A Nicked Form of Kinetoplast DNA in Leishmania

The mitoehondrial DNA of the protozoan Leishmania tarentolae, known as kinetoplast DNA, contains thousands of minicircles linked in a two-dimensional network. When kinetoplast DNA from exponentially growing cells is centrifuged to equilibrium in a CsCllethidium bromide gradient, it is resolved into two discrete components, Form I and Form II. Nearly all of the minicircles in Form I networks are covalently closed and all of those in Form II networks are open. These forms are indistinguishable from each other when examined by electron microscopy and they appear identical when analyzed by gel electrophoresis after diges- tion with the restriction enzymes Hue III or Hpa II. However, Form II networks sediment roughly 50% faster than Form I networks on a neutral sucrose gradient, indi-cating that Form II networks are larger in size or more compact in conformation, or both. Analysis of denatured Form II DNA by sedimentation or electron microscopy indicates that nearly all of its minicircles have one or more interruptions in both strands. the majority of the Form II minicircles can be closed by DNA ligase, most of these interruptions must be nicks. Experiments with S, nuclease indicate that some small may also

From the Department of Physiological Chemistry, Johns Hopkins School of Medicine, Baltimore, Maryland 21205 The mitoehondrial DNA of the protozoan Leishmania tarentolae, known as kinetoplast DNA, contains thousands of minicircles linked in a two-dimensional network. When kinetoplast DNA from exponentially growing cells is centrifuged to equilibrium in a CsCllethidium bromide gradient, it is resolved into two discrete components, Form I and Form II. Nearly all of the minicircles in Form I networks are covalently closed and all of those in Form II networks are open. These forms are indistinguishable from each other when examined by electron microscopy and they appear identical when analyzed by gel electrophoresis after digestion with the restriction enzymes Hue III or Hpa II. However, Form II networks sediment roughly 50% faster than Form I networks on a neutral sucrose gradient, indicating that Form II networks are larger in size or more compact in conformation, or both. Analysis of denatured Form II DNA by sedimentation or electron microscopy indicates that nearly all of its minicircles have one or more interruptions in both strands. Since the majority of the Form II minicircles can be closed by DNA ligase, most of these interruptions must be nicks. Experiments with S, nuclease indicate that some small gaps may also exist in Form II minicircles. 5'-Terminal nucleotide analysis of Form II kinetoplast DNA does not suggest that the interruptions are at specific locations in the minicircles.
The significance of the two forms of kinetoplast DNA has not yet been determined, but it is possible that Form II is an intermediate in replication of this DNA.
Kinetoplast DNA is an extremely complex structure which is found in the mitochondria of trypanosomes and related protozoa (1). Its major structural components are minicircles, which in the case ofLeishmania tarentolae are only about 1000 base pairs in size (2). Nearly all of the minicircles are linked together in massive networks, and each network contains roughly 15,000 minicircles (3). There appears to be only one network/mitochondrion, and only one mitochondrion/cell Although it was originally assumed that all minicircles within a network are identical, recent investigations involving restriction enzyme analysis and reassociation kinetics have revealed that networks contain numerous types of minicircles which have nonidentical but probably related nucleotide sequences (4-11). Analysis of kDNA' with restriction enzymes and also by electron microscopy has recently revealed a second type of structural component in networks (5,12,13). This component, known as the maxicircle, accounts for less than 5% of the total kDNA and has a molecular weight of about 22 x 10" in kDNA of Crithidia luciliae and about 13 x 10" in kDNA of Trypanosoma brucei (8). Although nothing is known about the genetic function of maxicircles and minicircles, Borst and co-workers have speculated that maxicircles carry the genes which are characteristic of mitochondrial DNA in other eukaryotic species (8).
Despite these recent advances in our understanding of the sequence complexity of kDNA, there is still very little known about the mechanism by which the circular components of kDNA replicate, about the mechanism by which the networks grow in size during the period of DNA synthesis, and about the way in which the kDNA segregates to form two networks within the daughter cells during cell division. These processes cannot be understood until the replicative intermediates of kDNA are identified and their structures are determined. As an initial approach to these questions, we have investigated the structure of a form of kDNA which usually accounts for about one-quarter of the kDNA in dividing cells, but which is absent in stationary phase cells (14). These networks, known as Form II, contain minicircles which are nicked or gapped, whereas the major species of kDNA, known as Form I, contains minicircles which are covalently closed. An additional difference between these two types of kDNA is that Form II networks sediment faster than Form I networks. We describe here our comparison of Form I and Form II kDNAs. Some of our conclusions have already been published in preliminary form (10). with the inoculum (about 3 x lo9 cells/5 liters of medium). Cells were harvested by centrifugation when they had grown to a cell concentration between 5 x lO'/ml and 2 x lO*/ml. They were growing exponentially even at the latter density. kDNA was isolated by differential centrifugation of a cell lysate essentially as described by Simpson and Berliner (31,and  II is roughly one-half of the amount of Form I.3 The DNA of the experiments described in the following paragraphs is to which bands between Forms I and II is usually in the range of compare the properties of these two forms of kDNA. 15 to 20% of the total. Form I DNA contains minicircles which Electron Microscopy of Form I and Form II DNA -When are nearly all covalently closed, and Form II DNA contains examined by electron microscopy, both forms appear as netopen minicircles with either nicks or gaps (10,141. The purpose works which are typical of kinetoplast DNA (Fig. 2). Both are preparations contain occasional free linear molecules of random length which are not joined to the networks, but these are probably contaminating nuclear DNA. After examining many micrographs of several different preparations of kDNA, we are unable to detect any characteristic differences between Forms I and II. Unfortunately, the sensitivity of Leishmania networks, to breakage makes it difficult to examine the size of intact networks by this technique. The only way that the two forms can be distinguished by electron microscopy is by spreading the DNA in the presence of ethidium bromide (Fig.  2). In this case nearly all (>80%) of the minicircles in Form I, which are covalently closed, become tightly twisted. In contrast, virtually all of the minicircles in Form II, which contain single strand breaks, remain untwisted.

Restriction
Enzyme Analysis of Forms Z and II-Because the minicircles in kDNA networks are heterogeneous in nucleotide sequence (4-ll), gel electrophoresis of restriction enzyme digests reveals complex patterns of fragments. We compared digests of Forms I and II kDNA for the purpose of determining whether the same pattern of fragments is obtained from both forms. As shown in Fig. 3, the two ZZae III digests are similar if not identical, and the same result was obtained with Hpa II digests. Most of the fragments shown in the Hue III digests (Fig. 3) derive from minicircles However, electrophoresis of Hpa II digests on 1.4% agarose gels reveals three fragments with a combined molecular weight of about 20 x 10" and which presumably derive from maxicircles. These fragments are also identical in both forms of kDNA. These results support the possibility that the same nucleotide sequences are present in the two forms of kDNA and that the different classes of circles are present in the same ratio.

Sedimentation of Form Z and Form ZZ kDNA in Neutral Sucrose
Gradients-One of the most striking properties of kDNA is its enormous sedimentation coefficient (3, 27). After zone sedimentation of Form I kDNA in a neutral sucrose gradient, most of the DNA is found in a single peak which sediments about 2.5 times faster than T7 phage. (The s*~,,~ of the phage is 487 S (28X) This peak probably consists of intact networks, and the slower sedimenting shoulder probably contains fragments of networks (Fig. 4A). Similar results with Form I DNA have been reported by Simpson and Berliner (3). Zone sedimentation of Form II DNA reveals that it sediments in a broad peak almost 4 times faster than T7 phage and about 1.5 times faster than Form I DNA (Fig. 4B).

Sedimentation
of Form Z and Form ZZ kDNA in Alkaline Sucrose Gradients -Zone centrifugation of Form II kDNA in an alkaline sucrose gradient reveals that virtually all of the DNA sediments as short fragments which are equal to or less than about 1000 nucleotides in size (Fig. 5S). When Form I DNA is centrifuged under the same conditions, most of the DNA is found in the pellet, and only a small fraction (5 to 20% in several preparations) sediments together with fragments of about 1000 nucleotides (Fig. 5.4). The experiments in Fig. 5 were performed with kDNA which was both uniformly labeled with 'IH and 3'-terminally labeled with =P, and the distribution of these isotopes after centrifugation reveals several facts about the structure of the DNAs. With Form I DNA, no 52P is found in the major component which sediments to the bottom of the tube. This absence of :%P was expected if the rapidly sedimenting DNA consists of networks of covalently closed circles which are not susceptible to terminal labeling. All of the "'P in Form I DNA is found in the peak sedimenting in the position of a marker of about 1000 nucleotides and, since the 4 Unpublished observations. ratio of "lP/"H is constant across this peak, nearly all the fragments released from Form I DNA must be about this size. With Form II DNA, both isotopes are found only in fragments which sediment together with or slower than the 1000 nucleotide marker. The breadth of the "H peak and the increasing "P/"H ratio in the trailing shoulder indicate that the fragments derived from Form II DNA are heterogeneous in size with a maximum of about 1000 nucleotides.
The experiment in Fig. 5A confirms Simpson and Berliner's observation (31 that Form I networks are stable in alkali but, since some linear fragments of about 1000 nucleotides are released, a small fraction of the minicircles within these networks must contain either one single strand break or one such break in each strand. In contrast, nearly every minicircle in Form II DNA contains single strand breaks and, after denaturation in alkali, each of the fragments sediments independently. The appearance of the small fragments in the gradients in Fig. 5, A and B, is not due to the presence of alkali-labile bonds in the kDNA minicircles because similar results were found with kDNA which was denatured by boiling and then centrifuged on a neutral sucrose gradient. Results identical to those in Fig. 5 were also obtained with kDNA which had been 5'-terminally labeled with polynucleotide kinase by a procedure similar to that described under "Methods."

Single
Strand Interruptions in Form ZZ kDNA -In agreement with the findings from the alkaline sucrose gradients, electron microscopy of Form II kDNA after heat denaturation reveals that nearly all of the DNA was converted to short single strands. Analysis of 1151 molecules on four micrographs showed that most of these strands were linear, heterogeneous in size, and shorter than a kDNA minicircle (1091 molecules; see histogram in Fig. 6 for a representative distribution of lengths). A very small fraction of the DNA was minicircular (34 molecules), another small fraction included longer linear molecules of various sizes which may have originated in kDNA networks or which may have been nuclear DNA contaminants (24 molecules), and virtually none were oligomeric minicircles (2 molecules). Since only a small fraction of circular structures survive heat denaturation, it can be concluded that most minicircles within Form II networks have interruptions in both strands. Because of the heterogeneity in size of the short linear strands formed by alkali (Fig. 5Z3) or by heat denaturation (Fig. 61, most of the minicircular strands probably have multiple interruptions.

Effect
of Ligase on Form ZZ kDNA -To test whether the interruptions in the minicircle strands of Form II kDNA are nicks or gaps, we treated this DNA with Escherichia coli DNA ligase. This enzyme seals nicks in a duplex DNA molecule, provided the apposing strands contain a 3'-hydroxyl and a 5'phosphate (29). We assayed the extent of ligation of Form II kDNA by CsCllethidium bromide centrifugation and by electron microscopy of molecules spread in the presence of ethidium bromide. Neither assay detects individual acts of ligation, but reveals only the ligation of all interruptions within a given minicircle.
The centrifugation assay (Fig. 7) shows that ligased Form II kDNA bands in a broad peak at a position near that of Form I DNA. This result indicates that many of the minicircles in all of the networks were covalently closed by ligase and therefore every interruption in those minicircles is a nick with appropriate termini. When the same preparation of ligased Form II kDNA was examined by electron microscopy in the presence of ethidium bromide, we found that about 60% of the minicircles had been covalently closed by the ligase treatment.
The at the site of nicks, the rate is less than that at singlestranded regions (31, 32). The effect of this enzyme on kDNA was first measured using sedimentation in neutral sucrose gradients (Fig. 8). S, clearly has an effect on Form I networks as shown by the conversion of the radioactivity to a more slowly sedimenting form (Fig.  8A). However, it has a much more striking effect on Form II networks in that about 65% of the radioactivity is converted to a form which sediments at a rate equal to or less than that of the T7 phage marker ( Fig.   W. S, digests of kDNA were also analyzed by gel electrophoresis (Fig. 9). Undigested kDNA, in the form of networks, is unable to enter gels and remains trapped on the upper surface. When an S, digest of Form I DNA was electrophoresed on an agarose gel, nearly all of the DNA remained on top but a single minor component entered the gel (Fig. 9A) A sample of each solution (30 ~1) was diluted with 60 ~1 of water and 10 ~1 of a solution containing 20% glycerol, 1% sodium dodecyl sulfate, and 0.03% bromphenol blue. The samples were then placed on 1.4% agarose tube gels (12 x 0.6 cm). The gel buffer contained 40 mM Tris/HCl (pH 8.0),5 mM sodium acetate, 1 rnM EDTA, and 0.4 pg/ml of ethidium bromide. Electrophoresis was at 75 V at room temperature and was stopped when the dye had migrated about 10 cm to the position marked D. The gels were then illuminated with an ultraviolet light and photographed with a Polaroid camera. Gel 1 is the control sample and Gel 2 is the sample containing S,. The arrow indicates the position of the faint band in Gel 2. B, Form II kDNA was incubated in solutions (25 ~1) containing 0.3 pg of kDNA and variable amounts of S, nuclease. Samples 1 to 7 contained 0, 0.05, 0.13, 0.50, 1.25, 5, and 15 units, respectively.
After 15 min at 37", the samples were treated with 5 ~1 of 50 mM EDTA, 50 ~1 of water, and 10 ~1 of a solution containing 20% glycerol, 1% sodium dodecyl sulfate, and 0.03% bromphenol blue. They were electrophoresed as described above until the dye had migrated about 8 cm to the position marked D The band at about 1 cm in all of the gels is primarily nuclear DNA which contaminated this Form II preparation. tained in each case (Fig. 9B). Some DNA always remained on the upper surface of the gel, but a family of bands also electrophoresed into the gel. Although these fragments have not been eluted and identified, an examination of an S, digest by electron microscopy (see below, Table I) indicates that these bands probably contain minicircles, 1000 base pair linear fragments, minicircular dimers, trimers, and higher oligomers. A comparison of the gels of digests of Form II kDNA prepared with different levels of S, reveals that the extent of digestion is not proportional to the level of enzyme used (Fig. 9B). For example the bands in Gel 4 seem only slightly less intense than those in Gel 7, even though the amount of enzyme used in these two digests differed by a factor of 30. Therefore, the pattern seen in Gel 4 must represent a nearly limit digest. The nonlinearity between extent of degradation and enzyme concentration indicates that some sites on the Form II network are cleaved readily at low enzyme concentrations, but that other sites are cleaved much more slowly. One possible interpretation of this result is that the very sensitive sites are gaps, and the less sensitive sites are nicks. Table I indicates the types of fragments which exist in an S, digest of Form II kDNA as determined by electron microscopy. Roughly 62% of the DNA in the digest remains as large oligomers and small networks (some with several hundred minicircle units), and about 28% is in the form of linear molecules. These linear molecules are rather homogeneous in length and are roughly the same size as minicircles (see histogram in Fig. 10). The rest of the digest consists of minicircles and small oligomers. All of the minicircles in this S, digest, whether free or part of an oligomer or small network, must contain single strand breaks, as none of them appear twisted when examined by electron microscopy in the presence of ethidium bromide. Virtually no acid-soluble material is released from Form II kDNA by S, nuclease (Table I). DISCUSSION Form I and Form II kDNAs are indistinguishable by several criteria. Both are organized as networks which appear identical in electron micrographs (Fig. 2), and both are degraded to a similar if not identical collection of fragments by the restriction enzymes Hae III and Hpa II (Fig. 3). However, as determined by equilibrium centrifugation and electron microscopy in the presence of ethidium bromide ( Figs. 1 and 21, by sedimentation in an alkaline sucrose gradient (Fig. 51, and by electron microscopy after heat denaturation (Fig. 6), virtually every minicircle in Form II kDNA contains one or more interruptions in both of its strands. In contrast, almost all of the minicircles in Form I kDNA are covalently closed. A second difference between Form I and Form II kDNAs is the fact that Form II sediments at neutral pH about 1.5 times faster than Form I kDNA. This large difference in sedimentation velocity could be due either to a more compact conformation or to a larger size of Form II networks.
Most of the interruptions in Form II minicircles are nicks which are susceptible to ligase ( Figure 7). However, the resistance of some minicircles to this enzyme, as well as the susceptibility of some to cleavage by low concentrations of S, nuclease, suggests that some minicircles may contain gaps as well. If gaps do exist, they must be very small, because S, digestion did not release significant amounts of acid-soluble nucleotides. It is likely that there is no more than one gap, or several tightly clustered gaps, in most minicircles because the linear fragments produced by S, digestion are nearly all about the same size as a minicircle (Fig. 10). The terminal nucleotide analysis of Form II 5'-[32P]kDNA did not suggest that the interruptions are at unique sites, but because of the presence of multiple types of minicircles and of multiple nicks within many of the minicircles, there remains a possibility that some of the nicks are at specific locations.
Several facts make it unlikely that the nicks or gaps in Form II DNA are caused by nucleolytic activity during isolation of the DNA. First, lysis of nonradioactive cells by the standard procedure (3) in the presence of Form I ["HlkDNA caused no change in the banding pattern of this DNA in a CsCl/ethidium bromide gradient. Second, direct assay of nicking activity, in extracts ofLeishmania prepared by sonication, revealed insignificant levels of such an enzyme even under conditions which could be expected to favor its activity." Finally, the fact that Form II kDNA sediments faster than Form I kDNA (Fig. 4) would not be a likely consequence of nicking during the isolation procedure. Simpson and Berliner have already shown that nicking of Form I kDNA with pancreatic DNase causes a gradual shift of all networks to a lower density position in a CsCl/ethidium bromide gradient, but this treatment does not affect the velocity of sedimentation of the networks through a neutral sucrose gradient (3). Futhermore, closure of the nicks in minicircles in Form II networks with E. coli ligase does not significantly alter its sedimentation rate in neutral sucrose gradients.' These facts, together with the argument that a nuclease which introduced random nicks during the isolation procedure would not be expected to attack all of the circles within some networks and to leave other networks virtually unscathed, indicate that Form II is not an artifact of isolation, but instead is a natural component of Very few facts are known about the mechanism of replication of kDNA, but some of the available information is summarized in the following paragraph. The Leishmania tarento-Zae cell cycle takes about 10 to 12 h, and kDNA synthesis occurs during a period of 3 to 4 h prior to cell division (15). Density shift experiments have indicated that each minicircle is replicated by a semiconservative mechanism once during each cell cycle and that recombination between minicircles may occur as well (14,34). These observations are consistent with the finding that the surface area of networks isolated from synchronized cultures of Crithidia fusciculata or Leishmania tarentolae, as estimated by light microscopy, approximately doubles in size during the period of DNA synthesis (14). Pulse labeling of Crithidia fasciculata or Leishmania tarentolae kDNA, followed by light microscope autoradiography, has indicated that DNA synthesis occurs at two sites on the periphery of the network which are situated 180" apart (35). Longer pulses result first in labeling of the entire periphery and then in uniform labeling of the entire network (14). Analysis of pulse-labeled kDNA in a CsCl/ethidium bromide gradient reveals that it bands in the position of Form II kDNA or in an intermediate position between Forms I and II, but that during a chase of several hours it is converted into a species which bands together with covalently closed minicircles. No radioactivity is found in individual covalently closed minicircles isolated from sonicated pulse-labeled networks until after a chase of 3 to 4 h (35), which indicates that there is a delay of several hours between the synthesis of a minicircle and its covalent closure.
We can speculate that prior to replication all minicircles in a network are covalently closed, but that after each minicircle is J Nicking activity was assayed in extracts of exponentially growing cells which were prepared bv sonication at 0" (3.3 x lo* cells in 0.5 ml of 50 mM Tri&Ci (pH S.i), 5 mM 2-mercaptoethanol). Assay mixtures (100 ~1) contained 0.5 fig of Form I U3HlkDNA. extract containing 117 gg of protein, 50'mM TrislHCl (pH 8.1), 5 rnra 2mercaptoethanol, and 10 mM MgCl,. The protein/kDNA ratio used in this assay is more than half that which exists in uiuo. After 30 min at 37", the solution was deproteinized by Sarkosylipronase treatment (3) followed by phenol extraction. One sample was centrifuged to equilibrium on a CsCUethidium bromide gradient, and the DNA banded in a position which was not distinguishable from untreated Form I kDNA. A second sample was tested for acid solubility of the r3HlkDNA, and none (cl% of the total) was.