Mechanism of mitochondrial DNA replication in mouse L-cells: RNA priming during the initiation of heavy-strand synthesis

The major form of mouse L-cell mitochondrial DNA contains a small displacement loop at the replication origin, created by synthesis of a 550 to 670-nucleotide portion of the heavy strand. These short heavy-strand segments remain hydrogen-bonded to the parental light strand and are collectively termed 7 S mitochondrial DNA. The unique location of these 7 S mitochondrial DNAs at the heavy-strand origin suggests that they may function as primers in the synthesis of full-length heavy strands. Ribonucleotides have been detected at the 5′-end of some of these molecules, which are most likely remnants of primer RNAs. Using 5′-end labeling in vitro, we have determined that these ribonucleotides occur at several discrete positions along the nucleotide sequence of the origin region, which suggests that there may be variability in the precise initiation point of RNA priming or in the location of the switchover from RNA priming to DNA synthesis. The length of 5′-end RNA was estimated by alkali treatment of mitochondrial DNA prior to end labeling. A range of one to ten ribonucleotides was hydrolyzed from the 5′-end of some 7 S mitochondrial DNA strands. This is the first evidence of RNA priming at a eukaryotic cell DNA replication origin.     

Structure of F-actin needles from extracts of sea urchin oocytes

The mouse L-cell line LD maintains its mitochondrial DNA genome in the form of a head-to-tail unicircular dimer of the monomeric 16,000 base-pair species. This situation permits a comparison of the mechanism of replication of this dimeric molecule with our previous studies of replication of monomeric mouse L-cell mitochondrial DNA. Whereas monomeric mitochondrial DNA requires about one hour for a round of replication, the dimeric molecule requires almost three hours. Denaturing agarose gel electrophoretic analyses of replicative intermediates reveals several discrete size classes of partially replicated daughter strands of dimeric mitochondrial DNA. This suggests that replication occurs with specific discontinuities in the rate of daughter strand synthesis. The strand specificity of these daughter strands was determined by hybridization with ³²P-labeled DNA representing either the heavy or light strand mitochondrial DNA sequence. The sizes and strand specificities of these discrete daughter strands indicate that the same set of control sequences is functional in both dimer and monomer mitochondrial DNA replication.Immediately following a round of replication, the majority of dimeric mitochondrial DNA molecules contain displacement loops, as assessed by their sensitivity to nicking within the displaced DNA strand by single-strand DNA specific S₁ nuclease under conditions which leave supercoiled DNA intact. This result is in contrast with the conformation of newly replicated monomeric mitochondrial DNA molecules, which lack both superhelical turns and displacement loops. This indicates that dimeric mitochondrial DNA proceeds through a different series of post-replicative processing steps than does monomeric mitochondrial DNA. We postulate that intermediates at late stages of dimeric mitochondrial DNA replication contain displacement loops which remain intact following closure of the full-length daughter strands.     

Mechanism of replication of human mitochondrial DNA. Localization of the 5' ends of nascent daughter strands

Human mitochondrial DNA contains two physically separate and distinct origins of DNA replication. The initiation of each strand (heavy and light) occurs at a unique site and elongation proceeds unidirectionally. Animal mitochondrial DNA is novel in that short nascent strands are maintained at one origin (D-loop) in a significant percentage of the molecules. In the case of human mitochondrial DNA, there are three distinct D-loop heavy strands differing in length at the 5' end. We report here the localization of the 5' ends of nascent daughter heavy strands originating from the D-loop region. Analyses of the map positions of 5' ends relative to known restriction endonuclease cleavage sites and 5' end nucleotides indicate that the points of initiation of D-loop synthesis and actual daughter strands are the same. In contrast, the second origin is located two-thirds of the way around the genome where light strand synthesis is presumably initiated on a single-stranded template. Mapping of 5' ends of daughter light strands at this origin relative to known restriction endonuclease cleavage sites reveals two distinct points of initiation separated by 37 nucleotides. This origin is in the same relative genomic position and shows a high degree of DNA sequence homology to that of mouse mitochondrial DNA. In both cases, the DNA region within and immediately flanking the origin of DNA replication contains five tightly clustered tRNA genes. A major portion of the pronounced DNA template secondary structure at this origin includes the known tDNA sequences.     

Lagging strand synthesis in coordinated DNA synthesis by bacteriophage t7 replication proteins

The proteins of bacteriophage T7 DNA replication mediate coordinated leading and lagging strand synthesis on a minicircle template. A distinguishing feature of the coordinated synthesis is the presence of a replication loop containing double and single-stranded DNA with a combined average length of 2600 nucleotides. Lagging strands consist of multiple Okazaki fragments, with an average length of 3000 nucleotides, suggesting that the replication loop dictates the frequency of initiation of Okazaki fragments. The size of Okazaki fragments is not affected by varying the components (T7 DNA polymerase, gene 4 helicase-primase, gene 2.5 single-stranded DNA binding protein, and rNTPs) of the reaction over a relatively wide range. Changes in the size of Okazaki fragments occurs only when leading and lagging strand synthesis is no longer coordinated. The synthesis of each Okazaki fragment is initiated by the synthesis of an RNA primer by the gene 4 primase at specific recognition sites. In the absence of a primase recognition site on the minicircle template no lagging strand synthesis occurs. The size of the Okazaki fragments is not affected by the number of recognition sites on the template.     

Synthesis of plus strands of retroviral DNA in cells infected with avian sarcoma virus and mouse mammary tumor virus

The vast majority of plus strands synthesized in quail cells acutely infected with avian sarcoma virus were subgenomic in size, generally less than 3 kilobases (kb). A series of discrete species could be identified after agarose gel electrophoresis by annealing with various complementary DNAs, indicating specificity in the initiation and termination of plus strands. The first plus strand to appear (within 2 h postinfection) was similar in length to the long redundancy at the ends of linear DNA (0.35 kb), and it annealed with complementary DNAs specific for the 3' and 5' termini of viral RNA (Varmus et al., J. Mol. Biol. 120:50-82, 1978). Several subgenomic plus-strand fragments (0.94, 1.38, 2.3, and 3.4 kb) annealed with these reagents. At least the 0.94- and 1.38-kb strands were located at the same end of linear DNA as the 0.35-kb strand, indicating that multiple specific sites for initiation were employed to generate strands which overlapped on the structural map. We were unable to detect RNA liked to plus strands isolated as early as 2.5 h postinfection; thus, the primers must be short (fewer than 50 to 100 nucleotides), rapidly removed, or not composed of RNA. To determine whether multiple priming events are a general property of retroviral DNA synthesis in vivo, we also examined plus strands of mouse mammary tumor virus DNA in chronically infected rat cells after induction of RNA and subsequent DNA synthesis with dexamethasone. In this case, multiple, discrete subgenomic DNA plus strands were not found when the same methods applied to avian sarcoma virus DNA were used; instead, the plus strands present in the linear DNA of mouse mammary tumor virus fell mainly into two classes: (i) strands of ca. 1.3 kb which appeared early in synthesis and were similar in size and genetic content to the terminally repeated sequence in linear DNA; and (ii) plus strands of the same length as linear DNA. A heterogeneous population of other strands diminished with time, was not found in completed molecules, and was probably composed of strands undergoing elongation. These two retroviruses thus appear to differ with respect to both the number of priming sites used for the synthesis of plus strands and the abundance of full-length plus strands. On the other hand the major subgenomic plus strand of mouse mammary tumor virus DNA (1.3 kb) is probably the functional homolog of a major subgenomic plus strand of avian sarcoma virus DNA (0.35 kb). The significance of this plus strand species is discussed in the context of current models which hold that it is used as a template for the completion of the minus strand, thereby generating the long terminal redundancy.     

Asymmetric Okazaki piece synthesis during replication of simian virus 40 DNA in vivo.

We have pulse-labeled simian virus 40 (SV40)-infected monkey cells with ³H-thymidine (³H-dThd) and have hybridized the viral Okazaki pieces (rapidly labeled short DNA chains found during DNA replication, < 250 nucleotides long) and SV40 “intermediate sized” DNA (longer nascent strands, up to full replicon size) to the separated strands of two SV40 DNA restriction fragments, one lying to either side of the origin of bidirectional DNA replication. As much as 5 fold more Okazaki piece DNA hybridized to one strand than to the other strand of each restriction fragment. The excess Okazaki piece DNA was in the strands oriented 3′ → 5′ away from the replication origin (the strands which are expected to be synthesized discontinuously). Neither the duration of the labeling period nor the temperature of the cells during labeling significantly altered this hybridization asymmetry. With respect to the hybridization of “intermediate sized” DNA, a reverse asymmetry was detected (1.7 fold more radioactivity in the strands oriented 5′ → 3′ away from the origin for a 1 min pulse label at 22°C). The effects on these hybridization asymmetries of preincubating the infected cells with FdUrd prior to pulse-labeling were also determined.We also measured the size of the Okazaki pieces using gel electrophoresis under denaturing conditons after releasing the pieces from the filter-bound DNA strands. The size distribution of the Okazaki piece DNA from each strand was the same (～ 145 nucleotides, weight average; 200–250 nucleotides, maximum size), indicating that the hybridization asymmetry resulted from a difference in the number rather than the size of the pieces in each strand.The simplest interpretation of our results is that SV40 DNA is synthesized semidiscontinuously: the strand with 3′ → 5′ orientation away from the origin is synthesized in short Okazaki pieces which are subsequently joined together, while the strand with 5′ → 3′ orientation away from the origin is synthesized continuously. Some models of two-strand discontinuous synthesis, however, cannot be ruled out.     

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

Examination of in vivo long-labeled, pulse-labeled and pulse-chase-labeled mitochondrial DNA has corroborated and extended the basic elements of the displacement model of replication. Mitochondrial DNA molecules are shown to replicate an average of once per cell doubling in exponentially growing cultures. Analysis of the separate strands of partially replicated molecules indicates that replication is highly asynchronous with heavy-strand synthesis preceding light-strand synthesis. Native and denatured pulse-labeled replicating molecules exhibit sedimentation properties predicted by the displacement model of replication. Pulse-label incorporated into molecules isolated in the lower band region of ethidium bromide/cesium chloride gradients is found primarily in heavy daughter strands. Pulse-label incorporated into molecules isolated in the upper band region is found primarily in light daughter strands. The results of a series of pulse-chase experiments indicate that the complete process of replication requires approximately 120 minutes. Both daughter molecules are shown to segregate in an open circular form. They are then converted to closed circular molecules having a superhelix density near zero. After closure, the 7 S heavy-strand initation sequence is synthesized, and this process is accompanied by nicking, unwinding and closing of at least one of the parental strands resulting in the formation of the D-loop structure. The 7 S heavy-strand initiation sequence of the D-loop structure is not stable and turns over with a half-life of 7·9 hours. We suggest that all in vivo forms of parental closed circular mitochondrial DNA have superhelix densities of near zero, and that the previously observed superhelix density of closed circular mitochondrial DNA, σ~ ?0·02, results from the loss of the 7 S heavy-strand initiation sequence from D-loop mitochondrial DNA molecules during isolation.     

Precise nucleotide location of the 5' ends of RNA-primed nascent light strands of mouse mitochondrial DNA

The mouse mitochondrial DNA origin of light-strand replication has been defined as a 32-nucleotide region located among five transfer RNA genes in the genomic sequence. A distinctive feature of this origin is its potential to form a perfectly complementary stem and 11-nucleotide loop structure. Previous studies have demonstrated that the 5′ ends of nascent light strands map within this region and a major trinucleotide ribosubstitution site in closed circular mouse mitochondrial DNA has been mapped within the stem sequence.Direct analysis and precise localization of the 5′ ends of nascent light strands indicate that essentially all 5′ ends are ribonucleotides mapping in the originspecific dyadic structure. The major 5′ end identified is the rG at position 5187 in the genomic sequence. Priming of replication most likely occurs within the loop portion of the potential dyad and continues for 2 to 16 nucleotides with a sharply defined switch to deoxyribonucleotide synthesis. This functional transition point is identical in map position to the trinucleotide ribosubstitution site in mature, closed circular mitochondrial DNA.     

Mechanism of mitochondrial DNA replication in mouse L-cells: localization and sequence of the light-strand origin of replication

In the novel replication mechanism of closed circular mouse L-cell mitochondrial DNA synthesis one strand of the duplex (the heavy-strand) is initiated at a defined origin and proceeds unidirectionally. Synthesis of the complementary light-strand is initiated at a different origin, located approximately two-thirds genome length from the heavy-strand origin, and also proceeds unidirectionally. The initiation of light-strand synthesis does not occur until synthesis of the heavy-strand has extended past the light-strand origin region. One intriguing consequence of this asynchrony is that the heavy-strand origin functions in a DNA duplex, while the light-strand origin functions as a single-stranded template. In order to obtain the precise location of the light-strand origin we have isolated replicative molecules in which light-strand synthesis has begun and subjected them to digestion by a combination of the single-strand specific nuclease S₁ and various restriction cndonucleases. By comparison of the sizes of the duplex fragments thus generated with those produced by cleavage of non-replicating molecules cleaved with the same enzymes we have located the 5′-end of daughter light-strands at a position 55 to 90 nucleotides from a HpaI cleavage site 0.67 genome length from the heavy-strand origin. The nucleotide sequence of a 318-base region surrounding this site, determined by chemical sequencing techniques, possesses the symmetry required for the formation of three hairpin loops. The most striking of these has a stem consisting of 12 consecutive basepairs and a 13-base loop. In the heavy-strand template, this loop contains 11 consecutive thymidine nucleotides. This light-strand origin region has been found to possess a remarkable degree of homology with several other prokaryotic and eukaryotic origin-related sequences, particularly those of the øX174 A region and the simian virus 40 EcoRII G fragment.It has previously been shown that mouse mitochondrial DNA contains alkali-labile sites, which are presumably due to the presence of ribonucleotides incorporated into the DNA. A cluster of sites, representing eight adjacent ribonucleotides, has been located in mature light strands at or near the origin of light-strand synthesis. The retention of ribonucleotides at this specific location may reflect inefficient removal of an RNA primer at the light-strand origin.     

DNA synthesis in a mitochondrial lysate of Xenopus laevis oocytes. H strand replication in vitro

Conditions for efficient replication in vitro of mitochondrial DNA L strand into H strand products have been established. Gel electrophoresis and hybridization analyses of the products show that neosynthesized H strands are progressively elongated from the D-loop region, and some of them are synthesized as full-length molecules. Evidence for initiation of these H strands de novo is presented. In contrast, there is no detectable L strand synthesis in vitro in this system. This may prove useful for analyzing the distinct molecular mechanisms operating at OH and OL. Use of specific inhibitors indicates that DNA synthesis in the mitochondrial lysate in vitro requires DNA polymerase gamma. These observations support the conclusion that replication in vitro in this system closely resembles the first steps of mitochondrial DNA replication in vivo.     

In vitro replication of heavy strand DNA in permeabilized human mitochondria.

We have characterized some of the experimental conditions that are essential for initiation of human mitochondrial DNA synthesis. Mitochondria were purified from HeLa cells and were permeabilized with Triton X-100. When supplied with rNTPs and dNTPs, the permeabilized mitochondria synthesized nucleic acids that ranged in size from about 600 to 2000 nucleotides. In vitro DNA synthesis occurred on endogenous DNA templates and required a continuous supply of ATP. Analyses of the synthetic products revealed that almost all of them were of heavy-strand sequence and included authentic 7S DNA. Most of the synthetic products had 5' ends that mapped to similar locations as those previously identified for nascent heavy-strand DNA. Identification of these parameters should facilitate our efforts to achieve in vitro replication of heavy-strand mitochondrial DNA.     

Template-directed arrest of mammalian mitochondrial DNA synthesis.

Mammalian mitochondrial DNA often contains a short DNA displacement loop at the heavy-strand origin of replication. This short nascent DNA molecule has been used to study site-specific termination of mitochondrial DNA synthesis in human and mouse cells. We examined D-loop strand termination in two distantly related artiodactyls, the pig and the cow. Porcine mitochondrial DNA was unique among mammals in that it contained only a single species of D-loop single-stranded DNA. Its 3' end mapped to a site 187 nucleotides from the 5' end of the proline tRNA gene. This site was 21 and 47 nucleotides 5' to two very similar sequences (5' ACATATPyATTAT 3') which are closely related to the human and mouse termination-associated sequences noted by Doda et al. (J. N. Doda, D. T. Wright, and D. A. Clayton, Proc. Nat. Acad. Sci. USA 78:616-6120, 1981). Bovine mitochondrial DNA contained three major D-loop DNA species whose 3' ends mapped to three different sites. These sites were not found in the porcine sequence. However, the bovine termination sites were located 60 to 64 base pairs 5' from sequences which were also very similar to the termination-associated sequences present in pigs and other mammals. These results firmly establish the concept that arrest of heavy-strand DNA synthesis is an event determined, at least in part, by template sequence. They also suggest that arrest is determined by sequences which are a considerable physical distance away from the actual termination site.