A mitochondrial DNA primase from the trypanosomatid Crithidia fasciculata.

We have purified to near homogeneity a DNA primase from a mitochondrial fraction of the trypanosomatid Crithidia fasciculata. The enzyme is a single polypeptide chain of 28 kDa. Using a poly(dT) template and ATP as a substrate, the enzyme makes oligonucleotides of which the vast majority are about 10 nucleotides in size or smaller. With a single-stranded M13 DNA template and the four rNTPs as substrates, the enzyme makes heterogeneous oligonucleotides in the same size range. These oligonucleotides efficiently prime the synthesis of DNA by the Klenow DNA polymerase. Immunolocalization with antibodies against the purified enzyme confirms that the primase is mitochondrial. Furthermore, the enzyme localizes to specific regions of the cell's single mitochondrion, above and below the condensed kinetoplast DNA. The primase does not co-localize with the mitochondrial topoisomerase II and DNA polymerase beta, both of which are associated with two protein complexes positioned on opposite sides of the kinetoplast disc. These localization studies have significant implications for the mechanism of kinetoplast DNA replication.

inserted into or deleted from the sequence, thus forming an open reading frame. Minicircles encode small guide RNAs, which control the specificity of editing (see Refs. 4 and 5 for reviews on editing).
The network structure of kDNA must require unusual mechanisms for its replication. For example, minicircles do not replicate while topologically linked to the network, but instead they are released from the network by a topoisomerase. After release they replicate as free minicircles, using a mechanism, and the progeny minicircles are then reattached to the network periphery. When all minicircles have replicated, the network has doubled in size. Then it splits in two, and the two progeny networks segregate into daughter cells during cell division (see Refs. 3, 6, and 7 for reviews).
A current goal in our laboratory is to study proteins and enzymes that are involved in the maintenance and replication of kDNA. These proteins are interesting not only in their own right, but also because in at least some cases they are assembled into mitochondrial protein complexes, which can be easily visualized by immunofluorescence. For example, both a topoisomerase II (8) and a DNA polymerase ␤ (9) co-localize within two discrete protein complexes (each roughly 0.4 m in diameter), which are situated in antipodal positions at the edge of the kinetoplast disc (the disc is approximately 1 m in diameter and 0.4 m thick). Since minicircles thought to be replication intermediates (as visualized by fluorescence in situ hybridization) are also present in these two complexes (9), it is likely that these structures are involved in minicircle replication (see Ref. 3

for review).
In this paper we report the purification of a C. fasciculata mitochondrial DNA primase, the enzyme responsible for making short RNA molecules that initiate the synthesis of a DNA strand. We found that the primase is a single polypeptide of 28 kDa, and we describe some of its enzymatic properties. Using immunofluorescence, we found unexpectedly that the primase does not co-localize with the topoisomerase II and the DNA polymerase ␤ in the two protein complexes that flank the kinetoplast disc. Instead it localizes above and below the disc. This finding has significant implications for our understanding of the mechanism of kDNA replication.
Primase Activity Assay-A coupled assay was used to follow the enzyme during purification (10). In this assay, using ATP as a substrate, the primase synthesized non-radioactive oligo(A) primers complementary to a poly(dT) template. The primers were then extended by Klenow DNA polymerase and [␣-32 P]dATP, and acid-insoluble radioactivity was measured. The 25-l reaction mixture contained 50 mM Tris-HCl, pH 7.5, 50 mM NaOAc, 5 mM MgCl 2 , 0.1 mg/ml BSA (the 10 mg/ml stock solution had been heated at 60°C for 15 min), 5% (v/v) glycerol, 1 mM DTT, 0.5 g of poly(dT), 1 mM ATP, 50 M [␣-32 P]dATP (about 2400 cpm/pmol), 0 -0.5 unit of primase, and 0.5 unit of Escherichia coli DNA polymerase I large fragment (New England Biolabs). After 30 min at 30°C, products of the reaction were precipitated by sequential addition of 10 l of BSA (10 mg/ml), 10 l of 0.1 M sodium pyrophosphate, 35 l of H 2 O, and 20 l of 50% (w/v) trichloroacetic acid. The acid-insoluble product was collected onto GF/C filters (Whatman) under vacuum and washed, first with 0.1 M HCl and 0.1 M sodium pyrophosphate, and then with 95% (v/v) ethanol. Radioactivity on the filters was measured by liquid scintillation counting. One DNA primase unit is defined as the amount required for conversion of 1 pmol of [ 32 P]dAMP/h into an acid-insoluble form under assay conditions. The activity is linear with enzyme concentration in the range of 0 -1 unit. The standard buffer conditions and temperature (30°C) were used in all primase reactions described in this paper.
Large Scale Growth of C. fasciculata and Preparation of a Mitochondrial Fraction-Parasites were grown in a 150-liter Fermatron fermenter (in the Department of Biochemistry, Johns Hopkins School of Hygiene and Public Health) at 26°C in medium containing 1.8% Deltown AE80M peptone, 0.45% yeast extract, 0.45% NaCl, 0.9% glucose, and 10 mg/liter hemin (11). Alternatively, they were grown at room temperature in 6-liter flasks (each containing 4 liter of medium) with vigorous shaking (200 rpm). Cells at a density of about 4 ϫ 10 7 /ml were harvested by centrifugation either using a Sharples continuous flow centrifuge or a Sorvall GS3 rotor (5000 rpm, 10 min, 4°C). They were washed in STE buffer (250 mM sucrose, 50 mM Tris-HCl, pH 7.5, 1 mM EDTA) (12). The yield of cells was 2-3 g (wet weight)/liter of culture. Two procedures, both conducted at 0 -4°C, were used for isolation of mitochondria. In Method A (modified from a previous procedure; Ref. 12), the cells (220 g, wet weight) were suspended in 1200 ml of cell disruption buffer (STE buffer supplemented with 0.15 M KCl and 5 mM MgCl 2 ). The cell suspension (in 600-ml aliquots) was transferred to a Parr cell disruption bomb. The bomb was pressurized with nitrogen to 1800 -2000 p.s.i. for 30 min, while the contents were stirred with a magnetic bar. The cells were disrupted as the suspension was released to atmospheric pressure through the discharge valve. To maintain the pressure through the discharge process, it was necessary to stop the flow, restore the nitrogen pressure to 1800 -2000 p.s.i., and then resume the discharge. Microscopy revealed that more than 95% of cells were disrupted. Fluorescence microscopy, after staining with DAPI (1 g/ml) to visualize the kDNA, showed that most of the mitochondria appeared as brightly fluorescent dots, indicating that the membranes enclosing the kDNA were still intact. To isolate a mitochondrial fraction, the suspension was centrifuged (Sorvall GS-3 rotor, 9000 rpm, 20 min, 4°C); the pellet was resuspended in 200 ml of cell disruption buffer and centrifuged again under the same conditions. This washing procedure was repeated four times. In Method B (modified from a scheme developed by Joseph Shlomai, Hebrew University), 2 the cells (450 g, wet weight) were suspended in 2000 ml of ice-cold 25 mM Tris-HCl, pH 7.5, 1 mM EDTA and then transferred to the Parr cell disruption bomb. The bomb was pressurized to 1000 p.s.i. for 30 min, while stirring with a magnetic bar. The cells were released from the discharge valve, which was connected to an 18-gauge needle. Microscopy indicated that almost all cells were broken and most mitochondria were intact. The cell homogenate was centrifuged (Sorvall GS-3 rotor, 2000 rpm, 5 min, 4°C) to remove unbroken cells, and then the supernatant was collected and centrifuged again (Sorvall GSA rotor, 12,500 rpm, 30 min, 4°C). The pellets from the second centrifugation, containing mitochondria, were pooled and used for the enzyme purification. Method B gives a better recovery of mitochondria, although it is probably a more crude preparation.
Small Scale Growth of C. fasciculata and Percoll Purification of Mitochondria-Cells were grown to about 4 ϫ 10 7 cells/ml with vigorous shaking (200 rpm) at room temperature in 8 liters of 3.7% (w/v) Brain Heart Infusion supplemented with 20 g/ml hemin using two 6-liter flasks. The cells were harvested as described above and disrupted in a Parr cell disruption bomb using Method A. After centrifugation of the lysate to sediment the mitochondria (Sorvall GSA rotor, 12,500 rpm, 30 min, 4°C), the pellet was washed twice with 200 ml of cell disruption buffer. The final pellet was gently resuspended with 125 ml of 100% Percoll (Pharmacia) and 125 ml of cell disruption buffer by means of a Dounce homogenizer with loose pestle (type B). The Percoll suspension was evenly divided into 10 tubes and centrifuged (Beckman Ti 50.2 rotor, 26,000 rpm, 45 min, 4°C), generating Percoll concentration gradients. In this gradient the mitochondrial fraction formed a turbid band about 1 cm thick, at about 2/3 of the distance from the top of each tube.
Crude mitochondrial lysate (Fraction 1, 500 ml) was loaded onto a 50-ml Q-Sepharose (Pharmacia) column at a flow rate of 1 ml/min. DNA primase activity does not bind to this column; however, this step was essential for removal of nucleic acids in the lysate. The flow-through fraction (Fraction 2, 450 ml) was loaded onto a 50-ml S-Sepharose (Pharmacia) column at 2 ml/min, which was washed with 250 ml of buffer A, then with 250 ml of 0.2 M KCl in buffer A, and finally with 300 ml of 0.5 M KCl in buffer A. DNA primase activity was eluted at the 0. The active fractions (Fraction 6, 1 ml) were dialyzed against buffer A containing 40% glycerol to reduce the volume to about 300 l. The final chromatographic procedure was gel filtration on a 30-ml Superose 12 FPLC column (Pharmacia) using buffer A at a flow rate of 0.5 ml/min. The active fractions were pooled and dialyzed against buffer A containing 40% glycerol (v/v) and stored at Ϫ80°C (Fraction 7, 0.8 ml).
In a second purification, we used a mitochondrial fraction isolated by Method B. The purification steps were identical to those described in the previous paragraph with the following exceptions: 1) a 150-ml phosphocellulose column, eluted with a 1200-ml gradient from 0 to 1 M KCl in buffer A, substituted for the DNA cellulose column; 2) column sizes (except for the Poros HS column) and volumes of elution buffers were increased proportionally to the amount of protein; 3) a 10-ml phenyl-Sepharose column substituted for the phenyl-Superose, and this column was run before the Poros HS column, and 4) the final Superose 12 FPLC step was omitted because the protein was pure after Poros HS chromatography.
Antibodies-Rabbit antibody to mitochondrial DNA polymerase ␤ was a gift from Dr. Al Torri (13). Antibody to the DNA primase was prepared by immunizing female BALB/c mice by intraperitoneal injections with purified DNA primase (Fraction 7, 2-5 g/injection). The initial inoculations were in Freund's complete adjuvant, and four subsequent boosts were at 3-week intervals with Freund's incomplete adjuvant. The serum (at 1:1000 dilution) was screened by Western blot for specific recognition of the 28-kDa protein. The antibody recognizes the homogeneous primase on a Western blot (Fig. 1, lane 12; see lane 10 for Coomassie-stained gel) and also recognizes a polypeptide of the same size in a preparation of isolated mitochondria (Fig. 1, lane 11; see lane 9 for Coomassie-stained gel). Preincubation of 5 l of primase antiserum with 5 l of primase solution (30 min, 4°C) resulted in 70% loss of primase activity in the standard assay. A control experiment with DNA polymerase ␤ antiserum resulted in loss of only 15% of primase activity. In an immunodepletion assay, antiserum bound to protein A-Sepharose beads was able to deplete 74% of the primase activity. In a control experiment with DNA polymerase ␤ antiserum, only 10% of the primase activity was depleted.
poly-L-lysine (Sigma, 0.1%). The slides were kept in a humidity chamber for 20 min to allow cells to adhere prior to a wash with PBS. The cells were first fixed with 2% paraformaldehyde (2 min at room temperature), a treatment followed by fixation in 100% methanol (overnight at Ϫ20°C). The fixed cells were washed with PBS and then treated with 20 mM glycine in PBS for 10 min to neutralize residual aldehyde groups. After washing again with PBS, the slides were treated for 1 h at room temperature with 10% (w/v) BSA, 0.5% (v/v) Tween 20 in PBS (blocking solution), and then incubated in a humidity chamber for 1 h at room temperature with mouse anti-primase serum diluted 1:250 in blocking solution. After washing in PBS, they were incubated for 1 h at room temperature with fluorescein isothiocyanate-conjugated goat antimouse IgG (Boehringer Mannheim) diluted 1:250 in blocking solution. The slides were washed in PBS and then in PBS with DAPI (0.1 g/ml). After washing with PBS for another 10 min, the slides were then mounted with Mowiol 4 -88 (Calbiochem) (14) containing p-phenylenediamine (1 mg/ml). To double-label the cells, the primary antibody reaction (1 h, room temperature) included rabbit antibody to DNA polymerase ␤ (1:250 dilution) and mouse antibody to DNA primase (1:250 dilution). The secondary antibody reaction (1 h, room temperature) included rhodamine-conjugated affinity-purified goat antibody to rabbit IgG (Boehringer Mannheim, 1:250 dilution) and fluoresceinconjugated affinity-purified goat antibody to mouse IgG (Boehringer Mannheim, 1:250 dilution). The slides were examined by fluorescence microscopy using a Zeiss Axioskop microscope, and the images captured on Kodak Tmax 400 film. The negatives were scanned with a Nikon 35-mm film scanner (LS-1000) and processed in Adobe Photoshop on a Macintosh computer. Table I presents a summary of the purification. Starting from the mitochondrial lysate supernatant (prepared by Method A), the DNA primase was purified 14,500-fold and the yield of activity was about 8%. Fig. 1 shows an analysis of Fractions 1-7 by SDS-PAGE and silver staining. Fraction 7 appears to contain only a single homogeneous protein of 28 kDa.

Purification of Primase-
The final yield of primase from mitochondria prepared by Method A was only about 10 g of protein (Table I). We considered the possibility that this mitochondrial isolation procedure had a low yield, due at least in part to the extensive washing of the mitochondrial fraction in Method A. We therefore did a second purification using a mitochondrial fraction isolated by Method B. In the preparation shown in Table I, we started with only 1750 mg of mitochondrial protein in a mitochondrial lysate prepared from 220 g of cells. Using Method B in a second purification, we started with 27,200 mg of mitochondrial protein from 450 g of cells. Using this procedure we obtained about 300 g of purified DNA primase in about 10% yield. Its specific activity was comparable to that shown in Table I. Fig. 1 (lane 8) shows an SDS-PAGE analysis of the final product. This most pure fraction, or the Fraction 7 enzyme in Table I, was used in all experiments described in this paper. The purified enzyme was stored at Ϫ80°C in buffer containing 25 mM Tris-HCl, pH 7.5, 100 mM KCl, 0.5 mM EDTA, 1 mM DTT, 40% glycerol. It maintains over 50% of its activity after 6 months of storage.
As shown in Fig. 2, the DNA primase activity (panel A) co-eluted from the Superose 12 column with the 28-kDa protein as indicated by the intensity of the silver-stained band on SDS-PAGE (panel B). Primase activity and the 28-kDa protein both peak in fraction 31. The 28-kDa protein also co-eluted with primase activity on the phenyl-Superose and Poros HS columns, and the ratio of activity to silver staining of the band was comparable to that of Fraction 7. These data provide strong evidence that the 28-kDa protein is responsible for activity. To assess the size of the native enzyme, we also compared its behavior in gel filtration (Superose 12 FPLC) with that of two reference single-chain proteins, carbonic anhydrase (29 kDa) and BSA (67 kDa). The elution position of the DNA primase was the same as that of carbonic anhydrase (see legend of Fig. 2A), providing strong evidence that the primase is a 28-kDa monomer.
Requirements for DNA Primase Activity-Maximal primase activity requires ATP and a divalent cation (Mg 2ϩ ) ( Table II). The enzyme may have a tightly bound metal ion, as 21% of maximal activity is detectable in the absence of added Mg 2ϩ . However, EDTA inhibits activity completely when added at a concentration equivalent to that of the divalent cation. Rifampicin and N-ethylmaleimide inhibit the activity only weakly. To determine whether in our standard assay (using a poly(dT) template) the incorporation of [ 32 P]dAMP into an acid-insoluble form depends on the activity of primase, we varied the ATP b These values may have significant error because of the low concentration of protein available.

FIG. 1. SDS-PAGE analysis of DNA primase purification and
Western blot to demonstrate specificity of the anti-primase antibody. Aliquots of each fraction were electrophoresed on a 12% SDSpolyacrylamide gel, which was then stained with silver (lanes 1-8) or Coomassie Blue (lanes 9 and 10). Lanes 1-7, fractions from the purification summarized in Table I were identical to lanes 9 and 10 but DNA primase was detected on a Western blot using anti-DNA primase antibody. The mitochondrial preparation used in lanes 9 and 11 was prepared by Michele Klingbeil and Tina Saxowsky using a new unpublished method involving cell breakage by the Stanstead Cell Disruptor and Percoll purification of the mitochondria. For Western blotting, proteins were transferred to PVDF Western blotting membranes (Boehringer Mannheim). The membrane was first blocked with 20% horse serum in TBS buffer (25 mM Tris-HCl, pH 7.5, 50 mM KCl, 0.5 mM EDTA) for 1 h and then incubated for 1 h at room temperature with anti-primase antibody diluted 1:1000 in blocking buffer. After washing three times for 10 min with TBS and 0.05% Tween 20, the membrane was incubated for 1 h at room temperature with 1:1000 diluted anti-mouse IgG monoclonal antibody conjugated with alkaline phosphatase (Boehringer Mannheim). After washing three times for 10 min with TBS containing 0.05% Tween 20, the blot was developed with bromochloroindolyl phosphate/nitro blue tetrazolium. The size markers on the left of the figure refer to lanes 1-7. concentration from 0 to 4 mM. As shown in Fig. 3A, there was no incorporation in the absence of ATP and incorporation was maximal at 1 mM (closed symbols). In a control experiment (open symbols), there was little effect of ATP on the incorporation of [ 32 P]dAMP when primase was omitted and oligo(dA) was used as a primer. In a related experiment we used singlestranded M13 DNA as a template (Fig. 3B). Again there was dependence upon the presence of rNTPs (compare lower line to upper line), and the magnitude of [ 32 P]dAMP incorporation depended upon the primase concentration.
The Products of the DNA Primase Reaction-To characterize the products, we conducted reactions (for times up to 45 min) containing poly(dT) template, [␣-32 P]ATP as substrate, and DNA primase. Analysis of the reaction products on a 20% polyacrylamide gel revealed a ladder of fragments ranging up to about 10 nucleotides in size (Fig. 4A). However, with a long exposure of the autoradiogram, there were very small amounts of larger oligonucleotides, up to about 15 nucleotides in size. We also found in a similar experiment using M13 DNA as a template and the 4 rNTPs that the primase synthesized a very heterogeneous population of oligonucleotides in the same size range (data not shown).
We next used a chase to determine whether the oligonucleotides synthesized by primase can prime DNA synthesis (Fig.   4B). We first synthesized primers using [␣-32 P]ATP as a substrate (lane 1), and the products are the characteristic ladder of fragments up to about 10 nucleotides in size. We then added Klenow DNA polymerase and increasing concentrations of dATP to the reaction. As shown in lanes 2-4, the primers were elongated and the products appeared in or near the slot. The products of DNA primase can also be elongated by the purified C. fasciculata mitochondrial DNA polymerase ␤ (15). However, because of the low processivity of this enzyme, the primer extensions occurred with much lower efficiency (data not shown).
Intracellular Localization of DNA Primase-Four lines of evidence support the mitochondrial localization of this primase. First, in a small scale purification (from about 25 g of cells), we used Percoll gradient-purified mitochondria to purify the primase through the phenyl-Superose fraction. The primase obtained from the purified mitochondria had activity and chromatographic behavior similar to that obtained from the crude mitochondrial fraction. 3 Second, the activity in the crude mitochondrial fraction was latent, requiring release from vesicles by treatment with a buffer containing 0.25% Nonidet P-40 and 0.5 M KCl. In a control assay in which the detergent/KCl treatment was omitted, the amount of detectable primase activity was only about 10% of that obtained with detergent/KCl. 3 Third, a Western blot of proteins from a mitochondrial fraction revealed a 28-kDa polypeptide that reacted with anti-primase antibody (Fig. 1, lanes 9 and 11). Finally, immunolocalization experiments proved conclusively that the primase is localized within the single mitochondrion of this parasite.
For immunolocalization we used a polyclonal antibody against the primase. The results indicate that the primase is indeed situated within the mitochondrion in discrete sites near the kDNA network. Fig. 5A shows two C. fasciculata cells visualized by Nomarski optics. Fig. 5B shows the same cells stained with DAPI, which brightly stains the kinetoplasts (the nuclei stain poorly under these conditions and are barely visible). The kinetoplast is a disc-shaped structure, oriented per-3 C. Li and P. T. Englund, unpublished data.   3. Requirement for rNTPs in the primase reaction. Panel A, primase was assayed using the standard coupled assay conditions (25 l) with a poly(dT) template, 1 unit of primase, and variable concentrations of ATP (solid line). In a control reaction, the primase was omitted and oligo(dA) 5 (Sigma) (1:10 ratio to poly(dT)) was used as a primer (dashed line). Panel B, primase was assayed using a similar coupled assay with a single-stranded M13 DNA template. The reactions (25 l) were identical to the standard assay conditions (see "Experimental Procedures"), but they contained 30 ng of single-stranded M13mp19 pendicular to the flagellum, and this image shows the edges of the discs. Fig. 5C shows the primase in the same cells, localized above and below the kinetoplast disc. Somewhat more fluorescence is observed on the side of the disc nearest the flagellum, a distribution observed in many cells. We were concerned that the images of primase and DAPI staining might have been misaligned and that both components of the primase fluorescence may have been on the same side of the kinetoplast disc. This possibility is highly unlikely, given that the two cells are oriented in opposite directions and that the distances between the DAPI stains and the primase stains in the two cells are identical. Therefore we conclude that the primase forms a sandwich-like structure around the kinetoplast disc. DNA primase is localized in the same positions in the majority of log phase cells (90% or more). In another experiment, Fig. 5

DISCUSSION
Starting with a mitochondrial fraction from C. fasciculata, we have purified a DNA primase to near homogeneity. The purified enzyme has a molecular mass of 28 kDa, and its active form is a single polypeptide chain. Assuming that the recovery of mitochondria used for purification was 100%, we can make a rough estimate that there are about 11,000 molecules of primase within the parasite's single mitochondrion. Using a poly(dT) template, the primase produces a ladder of homopolymeric products, with a maximum size of about 10 nucleotides. Using single-stranded M13 DNA as a template, the primase makes products in roughly the same size range (data not shown). However, due to sequence heterogeneity, the M13 products are not resolved into a uniform ladder by gel electrophoresis, a finding which implies that initiation on M13 occurs at multiple sites.
The primase products can be efficiently elongated by Klenow DNA polymerase in the presence of the appropriate dNTPs. However, when poly(dT) is used as a template, the only primers that are efficiently extended by Klenow polymerase are the 10-mers and, to a lesser extent, the 9-mers (see Fig. 4B, lanes  3 and 4). Shorter oligonucleotides are not elongated efficiently by the DNA polymerase. A similar effect has been observed with mammalian primases (16). A chase experiment, in which non-radioactive ATP was added after synthesis of primers with [␣-32 P]ATP, did not result in the shorter oligonucleotides being extended (data not shown). It is possible that many of the shorter oligonucleotides had dissociated from the dT template and therefore could not serve as intermediates in primer synthesis. In another experiment, we found that the primers were poorly extended by the C. fasciculata mitochondrial DNA polymerase ␤, a result not surprising given that that enzyme may not be the major replicative enzyme in this organelle (see below).
Intracellular enzyme localization studies have been valuable in clarifying our understanding of kDNA replication. The C. fasciculata kinetoplast system is ideal for these studies because the non-replicating cell has only one kDNA network, which resides within its single mitochondrion. The network in vivo is condensed into a characteristic disc-shaped structure, about 1 m in diameter and 0.4 m thick, well within the resolution of fluorescence microscopy. Previous immunolocalization studies had demonstrated that a mitochondrial topoisomerase II (8) and a mitochondrial DNA polymerase ␤ (9) co-localize within two protein complexes, which are situated in antipodal positions adjacent to the kinetoplast disc (see localization of DNA polymerase ␤ in Fig. 5G). Based on the presence of two enzymes involved in replication, and other evidence (9), we had hypothesized that these two protein complexes are involved in minicircle replication. Another topoisomerase II (7), as well as some histone-like DNA binding proteins (17), localize within the kinetoplast disc. An hsp70 heat shock protein surrounds but does not appear to penetrate the kinetoplast disc (18,19). We now report that the primase has a distinct localization, being situated both above and below the kinetoplast disc. Little if any appears to flank the edge of the disc or is associated with the two complexes containing DNA polymerase ␤ and topoisomerase II. We have not yet determined whether the primase covers the entire upper and lower surfaces of the kinetoplast disc or if it covers only part of these surfaces. We also do not know if the enzyme is completely excluded from the kinetoplast disc or if it penetrates the upper and lower regions of the disc. Studies at higher resolution, involving confocal fluorescence microscopy or immunoelectron microscopy, will be needed to address these issues.
What is the significance of the primase localization to the mechanism of minicircle replication? A current view of this mechanism is shown in Fig. 6. The diagram shows a section through the kinetoplast disc, with interlocked minicircles forming a monolayer (see Ref. 3 for further information about this arrangement and about the replication scheme). Covalently closed minicircles are released from the central region of the network by a topoisomerase II, which is situated within the disc. Ultimately these free minicircles migrate to one of the two protein complexes (containing topoisomerase II and DNA polymerase ␤) that flank the kinetoplast disc. The newly replicated progeny minicircles, containing gaps, are attached to the network periphery adjacent to these protein complexes, in another topoisomerase reaction (20). The localization of the primase above and below the disc raises the possibility that replication may not actually occur within the antipodal protein complexes. Instead, minicircle replication could occur in the region above or below the kinetoplast disc. A replicative DNA polymerase in this location (not yet discovered but possibly related to DNA polymerase ␥) could have the major responsibility for DNA synthesis. The minicircle progeny could then migrate to one of the two protein complexes where many of the gaps in the discontinuously synthesized strand could be repaired immediately prior to network attachment (21). The DNA polymerase ␤ is ideally situated to carry out this reaction, and it is known to have a preference for gap filling. The progeny minicircles could then be attached to the network periphery by topoisomerase II. In an alternative scheme, the covalently closed minicircle, immediately after release from the network, could associate with primase and other proteins to form a replication initiation complex. It could then migrate to one of the two antipodal protein complexes to complete its replication. Further studies, including the localization of additional replication enzymes, will be needed to distinguish between these possibilities. 3, shows a section through the kinetoplast disc, with interlocked minicircles aligned in a monolayer. The disc is flanked by two antipodal protein complexes containing DNA polymerase ␤ and topoisomerase II. Primase is localized above and below the kinetoplast disc. Covalently closed free minicircles are released from the central region of the disc. After replication, the progeny minicircles, containing gaps, are attached to the network periphery adjacent to the two protein complexes. The newly replicated minicircles are drawn in bold. See "Discussion" for further discussion of the role of primase in this process.