Base composition at mtDNA boundaries suggests a DNA triple helix model for human mitochondrial DNA large-scale rearrangements

Molecular Genetics and Metabolism 76 (2002) 123–132 www.academicpress.com

Base composition at mtDNA boundaries suggests a DNA triple helix model for human mitochondrial DNA large-scale rearrangements Christophe Rocher,a Thierry Letellier,a William C. Copeland,b and Patrick Lestiennea,* a

EMI 99.29 INSERM, G
en
etique Mitochondriale, Universit
e Victor Segalen Bordeaux 2, 146 rue L
eo Saignat, 33076 Bordeaux Cedex, France b Laboratory of Molecular Genetics, National Institute of Health, National Institute of Environmental Health Sciences, P.O. Box 12233, Research Triangle Park, NC 27709, USA Received 22 January 2002; received in revised form 26 March 2002; accepted 26 March 2002

Abstract Different mechanisms have been proposed to account for mitochondrial DNA (mtDNA) instability based on the presence of short homologous sequences (direct repeats, DR) at the potential boundaries of mtDNA rearrangements. Among them, slippagemispairing of the replication complex during the asymmetric replication cycle of the mammalian mitochondrial DNA has been proposed to account for the preferential localization of deletions. This mechanism involves a transfer of the replication complex from the first neo-synthesized heavy (H) strand of the DR1, to the DR2, thus bypassing the intervening sequence and producing a deleted molecule. Nevertheless, the nature of the bonds between the DNA strands remains unknown as the forward sequence of DR2, beyond the replication complex, stays double-stranded. Here, we have analyzed the base composition of the DR at the boundaries of mtDNA deletions and duplications and found a skewed pyrimidine content of about 75% in the light-strand DNA template. This suggests the possible building of a DNA triple helix between the G-rich neo-synthesized DR1 and the base-paired homologous G.C-rich DR2. In vitro experiments with the purified human DNA polymerase c subunits enabled us to show that the third DNA strand may be used as a primer for DNA replication, using a template with the direct repeat forming a hairpin, with which the primer could initiate DNA replication. These data suggest a novel molecular basis for mitochondrial DNA rearrangements through the distributive nature of the DNA polymerase c, at the level of the direct repeats. A general model accounting for large-scale mitochondrial DNA deletion and duplication is proposed. These experiments extend to a DNA polymerase from an eucaryote source the use of a DNA triple helix strand as a primer, like other DNA polymerases from phage and bacterial origins. Ó 2002 Elsevier Science (USA). All rights reserved. Keywords: Slipped mispairing; Deletion; Duplication; Mitochondrial DNA

1. Introduction The 16.568 bp of the naked circular human mitochondrial DNA replicate asymmetrically during a slow process lasting about 2 h in vivo. The H (heavy)-strand DNA (leading strand) is synthesized first by DNA polymerase gamma from the heavy-strand origin of replication (OH ) upon elongation of a primer RNA generated from the cleavage of a primary transcript *

Corresponding author. Fax: +33-5-57-57-13-91. E-mail address: [email protected] (P. Lestienne).

synthesized from the light-strand promoter. When the replication complex has displaced the parental H strand at the light-strand origin of replication (OL ) located about 10 kb from OH , the light-strand origin of DNA replication is then primed by a DNA primase activity and the light strand (lagging strand) is synthesized counterclockwise. Upon completion of each synthesis, daughter DNA strands are then ligated and the superhelicity of the molecules is increased by a DNA gyrase in the presence of ATP [1]. Large-scale heteroplasmic mitochondrial DNA rearrangements were first disclosed in several progressive neuromuscular disorders including Kearns–Sayre

1096-7192/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S 1 0 9 6 - 7 1 9 2 ( 0 2 ) 0 0 0 2 4 - 0

124

C. Rocher et al. / Molecular Genetics and Metabolism 76 (2002) 123–132

syndrome, mitochondrial myopathies, chronic progressive external ophthalmoplegia, and in endocrine dysfunctions. Due to their high incidence in pathology, it is mostly deletions which have been studied. Their mapping showed that in almost all of the cases, they extend in the large arc spanning from OH to OL , a region which is replicated first, leading to a displaced single-stranded H DNA. The origins of large-scale mitochondrial DNA rearrangements causing diseases, including deletions and duplications, have remained an intriguing enigma since their discovery in 1988 (cf. [2,3], for review see [4], and [5]). The determination of the sequences occurring at deletion boundaries led to the assumption that different molecular mechanisms may be responsible for these rearrangements, as three classes of boundaries were found: in most cases (60%), the deletions occur at the level of two perfectly homologous direct repeats (DR) in the mitochondrial genome, one of them being eliminated during this process. In this case (class I), a common deletion of 5 kb presenting a direct repeat of 13 bp is encountered in about 40% of cases. Class II is characterized by imperfect direct repeats and is encountered in 30% of cases, while class III is lacking direct repeats and accounts for 10% of cases [6,7]. Due to various rearrangement breakpoints presenting either direct repeats of up to 13 bp, or partial direct repeat or no repeats, several mechanisms have been proposed including a kinetic model [8], slipped mispairing [9], topoisomerase cleavage [10], double-strand break annealing-repair [11], and recombination [7]. Abnormal recombination events were postulated from the presence of these short homologous sequences at the place where rearrangements occur. So far, except from in vitro [12] and ex vivo experiments [13], no firm data have confirmed any mammalian recombination pathway [14,15]. Bent DNA due to the high pyrimidine content in the DR of the common deletion, leading to possible mispairing and strand invasion during replication, was also suggested to promote intramolecular recombination [16]. On the other hand, due to their principal localization between the heavy- and light-strand DNA origins of replication (OH and OL ) [17], Wallace et al. [9] proposed that mtDNA deletions occur through slipped mispairing from the DR1 located near OH to the DR2 located nearby OL . This model, which is based on the presence of a 30 nucleotide outside DR1 in the deleted molecule, involves the pairing of the displaced DR1 of the H strand with the single-L strand of DR2, then nuclease digestion, and finally ligation of the intervening single H-strand DNA. However, as the DR2 is thought to be double-stranded far beyond the DR1, the molecular basis of the interactions between the DR1 of the H strand and the base-paired DR2 remains unknown.

We have already described the skewed pyrimidine composition of the DR involved in a duplication– triplication of a patient presenting maternally inherited diabetes and deafness [18], and suggested the possible transient formation of a triple helix between the neo-synthesized G-rich sequence, synthesized at the level of DR1, and the second homologous duplex DR [4,18]. More recently, we have reported that several phage and bacterial DNA polymerases may elongate a DNA forming triple helix as a primer for DNA replication [19]. Here we have studied the base composition of the DR at mtDNA rearrangement boundaries, providing extension to the initial observation [16,18], and have tested DNA polymerase c which also belongs to class A of DNA polymerases. These experiments lead us to propose, as molecular basis for large-scale mtDNA rearrangements, that formation of a transient DNA triple helix may guide slipped mispairing of the replication complex, leading to deletions and/or duplications at the level of the template with short pyrimidine-rich homologous sequences from which replication might then resume. 2. Materials and methods HPLC purified oligonucleotides were purchased from Eurogentec (Seraing, Belgium). The single-stranded DNA sequence used as a control for DNA replication, Dupl 50, complementary at its 30 end (bold) to the primers TFO Dupl is as follows: 50 -AGCTGTATCACGTTCAGTCCTCGCTCTTCTACTATGAACC CCCCTCCCCA-30 The oligonucleotide sequences used otherwise are represented in Fig. 1. The human DNA polymerase gamma subunits a and b were purified according to [20] and [21], respectively. 2.1. Analysis of the base composition of the direct repeats involved in mtDNA rearrangement The human mitochondrial DNA sequence and the sequences of the direct repeats involved in mtDNA rearrangements were obtained on the MITOMAP web site: (http://www.gen.emory.edu/mitomap.html) [5]. The human mitochondrial DNA ‘‘Cambridge’’ sequence is L-strand Genbank’s updated version as of April 18, 1997 (Accession #J01415) with the exception that further corrections have been made and annotated in accordance with the recent reanalysis by Andrews et al. [22]. The search for homologous sequences of more than 10 bases in the human mitochondrial DNA sequence was performed by computer analysis using DNASIS (Sequence Analysis Software—Hitachi Soft-

C. Rocher et al. / Molecular Genetics and Metabolism 76 (2002) 123–132

125

Fig. 1. (A) Experimental design of primer elongation reported in this study. 1—The control elongation reaction requires a 50 32 P end labeled primer (*) (here referred to as TFO Dupl 2) able to base-pair with a complementary sequence (bold in Dupl 50) of the template. Addition of unlabeled dNTP and of DNA polymerase allows the extension of the primer to 50 nucleotides. 2—Since DNA polymerases are unable to melt a double-stranded DNA, we used a partial double helix (hairpin) to which the TFO might possibly bind. However, as the TFO and the 30 part of the hairpin are identical, we had to introduce a 30 end modification (#) in the hairpin thus forming a double helix by the presence of a 20 ; 30 dideoxynucleotide. This was therefore unable to serve as primer in the in vitro elongation assay (even though unlabeled dNTP were used). If the TFO binds the double helix at the right place to enable access of its 30 OH end to the DNA polymerase, then in the presence of dNTP and DNA polymerases, extension from the 32 P 50 end labeled TFO Dupl [1–3] should yield a product 30 nucleotides long using Dupl 44 as template (as represented here), or extension from TFO Del [1–3] should give a product of 33 nucleotides long using Del 50 as template. It is to be noted that the TFO is not expected to base pair with the complementary region in the double helix, as would seem from this scheme, but rather to lie in the major groove of the duplex strand through Hoogsteen bonds [19,26]. 3—The possibility for the TFO to base-pair with the complementary sequence of the template, by displacing the homologous sequence of the hairpin, and thus to be elongated by the DNA polymerase, was tested by substituting the deoxyguanosines of the hairpin by N-7 deaza-20 -deoxyguanosines (dotted line). If this strand displacement occured, then elongation of the 5 end 32 P labeled TFO would still be observed due to the unmodified Watson–Crick base pairings. In contrast, no TFO elongation would be observed if it has to bind as a triple helix through its N2 of the deoxyguanosine residues to those of the N7 of the N-7 deazaguanosines of the hairpin [28]. (B) Representation of the oligonucleotides used in this study. Complementary nucleotides of the DR are shown in bold, 20 ; 30 dideoxynucleotides are shown by #, and N-7 deaza 20 -deoxyguanosines are shown as Go . The Dupl 44 sequence corresponds to the second direct repeat involved in a duplication–triplication [18]. Del 50 represents the sequence of the second direct repeat involved in the common deletion of 5 kb [16]. The numbers indicate the position of these sequences in the human mitochondrial DNA [23].

ware) and the calculation of the (C + T) composition was performed using Microsoft Excel. The analyzed sequence of the direct repeats involved in mtDNA rearrangements was seven base sequences reported in the submenus ‘‘mitomap: (1) Reported mtDNA Deletions. (2) Reported Multiple mtDNA Deletions Within Individuals. (3) mtDNA Simple Insertions. (4) mtDNA Complex Rearrangements.’’ The 48 selected sequences, their positions, and calculated pyrimidine percentages are given in Table 1. 2.2. Oligonucleotide labeling Oligonucleotides TFO were 50 end-labeled by 10 U of T4 polynucleotide kinase and 50 lCi of [c-32 P]ATP

(ICN, Orsay, France). Non-incorporated ATP was separated by Sephadex G50 spin filtration. 2.3. Replication assay One to two pmoles of 32 P 50 end-labeled TFO were incubated in 10 lL of 100 mM Tris–HCl, pH 7.5, 10 mM MgCl2 , 15 mM dithiothreitol, 0.2 mM dNTP, and 70 pmoles of template. Reactions were started with the addition of 0.1 lM human DNA polymerase gamma subunits. After 30 min incubation at 37 °C, samples were heat-denatured, then subjected to electrophoresis on a 20% polyacrylamide gel containing 8 M urea in 50 mM Tris–borate 2 mM EDTA buffer, pH 7.5, and subjected to autoradiography.

126

C. Rocher et al. / Molecular Genetics and Metabolism 76 (2002) 123–132

Table 1 Repeat location Single deletions 1 471–483/5152–5163 2 504–512/5443–5451 3 536–548/4430–4442 4 3166–3173/14153–14160 5 6076–6084/13799–13807 6 6218–6226/13447–13455 7 6326–6341/13989–14004 8 6325–6340/13989–14005 9 6331–6341/13994–14004 10 6466–6476/14135–14145 11 7484–7491/10996–11003 12 7662–7669/15429–15436 13 7816–7823/15381–15388 14 7975–7982/15496–15053 15 8470–8482/13447–13459 16 8547–8587/15727–15737 17 8617–8624/13877–13885 18 8637–8648/16073–16084 19 8814–8823/15845–15854 20 8829–8838/14896–14095 21 9137–9144/13808–13815 22 9345–9361/13851–13868 23 9998–10006/15895–15903 24 10191–10198/13753–13760 25 10745–10754/14124–14133 26 10942–10951/15362–15371 27 10953–10961/15837–15845 28 11234–11242/13981–13989 29 12103–12112/14412–14421 Duplications 30 302–308/567–573 31 10666–10676/14856–14866 32 3318–3337/13445–13462

Sequence

Flanking repeat

TACTCTAATCTC/TACTACTATCTC TCCTACCCA/TCCTCCCCA CCATACCCCGAAC/CCATACC CCGAAC TTCCCCCG/TTCCCCCG TCACAGCCC/TACAGCCC ACCTCCCTC/ACCTCCCTC CTCCGTAGACCTAACC/CCTCCTAGACCTAACC CCTCCGTAGACCTAAC/CCTCCTAGACCTAACCT TAGACCTAACC/TAGACCTAACC TCCTAATCACA/TCCTAATCACA AAGCCAAC/AAGCCAAC ACGCCCTC/ACGCCCTC CCTCCCAT/CCTCCCAT AGGCGACC/AGGCGACC ACCTCCCTCACCA/ACCTCCCTCACCA CCTACCCGCCGCAG/CCTAGCCGCAG ATCCCCAC/ATCCCCAC CATCAACAACCG/CATCAACAACCG CCAACTATCT/CCAACTATCT CCTAGCCATG/CCTAGCCATG TCGCTGTC/TCGCTGTC CTACTAACCAACACACT/CTACCTAACCAACAAACT TAGTATAAA/TAGTATAAA TCCCCCGC/TCCCCCGC CCTAAACCTA/CCTAACCCTA AACAACCCCC/AACAACCCCC TCCTAATAC/TCCTAATAC CCCCTACTC/CCCCTACTC CCTCAACCCC/CCTCAACCCC

12/13 8/9 13/13 8/8 9/9 9/9 16/17 17/18 11/11 11/11 8/8 8/8 8/8 8/8 13/13 11/14 8/8 12/12 10/10 10/10 8/8 16/18 9/9 8/8 9/10 10/10 9/9 9/9 10/10

67 75 62 88 67 89 63 65 55 64 38 75 88 38 77 82 75 50 70 60 75 63 33 88 67 60 67 89 80

7/7 10/11 16/19

86 70 74

10/10 10/10 9/9 7/7

90 80 89 100

7/8 7/8 12/12 13/13 8/8 10/10 8/8 12/12 9/9 9/10 8/8 8/8

14 14 50 77 75 30 63 58 67 56 88 75

33 34 35 36

3568–3577/15537–15546 4389–4398/14812–14821 8272–8280/8281–8289 9495–9501 /9580–9586

ACCCCCC/ACCCCCC TTGCCGCCTGC/TTGGCGCCTGC CAACCTCCTACTCCTCATTG/CAACCTCCCTCACC ATTG CCCCTCCCCA/CCCCTCCCCA CACCCCATCC/CACCCCATCC CCCCCTCTA/CCCCCTCTA TTTTTCT/TTTTTCT

Multiple 37 38 39 40 41 42 43 44 45 46 47 46

deletions 7850–7857/15762–15769 7924–7931/15760–15767 8637–8648/16073–16084 8470–8482/13447–13459 9274–9281/13802–13809 9431–9440/14072–14082 9437–9444/13956–13963 9486–9497/13723–13734 9579–9587/13944–13952 9833–9844/16061–16070 10751–10758/14129–14137 13754–13761/14244–14251

GAGGTCAA/GAGGACAA CGGCGGAC/CGGAGGAC CATCAACAACCG/CATCAACAACCG ACCTCCCTCACCA/ACCTCCCTCACCA CAGCCCTC/CAGCCCTC CCAAAAAGGC/CCCAAAAAGGC AGGCCTTC/AGGCCTTC TTCGCAGGATTT/TTCGCAGGATTT AATCCCCTA/AATCCCCTA TATTGGCTCAAC/TATTGACTCA CCTACTCC/CCCTACTCC CCCCCGCA/CCCCCGCA

3. Results 3.1. Skewed base composition of the direct repeats involved in mtDNA rearrangements While the overall human mtDNA L-strand template is characterized by a 56% C + T content [23], a

% Pyrimidine

search for homologous sequences by computer analysis revealed the presence of 332 DR of more than 10 bases, which had an average (C + T) composition of 61%. When comparing the base composition of the deletion breakpoints with a DR of more than seven bases, we found that only 7 out of the 29 reported breakpoints

C. Rocher et al. / Molecular Genetics and Metabolism 76 (2002) 123–132

127

display a pyrimidine content below the average content of the DR of 61% (Table 1). In contrast, nearly twothirds of them (17 out of 29) displayed a pyrimidine content ranging from 67% to 100% with a mean value of 78% (SD  17%). Seven deletion breakpoints contained between 60% and 66% pyrimidine, and only five of them contained between 33% and 55%. In the case of duplications, this skewed base composition was even more striking with a range of 70–100% and a mean value of 84% (SD  10%). On the other hand, a range of 14–88% with a mean value of 56% was found in multiple deletions (SD  24%). This perusal of the composition of the DR at mtDNA breakpoint boundaries therefore extends the observation made on the common deletion [16] and in a duplication–triplication [18] showing a significant increase in the C + T content at the level of the DR of the rearrangement breakpoints as compared with the C + T content of the DR of the mtDNA. The latter is also skewed with respect to the overall C + T mtDNA content. 3.2. Unusual properties of polypurine polypyrimidine-rich sequences With the exception of multiple deletions, this skewed base composition at mtDNA rearrangement boundaries suggested a molecular basis for large-scale rearrangements involving pyrimidine-rich template sequences. It has already been shown that oligonucleotides may form a stable triple helix when a double-stranded DNA formed by a purine-rich sequence (e.g., W strand in Fig. 2) paired with a pyrimidine-rich one (C strand) binds with either a third polypurine strand (H strand of W sequence) or with a polypyrimidine strand (H strand of C sequence) with an antiparallel orientation to the W or C sequence [24]. However, parallel orientation of the Grich third strand has been reported from vibrational spectroscopy [25] and X-ray diffraction studies [26]. This observation led us to propose that the template DNA ensures the synthesis of a purine-rich sequence at the level of the DR1, which can match with the duplex sequence (DR2) containing the homologous strand of parallel orientation through the formation of a transient triple helix from which replication could resume.

Fig. 2. Schematic representation of the slipped mispairing model mediated through the formation of a transient DNA triple helix. The arrows indicate the single strand of the direct repeats and the orientation of DNA replication. The W represents the Watson strand or heavy strand of the mtDNA. The C represents the Crick strand or light strand of the mtDNA. The H represents the Hoogsteen strand identical to the W (or heavy) strand. The W strand is not base-paired with the C strand at the level of DR1, but could be at the level of DR2, far away from the replication fork. 1—Synthesis of the polypurine tract (Heavy-strand sequence) at the level of DR1 followed by the dissociation of the replication complex from the DNA template. 2–Binding of the replication complex to the base-paired DR2 through Hoogsteen base pairings, forming a transient triple helix with the homologous base-paired poly (purine.pyrimidine)-rich sequence. This complex is represented by the hairpin template used in these studies with the neosynthesized H strand serving as a primer (TFO) for further DNA replication. 3—Reinitiation of DNA synthesis from the TFO leading to a deletion between repeats when the newly synthesized H strand is replicated. The Watson–Crick strands might then be unwound by the helicase activity of the replication complex, and the DNA polymerase would resume synthesis using the C-strand template leading to a deleted molecule.

3.3. In vitro replication from primers forming a triple helix by the human DNA polymerase gamma We have previously reported that the neo-synthesized H strand corresponding to the DR1 could specifically bind to the double-stranded sequence of the DR2 by gel shift assays [19], indicating that they form a triple helix and could initiate DNA replication if the 30 OH available end is located in the catalytic centre of the DNA polymerase in parallel orientation, like the homologous

strand of the DR2. The experimental design of the oligonucleotides used for these studies is outlined in Fig. 1A. We therefore investigated the possibility that the third strand might form a triple helix of the length of the DR (TFO) to prime DNA replication in vitro. For this purpose, the hairpin DNA Dupl 44 and Del 50 with a 50 end 20 nucleotide extension were used since DNA

128

C. Rocher et al. / Molecular Genetics and Metabolism 76 (2002) 123–132

polymerases are unable to melt a double-stranded structure but require a DNA helicase activity. Furthermore, their 30 end was modified with a 20 ; 30 dideoxynucleotide to prevent their elongation by DNA polymerase (Fig. 1B). The 50 32 P end-labeled TFO was incubated under replication condition, with their corresponding hairpin template, in the presence of deoxyribonucleotide triphosphates and with catalytic amounts of the recombinant human DNA polymerase c subunits. The fate of the 50 32 P end labeled TFO was studied by their possible elongation in the assay upon incubation and electrophoresis under denaturing conditions, followed by autoradiography as shown in Fig. 3. In the first experiments lacking the DNA polymerase c, spots were evidenced, migrating at the position of about 20 nucleotides and shown by a star in Lanes 1–3. At present, we have no clear explanation for the occurrence of these spots which were still detected upon heat denaturation, in different chemical syntheses, and when purity was checked by HPLC. Their apparant size might correspond to the dimers of TFO (Fig. 3A), resulting in their richness in deoxyguanosines [27]. Addition of DNA polymerase c enabled the formation of the expected product of 30 nucleotides in length (Lanes 5–7) corresponding to the length of the TFO Dupl (9–11 nucleotides), and that of the 50 end extension of template Dupl 44 to be replicated (20 nucleotides), as shown in Fig. 3A. Such an elongation product corresponds to the sequenced replication product using T7 DNA polymerase [19]. Comparable studies were performed with the system mimicking the common deletion using the TFO Del

1–TFO Del 3 as primers, and the Del 50 as a template. Fig. 3B shows that the addition of the DNA polymerase c to the assays resulted in the elongation of the TFO to the expected size of 33 nucleotides (Lanes 5–7). Due to their homologous sequence at their 30 end forming the hairpin, it could be argued that the TFO displaces the homologous sequence and base-pair with the complementary sequence of the template, although their relative concentration is by a 35–70-fold lower than that of the hairpin template. We indirectly tested this possibility by substituting the 20 deoxyguanosine residues of the 30 end hairpin by 7-deaza-20 -deoxyguanosine residues. This modification prevents binding between the homologous strand forming the triple helix, thereby reducing the affinity of the third strand for the duplex [28]. As shown on Lane 8 in Fig. 3B, no elongation could be detected under these conditions, thus indicating that the primer elongation needs bonds with the 30 end hairpin homologous sequence of the template and does not result from a simple displacement of the homologous strand. Depending on the dissociation constant between the TFO primers and the corresponding double helix, mismatched TFO may also serve as primers. Additional experiments with pyrimidine-rich primers failed to be elongated by the DNA polymerase c (data not shown). Taken together, our data are the first to show that the human mitochondrial DNA polymerase c can elongate a DNA primer to form a triple helice. They thus support the concept that the direct repeats studied here, which correspond to rearrangement breakpoints, may trigger DNA replication by priming it from the neo-synthesized

Fig. 3. Elongation of the TFO primers in the presence of the hairpin templates with the human DNA polymerase gamma. (A) shows the reactions in the system of Dupl 44 with the TFO Dupl as 50 end 32 P labeled primers. Lanes 1–3 show the migration rates of the TFO Dupl 1, TFO Dupl 2, and TFO Dupl 3 alone in the replication assay without DNA polymerase. The star shows the abnormal spots of TFO. Lane 4 shows control reaction using the single-stranded template Dupl 50 (see Section 2) and TFO Dupl 2 as primer with an expected replication product 50 nucleotides long. Lanes 5–7 show elongation of the TFO Dupl 1 (Lane 5), TFO Dupl 2 (Lane 6), and of TFO Dupl 3 (Lane 7) in the replication assay by the human DNA polymerase gamma subunits, in the presence of Dupl 44. (B) shows the elongation of the TFO Del in the replication assay directed by Del 50. Lanes 1, 2, and 3 show the migration rates of the TFO Del 1, Del 2, and TFO Del 3, respectively, in the replication assay minus the human gamma polymerase subunits. Lane 4 is the control replication reaction using TFO Dupl 2 and Dupl 50 as template. Lanes 5, 6, and 7 show elongation of the TFO Del 1, TFO Del 2, and TFO Del 3, respectively, using the hairpin Del 50 as template. Lane 8 uses TFO Del 2 as primer and Deaza Del 50 as template.

C. Rocher et al. / Molecular Genetics and Metabolism 76 (2002) 123–132

deoxyguanosine-rich polynucleotide of the Heavystrand DNA sequence.

4. Discussion 4.1. Analysis of mitochondrial DNA sequences Following our previous observation of the high pyrimidine content in the DR involved in a duplication– triplication of the mitochondrial genome associated with maternally transmitted diabetes and deafness [18], we studied the base composition of 48 repeats of the mtDNA identified as breakpoints. We found that with the notable exception of nuclear-driven multiple deletions, a skewed base composition in the range of 75% and favoring the pyrimidine content of the DR in the template strand was observed in nearly two-thirds of the reported cases of deletions with DR. Five out of the 41 direct repeats of the deletion breakpoints (No. 11, 14, 23, 37, and 38) present a much lower (C + T) content in the template strand (14–38%), thus reflecting possible mirror image of the proposed model. Indeed, the C + Trich replicated strand might also form a triple helix through protonation of cytosine residues under acidic conditions [24]. However triple helices formed by polypyrimidines have to be of antiparallel orientation to the homologous strand to bind a double helix [24]. The large number of 10–15 bp DR (332) in human mtDNA is suggestive of ancient duplication events. In the case of major large-scale deletions, we found this important skewed base composition at breakpoints presenting DR. However, not all the 332 DR analyzed have been associated with mtDNA rearrangements. Indeed, 131 of these might no longer have the origins of replication, particularly OL , which when deleted might prevent any DNA replication. Yet there are still 70 DR with a pyrimidine content above 67% and these may be potential sources of rearrangements. The much higher incidence of the 5 kb deletion may be due to the long skewed pyrimidine composition of its direct repeat. Indeed, computer analysis showed the presence of four 13 bp repeats. Two of them are separated either by OH or by OL , one presents a 13 bp DR with 62% pyrimidine content, and the fourth one with the 77% content encountered in the common deletion studied here which represents about 40% of the reported deletions with DR. The mean base composition of the DR involved in duplication is even more striking with a value of 84%. 4.2. A triple helix intermediate These observations lead us to propose a molecular basis to account for these rearrangements, based on the richness of C + T at the level of the template leading strand of the DR. Here the DNA polymerase c might

129

synthesize the purine-rich H strand of the repeat, stall and leave its template to associate with a homologous duplex sequence through the formation of a transient triple helix from which replication could reinitiate from the available 30 OH end of the neo-synthesized H strand of the DR1 (Fig. 4A). Another implication of the model may be a common step in the formation of a duplicated molecule which has been described in several patients [29,30] and presents DR. As symbolized in Fig. 4B, the duplicated region might result from a second round of replication between the DR, and then reinitiation of the complete mtDNA replication cycle. An additional round of replication would provide a tandem triplication. This mechanism may also allow a certain wobble in the pairings between the third strand and the direct repeat, thus accounting for the presence of imperfect repeat, and several different breakpoints within a single patient [6]. 4.3. Pathological relevance In contrast with chromatin, mitochondrial DNA is naked within the matrix, and the a subunit of the DNA polymerase c displays a moderate processivity of 50–75 nucleotides elongated per binding event [21]. This is improved by the binding of the b subunits [31,32]. Pausing of DNA replication has been shown at the level of the structure forming the triple helix by the DNA polymerase c of Xenopus laevis [33]. Triple helix structures may occur with as few as seven nucleotides [34], a comparable length to the direct repeats examined here, thus indicating that they may occur through the mtDNA replication cycle under normal circumstances during oogenesis where the mtDNA copy number increases from about 2000 to 100,000 copies [35]. The common deletion of 4977 bp was detected in 25 of the 100,000 copies of the normal mature oocyte [36]. Additionally, repopulation of Hela cells devoid of mtDNA by normal mitochondria revealed that the common deletion was detected at a rate of 6  108 mtDNA molecule per replication cycle [37] upon several doubling times. Moreover, the common deletion has been reported in aged healthy people [38]. The model may account for these results through replication errors due to the variable processivity of the DNA polymerase c depending on its subunit composition, since the a and b subunits appear to be differentially regulated [39]. Mutant of mtDNA polymerase decreasing its replication rate and processivity are thus expected to be found in association with mtDNA deletion or duplication, as they would increase mispairing and thus the number of mtDNA rearrangements, since in vitro experiments with purified subunits suggest that homopolymeric tracts may be more prone to replication errors [40].

130

C. Rocher et al. / Molecular Genetics and Metabolism 76 (2002) 123–132

Fig. 4. Overall model for large-scale mtDNA rearrangements leading either to deletions (A) or to duplication and triplication (B) according to the triple helix slip-mispairing hypothesis. (A) 1—Synthesis of the H strand up to the DR1 and slippage mispairing (dotted line) to the DR2 through the formation of a transient DNA triple helix (star). 2—Further elongation of the H strand up to OH and initiation of L-strand replication at OL . 3— Replication of the L strand providing a deleted molecule and a normal one. (B) 1—H strand DNA synthesis up to the DR1 followed by slippedmispairing to the DR2 and rolling replication through OH to the DR1, thus providing the first partial duplicated sequence. 2—If one additional transfer of the replication complex occurs from DR1 to DR2, then a second partial tandem duplication would lead to a partially triplicated H-strand DNA. 3—The replication complex might continue normally through OL allowing replication of the light strand, leading either to a tandem duplicated molecule (dp 3a) or to a tandem triplicated molecule (tp 3b) and a normal mtDNA molecule.

This analysis also indicates that in contrast with single deletion or duplication, the base composition of the DR found in nuclear-driven multiple deletions does not appear to be skewed but rather random. Linkage analysis have pointed to the presence of several mutations in nuclear-encoded gene products playing a role in mitochondrial DNA metabolism. Mutations in the gene encoding for the thymidine phosphorylase are associated with MNGIE [41], and in Adenine Nucleotide Translocator 1 associated with autosomal dominant progressive external ophthalmoplegia characterized by multiple deletions [42]. This indicates a mechanism based on the imbalance of the nucleotide pool level. Furthermore, mutations in the catalytic subunit of the human DNA polymerase c [43] and in C10orf2 encoding for a human mitochondrial DNA helicase, have recently been associated with multiple mtDNA deletions [44]. The human DNA polymerase c catalytic subunit a thus behaves like the phage T4, T7, and Klenow fragment of Escherichia coli DNA polymerase I by using a third deoxyguanosine-rich strand as a primer. This points to their common catalytic processes, their position to the class A of procaryotic DNA polymerases, and perhaps their phage-derived origin. It is interesting

to note that the mitochondrial RNA polymerase, and DNA helicase also display ressemblances with the phage enzymes, thus supporting the idea of a progressive replacement of enzymes involved in mitochondrial DNA metabolism of metazoa by phage-derived genes. The unusual property of using a third strand as a primer was modeled and shown to fit the catalytic center of the T7 DNA polymerase [19], thus providing the first molecular basis for slippage mispairing from a primer forming a triple helix. These data may now be extended with the recent identification of mitochondrial DNA helicase [44], of mutants of DNA polymerase subunit a particularly in the DNA polymerization domain [43], and in the proofreading domain [45,46], and of deletion mutants in the accessory subunit b [47–49]. Indeed, kinetic studies using a longer double helical DNA template to detect dissociation of the enzyme, may now be possible with additional proteins participating in the mtDNA replication process, such as the mitochondrial single-stranded DNA binding protein. Triple helix DNA binding proteins have been characterized from several eukaryotes. Although their function is not clear, we propose that they might play a

C. Rocher et al. / Molecular Genetics and Metabolism 76 (2002) 123–132

role in genome rearrangements. Indeed, they may favor alternative replication modes and trigger the proposed mechanism. Whether the auxiliary subunit of the polymerase c modulates the replication processivity, or a hypothetical mitochondrial triple helix DNA binding protein, plays a role in the genesis of mammalian mitochondrial DNA alterations and explains the fine mechanism of triple helix formation, remains to be explored in diseases.

Acknowledgments We wish to thank D. Bogenhagen for a gift of DNA polymerase c subunits, E.A. Schon for comments and Ray Cooke for text editing. This work was supported by the Institut National de la Sant
e et de la Recherche M
edicale.

References [1] D.A. Clayton, Replication of animal mitochondrial DNA, Cell 28 (1982) 693–705. [2] P. Lestienne, G. Ponsot, Kearns–Sayre syndrome with muscle mitochondrial DNA deletion, Lancet 1 (1988) 805. [3] I.J. Holt, A. Harding, J.A. Morgan-Hughes, Deletion of muscle mitochondrial DNA in patients with mitochondrial myopathies, Nature 331 (1988) 717–719. [4] P. Lestienne, M.F. Bouzidi, I. Desguerre, G. Ponsot, Molecular basis of mitochondrial diseases, in: P. Lestienne (Ed.), Mitochondrial Diseases. Models and Methods, Springer, Paris, Heidelberg, 1999, pp. 33–58. [5] D.C. Wallace, M.T. Lott, M.D. Brown, K. Huoponen, A. Torroni, Report of the committee on human mitochondrial DNA, in: A.J. Cuticchia (Ed.), Human Gene Mapping 1995: A Compendium, John Hopkins University Press, Baltimore, 1995. Available from , pp. 910–954. [6] F. Degoul, I. Nelson, S. Amselem, N. Romero, B. ObermaierKusser, G. Ponsot, C. Marsac, P. Lestienne, Different mechanisms inferred from sequences of human mitochondrial DNA deletions in ocular myopathies, Nucleic Acids Res. 19 (1991) 493–496. [7] S. Mita, R. Rizzuto, C.T. Moraes, S. Shanske, E. Arnaudo, G.M. Fabrizi, Y. Koga, E.A. Schon, S. diMauro, Recombination via flanking direct repeats is a major cause of large-scale deletions of human mitochondrial DNA, Nucleic Acids Res. 18 (1990) 561– 567. [8] I. Nelson, L. d’Auriol, F. Galibert, G. Ponsot, P. Lestienne, Nucleotidic mapping and kinetic model for a 4666 base pairs heteroplasmic mitochondrial DNA deletion in Kearns–Sayre syndrome, C. R. Acad. Sci. Paris, S
erie III 309 (1989) 403–407. [9] J.M. Shoffner, M.T. Lott, A.S. Voljavec, S.A. Souedan, D.A. Costigan, D.C. Wallace, Spontaneous Kearns–Sayre: chronic external ophthalmoplegia plus syndrome associated with a mitochondrial DNA deletion: a slip-replication model and metabolic therapy, Proc. Natl. Acad. Sci. USA 86 (1989) 7952–7956. [10] R.B. Block, D.R. Thorburn, G.N. Thompson, H.H. Dahl, A topoisomerase II cleavage site is associated with a novel mitochondrial DNA deletion, Human Genet. 95 (1995) 75–81. [11] G. Coffey, U. Lakshmipathy, C. Campbell, Mammalian mitochondrial extract possess DNA end-binding activity, Nucleic Acids Res. 27 (1999) 3348–3354.

131

[12] R.A. Thyagarayan, C. Padua, C. Campbell, Mammalian mitochondria possess homologous recombination activity, J. Biol. Chem. 271 (1996) 27536–27543. [13] R. Paul, R. Dalibart, S. Lemoine, P. Lestienne, Expression of E. coli RecA targeted to mitochondria of human cells, Mutat. Res. 486 (2001) 11–19. [14] J.L. Elson, R.M. Andrew, P.F. Chinnery, R.N. Lightowlers, D.M. Turnbull, N. Howell, Analysis of European mtDNAs for recombination, Am. J. Hum. Genet. 68 (2001) 145–153. [15] T. Kivisild, R. Villems, Questioning evidence for recombination in human mitochondrial DNA, Science 288 (2000) 1931. [16] E.A. Schon, R. Rizutto, C.T. Moraes, H. Nakase, M. Zeviani, S.A. diMauro, Direct repeat is a hot spot for large-scale deletions of human mitochondrial DNA, Science 244 (1989) 346–349. [17] I. Nelson, F. Degoul, B. Obermaier-Kusser, N. Romero, C. Borrone, C. Marsac, J.L. Vayssiere, K. Gerbitz, M. Fardeau, G. Ponsot, P. Lestienne, Mapping of heteroplasmic mitochondrial DNA deletions in Kearns–Sayre syndrome, Nucleic Acids Res. 17 (1989) 8117–8124. [18] M.-L. Martin-N
egrier, M. Coquet, B. Teissier-Moretto, J.-Y. Lacut, M. Dupon, B. Bloch, P. Lestienne, C. Vital, Partial triplication of the mtDNA in maternally transmitted diabetes and deafness, Am. J. Hum. Genet. 63 (1998) 1227–1232. [19] C. Rocher, R. Dalibart, T. Letellier, G. Pr
ecigoux, P. Lestienne, Initiation of DNA replication by DNA polymerases from primers forming a triple helix, Nucleic Acids Res. 29 (2001) 3320–3326. [20] P.A. Ropp, W.C. Copeland, Cloning and characterization of the human mitochondrial DNA polymerase, DNA polymerase c, Genomics 36 (1996) 449–458. [21] S.E. Lim, M.J. Longley, W.C. Copeland, The mitochondrial P 55 accessory subunit of human DNA polymerase c enhances DNA binding, promotes processive DNA synthesis, and confers NEthylmaleimide resistance, J. Biol. Chem. 274 (1999) 38197– 38203. [22] R.M. Andrews, I. Kubacka, P.F. Chinnery, R.N. Lightowlers, D.M. Turnbull, N. Howell, Reanalysis and revision of the Cambridge reference sequence for human DNA, Nat. Genet. 23 (1999) 147. [23] S. Anderson, A.T. Bankier, B.G. Barrell, M.H. Debruijn, A.R. Coulson, J. Drouin, I.C. Eperon, D.P. Nierlich, B.A. Roe, F. Sanger, P.H. Schreier, A.J.H. Smith, R. Staden, I.G. Young, Sequence and organization of the human mitochondrial DNA, Nature 290 (1981) 457–465. [24] M.D. Frank-Kamenetskii, S.M. Mirkin, Triplex DNA structures, Ann. Rev. Biochem. 64 (1995) 65–95. [25] M. Ouali, S. Letellier, J.-S. Sun, A. Akhebat, F. Adnet, F. Liquier, E. Taillandier, Determination of G-G.C triple helix structure by molecular modeling and vibrational spectroscopy, J. Am. Chem. Soc. 115 (1993) 4264–4270. [26] D. Vlieghe, L.V. Meervelt, A. Dautant, B. Gallois, G. Pr
ecigoux, O. Kennard, Parallel and antiparallel (G.GC)2 triple helix fragment in a crystal structure, Science 273 (1996) 1702–1705. [27] E. Henderson, C.C. Hardin, S.K. Walk, I. Tinocco, E. Blackburn, Telomeric DNA oligonucleotides form novel intramolecular structures containing Guanine.Guanine base pairs, Cell 51 (1987) 899–908. [28] M.G. Kim, V.B. Zhurkin, R.L. Jernigan, R.D. Camerini-Otero, Probing the structure of a putative intermediate in homologous recombination: the third strand in the parallel DNA triplex is in contact with the major groove of the duplex, J. Mol. Biol. 247 (1995) 874–8823. [29] G. Manfredi, T. Vu, E. Bonilla, E.A. Schon, S. DiMauro, E. Arnaudo, L. Zhang, L. Rowland, M. Hirano, Association of myopathy with large-scale mitochondrial DNA duplication and deletion: which is pathogenic?, Ann. Neurol. 42 (1997) 180–188. [30] J. Poulton, K.J. Morten, D. Marchington, K. Weber, G.K. Brown, A. R€ otig, L. Bindoff, Duplications of mitochondrial DNA

132

[31]

[32]

[33]

[34] [35]

[36]

[37]

[38]

[39]

[40]

C. Rocher et al. / Molecular Genetics and Metabolism 76 (2002) 123–132 in Kearns–Sayre syndrome, Muscle Nerve 3 (Suppl.) (1995) S154– S158. M.J. Longley, P. Ropp, S. Lim, W.C. Copeland, Characterization of the native and recombinant catalytic subunit of human DNA polymerase c: identification of residues critical for exonuclease activity and dideoxynucleotides sensitivity, Biochemistry 37 (1998) 10529–10539. J.A. Carrodeguas, R. Kobayashi, S. Lim, W.C. Copeland, D.F. Bogenhagen, The accessory subunit of Xenopus laevis mitochondrial DNA polymerase c increases processivity of the catalytic subunit of human DNA polymerase c and is related to class II Aminoacyl-tRNA synthetases, Mol. Cell. Biol. 19 (1999) 4039– 4046. V.S. Mikhailov, D.F. Bogenhagen, Termination within oligo(dT) tracts in template DNA by DNA polymerase c occurs with formation of a DNA triplex structure and is relieved by mitochondrial single-stranded DNA-binding protein, J. Biol. Chem. 271 (1996) 30774–30780. V. Sklenar, J. Feigon, Formation of a stable triplex from a single DNA strand, Nature 354 (1990) 836–838. J.P. Jenuth, A.C. Peterson, K. Ku, E.A. Shoubridge, Random genetic drift in the female germ-line explains the rapid segregation of mammalian mitochondrial DNA, Nat. Genet. 13 (1996) 146– 151. X. Chen, R. Prosser, S. Simonetti, J. Sadlock, G. Jagiello, E.A. Schon, Rearranged mitochondrial genomes are present in normal ovocytes, Am. J. Hum. Genet. 57 (1995) 239–247. R. Shenkar, W. Navidi, S. Tavar
e, M.H. Dang, A. Chomyn, G. Attardi, G. Cortopassi, N. Arnheim, The mutation rate of the human mtDNA deletion mtDNA 4977, Am. J. Hum. Genet. 59 (1996) 772–780. G.A. Cortopassi, N. Arnheim, Detection of specific mitochondrial DNA deletion in tissues of older humans, Nucleic Acids Res. 18 (1990) 6927–6933. E. Lefai, M.A. Fernandez-Moreno, A. Alahari, L.S. Kaguni, R. Garesse, Differential regulation of the catalytic and accessory subunits genes of Drosophila mitochondrial DNA polymerase, J. Biol. Chem. 275 (2000) 33123–33133. M.J. Longley, D. Nguyen, T.A. Kunkel, W.C. Copeland, The fidelity of human DNA polymerase c with and without exonu-

[41]

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

cleotidic proofreading and the P55 accessory subunit, J. Biol. Chem. 276 (2001) 38555–38562. I. Nishimo, A. Spinazzolla, M. Hirano, Thymidine phosphorylase gene mutations in MNGIE, a human mitochondrial disorder, Science 283 (1999) 689–692. J. Kaukonen, J.K. Juselius, V. Tiranti, A. Kyttala, M. Zeviani, G.P. Comi, S. Keranen, L. Peltonen, A. Suomalainen, Role of adenine nucleotide translocator 1 in mtDNA maintenance, Science 289 (2000) 782–785. G.V. Van Goethem, B. Dermaut, A. L€ ofgren, J.-J. Martin, C. Van Broeckhoven, Mutations in POLG is associated with progressive external ophthalmoplegia characterized by mtDNA deletions, Nat. Genet. 28 (2001) 211–212. J.N. Spelbrink, F.-Y. Li, V. Tiranti, K. Nikali, Q.-P. Yuan, M. Tarip, S. Wanrooij, N. Garrido, G. Comi, L. Morandi, L. Santoro, A. Toscano, G.-M. Fabrizi, H. Somer, R. Croxen, D. Beeson, J. Poulton, A. Suomalainen, H.T. Jacobs, M. Zeviani, C. Larsson, Human mitochondrial DNA deletions associated with mutations in the gene encoding twinkle, a phage T7 gene-like protein localized in mitochondria, Nat. Genet. 28 (2001) 223–231. J.N. Spelbrink, J.M. Toivonen, G.A.J. Hakkaart, J.M. Kurkela, H.M. Cooper, S.K. Lehtinen, N. Lecrenier, J.W. Back, D. Speijer, F. Foury, H.T. Jacobs, In vivo functional analysis of the human mitochondrial DNA polymerase PolG expressed in cultured human cells, J. Biol. Chem. 275 (2000) 24818–24828. D. Zhang, J.L. Mott, S.-H. Chang, G. Denniger, Z. Feng, H.P. Zassenhaus, Construction of transgenic mice with tissue-specific acceleration of mitochondrial DNA mutagenesis, Genomics 69 (2000) 151–161. J.A. Carredoguas, D.F. Bogenhagen, Protein sequences conserved in prokaryotic aminoacyl-tRNA synthetases are important for the activity of the processive factor of human mitochondrial DNA polymerase, Nucleic Acids Res. 28 (2000) 1237–1244. J.A. Carrodeguas, K. Theis, D.F. Bogenhagen, C. Kisker, Crystal structure and deletion analysis show that the accessory subunit of mammalian DNA polymerase c, PolcB, functions as a homodimer, Mol. Cell. 7 (2001) 43–54. L. Fan, L. Kaguni, Multiple regions of subunit interaction in Drosophila mitochondrial DNA polymerase: three functional domains in the accessory subunit, Biochemistry 40 (2001) 4780–4791.