In Vivo Determination of Replication Origins of Human Mitochondrial DNA by Ligation-mediated Polymerase Chain Reaction*

A large part of replication is aborted in human mitochondria, the result being a D-loop. As few attempts have been made to distinguish free 5′ ends of true replicate from those of abortive ones, we examined the 5′ ends of true replicate of human mitochondrial DNA at one nucleotide resolution in vivo by making use of ligation-mediated polymerase chain reaction. The distribution and relative amounts of origins of the true replicate are exactly the same as those of total newly synthesized heavy strands, which means that the abortion of replication is independent of 5′ ends. Treatment of DNA with RNase H frees 5′ ends on both heavy and light strands. This is the first in vivo evidence for covalently attached primer RNA to nascent strand in human mitochondrial DNA.

A large part of replication is aborted in human mitochondria, the result being a D-loop. As few attempts have been made to distinguish free 5 ends of true replicate from those of abortive ones, we examined the 5 ends of true replicate of human mitochondrial DNA at one nucleotide resolution in vivo by making use of ligation-mediated polymerase chain reaction. The distribution and relative amounts of origins of the true replicate are exactly the same as those of total newly synthesized heavy strands, which means that the abortion of replication is independent of 5 ends. Treatment of DNA with RNase H frees 5 ends on both heavy and light strands. This is the first in vivo evidence for covalently attached primer RNA to nascent strand in human mitochondrial DNA.
Replication of mitochondrial DNA (mtDNA) begins with synthesis of the heavy strand (H strand) 1 from the replication origin O H , following transcription of the light strand (L strand). Transcription of the L strand is regulated by the L strand promoter (LSP) sequence. When synthesis of the H strand has proceeded about two-thirds, synthesis of the L strand begins from replication origin O L . Thus, the total replication rate of mtDNA is determined by H strand synthesis (for reviews, see Refs. [1][2][3]. The multiple replication origins of human mitochondrial H and L strands have been determined as free 5Ј ends of mtDNA extracted from prepared mitochondria. The molecular mechanism and physiological significance of these events are poorly understood. A large part of the synthesis of H strand is aborted, leaving a displacement loop ((D-loop) or 7 S DNA). For example, over 95% of the newly synthesized H strand is a D-loop in the case of mouse L cells (4). This makes it difficult to precisely determine origins of the true replication form (nascent H strand) (5). Few investigators have distinguished the free 5Ј end of the true replicate form from that of the D-loop even though selective detection of the free 5Ј end of the true replicate form is critical to determine replication origin. Although the 5Ј ends of the nascent H strand are suggested to be the same as those of D-loop in human mitochondria (5), it is uncertain whether the distribution and relative amount of the 5Ј ends of nascent H strand are exactly the same as those of the D-loop. In addition, the free 5Ј ends of human mtDNA are not set at one nucleotide resolution (6). To precisely determine the replication origin but not the simple free 5Ј end, it is required to selectively detect the replication origin at a higher resolution.
We made use of the ligation-mediated polymerase chain reaction (LMPCR) to detect free 5Ј ends (7,8), an approach which makes feasible use of total DNA extracted directly from whole cells. By selective amplification of nascent H (not D-loop), we determined precise and comprehensive replication origins of the H strand in vivo. Here, we describe sites of free 5Ј ends and the transition sites of RNA to DNA for both H and L strands at one nucleotide resolution.

Materials
BamHI, RNase inhibitor, and T4 DNA ligase were purchased from Takara (Seta, Japan). RNase A and diethylpyrocarbonate were from Sigma. Proteinase K was from Boehringer Mannheim (Mannheim, Germany). Long Ranger® was from FMC® BioProducts (Rockland, ME). Vent DNA polymerase was from New England Biolabs (Beverly, MA). Other reagents were of analytical grade.

Preparation of DNA
HeLa MRV11 and Jurkat (human T cell leukemia line) cells were cultured as described (9) and were harvested in logarithmic proliferation phase. The cells (about 10 6 cells) were centrifuged, and the pellets were rapidly denatured and solubilized in 100 l of denaturing buffer containing 50 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1% SDS, and 0.1 mg/ml proteinase K at 50°C for 1 h. Total DNA was isolated by two extractions with phenol/chloroform (1/1), and then DNA was ethanol precipitated. The pellets were dried, solubilized in 100 l of distilled water, and treated with RNase A (0.1 g/l) and BamHI (0.1 unit/l). DNA was extracted with phenol/chloroform (1/1), ethanol precipitated, and solubilized in water. The amount of the total DNA was determined by measuring A 260 . In some cases, the initially extracted DNA was solubilized in diethylpyrocarbonate-treated water containing RNase inhibitor (0.7 units/l) and digested with BamHI alone. After extracting DNA with phenol/chloroform (1/1), precipitating with ethanol, and solubilization in water, DNA was treated with either RNase A (0.1 g/l) alone or RNase A plus RNase H (6 units/l) before LMPCR.

Preparation of Mitochondria
Mitochondria of HeLa cells were prepared by differential centrifugation (9). Briefly, the cells suspended in buffer (TES) containing 0.25 M sucrose, 10 mM Tris-HCl, pH 7.4, and 0.1 mM EDTA were homogenized with a Potter-Elvehjem homogenizer and centrifuged at 600 ϫ g for 10 min at 4°C. The supernatant was centrifuged at 7,000 ϫ g for 10 min. The pellets were washed three times with TES. The mitochondrial fraction was allowed to proceed in in vitro replication at 37°C for 1 h, as described by Koike et al. (10,11) prior to DNA extraction.

Ligation-mediated PCR
The primer sets used in this study are shown in Tables I and II. A unidirectional linker was prepared by hybridizing LMPR1 (5Ј-gcggtgac-ccgggagatctgtattc-3Ј) and LMPR2 (5Ј-gaatacagatc-3Ј). DNA was chemically modified for the sequence ladder according to the method of Maxam and Gilbert (12). LMPCR was performed essentially according to the method of Mueller et al. (13) as follows.
Primer Extension-Primer 1 was extended in 30 l of the first strand synthesis reaction mixture consisting of 40 mM NaCl, 10 mM Tris-HCl, pH 8.9, 5 mM MgSO 4 , 0.01% gelatin, 0.3 pmol primer 1, 0.2 mM of each dNTP, 0.5 units of Vent DNA polymerase, and DNA (0.2 g for sequence ladder and 0.4 g for detection of 5Ј ends). DNA was denatured at 95°C for 5 min, and the primer was annealed at 54 -60°C for 30 min, after which polymerase reaction was performed at 76°C for 10 min.
Ligation-After cooling on ice, 45 l of ice-cold ligation mix was added, and the ligation reaction was performed at 16°C for more than 6 h. The ligation mix consisted of 77 mM Tris-HCl, pH 7.5, 13.3 mM MgCl 2 , 33.3 mM dithiothreitol, 8.3 mg/ml bovine serum albumin, 1.7 mM ATP, 100 pmol of unidirectional linker, and 3 Weiss units of T4 DNA ligase. DNA was precipitated with 9.4 l of ice-cold precipitation mix (2.7 M sodium acetate, pH 7.0, and 1 mg of tRNA) and 220 l of ice-cold ethanol, and then the entire mixture was kept at Ϫ20°C for 2 h.
PCR Amplification-DNA pellets were solubilized in 66.5 l of water, and then 33.5 l of amplification mix (123 mM NaCl, 61.5 mM Tris-HCl, pH 8.9, 15 mM MgSO 4 , 0.03% gelatin, 0.3% Triton X-100, 10 pmol LMPR1, 20 pmol primer 2, 0.67 mM of each dNTP, and 1 unit of Vent DNA polymerase) was added to the DNA solution. DNA was denatured initially at 95°C for 3 min, and then the reaction underwent 20 PCR cycles of 95°C for 1 min, 60 -65°C for 2 min, and 76°C for 3 min plus an extra 5 s for each cycle. Final extension was allowed to proceed at 76°C for 10 min. Usually, the 5Ј end of primer 2 was fluorescein isothiocyanate-labeled, and this step was followed by the DNA extraction and analysis as described below.
To distinguish the nascent H strand from the total newly synthesized H strand, a three-primers system was adopted. H1 and H2 were used as primers 1 and 2, respectively, for the nascent H strand. For the total newly synthesized H strand, D1 and D2 were used as primers 1 and 2, respectively. Fifty l of end-labeling mix (40 mM NaCl, 20 mM Tris-HCl, pH 8.9, 5 mM MgSO 4 , 0.01% gelatin, 0.1% Triton X-100, 5 pmol LMPR1, 20 pmol FD6 as primer 3, 0.2 mM of each dNTP, and 0.5 units of Vent DNA polymerase) was added to a 50-l aliquot of each PCR mixture after amplification for end labeling. After initial denaturation at 95°C for 3 min, 2 cycles of reaction were run at 95°C for 1 min, at 65°C for 2 min, and at 76°C for 3 min, followed by a final extension of 76°C for 5 min.
DNA Extraction and Analysis-DNA was extracted with phenol/ chloroform (1/1), ethanol precipitated, dried, and finally solubilized in 10 l of loading buffer (80% formamide, 45 mM Tris base, 45 mM boric acid, and 1 mM EDTA). After heat denaturation, 3 l of sample was electrophoresed through a 5% Long Ranger® 7 M urea gel. The products were scanned with FluorImager SI (Molecular Dynamics, Sunnyvale, CA) and quantified with ImageQuant TM (Molecular Dynamics, Sunnyvale, CA) software.
Sequencing D-loop Region-The D-loop region was amplified by using primers D6 and RL665D9 (5Ј-gctaggaccaaacctatttg-3Ј). The PCR product was resolved on a 1% agarose gel and then extracted. The sequence was determined on a 5% Long Ranger® 7 M urea gel using an ABI PRISM TM 377 DNA sequencer by the dye terminator cycle sequencing (Perkin Elmer) method. Primer L77 (5Ј-acgcgatagcattgcgagac-3Ј) was used for the L strand or RL665D9 for the H strand.

RESULTS
Origins for H Strand Replication-Free 5Ј ends of the H strand were determined at one nucleotide resolution. A signal at a particular site indicates that a free 5Ј end is located at the 3Ј side of the signal by one base (e.g. a signal at nucleotide 192   indicates that nucleotide 191 is a 5Ј end on the H strand). The free 5Ј ends were clustered from nucleotides 100 to 200 (Fig. 1).
Although the signal at nucleotide 152 was weaker in Jurkat cells than in HeLa cells, distribution of the signals was essentially the same between the two cell lines (Fig. 1). The sequence at nucleotide 150 on the H strand is G in HeLa MRV11 cells and A in Jurkat cells. This difference might affect usage fre-quency of a replication origin at nucleotide 151. The major signals were grouped into 4 regions around nucleotides 110, 150, 170, and 190. The locations were essentially the same as those reported by Chang and Clayton (6). These same authors (6) have reported two bands of 5Ј ends (around nucleotides 220 and 310) that almost correspond to the 3Ј ends of possible primer RNAs and another relative strong band at nucleotide 440. We did not detect the three bands by using primer sets D-7 and D-8 (results not shown).
Selective Detection of the Nascent H Strand-As a large part of the synthesis of H strand is aborted as D-loop, there is the possibility that the 5Ј ends described above do not reflect origins of the true replicate (nascent H strand). The 3Ј termini of the human D-loop have been mapped to nucleotides 16104 -16106 (14). This indicates that the newly synthesized H strand exceeding this region is the nascent H strand. Hence, we amplified the nascent H strand using primers set outside of the D-loop. We designed the other primers on the inside of the D-loop to amplify the total newly synthesized H strand (i.e. D-loop and nascent H strand). Intensity of the signals for the nascent H strand was about 40% of that for the total newly synthesized H strand (results not shown). When the conditions were selected to match the apparent signal intensity between the nascent H strand and total newly synthesized H strand, the distribution and relative amounts of signals for the nascent H strand were much the same as those of the total newly synthesized H strand (Fig. 2).
Detection of RNA Covalently Attached to the Newly Synthesized Strand-It is considered that replication of the H strand is initiated by cleavage of the L strand transcript by RNase H-like activity. In human cells, RNA covalently attached to the newly synthesized H strand is not detectable in vivo (5,15) Fig. 2. B, to determine the accurate location of the new free 5Ј ends on the H strand, LMPCR was performed using primer set D-7. C, to detect free 5Ј ends on the L strand, LMPCR was carried out using primer set O L . The signals on the right two lanes were enhanced on computer.
although such RNA is noted in mouse cells (16). We did not detect the 5Ј ends at areas where the 3Ј ends of free RNA are mapped (6). Considering the possibility that RNA is not cleaved at the sites in vivo, we treated DNA with RNase H and then performed LMPCR. This treatment led to new free 5Ј ends at nucleotides 297 and 302-309 (Fig. 3, A and B) but not around nucleotide 220, corresponding to conserved sequence block (CSB) I (Fig. 3A). The appearance of new 5Ј ends after treatment with RNase H indicates the covalent attachment of RNA to DNA. This is apparently the first demonstration of the covalent attachment of RNA to a newly synthesized H strand in human cells.
The free 5Ј ends on the L strand were also detected at nucleotides 5772, 5775, 5776, 5778, 5779, 5780, and 5762 by LMPCR (Fig. 3C). Although covalent attachment of RNA to the L strand is noted in an in vitro system (15), such has not been demonstrated in vivo. We obtained new free 5Ј ends at nucleotide 5770 and to a lesser extent at 5768, 5769, and 5774 after treatment with RNase H (Fig. 3C). These new 5Ј ends are located at the base of the stem portion in the proposed stem-loop structure near O L (15). This is the first example of the existence of covalently attached RNA to a human nascent L strand in vivo.
DNA from Prepared Mitochondria-Chang and Clayton (6) have detected free 5Ј ends in the area of nucleotide 310, without using RNase H, whereas we detected 5Ј ends only after RNase H treatment. Because they extracted DNA from mitochondria, it may be that their free 5Ј ends around nucleotide 310 are the result of activation of an endogenous RNase H-like processing enzyme during mitochondria preparations. We carried out LMPCR using DNA extracted from mitochondria and detected free 5Ј ends around nucleotide 310 without using RNase H treatment (Fig. 4), suggesting that the primer RNA attached to DNA can be processed at these sites by an endogenous RNase H-like activity. DISCUSSION We selectively detected free 5Ј ends of the true nascent H strand by LMPCR. We found no difference in the distribution and relative amounts of 5Ј ends between the nascent H strand and the total newly synthesized H strand. The same distribution pattern of 5Ј ends of nascent H strand as that of D-loop suggests that the abortion of replication is not affected by the origin.
A transcript must form a persistent RNA-DNA hybrid to prime replication. We observed new free 5Ј ends only near the CSB II region after RNase H treatment of DNA (Fig. 3). This suggests that the primer RNAs extend to CSB II and covalently attach to the nascent H strand. Chang and Clayton (6) have reported that the 3Ј termini of free RNA are mapped to three CSB regions (CSBs I-III). Our observations suggest that the RNA-DNA hybrid at CSB II is more stable than those at the other CSBs in vivo. Consistent with this, a persistent RNA-DNA hybrid is reported to be formed under an in vitro system containing the GC-rich CSB II region downstream FIG. 4. LMPCR using DNA extracted from mitochondria. The mitochondrial fraction was prepared from 2 ϫ 10 7 of HeLa MRV11 cells, and DNA was extracted as described under "Materials and Methods" (Mt). The total DNA was extracted directly from 2 ϫ 10 7 whole cells (Cell). mtDNA corresponding to the total DNA from the same number of the cells was used for LMPCR. Lane D, primer set D; lane H, primer set H.  (20). The actual sequence in CSB II is 5Ј-TTTGGGGGGGGAGGGGGG-3Ј in HeLa MRV11 cells. The large circles indicate major nucleotides with free 5Ј ends, and the small circles indicate minor nucleotides with free 5Ј ends. The arrows indicate nucleotides of which the 5Ј ends appear only after treatment with RNase H. Three CSBs are underlined. The LSP region is expressed as shadowed and underlined letters. B, the sequence on L strand is shown. The large arrow indicates major free 5Ј ends that appeared after RNase H treatment, and the small arrows indicate minor free 5Ј ends that appeared after the treatment. Other marks are the same as in panel A. of a promoter (17,18).
Although we noted a small amount of covalently attached RNA to nascent H strand near the CSB II region, almost all free 5Ј ends were confined downstream of the CSB I or the region between nucleotides 100 and 200. These 5Ј ends were free of RNA. Because the 5Ј ends in the region of nucleotides 100 to 200 should have been primed with RNA, the RNA-DNA hybrid should extend to nucleotides 100 -200 from the LSP region (around nucleotide 400) and be processed at the region. A large part of free RNA should have free 3Ј ends at nucleotides 100 to 200, while most of 3Ј ends of free RNA are clustered in three CSBs (6). The reason for the discrepancy remains to be explained.
We extensively determined the 5Ј ends of the true replicate of H strand by LMPCR (Fig. 5). This approach is sensitive, rapid, facilitated, and precise. Detection of a free 5Ј end equals to detection of the strand with a free 5Ј end (i.e. nascent strand), therefore, the signal intensity reflects the amount of nascent strand. Hence, it is possible to estimate the steady-state level of replication in cells using LMPCR (19).