Characterization of DNA synthesis catalyzed by bacteriophage T4 replication complexes reconstituted on synthetic circular substrates

Abstract

Replication complexes were reconstituted using the eight purified bacteriophage T4 replication proteins and synthetic circular 70-, 120- or 240-nt DNA substrates annealed to a leading-strand primer. To differentiate leading strands from lagging strands, the circular parts of the substrates lacked dCMP; thus, no dCTP was required for leading-strand synthesis and no dGTP for lagging-strand synthesis. The size of the substrates was crucial, the longer substrates supporting much more DNA synthesis. Leading and lagging strands were synthesized in a coupled manner. Specifically targeting leading-strand synthesis by decreasing the concentration of dGTP decreased the rate of extension of leading strands. However, blocking lagging-strand synthesis by lowering the dCTP concentration, by omitting dCTP altogether, by adding ddCTP, or with a single abasic site had no immediate effect on the rate of extension of leading strands.

INTRODUCTION

Mechanisms of DNA replication and functional components of the DNA replication machinery are conserved throughout evolution (1). The replication systems of Escherichia coli, phage T7 and phage T4 are the best understood. Eight bacteriophage T4 proteins are required to reconstitute replication forks able to catalyze both leading-strand and lagging-strand DNA synthesis (2). These proteins are the DNA polymerase (gp43), the processivity clamp (gp45), the clamp loader (gp44/gp62), the single-stranded DNA (ssDNA)- binding protein (gp32), the 5′-3′ helicase (gp41), the primase (gp61) and the helicase loader (gp59) (2–5). Two other T4 proteins, DNA ligase (gp30) and RNase H, are required later to process and seal Okazaki fragments (3). Pioneering studies of reconstituted T4 DNA replication suggested a mechanism based on coupling between the synthesis of leading and lagging strands at a replication fork (4,5): upon loading onto template DNA, the T4 replication proteins would form a replication complex held together by protein–protein interactions. Two DNA polymerases, one for leading-strand synthesis and another for lagging-strand synthesis, would be tethered by protein–protein interactions.

The reconstituted E.coli replication system was the first to reveal coupled synthesis of leading and lagging strands at both the structural and functional levels (6–9). Gel-filtration studies showed that E.coli DNA polymerase III forms a tight and stable complex of 14 polypeptides, including two polymerase cores (7). Dilution experiments with pre-formed E.coli replication complexes demonstrated that lagging-strand synthesis resists dilution when the dilution buffer is supplied with primase, ssDNA-binding protein and β-clamp (6,8). Reconstitution of T7 DNA replication on a 70-nt circular substrate showed that synthesis of both strands is coordinated (i.e. the rate of lagging-strand synthesis equals the rate of leading-strand synthesis), with recycling of the lagging-strand polymerase (10). Moreover, in the T7 system, the DNA loops postulated to arise in coupled DNA replication were observed in about half of the molecules (10).

In early studies of phage T4 DNA replication in vitro, lacking only the then-undiscovered gp59 helicase loader, varying the polymerase concentration over a wide range did not change the average size of Okazaki fragments (4). More recently, reconstitution of T4 replication forks on a 70-nt circular substrate using an exonuclease-deficient DNA polymerase showed that DNA synthesis of leading and lagging strands is coordinated under those conditions (11). Reconstitution of T4 replication forks on M13 DNA followed by dilution showed that leading-strand and lagging-strand synthesis remain coordinated provided the gp44/gp62 clamp loader, the gp45 clamp and the gp32 ssDNA binding protein are provided in the dilution buffer (12).

Blocks to DNA synthesis are barriers to DNA replication unless they are repaired or circumvented. Several models have been suggested to explain how replication forks could bypass blocks in phage T4 (5,13,14), E.coli (15), mammals (16,17) and SV40 (18–20). Recombinational rescue of stalled forks has been described in E.coli (15). Other models suggest how a replication fork could simply continue past a block, either leaving a gap or abandoning one progeny strand (17,18,20). One model suggests that a replication complex could uncouple, leaving one part stalled at the lesion and permitting the other to continue (14,20). Another model, based on the semi-discontinuous nature of a replication fork, holds that blocks to lagging-strand synthesis could be bypassed, leaving a gap opposite the lesion (17,18).

The bacteriophage T4 in vitro replication system is an excellent model for studying the effects of DNA blocks on DNA replication. T4 replication complexes in vitro were used previously to study the collision of a replication complex with a transcription complex moving in either the same or the opposite direction (21–23). In those studies, DNA replication and transcription were reconstituted on circular double-stranded DNA molecules using a nick to prime DNA replication and an appropriately oriented E.coli σ⁷⁰ promoter. Regardless of whether the two complexes moved in the same or the opposite direction, the extension of leading strands slowed, implying that leading-strand extension pauses briefly as a result of the interaction between the complexes (21–23).

In this study, we used circular 70-, 120- and 240-nt DNA substrates to reconstitute T4 replication forks in the presence of all eight replication proteins. The circular parts of these substrates lacked dCMP residues in order to differentiate leading-strand and lagging-strand synthesis. The size of the substrates strongly affects replication, allowing more DNA synthesis with the longer substrates: the extension rate of leading strands supported by the 70-nt circle is ∼50 nt/s, whereas it becomes ∼350 nt/s with the 120- and 240-nt circles. As expected for the coupled synthesis of leading and lagging strands, synthesis of both strands on 240-nt DNA is highly coordinated in undiluted reactions and remains coordinated after a 64-fold dilution of pre-formed replication complexes in a buffer lacking the phage T4 DNA polymerase and the primosomal gp41, gp61 and gp59 proteins. Because leading-strand synthesis on the 240-nt circle does not require dCTP, we could selectively block lagging-strand synthesis with either low concentrations or the complete absence of dCTP, or by adding ddCTP. We could also block lagging-strand synthesis with a single abasic site. None of these blocks decreased the rate of extension of leading strands, whereas decreasing the dGTP concentration did. Thus, contrary to the situation when T4 replication complexes encounters E.coli transcription complexes (21–23), blocks to lagging-strand DNA synthesis do not slow the rate of extension of leading strands. We suggest that leading-strand synthesis in T4 replication forks is able to continue despite blocks to lagging-strand DNA synthesis.

MATERIALS AND METHODS

Chemicals, oligonucleotides and proteins

ATP, GTP, CTP, UTP, dATP, dGTP, dCTP, dTTP and ddCTP were from Amersham Pharmacia Biotech.

The following oligonucleotides were supplied by Oligo etc. Inc. in a gel-purified form: (#1) 120mer (5′-TAACCTACTACTATTATAAATTATCCACCTTTTCACTCCCAAATAAAATATACTAAACATTATTACCATTTCAAATTATCTAATTCAAACTAAATCTACTCTTCCAAATTAATCAACTCC-3′); (#2) 120mer (5′-CATAATAAACTTCCAACACCTACTTTATTCATATTTAAAACATTTACTACACACCAATTCACAATTAACTCTAACCATCCCAAAATACCTCTTATCAAAACAATTAAATACTCTCTAATC-3′); (#3) 120mer (5′-GAGGTTGGTTTATATGGTAAGAATTTGTATAAGATATGATAGTAGATTTAGTTTGAATTAGATAATTTGAAATGGTAATAATGTTTAGTATATTTTATTTGGGAGTGAAAAGGTGGATAA-3′); (#4) 120mer (5′-GAGGTTGGTTTATAT GGTAAGAATTTGTATAAGATATGATAGTAGATTTAGTTTGAATTAGATAATTTGAAATGGTAATAATGTTTAGTATATTTTATxTGGGAGTGAAAAGGTGGATAA-3′ where x indicates a tetrahydrofuran moiety mimicking an abasic site); (#5) 80mer (5′-GAGGTTGGTTTATATGGTAAGAATTTGTATAAGATATGATTTTATTTGGGAGTGAAAAGGTGGATAATTTATAATAGTAG-3′); (#6) 70mer (5′-TAACCT ACTACTATTATAAATTATCCACCTTTTCACTCCCAAATAAAATATTTCCAAATTAATCAACTCC-3′); and (#7) 21mer (5′-TTATCCACCTTTTCACTCCCA-3′).

The following oligonucleotides were from Research Genetics Inc.: (#8) 40mer (5′-AGGTGTTGGAAGTTTATTATGGGAGTTGATTAATTTGGAA-3′); (#9) 40mer (5′-TTTATAATAGTAGTAGGTTAGATTAGAGAGTATTTAATTG-3′); and (#10) 39mer (5′-TTATAATAGTAGTAGGTTAGGAGTTGATTAATTTGGAAG-3′).

[α-³²P]dGTP (3000 Ci/mmol) was from NEN. [α-³²P]dCTP (3000 Ci/mmol) and [α-³²P]dTTP (3000 Ci/mmol) was from Amersham Pharmacia Biotech. T4 DNA ligase, DraI and polynucleotide kinase were from New England Biolabs. Gp43, gp43D219A, gp44/gp62, gp45, gp32, gp41, gp61 and gp59 were purified and their concentrations were estimated as described (12).

DNA substrates

The complementary strand of 7196-nt circular M13mp2 ssDNA was synthesized as described (12).

The 240-nt circular substrate was constructed as follows. Oligonucleotides #1 and #2 were phosphorylated separately in 200 µl of 1× polynucleotide-kinase buffer (New England Biolabs) containing 4 µM of either oligonucleotide, 1 mM ATP, and 150 U polynucleotide kinase for 60 min at 37°C followed by incubation at 75°C for 10 min. The 200 µl of phosphorylated 120mer #1 were supplemented with NaCl to 100 mM and with scaffold 40mer #8 to 4 µM. The mixture was incubated for 20 min at 45°C and then mixed with the 200 µl of phosphorylated 120mer #2 supplemented with 100 mM NaCl, followed by incubation for 30 min at 45°C. The mixture was then mixed with 600 µl of 1× polynucleotide-kinase buffer containing 1 mM ATP, 100 mM NaCl and 1.33 µM of scaffold 40mer #9 and the resulting mixture was incubated for 2 h at 45°C and for 5 min at room temperature. Then, 120 Weiss units of T4 DNA ligase were added and the mixture was held at 16°C for 19 h, followed by incubation at 75°C for 10 min. To digest 3′ ends, T4 DNA polymerase was added to 0.44 µM and the mixture was incubated at 37°C for 1 h and then at 75°C for 10 min.

To anneal the circular 240-nt ssDNA with 120mer #3 whose 80 nt at the 3′ end are complimentary to the circle, 120mer #3 was added to 0.8 µM and the mixture was incubated at 55°C for 15 min. The 240-nt DNA was purified through a 6% polyacrylamide gel in 1× Tris-borate buffer (0.09 M Tris-OH, 0.09 M boric acid, 1 mM EDTA), followed by gel filtration using micro-spin columns (Bio-Rad).

Circular 120-nt ssDNA annealed with 120mer #3, whose 80 nt at the 3′ end are complimentary to the 120-nt circle, and circular 70-nt ssDNA annealed with 80mer #5, whose 40 nt at the 3′ end are complimentary to the 70-nt circle, were obtained similarly with the following changes. Phosphorylated 120mer #1 or phosphorylated 70mer #6 were circularized in the presence of the scaffold 39mer #10 for 30 min at 45°C. After ligation and digestion, the 120-nt circular DNA was annealed with 120mer #3 and the 70-nt circle was annealed with 80mer #5, followed by gel-purification and gel-filtration.

DNA replication assays

DNA replication reactions catalyzed by T4 proteins were performed in a final volume of 40 µl in a standard replication mixture containing 20 mM Tris acetate, pH 7.8, 50 mM K glutamate, 17.5 mM KCl, 8 mM Mg acetate, 5 mM DTT, 8.7% glycerol (v/v), 500 µg/ml BSA, 0.2 mM dATP, 0.2 mM dTTP, 0.1 mM dGTP, 0.1 mM dCTP, 1.5 mM ATP, 1.5 mM GTP, 0.4 mM CTP, 0.4 mM UTP, 69 µCi/ml [α-³²P]dNTP (3000 Ci/mmol) and 4 nM DNA. When indicated, dCTP or CTP and UTP were omitted from the reaction mixture, or dGTP and dCTP concentrations were lowered, and/or ddCTP was included. Unless otherwise noted, the standard replication mixture was supplemented with 9.4 nM gp43, 16.5 nM gp44/gp62, 14.2 nM gp41 (as a hexamer), 103 nM gp45 (as a trimer), 900 nM gp32, 16 nM gp61 and 9 nM gp59. These concentrations correspond to molar ratios of DNA:gp43:gp44/gp62: gp45:gp32:gp41:gp61:gp59 of 1:2.3:4.1:26:225:3.5:4:2.2. Reaction mixtures without template DNA but with all T4 proteins except gp43 and gp32 were first incubated at room temperature for 3 min. Then, gp43 and gp32 were added, the mixtures were transferred to a 37°C water bath for 1 min, pre-warmed template DNA was added at time zero, and reactions were run at 37°C. Samples (5 µl) were withdrawn at the indicated times and were mixed with 25 µl of 75 mM EDTA, 30 mM NaOH. Samples of the diluted reaction products were separated in 0.6% alkaline agarose gels in 33 mM NaOH, 2 mM EDTA. In the cases of the alkaline agarose gel separations of labeled leading-strand products along with labeled lagging-strand products, both of which were obtained in the same conditions but in the presence of a different marker— [α-³²P]dGTP to label products of leading-strand synthesis and [α-³²P]dCTP to label products of lagging-strand synthesis—amounts of [α-³²P]dGTP and [α-³²P]dCTP radioactivity were equalized before loading. To determine the total amount of [α-³²P]dNTP, the reaction products were separated on PEI-plates (Merck) in 1.3 M LiCl, 1.5 M acetic acid and then quantified.

DNA synthesis was quantified by separating the reaction products with thin-layer chromatography on PEI-plates (Merck) in 1.3 M LiCl, 1.5 M acetic acid as described (12). The data presented are averages from at least four experiments. The ranges of the values were <10%.

For the experiments displayed in Figure 6A, conditions were the same as above but gp41, gp61 and gp59 were omitted, the 4-nM DNA substrates were ³²P-kinase-labeled 21mer #7 annealed with either 120mer #3 or 120mer #4, and gp32 and gp45 were at 75 and 34.3 nM (as a trimer), respectively. In Figure 6B, the products of 30-s standard reactions carried out in the presence of [α-³²P]dCTP and 240mer ssDNA annealed with 120mer oligomer with (#4) or without (#3) an abasic site were stopped by adding EDTA to 125 mM, followed by heating at 75°C for 15 min. The products were passed through micro-spin columns (Bio-Rad) and 15 µl were cleaved with 5 U DraI in 1× NEB buffer (New England Biolabs) for 15 min at 37°C followed by heat inactivation.

Figure 6.

Effect of an abasic site on DNA synthesis by reconstituted phage T4 replication complexes. (A) An abasic site blocks DNA synthesis by T4 DNA polymerase holoenzyme. ³²P-end-labeled 21mer #7 annealed with 120mer #3 (lanes 1, 3, 5, 7, 9, 11, 13 and 15) or with the same oligomer containing an abasic site at position 99 (120mer #4) (lanes 2, 4, 6, 8, 10, 12, 14 and 16) was incubated with either T4 gp43 (lanes 5, 6, 13 and 14) or the exonuclease-deficient gp43D219A (lanes 7, 8, 15 and 16) in the presence of gp44/62, gp45 and gp32 for the indicated times, and the products were separated in a 10% polyacrylamide gel + 6 M urea. Note that in the substrate with an abasic site, the first nucleotide to be incorporated into the ³²P-end-labeled 21mer is opposite the abasic site. (B) Block to lagging-strand synthesis imposed by the abasic site. Products of DNA synthesized by reconstituted T4 replication complexes in the presence of [α-³²P]dCTP and 240mer circular ssDNA annealed with the 120mer ssDNA without (lanes 1 and 2) or with (lanes 3 and 4) abasic site were digested (lanes 2 and 4) or not (lanes 1 and 3) by DraI and separated in a 6% polyacrylamide gel + 6 M urea along with marker DNA. Note that the 240mer circular ssDNA has a unique DraI site. Upon DraI cleavage of the replication products, the major 240mer fragment is formed along with other fragments that are located between the DraI site and either the 3′ or 5′ ends of Okazaki fragments. Arrows show positions of the major 240-nt DraI fragment and the 126-nt fragment located between the DraI site and the 3′ end of the nascent fragment terminated by the abasic site. (C) DNA products were synthesized by reconstituted T4 replication complexes in the presence of [α-³²P]dTTP and of 240mer circular ssDNA annealed with 120mer ssDNA without (lanes 1–3) or with (lanes 4–6) an abasic site, and separated in a 0.6% alkali gel.

Dilutions of pre-existing replication complexes

The standard dilution mixture contained 20 mM Tris-acetate, pH 7.8, 50 mM K glutamate, 17.5 mM KCl, 8 mM Mg acetate, 5 mM DTT, 8.7% glycerol (v/v), 500 µg/ml bovine serum albumin, 0.2 mM dATP, 0.2 mM dTTP, 0.1 mM dGTP, 0.1 mM dCTP, 1.5 mM ATP, 1.5 mM GTP, 0.4 mM CTP, 0.4 mM UTP, 69 µCi/ml [α-³²P]dNTP (3000 Ci/mmol), 16.5 nM gp44/gp62, 103 nM gp45 (as a trimer) and 43.8 nM gp32. The standard replication reaction was carried out in the presence of 4 nM of the 240-nt circular DNA. After 1 min, 2 µl of the mixture were added to 126 µl of pre-warmed standard dilution mixture. The reaction was incubated for 4 min at 37°C and then terminated by adding EDTA to 50 mM followed by heating at 75°C for 10 min. Samples (2 µl) of the standard replication mixture mixed with 126 µl of 50 mM EDTA after 1 and 5 min were used as controls to estimate levels of DNA synthesis before and without 64-fold dilution, respectively. Samples of all these reactions were centrifuged through Bio-Rad micro-spin columns to remove unincorporated [α-³²P]dNTP and were analyzed by electrophoresis in a 0.6% alkaline agarose gel as described above.

Volume integration of DNA products longer than the 9.4-kb marker DNA in a reaction containing [α-³²P]dGTP and shorter than 9.4-kb marker DNA in a reaction containing [α-³²P]dCTP was used to quantify leading-strand and lagging-strand synthesis, respectively, after 64-fold dilution. The volume data of a particular diluted reaction were subtracted from those of a 1-min reaction, the time at which dilution was performed.

RESULTS

The use of a 70-nt synthetic circular substrate to characterize replication complexes reconstituted in the presence of the exonuclease-deficient D219A gp43 and the other seven wild-type T4 replication proteins was reported previously (11). In order to further characterize the behavior of bacteriophage T4 replication complexes on minicircular DNA substrates, we examined the impact of circle size on the rate of DNA synthesis. We first compared rates of total DNA synthesis by wild-type T4 replication complexes using circular 70-, 120- and 240-nt substrates with synthesis using a well-characterized large substrate, circular 7196-nt M13mp2 DNA (Fig. 1A). The rate of total DNA synthesis is 2–5-fold lower with the 70-nt substrate than with the other substrates, the rates becoming progressively higher with increasing substrate size (Fig. 1B). Incorporation begins approximately linearly with the small substrates but shows a distinct lag with the large M13 substrate. This reflects the tendency of (at least) the helicase loading protein gp59 to bind to double-stranded as well as to ssDNA, effectively lowering its local concentration at the replication fork and thus slowing primosome loading (2).

Figure 1.

DNA synthesis by T4 replication complexes reconstituted on M13, 240-, 120- and 70-nt DNA substrates. (A) Schematic representation of DNA substrates used to reconstitute T4 replication complexes. With M13 DNA, the 5′ tail is ∼150 nt. With the circular 70-nt substrate, 40 nt at the 3′ end of the 80-nt linear strand are complimentary to the 70-nt circle. With the circular 120- and 240-nt substrates, 80 nt at the 3′ end of the 120-nt linear strand are complimentary to the 120- or 240-nt circles. The circular parts of the 70-, 120- and 240-nt substrates are devoid of dC residues. (B) Total DNA synthesis in standard DNA replication reactions containing as template DNA M13 mp2 (squares), the 240-nt circle (diamonds), the 120-nt circle (circles), or the 70-nt circle (triangles). To obtain the total value for M13 DNA, µM of dGMP incorporated were multiplied by 4.73, a coefficient based on the G·C content of the double-stranded DNA. To obtain the values for the 70-, 120- and 240-nt substrates, µM of dGMP incorporated and µM of dCMP incorporated were added and multiplied by 3.68, 4.0 and 3.64, coefficients based on the G·C contents of their respective double-stranded DNAs. (C) Products of standard DNA replication reactions obtained at the indicated times on different templates were separated on 0.6% agarose gels along with molecular-mass markers consisting of a HindIII digest of ³²P-labeled λ DNA. The reaction with M13 DNA (lanes 1–3) included [α-³²P]dGTP which labeled both leading-strand and lagging-strand products. With the 240- (lanes 4–9), 120- (lanes 10–15) and 70-nt (lanes 16–21) circular DNAs, reactions included either [α-³²P]dGTP to label products of leading-strand synthesis or [α-³²P]dCTP to label products of lagging-strand synthesis. (D) Influence of some reaction components on the length of leading strands formed by T4 replication complexes reconstituted on a 70-nt circle. Products were obtained at the indicated times in reactions that included [α-³²P]dGTP, which labels leading-strand products, and were separated in a denaturing 0.6% agarose gel. Lanes 1–3 and lanes 7–9 contain products of a reaction in which gp41 helicase, gp61 primase and gp59 helicase loader were omitted. Lanes 4–6 contain products of a standard reaction. Lanes 7–9 and lanes 10–12 contain products of reactions in which wild-type gp43 was replaced by an exonuclease-deficient D219A enzyme at 9.4 nM.

During such reactions, total DNA synthesis reflects two processes: strand extension by established replication complexes and the formation of new complexes. We therefore enquired how the rate of extension of leading strands is affected by the size of the circle. In order to differentiate leading-strand and lagging-strand synthesis, the circular parts of the 70-, 120- and 240-nt substrates (which template leading-strand synthesis) were designed to lack dCMP. With the 70-, 120- and 240-nt substrates, lagging strands labeled with [α-³²P]dCMP have a median size of 1.5–3 kb. With the 120- and 240-nt substrates, the longest leading strands at 1 min were up to 23 kb, implying extension rates of up to 380 nt/s. The rate of extension of leading strands formed with the 70-nt substrate is ∼50 nt/s, 5–7-fold less than with the other substrates (Fig. 1C). We next studied the influence of several factors on the rate of synthesis of leading strands formed with the 70-nt substrate (Fig. 1D). Omitting the primosomal proteins slightly decreased the lengths of leading strands, showing that the primosome only slightly affects leading-strand synthesis; this result suggests that the 70mer lacks space sufficient to accommodate both the T4 primosome and the DNA polymerase holoenzyme. Replacing the wild-type gp43 with an exonuclease-deficient variant (24) has no strong effect on leading-strand synthesis in the complete reaction. When the primosomal proteins were then omitted from this reaction, the length of leading-strand products was even greater than in the complete reaction. This result is consistent with a previous observation that the T4 exonuclease-deficient DNA polymerase conducts proficient strand-displacement synthesis (25).

Taken together, these results show that T4 replication complexes formed on the 70-nt circle are characterized by slow extension of leading strands and a resulting low rate of total DNA synthesis. Thus, T4 replication complexes prefer templates above a minimum size to synthesize DNA most efficiently in vitro, the minimum being ∼240 nt for a circular substrate. Accordingly, we used 240-nt circles in subsequent experiments.

As expected of coupled synthesis, synthesis of >23-kb leading strands and 1.5–3-kb lagging strands on 240-nt circles occurs in a 1:1 ratio (Fig. 2A). We then tested the effect on lagging-strand synthesis of diluting pre-formed complexes 64-fold in buffer containing neither DNA polymerase nor the T4 primosomal proteins (Fig. 2B). The amount of DNA accumulated after 4 min in the diluted reactions increased by 3.8-fold, an amount 2.2 times less than in undiluted reactions, presumably because no new complexes can form after dilution; this result is consistent with the observation that no DNA synthesis was detected in reactions where DNA polymerase, gp41 helicase, gp61 primase, gp59 helicase-loading protein and template DNA were pre-diluted in a dilution mixture containing gp32, gp44/62 and gp45 and then incubated for 5 min (Fig. 2B, lanes 1 and 8). The average size of Okazaki fragments was unaffected when replication complexes formed in the first minute were diluted 64-fold. The ratio of lagging strands to all products in the diluted reaction was 49.4%, indicating that coordination of lagging-strand synthesis and leading-strand synthesis resists this high dilution. These results show that T4 replication complexes reconstituted on the 240-nt substrate synthesize leading and lagging strands in a coupled manner.

Figure 2.

Coordination of leading-strand and lagging-strand synthesis at T4 replication forks reconstituted on a 240-nt substrate. (A) Standard replication reactions were carried out in the presence of either [α-³²P]dGTP (squares) or [α-³²P]dCTP (diamonds) to estimate leading-strand and lagging-strand synthesis, respectively. (B) Lagging-strand synthesis resists 64-fold dilution in dilution buffer supplemented with the gp44/gp62 clamp loader, the gp45 clamp and the gp32 ssDNA-binding protein. In lanes 1 and 8, template DNA, gp43, gp32, gp41, gp61 and gp59 were pre-diluted to obtain final concentrations of the proteins in the reaction identical to those in the reaction diluted 64-fold (lanes 4 and 5), and were then incubated in the standard dilution mixture for 5 min. Lanes 2 and 3 show products of DNA synthesis after 1 min incubation in standard replication buffer but diluted 64-fold for comparison with other diluted reactions. Lanes 4 and 5 show products of DNA synthesis formed during the first 1 min of incubation in the standard replication buffer and during the next 4 min after 64-fold dilution into the standard dilution mixture. Lanes 6 and 7 show products of 5 min standard replication reactions after 64-fold dilution for comparison with other diluted reactions. DNA products were labeled with [α-³²P]dCTP in lanes 1, 2, 4 and 6 and with [α-³²P]dGTP in lanes 3, 5, 7 and 8.

We next sought to determine how leading-strand synthesis responds to blocking ongoing lagging-strand DNA synthesis. To create such blocks, we omitted the dCTP that is required exclusively for lagging-strand synthesis. The T4 primosome synthesizes predominantly 5′-rArCrNrNr-3′ primers (26,27) and an analysis of the sequence of the 240-nt DNA substrate showed that omitting dCTP should block the lagging-strand polymerase on average at the third dNMP to be added to 5′-rArCrNrNrNdNdN-3′ during the synthesis of each Okazaki fragment. If blocks to lagging-strand synthesis also block leading-strand synthesis, then the extension of leading strands should either cease or, perhaps, pause and then proceed. Alternatively, if blocks to lagging-strand synthesis have no effect on leading-strand synthesis, then the rate of extension of leading strands should be unaffected. Liu et al. (21–23) used a similar approach to study the effects of an encounter between T4 replication complexes and E.coli transcription complexes and reported a small decrease in the rate of extension of leading strands.

When we examined the effect of blocks to lagging-strand synthesis, the rate of extension of leading strands at early times was the same in reactions with and without dCTP: compare the mobility of the major bands in lanes 1–3 with those in lanes 7–9 of Figure 3A. The integrated intensity of the bands formed in a 1-min reaction without dCTP (Fig. 3A, lane 7) was 90% of that of the complete reaction (Fig. 3A, lane 1), further indicating that most leading-strand synthesis is at least initially unaffected by blocks to lagging-strand synthesis. [On the other hand, decreasing the dGTP concentration from 100 to 10 µM considerably decreased the rate of extension of leading strands (Fig. 4A), as expected. This result indicates that our replication conditions allow us to detect blocks to the extension of leading strands.] To synthesize leading strands of 9.4–23.1 kb (Fig. 3A) during coupled synthesis of both strands, a replication complex must synthesize from five to eleven 2-kb Okazaki fragments under standard conditions. Thus, in the reaction without dCTP, a replication complex will encounter a block to lagging-strand synthesis from five to eleven times (or only once if the block uncouples replication), whereas the rate of extension of leading strands is in fact unaffected. As controls we used reactions where lagging-strand synthesis was blocked during RNA primer synthesis by omitting either primase or CTP + UTP. There is no RNA primer synthesis in reactions without primase, and omitting CTP + UTP almost completely abolishes the synthesis of RNA primers and uncouples lagging-strand synthesis by dramatically increasing the average size of Okazaki fragments (Fig. 3A, lanes 13–15) and decreasing the ratio of lagging-strand synthesis from 50 to 17% of total synthesis. Omitting CTP + UTP (Fig. 3A, lanes 10–12) or primase (data not shown) has no effect on the rate of extension of leading strands.

Figure 3.

Effect of omitting dCTP on DNA synthesis catalyzed by T4 replication complexes reconstituted on a 240-nt circle. (A) DNA replication products synthesized in 1, 2 and 4 min were separated in a 0.6% alkali agarose gel. The reactions were carried out either under standard conditions (lanes 1–6), without CTP and UTP (lanes 7–12), or without dCTP (lanes 13–18). [α-³²P]dGTP or [α-³²P]dCTP was used to label products of leading-strand or lagging-strand synthesis, respectively. (B) DNA replication reactions were carried out either under standard conditions (closed squares), or omitting the gp61 primase (closed circles), or omitting CTP and UTP (triangles), or omitting dCTP (diamonds), or with 10 µM dCTP (open squares) or with 2 µM dCTP (open circles). To monitor leading-strand synthesis, [α-³²P]dGTP was included. (C) DNA replication was conducted in standard reactions in the presence of [α-³²P]dGTP, 2.3 nM polymerase and 225 nM gp32 (squares), or without CTP and UTP (circles) or without dCTP (diamonds). Reaction products were separated in 0.6% alkali agarose gels and quantified. Leading-strand synthesis is expressed in percent, where 100% is incorporation at 8 min with CTP, UTP and dCTP present.

Figure 4.

Effect of decreased concentrations of dCTP on DNA synthesis catalyzed by T4 replication complexes reconstituted on a 240-nt circle. (A) DNA replication products synthesized in 1, 2 and 4 min were separated in a 0.6% alkali agarose gel. The reactions were carried out either under standard conditions (lanes 7–9), or without dCTP (lanes 1–3), or with 10 µM dGTP (lanes 4–6), or with 10 µM dCTP (lanes 10–12), or with 2 µM dCTP (lanes 13–15), or with 10 µM dCTP and 0.6 mM ddCTP (lanes 16–18); [α-³²P]dTTP was used to label products of leading-strand and lagging-strand synthesis. (B) Relative positions of Okazaki-fragment peaks formed by 4 min (a) in the standard reaction; (b) in a reaction with 10 µM dCTP; (c) in a reaction with 2 µM dCTP; (d), in a reaction with 10 µM dCTP and 0.6 mM ddCTP; (e) positions of 2.3-, 2.0- and 0.6-kb peaks of the molecular weight standard.

We also quantified the effects of blocking lagging-strand synthesis (by omitting dCTP, CTP + UTP or primase) on the rate of total synthesis of leading strands for up to 8 min (Fig. 3B). The rates of total leading-strand synthesis were decreased by 1.3- and 1.2-fold after 2 min and 2.2- and 1.7-fold after 8 min in reactions where lagging-strand synthesis was blocked at the RNA-priming step and at the DNA-synthesis step, respectively. Thus, the rates of total leading-strand synthesis are modestly reduced at 2 min, when leading-strand products extend up to 40 kb (Fig. 3A and data not shown), but are progressively reduced at later times (Fig. 3B). The rate of total leading-strand synthesis equals the rate of extension at an average replication fork multiplied by the number of replication forks. Because the rates of extension of leading-strands were unchanged (Figs 3A and 4A), we conclude that blocks to lagging-strand synthesis at both the RNA-primer step and the DNA-synthesis step inhibit the increase in the number of replication forks that occurs in the standard reaction. Furthermore, this inhibition is not due merely to the collapse of replication forks that encounter blocks to lagging-strand synthesis because, in the case of blocks to RNA priming, the lagging-strand polymerase is not yet loaded and replication complexes cannot be stalled or collapsed by blocks to RNA priming. We showed previously that replication complexes continue to form throughout an 8-min reaction (12) so that an excess of ssDNA, which forms in large amounts when lagging-strand synthesis is blocked, is expected to bind T4 proteins and thus inhibit the ongoing formation of replication complexes. Indeed, adding 1–4 nM M13mp2 ssDNA to a reaction without CTP + UTP further decreases the rate of leading-strand synthesis (data not shown). In addition, we explored the effects of reducing total synthesis of both strands ∼8-fold by decreasing the concentrations of gp43 and gp32 by 4-fold. Under these conditions, synthesis of leading and lagging strands remains coordinated and resists dilution (data not shown). When lagging-strand synthesis was then blocked, by omitting either dCTP or CTP + UTP, there was little or no inhibition of leading-strand synthesis (Fig. 3C). Thus, under these conditions, which we surmise reduce the amount of ssDNA, inhibiting coordinated lagging-strand synthesis at either the RNA-priming step or the DNA-synthesis step hardly affected leading-strand synthesis. Taken together, these results indicate that extremely early blocks to lagging-strand DNA synthesis (1–3 nt beyond the primer) usually do not block ongoing leading-strand synthesis, although they may inhibit the formation of new replication complexes.

Omitting dCTP blocks lagging-strand synthesis almost immediately after the polymerase loads onto a new RNA primer and begins to incorporate the first dNMPs. In order to examine how leading-strand synthesis responds to blocked lagging-strand synthesis when the polymerase encounters blocks randomly throughout the synthesis of a 1.5–3-kb Okazaki fragment, we simply decreased the dCTP concentration from 100 µM to either 10 or 2 µM in our standard replication buffer. With 10 or 2 µM dCTP, lagging-strand synthesis decreased from 50 to 43% or 29% of total synthesis, respectively, and the average size of Okazaki fragments also decreased (Fig. 4A and B), indicating that lagging-strand synthesis was inhibited. The rate of extension of leading strands did not change when the dCTP concentration was thusly decreased (Fig. 4A). Rates of total leading-strand synthesis in reactions containing 10 µM dCTP and 2 µM dCTP were higher than in a reaction lacking dCTP as well as in reactions without CTP + UTP or without the gp61 primase (Fig. 3B). Thus, lagging-strand blocks brought about by decreasing the dCTP concentration reduce leading-strand synthesis less than the very early blocks formed by omitting dCTP (Fig. 3B).

ddCTP can incorporate only into lagging strands in this minicircle system. Our first experiments, conducted in the presence of 100 µM dCTP plus 2.4 mM ddCTP, showed that a 24-fold molar excess of ddCTP relative to dCTP was required to substantially inhibit the synthesis of leading strands (Fig. 5A) and lagging strands (data not shown). Okazaki fragment size was unaffected by ddCTP (data not shown). All these results are consistent with the inclusion of gp43 in the Pol α family, whose members discriminate strongly against ddNMP incorporation (1). The same effects were observed when wild-type enzyme was replaced with the exonuclease-deficient D219A gp43 (data not shown), indicating that the 3′–5′ exonuclease of T4 DNA polymerase is not responsible for the discrimination of the enzyme against ddCTP. Surprisingly, we found that ddCTP similarly inhibited leading-strand synthesis in a reaction lacking gp61, where no lagging-strand synthesis occurs (Fig. 5A): in the presence of 100 µM dCTP, 2.4 mM ddCTP promptly inhibits leading-strand synthesis, presumably by reacting with the leading-strand polymerase complex. To reduce this inhibition, we lowered the dCTP concentration from 100 to 10 µM, whereupon 0.6 mM ddCTP inhibited leading-strand synthesis only slightly early (2 min) in a reaction lacking gp61 primase (Fig. 5B). Under these conditions, ddCTP decreased the fraction of lagging-strand synthesis from 43 to 23% of total synthesis without much reducing the average size of Okazaki fragments (Fig. 4B), and decreased total leading-strand synthesis (Fig. 5B). Note that the decrease in rates of total leading-strand synthesis was about the same when 0.6 mM ddCTP was added at 0 min or at 1.5 min (Fig. 5B). However, 0.6 mM ddCTP + 10 µM dCTP decreased the rate of leading-strand synthesis no more than did omitting dCTP. Moreover, blocking lagging-strand synthesis with 0.6 mM ddCTP + 10 µM dCTP did not decrease the rate of extension of leading strands (Fig. 4A). Thus, experiments with decreased concentrations of dCTP and ddCTP further support the conclusion that leading-strand synthesis in the T4 replication fork can continue despite blocks to lagging-strand synthesis.

Figure 5.

Effect of ddCTP on leading-strand synthesis catalyzed by T4 replication complexes reconstituted on a 240-nt circle. (A) Replication reactions were either standard (squares), or in the presence of 2.4 mM ddCTP (triangles), or without the gp61 primase (circles), or without the gp61 primase in the presence of 2.4 mM ddCTP (diamonds). (B) Replication reactions were carried out under standard conditions with the following changes: 10 µM dCTP (circles), 10 µM dCTP and 0.6 mM ddCTP (triangles), 10 µM dCTP with 0.6 mM ddCTP added at 1.5 min (crosses), no gp61 primase and 10 µM dCTP (diamonds), no gp61 primase, 10 µM dCTP and 0.6 mM ddCTP (closed squares), and dCTP omitted (open squares).

We next asked whether an abasic site blocks DNA synthesis. The data presented in Figure 6A show that a single abasic site introduced into 120mer linear (rather than circular) ssDNA at a position 99 nt downstream from its 5′ end is indeed a strong block for holoenzyme-catalyzed DNA extension. A block was also observed when the exonuclease-deficient polymerase was used (Fig. 6A), but after prolonged incubation (3 min versus 10 s), translesion products were observed, a result consistent with the idea that the 5′–3′ proofreading activity limits the ability of wild-type gp43 to perform translesion synthesis. Note that in the above substrate, the first site for deoxyribonucleotide addition to the ³²P-labeled 21mer is opposite the abasic site. However, the results were the same when the abasic site was in position +9 (data not shown). These results show that an abasic site can be used to selectively block lagging-strand synthesis.

In order to study the effects of blocking lagging-strand synthesis with an abasic site on the rate of extension of leading strands, we first constructed a substrate the same as the 240-nt circle used in our other experiments, but containing an abasic site in the lagging-strand template. With this substrate, lagging-strand synthesis should be blocked towards the end of the first Okazaki fragment. We first tested whether the abasic site indeed blocked lagging-strand synthesis. Fragments of ∼125 nt appeared upon DraI cleavage of these lagging-strand products (Fig. 6B). Because this fragment was absent when the normal substrate was used, and the fragment size is very close to its expected value of 126 nt, we conclude that synthesis of the first Okazaki fragment is indeed blocked. When the rate of extension of leading-strand products was compared on substrates with and without an abasic site, no difference was found (Fig. 6C). These results strongly support the idea that leading-strand synthesis in the T4 replication fork can continue despite blocks to lagging-strand synthesis.

DISCUSSION

T4 DNA replication reconstituted in vitro is a powerful system. Many aspects of DNA replication are well conserved from phages to eukaryotes and the T4 system is particularly well characterized. In addition, T4 replication proteins are available in highly purified milligram amounts (12,28). In this study, we characterized DNA synthesis catalyzed by T4 replication complexes reconstituted on synthetic circular substrates of different lengths already primed for leading-strand synthesis. Previously, 70-nt circles were used to reconstitute T4 and T7 replication complexes (10,11). However, it was not clear in those experiments whether the 70-nt substrates were used efficiently. An important related parameter is the rate of extension of leading-strands. In the T7 system, the rate was 300 nt/s, the same as when 7.2-kb M13 DNA was used (10). In the T4 system, the rate of extension was not reported (11). We found that total DNA synthesis and the rate of extension of leading strands are substantially lower with a 70mer circular substrate than with 120-, 240- or 7196-nt circles (Fig. 1B and C). Because the eight T4 replication proteins cannot form as potent replication complexes on the 70mer as they can on the longer substrates, we suspect that the circular part of the 70mer lacks enough space for the proteins to form efficient complexes.

Comparing total DNA synthesis on different substrates shows that T4 replication forks reconstituted on a 240-nt circular substrate catalyze DNA synthesis of both strands with about the same efficiency as with M13 DNA, a substrate used in previous studies to characterize reconstituted T4 replication complexes (12,25,29). Moreover, synthesis of 1.5–3-kb lagging strands by T4 replication forks reconstituted on the 240-nt circle was coupled with leading-strand synthesis. Thus, T4 replication forks reconstituted on a 240-nt circular substrate have the same properties as those described using much larger substrates (4,12).

Characterizing reconstituted T4 replication complexes showed that the lagging-strand polymerase is coupled to the rest of the complex (4,11,12; this study). The structural basis of this coupling remains unclear. It was initially suggested that two DNA polymerase molecules interact directly (4,5). However, the T4 DNA polymerase forms no detectable dimers in solution as judged by ultra-centrifugation or gel filtration (12,30). While some gp43–gp43 interactions have been detected (4,11), they could have been DNA mediated. For instance, ssDNA is required for the strong gp41–gp61 interaction (31) and for interactions between T7 DNA polymerase and DNA helicase (32).

T4 replication complexes reconstituted on the 240-nt circle provided a unique opportunity to measure the impact of blocks to lagging-strand DNA synthesis upon leading-strand synthesis during efficient coupled synthesis. Analyses of leading-strand products in reactions with lagging-strand synthesis blocked (by adding 0.6 mM ddCTP + 10 µM dCTP, or omitting dCTP, or decreasing the dCTP concentration from 100 to 10 or 2 µM, or with a single abasic site) showed that the rate of extension of leading strands was largely unaffected. (As expected, the rate extension of leading strands fell when the dGTP concentration was decreased from 100 to 10 µM.) One concern in interpreting this body of results is that, in reactions where lagging-strand DNA synthesis was inhibited, two classes of replication forks might exist, one synthesizing only leading strands and the other synthesizing both leading and lagging strands. However, this seems improbable for several reasons. First, when no lagging-strand blocks are imposed, synthesis of both strands is coupled in the large majority of forks (Fig. 2). Secondly, blocking lagging-strand synthesis at the level of DNA synthesis (such as by omitting dCTP or introducing an abasic site into the template) does not change the number of coupled replication forks before blocks are encountered, because loading the lagging-strand polymerase on the RNA primer is unaffected. Therefore, we conclude that most leading-strand synthesis neither stops nor pauses at blocks to lagging-strand synthesis; otherwise, the rate of leading-strand extension should have decreased substantially.

Nevertheless, blocking lagging-strand synthesis reduced total leading-strand synthesis slightly to moderately at late times in most reactions (Figs 3A and C and 5B), the strongest inhibition resulting when dCTP was omitted or when 0.6 mM ddCTP + 10 µM dCTP were included. Because total leading-strand synthesis was similarly decreased in the absence of blocks to ongoing lagging-strand DNA synthesis simply by omitting either CTP and UTP or the gp61 primase, most of the decrease in total leading-strand synthesis results from the absence of lagging-strand synthesis but is not mediated by blocking ongoing lagging-strand DNA synthesis. One possible explanation is that ssDNA tails accumulate in the absence of lagging-strand synthesis and sequester the T4 replication proteins (which have strong affinities for ssDNA), thus decreasing the number of replication complexes and lowering total DNA synthesis. Another possibility is that fully processive leading-strand synthesis requires the presence of concurrent lagging-strand synthesis, so that inhibiting lagging-strand synthesis in any way decreases the processivity of leading-strand synthesis. Because the amounts of leading-strand synthesis in reactions without dCTP or with 0.6 mM ddCTP + 10 µM dCTP were slightly less than in reactions lacking CTP + UTP or primase, T4 replication complexes may occasionally collapse upon encountering a lagging-strand block, but this would not affect our overall conclusion.

A previous study of T4 replication forks reconstituted on a 70-nt circular substrate showed that adding 0.6 mM ddCTP (which could incorporate only into lagging strands if at all) to a replication mixture containing 1.2 mM dCTP completely stalled leading-strand synthesis within 100 s (11). The different impacts of ddCTP in these two systems appear to reflect the different DNA substrates. The dG content of our 240-nt substrate (26.2%) is ∼2-fold lower than that of the 70-nt substrate (50%) used in the other study. With the high-dG substrate, lagging-strand synthesis would be interrupted more frequently by ddCTP; on average, ddCTP would compete with dCTP for every second polymerization event. This condition may render replication complexes particularly likely to stall or collapse.

The mechanism by which coupled leading strand synthesis continues despite blocks to lagging-strand synthesis remains unknown. According to one model, stalling a lagging-strand polymerase might uncouple it from the leading-strand complex, allowing the latter to continue (2,9). As a result, a long ssDNA gap should form beyond the block, which could later be filled by recombination-primed replication (33). The discontinuous nature of lagging-strand synthesis suggests another possibility: upon encountering a block to lagging-strand synthesis, the replication fork might bypass the lesion by dissociating the lagging-strand polymerase from the block while retaining it as part of the replication complex, and then reloading it onto the next RNA primer (17,18). As a result, the replication fork would proceed beyond the block, leaving a ssDNA gap that could be repaired later by a specialized pathway. This model is consistent with the observation that SV40 origin-dependent DNA synthesis catalyzed by human HeLa cell extracts is only blocked by a DNA lesion in the leading-strand template (34,35). This model resembles the completion of a cycle of Okazaki-fragment synthesis, in which the lagging-strand polymerase dissociates from the completed Okazaki fragment and is recycled by reloading onto the next RNA primer via protein–protein interactions with the rest of the replication complex. What, then, would make a stalled lagging-strand polymerase dissociate from the primer– template junction? One potential explanation is that the polymerase uses some of the energy of dNTP hydrolysis to maintain a conformation that favors polymerase binding to the primer–template junction. While stalled at a lesion, the enzyme would either add no dNMPs or would idle (incorporating and excising dNMPs), thus stopping or sputtering the energy of dNTP hydrolysis and promoting a polymerase conformation favoring dissociation. Alternative explanations are that the incorporation of any dNMP opposite a lesion is interpreted by the polymerase as a misinsertion that causes the polymerase to dissociate from the primer–template junction, or that protein–protein contacts tethering the lagging-strand polymerase to the rest of the complex cause the stalled polymerase to dissociate from the primer–template junction.