. Elongation of the endogenous primer sequence in D loop mitochondrial DNA by human DNA polymerase beta. J Biol Chem 252:78887893.
  • Fish J, Raule N, Attardi G. 2004. Discovery of a major D-loop replication origin reveals two modes of human mtDNA synthesis. Science 306:20982101.
  • Fromenty B, Carrozzo R, Shanske S, Schon EA. 1997. High proportions of mtDNA duplications in patients with Kearns–Sayre syndrome occur in the heart. Am J Med Genet 71:443452.
  • Guo X, Popadin KY, Markuzon N, Orlov YL, Kraytsberg Y, Krishnan KJ, Zsurka G, Turnbull DM, Kunz WS, Khrapko K. 2010. Repeats, longevity and the sources of mtDNA deletions: evidence from ‘deletional spectra’. Trends Genet 26:340343.
  • Hao H, Manfredi G, Moraes CT. 1997. Functional and structural features of a tandem duplication of the human mtDNA promoter region. Am J Hum Genet 60:13631372.
  • He J, Mao CC, Reyes A, Sembongi H, Di RM, Granycome C, Clippingdale AB, Fearnley IM, Harbour M, Robinson AJ, Reichelt S, Spelbrink JN, et al. 2007. The AAA+ protein ATAD3 has displacement loop binding properties and is involved in mitochondrial nucleoid organization. J Cell Biol 176:141146.
  • Hixson JE, Wong TW, Clayton DA. 1986. Both the conserved stem-loop and divergent 5’-flanking sequences are required for initiation at the human mitochondrial origin of light-strand DNA replication. J Biol Chem 261:23842390.
  • Holt IJ, Dunbar DR, Jacobs HT. 1997. Behaviour of a population of partially duplicated mitochondrial DNA molecules in cell culture: segregation, maintenance and recombination dependent upon nuclear background. Hum Mol Genet 6:12511260.
  • Holt IJ, Harding AE, Morgan-Hughes JA. 1988. Deletions of muscle mitochondrial DNA in patients with mitochondrial myopathies. Nature 331:717719.
  • Hou JH, Wei YH. 1996. The unusual structures of the hot-regions flanking large-scale deletions in human mitochondrial DNA. Biochem J 318(Pt 3):10651070.
  • Ikebe S, Tanaka M, Ohno K, Sato W, Hattori K, Kondo T, Mizuno Y, Ozawa T. 1990. Increase of deleted mitochondrial DNA in the striatum in Parkinson's disease and senescence. Biochem Biophys Res Commun 170:10441048.
  • Johns DR, Rutledge SL, Stine OC, Hurko O. 1989. Directly repeated sequences associated with pathogenic mitochondrial DNA deletions. Proc Natl Acad Sci 86:80598062.
  • Kajander OA, Rovio AT, Majamaa K, Poulton J, Spelbrink JN, Holt IJ, Karhunen PJ, Jacobs HT. 2000. Human mtDNA sublimons resemble rearranged mitochondrial genomes found in pathological states. Hum Mol Genet 9:28212835.
  • Kearns TP, Sayre G. 1958. Retinitis pigmentosa, external ophthalmophegia, and complete heart block: unusual syndrome with histologic study in one of two cases. AMA Arch Ophthalmol 60:280289.
  • Khrapko K, Bodyak N, Thilly WG, van Orsouw NJ, Zhang X, Coller HA, Perls TT, Upton M, Vijg J, Wei JY. 1999. Cell-by-cell scanning of whole mitochondrial genomes in aged human heart reveals a significant fraction of myocytes with clonally expanded deletions. Nucleic Acids Res 27:24342441.
  • Khrapko K, Vijg J. 2009. Mitochondrial DNA mutations and aging: devils in the details? Trends Genet 25:9198.
  • Kleinle S, Wiesmann U, SupertiFurga A, Krahenbuhl S, Boltshauser E, Reichen J, LiechtiGallati S. 1997. Detection and characterization of mitochondrial DNA rearrangements in Pearson and Kearns–Sayre syndromes by long PCR. Hum Genet 100:643650.
  • Kraytsberg Y, Kudryavtseva E, McKee AC, Geula C, Kowall NW, Khrapko K. 2006. Mitochondrial DNA deletions are abundant and cause functional impairment in aged human substantia nigra neurons. Nat Genet 38:518520.
  • Krishnan KJ, Reeve AK, Samuels DC, Chinnery PF, Blackwood JK, Taylor RW, Wanrooij S, Spelbrink JN, Lightowlers RN, Turnbull DM. 2008. What causes mitochondrial DNA deletions in human cells? Nat Genet 40:275279.
  • Krishnan KJ, Blackwood JK, Reeve AK, Turnbull DM, Taylor RW. 2010. Detection of mitochondrial DNA variation in human cells. Methods Mol Biol 628:227257.
  • Kujoth GC, Bradshaw PC, Haroon S, Prolla TA. 2007. The role of mitochondrial DNA mutations in mammalian aging. PLoS Genet 3:e24.
  • Lakshmanan LN, Gruber J, Halliwell B, Gunawan R. 2012. Role of direct repeat and stem-loop motifs in mtDNA deletions: cause or coincidence? PLoS One 7:e35271.
  • Linnane AW, Baumer A, Maxwell RJ, Preston H, Zhang CF, Marzuki S. 1990. Mitochondrial gene mutation: the ageing process and degenerative diseases. Biochem Int 22:10671076.
  • Lopez-Gallardo E, Lopez-Perez MJ, Montoya J, Ruiz-Pesini E. 2009. CPEO and KSS differ in the percentage and location of the mtDNA deletion. Mitochondrion 9:314317.
  • Manfredi G, Vu T, Bonilla E, Schon EA, DiMauro S, Arnaudo E, Zhang L, Rowland LP, Hirano M. 1997. Association of myopathy with large-scale mitochondrial DNA duplications and deletions: which is pathogenic? Ann Neurol 42:180188.
  • Mita S, Rizzuto R, Moraes CT, Shanske S, Arnaudo E, Fabrizi GM, Koga Y, DiMauro S, Schon EA. 1990. Recombination via flanking direct repeats is a major cause of large-scale deletions of human mitochondrial DNA. Nucleic Acids Res 18:561567.
  • MITOMAP. 2011. A Human Mitochondrial Genome Database.
  • Moraes CT, DiMauro S, Zeviani M, Lombes A, Shanske S, Miranda AF, Nakase H, Bonilla E, Werneck LC, Servidei S. 1989. Mitochondrial DNA deletions in progressive external ophthalmoplegia and Kearns–Sayre syndrome. N Engl J Med 320:12931299.
  • Moslemi AR, Lindberg C, Oldfors A. 1997. Analysis of multiple mitochondrial DNA deletions in inclusion body myositis. Hum Mutat 10:381386.
  • Muraki K, Sakura N, Ueda H, Kihara H, Goto Y. 2001. Clinical implications of duplicated mtDNA in Pearson syndrome. Am J Med Genet 98:205209.
  • Nakase H, Moraes CT, Rizzuto R, Lombes A, DiMauro S, Schon EA. 1990. Transcription and translation of deleted mitochondrial genomes in Kearns–Sayre syndrome: implications for pathogenesis. Am J Hum Genet 46:418.
  • Nardi F, Carapelli A, Frati F. 2012. Repeated regions in mitochondrial genomes: distribution, origin and evolutionary significance. Mitochondrion 12:483491.
  • Nishigaki Y, Marti R, Hirano M. 2004. ND5 is a hot-spot for multiple atypical mitochondrial DNA deletions in mitochondrial neurogastrointestinal encephalomyopathy. Hum Mol Genet 13:91101.
  • Nishino I, Spinazzola A, Hirano M. 1999. Thymidine phosphorylase gene mutations in MNGIE, a human mitochondrial disorder. Science 283:689692.
  • Odoardi F, Rana M, Broccolini A, Mirabella M, Modoni A, D'Amico A, Papacci M, Tonali P, Servidei S, Silvestri G. 2003. Pathogenic role of mtDNA duplications in mitochondrial diseases associated with mtDNA deletions. Am J Med Genet A 118A:247254.
  • Oliveira P, da Silva C, Cabral J. 2013. An appraisal of human mitochondrial DNA instability: new insights into the role of non-canonical DNA structures and sequence motifs. PLoS One 8:e59907.
  • Park CB, Larsson NG. 2011. Mitochondrial DNA mutations in disease and aging. J Cell Biol 193:809818.
  • Pearson HA, Lobel JS, Kocoshis SA, Naiman JL, Windmiller J, Lammi AT, Hoffman R, Marsh JC. 1979. A new syndrome of refractory sideroblastic anemia with vacuolization of marrow precursors and exocrine pancreatic dysfunction. J Pediatr 95:976984.
  • Pereira F, Soares P, Carneiro J, Pereira L, Richards MB, Samuels DC, Amorim A. 2008. Evidence for variable selective pressures at a large secondary structure of the human mitochondrial DNA control region. Mol Biol Evol 25:27592770.
  • Pevzner P, Tesler G. 2003. Human and mouse genomic sequences reveal extensive breakpoint reuse in mammalian evolution. Proc Natl Acad Sci 100:76727677.
  • Poulton J. 1992. Duplications of mitochondrial DNA: implications for pathogenesis. J Inherit Metab Dis 15:487498.
  • Poulton J, Deadman ME, Bindoff L, Morten K, Land J, Brown G. 1993. Families of mtDNA re-arrangements can be detected in patients with mtDNA deletions: duplications may be a transient intermediate form. Hum Mol Genet 2:2330.
  • Poulton J, Deadman ME, Gardiner RM. 1989b. Tandem direct duplications of mitochondrial DNA in mitochondrial myopathy: analysis of nucleotide sequence and tissue distribution. Nucleic Acids Res 17:1022310229.
  • Poulton J, Deadman ME, Gardiner RM. 1989a. Duplications of mitochondrial DNA in mitochondrial myopathy. Lancet 1:236240.
  • Poulton J, Morten KJ, Weber K, Brown GK, Bindoff L. 1994. Are duplications of mitochondrial DNA characteristic of Kearns–Sayre syndrome? Hum Mol Genet 3:947951.
  • Reeve AK, Krishnan KJ, Elson JL, Morris CM, Bender A, Lightowlers RN, Turnbull DM. 2008. Nature of mitochondrial DNA deletions in substantia nigra neurons. Am J Hum Genet 82:228235.
  • Robberson DL, Kasamatsu H, Vinograd J. 1972. Replication of mitochondrial DNA. Circular replicative intermediates in mouse L cells. Proc Natl Acad Sci USA 69:737741.
  • Rogounovitch TI, Saenko VA, Shimizu-Yoshida Y, Abrosimov AY, Lushnikov EF, Roumiantsev PO, Ohtsuru A, Namba H, Tsyb AF, Yamashita S. 2002. Large deletions in mitochondrial DNA in radiation-associated human thyroid tumors. Cancer Res 62:7031.
  • Rotig A, Bourgeron T, Chretien D, Rustin P, Munnich A. 1995. Spectrum of mitochondrial-Dna rearrangements in the Pearson Marrow-Pancreas Syndrome. Human Mol Genet 4:13271330.
  • Rotig A, Colonna M, Bonnefont JP, Blanche S, Fischer A, Saudubray JM, Munnich A. 1989. Mitochondrial DNA deletion in Pearson's marrow/pancreas syndrome. Lancet 1:902903.
  • Sadikovic B, Wang J, El-Hattab A, Landsverk M, Douglas G, Brundage EK, Craigen WJ, Schmitt ES, Wong LJ. 2010. Sequence homology at the breakpoint and clinical phenotype of mitochondrial DNA deletion syndromes. PLoS One 5:e15687.
  • Samuels DC, Schon EA, Chinnery PF. 2004. Two direct repeats cause most human mtDNA deletions. Trends Genet 20:393398.
  • Sbisa E, Tanzariello F, Reyes A, Pesole G, Saccone C. 1997. Mammalian mitochondrial D-loop region structural analysis: identification of new conserved sequences and their functional and evolutionary implications. Gene 205:125140.
  • Schon EA, Bonilla E, DiMauro S. 1997. Mitochondrial DNA mutations and pathogenesis. J Bioenerg Biomembr 29:131149.
  • Schon EA, Rizzuto R, Moraes CT, Nakase H, Zeviani M, DiMauro S. 1989. A direct repeat is a hotspot for large-scale deletion of human mitochondrial DNA. Science 244:346349.
  • Schon EA, DiMauro S, Hirano M. 2012. Human mitochondrial DNA: roles of inherited and somatic mutations. Nat Rev Genet 13:878890.
  • Sembongi H, Di Re M, Bokori-Brown M, Holt IJ. 2007. The yeast Holliday junction resolvase, CCE1, can restore wild-type mitochondrial DNA to human cells carrying rearranged mitochondrial DNA. Hum Mol Genet 16:23062314.
  • Spelbrink JN, Li FY, Tiranti V, Nikali K, Yuan QP, Tariq M, Wanrooij S, Garrido N, Comi G, Morandi L, Santoro L, Toscano A, et al. 2001. Human mitochondrial DNA deletions associated with mutations in the gene encoding Twinkle, a phage T7 gene 4-like protein localized in mitochondria. Nat Genet 28:223231.
  • Srivastava S, Moraes CT. 2005. Double-strand breaks of mouse muscle mtDNA promote large deletions similar to multiple mtDNA deletions in humans. Hum Mol Genet 14:893902.
  • Tang S, Wang J, Zhang VW, Li F, Landsverk M, Cui H, Truong CK, Wang G, Chen LC, Graham B. 2013. Transition to next generation analysis of the whole mitochondrial genome: a summary of molecular defects. Hum Mutat 34:882893.
  • Tang Y, Manfredi G, Hirano M, Schon EA. 2000. Maintenance of human rearranged mitochondrial DNAs in long-term cultured transmitochondrial cell lines. Mol Biol Cell 11:23492358.
  • Tengan CH, Ferreiro-Barros C, Cardeal M, Fireman MA, Oliveira AS, Kiyomoto BH, Gabbai AA. 2002. Frequency of duplications in the D-loop in patients with mitochondrial DNA deletions. Biochim Biophys Acta 1588:6570.
  • Tengan CH, Moraes CT. 1998. Duplication and triplication with staggered breakpoints in human mitochondrial DNA. Biochim Biophys Acta 1406:7380.
  • Torroni A, Lott MT, Cabell MF, Chen YS, Lavergne L, Wallace DC. 1994. mtDNA and the origin of Caucasians: identification of ancient Caucasian-specific haplogroups, one of which is prone to a recurrent somatic duplication in the D-loop region. Am J Hum Genet 55:760776.
  • Tuppen HA, Blakely EL, Turnbull DM, Taylor RW. 2010. Mitochondrial DNA mutations and human disease. Biochim Biophys Acta 1797:113128.
  • van Goethem G, Dermaut B, Lofgren A, Martin JJ, van BC. 2001. Mutation of POLG is associated with progressive external ophthalmoplegia characterized by mtDNA deletions. Nat Genet 28:211212.
  • Walberg MW, Clayton DA. 1981. Sequence and properties of the human KB cell and mouse L cell D-loop regions of mitochondrial DNA. Nucleic Acids Res 9:54115421.
  • Wanrooij S, Luoma P, van GG, van BC, Suomalainen A, Spelbrink JN. 2004. Twinkle and POLG defects enhance age-dependent accumulation of mutations in the control region of mtDNA. Nucleic Acids Res 32:30533064.
  • Wei YH, Pang CY, You BJ, Lee HC. 1996. Tandem duplications and large-scale deletions of mitochondrial DNA are early molecular events of human aging process. Ann N Y Acad Sci 786:82101.
  • Yasukawa T, Yang MY, Jacobs HT, Holt IJ. 2005. A bidirectional origin of replication maps to the major noncoding region of human mitochondrial DNA. Mol Cell 18:651662.
  • Zeviani M, Moraes CT, DiMauro S, Nakase H, Bonilla E, Schon EA, Rowland LP. 1988. Deletions of mitochondrial DNA in Kearns–Sayre syndrome. Neurology 38:13391346.
  • Zeviani M, Servidei S, Gellera C, Bertini E, DiMauro S, DiDonato S. 1989. An autosomal dominant disorder with multiple deletions of mitochondrial DNA starting at the D-loop region. Nature 339:309311.
  • Zhang F, Carvalho CM, Lupski JR. 2009. Complex human chromosomal and genomic rearrangements. Trends Genet 25:298307.

Supporting Information

  1. Top of page
  3. Introduction
  4. Data Source and Analysis
  5. Major Findings
  6. Discussion
  7. Conclusions
  8. Acknowledgment
  9. References
  10. Supporting Information

Disclaimer: Supplementary materials have been peer-reviewed but not copyedited.


Figure S1. Possible locations of the junction site in the presence of a perfect direct repeat (bold underlined nucleotides) at 5’ and 3’ breakpoints

Figure S2. Distribution of deletions and duplications breakpoints throughout the human mtDNA.

Figure S3. Two-dimensional scatterplots showing the location of the 730 mtDNA deletions (black dots) and 37 mtDNA duplications (red dots) in the seven groups.

Figure S4. Two-dimensional scatterplots showing the breakpoints of the 730 mtDNA deletions (A and B) and 37 duplications (C and D).

Figure S5. Distribution of 5’ and 3’ deletion (blue and red bars) and duplication (yellow and green bars) breakpoints in the human mtDNA. The frequency of missing sites (grey area) is defined as the number of times that each mtDNA L-strand position is missing in the 730 different mtDNA deletions.

Figure S6. The distribution of deletions (top graph) and duplications (bottom graph) is depicted for all combination of 500-nt windows from the mitochondrial genome.

Table S1. General features of mtDNA deletions and duplications

Table S2. Description of the seven mtDNA duplications exclusively located inside the control region

Table S3. Description of the seven mtDNA deletions and duplications with equal breakpoints

Table S4. Distribution of mtDNA deletions and duplications according to the length of the perfect direct repeat (DR) present at breakpoints

Table S5. List of all mtDNA deletions detected in four or more groups

Table S6. Length of perfect direct repeats (DR) present in all combinations of flanking regions of the 730 deletions and 37 duplications reported in human mtDNA

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.