Complete Replication of Templates by Escherichia coli DNA Polymerase I11 Holoenzyme*

DNA polymerase I11 holoenzyme (holoenzyme) processively and rapidly replicates a  primed singlestranded DNA circle to produce a duplex with an interruption in the synthetic strand. The precise nature of this discontinuity in the replicative form (RF 11) and the influence of the 5’ termini of the DNA and RNA primers were analyzed in this study. Virtually all (90%) of the RF I1 products primed by DNA were nicked structures sealable by Escherichia coli DNA ligase; in 10% of the products, replication proceeded one nucleotide beyond the 5‘ DNA terminus displacing (but not removing) the 5’ terminal nucleotide. With RNA primers, replication generally went beyond the available single-stranded template. The 5’ RNA terminus was displaced by 1-5 nucleotides in 85% of the products; a minority of products was nicked (9%) or had short gaps (6%). Termination of synthesis on a linear DNA template was usually (85%) one base shy of completion. Thus, replication by holoenzyme utilizes all, or nearly all, of the available template and shows no significant 5’+3’ exonuclease action as observed in primer removal by the “nick-translation” activity of DNA polymerase I.

DNA polymerase I11 holoenzyme (holoenzyme) processively and rapidly replicates a primed singlestranded DNA circle to produce a duplex with an interruption in the synthetic strand. The precise nature of this discontinuity in the replicative form (RF 11) and the influence of the 5' termini of the DNA and RNA primers were analyzed in this study. Virtually all (90%) of the RF I1 products primed by DNA were nicked structures sealable by Escherichia coli DNA ligase; in 10% of the products, replication proceeded one nucleotide beyond the 5' DNA terminus displacing (but not removing) the 5' terminal nucleotide. With RNA primers, replication generally went beyond the available single-stranded template. The 5' RNA terminus was displaced by 1-5 nucleotides in 85% of the products; a minority of products was nicked (9%) or had short gaps (6%). Termination of synthesis on a linear DNA template was usually (85%) one base shy of completion. Thus, replication by holoenzyme utilizes all, or nearly all, of the available template and shows no significant 5'+3' exonuclease action as observed in primer removal by the "nick-translation" activity of DNA polymerase I. DNA polymerase I11 holoenzyme, the multisubunit replicative enzyme of Escherichia coli, rapidly and processively converts a primed, single-stranded DNA circle to the duplex form (1-4). The nature of the discontinuity in the synthetic strand of the replicated product (RF 111) has remained uncertain. How closely the huge holoenzyme can approach the 5' terminus of an RNA or DNA primer, the influence of substituents at the 5' terminus, and the extent of nucleolytic or strand displacement activity of the enzyme on 5' termini have been important and unresolved questions. With more data available about the dynamics of holoenzyme movements on a template ( 5 ) , the need for information about termination of replication by holoenzyme has become more pressing. The present studies establish that the holoenzyme almost invariably completes the replication of all available template residues, leaving the 5' end of the initiating primer strand intact.
* This work was supported by grants from the National Institutes of Health and the National Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. of bovine serum albumin, and 5 mM dithiothreitol. All other chemicals, nucleic acids, and enzymes not described here were from sources or prepared as in the previous paper (5).
Oligodeoxynucleotide-primed Ml3Goril DNA-An oligodeoxyribonucleotide (15-mer), 5'-GCTTTCGCCGTCCAT-3', was synthesized by the solid phase triester method using protected nucleotide dimers for successive coupling reactions (6), separated from incomplete products by electrophoresis in a 20% polyacrylamide sequencing gel, and extracted from the gel by the crush and soak method (7). The 15-mer was further purified by binding to NACS-52 in 10 mM Tris-HC1 (pH 7.2), 1 mM EDTA, and 0.25 M ammonium acetate followed by washing with the same buffer and elution with 4.0 M ammonium acetate, 10 mM Tris-HC1 (pH 7.2), 1 mM EDTA, and freed of salt by repeated lyophilizations; its sequence was confirmed by analysis (8). Hybridization of the synthetic DNA 15-mer to 4500 pmol (as nucleotide) of Ml3Goril ssDNA was in 10 mM Tris-HC1, 0.3 M NaCl, and 0.03 M sodium citrate (final pH 8.5) using a 10-fold molar excess of the DNA primer (extinction coefficient at 260 nm of 160,000 M-' cm-'). Hybridization was complete within 1 h at 30 "C as shown by replication (90%) of the input ssDNA to RF I1 DNA by holoenzyme.
Replication of DNAand RNA-primed Ml3Goril DNA-The 200pl reactions contained 3500 pmol (as nucleotide) of Ml3Goril DNA, 10.4 pg of SSB, 1.0 mM ATP, 10 ~L M each of CTP, GTP, and UTP, 50 p~ each of dCTP, dGTP, and dATP, and 20 p~ [cx-~'P]~TTP (5000 cpm/pmol) in buffer A. For M13Goril DNA primed with the synthetic DNA 15-mer, replication was initiated with 4.8 pg of holoenzyme (approximately one-half the molar concentration of circles) and quenched after 5 min at 30 "C with 10 pl of 10% SDS. In the case of RNA priming coupled to DNA replication, 2 pg of primase (50fold molar excess over circles) and 4.8 pg of holoenzyme were incubated with M13Goril DNA for 5 min at 30 "C before quenching with 10 p1 of 10% SDS. Replication of the samples was complete as measured by trichloroacetic acid-precipitable counts.
Preparation of PuuII/AuaII Fragments-DNA products from replication reactions of DNA or RNA primed circular M13Goril DNA were extracted with phenol/chloroform (l:l), precipitated with ethanol, and resuspended in 30 p1 of 10 mM Tris-HC1 (pH 7.41, 10 mM MgS04, 1 mM dithiothreitol, and 50 mM NaC1. Each sample was digested with AuaII (10 units) and PuuII (5 units) for 30 min at 37 "C, quenched with 1.5 pl of 10% SDS, and analyzed by electrophoresis in a 2.0% agarose gel in 90 mM Tris borate (pH 8.11, 2.5 mM EDTA, and 1 pg/ml of ethidium bromide. After resolution from the 384-bp fragment, the 155-bp fragment was electroeluted into a NA45 DEAE membrane and removed from the membrane by three successive extractions with 100 pl of 20 mM Tris-HC1 (pH 8.01, 1 M NaCl, and 0.1 mM EDTA at 70 "C for 20 min. A PuuIIIAuaII fragment uniquely labeled with 32P at the PuuII 5' terminus was prepared from M13Goril RF I DNA. The M13Goril RF I DNA (15 pg) was separatedfrom small nucleic acid contaminants by electrophoresis as described above except in a 0.8% low melting agarose gel. The RF I DNA was eluted from the gel, purified by phenol/chloroform extractions, precipitated with ethanol, resuspended in 20 pl of 10 mM Tris-HC1 (pH 7.4), 10 mM MgS04, 50 mM NaCl, and 1 mM dithiothreitol, and linearized by incubating 1 h a t 37 "C with PuuII (12 units). The DNA was extracted with phenol/ chloroform, precipitated with ethanol, and resuspended in 50 pl of 50 mM Tris-HC1 (pH 9.0), 1.0 mM MgC12, 0.1 mM ZnClz, and 1 mM spermidine; the 5' phosphate was removed by treatment with calf intestinal alkaline phosphatase (0.5 units) for 5 min at 37 "C. The dephosphorylated DNA was extracted with phenol/chloroform, precipitated with ethanol, redissolved in 20 pl of 50 mM Tris-HC1 (pH 7.6), 2 pM [-y-32P]ATP (10' cpm/pmol), 10 mM MgC12, 5 mM dithiothreitol, 0.1 mM spermidine, and 0.1 mM EDTA, and incubated with T4 polynucleotide kinase (15 units) for 30 min at 37 "C to phosphorylate the 5' termini. The labeled DNA was extracted with phenol/ chloroform, precipitated with ethanol, redissolved in 30 p1 of 10 mM Tris-HC1 (pH 7.4), 10 mM MgS04, 50 mM NaCl, and 1 mM dithiothreitol, and digested with AuaII (24 units) for 1 h at 37 "C. The reaction was quenched with 1.5 pl of 10% SDS and subjected to electrophoresis in a native 2.0% agarose gel; the 155-bp AuaII/PuaII fragment was purified using a NA45 DEAE membrane as described above.
Replication of DNA-primed Linear $X DNA-The $X ssDNA (19 pmol, as circles; 105 nmol as nucleotide) was primed with 320 pmol (as 15-mer) of a primer (primer 1) (5) that hybridizes over the unique NciI site at position 2802 in 30 p1 of 10 mM Tris-HC1, 0.3 M NaCl, 0.03 M sodium citrate (final pH 8.5). The primed circular $X DNA (9000 pmol as nucleotide) was linearized by nuclease NciI as described in Ref. 5 and then extracted with phenol/chloroform (1:l). The linear DNA (4300 pmol as nucleotide), from which the fragments of the cleaved 15-mer have dissociated, was primed again by incubating at 30 "C for 20 min with a 200-fold molar excess of another 15-mer (5) (primer 2, which anneals at position 4047) in 115 p1 of buffer A containing 12.8 pg of SSB, 0.5 mM ATP, 50 mM each of dCTP, dGTP, and dATP, and 20 p~ [cx-~'P]~TTP (5000 cpm/pmol). Replication was initiated with 5 pg of holoenzyme and quenched after 5 min at 30 "C with 6 p1 of 10% SDS and 5 p1 of 0.5 M EDTA. A size standard for position of termination on linear $X DNA was prepared from circular $X ssDNA (8600 pmol as nucleotide) primed with a 200-fold molar excess of primer 2 (see above) at 30 "C for 20 min in 230 pl of buffer A containing 25.6 pg of SSB, 0.5 mM ATP, 50 p~ each of dCTP, dGTP, and dATP, and 20 p~ [cx-~'P]~TTP (5000 cpm/pmol). Replication was initiated upon adding 10 pg of holoenzyme and quenched after 5 min at 30 "C with 12 pl of 10% SDS and 10 pl of 0.5 M EDTA. Replication of the linear and circular DNA was essentially complete as measured by acid-insoluble radioactivity.
Preparation of SacII/NciI Fragments-DNA products from replication reactions of DNA-primed linear and circular $X DNA were extracted with phenol/chloroform (Ll), precipitated with ethanol, and resuspended in 100 pl of 10 mM Tris-HC1 (pH 7.4), 10 mM MgSOI, and 1 mM dithiothreitol.  (9); the G, C, and C + T reactions were at 0°C for 20 min and the G + A reaction was at 37 "C for 20 min. Autoradiography of dried gels and quantitation of autoradiograms were as described (5).

RESULTS
Replicated Circular DNA Primed with a DNA 15-mer Can Be Covalently Closed by E. coli DNA Ligase-A synthetic DNA 15-mer complementary to the 3' terminal portion of the 28-residue RNA primer synthesized by primase on G4 DNA (10, 11; see "Experimental Procedures") was labeled at its 5' terminus (with T4 polynucleotide kinase and [-p3'P]ATP) and hybridized to M13Goril viral ssDNA. (The latter is a chimera of M13 DNA, 6407 residues, and G4 DNA, 2216 residues, that contains the G4 origin (12).) The primed DNA was replicated by DNA polymerase I11 holoenzyme and the purified products (98% conversion to RF 11; Fig. 1, lane 2) upon treatment with E. coli DNA ligase were covalently closed to RF I DNA; the latter becomes supercoiled upon intercalation of ethidium bromide. Over half of the RF I1 was sealed by ligase within 5 min, consistent with the amount of ligase present; the reaction neared completion within 20 min with a limit of 92% conversion to RF I by 60 min. Trials with other preparations of E. coli ligase gave similar results. Inasmuch as this ligase typically seals DNA at a nick in a strand of duplex DNA (13), it may be presumed that the predominant product of holoenzyme action in this instance has a nick in the synthetic strand.
The Terminus of Replicated DNA Circles Primed with DNA-The position at which holoenzyme terminates replication was determined by polyacrylamide gel analysis of the size of the product digested by restriction nucleases (Fig. 2). M13Goril DNA, primed with the synthetic DNA 15-mer, was replicated by holoenzyme. Digestion of the products with both PuuII (single cut at position 6910) and AuaII (cuts at 6756 and 7296) yields fragments of 7928, 384, and 155 bp. The smallest fragment (separated from the larger ones by electrophoresis) contains the G4 origin and thus the primer and presumably the terminus of the synthetic strand (Fig. 2B).
The discontinuity in the synthetic strand of the 155-bp fragment creates a short arm of 47 nucleotides that includes the synthetic primer and a long arm that includes the terminus. The length of the long arm was determined in a polyacrylamide sequencing gel relative to fragments of known length (Fig. 3). The major species (about 90%) of the long arm of the synthetic DNA (Fig. 3, lane 3 ) coincides in its migration with a fragment of 108 residues ending at guanine 6802. Whereas the reference fragments produced by the base-specific cleavage reactions have a phosphate at both 3' and 5' termini, the long arm of the 155-bp fragment has phosphate only at the 5' end. Hence, this fragment migrates slower and really coincides with a standard that has one less nucleotide (e.g. 107 residues). Thus, holoenzyme terminates replication of the template by including the G residue at 6803 and abutting it to the G residue (6802) at the 5' end of the primer. With respect to the minor species (about 10%; Fig. 3, lane 3), its migration to a position one nucleotide longer than the major species indicates that the synthetic strand terminates at G 6802.
To account for the extra length of this small fraction of the product several possibilities may be considered: (i) displacement of the 5' terminal residue of the primer (G 6802), (ii) nucleolytic removal of G 6802, and (iii) contamination by a 14-mer in the 15-mer primer preparation. Explanation iii is unlikely because in the synthesis of the 15-mer, dimeric nucleotides are successively coupled to a single nucleotide 3'linked to a resin and so should yield only odd numbered products. The virtual absence of 5'+3' exonuclease activity from holoenzyme preparations' leaves displacement of the 5' terminus as the most probable basis for extension of the synthetic strand by an extra nucleotide.
H. Maki and A. Kornberg, personal communication.

C C+T 4 5 6
ti-  Direct measurement of the length of the short arm of the strand in the 155-bp fragment should disclose a minor species with 46 residues (in addition to a major one of 47 residues) were the 5' end of the primer absent at the outset or removed later. As determined by electrophoresis (Fig. 4, lane 3 ) , only a single major band was observed even after a prolonged film exposure that would have detected a minor species (e.g. 46 nucleotides). The unique band migrated to the position of T 6803 (48 nucleotides) consistent with a strand of 47 nucleotides; the lack of 5' phosphate residues on the short arm strand partially accounts for this somewhat slower migration. Thus, holoenzyme in approaching the 5' terminus of a DNA primer in its path leaves no gap in most instances, but occasionally extends the strand by displacing the 5' terminal residue (Table I).

T -
The Terminus of Replicated DNA Circles Primed with RNA-E. coli primase synthesizes an RNA primer on G4 DNA at the complementary strand origin from a unique point (10, 11). When coupled to replication, the length of RNA synthesized by primase depends on the concentrations of rNTPs and dNTPs (15,16) and varies from 2 to 9 nucleotides under the conditions used here. The RF I1 product of the reaction was treated with PuuII and AuaII to generate the 155-bp fragment as for the DNA-primed reaction (above). The lengths of the long arm of the synthetic strand determined by electrophoresis on a sequencing gel (Fig. 3, lane 4 ) are shorter than those from the DNA-primed product, consistent with the start of the 5' RNA primer terminus (Fig.  2C). A major species (67%) of 96 nucleotides terminates at position A 6815 (corrected for the lower charge to length ratio relative to the standard fragments), presumably displacing the 5' RNA terminal nucleotide. Slower migrating products indicate that holoenzyme can replicate and displace even further beyond the 5' RNA terminus. Fragments migrating faster than the major species indicate terminations of replication that generate a nick or leave a small gap ( Table I ) .
As was true for the DNA-primed reaction, the short arm appears as a single band indicative of a unique start (Fig. 4,  lane 4 ) ; no other bands were detected on longer exposure (data not shown). The short arm, with a 5'-ribonucleoside triphosphate and a 3'-hydroxyl moiety has the same net charge of its termini (minus four) as the reference basespecific cleavage fragments, The fragment migrates to a position between A 6849 and A 6850, 60 and 61 nucleotides in length, respectively; a short arm initiated at A 6815, the primase origin, and ending at the PuuII site should be 60 nucleotides long. Thus, holoenzyme generally continues replication until one or a few of the 5' terminal nucleotides of an RNA primer are displaced or, less frequently, a nick or small gap is left (Table I).
The Terminus of Replicated Linear Templates-The position at which holoenzyme terminates replication on a linear template was determined by polyacrylamide gel analysis of the size of product at the terminus of a linear template. A 6x174 ssDNA circle was linearized at position 2803 by annealing a synthetic DNA 15-mer at the unique NciI site followed by restriction nuclease digestion (5); fragments of the DNA 15-mer produced by nuclease cleavage dissociate from the linearized +X DNA. The linear DNA, primed by annealing a DNA 15-mer at position 4047, was replicated by holoenzyme and the product was cleaved with SacII which removes a small fragment from the terminus (Fig. 5). The size of the SacII terminal fragment was determined by comparing its migration in a sequencing gel (Fig. 5, lane 3 ) to that of a 58-base fragment prepared by SacII and NciI digestion of replicated circular +X DNA primed at position 4047 (lane 1 ); a size ladder was made by slight digestion of the reference fragment with exonuclease I11 (lane 2). The major product (85%) of replication of the linear DNA (85%) was 58 nucleotides long showing that the holoenzyme generally stops one base short of full use of the template. The other products were full length (59 bases) and 57-base fragments.

DISCUSSION
These studies demonstrate that DNA polymerase I11 holoenzyme can replicate an available template flush to the 5' terminus of a pre-existing duplex DNA or an RNA primertemplate hybrid. This property of the enzyme to create a complete duplex structure fits nicely with recent information about the dynamics of holoenzyme movements on DNA (5). Holoenzyme, in an activated complex with a primer-template, carries on rapid, processive replication. Retaining a firm grip on the template strand, the holoenzyme diffuses readily over duplex DNA (or an RNA-DNA hybrid duplex) until it finds the next available primer terminus and thus replicates all available templates downstream. However, diffusion over single-stranded DNA does not occur at a significant rate, and thus holoenzyme fails to exploit an upstream primer separated by a single-stranded region. Another indication that interaction with single-stranded template downstream is not crucial to holoenzyme action is the capacity of the enzyme to replicate a linear template to one nucleotide short of its end. The mode of termination on a primed circular DNA has proved to be slightly different depending on whether the primer is DNA or RNA (Table I). With a synthetic DNA 15mer containing a 5'-hydroxyl terminus, the replication product was predominantly (90%) a nicked structure; in a minority of products, the chain was lengthened an extra nucleotide by displacing the 5' terminal residue. When the DNA 15-mer contained a 5'-phosphoryl terminus, the duplex circular replication product was almost completely (92%) convertible to a covalently closed structure by E. coli DNA ligase. Thus, termination of replication by holoenzyme completes the circular template, whether it meets a hydroxyl or a phosphate at the 5' end of the primer DNA strand. Two other prokaryotic DNA polymerases, DNA polymerase I (16) and bacteriophage T7 DNA polymerase (17) completely replicate primed viral ssDNA circles to form a structure sealable by ligase.
An RNA primer synthesized by primase, in the presence of holoenzyme and at the concentrations of nucleotides used here, is 2-9 residues long (18) and was generally (85%) extended beyond the open template; one or a few additional nucleotides were incorporated into the synthetic strand of the circular template by displacing residues at the 5' end of the initiating primer. In a minority of instances, the replication product was completed exactly to generate a nicked product (RF 11) or was slightly shy of completion to leave a gap of 2 or 3 residues. A similar profile of products was observed with an RNA primer 28 residues long (produced by primase action not coupled to replication) as found with the short RNA primers (Table I).3 The tendency to displace the 5' end of the RNA primer and extend the synthetic strand may be due to one or more features of the primer: (i) the triphosphate substituent on the 5' end; (ii) the lower stability of the 5' terminal rA.dT pair end compared to the dG. dC base pair of the synthetic DNA primer; and (iii) pausing of holoenzyme upon encountering an RNA primer uersus a DNA primer, as suggested by the 3s transfer time for holoenzyme across an RNA primer compared to a 1-s transfer time across a DNA primer (5). Another possible explanation for the multiple forms of the terminal residue of the replication product, that include incomplete as well as overextended strands (Table I), may be the heterogeneity of subassemblies present in the multisubunit holoenzyme preparations.
In these holoenzyme studies, none of the 5'+3' exonuclease activity that enables DNA polymerase I to remove RNA primers and cleave nucleotides from the 5' terminus of a DNA duplex was detected (19). This is consistent with direct measurements that show no significant level of this exonuclease activity in preparations of DNA polymerase I11 core or holoenzyme.' For example, the ratio of 5 polymerase activity in DNA polymerase I is about 1 X but the ratio for DNA polymerase 111 core or holoenzyme is less than 2 x 1O"j (19).* These findings are in keeping with the physiological requirement for DNA polymerase I in removing RNA primers in replication (20) and mismatched residues in repair of DNA (21) and the implied failure of DNA polymerase 111 holoenzyme to perform these functions.
These studies together with those of the previous paper demonstrate that once holoenzyme is bound to an available ssDNA template the enzyme replicates the template to the very last nucleotide without dissociating, and will also use any available stretches of duplex DNA downstream as primers.