Bacteriophage T7 DNA replication. Synthesis of lagging strands in a reconstituted system using purified proteins.

Replication of the lagging strand of bacteriophage T7 DNA occurs in a discontinuous fashion that requires RNA-primed DNA synthesis, the removal of the RNA primers, the replacement of the ribonucleotides with deoxyribonucleotides, and the covalent joining of adjacent DNA fragments. We have examined each of these steps as well as the whole process through the use of model substrates and partial reactions using purified proteins. Tetraribonucleotides (pppACCC or pppACCA), synthesized by the T7 gene 4 protein on single-stranded DNA, are used as primers by T7 DNA polymerase to yield RNA-terminated DNA fragments. The removal of the RNA primers is catalyzed by the 5' to 3' hydrolytic activities of either Escherichia coli DNA polymerase I or the T7 gene 6 exonuclease. The products of hydrolysis are pppApC, ATP, and nucleoside 5'-monophosphates or ATP and nucleoside 5'-monophosphates, respectively. The requirement for DNA synthesis to fill the gap between adjacent DNA fragments can be fulfilled by Form II of T7 DNA polymerase but not by Form I. DNA synthesis catalyzed by Form II of T7 DNA polymerase eliminates gaps to create a substrate for DNA ligase whereas strand displacement synthesis catalyzed by Form I creates an aberrant structure that cannot be joined. Either the host or phage DNA ligase can effect the final covalent joining. All steps in the replication of a lagging strand have been coupled in a model system that catalyzes the formation of covalently closed, circular, double-stranded DNA molecules using single-stranded viral DNA as template. A combination of four bacteriophage proteins, gene 4 protein, Form II of T7 DNA polymerase, gene 6 exonuclease, and DNA ligase, can accomplish this overall reaction.

Replication of the lagging strand of bacteriophage T7 DNA occurs in a discontinuous fashion that requires RNA-primed DNA synthesis, the removal of the RNA primers, the replacement of the ribonucleotides with deoxyribonucleotides, and the covalent joining of adjacent DNA fragments. We have examined each of these steps as well as the whole process through the use of model substrates and partial reactions using purified proteins.
Tetraribonucleotides (pppACCC or pppACCA), synthesized by the T7 gene 4 protein on single-stranded DNA, are used as primers by T7 DNA polymerase to yield RNA-terminated DNA fragments. The removal of the RNA primers is catalyzed by the 5' to 3' hydrolytic activities of either Escherichia coli DNA polymerase I or the T7 gene 6 exonuclease. The products of hydrolysis are pppApC, ATP, and nucleoside 5'-monophosphates or ATP and nucleoside 5'-mOnOphosphates, respectively. The requirement for DNA synthesis to fill the gap between adjacent DNA fragments can be fulfilled by Form I1 of T7 DNA polymerase but not by Form I. DNA synthesis catalyzed by Form I1 of T7 DNA polymerase eliminates gaps to create a substrate for DNA ligase whereas strand displacement synthesis catalyzed by Form I creates an aberrant structure that cannot be joined. Either the host or phage DNA ligase can effect the final covalent joining.
All steps in the replication of a lagging strand have been coupled in a model system that catalyzes the formation of covalently closed, circular, double-stranded DNA molecules using single-stranded viral DNA as template. A combination of four bacteriophage proteins, gene 4 protein, Form I1 of T7 DNA polymerase, gene 6 exonuclease, and DNA ligase, can accomplish this overall reaction.
The catalytic properties of DNA polymerase along with the antiparallel structure of duplex DNA molecules dictate that the leading and lagging strands be replicated differently. The replication of the lagging strand poses by far the most difficulty; the discontinuous DNA synthesis (2) that accounts for * This investigation was supported by United States Public Health Service Grant AI-06045 and Grant NP-1L from the American Cancer Society, Inc. This is Paper 26 in a series entitled "Replication of Bacteriophage T7 Deoxyribonucleic Acid." The previous paper is Ref. 1. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$Recipient of National Institutes of Health Fellowship 5 F32 AI05803. Present address, Department of Biochemistry and Molecular Biology, University of Texas Medical School at Houston, Houston,

TX.
its replication requires a number of steps and proteins that are not involved in leading strand synthesis. Bacteriophage T7 DNA replication, requiring relatively few proteins, provides an opportunity to understand the enzymatic events responsible for lagging strand synthesis.
In vivo and in vitro studies have provided considerable insight into the process of discontinuous synthesis on the lagging strand of T7 DNA. Analysis of newly synthesized bacteriophage T7 DNA, isolated from phage-infected cells, reveals the presence of short fragments of DNA (Okazaki fragments) having lengths of from 1000 to 6000 nucleotides (3, 4). These nascent fragments are terminated a t their 5' ends with oligoribonucleotides having the sequence pppA(C) (N)2-3 in which N is mainly A and C, and the chain length is predominantly four ribonucleotides (5,6). Mapping of initiation sites on newly replicated T7 DNA indicates that the recognition sequences for RNA primer synthesis are 3'-CTGGN-5' and 3'-CTGTN-5' (7).
From in vitro studies, it is known that the gene 4 protein of phage T7 catalyzes the synthesis of oligoribonucleotides on single-stranded DNA (8)(9)(10)(11)(12). This primase activity of the gene 4 protein requires the presence of a single-stranded DNA template and ribonucleoside 5"triphosphates. The oligoribonucleotides synthesized in this reaction serve to initiate DNA synthesis when T7 DNA polymerase and the deoxyribonucleoside 5"triphosphates are present. The chains of newly synthesized DNA are terminated at, their 5' ends by covalently linked tetraribonucleotides whose sequences are, as is the case in uiuo, predominantly pppACCC and pppACCA (9,11). The primers are synthesized on the single-stranded DNA at specific sites that are complementary to the primer sequence (12). These recognition sites for the gene 4 protein share the common sequence 3'-CTGGG-5' or 3'-CTGGT-5'. Thus, a cytosine residue is required for recognition but is not copied into the primer, a finding that is consistent with the in vivo studies mentioned above.
In the preceding paper (l), a DNA molecule containing a preformed replication fork was used to define the requirements for leading strand synthesis during the replication of a duplex DNA molecule. In this phase of DNA replication, two phage proteins were found to be required the T7 gene 4 protein and T7 DNA polymerase. Here the gene 4 protein functions as a helicase (13) to "open up" the duplex region ahead of the DNA polymerase at the replication fork. Underlying both the primase and helicase activities of the gene 4 protein is its ability to translocate unidirectionally in a 5' to 3' direction along a single strand (12), a process that requires the hydrolysis of NTPs (12, 13). ' In this paper, we describe the requirements for the synthesis of a continuous lagging strand in a reaction separate from leading strand synthesis. In this process, several steps are S. W. Matson

Lagging Strand
DNA Synthesis required that are unique to lagging strand synthesis: RNA primers must be synthesized and extended by DNA polymerase, the RNA primers must be removed and replaced by deoxyribonucleotides, and the resulting fragments must be joined to yield an uninterrupted strand of DNA.
Of these steps, the synthesis of RNA primers by the gene 4 protein has, to date, been most thoroughly documented. In the phage-infected cell, two enzymes exist that could remove RNA primers, the 5' to 3' exonucleolytic activity of Escherichia coli DNA polymerase I (14) and the gene 6 protein of the phage itself. The latter enzyme is a 5' to 3' doublestranded DNA exonuclease (15,16) that also hydrolyzes RNA in RNA/DNA hybrids (RNase H activity) (17). In the absence of a functional gene 6 protein, RNA-terminated DNA fragments accumulate in infected cells. When both the gene 6 exonuclease and E . coli DNA polymerase I are deficient, a further accumulation is observed (18). However, the major activity responsible for primer removal appears to be the gene 6 protein since, in the absence of DNA polymerase I, there is no detectable accumulation of RNA-terminated DNA fragments (18).
The most likely enzyme that could serve to fill the gaps created by primer removal is the T7 DNA polymerase since it should be positioned at the 3' ends of the nascent DNA fragments. However, our earlier studies (10) on RNA-primed DNA synthesis, using purified proteins, suggested a requirement for additional proteins or another DNA polymerase to correctly fill gaps. In these studies in which T7 gene 4 protein and Form I of T7 DNA polymerase were used, synthesis that had initiated at one primer site did not stop when it reached the 5' terminus of the next RNA-terminated DNA fragment. Instead, synthesis continued, resulting in the displacement of a single-stranded, RNA-terminated DNA fragment. This displaced fragment, in turn, provided additional sites for primer synthesis, leading to repeated initiation events. In this paper, we show that Form I of T7 DNA polymerase catalyzes strand displacement synthesis so rapidly that, even in the presence of DNA ligase, polymerization of nucleotides continues even after the gap is filled. Consequently, DNA ligase is unable to join adjacent fragments. In searching for activities that might circumvent this problem, we found that Form I1 of T7 DNA polymerase, which does not catalyze strand displacement synthesis (19), completes gap filling properly.
From in uiuo studies, it is apparent that the final covalent joining of adjacent fragments can be catalyzed by either the hostor phage-encoded DNA ligase. Mutants of phage T7, defective in T7 DNA ligase, grow normally in wild type E . coli, and wild type T7 phages grow normally in E . coli strains deficient in the host ligase (3); in neither case is there any apparent defect in T7 DNA replication. However, ligasedeficient E. coli are unable to support infection by T7 ligase mutants. In this case, short fragments of newly synthesized DNA accumulate (3,20,21).
In this paper, we present the characterization of each of the individual partial reactions that constitute lagging strand synthesis. Based on this analysis, we have reconstituted lagging strand synthesis using four purified phage enzymes: gene 4 protein, Form I1 of T7 DNA polymerase, gene 6 exonuclease, and DNA ligase.

Materials
Bacterial Strains and Bacteriophages-E. coli Dl10 Suthy end polAl has been previously described (22 DNA-Bacteriophage 6x174 an3 DNA was isolated by the procedure of Hutchison and Sinsheimer (26). Duplex pMB9 DNA containing a single nick was prepared by incubation of pMB9 DNA with pancreatic DNase in the presence of ethidium bromide (27) as described in the accompanying paper (28). DNA of bacteriophage M13 was generously provided by S. W. Matson and S. Tabor (Harvard Medical School). Plasmid pBR322 DNA was prepared as previously described (29). SV40 13H]DNA was a gift from Dr. Paul Wassarman (Harvard Medical School).
Nucleotides-Unlabeled nucleotides were purchased from P-L Biochemicals. All labeled nucleotides were obtained from New England Nuclear.
Enzymes-Gene 4 protein of bacteriophage T7 was Fraction V (50% pure) purified as previously described (30). T7 DNA ligase (95% pure) was purified by a procedure to be published elsewhere. '  . Bacterial alkaline phosphatase was purchased from Millipore Corp. and further purified as previously described (28).
Other Materials-Polyethyleneimine cellulose thin layer plates were purchased from Brinkmann. Ultrapure ammonium sulfate was from Schwarz/Mann.

Methods
Enzyme Assays-T7 DNA ligase was assayed using the ATP-PP, exchange assay previously described for T4 DNA ligase (33). T7 DNA polymerase and gene 4 protein were assayed as described in the accompanying paper (19). E. coli DNA polymerase I was assayed as previously described (34). Gene 6 exonuclease of phage T7 was assayed essentially as described by Kerr and Sadowski (15) except that HaeIII-digested SV40 [3H]DNA (1.5 nmol/reaction mixture) was used as substrate. The specific activity of the DNA was 18,000 cpm/ nmol. Enzyme activity was proportional to enzyme concentration between 0.05 and 0.6 unit/reaction mixture (0.1 ml). One unit of gene 6 exonuclease activity is defined as that amount that catalyzes the release of 1 nmol of acid-soluble nucleotide after a 15-min incubation at 37 "C under the conditions of the standard assay. E. coli exonuclease VI1 was assayed as previously described (35).
The 5' to 3' hydrolytic activities of E. coli DNA polymerase I and of T7 gene 6 exonuclease were, in some experiments, determined in an assay that measured the conversion of acid-insoluble 32P04 ester at the 5' terminus of DNA to an acid-soluble form. The reaction mixture (0.1 ml) for the assay of DNA polymerase I contained T7 [5'-32P]DNA (1.8 pmol of 5"terminal phosphorus; 6.1 X lo3 cpm/ pmol), 70 mM potassium phosphate buffer (pH 7.4), 7 mM MgC12, 1 mM 2-mercaptoethanol, 33 /*M each dATP, dTTP, dCTP, and dGTP, and the indicated amounts of E. coli DNA polymerase I. The reaction mixture (0.1 ml) for the assay of gene 6 exonuclease contained T7 [5'-"P]DNA (1.8 pmol of 5"terminal phosphorus; 6.1 X lo3 cpm/ pmol), 50 mM Tris-HC1 (pH 8.0), 5 mM MgCL, 1 mM dithiothreitol, 20 mM KC1, and the indicated amounts of T7 gene 6 exonuclease. In both assays, after incubation for 10 min at 37 "C, 0.2 ml of salmon sperm DNA (2.5 mg/ml) and 0.5 ml of 0.7 N trichloroacetic acid were added. After centrifugation at 10,000 X g for 10 min, an aliquot of the supernatant fluid was removed and the radioactivity was determined. For both enzymes, one unit of 5"terminal nucleotide activity is defined as the amount of enzyme catalyzing the conversion of 1 pmol of 5"terminal phosphate to an acid-soluble form in 10 min at 37 "C.
Preparation of Radioactively Labeled DNA and RNA-DNA Substrates-T7 [5'-32P]DNA was prepared as described by Chase and Richardson (35). T7 DNA was partially degraded by sonic irradiation, dephosphorylated by incubation with bacterial alkaline phosphatase at 65 "C, and then phosphorylated using T4 polynucleotide kinase in the presence of [y3'P]ATP (6.1 X lo3 cpm/pmol). The  http://www.jbc.org/ Downloaded from was purified on a benzoylated DEAE column as previously described (35).
Fragments of T7 DNA bearing radioactively labeled tetraribonucleotides at their 5' termini were synthesized in a reaction containing T7 gene 4 protein and T 7 DNA polymerase as previously described (10). The tetraribonucleotide primers synthesized by the T7 gene 4 protein were radioactively labeled with 32P using [y-32PJATP and [a-32P]CTP; the DNA portion of the molecule was labeled with 3H using [3H]dTTP. The reaction mixture (0.5 ml) contained 40 mM Tris-HCI (pH 7.5), 10 mM MgC12, 10 mM dithiothreitol, 0.3 mM each dNTP, including [3H]dTTP and either [y-32P]ATP (2.4 X lo5 cpm/pmol) or [a-"P]CTP (1.8 X lo5 cpm/pmol), or both "P labels, 30 nmol of duplex T7 DNA, and 0.10 ml of a solution containing 2 units of T 7 DNA polymerase (Form I), 2 units of gene 4 protein, and 5 pg of E. coli DNA-binding protein in 10 mM Tris-HC1 (pH 7.5). 10 mM 2mercaptoethanol, 0.5 mg/ml of bovine serum albumin. After incubation a t 30 "C for 10 min, the reaction was stopped and the RNA-DNA product was isolated as previously described (10). The average length of the RNA-primed DNA was 4000 nucleotides, calculated on the basis of pppACCC being the predominant primer synthesized in this reaction (11).
Assay for Removal of RNA Primers-The assay for enzymatic removal of a gene 4 protein tetraribonucleotide primer from the 5' were incubated with either E. coliDNA polymerase I (0.07 5"terminal nucleotide unit) or T7 gene 6 exonuclease (1.4 5"terminal nucleot,ide units) for 20 min a t 37 "C under the conditions used for assay of hydrolysis of RNA primers. The reaction was stopped by the addition of 20 pl of tRNA (4 pg/ml) and 40 pl of a solution containing 10 M urea, 0.05% xylene cyanol, and 0.05% bromphenol blue. One-half (40 pl) of each sample was applied to a 23% polyacrylamide gel containing 7 M urea and electrophoresed as previously described (11). The radioactive species were cut from the gel, crushed, and soaked in water to remove the products. The products were adsorbed to Norit, washed, and eluted. The eluted products were then analyzed, along with markers of ATP, CTP, ADP, CDP, AMP, and CMP, by chromatography on polyethyleneimine cellulose thin layer plates developed with 0.5 M LiCI, 1 M formic acid.
Assay of Repair of Gaps and Nicks-The extent of gap filling by T7 DNA polymerase was measured by determining the number of covalenty closed circular DNA molecules when circular, duplex DNA containing a gap was incubated with T7 DNA polymerase and T7 DNA ligase. The reaction mixture (10 pl) contained 50 mM Tris-HC1 (pH 7.5), 10 mM MgC12, 150 p~ each ATP, GTP, UTP, CTP, dATP, dCTP, dTTP, and dCTP, 10 mM dithiothreitol, 20 mM KCl, 0.5 nmol of pMB9 DNA containing a single nick, and 0.02 unit of E. coli exonuclease 111. After incubation a t 37 "C for 30 min, the reaction mixture was incubated at 43 "C for 10 min, and then heated to 65 "C for 10 min in order to inactivak the exonuclease (36). The reaction mixture was placed in an ice bath until the remaining enzymes were added. Approximately 5% of the nucleotides in the nicked strand were hydrolyzed under these conditions and no covalently closed duplex circles were detectable even after incubation with DNA ligase. The indicated amounts of T7 DNA polymerase and T7 DNA ligase were then added and the reaction mixture was incubated a t 30 "C for 30 min. The reaction was stopped by chilling the reaction mixture to 0 "C and the addition of EDTA to 20 mM. The entire reaction mixture was then subjected to electrophoresis at 100 V for 15 h on a 0.8% agarose gel containing ethidium bromide (0.06 pglml) as described in the accompanying paper (28).
After electrophoresis, the gel was stained with ethidium bromide and photographed. Proper gap filling and covalent joining of the resulting nick lead to the formation of covalently closed circular molecules.
Assay for Conversion of Single-stranded Circular DNA to Covalently Closed Duplex Circular DNA-A model system for the total reconstitution of lagging strand synthesis is the conversion of a singlestranded circular DNA molecule to a covalently closed duplex circular molecule. Such a conversion requires RNA-primed DNA synthesis, primer removal, and covalent joining. The reaction mixture (10 p1) contained 50 mM Tris-HC1 (pH 7.5), 10 mM MgCI,, 150 p M ATP, GTP, UTP, CTP, [o~-~'P]~ATP (100 cpm/pmol), dGTP, dTTP, and dCTP, 20 mM KC1, 10 mM dithiothreitol, and 0.25 nmol of singlestranded circular DNA (M13 or 6x174 DNA as indicated), and the indicated amounts of T7 gene 4 protein, T7 DNA polymerase, T7 gene 6 protein, and T7 DNA ligase. After incubation at 30 "C for 1 h, the reaction was stopped and subjected to electrophoresis on an 0.8% agarose gel containing ethidium bromide, and the gel was photographed as described above for the assay of repair of gaps and nicks. The agarose gel was dried onto a sheet of Whatman 3MM filter paper and the radioactivity in newly synthesized DNA was located by radioautography. The dried gel was cut into strips and the radioactivity in covalently closed duplex circles was measured in a liquid scintillation counter.
In experiments designed to identify proteins or factors that would promote the overall conversion of single-stranded circles to covalently closed duplex circles, the assay was modified as follows. The reaction mixture (25 pl) contained 0.5 nmol of single-stranded 4x174 DNA, 0.07 unit of T7 DNA polymerase, 0.07 unit of gene 4 protein, 0.35 unit of T7 DNA polymerase, 0.6 unit (5"terminal nucleotide hydrolysis activity) of T7 gene 6 protein, 0.13 unit of T7 DNA ligase, and the indicated fractions or proteins to be assayed. After incubation at 30 "C for 1 h, the enzymes were inactivated by incubation a t 60 "C for 10 min. Then 0.7 unit of E. coli exonuclease I and 5.6 units of E. coli exonuclease 111 were added and the reaction incubated at 37 "C for 30 min. Incubation with the combination of E. coli exonucleases I and I11 hydrolyzes all newly synthesized DNA that has not been covalently joined into a continuous strand, thus leaving only covalently closed circular duplex molecules. The reaction was stopped by the addition of EDTA to 20 mM and the products of the reaction were separated by agarose gel electrophoresis as described above.
Purification of Gene 6 Exonuclease of Bacteriophage T7"The gene 6 exonuclease of phage T7 was purified from E. coli HMS151 cells infected with a T7 mutant carrying mutations in genes 1.3, 3, and 5 by a modification of the procedure of Kerr and Sadowski (15). The use of the mutant phage ensures the absence of T7 DNA polymerase, gene 3 endonuclease, and T7 DNA ligase. The latter enzyme had been a major contaminant in earlier preparations of gene 6 exonuclease. In addition, the E. coli strain used for the infection carries a mutation, polAexl, that eliminates the 5' to 3' hydrolytic activity of DNA polymerase I. The results of a typical purification procedure are presented in Table 1. E. coli HMS151 was grown and infected with T7 L.3,3,5as previously described (30). Cell paste (26 g) was suspended in 104 ml of 50 mM Tris-HC1 buffer (pH 7.5), 10% sucrose, frozen in liquid nitrogen, and stored at -85 "C. Frozen cells were thawed overnight in ice and 2.48 ml of 1 M NaCl and 2.48 ml of a solution of lysozyme (10 mg/rnl), 50 mM Tris-HC1 (pH 7.5), 10% sucrose were added. After 45 min a t 0 "C, the mixture was heated to 20 "C in a 37 "C water bath with constant stirring, and then chilled to 5 "C in an ice bath with stirring. The lysates were then centrifuged a t 35,000 rpm for 30 min in a Spinco 45-Ti rotor.  As shown in Table I, this procedure resulted in an 80-fold purification of the activity over the starting material, with an overall yield of 40%. The specific activity of the purified enzyme was 93,000 units/ mg. Electrophoresis of the denatured and reduced gene 6 exonuclease (Fraction V) through polyacrylamide gels containing sodium dodecyl sulfate revealed a single band corresponding to a protein of M, = 3 1,000.

Removal of RNA Primers
T7 DNA isolated from the virion is composed of two uninterrupted strands, neither of which contains any detectable ribonucleotides (37,38). Furthermore, E. coli DNA ligase, an enzyme that can replace the T7 DNA ligase in vivo (3), cannot catalyze the joining of a 5'-RNA-terminated fragment to a DNA fragment (39) even if the 5' terminus is converted to a monophosphate ester. Therefore, an efficient mechanism must be available in T7-infected cells to remove the many 5'tetraribonucleotides that terminate lagging strand Okazaki fragments.  (14,17). In addition, the 5' to 3' hydrolytic activity of E. coli DNA polymerase I has been shown to be essential for the processing of Okazaki fragments generated during the replication of the bacterial chromosome in uiuo (40,41).
In order to examine the removal of RNA primers from the 5' termini of DNA, we have chosen for a model substrate the newly synthesized DNA made in a reaction in which T7 gene 4 protein and T7 DNA polymerase act on a duplex T7 DNA template in the presence of radioactively labeled rNTPs and dNTPs. In this reaction, DNA fragments 5000 to 6000 nucleotides in length are produced. Each of these has a tetraribonucleotide (pppACCC or pppACCA) covalently linked to its 5' terminus, the result of multiple priming events catalyzed by the primase activity of the gene 4 protein (10,ll). In order to follow the removal of the primers they were radioactively labeled with [32P]CMP; the DNA was labeled with 3H.
As shown in Fig. 1, both E. coli DNA polymerase I and T7 gene 6 exonuclease render the radioactively labeled RNA primers acid-soluble. The addition of 2-fold additional E. coli DNA polymerase I resulted in greater than 90% removal in a 40-min incubation. However, the addition of 2-fold more gene 6 exonuclease did not result in the removal of more than 75% of the primer. In addition, considerable hydrolysis of the 3Hlabeled, newly synthesized DNA accompanied primer removal by the gene 6 exonuclease (Fig. 1). The Fig. 2. When [y-"PIATP-labeled primers were incubated with E. coli DNA polymerase I, two radioactive products were observed (Fig. 2). The major, slower moving product is the triphosphate-terminated dinucleotide pppApC; the minor product is ATP (see Fig. 2 for identification). Two radioactively labeled products were also found when the ["2P]CMPlabeled primers were used as substrate. Again, one is pppApC while the other is CMP. Thus, the 5' to 3' hydrolytic activity of E. coli DNA polymerase I hydrolyzes the primer by cleaving the first or second phosphodiester linkage from the 5' terminus, releasing ATP or pppApC, respectively. The analogous products have also been observed when the enzyme hydrolyzes a DNA molecule bearing a 5"triphosphate (42). In these earlier studies, kinetic data indicated that the dinucleotide observed represented the initial cleavage product of the 5' to 3' hydrolytic activity and was not the product of cleavage of a larger oligonucleotide by the 3' to 5' hydrolytic activity of DNA polymerase I. The ATP observed must be a hydrolysis product of the 5' to 3' activity since pppTpTpT is not a substrate for the 3' to 5' hydrolytic activity of DNA polymerase I (42). However, the CMP residues observed may be subsequent cleavage products of the 5' to 3' hydrolytic activity, or the products of the combined activities of the 5' to 3' and 3' to 5' hydrolytic activities.
When the analysis was carried using gene 6 exonuclease, only a single radioactive product, ["P]CMP, was produced when [:''P]CMP-labeled primer was used as substrate. Similarly, only a single radioactive product, ["PIATP, was released from the [-y-:"P]ATP-labeled primer. We conclude that gene 6 exonuclease removes single nucleotide residues exclusively even if a 5"terminal triphosphate is present. A similar result was obtained by Shinozaki and Okazaki (17) using a 5"triphosphate-terminated RNA transcript covalently attached to DNA. were hydrolyzed with either E. coli DNA polymerase I or T7 gene 6 exonuclease. The RNA primers were radioactively labeled with (y-"PIATP or [32P]CMP as described under "Experimental Procedures." Radioactively labeled products were analyzed by electrophoresis through a 23% polyacrylamide gel containing 7 M urea and located by autoradiography. The identity of the indicated products was determined as described under "Experimental Procedures." The species containing both ATP and CMP was identified as the terminal pppApC by alkaline hydrolysis followed by thin layer chromatography; greater than 90% of the radioactivity was found in pppAp. Elimination of Gaps and Resulting Nicks In the accompanying papers (1, 19, 28), we showed that Form I of T 7 DNA polymerase catalyzes limited strand displacement synthesis at a nick, with subsequent strand switching and additional synthesis giving rise to branched duplex structures (panhandles). Such an aberrant structure prevents joining of adjacent DNA fragments by DNA ligase (19). If Form I of T 7 DNA polymerase catalyzes a similar strand displacement reaction after filling in the gap between adjacent Okazaki fragments, then gap filling could specifically require Form I1 of T 7 DNA polymerase, a form that does not catalyze strand displacement synthesis.
Although it seemed likely that Form I of T 7 DNA polymerase would also catalyze strand displacement synthesis at the nick generated by gap filling, one cannot a priori assume that such a reaction would occur. For example, the blunt ends generated by the 5' to 3' hydrolytic activity of E. coli DNA polymerase 111 acting on duplex DNA molecules bearing 5'single-stranded tails are not recognized as such by the enzyme (43). The enzyme continues its hydrolysis into the duplex region even though it cannot initiate hydrolysis at blunt ends normally. Thus, Form I of T 7 DNA polymerase, once engaged in the process of DNA synthesis, might not recognize the nick it creates.
In the experiment shown in Fig. 3   containing ethidium bromide (Fig. 3, lanes 1 and 2). As shown in the accompanying paper (28), molecules containing duplex branched structures (panhandles) can be identified by their slightly lower mobility relative to covalently closed duplex circles. DNA synthesis catalyzed by Form I of T 7 DNA polymerase leads to the formation of panhandles (Fig. 3, lanes  6, 7, and 8); few covalently closed circular molecules formed. On the other hand, DNA synthesis catalyzed by Form I1 of DNA polymerase completely fills the gaps created by exonuclease 111; the resulting nicked DNA is a substrate for T 7 DNA ligase (Fig. 3, compare lanes 2 and 11). Finally, a mixture of the two forms of T 7 DNA polymerase yields both products, molecules containing panhandles and covalently closed duplex circles (Fig. 3, lanes 12 and 13). It

DNAs: a Model System for Lagging Strand Synthesis
Our earlier attempts to reconstitute lagging strand synthesis (10) using duplex DNA as a template were complicated not only by the simultaneous leading strand synthesis, but also by the occurrence of aberrant strand displacement and strand switching that gave rise to excessive DNA synthesis (see the Introduction). In this section, we report the conversion of single-stranded, circular DNA to covalently closed, duplex circular DNA as an in vitro system for the synthesis and processing of Okazaki fragments. As depicted schematically in Fig. 4, the initial step in the synthesis of the complementary strand by T 7 proteins is the synthesis of a tetraribonucleotide at one of 13 sites on the 4x174 DNA molecule (12). T 7 DNA polymerase then uses the tetraribonucleotide as a primer to initiate DNA synthesis, a reaction that is analogous to the synthesis of RNA-primed DNA on the lagging strand at a replication fork. The removal of the RNA primer, the completion of DNA synthesis (gap filling), and the ligation of the adjacent ends are equivalent to the stages involved in the processing of Okazaki fragments during replication. Although E. coli DNA polymerase I can also catalyze the removal of the RNA primers (see above), in these studies we have used the T7 gene 6 exonuclease since in vivo studies suggest that it is the major activity responsible for primer removal (18).

Requirements for Complementary Strand Synthesis-Syn-
thesis of the complementary strand of 4x174 single-stranded RNA Primer, viral DNA to yield a covalently closed, circular duplex is shown in Fig. 5. The products of the reaction are separated by electrophoresis through an agarose gel containing ethidium bromide. The appearance of covalently closed, circular DNA is seen only in the presence of all four enzymes: T 7 gene 4 protein, Form I1 of T 7 DNA polymerase, T 7 gene 6 exonuclease, and T7 DNA ligase (Fig. 5, lane 4). In the absence of gene 4 protein or T 7 DNA polymerase, no DNA synthesis occurs; only the input 4x174 single-stranded template is observed.
Both gene 6 exonuclease (lane 3 ) and DNA ligase (lane 5 ) are required for the formation of covalently closed, circular molecules; in the absence of either, nicked duplex, circular molecules accumulate. Since the gene 6 exonuclease is needed, sealing by T 7 DNA ligase must require that RNA primers first be removed. No DNA containing duplex branches (panhandles) is present in any of the reaction mixtures, attesting to the absence of Form I of T 7 DNA polymerase. Panhandle DNA is generated by Form I of T 7 DNA polymerase in a reaction requiring neither DNA ligase nor gene 6 exonuclease (data not shown).
In order to obtain a more quantitative evaluation of the data presented in Fig. 5, the quantity of newly synthesized DNA in each species of product molecule was determined by  Table I1 summarize the amount of newly synthesized DNA in nicked, duplex circles, linear duplexes, covalently closed, circular molecules, and single-stranded circles. Clearly, only in the presence of all four enzymes is there any significant amount of covalently closed product synthesized; of the 250 pmol of 6x174 DNA template added to the reaction, 61 pmol were converted to covalently closed, circular duplex molecules.
In the absence of gene 4 protein (the primase), synthesis is reduced 100-fold. In the absence of either gene 6 exonuclease or DNA ligase, nicked duplex, circular molecules accumulate. The presence of gene 6 exonuclease at the concentration used here reduces the net amount of DNA synthesized by approximately 25%.
Form I of T7 DNA Polymerase-Our initial attempts to reconstitute lagging strand synthesis involved the use of Form I of T 7 DNA polymerase. In light of the studies described in the accompanying papers (1,19,28) and those presented in the section on gap filling above, we can now explain our inability to obtain covalent closure of the product molecules in the four-enzyme system just described. However, our attempts to identify factors that would allow the formation of covalently closed circles, perhaps by preventing panhandle formation, by Form I of T 7 DNA polymerase led to the purification of Form I1 of T 7 DNA polymerase and to the discovery that another protein, E. coli exonuclease VII, could partially overcome aberrant strand displacement and strandswitching reactions by Form I of T 7 DNA polymerase.
In order to facilitate distinguishing between covalently closed DNA molecules and topologically restrained molecules containing panhandles, we have used the assay described in Fig. 6. The combination of Form I of T 7 DNA polymerase, gene 4 protein, gene 6 exonuclease, and T7 DNA ligase results in extensive synthesis of DNA on single-stranded, circular DNA templates (Fig. 6, lune 1 ) . Although a significant portion of the newly synthesized DNA has a mobility similar to but not identical with that of covalently closed, circular molecules in ethidium bromide, it can be distinguished from the latter form by its susceptibility to a combination of E. coli exonucleases I and 111. As shown in lune 2, essentially all of the newly synthesized DNA, including the rapidly migrating species, is sensitive to exonuclease treatment. Covalently closed, circular duplex molecules are resistant to these exonucleases. The detailed characterization of these panhandle structures has been described in the second paper of this series (28). Varying the concentration of each of the four enzymes did not yield any significant exonuclease-resistant, newly synthe-

TABLE I1
Requirements for synthesk of covalently closed, circular duplex DNA The agarose gel shown in Fig. 5 was dried onto filter paper and the "P-labeled newly synthesized DNA was located by autoradiography. Regions of the gel corresponding to nicked circles, duplex linears, covalently closed circles, and single-stranded DNA were cut out, and the amount of newly synthesized DNA in each species was determined by liquid scintillation counting. prior to electrophoresis through an agarose gel in the presence of ethidium bromide. After photography, the gel was prepared for fluorautography by first soaking in 95% ethanol for 2 h and then in 3% 2,5-diphenyloxazole in 95% ethanol for 2 h. The gel was transferred to H20, soaked for 12 h at 4 "C, and then dried onto Whatman 3" filter paper after which it was analyzed by fluorautography. A concentrated fraction of protein was prepared from T71-infected E. coli D110. Lysis and removal of debris were carried out as previously described (19). A portion of the extract (0.35 ml) was passed through a DE52 column (0.75 ml) equilibrated with 0.25 M (NH,)*SO,, 50 mM Tris-HCI (pH 7.5), and 1 mM dithiothreitol in order to remove the bulk of the nucleic acids. The protein passing through the column was collected as a single fracton and assayed for its ability to promote covalent closure of the newly synthesized DNA in the assay described above. Lane 1, no exonuclease treatment following the DNA synthesis reaction; lane 2, treatment with exonucleases following the DNA synthesis reaction; lane 3, extract of T7 I-infected cells (2 pl of DE52 fraction added to DNA synthesis reaction, no exonuclease treatment following DNA synthesis reaction; lane 4, extract of T7,-infected cells (2 pl of DE52 fraction added to DNA synthesis reaction, treatment with exonucleases following DNA synthesis reaction. sized DNA (data not shown). Furthermore, although the amount of DNA synthesized is greater than the amount of template added, the addition of DNA ligase is without effect. The possibility of a missing component in the reaction catalyzed by the four T7 proteins led us to examine extracts of T7-infected cells for an activity that would stimulate covalent joining of the newly synthesized, complementary strand. As shown in Fig. 6, (lanes 3 and 4), such extracts do stimulate synthesis of covalently closed, circular DNA molecules. Purification of the stimulatory activity led to the identification of two proteins that promoted the synthesis of covalently closed, circular DNA molecules. One protein purified together with an EDTA-resistant, 5'-3'-single-stranded DNA-specific exonuclease that was also present in uninfected cells (data not shown). The chromatographic and enzymatic properties of this exonuclease are identical with those of E. coli exonuclease VI1 (35). The second stimulatory protein, found only in extracts of T7-infected E. coli had chromatographic and enzymatic properties similar to Form I1 of T 7 DNA polymerase (19). The results of this study are summarized in Fig. 7. An extract of T7-infected cells, after preliminary fractionation with streptomycin sulfate and ammonium sulfate, is chromatographed on DEAE-cellulose. E. coli exonuclease VI1 and T7 DNA polymerase activities were measured and, as can be seen in A , the stimulatory activity spans the fractions encompassing both enzymatic activities. Furthermore, the DEAE fraction of Form I1 of T 7 DNA polymerase is most effective in promoting the synthesis of covalently cloned molecules when Form I is omitted from the reaction mixture (Fig. 7B).

DISCUSSION
Based on the known activities of the T7 gene 4 protein, T 7 DNA polymerase, T 7 gene 6 exonuclease, and T 7 DNA ligase, we initially expected that a combination of these proteins would readily carry out the multiple steps of lagging strand synthesis. However, the first difficulty encountered arose when only two of the proteins, T7 gene 4 protein and T7 DNA polymerase, were used to carry out the first steps in lagging strand synthesis, i.e. RNA primer synthesis and the extension of the primer by polymerization of deoxyribonucleotides. We found that extensive synthesis occurred on the lagging strand, the result of strand displacement synthesis and multiple priming events (10,11). We erroneously attributed this aberrant reaction to the ability of the helicase activity of gene 4 protein to allow the T7 DNA polymerase to catalyze strand displacement synthesis at a nick. We now know that Form I of T 7 DNA polymerase used in these earlier studies could itself catalyze strand displacement synthesis.
Again, we assumed that the presence of an RNase H activity to remove the primers would enable DNA ligase to join the ends of Okazaki fragments and thus prevent strand displacement synthesis. In this paper, we have shown that the T7 gene 6 exonuclease, as well as the 5' to 3' hydrolytic activity of E. coli DNA polymerase I, will remove the RNA primer synthesized by the gene 4 protein from the 5' termini of Okazaki fragments synthesized in uitro. However, attempts to achieve proper lagging strand synthesis in the presence of gene 6 exonuclease and DNA ligase were unsuccessful. Essentially the identical amount of synthesis occurred on the lagging strand in the presence or absence of these enzymes.
In an attempt to identify factors that would circumvent this problem, we discovered Form I1 of T 7 DNA polymerase whose purification and properties are described in the first paper of this series (19). We now realize that the inability to achieve proper lagging strand synthesis in these earlier studies resided in the properties of Form I of T 7 DNA polymerase.
Its ability to catalyze rapid strand displacement synthesis at a nick prevents ligation (28). However, Form I1 of T 7 DNA, a form that is unable to catalyze even a few nucleotides at a nick or at a preformed replication fork (l), completely fills the gap and permits ligation of adjacent fragments to occur. An additional enzymatic activity was also found that enabled the Form I of T 7 DNA polymerase to fill gaps to yield nicks that could be repaired by T 7 DNA ligase. Upon purification, this activity was identified as E. coli exonuclease VI1 and it functioned equally well in the presence or absence of gene 6 exonuclease. We propose that the single strand-specific exonuclease VI1 hydrolyzes the displaced single strand before strand switching by the polymerase occurs. Thus, the RNA primer is removed by exonuclease VII, and DNA ligase can seal the duplex generated by the complete excision of the displaced strand. Whether this is ever a physiological role of exonuclease VI1 is not known. However, it is interesting to note that mutants defective in exonuclease VI1 are phenotypically hyper-Rec (45) as are other E. coli mutants perceived to increase the occurrence of single-stranded tails (46). That exonuclease VI1 mutants support normal infection by bacteriophage T 7 map indicate that this repair pathway is seldom needed when Form I1 of T 7 DNA polymerase is present.
After characterization of the partial reaction of lagging strand synthesis, we have reconstituted a major portion of the process using a model system: the synthesis of the complementary strand of a single-strand viral DNA. In this reaction, the four proteins discussed above synthesize covalently closed circular duplexes. Such a system has also been used to characterize the conversion of 4x174 DNA to a supercoiled molecule using purified E. coli proteins (47), a reaction that requires not four but a t least 12 proteins.
At present, we consider our understanding of lagging strand synthesis only partially complete. First, the reconstituted system lacks the tight regulation necessary for complete conversion of single-stranded circular DNA into covalently closed duplex circles. One problem is the potent activity of gene 6 exonuclease on duplex DNA. DNA hydrolysis invariably accompanies primer removal and suggests the presence of other factors or perhaps a complex of replication enzymes that are under tighter regulation. In this regard, it is important to note that the lagging strand synthesis characterized in this paper is uncoupled from leading strand synthesis. The fact that the helicase activity of the gene 4 protein is specific for T 7 DNA polymerase (1, 30) suggests a physical association of a t least these two enzymes at the replication fork. This in turn raises the possibility that leading and lagging strands are replicated complex (48,49). In this structure, the use of potential primer U. S. A