Functional Properties of Replication Fork Assemblies Established by the Bacteriophage λ O and P Replication Proteins*

We have used a set of bacteriophage λ andEscherichia coli replication proteins to establish rolling circle DNA replication in vitro to permit characterization of the functional properties of λ replication forks. We demonstrate that the λ replication fork assembly synthesizes leading strand DNA chains at a physiological rate of 650–750 nucleotides/s at 30 °C. This rate is identical to the fork movement rate we obtained using a minimal protein system, composed solely of E. coli DnaB helicase and DNA polymerase III holoenzyme. Our data are consistent with the conclusion that these two key bacterial replication proteins constitute the basic functional unit of a λ replication fork. A comparison of rolling circle DNA replication in the minimal and λ replication systems indicated that DNA synthesis proceeded for more extensive periods in the λ system and produced longer DNA chains, which averaged nearly 200 kilobases in length. The higher potency of the λ replication system is believed to result from its capacity to mediate efficient reloading of DnaB helicase onto rolling circle replication products, thereby permitting reinitiation of DNA chain elongation following spontaneous termination events. E. coli single-stranded DNA-binding protein and primase individually stimulated rolling circle DNA replication, but they apparently act indirectly by blocking accumulation of inhibitory free single-stranded DNA product. Finally, in the course of this work, we discovered thatE. coli DNA polymerase III holoenzyme is itself capable of carrying out significant strand displacement DNA synthesis at about 50 nucleotides/s when it is supplemented with E. colisingle-stranded DNA-binding protein.

Investigations of the biochemical mechanisms involved in the initiation of bacteriophage DNA replication have been aided by the reconstitution of the initiation reaction with a defined set of purified and Escherichia coli proteins (1,2). These studies have demonstrated that an ordered series of nucleoprotein structures are assembled at ori, the viral replication origin, and subsequently partially disassembled during the establishment of the apparatus responsible for replication fork propagation (2)(3)(4)(5).
The first step in the initiation pathway consists of the binding of multiple copies of the O protein to the viral origin and the subsequent self-assembly of this replication initiator into a nucleoprotein structure called the O-some (6,7). The O-some both localizes the origin and serves as the foundation for the assembly of more complex nucleoprotein structures. Simultaneously, in solution, several molecules of the P protein bind to the hexameric E. coli DnaB helicase, to form a P⅐DnaB complex (8 -10). One or more molecules of this latter complex binds to the O-some, as a consequence of specific protein-protein interactions between O and P, to form a second stage nucleoprotein structure containing O, P, and DnaB. Protein-protein interactions within this ori⅐O⅐P⅐DnaB structure prevent release and activation of the helicase activity of the DnaB molecule or molecules present in the complex (7). Binding of the E. coli DnaJ protein to the ori⅐O⅐P⅐DnaB preinitiation structure (3) sets the stage for the ATP-dependent partial disassembly of the complex by the host DnaK (Hsp70) molecular chaperone (4,5,11), a reaction that is aided by the GrpE protein (2,4). This protein turnover reaction results in the release of a portion of the bound P and DnaJ proteins, as well as the liberation of DnaB and activation of its intrinsic helicase activity (4,5,11,12). If the ori template DNA is sufficiently negatively supercoiled, the released DnaB is inserted between the two DNA strands, presumably at the A ϩ T-rich region of ori (7,12,13), and becomes active as a replicative helicase.
The transfer of DnaB helicase onto the DNA template completes the initiation phase of DNA replication and triggers a second series of protein-protein and protein-DNA interactions that culminate in assembly of the replication fork apparatus. The precise assembly pathway remains to be defined, but it is believed to include the following course of events. Singlestranded DNA (ssDNA) 1 created by DnaB helicase action is stabilized by the binding of stoichiometric quantities of the E. coli single-stranded-DNA binding protein (SSB) (3,12,14). The E. coli primase (DnaG protein) binds transiently to the DnaB helicase (15,16) and subsequently synthesizes short RNA chains that serve as primers for synthesis of the leading and lagging DNA strands by the DNA polymerase III holoenzyme (DNA pol IIIh) (3,15).
Biochemical studies of DNA replication in vitro have demonstrated that detectable levels of O, P, and DnaK remain associated with the ori template DNA following the chaperone-mediated partial disassembly reactions that bring about initiation of DNA replication (4,11,17). It is still uncertain if one or more of these proteins is associated with the replication fork apparatus. Although none of the evidence available to date suggests a direct role for the O and P replication proteins or the DnaK/DnaJ/GrpE chaperone system in the propagation of replication forks during DNA replication (4), there remains a possibility that one or more of these proteins that act during the initiation phase of the reaction have an auxiliary role during the fork movement phase.
We wished, therefore, to characterize the properties of replication forks established by the O and P proteins for comparison to the behavior of replication forks established with E. coli replication proteins. To study replication fork assembly and movement, we adapted the rolling circle replication assay pioneered by Lechner and Richardson (18), in which a tailed, nicked circular (TNC) DNA molecule serves as the replication template. With this assay, we can take advantage of the capacity of the O and P proteins to mediate the transfer of DnaB helicase onto any ssDNA, even if the ssDNA is stoichiometrically coated with SSB (19,20). A primary benefit of the rolling circle replication system for studies of the propagation of replication forks is that there are no topological constraints to encumber fork movement. This is in sharp contrast to the situation encountered with the standard ori plasmid replication assay, in which the rate of fork movement is probably limited by the rate at which topoisomerases relax the positive supercoils generated during replication of a covalently closed, circular template (21,22).
DNAs and Phages-DNA and /HindIII DNA size standards were All and E. coli replication proteins used were estimated to be greater than 95% pure. Purification and specific activities of the O and P replication proteins and the E. coli DnaJ, DnaK, DnaG primase, and DNA polymerase III holoenzyme proteins have been described previously (1). Some preparations of DNA pol IIIh (3 ϫ 10 5 units/mg) were purified from SK6776 (uvrD) essentially as described (29). DNA polymerase III* was purified by a modification of the published protocol (30), using a procedure developed by Dr. M. O'Donnell (Rockefeller University). E. coli SSB (7 ϫ 10 4 units/mg) was purified from strain RLM727 (HfrH/pRLM55) by a modification of the protocol of LeBowitz (25). Briefly, the blue dextran-Sepharose column was replaced by a ssDNA cellulose column prepared as described by Alberts (31). Approximately 57.5 mg of fraction II protein (Ͼ95% SSB) was diluted with buffer A (minus NaCl) to a conductivity equivalent to buffer A and applied at 3 column volumes/h to an ssDNA cellulose column (150 ml, 5.75 mg ssDNA/ml) that had been equilibrated with buffer A. The column was washed with buffer A containing 0.25 M NaCl until the absorbance at 280 nm approached zero. Bound SSB was eluted with buffer A containing 2 M NaCl and purified further by chromatography on hydroxyapatite as described (25). E. coli GrpE protein (6.4 ϫ 10 5 units/mg) was purified essentially as described by Zylicz et al. (32), except that the strain used was C600dnaK103/pJK23 (23).
E. coli DnaB protein was purified from 300 g of thermally induced RLM861 by a modification 2 of the protocol of Ueda et al. (33). Following the 0.24 ammonium sulfate and two 0.20 ammonium sulfate backwashes of the 0 -40% ammonium sulfate precipitate, the residual protein precipitate was resuspended in 18 ml of buffer A (fraction II, 27 ml, 1070 mg, 4.7 ϫ 10 7 units). A 4-ml portion of fraction II (160 mg of protein) was diluted 3-fold with buffer A and applied to a 75-ml ATP-Sepharose column that had been prepared by the technique of Lamed (34), except that the amount of adipic dihydrazide used as a linker ligand was reduced 10-fold to 0.9 g/100 ml. This modification resulted in lowered amounts of ATP covalently linked to the Sepharose matrix (typically 1 mol of ATP/ml of gel matrix), a reduction that correlated with improved stability of DnaB protein eluted from the ATP affinity column. 3 The ATP-Sepharose column was washed with 5 column volumes of buffer A, followed by 10 column volumes of buffer A containing 10 mM MgCl 2 and 10 mM AMP. DnaB protein was subsequently eluted with buffer A with 20 mM sodium pyrophosphate (fraction III, 11 mg, 2.3 ϫ 10 6 units). Fraction III was applied to a 7-ml DEAE-Sephacel column that had been equilibrated with buffer A. The column was 2 (27). M13mp19oriL was propagated in the E. coli strain 71-18; phage particles were isolated from the cell-free supernatant as described (35). Phage DNA was isolated free of contaminating protein and RNA by using a modification of the plasmid isolation protocol of Birnboim and Doly (36). Phage prepared from a 1-liter culture of infected E. coli, as described above, were resuspended in 5 ml of TE buffer and denatured by the addition of 10 ml of a solution containing 0.2 M NaOH and 1% SDS. Contaminating protein and RNA was precipitated by the addition of 7.5 ml 3 M sodium acetate, pH 4.8. Following centrifugation, the DNA remaining in the supernatant was precipitated by an equal volume of ethanol. The pEMBL130(ϩ) ssDNA prepared by this technique is suitable for digestion with restriction enzymes.
For some preparations of the tailed rolling circle DNA template, the circular single-stranded pEMBL130(ϩ) plasmid DNA was isolated free of helper phage DNA by sucrose gradient centrifugation. The DNA mixture was diluted with high salt gradient buffer and layered onto an 11-ml 5-30% sucrose gradient prepared in high salt gradient buffer (each tube contained a 1-ml cushion of 50% sucrose). The DNA samples were centrifuged in a Beckman SW41 rotor at 25,000 rpm for 16 h at 20°C. Fractions (0.5 ml) were collected from the bottom, and portions of each fraction were analyzed for the presence of DNA by neutral agarose gel electrophoresis. The fractions free of helper phage DNA were pooled, and the pEMBL130(ϩ) DNA was concentrated by ethanol precipitation. The DNA sample (350 g) was resuspended in 0.5 ml of TE buffer, adjusted to 0.1 M NaOH and applied to a 12-ml Bio-Gel A-15 M column that had been equilibrated with 0.1 mM Tris base. Fractions (0.5 ml) were collected and subsequently analyzed by neutral agarose gel electrophoresis. The fractions containing the peak of pEMBL130(ϩ) ssDNA were pooled, neutralized by the addition of 1 ⁄19 volume of 1 M Tris-HCl, pH 8, and precipitated with ethanol. The precipitate was collected by centrifugation and resuspended in TE buffer, and the purified circular ssDNA was used for the preparation of tailed DNA circles.
Synthesis and Isolation of the Rolling Circle DNA Template-The ssDNA 800-nucleotide primer (800-mer) used to form the 5Ј-tail of the TNC DNA template and helicase substrate was prepared by digesting ssDNA at restriction sites made duplex by the presence of hybridized oligonucleotides. 500 g of mp19oriL ssDNA was suspended in 3.2 ml of core buffer and mixed with 42.3 g of oligonucleotide RM34, 17.3 g of oligonucleotide RM35, PstI (600 units), and SacI (800 units); the mixture was incubated at 37°C for 60 min. The 800-mer was isolated free of linear M13mp19 DNA by neutral agarose gel electrophoresis and was recovered from the gel by electroelution using a Schleicher and Schuell Elutrap device. Some preparations of the 800-mer were treated with calf intestine phosphatase before purification by gel electrophoresis. This permits efficient end labeling of the linear strand of the helicase substrate/rolling circle DNA template by T4 polynucleotide kinase for preparation of a labeled size marker.
Hybridization of the 800-mer to the pEMBL130(ϩ) ssDNA (the 800mer contains at its 3Ј terminus a 42-nucleotide-long sequence that is complementary to the pEMBL130(ϩ) polylinker) was accomplished by mixing 250 g of circular ssDNA with 25 g of the 800-mer in TE buffer containing 0.5 M NaCl in a final volume of 1.25 ml. The mixture was incubated at 65°C for 2 min and then immediately supplemented with 1.37 mg of E. coli SSB and incubated for an additional 60 min at 37°C. The final DNA concentration was 0.14 mg/ml and the molar ratio of pEMBL130(ϩ) ssDNA to 800-mer was 2:1. The hybrid was diluted 7.67-fold into X EDB containing 2.5 mM ATP, 13 mM MgCl 2 , 2.5 mM spermidine chloride, 100 M each dATP, dCTP, dGTP, and dTTP, 16 g of DNA pol IIIh ␤-subunit, 81 g of SSB, and 67,500 units of DNA polymerase III* (X EDB replication mix). Some template batches were prepared using partially purified DNA pol IIIh ϳ(2 ϫ 10 5 units/mg; Ref. 29) instead of DNA polymerase III* and ␤-subunit. DNA synthesis was carried out during a 30-min incubation at 30°C and was terminated by chilling the mixture to 0°C. In some instances, radiolabeled TNC DNA template was prepared. On these occasions, the replication mixture contained [ 3 H]dNTPs or [␣-32 P]dNTPs at approximately 100 -200 cpm/ pmol of dNTP. The TNC DNA product was concentrated 10-fold using an Amicon Centricon 30 device. Replication proteins and unreplicated pEMBL130(ϩ) ssDNA were removed from the TNC DNA template by sucrose gradient centrifugation. The DNA sample was supplemented with SDS to 1% (w/v) and diluted with 0.25 volume of 5 ϫ high salt gradient buffer. Portions were layered onto neutral 5-30% sucrose gradients and centrifuged and analyzed as described above. The fractions containing double-stranded DNA were pooled, and the TNC DNA product was concentrated by ethanol precipitation.
For those instances in which helper phage DNA was first removed from pEMBL130(ϩ) ssDNA by sucrose gradient centrifugation before synthesis of the TNC DNA template, a simplified protocol was used for synthesis and purification of the template. The 800-mer was mixed with pEMBL130(ϩ) ssDNA at a molar ratio of 1.5 (800-mer:circle) and hybridized as described above. Once formed, the DNA hybrid was diluted into X EDB replication mix as described above, except that the mix contained 100,000 units of DNA pol IIIh instead of DNA polymerase III* and ␤-subunit. The TNC DNA replication product prepared in this manner had negligible amounts of contaminating single-stranded DNA. The TNC DNA sample was incubated for 4 h at 55°C with proteinase K (1 mg/ml in 0.5% N-lauroylsarcosine) and extracted three times in succession with neutralized phenol/chloroform/isoamyl alcohol (24:24: 1). DNA product in the aqueous phase was precipitated with ethanol, resuspended in TE buffer, and used directly as a substrate/template in rolling circle DNA replication assays.
DNA Replication Fork Movement Assays (Rolling Circle Replication (RCR) Assay and RCR Assay)-The reaction mixture (25 l) for the standard RCR assay (also denoted the minimal RCR system) contained 50 mM Hepes-KOH, pH 7.6, 10 mM magnesium acetate, BSA (200 g/ml), 10 mM dithiothreitol, 2.5 mM ATP, 42 fmol TNC DNA, 100 M concentration each of dATP, dCTP, dGTP, and TTP with [␣-32 P]dCTP at a specific activity of 100 -2500 cpm/pmol total deoxynucleotide or 172 M concentration each of dATP, dGTP, and dCTP and 72 M dTTP with [methyl-3 H]dTTP at a specific activity of 70 -100 cpm/pmol total deoxynucleotide, 200 units DNA pol IIIh, 0.043-1.9 g (0.14 -6 pmol, as hexamer) DnaB protein, and, where indicated, 2.2 g of SSB. Typically, the reaction mixtures for rolling circle DNA replication were assembled at 0°C; following the addition of dNTPs and DNA pol IIIh, DNA synthesis was initiated by a 30-min incubation at 30°C. In some experiments, however, DnaB was mixed with the TNC DNA template and 200 M AMP-PNP at 0°C, and the mixture was subsequently incubated at 30°C for 5 min to stimulate binding of DnaB to the ssDNA tail of the TNC template. The mixture was then returned to 0°C, supplemented with the other reaction components, including ATP, and incubated at 30°C for 30 min.
In certain other experiments, a third approach (denoted the RCR assay or system) was used to load DnaB onto the TNC DNA template. DnaB was assembled on the ssDNA tail of the TNC template using the phage and host proteins required for initiation of bacteriophage DNA replication (1,20). In these instances, the standard reaction mixture (25 l) contained: 53 ng of O, 80 ng of P, 86 ng of DnaB, 144 ng of DnaJ, 6.9 g of DnaK, 80 ng of GrpE, 200 units of DNA pol IIIh, and, where indicated, 2.2 g of SSB. To repress the intrinsic functional activities of DnaB (10), P and DnaB were mixed together (at a 2:1 molar ratio of P to DnaB) before the addition of the DNA template and other proteins. Reaction mixtures were assembled at 0°C, and rolling circle DNA synthesis was initiated by warming the reaction mixture to 30°C. For all types of rolling circle replication reactions, DNA synthesis was terminated by chilling the reaction mixture to 0°C. However, if the DNA replication reaction mixtures were to be analyzed for the sizes of newly synthesized DNA chains by electrophoresis in an alkaline agarose gel, DNA synthesis was terminated by the addition of EDTA to a final concentration of 50 mM. To conserve TNC DNA template, the reaction volumes and components were reduced by half in certain experiments. This had no detectable effect on the properties of the system. In these cases, the DNA synthesis values obtained were normalized (multiplied by 2) to permit direct comparison with results obtained with standard reactions.
The amount of DNA synthesis was usually determined by measuring the level of labeled deoxynucleotide incorporated into trichloroacetic acid-precipitable material, which was collected by filtration through glass fiber filters (Whatman, 934-AH) and then washed, dried, and counted in a liquid scintillation spectrometer as described previously (1). BSA (200 g) was added as a carrier to all reaction mixtures, before the addition of trichloroacetic acid, to aid precipitation of DNA. If the reaction volume was limiting, however, the amount of DNA synthesis was determined by spotting a portion of the reaction mixture onto Whatman DE81 DEAE-cellulose discs, which were washed and counted as described previously (37).
Where noted, SSB and/or primase was added to the various types of rolling circle DNA replication assays. When primase was included in the reaction mixture, CTP, UTP, and GTP were also added, each to a final concentration of 67 M. We measured the sensitivity of newly synthesized DNA to a nuclease specific for single-stranded DNA to determine whether the lagging strand had been synthesized when the reaction mixture also contained primase and rNTPs. Each reaction mixture was brought to 0.8% SDS and heated for 10 min at 65°C. S1 nuclease buffer (13 volumes) and 300 units of S1 nuclease were added to all samples, which were incubated for 30 min at 37°C and then analyzed for the presence of acid-insoluble radioactivity as described above.
Highly viscous DNA products were produced when primase or primase and SSB were present during rolling circle DNA synthesis that was initiated with the assistance of the O and P proteins. Such DNA products, if labeled with 3 H, needed to be digested partially with DNase I to obtain optimal counting efficiencies. Following completion of DNA synthesis, reaction mixtures were supplemented with DNase I (0.001 unit/25 l of reaction mixture) and incubated for an additional 5 min at 30°C before analysis.
The efficiency of initiation of rolling circle DNA replication was determined using radioactively labeled TNC DNA as the template. DNA replication was carried out as described above using unlabeled dNTPs. Replication products were resolved from unutilized template DNA by electrophoresis in a neutral agarose gel. Following electrophoresis, the gel was dried onto Whatman DE 81 paper and autoradiographed (see below). The band containing the TNC DNA template and the band containing tailless nicked circular DNA were excised, using the autoradiograph as a guide. The remainder of each lane was cut into 8 equivalent segments. The bottom 1-cm segment had no detectable DNA product and was used to determine the background level of radioactivity (B). The individual segments of paper were counted in a liquid scintillation spectrometer. The amount of template utilized at each time point was calculated by the following formula: where S represents the amount of label in the TNC DNA band, B represents the amount of background label, and T represents the total amount of label, corrected for background, in the entire lane. The proportion of tailless nicked circular DNA in each lane was also determined.
Electrophoresis of DNA-DNA from restriction digests and products of rolling circle DNA replication were electrophoresed under native conditions in 1% agarose gels (13.5 cm long) in TGE buffer. Electrophoresis was carried out for 4 h at 4.25 V/cm. If a specific DNA species was to be recovered by electroelution, the agarose gel was stained briefly with TGE buffer containing 5 g/ml ethidium bromide, and the desired DNA band was visualized under 300-nm ultraviolet light and excised. DNA was electroeluted from the agarose slice in TGE buffer in an Elutrap apparatus (Schleicher and Schuell) as detailed by the manufacturer. For autoradiography, agarose gels were dried onto DEAEcellulose paper (Whatman DE81) in a slab gel dryer (Hoefer Scientific Instruments) and then exposed to Kodak XAR-5 x-ray film at Ϫ80°C in the presence of a DuPont Cronex Lightning-Plus intensifying screen.
The chain lengths of replication products were determined from their electrophoretic mobilities in 30-cm alkaline agarose gels. The gels contained 30 mM NaOH, 2 mM EDTA, 0.1% SDS, and 0.4% agarose. The DNA samples were denatured by the addition of alkaline gel sample buffer (0.25 volume) and were electrophoresed at room temperature at 1.6 V/cm for 60 h. The gels were transferred immediately onto DEAEcellulose paper, without neutralization or staining, and dried and exposed to film as described above. The sizes of the largest ssDNA molecules in a population were determined by measuring the relative mobility (R m ) at the back edge of the slowest migrating products relative to an internal standard and comparing the R m obtained with the relative mobilities of known molecular weight standards present on the same agarose gel. The internal standard used in these experiments was pEMBL130(ϩ) DNA that had been linearized with BamHI and endlabeled with 32 P.
Other Methods-DNA concentrations were measured by absorbance at 260 nm in a spectrophotometer or by a dye-binding assay, using Hoechst 33258, in a TK100 DNA fluorimeter (Hoefer Scientific Instruments) carried out according to the manufacturer's instructions. The DNA molecular weight markers and the DNA internal standard used for the alkaline agarose DNA gels were treated with calf intestine phosphatase followed by end labeling with T4 polynucleotide kinase and [␥-32 P]ATP (38).

RESULTS
Construction of a DNA Template for Assessment of Replication Fork Movement-During the initiation of bacteriophage DNA replication, the phage-encoded O and P replication pro-teins participate in the assembly of the replication fork apparatus by transferring one or more molecules of DnaB helicase onto the viral chromosome. To investigate if the O and P proteins have any detectable effects on replication forks that they help establish, we decided to compare the functional properties of such forks with those of a minimal replication fork apparatus composed of DnaB helicase and DNA pol IIIh. We chose to use TNC DNA molecules as the replication template in these studies, since Mok and Marians (39) had earlier shown their effectiveness as templates for the establishment of rolling circle DNA replication accomplished by the precisely coordinated action of the helicase activity of DnaB and the DNA synthetic activity of DNA pol IIIh. The TNC DNA template used throughout this study is a 4.0-kb nicked circular duplex DNA carrying a 758-nucleotide single-stranded DNA tail at the 5Ј terminus of the nicked strand. It was prepared in several steps. Initially, an 800-nucleotide ssDNA fragment was released from M13mp19oriL DNA. This was accomplished by digestion of this single-stranded circular phage DNA with PstI and SacI after first converting the unique recognition sequences for these restriction nucleases into a double-stranded form by the hybridization of a short complementary oligonucleotide. The 800-mer ssDNA fragment was isolated and hybridized to pEMBL130(ϩ) DNA via a 42-nucleotide region at the 3Ј terminus of the fragment that is complementary to the polylinker sequence present on pEMBL130. The primed ssDNA circle that results was converted to a duplex form, the TNC DNA template, by replication with DNA pol IIIh in the presence of SSB.
Transfer of DnaB Helicase onto the TNC DNA Template before the Initiation of Rolling Circle DNA Synthesis-An important early step in the establishment of rolling circle DNA synthesis on a TNC DNA template in our in vitro system is the assembly of a molecule of DnaB helicase on the 5Ј-ssDNA tail carried by the TNC template. This ssDNA tail was purposely made long to maximize the opportunity for DnaB to interact with the TNC template. Once transferred onto the (Ϫ)-strand ssDNA tail, DnaB will, in the presence of ATP, translocate along the strand in the 5Ј 3 3Ј direction and initiate unwinding of the duplex portion of the TNC DNA template (14). This unwinding mediated by DnaB can be coupled by protein-protein interactions (40) to DNA chain elongation from the 3Ј terminus of the (Ϫ)-strand by DNA pol IIIh, producing processive and rapid rolling circle DNA replication ( Fig. 1) (39).
A large molar excess of DnaB over template is required for optimal transfer of DnaB from solution onto ssDNA under standard replication conditions (14). Therefore, we initially explored various approaches for assembling DnaB helicase molecules on the 800-base ssDNA tail of the TNC DNA template, since the relative efficiency of this step might determine the overall efficiency of rolling circle DNA replication on the TNC DNA template. In the standard assay, DnaB is simply mixed with the DNA template in the presence of ATP at 0°C (Fig. 1A), and DNA synthesis is initiated by the addition of DNA pol IIIh and dNTPs and warming of the reaction mixture to 30°C. In some experiments, we included 200 M AMP-PNP in the DnaB-binding step, since it has been reported that the binding of DnaB to ssDNA is increased when AMP-PNP is substituted for ATP (41). Although AMP-PNP is incapable of supporting the helicase activity of DnaB, it does not interfere significantly with the binding of ATP to DnaB (14,42). Thus, once DnaB is bound to ssDNA in the presence of AMP-PNP, the addition of sufficient quantities of ATP permits helicase function by DnaB and the establishment of the elongation phase of the rolling circle replication reaction (Fig. 1A).
An alternate pathway for the transfer of DnaB onto ssDNA or TNC DNA templates involves the use of the O and P proteins (19,20). This multistep transfer reaction is relatively independent of DNA sequence but does require the action of the DnaK/DnaJ/GrpE molecular chaperone system. Additionally, unlike the binding of DnaB alone, the O-and P-mediated transfer reaction (Fig. 1B) is not blocked by stoichiometric coating of ssDNA by SSB (19).
In the absence of DnaB, only very modest amounts of DNA synthesis were obtained (about 50 pmol) when the TNC DNA template was incubated with DNA pol IIIh and dNTPs alone (Fig. 2). Extensive DNA synthesis was obtained, however, when DnaB helicase was included along with DNA pol IIIh in the reaction mixture (Fig. 2, squares). Maximal DNA synthesis under these conditions occurred when the reaction mixture contained approximately 60 molecules of DnaB hexamer/molecule of DNA template. This is essentially identical to the DnaB stoichiometry required for maximal strand displacement in the standard helicase assay (14). Preincubation of DnaB with the TNC DNA template for 5 min at 30°C in the presence of AMP-PNP had little noticeable effect on the level of DnaB required to saturate the RCR assay (data not shown).
Significantly reduced amounts of DnaB were found to be required for saturation of the RCR assay when DnaB was transferred onto DNA with the aid of the O and P replication initiation proteins (Fig. 2, open and closed circles). For simplicity, the RCR assay that uses the O and P proteins to load DnaB onto the TNC template will be referred to as the RCR assay. In such assays, DnaB was first converted into a P⅐DnaB complex with the addition of excess P protein before its addition to the TNC DNA. This mixture was subsequently supplemented with O, the DnaK, DnaJ, and GrpE molecular chaperones, and DNA pol IIIh. Preincubation of DnaB with P eliminates the possibility that free DnaB helicase binds to the TNC DNA (10) and contributes to the observed DNA synthesis.
Although the reaction mixtures contained substantially fewer DnaB molecules when the RCR assay was used, maximal DNA synthesis was found to increase 2-3-fold (Fig. 2, open and closed circles). Nevertheless, the presence of excess P⅐DnaB in the RCR assay was found to be strikingly inhibitory. In contrast, excess DnaB was not harmful to the standard RCR assay (Fig. 2, squares). The presence of E. coli SSB in the RCR assay reproducibly stimulated maximal DNA synthesis by approximately 50% (Fig. 2, closed circles).
DNA synthesis in the RCR assay requires, in addition to the P⅐DnaB complex, O, DNA pol IIIh, and the E. coli DnaJ and DnaK heat shock proteins (Table I). GrpE, a member of the DnaK molecular chaperone system, stimulates DNA synthesis in the RCR assay (Table I) but becomes absolutely essential in both this assay and the SS replication system (20) only at lower DnaK concentrations (data not shown). These latter findings are consistent with earlier observations that GrpE is not required for initiation of DNA replication in vitro unless DnaK concentrations are lowered to suboptimal levels (2,4).
Kinetics of Rolling Circle DNA Synthesis-We examined the influence of various reaction conditions on the time course of DNA synthesis in the RCR assay. Previous studies have demonstrated that DNA pol IIIh has negligible strand displacement activity, displacing no more than 1-5 nucleotides (43) when it encounters in its path a 5Ј terminus of a polynucleotide chain base paired to the template strand. Not surprisingly, minimal DNA synthesis was obtained when DNA pol IIIh alone was incubated with the TNC template in the presence of dNTPs and ATP for 1 h (Fig. 3A, open circles). The inclusion of SSB in FIG. 1. Schematic diagrams of the rolling circle DNA replication assays used to study replication fork movement. Rolling circle replication reactions were performed as described under "Experimental Procedures." A, standard RCR assay. Reaction mixtures containing DnaB helicase, DNA pol IIIh, ATP, and dNTPs were assembled on ice, and DNA replication was initiated by warming the mixture to 30°C. In certain instances, DnaB helicase was incubated with 200 M AMP-PNP and TNC template at 30°C for 5 min to permit binding of DnaB to the single-stranded DNA tail before the addition of DNA pol IIIh, ATP, and dNTPs. B, the RCR assay. The O and P replication proteins and the E. coli DnaJ, DnaK, and GrpE molecular chaperones were used to transfer DnaB helicase onto the tail of the TNC DNA template (in either the presence or absence of SSB) before the initiation of rolling circle DNA replication. For each type of assay, the first round of leading strand DNA synthesis, accompanied by DnaB-mediated displacement of the 5Ј-tail, is pictured. The replication fork apparatus is shown as consisting of a molecule of DnaB helicase (hexameric ring), which translocates 5Ј 3 3Ј along the linear strand and interacts with a molecule of DNA pol III that is simultaneously synthesizing the growing DNA chain as it moves 3Ј 3 5Ј along the circular template strand. Newly synthesized DNA is represented by a thicker strand.

FIG. 2.
Optimization of the levels of DnaB and the P⅐DnaB complex in the RCR and RCR assays, respectively. RCR assays were performed as described under "Experimental Procedures," except that the amount of DnaB, or, in the case of the RCR assay, the amount of the P⅐DnaB complex, was varied as indicated. DNA synthesis was measured after a 30-min incubation at 30°C. f--f, RCR assay with no DnaB preincubation; E--E, RCR assay in the absence of SSB; q--q, RCR assay in the presence of SSB. the reaction mixture, however, stimulated DNA synthesis significantly (Fig. 3A, closed circles). To determine whether DNA pol IIIh has an intrinsic strand displacement activity in the presence of SSB or is simply filling in gaps in the TNC template, the replication products produced at early time points in the presence and absence of SSB were analyzed by alkaline agarose gel electrophoresis. In the absence of SSB, DNA pol IIIh appears to label the linear strand of the TNC template by an idling reaction at the primer termini. Labeled DNA strands of 5.5 and 4.7 kb were detected, but no significant extension of these strands occurred during the 1-min incubation (Fig. 4A). When SSB is added, a modest strand displacement DNA synthesis activity is seen (Fig. 4B) that produces a fork movement rate of approximately 50 bp/s (Fig. 4C). It is apparent that the strand displacement activity of DNA pol IIIh observed in the presence of SSB is relatively feeble. DNA chains with a length of approximately 5.3 kb accumulate during the incubation, an indication that the DNA polymerase may pause for extensive periods at certain sites or possibly even terminate DNA synthesis at such sites. The 8.8-kb band observed in Fig. 4B may result from DNA polymerase-mediated end-labeling at the 3Ј termini of broken M13K07 helper phage DNA molecules (8.7 kb), which are present in trace amounts in the TNC preparation.
It has been found that many preparations of DNA pol IIIh contain significant amounts of DNA helicase II (UvrD protein) (44). For this reason, we determined if the strand displacement activity observed when DNA pol IIIh and SSB were incubated with the TNC template resulted from UvrD helicase contamination of the DNA pol IIIh preparation. This was found not to be the case. The strand displacement activity obtained with DNA pol IIIh purified from an E. coli strain carrying a deletion of the uvrD gene (SK6776) was essentially identical to that displayed by DNA polymerase isolated from wild type cells (data not shown).
The effect of DnaB helicase on the time course of DNA synthesis by DNA pol IIIh on the TNC template is illustrated in Fig. 3B. If DnaB is permitted to bind to the ssDNA tail of the TNC template before the addition of DNA pol IIIh, DNA synthesis remains linear for approximately 10 min before slowing almost to a halt 30 min after initiation (Fig. 3B, open circles). In contrast, if SSB is added to the reaction mixture after DnaB has first been permitted to bind to the TNC template, DNA synthesis was much more extensive and was linear for a more prolonged period of time (Fig. 3B, closed circles). Nonetheless, SSB also demonstrated an inhibitory effect on RCR DNA synthesis when it was added simultaneously with DnaB to the TNC template (Fig. 3B, squares). This inhibition presumably reflects the capacity of SSB to bind rapidly to the ssDNA tail of the TNC template and interfere with the loading of DnaB onto the DNA chain (45).
We examined the effect of a preincubation of DnaB with the TNC template on the early DNA synthesis kinetics in the RCR assay. Preincubation of DnaB with the template in the presence of the ATP analogue AMP-PNP permitted constant DNA synthesis from the earliest time points after the initiation of DNA replication (Fig. 5, open circles). On the other hand, a striking lag in the kinetics of DNA synthesis was observed when no nucleotide was present during the DnaB preincubation with the template (Fig. 5, closed circles). We infer that transfer of DnaB onto the TNC template and its activation as a helicase is relatively slow, since preincubation of the TNC template with DnaB and AMP-PNP provided for more synchronous establishment of maximal DNA synthesis rates in the RCR assay.
We also examined the effect on the DNA synthesis kinetics of using the O and P proteins to transfer DnaB onto the TNC template. The kinetics of DNA synthesis in the RCR assay, carried out in the presence and absence of SSB, is depicted in Fig. 3C. In the absence of SSB (Fig. 3C, open circles), DNA synthesis proceeds in a linear fashion for approximately 30 min after a short initial lag. Both the rate and extent of DNA synthesis in the RCR assay are elevated if SSB is present, and the early lag is less prominent (Fig. 3C, closed circles).
Rates of Replication Fork Movement in the RCR and RCR Assays-The foregoing results indicate that the kinetics and extent of DNA synthesis in the rolling circle DNA replication assays are dependent upon the manner by which the DnaB helicase is loaded onto the TNC template. In view of this, we examined if the mechanism used to load DnaB affected the rate of DNA chain elongation during rolling circle DNA replication. We first compared the rate of DNA chain elongation obtained when DnaB was loaded onto the TNC template with the aid of AMP-PNP (Fig. 6A) with the rate yielded when the O and P replication proteins were used to deliver DnaB onto the template (Fig. 6B). The sizes of product DNA chains were estimated from their electrophoretic mobilities in alkaline agarose gels as compared with the mobilities of DNA chains of known length. To assess the maximal rate of DNA chain elongation, we measured the sizes of the longest DNA chains in each population of replication products and plotted these sizes against the reaction time (Fig. 6C). Analysis of this data indicated that the fork movement rate was 720 -750 base pairs/s at 30°C in both the RCR assay (Fig. 6A) and the RCR assay (Fig. 6B). We conclude that the rate of replication fork movement is independent of the manner used to load DnaB onto the DNA.
We also examined the effect of SSB on the rate of replication fork movement in both the RCR assay and the RCR assay.
Our results indicate that SSB has no discernible effect on the movement rate of replication forks in either type of assay (Fig.  6C). Nevertheless, we did find with the RCR assay that the addition of SSB to the TNC template before the addition of DnaB helicase strongly inhibited DNA synthesis. Furthermore, replication forks assembled in this manner moved at the slow rate (40 -50 bp/s) observed when DNA pol IIIh and SSB were present but DnaB was absent ( Fig. 4B and data not shown). This result was anticipated, since complete coating of the single-stranded DNA tail with SSB would be expected to block loading of the DnaB helicase onto the TNC template (19,45).
Effect of DNA Helicase II (UvrD) on Assembly and Movement of Replication Forks-DNA pol IIIh, the replicative DNA polymerase used in the RCR and RCR assays, is a complex enzyme consisting of at least 10 subunits (30,46,47). As mentioned earlier, it has been reported that many preparations of DNA pol IIIh contain significant levels of DNA helicase II, the product of the E. coli uvrD gene (44). Western blot analysis FIG. 4. Alkaline agarose gel electrophoretic analysis of DNA chain elongation by DNA pol IIIh in the presence and absence of SSB. A and B, replication fork movement assay mixtures were assembled as described under "Experimental Procedures" (using [␣-32 P]dCTP), except that DnaB helicase was omitted and, for B only, 1.1 g of SSB was present per 25 l of reaction volume. For these experiments, the required reaction components were initially split between two mixtures, with the TNC DNA template in one mixture and the DNA pol IIIh, SSB (when present), ATP, and dNTPs in the other. Both mixtures were prewarmed to 30°C and then mixed to initiate DNA replication. The combined mixture was incubated at 30°C. At 10, 20, 30, 40, 50, and 60 s (lanes 2-7, respectively) following mixing, 7-l portions were removed, mixed with an equivalent volume of alkaline gel sample buffer, and electrophoresed through a 0.8% alkaline agarose gel. Lane 1, /HindIII DNA molecular weight standards (end-labeled with 32 P). C, determination of the rate of strand displacement DNA synthesis catalyzed by DNA pol IIIh in the presence of SSB. The sizes of the longest DNA chains synthesized from the TNC template in B (not including the labeled 8.8-kb product) were determined and plotted versus incubation time. The rate of chain elongation was calculated from a linear regression analysis of the data. of our DNA pol IIIh preparation with antibody specific for UvrD confirmed the presence of DNA helicase II and indicated further that approximately 20 -40 ng of UvrD was being added along with DNA polymerase IIIh to individual RCR assays. It was of interest, therefore, to determine if the presence of UvrD helicase in the RCR assays influenced the results reported here. We purified DNA pol IIIh from an E. coli strain carrying a deletion of the uvrD gene and assessed the rates of replication fork movement when the UvrD-free DNA pol IIIh served as the replicative DNA polymerase in the RCR and RCR assays. We found that removal of UvrD from the DNA pol IIIh preparation had no detectable effect on fork movement rates (data not shown). However, the two DNA pol IIIh preparations did behave differently in one respect. Preincubation of DnaB with the TNC template in the presence of AMP-PNP resulted in a 2-or 3-fold stimulation of DNA synthesis in the standard RCR assay (30-min incubation) when the DNA pol IIIh was from uvrD ϩ cells (data not shown; also see Fig. 5). In contrast, the preincubation of DnaB with the TNC template and AMP-PNP did not stimulate DNA synthesis when UvrD Ϫ DNA pol IIIh was used ( Fig. 7; compare closed and open circle points with no added UvrD). Similarly, the absence of UvrD in the DNA pol IIIh preparation was associated with a nearly complete elimination of the lag in early DNA synthesis kinetics previously seen (Fig. 5) when this AMP-PNP preincubation was omitted (data not shown). These results suggest that loading of the DnaB helicase onto the TNC template may be inhibited by the presence of UvrD in the DNA pol IIIh preparation. To examine this possibility, we determined the effect of adding purified UvrD protein to rolling circle replication reactions carried out by UvrD Ϫ DNA pol IIIh (Fig. 7). In control experiments, UvrD stimulated DNA synthesis only slightly in the absence of DnaB (Fig. 7, squares). Added UvrD, however, had a striking inhibitory effect on DnaB-dependent RCR DNA synthesis (Fig. 7, plotted against the reaction time. A, q, RCR assay described for panel A, which included a preincubation of DnaB with the TNC template in the presence of AMP-PNP; E, RCR replication as described for panel A, except that a lower concentration of DnaB was used (1 pmol of DnaB/25 l of reaction volume), and DNA chain elongation was carried out in the presence of SSB (2.2 g/25 l of reaction volume) (gel not shown); f, RCR assay as described under "Experimental Procedures" (gel not shown); Ⅺ, RCR assay described for panel B, in which SSB (2.2 g/25 l of reaction volume) was present during DNA chain elongation. open circles). This inhibitory effect of UvrD was partially ameliorated if DnaB was allowed to bind to the tailed DNA template, during a preincubation with AMP-PNP, before the addition of UvrD to the RCR reaction (Fig. 7, closed circles). Thus, the severalfold stimulatory effect noted above of the AMP-PNP preincubation on RCR DNA synthesis would be explained if the DNA pol IIIh (from uvrD ϩ cells) added to the reaction mixture contained between 20 and 100 ng of UvrD protein.
Effect of Primase on DNA Synthesis Levels and Fork Movement Rates in the RCR Assay-Previous studies have demonstrated that primase, the product of the E. coli dnaG gene, primes synthesis of all DNA strands during replication of the chromosomes of both E. coli and phage (48). Synthesis of RNA primers is highly dependent on transient interactions of primase with DnaB helicase bound to the template strand (15,45,49,50). Accordingly, it was of interest to examine the effect of primase on DNA synthesis levels and fork movement rates in the RCR and RCR assays. As expected, in the absence of DnaB, primase had no effect on the levels of DNA synthesis obtained in the RCR and RCR assays, whether or not SSB was present (Table II, lines 1-4 and 9 -12). We conclude that primase does not affect the strand displacement activity of DNA pol IIIh. SSB stimulated DNA synthesis in the RCR assay when DnaB was absent (Table II, lines 1 and 3), as noted earlier (Fig. 4). On the other hand, in the case of the RCR assay, SSB failed to stimulate DNA synthesis when DnaB was absent (Table II, lines 9 and 11). Additional studies indicated that the presence of O was responsible for blocking the SSB-mediated stimulation of strand displacement DNA synthesis in this situation (data not shown).
The TNC template used in the RCR assays has a 3Ј-primer terminus that can be used to prime leading strand DNA synthesis. Hence, it would be anticipated that primase would contribute primarily to synthesis of the lagging strand Okazaki fragments and would produce, ideally, about a 2-fold elevation in total DNA synthesis. This was roughly the result obtained when primase was included in the RCR assay (Table II, lines [13][14][15][16]. But in the standard RCR assay, the degree of stimulation elicited by the presence of primase ranged from 1.2-to 3.8-fold, depending on whether SSB was also present (Table II, lines [5][6][7][8]. Interpretation of these effects of primase on the RCR assays is complicated by the fact that SSB alone stimulated strand displacement synthesis (see below).
To determine if the presence of primase permits replication of the leading strand DNA produced in the RCR and RCR assays, we treated the DNA products with a single strandspecific nuclease, S1 nuclease, as described under "Experimental Procedures." S1 nuclease digestion converted 80 -90% of the DNA products into an acid-soluble form in those assays where primase was not included in the reaction mixture (data not shown). In contrast, the DNA products of RCR and RCR assays that contained primase were 80 -100% resistant to S1 nuclease digestion. We conclude that primase permits synthesis of duplex DNA products in the RCR and RCR assays.
Alkaline gel electrophoresis was used to analyze the chain lengths of products made at early times in the RCR assay when primase was present. The DNA chains synthesized in the presence of primase (Fig. 8A) resembled those made in its absence (Fig. 6A), except that the fraction of DNA chains shorter than 6 kb in length increased markedly. The production of smaller fragments is another indication that primase action initiates the synthesis of Okazaki fragments on the leading-strand DNA chain produced by rolling circle DNA replication. Thus, it is likely that the DnaB helicase component of the replication fork apparatus must repeatedly interact with primase every few seconds to permit initiation of RNA primer synthesis (51). We examined if these multiple protein-protein interaction events affect the rate of replication fork movement during rolling circle DNA replication. The lengths of the longest chains present at each time point were determined and used to calculate the rate of replication fork movement (Fig. 8B). In the presence of primase, the rate of replication fork movement was measured at 650 bp/s, whether or not SSB was included in the reaction mixture. An identical rate was obtained for replication fork movement in the absence of primase with these preparations of DNA pol IIIh and TNC DNA template (data not shown). Thus, the repeated interactions between primase and the DnaB helicase component of the replication fork apparatus have no discernible effect on the net rate of movement of the replication fork apparatus.
Kinetics and Efficiencies of Template Utilization in the RCR and RCR Assays-Our results indicate that each SSB and primase stimulate overall DNA synthesis, both for the RCR assay and the RCR assay. This increased DNA replication could result from a stimulation in the efficiency of utilization of the TNC template, from enhanced propagation of replication forks following initiation, or from both. To decide between these possibilities, we measured both the kinetics and efficiencies of utilization of the input TNC template molecules for a variety of protein mixtures. The TNC template molecules used in these template utilization experiments were labeled with 32 P, which was accomplished by supplementing the reaction mixture used TABLE II Effects of SSB and primase on the level of rolling circle DNA replication All RCR assays that contained DnaB included a preincubation of DnaB with the TNC template DNA in the presence of AMP-PNP as described under "Experimental Procedures." Where indicated, primase (100 ng) was also included in the preincubation mixture. SSB (2.2 g) was added, where specified, to each mixture just before the initiation of DNA synthesis. For the RCR assays, primase (100 ng) and/or SSB (2.2 g) was added, where indicated, along with the other required proteins and mixed with the TNC DNA template and incubated for 5 min at 30°C in the absence of ATP. DNA synthesis was initiated by the addition of ATP. All reaction mixtures also contained rNTPs (67 M each), which were added just before the initiation of DNA synthesis. All reaction mixtures were incubated for 15 min at 30°C and then processed to determine the level of DNA synthesis. The values listed represent the average of duplicate samples. The average length of the DNA product molecules is calculated by dividing the amount of DNA synthesis by the total amount of template utilized in the assay (ϳ0.029 pmol of TNC DNA template; e.g. see Fig.  10).
b The average DNA product length was calculated with the assumption that the lagging strand is being synthesized (see "Experimental Procedures" for details).
c This sample was treated briefly with DNase I as described under "Results." for synthesis of the complementary linear strand with [␣-32 P]dCTP. For all experiments that used radiolabeled DNA templates, rolling circle replication was carried out with unlabeled dNTPs.
Replicated TNC template molecules were resolved from unreplicated template molecules by electrophoresis through neutral agarose gels. The input TNC template DNA migrated as two distinct subpopulations (Fig. 9, A and B, lane 1). The upper band and the broad smear below it, labeled S in Fig. 9, A and B, apparently consist of nicked circular or gapped circular template molecules with single-stranded DNA tails. The lower band and the faint smear below it (labeled NC in Fig. 9, A and  B) apparently are composed of nicked or gapped circular molecules that have lost their single-stranded tails during handling. The latter molecules, because they do not contain a single-stranded DNA tail, should not be effective substrates for DnaB helicase action (14). Our results are fully consistent with this conclusion, since for all reaction conditions tested, the amount of radioactivity present in the region labeled NC remained relatively constant during the time course of the rolling circle replication reaction ( Fig. 9 and data not shown). In the presence of DnaB helicase, the replication products made in the  9. Kinetics of template utilization in rolling circle replication assays as analyzed by neutral agarose gel electrophoresis. Rolling circle replication assay reaction mixtures (62.5 l each) were assembled using 32 P-labeled TNC DNA template (ϳ2 ϫ 10 5 cpm) and unlabeled dNTPs. A, the RCR assay mixture contained DNA pol IIIh and SSB (2.75 g) but no DnaB helicase. B, the RCR assay mixture contained the following proteins: O, P, DnaB, DnaJ, DnaK, GrpE, and DNA pol IIIh as described under "Experimental Procedures," except that the amounts of each protein were increased 2.5-fold. A 5-l portion was removed from each reaction mixture before the initiation of DNA synthesis and mixed with an equal volume of 0.1 M EDTA (lanes 1). Portions (5 l) of the reaction mixture were removed 10 s, 1 min, 2 min, 3 min, 5 min, 10 min, 15 min, and 30 min (lanes 2-9, respectively) after the initiation of DNA replication and mixed with EDTA to terminate DNA synthesis. The samples were electrophoresed in 1% neutral agarose gels as described under "Experimental Procedures" and visualized by autoradiography. P, rolling circle DNA replication products; S, TNC DNA template; NC, nicked circular or gapped circular duplex DNA that contains no ssDNA tail or a highly truncated ssDNA tail; W, well. reaction mixture, a nearly identical profile of DNA chain length distribution was obtained (data not shown). B, determination of the rates of replication fork movement in an RCR assay containing primase (OE) or primase and SSB (E). The lengths of the longest DNA chains in the time course depicted in Fig. 8A and for a similar time course carried out in the absence of SSB were determined. RCR and RCR assays moved very slowly during neutral agarose gel electrophoresis. Whether or not SSB was present in the reaction mixture, a substantial fraction of the product rolling circle molecules accumulated near the well for both the RCR assay (Fig. 9B) and the standard RCR assay (data not shown). The low electrophoretic mobilities of these molecules presumably reflect their tremendous size ( Fig. 6B and Table  II). In the absence of DnaB helicase, however, the TNC DNA molecules replicated by the combination of DNA pol IIIh and SSB migrated more quickly during electrophoresis (Fig. 9A), as expected from the smaller sizes of leading strand products produced under these conditions ( Fig. 4B and Table II). These rolling circles migrated as a broad smear (labeled P in Fig. 9A) behind the unreplicated TNC template molecules, but it is notable that all of the labeled product molecules migrated substantially into the gel. It is also apparent from the electrophoretic mobilities of the labeled TNC templates/products at early time points that the TNC templates were replicated somewhat more synchronously when just DNA pol IIIh and SSB were present in the reaction mixture (Fig. 9A) as compared with when strand displacement was mediated by DnaB helicase, as in the RCR assay (Fig. 9B) or in the standard RCR assay (data not shown).
The proportion of TNC template utilized in various rolling circle replication assays was calculated by determining the amount of radiolabeled template remaining at specific time points that comigrated with the input template DNA. The results, depicted in Fig. 10, indicate that a minimum of 60 -70% of the starting template molecules were utilized as substrate within the initial 5 or 10 min of incubation under a variety of RCR assay conditions. This is a minimum estimate for template utilization, since we have not corrected for the presence of damaged and apparently unproductive tailless template molecules in the preparation (which accounted for about 25% of the radiolabel in the template preparation). However, a clear ex-ception to this general trend of efficient template utilization was obtained when DNA pol IIIh was the only protein added to the replication assay mixture. In this type of assay, template utilization was very poor, peaking at approximately 10%, consistent with the feeble DNA synthesis observed under these conditions (Fig. 3A). The kinetic analysis (Fig. 10) demonstrated that the half-time for template utilization was about 1 min when DNA pol IIIh and SSB were present (open circles) or when DnaB helicase and DNA pol IIIh were included in the RCR assay mixture (open squares). Significantly slower kinetics of template utilization were obtained in the latter situation, however, when the preincubation of DnaB with the TNC template in the presence of AMP-PNP was omitted (Fig. 10, closed  squares). Surprisingly, in the presence of ATP, the O and P replication proteins and the DnaK/DnaJ/GrpE molecular chaperone group cooperated to load DnaB helicase onto the TNC DNA template and initiate DNA replication somewhat more quickly than when DnaB alone was present ( Fig. 10; compare closed triangles and closed squares). Initiation of replication of the TNC template in the RCR system, in fact, lagged only slightly behind the initiation rate found when DnaB helicase was prebound to the ssDNA tail of the TNC template in the presence of AMP-PNP. Finally, we determined that the addition of primase alone, SSB alone, or primase and SSB together had no discernible impact on either the kinetics or the efficiencies of template utilization in the standard RCR assay (DnaB and DNA pol IIIh) or the standard RCR assay (data not shown).
Processivity of the Replication Fork Apparatus during Rolling Circle Replication-The data presented in this report demonstrate that the efficiencies and kinetics of template utilization in the RCR and RCR assays are similar and that the replication forks established by each system on the TNC template synthesize DNA at the maximum rate of 650 -750 bases/s. Thus, the variations in total DNA synthesis yielded after a 15-min incubation by the various protein mixtures that contain DnaB (Table II) must therefore reflect a difference in the average amount of DNA synthesis obtained per primary initiation event. Since it is not possible to obtain an accurate determination of the size of DNA chains longer than 50 kb by alkaline agarose gel electrophoresis, we estimated the average chain length of the leading strand produced by rolling circle DNA replication under each condition. This was accomplished by calculating the amount of DNA synthesis that was obtained for each TNC template molecule that was utilized in the assay (Table II). When both primase and DnaB were present in the assay mixture, the chain length values obtained were halved to compensate for the presence of lagging strand DNA synthesis. These values are minimal estimates of the average leading strand chain length, since they have not been corrected for the fact that template utilization was asynchronous and was not completed until about 5 min after initiation of the replication reaction (Fig. 10). The longest leading strand DNA chains, averaging 160 -200 kb, were produced in the RCR assay when SSB was present (Table II, lines 15 and 16). We also found that the presence of SSB or primase in the minimal RCR assay resulted in the synthesis of longer leading strands (Table II,  lines 5-7). DISCUSSION In this report we have characterized the properties of the replication fork apparatus assembled under the direction of the phage -encoded O and P replication proteins. We have taken advantage of the capacity of these viral proteins, assisted by the DnaK/DnaJ/GrpE molecular chaperone system, to mediate the transfer of E. coli DnaB helicase onto the ssDNA tail of a TNC DNA template. As previously shown by Mok and Marians FIG. 10. Time courses of template utilization in various RCR and RCR assays. RCR assays and RCR assays were performed, essentially as described in the legend to Fig. 9, and the DNA products present at the times indicated were resolved from the labeled TNC DNA template by electrophoresis in 1% neutral agarose gels. The amount of radioactive template DNA present at each time point was determined and used to calculate the percentage of TNC template utilized (see "Experimental Procedures"). The results were normalized to 0% template utilized at time 0. The first two measurements were taken 10 s and 1 min after the initiation of DNA replication. q, RCR assay carried out in the absence of DnaB and SSB; E, RCR assay carried out in the absence of DnaB (contained 1.1 g of SSB/25 l of reaction volume); f, standard RCR assay (no preincubation) containing 1.35 pmol of DnaB/25 l of reaction volume; Ⅺ, RCR assay (1.35 pmol of DnaB/25 l of reaction volume) with a 5-min preincubation of DnaB and the TNC template at 30°C in the presence of AMP-PNP; OE, RCR assay carried out in the absence of SSB. (39) and confirmed here, the presence of DnaB helicase is essential to the assembly of a robust replication fork apparatus. We have demonstrated that the replication fork apparatus assembled in the foregoing manner is capable of carrying out extensive rolling circle DNA replication. Replication products averaging up to 190 kilobases in length can be produced within a few minutes in the multiprotein RCR assay system. Previous studies of DNA replication in vivo have identified a rolling circle DNA replication mode as being prominent late during the viral replication cycle (52)(53)(54)(55). Electron microscopic analysis has shown that the linear duplex concatameric tails of such molecules generally reach multiple genomes in length (i.e. 100 -500 kb). We conclude that the in vitro rolling circle DNA replication system described here is physiologically relevant to DNA replication in vivo, both with respect to the phage and bacterial proteins required for this mode of replication (Table I) and with regard to the lengths of the concatameric DNA molecules produced.
The precise nature of the replication fork assembly established in the RCR system is unknown, however. With current technology, it is a particularly daunting problem to determine the exact composition and stoichiometry of mobile and transient replication fork assemblies. Nevertheless, as a step toward this ultimate goal, we sought to define the basic properties of the replication fork apparatus. It was of particular interest to determine if the unique assembly and disassembly reactions used to establish a replication fork apparatus had any measurable impact on the functional properties of the replication fork assembly. Substantial differences between the fundamental properties of the and minimal replication fork assemblies would provide evidence for a difference in their underlying structure. For this reason, the properties of replication fork establishment and movement in the RCR system were compared with those manifested by a minimal RCR system in which the replication fork apparatus is known to be composed solely of DnaB helicase and DNA polymerase III holoenzyme.
Rolling circle DNA replication established by the RCR multiprotein system in the absence of SSB was substantially more vigorous than that yielded by the two-protein minimal RCR system. Moreover, the enhanced DNA synthesis capacity of the RCR system was achieved with amounts of DnaB severalfold lower than required for the optimized minimal RCR system. These results would be readily explained if either the RCR system utilized a higher percentage of the input TNC template DNA than the minimal RCR system or if the system established a replication fork apparatus that synthesized DNA at a higher rate. Neither explanation, however, was supported by our results. The efficiencies and time courses of template utilization were roughly comparable between the two RCR systems, particularly so when DnaB protein present in the minimal RCR system was permitted to bind to the TNC template DNA before the initiation of DNA synthesis. Additionally, the rate of replication fork movement, approximately 720 -750 bp/s at 30°C, was identical in the and minimal RCR systems. These results point to an alternative explanation for the greater levels of DNA replication provided by the RCR system. Simply, more DNA was synthesized per primary initiation event in the RCR assay system than in the standard minimal RCR system (e.g. compare lines 5 and 13 in Table II). This conclusion is consistent with the finding that DNA synthesis continued in a linear fashion for a significantly prolonged period of time with the RCR system but slowed quickly in the minimal RCR system when only DnaB helicase and DNA pol IIIh were present (compare open circles in Fig. 3, B and C).
There are at least two possible reasons why the RCR system is more potent than the minimal RCR system. The system may create replication fork assemblies that have a higher processivity than the basic assembly composed of a molecule of DnaB helicase coupled to DNA pol IIIh. For example, the replication fork apparatus assembled in the RCR system could potentially contain additional proteins or have a different arrangement of proteins and DNA at the fork that enhance processivity. While we cannot rule out this possibility with the data presently available, we favor the notion that the O and P initiators establish a replication fork apparatus identical to the minimal apparatus composed of DnaB and DNA pol IIIh. This view is consistent with the finding that the two types of replication fork assemblies move at identical rates. If the replication fork apparatus has processivity equivalent to that of the minimal fork assembly, then the elevated DNA synthesis produced in the RCR assay may reflect more frequent reinitiation of strand displacement DNA synthesis following spontaneous termination of DNA chain elongation. It is most likely relevant in this regard that we have recently demonstrated that the O and P replication proteins each contain an intrinsic, but cryptic, single-stranded DNA binding activity (56). These activities apparently play important roles in the efficient transfer of DnaB helicase onto a single-stranded DNA template. Moreover, the replication system described in this report is even capable of transferring DnaB onto SSB-coated ssDNA templates (19,20). This capacity permits the efficient reloading of DnaB onto the long displaced single strand of a replicated TNC DNA template and presumably ensures an eventual resumption of strand displacement DNA synthesis following a replication termination event.
It would be anticipated that when SSB is absent, even the minimal RCR system should be capable of reassembling, subsequent to a termination event, a replication fork apparatus that contains DnaB. However, reloading of DnaB onto the TNC template following termination should be blocked when the displaced ssDNA tail is coated with SSB (19,45). Nevertheless, DNA synthesis by the minimal RCR system increased sharply when SSB was present during the DNA synthesis reaction (Fig.  3B). How could the presence of SSB stimulate DNA synthesis under conditions where SSB prevents reinitiation of DNA chain elongation after dissociation of the replication fork apparatus from the template DNA? It is conceivable that binding of SSB to the displaced DNA chain during rolling circle DNA replication somehow heightens the processivity of the replication fork apparatus. Such a mechanism appears unlikely, however. For example, our findings indicate that the presence of SSB does not affect the rate of movement of replication forks established by either the minimal or RCR systems. A similar lack of influence of SSB on movement rates was reported previously for replication fork assemblies established by the E. coli primosomal proteins (39); nor have there been any reports that SSB directly interacts with the proteins that constitute the leading strand replication fork assembly (i.e. DnaB helicase and DNA pol IIIh). Based on these considerations, we tentatively conclude that SSB has no direct impact on the processivity of leading strand DNA chain elongation during rolling circle DNA replication.
We suggest instead that SSB may act to stimulate DNA synthesis through an indirect mechanism by binding and sequestering ssDNA chains that are produced by rolling circle DNA replication. Concentrations of ssDNA reach very high levels in the RCR assays, attaining concentrations of greater than 100 M (as nucleotide) within 15 min following the initiation of DNA synthesis (Table II). It is possible that the replication fork apparatus present in the minimal RCR assay is particularly susceptible to the presence of high concentrations of free ssDNA chains. Consistent with this hypothesis, we have found in preliminary experiments that DNA synthesis by the minimal RCR replication fork apparatus was strongly inhibited by the addition of dT 200 to the assay at a concentration of approximately 50 M. 3 The DnaB helicase molecule at the fork, believed to be a hexameric protein (57)(58)(59), might exist in certain states during its action as a processive DNA helicase that leaves one or more of its subunits available for interaction with ssDNA chains. We speculate that such aberrant interactions could lead to the dissociation of DnaB from the replication fork assembly and bring about immediate termination of DNA synthesis. In this scenario, SSB enhances the processivity of DNA synthesis in the RCR assay by sequestering the ssDNA product, thus minimizing the opportunity for anomalous interactions of ssDNA with the replication fork apparatus.
The presence of SSB also stimulated DNA synthesis in the RCR assay (Fig. 3C), but only about 50%, as compared with an effect of more than 3-fold in the standard RCR assay (Fig. 3B and Table II). The diminished impact of SSB on the RCR assay is mostly attributable to the fact that DNA synthesis is much more vigorous in the RCR assay, as compared with the minimal RCR assay, when SSB is absent (compare open circles in Fig. 3, B and C). Nevertheless, as discussed earlier, we believe that the replication fork apparatus assembled in the minimal RCR assay is most likely identical to that formed in the RCR assay. We infer further that, when SSB is absent, it is the capacity of the RCR system to efficiently reload DnaB onto DNA template/product molecules that permits this system, but not the minimal RCR system, to overcome the probable inhibitory effects of the ssDNA tails produced by rolling circle DNA replication.
In the absence of primase, only leading strand DNA synthesis can be established by the and minimal RCR systems. This was confirmed by the nearly complete sensitivity of the replication products to digestion with S1 nuclease. When primase is present, the resulting primase-DnaB interactions should lead to the synthesis of RNA primers on the displaced singlestranded DNA tail of the TNC DNA template, which, in turn, should permit synthesis of lagging strand DNA chains. This was verified by the resistance of the vast majority of the DNA product to S1 nuclease digestion. It was anticipated, therefore, that the synthesis of lagging strand DNA chains when primase was present in the RCR assays would lead to an approximate doubling in the level of DNA synthesis. This expectation was met when primase supplemented RCR assays performed in the presence of SSB (Table II, lines 15 and 16). Surprisingly, however, inclusion of primase produced an unexpectedly large 4-fold stimulation of DNA synthesis in the minimal RCR assay when SSB was absent (Table II, lines 5 and 6). Thus, it is apparent that primase, in addition to priming lagging strand DNA synthesis, must also stimulate leading strand DNA synthesis in the minimal RCR assay when SSB is absent. Stimulation of leading strand DNA synthesis by primase, however, is not consistent with previous evidence indicating that primase acts distributively during replication fork propagation mediated by DnaB helicase and DNA pol IIIh (51,60). One possible resolution to this apparent paradox would be to suppose that primase acts indirectly to stimulate DNA synthesis by preventing the accumulation of ssDNA product. If it is presumed that excess ssDNA product is inhibitory to replication fork propagation in the minimal RCR assay, as suggested earlier, then primase might stimulate RCR DNA replication by virtue of its capacity to initiate conversion of ssDNA to an innocuous duplex form. This hypothesis would also explain why stimulation by primase is more prominent when SSB is absent, since SSB coating of the ssDNA product is apparently sufficient by itself to interfere with interactions of ssDNA with DnaB helicase (45). Thus, primase and SSB may each indirectly stimulate RCR DNA synthesis in a similar fashion, primase by eliminating, and SSB by sequestering, free ssDNA.
The functional properties reported here for the replication fork apparatus are reminiscent of those described for replication forks assembled with the "X-type" primosomal proteins (39). In the latter system, loading of DnaB helicase onto the ssDNA tail of TNC templates was accomplished with the aid of several E. coli preprimosomal proteins, including DnaC, DnaT, PriA, PriB, and PriC. Although the X-type replication fork apparatus assembled in this fashion is believed to contain PriA, a 3Ј 3 5Ј DNA helicase (61), in addition to DnaB and DNA pol IIIh, recent findings suggest that the helicase activity of PriA is not required for replication fork propagation in this system (62). The presence of the X-type primosomal proteins yielded more extensive rolling circle DNA synthesis, and at significantly lower concentrations of DnaB, than produced by a minimal enzyme system of DnaB and DNA pol IIIh (39). In this respect, therefore, the X-type primosomal proteins apparently act like the O and P initiators to assemble the replication fork apparatus in a more efficient manner. Moreover, once assembled, the replication fork apparatus assembled by the X-type primosomal proteins moved at approximately the same rate (730 nucleotides/s at 30°C) (39) as both the replication fork apparatus and the minimal replication fork apparatus composed of DnaB and DNA pol IIIh. We infer from the similar functional properties of the replication fork assemblies formed in these three diverse systems that DnaB helicase and DNA pol IIIh in each case form the basic enzymatic machinery responsible for DNA unwinding and DNA synthesis. Although we cannot strictly rule out the presence of O, P, or any of the molecular chaperones in the replication fork apparatus formed in our reconstituted system, previous and ongoing biochemical studies suggest that these proteins are required only for the transfer of DnaB onto the DNA template 4 (4,11,63) and that they do not play a direct role in replication fork propagation. One note of caution, however, comes from published observations that late DNA replication in vivo is markedly reduced (64) or aberrant (65) when cells infected with certain thermosensitive O mutants are shifted to a nonpermissive temperature during the rolling circle phase of DNA replication.
Two other aspects of this work require further comment. First, we have found that the DNA polymerase III holoenzyme is capable of strand displacement DNA synthesis. This activity is strictly dependent on the presence of SSB and proceeds at least 10 times more slowly than chain elongation by DNA pol IIIh during replication of SSB-coated ssDNA (66) or during rolling circle DNA replication performed in conjunction with DnaB helicase (39). An earlier investigation showed that DNA pol IIIh usually terminates DNA chain elongation when it encounters the 5Ј-end of an annealed DNA or RNA chain in its path (43). When this finding is considered together with our results, it suggests that strand displacement DNA synthesis by DNA pol IIIh may be facilitated by the presence of a singlestranded 5Ј-tail in the strand that will be displaced. If so, it may explain why the strand displacement activity had not been detected in previous studies of DNA pol IIIh. Strand displacement activity by DNA pol IIIh is presumably not essential for replication fork propagation, since DnaB helicase apparently carries out this function during coupled synthesis of leading and lagging strands. Nevertheless, the strand displacement activity of DNA pol IIIh conceivably could augment DNA un-4 B. Learn and R. McMacken, unpublished data. winding by DnaB helicase. In the absence of DNA pol IIIh, DnaB-mediated unwinding proceeds slowly (ϳ50 bp/s) 5 (14), so it is possible that a strand displacement activity of DNA pol IIIh contributes to the rapid rate of DNA unwinding and chain elongation observed during propagation of and E. coli replication forks in vivo and in vitro. It is also possible that the strand displacement activity of DNA pol IIIh plays a role in other stages of chromosomal DNA replication. For example, if DNA helicase action is spatially restricted or missing altogether, strand displacement synthesis by DNA pol IIIh could serve to enlarge DNA "bubbles" during initiation of replication, or it could enable complete replication of parental DNA during the terminal stages of chromosomal DNA replication.
A second matter of concern is the presence of DNA helicase II (UvrD protein) in preparations of DNA pol IIIh purified from wild type cells (44). We, therefore, characterized the functional properties of DNA pol IIIh purified from an E. coli strain deleted for the uvrD gene. We found that the presence of DNA helicase II had no detectable influence on the properties of replication fork movement during rolling circle DNA replication and that DNA helicase II-free preparations of DNA pol IIIh retained their capacity to carry out strand displacement DNA synthesis. Nevertheless, we cannot absolutely rule out the possibility that the strand displacement activity observed in our preparations of DNA pol IIIh is due to the action of some other contaminating DNA helicase. Studies of the properties of DNA pol IIIh reconstituted from highly purified subunit polypeptides may be required to certify that the strand displacement activity reported here is indeed intrinsic to this multisubunit DNA polymerase. The presence of contaminating DNA helicase II in the DNA pol IIIh preparation was found to interfere with the assembly of replication forks in the standard RCR assay. The UvrD-mediated inhibition of rolling circle DNA replication was partially circumvented when DnaB was permitted to bind to the TNC DNA template before the addition of DNA pol IIIh to the reaction mixture. If this finding is taken together with the fact that DNA helicase II acts stoichiometrically to bring about DNA unwinding (67), then it appears likely that DNA helicase II inhibits rolling circle DNA replication by binding in multiple copies to the ssDNA tail of the TNC template. Coating of the ssDNA tail with DNA helicase II could, in a fashion similar to SSB (19,45), block binding of DnaB helicase to the ssDNA tail. It is also possible that bound molecules of DNA helicase II interfere with replication fork assembly at a step after DnaB binding, e.g. by blocking translocation of DnaB 5Ј 3 3Ј along the ssDNA tail and reducing access of DnaB to the preformed fork (single strand/double strand junction) on the TNC template.