The Application of a Minicircle Substrate in the Study of the Coordinated T4 DNA Replication*

A reconstituted in vitro bacteriophage T4 DNA replication system was studied on a synthetic 70-mer minicircle substrate. This substrate was designed so that dGMP and dCMP were exclusively incorporated into the leading and the lagging strand, respectively. This design allows the simultaneous and independent measurement of the leading and lagging strand synthesis. In this paper, we report our results on the characterization of the 70-mer minicircle substrate. We show here that the minicircle substrate supports coordinated leading and lagging strand synthesis under the experimental conditions employed. The rate of the leading strand fork movement was at an average of ∼150 nucleotides/s. This rate decreased to less than 30 nucleotides/s when the helicase was omitted from the reaction. These results suggest that both the holoenzyme and the primosome can be simultaneously assembled onto the minicircle substrate. The lagging strand synthesized on this substrate is of an average of 1.5 kb, and the length of the Okazaki fragments increased with decreasing [rNTPs]. The proper response of the Okazaki fragment size toward the change of the priming signal further indicates a functional replisome assembled on the minicircle template. The effects of various protein components on the leading and lagging strand synthesis were also studied. The collective results indicate that coordinated strand synthesis only takes place within certain protein concentration ranges. The optimal protein levels of the proteins that constitute the T4 replisome generally bracket the concentrations of the same proteins in vivo. Omission of the primase has little effect on the rate of dNMP incorporation or the rate of the fork movement on the leading strand within the first 30 s of the reaction. This inhibition only becomes significant at later times of the reaction and may be associated with the accumulation of single-stranded DNA leading to the collapse of active replisomes.

The bacteriophage T4 DNA replication system has served as a successful working model for eukaryotic systems. A total of eight T4 proteins are required to reconstitute an in vitro rep-lication fork that carries out efficient and coordinated leading and lagging strand synthesis (1)(2)(3)(4). These eight proteins form the T4 DNA replisome that can be subdivided into one primosome complex and two holoenzyme complexes based on their different functionalities. These subassemblies are believed to be structurally and functionally integrated through proteinprotein interactions within the replisome.
Three protein components are involved in the assembly of the holoenzyme complex: gp43 1 (the polymerase), gp45 (the clamp protein), and gp44/62 (the clamp loader protein complex). The gp43 polymerase catalyzes the addition of a nucleotide monophosphate onto the 3Ј end of the nascent DNA chain and also contains an editing 3Ј-5Ј exonuclease activity (5). In the presence of the clamp protein, gp43 switches from a distributive to a processive enzyme (6 -8). gp45 is a trimeric protein that forms a sliding clamp circumscribing the primer/ template junction and increases the binding between gp43 and the DNA template (7). gp45 itself is loaded onto DNA by gp44/ 62, a molecular motor protein that utilizes energy from ATP hydrolysis for clamp loading (9 -11). The detailed kinetics of the multistep clamp loading process have been carefully studied by a series of pre-steady-state kinetics and stop-flow fluorescence resonance energy transfer experiments (12)(13)(14).
The primosome complex constitutes another important component of the T4 replisome consisting of the helicase (gp41), the primase (gp61), and the helicase accessory protein (gp59). gp41 helicase forms a ring-shaped hexameric structure in the presence of ATP/GTP (15). This gp41 hexamer is assembled on the lagging strand and unwinds the double-strand DNA in front of the moving fork in the 5Ј to 3Ј direction (16 -18). In the presence of gp32, gp59 facilitates the loading of the helicase through its interactions with both proteins (19 -22). gp61 is required to synthesize pentaribonucleotide primers for Okazaki fragment synthesis on the lagging strand. The recognition sequences for the T4 primase are 5Ј-GTT and 5Ј-GCT (23), with the former being the recognition site in vivo (24,25). The primase activity is modulated by the presence of other proteins. For example, gp61 activity is greatly stimulated by the gp41 helicase (24) and is further enhanced by the presence of both gp59 and gp32 (26). Recently, strong evidence for the formation of a hexameric ring structure of gp61 within the primosome has been provided through both kinetic and biophysical studies (26).
The single-stranded DNA-binding protein, gp32, is an important component of the T4 replisome. It exhibits strong cooperative single-stranded DNA binding ability and is presumably functioning in stabilizing the loop structure formed during replication (27). gp32 also interacts with a number of other T4 replication proteins in the replisome including gp59 and gp61 (22, 28 -29). Study of the C-terminal deletion mutant of gp32 suggested the role of gp32 in the coordination and stabilization of the T4 replisome (29). One proposed activity of gp32 during replication is to inhibit the random priming by the primase on the lagging strand so that the priming event is only possible when a primer is required for the synthesis of the Okazaki fragment (30).
In vivo DNA synthesis requires the coordinated action of all the functional units within the replisome. It was first suggested by Alberts and co-workers (31) that the two T4 holoenzyme complexes must be coupled during replication in order to explain how the tethered holoenzyme complexes carry out the synthesis of anti-parallel DNA double strands. Their trombone model was later adopted in the Escherichia coli (32) and T7 replication system (33). Studies of the E. coli system have provided the most convincing evidence for the presence of asymmetric and dimeric holoenzyme complexes (34 -36). A subunit has been shown to interact with both polymerases within the E. coli replisome (37). Although a direct physical link between the two polymerases has not been identified in the T7 system, dilution experiments did suggest that the lagging strand polymerase was highly processive and was recycled during repetitive cycles of the Okazaki fragment synthesis (38). Recent studies (39,40) likewise support a dimeric polymerase in the bacteriophage T4 system. Besides a link between the two polymerases, there must be a coupled movement between the holoenzyme complex and the primosome as well. An interaction between the C terminus of the gp4 helicase/primase and the polymerase was identified in the T7 system (41). Cha and Alberts (17) demonstrated that the T4 holoenzyme could perform rapid and processive synthesis with only the helicase present. Dong et al. (42) further showed that in the presence of a macromolecular crowding agent, rapid and processive synthesis can be carried out with a T4 polymerase/helicase "twoprotein" system. These results strongly suggested an interaction between the leading strand holoenzyme and components of the primosome.
The T4 DNA replication system has been studied on a tailed replicative form II (TRFII) DNA template constructed on M13 ssDNA. Only recently was a minicircle substrate utilized for more quantitative analyses (39,43). Compared with the TRFII M13 template, the minicircle system offers a number of advantages. With the minicircle substrate, the manipulation of the DNA sequence becomes possible. The design of the 70-mer minicircle substrate in this study enables the strand-specific incorporation of dGMP and dCMP into the leading and lagging strands, respectively. Two priming sites ϳ40 nucleotides apart on the lagging strand also are present in this substrate. Another advantage of using a small circular substrate is that the smaller size of the minicircle provides a higher fork number/ nucleotides ratio. As a result, a high concentration of the replication forks can be achieved in the reaction mixture. This allows the replication reactions to be carried out with the fork concentrations that are near or surpass the protein concentrations.
There are concerns, however, about using such a substrate. One of the potential problems with the 70-mer minicircle is that the small size of the substrate may impose steric constraints on the loading of all replisome components. In particular, the simultaneous loading of both the holoenzyme and the primosome may be hindered on a 70-mer DNA substrate (44). Should this be the case, the coordination between leading and lagging strand synthesis would be disrupted. In this paper, we assess the feasibility of the minicircle substrate as a fork template for the study of the coordinated T4 DNA replication.
We have monitored the replication fork movement under various experimental conditions. The rate of fork movement varied from the highest measured rate of 250 nt/s to an average rate around 150 nt/s. We have developed a dual-label method that enables us to quantify simultaneously the amount of leading and lagging strand synthesis in the same reaction mixture. This greatly reduces the experimental error associated with the study of coordinated synthesis in which leading and lagging strand synthesis is measured in separate reactions. Our results support the assembly of an intact and functional replisome on the 70-mer substrate that established coordinated strand synthesis and responded to the variation of individual replisome protein concentrations. The protein concentrations that support coordinated synthesis agree well with their in vivo concentrations. Furthermore, we studied the effect of blocking lagging strand synthesis on the leading strand. Preventing primer synthesis by omitting gp61 did not affect the rate of the leading strand synthesis at early reaction times, but the effect became significant at later reaction times. A possible role of the interaction between the accumulated ssDNA and T4 proteins such as gp32 is also discussed. In the accompanying paper (57) (26), and gp59 (22), were purified as described previously. All other chemicals were of analytical grade or better. The complex buffer used in all experiments was 25 mM Tris-OAc (pH 7.5), 150 mM KOAc, and 10 mM Mg(OAc) 2 .
The Standard Replication Conditions and the Filter Binding Assay-Replication reactions were carried out in a complex buffer containing 25 mM Tris acetate (pH 7.5), 125 mM KOAc, and 10 mM Mg(OAc) 2 . The standard replication conditions used in all minicircle reactions consisted of 100 nM minicircle substrate, 240 nM each of gp43, gp45 (as trimer), and gp44/62, 600 nM each of monomer gp41, gp61, and gp59, 4.5 M of gp32, 100 M each of CTP, GTP, and UTP, 2 mM ATP, 100 M each of dATP, dGTP, dCTP, and dTTP, [8-3 H]dGTP (6.9 Ci/mmol), and [␣-32 P]dCTP (3000 Ci/mmol), in a typical reaction volume of 100 l. Unless otherwise noted, the DNA polymerase holoenzymes (gp43, gp45, and gp44/62) were first preincubated at 37°C with the minicircle DNA template in the presence of 2 mM ATP for 30 s followed by the addition of the primosome (gp41, gp61, and gp59) and gp32 along with dNTPs, rNTPs, and 2 mM ATP. 10 -20 l aliquots were removed at various time points and quenched with an equal volume of 0.5 M EDTA. The quenched reaction aliquots were then spotted onto DE81 filter paper. All filter papers were allowed to air-dry and then washed in 300 mM ammonium formate buffer (pH 8.0) until no radioactivity was detected in the wash. The filter papers were washed twice with 95% ethanol and allowed to air-dry in a hood. The dried filter papers were placed in LSC vials with 5 ml of Ecoscint LSC mixture in each vial and counted with dual-channel liquid scintillation counting (channel settings are based on an arbitrary scale of 0 -1000. Channel 1 is set for 0 -400 and channel 2 for 400 -1000). The specific activities of both tritium and 32 P were calculated by direct counting of a known volume of the reaction mixture in LSC mixture. Control experiments showed that there was no quenching effect of 32 P on the filter paper. However, there was a 70% quenching of tritium by the filter paper. This effect was corrected in the calculation of the tritium-specific activity. The spill-over of 32 P radioactivity into channel 1 was determined to be 2.2% of the total 32 P radioactivity. The amount of dNMP incorporation (cpm) in the leading and the lagging strand was calculated according to Equations 1 and 2, 32 P ϭ 100 ϫ B/97. 8 (Eq. 1) where A and B are the counts in channel 1 and 2, respectively. The Rate of the Replication Fork Movement-Unless otherwise specified, replication reactions were carried out under the standard conditions with either [␣-32 P]dGTP (3000 Ci/mmol) or [␣-32 P]dCTP (3000 Ci/mmol). Aliquots of the reaction mixture were sampled at various time intervals and quenched with an equal volume of 0.5 M EDTA, pH 8.0. The DNA products were analyzed either through 0.8% alkalineagarose gel electrophoresis (30 mM NaOH and 5 mM EDTA) or through denaturing 4% PAGE. At the end of the separation, the alkaline gels were neutralized with 1 liter of TBE buffer, dried onto Whatman DE81 filter paper at room temperature for 12 h, and then dried under vacuum at 55°C for 1 h. Autoradiography was obtained using a Molecular Dynamics Storm 800 PhosphorImager system (Amersham Biosciences). The length of the DNA was determined using Quantity One Quantitation Software (Bio-Rad) and comparing it to 32 P-labeled DNA markers.
[␣-32 P]dCTP (3000 Ci/mmol) was included in the reactions for the detection of the lagging strand synthesis and was added 1 min after the initiation of the reaction. The reactions were allowed to proceed for another 3 min before being quenched in equal volume of 0.5 M EDTA, pH 8.0. Reaction products were analyzed by electrophoresis on a 0.8% alkaline-agarose gel, and the sizes of the Okazaki fragments were analyzed as described above.
Pre-steady-state DNA Replication Reactions-Pre-steady-state reactions were performed using a KinTek rapid chemical quench-flow instrument. All concentrations are after initiating the reaction. The reaction buffer was as described above. One sample syringe contained 100 nM minicircle template, 4.5 M gp32, 1 mM ATP, 100 M dNTPs, [␣-32 P]dGTP (3000 Ci/mmol), with or without 600 nM gp61, and with or without 100 M rNTPs. The other sample syringe contained 240 nM gp43, 240 nM gp45 (as trimer), 240 nM gp44/62, 600 nM gp41 (as monomer), and 600 nM gp59. All reaction mixtures were pre-equilibrated at 37°C. The reactions were initiated by rapid mixing of the contents in two syringes. The replication reaction was allowed to proceed for various time lengths, after which it was stopped by rapid addition of quench solution (pH 8.0, 500 mM EDTA). Reaction aliquots were spotted onto DE81 filter paper for determining the amount of incorporated radioactivity or analyzed by electrophoresis for assessing the rate of the fork movement as described above.

RESULTS AND DISCUSSION
The Rate of the Fork Movement-One of the concerns with using a 70-mer minicircle ( Fig. 1) substrate is whether it has sufficient space for the proper assembly of the replisome (44). The rate of the leading strand fork movement is a useful parameter for assessing the replisome assembly. Based on earlier results about the structure and stoichiometry of the T4 replisome, we set the following standard reaction condition to study replication on the 70-mer minicircle substrate: 100 nM minicircle substrate, 240 nM each of the holoenzyme components (gp43, gp45 trimer, and gp44/62), 600 nM each of the primosome components (gp41, gp61, and gp59, monomer concentration), and 4.5 M gp32 with the stoichiometry of DNA: gp43:gp45:gp44/62:gp41:gp61:gp59:gp32 of 1:2.4:2.4:2.4:6:6:6:45. The rate of the leading strand fork movement was measured under these conditions. Fig. 2A shows a gel of the leading strand fork movement (lanes 1-3). The highest fork rate measured was around 250 nt/s with a KinTek rapid quench instrument (data not shown). The average rate by manual mixing was 150 nt/s ( Fig. 2A). The higher rate probably reflects the capture of earlier events by the rapid quench method. The average rate is similar to the fork rate observed on a TRFII M13 template synthesized in our laboratory (57). Taken together, it seems that the fork rate on a minicircle substrate approaches that observed on a larger circular substrate. Other published reports describing T4 DNA replication on a 70-mer minicircle template showed that the fork extended only at a rate of 50 nt/s. Moreover, this rate was not affected by the presence or absence of the primosome, indicating that the loading of the primosome onto the 70-mer minicircle was hampered (44). We note that these reactions were carried out at lower concentrations of both the minicircle substrate and the T4 replication proteins. It therefore appears that for the 70-mer minicircle substrate, higher concentrations of both the minicircle and the replication proteins are needed to assemble the functional replisome.
The need for higher protein component concentrations for the minicircle substrate suggests decreased protein/protein and protein/DNA affinities within the assembled replisome. In particular, the assembly of the primosome and its contacts with the holoenzymes might be altered. We next determined the effect of the primosome components on the rate of the fork movement to test for the capability of the minicircle substrate to accommodate both the holoenzyme and the primosome complexes. The rate of the fork movement was measured for the omission of either gp59 alone or both gp41 and gp59. The order of addition of proteins features the assembly of the holoenzyme prior to that of the primosome. If the size of the minicircle substrate is not big enough for loading both the holoenzyme and the primosome, one would expect that the latter would not assemble. This was clearly not the case under our experimental conditions. Omission of either gp59 or both gp41/gp59 markedly decreased the rate of the fork movement, as compared with the reaction in which all eight proteins were present (Fig.  2, A and B).
As shown in Fig. 2A (lanes 4 -9), when gp59 was lowered to 10 nM or was omitted from the reaction, two populations of forks were observed that moved at different rates. The major population moves slowly at a rate of less than 30 nt/s. The minor population moves at a much faster rate of ϳ150 nt/s. In order to test whether or not the slower moving fork was due to the absence of the helicase in the replisome, both gp59 and gp41 were omitted from the reaction mixture, and the leading strand fork movement was measured under this condition. The faster moving fork fraction was now completely abolished, and all the forks moved at a rate of less than 30 nt/s (Fig. 2B). This result indicated that the faster moving minor population had gp41 in its replisome spontaneously loaded in the absence or at low concentrations of gp59. Correspondingly, the slower moving forks were defective in gp41 loading and performed DNA synthesis at a much slower rate. This result not only reinforces the role of gp59 in gp41 loading but also demonstrates the assembly of an intact replisome containing both the holoenzyme and the primosome complexes on the 70-mer minicircle DNA in the presence of all eight T4 proteins. Therefore, the strains mentioned above, if present, are not severe enough to affect the proper assembly of the replisome on the minicircle. Note also that the fast moving forks assembled in the absence of gp59 still travel at the same rate as the completely assembled forks, suggesting that gp59 does not affect the fork movement rate on the leading strand.
Effect of rNTPs Concentrations on the Size of the Okazaki Fragments-It has been shown in the E. coli replication system that the size of the Okazaki fragments can be altered under conditions that modify the priming on the lagging strand (49). For example, in both the T4 and the E. coli systems, an increase of the size of the Okazaki fragments was observed as the rNTPs concentration decreased (49,50). A rationale for this response is a slowing in the priming frequency at different rNTPs concentrations (51). Because Okazaki fragment synthesis can only initiate when there is an RNA primer available, a decrease in the primer synthesis rate leads to an increase in the size of the Okazaki fragment.
We studied the size variation of the Okazaki fragments upon dilution of rNTP concentration to test whether or not such a response existed with the minicircle substrate. As shown in Fig. 3, decreasing the rNTPs concentration from 200 to 12.5 M caused a shift of the average Okazaki fragment size from 1.3 to 2.5 kb. Furthermore, Okazaki fragments of 5-6 kb long were observed at 12.5 M rNTPs. Such long Okazaki fragments were absent at 200 M rNTPs.
As presented in the accompanying paper (57), varying gp45, gp44/62, or gp61 concentrations also has a marked effect on the length of the Okazaki fragments in the minicircle system. Priming events had been proposed to govern the repetitive cycle of the Okazaki fragment synthesis through temporally and spatially controlled protein-protein interactions within the replisome (49). The correctly assembled replisome therefore should have the ability to respond to different conditions that modulate the priming events. These results provide additional  1-3), were used in these reactions. Aliquots were analyzed on an alkaline-agarose gel.

FIG. 2. The leading strand fork rate.
A, the leading strand fork rate on the minicircle template. Reactions were carried out under the standard conditions indicated under "Materials and Methods" with [␣-32 P]dGTP to monitor the leading strand synthesis. Three gp59 concentrations were used in these reactions. Lanes 1-3, 600 nM gp59; lanes 4 -6, 10 nM gp59; and lanes 7-9, 0 nM gp59. For each reaction, 3 aliquots were taken at 20, 40, and 60 s. The samples were treated and analyzed on an alkaline-agarose gel as described under "Materials and Methods." B, the leading strand fork rate on the minicircle template in the absence of gp41 and gp59. The reaction was carried out under the standard conditions indicated under "Materials and Methods" with [␣-32 P]dGTP to monitor the leading strand synthesis. Both gp41 and gp59 were omitted from the reaction. Five aliquots were withdrawn from the reaction at 20, 40, 60, 90, and 120 s (lanes 1-5). The samples were treated and analyzed on an alkaline-agarose gel as described under "Materials and Methods." evidence for an active and properly assembled replisome on the minicircle substrate.
Characterization of the Coordinated Replication Process-Leading and lagging strand synthesis on a minicircle substrate can be monitored and quantified independently and simultaneously. Because the leading strand template is devoid of guanine residues and the lagging strand template cytosine residues, dGMP is only incorporated into the leading strand. Likewise, dCMP is only incorporated into the lagging strand. The amount of the leading and lagging strand synthesis can thus be monitored independently by using different radiolabeled deoxyribonucleoside triphosphates (dGTP or dCTP). The leading and lagging strand syntheses were measured in a single reaction mixture by determining the incorporation of [ 3 H]dGTP and [␣-32 P]dCTP via dual-channel liquid scintillation counting. This method allows a direct comparison of leading and lagging strand synthesis in the same reaction mixture and a better assessment of the coordinated replication with greatly reduced experimental error. Under the optimal conditions employed in this study, the minicircle substrate supports coordinated leading and lagging strand synthesis. A typical coordinated strand synthesis is shown in Fig. 4.
As has been observed in the T7 system (38), two features are noted for coordinated synthesis in the T4 system. First, the rates of the leading and lagging strand synthesis are identical, as shown by the same slope for both strand synthesis (1.37 and 1.23 pmol/s), suggesting that although the functions of the holoenzyme on the leading and lagging strand are asymmetric, they replicate in a coordinated fashion. Second, there is a distinct lag phase for the lagging strand synthesis. This lag phase presumably represents the time required for the leading strand synthesis to displace a critical length of the singlestrand template in order for the lagging strand holoenzyme to assemble and replicate. This lag phase was further characterized by measuring the rate of both strand syntheses on a shorter time scale. As shown in Fig. 5, a lag of ϳ15 s was identified for the lagging strand synthesis. We concluded that the minicircle system would support coordinated leading and lagging strand syntheses and turned to further optimize the concentrations of the component proteins.
Effects of Varying the Holoenzyme Components-When the holoenzyme components were varied individually, they revealed marked effects on the amount of DNA synthesis as well as on the coordination between leading and lagging strand synthesis (Table I). In these experiments, five gp43 concentrations (24, 56, 120, 240, and 480 nM) and three gp45 (as trimer) and gp44/62 concentrations (28,84, and 252 nM) were studied. When the concentration of either gp43, gp45, or gp44/62 was decreased, the amount of synthesis of both strands decreased correspondingly (Fig. 6).
For leading strand synthesis, a 9-fold dilution of [gp45] from 252 to 28 nM decreased the rate proportionally 10-fold from 1.9 to 0.20 pmol/s (Fig. 6A). For the clamp loader protein, however, the rate of leading strand synthesis decreased by only 2-fold (from 2.13 to 1.10 pmol/s) after a 9-fold dilution from 252 to 28  nM (Fig. 6C). The slight effect of [gp44/62] on the synthesis rate is consistent with its catalytic function during holoenzyme assembly (9 -11). An optimal leading strand synthesis rate of ϳ1.5 pmol/s was achieved at 240 -480 nM gp43, and the rate decreased only slightly at 120 nM gp43 (ϳ1.38 pmol/s). Further lowering the [gp43] to 24 nM, a 10-fold dilution, decreased the rate to 0.42 pmol/s or 3.5-fold (Fig. 6E).
The difference in behavior between the clamp and polymerase proteins cannot be simply attributed to the instability of the former as a trimer at low protein concentrations. The K D value for the trimer 7 monomer equilibrium has been measured by fluorescence resonance energy transfer (52) providing a value of 0.08 M 2 assuming cooperative binding. Within the concentration range of 252 to 28 nM, the gp45 trimer would decrease ϳ40-fold from 175 to 4.4 nM. The presence of all replisome proteins must stabilize the gp45 trimer. The decrease in the rate of leading strand synthesis with gp45 concentration hence appears to correlate with a decrease in the concentration of the active leading strand holoenzyme. The effect of changing levels on polymerase, on the other hand, should be primarily reflected in its affinity for DNA and suggests that at 24 nM gp43, the concentration of the leading strand holoenzyme only decreases moderately.
Next, we examined the effect of changing the concentration of the holoenzyme components on lagging strand synthesis as well as on the coupling between two strand synthesis. The rate of lagging strand synthesis decreased by ϳ370-fold from 1.88 to 0.005 pmol/s after only a 9-fold dilution of [gp45] from 252 to 28 nM (Fig. 6B). Coordinated leading/lagging strand synthesis was observed at the highest [gp45] (252 nM) in this experiment but was already disrupted when [gp45] was dropped to 84 nM (0.65 and 0.26 pmol/s for the leading and lagging strand, respectively).
The effect of changing gp45 levels contrasted sharply with that of gp44/62 (Fig. 6D) and gp43 (Fig. 6F). A 9-fold dilution of  the clamp loader protein decreased the rate of lagging strand synthesis by only 3.8-fold (from 2.0 to 0.53 pmol/s). Coordinated leading/lagging strand synthesis, which was observed at 252 nM gp44/62, was only slightly disrupted at 84 nM gp44/62 when the rate of leading and lagging synthesis was 1.71 and 1.43 pmol/s, respectively. The effect of gp43 levels on lagging strand synthesis was very similar to that on the leading strand (Fig. 6,  E and F). The rate decreased ϳ2-fold (1.53 to 0.81 pmol/s) over a 9-fold change in gp43 levels (480 to 56 nM). The lagging strand synthesis was affected significantly only at a very low gp43 concentration (24 nM) with the rate decreased to 0.21 pmol/s. Coordinated synthesis was maintained over a wide range of gp43 levels from 56 to 480 nM. Apparent K m values of 55 and 81 nM for gp43 were obtained for the leading and lagging strand, respectively. A V max of 1.8 pmol/s was obtained for both the leading and lagging strand synthesis.
Collectively, our results indicate that gp45, when compared with gp43 and gp44/62, has a much more significant effect on lagging strand synthesis. In the accompanying paper (57), we demonstrated that the clamp protein was recruited from solution during repetitive Okazaki fragment synthesis. Presuming that the clamp protein is somehow involved in RNA primer capture, a decrease in the gp45 level would severely affect the efficiency of primer utilization during the lagging strand synthesis. The difference in the stringency for the requirement of gp45 and gp44/62 during replication can also be seen from the effect of changing their levels on the size of the Okazaki fragments. Lowering [gp45] produced a much longer Okazaki fragment than lowering [gp44/62] by the same fold (57).
When the holoenzyme concentration was varied as a whole, i.e. all components gp44/62, gp45, and gp43 were kept at a fixed ratio of 1:1:1 but their level was changed (30, 60, 120, 240, and 480 nM), the effect on the leading and lagging strand synthesis was largely determined by the effect of gp45 (Fig. 6, G and H) which parallels the behavior noted in Fig. 6, A and B. A decrease by a factor of 10 was observed in the rate of the leading strand synthesis as the holoenzyme concentration dropped from 240 to 30 nM. At 30 nM, the lagging strand synthesis was almost completely eliminated. These results further demonstrate that gp45 is the limiting factor in the assembly of the holoenzyme complex.
Effects of Varying the Primosome Components-Next, the primosome components were varied individually to study their effects on the leading and lagging strand synthesis. As expected, when the concentration of either of the primosome components (gp41, gp61, gp32, and gp59) decreased, the lagging strand synthesis also decreased. When they were omitted from the reaction mixture, lagging strand synthesis was largely or entirely eliminated (Table II). All the concentrations below represent the monomeric protein concentrations.
Between 100 and 600 nM, the effects of gp41, gp61, and gp59 were similar on the leading strand synthesis (Fig. 7, A, C, and  E). For example, 150 nM of gp59 induced an optimal synthesis rate at the leading strand (1.8 pmol/s), and a further increase to 600 nM did not change the rate significantly. The same pattern was observed with gp61. For gp41, varying the level from 144 to 600 nM only induced less than a 2-fold increase in the leading strand rate. For all three proteins, coordinated synthesis was maintained between ϳ100 and 600 nM. The only notable difference on leading strand synthesis was when any of these proteins was omitted from the reaction. Omission of gp41 from the reaction decreased the rate of the leading strand synthesis to 0.26 pmol/s and could be attributed to the slowing of the replication fork in the absence of gp41. Because gp59 functions primarily through facilitating gp41 loading, its absence also inhibits leading strand synthesis (0.40 pmol/s). Changing the levels of gp61, on the other hand, did not change significantly the rate of the leading strand synthesis over the entire concen-tration range studied. These results support the observation that among the primosome components, only the helicase, gp41, is needed to establish the rapid and processive replication by the holoenzyme (17).
The effects of gp41, gp61, and gp59 were also quite similar on the lagging strand synthesis (Fig. 7, B, D, and F). There was virtually no lagging strand synthesis when either gp41 or gp61 was omitted from the reaction. The residual rate observed was probably due to the misincorporation of [ 32 P]dGTP into the lagging strand. Lagging strand synthesis was observed at a very low rate of 0.06 pmol/s in the absence of gp59, a synthesis possibly induced by spontaneous gp41 loading. The rate of lagging strand synthesis increased from a residual rate of 0.0045 pmol/s in the absence of gp41 to 1.04 pmol/s at 144 nM gp41. Further raising the gp41 concentration ϳ4-fold to 600 nM increased the rate of the lagging strand synthesis by only Replication reactions were carried out essentially as described in Fig. 8 except that only the effect of primase was studied. The rNTP concentration was kept at 100 M in all reactions. A, reaction aliquots were analyzed by the filter binding assay to determine the amount of total DNA synthesis in the leading strand. OE, reaction without primase; f, reaction with 600 nM primase. B, for each reaction, three time points (15,30, and 45 s) were taken, and the samples were analyzed on a 1% alkaline-agarose gel for the measurement of the fork extension rate. ϳ1.5-fold (Fig. 7B). Similar effects were observed when [gp61] was varied (Fig. 7D). The rate of lagging strand synthesis increased from 0.0054 pmol/s in the absence of gp61 to 0.81 pmol/s at 114 nM primase. This rate, however, did not change significantly when the primase concentration was further raised to 600 nM (0.79 pmol/s). This rate was lower than the optimal rate of ϳ1.8 pmol/s because less gp43 (80 nM) was used in this reaction. The rate of lagging strand synthesis remained constant and averaged ϳ1.75 pmol/s when [gp59] was between 150 and 600 nM (Fig. 7F). Higher concentration of either gp61 or gp59 inhibited both strand syntheses.
From the above study, it can be seen that for all three proteins (gp41, gp61, and gp59), a concentration around 100 nM induces near-optimal synthesis in both strands. Moreover, in all three cases, a further increase to 600 nM does not change the rate of strand synthesis significantly. This result provides indirect evidence for a stoichiometry of gp41:gp61:gp59 ϭ 1:1:1 during primosome assembly, an observation made in our laboratory from biophysical studies of the primosome assembly process. 3 One interesting observation from the above study was that when gp61 was omitted from the reaction, DNA replication rate of the leading strand did not deviate from its optimal rate until after ϳ60 s (Fig. 7C, inset). Similar results have been observed on a larger (240-mer) minicircle substrate (44). To characterize this phenomenon further, we studied the effect of blocking the lagging strand on the leading strand synthesis within a shorter time frame. The leading strand synthesis was monitored from 0 to 10 s (Fig. 8) and from 10 to 30 s (Fig. 9) under conditions described in the figure legends.
The results indicate that the omission of either the primase or rNTPs has no significant effect on the leading strand synthesis during the first 30 s of the reaction. Both the rate of the total nucleotide incorporation and the rate of the leading strand fork movement remained the same regardless of the presence or absence of the lagging strand synthesis. During the steady-state phase (Ͼ60 s), however, the leading strand synthesis started to be inhibited in the absence of the primase (Fig.  7C, inset). This inhibitory effect became more significant as the reaction proceeded. It is possible that during the first 10 s of the reaction, the lagging strand holoenzyme is not assembled, and the replisome contains only the leading strand holoenzyme and the primosome. However, from 10 to 30 s, the assembly of the lagging strand holoenzyme should take place because lagging strand synthesis initiates at ϳ15 s (Fig. 5). Within both time frames, the leading strand synthesis was not affected by the presence or the absence of the lagging strand synthesis. These results suggest that, in the presence of the primosome, the function of the leading strand holoenzyme is not inhibited, nor is it stimulated, by the assembly of the third member of the replisome complex, the lagging strand holoenzyme.
The inhibition of leading strand synthesis in the absence of lagging strand after 1 min of the reaction can be a result of the accumulated ssDNA in the reaction mixture which could interact with and sequester the T4 replication proteins (44). Because the fork rate remains constant under these conditions, the inhibition must be due to a decrease in the amount of active complexes in the reaction mixture. Because no significant amount of active complex formation after 1 min of the reaction was observed (t1 ⁄2 ϭ 24 s, see accompanying paper (57)), we suggest that under our conditions it is primarily the collapse of the active forks that contributes to the inhibition of the leading strand synthesis. With the accumulation of ssDNA, gp32 is most likely to be the protein that is trapped. Our DNase I footprinting experiments show that among the four primosome proteins (gp41, gp61, gp32, and gp59), only gp32 forms a stable complex with ssDNA and protects it from nuclease digestion (data not shown). Although gp41, gp61, and gp59 are all known to bind ssDNA, they may have fast off rates so that no footprinting was observed.
Several lines of evidence suggested that gp32 was involved in the organization of the replisome. Indeed, gp32 has interactions with a number of T4 replication proteins (22, 28 -29) and has an active role in facilitating holoenzyme assembly and improving its stability (53,54). An earlier study on gp32 demonstrated the importance of this protein on the replication of the minicircle (39). Data generated in this study show that lowering [gp32] decreased significantly the rate of not only the lagging strand but also the leading strand synthesis (Fig. 7, G  and H). Omission of gp32 almost eliminated entirely the lagging strand synthesis (0.0089 pmol/s) and decreased the leading strand rate by ϳ5-fold. The trapping of gp32 from the replisome could disrupt the protein-protein interactions within the replisome and lead to its collapse. Further studies will likely reveal the detailed functions of gp32 within the replisome.
Results from the above experiments indicate that the rates of the leading and lagging strand synthesis vary at different protein concentrations, and the coordinated synthesis can be supported only within certain protein concentration ranges. Specifically, about 200 nM each of the holoenzyme components (gp43, gp45, and gp44/62) and between 100 and 600 nM each of the primosome components gp41, gp61, and gp59 are needed for coordinated synthesis on 100 nM of minicircle template. Higher concentrations of primosome proteins generally inhibit replication, whereas lower concentrations of replisome proteins disrupt coordinated synthesis. Our results also suggest that the trapping of proteins such as gp32 by ssDNA could result in the collapse of the active replisome. DNA replication requires the coordinated actions of multiple macromolecular com-  ponents. The proper interactions among these components are the key to the successful execution of the replication process. Thus, maintaining optimal system equilibria is important for efficient and coordinated DNA synthesis. Disruption of such equilibria by the imbalance of protein concentration(s) affects significantly the coordinated strand synthesis.
To this end, it is of interest to compare the optimal protein concentrations obtained in our studies on the minicircle substrate with their cellular concentrations (Table III) (55). The in vivo fork concentration is around 60 nM, which is close to the 100 nM fork concentration employed in the current studies. As shown in Table III, the cellular concentrations of most of the replication proteins are generally high, being in the upper nanomolar to micromolar range. These concentrations should reflect favorable system equilibria inside a cell dictated by individual K D values. The protein concentrations determined in the current studies in general agree with those found in vivo. The only notable difference is that gp45 has a higher in vivo concentration around 10 M. We observed no significant increase in the synthesis rate of both strands at higher gp45 concentration (500 nM) in our in vitro assay on the minicircle substrate. The high in vivo gp45 concentration is likely due to the fact that gp45 is also actively involved in the transcriptional activation of bacteriophage T4 (56). CONCLUSIONS Reconstituted in vitro T4 bacteriophage DNA replication system with eight replication proteins on a 70-mer minicircle template is shown here to be active and capable of coordinated synthesis of both the leading and lagging strand. The coordination between leading and lagging strand synthesis was shown by the identical rate of the nucleotide incorporation into both strands, as well as by the presence of a distinct lag phase of lagging strand synthesis. Under optimal conditions, the rate of the replication fork extension averages at 150 nt/s among different measurements with the highest measured rate of 250 nt/s. The average fork rate is similar to that measured on a TRFII M13 replication fork in our laboratory. It is possible that the small size of the minicircle substrate may create strains within the replisome. However, these strains, if present, are not severe enough to cause the collapse of the replisome complex. Our results have shown that both the holoenzyme and the primosome can be loaded simultaneously onto the 70-mer minicircle DNA. Furthermore, the replication fork responds to the change in rNTPs concentrations by producing longer Okazaki fragments at lower [rNTPs]. This result suggests functional lagging strand synthesis executed by the replisome complex reconstituted on the minicircle substrate. As shown in the accompanying paper (57), the size of the Okazaki fragments also varies with varying concentrations of certain other replisome proteins. All of these results suggest that a 70-mer minicircle template is a useful substrate for the study of the coordinated DNA synthesis in the T4 system.
Both leading and lagging strand synthesis respond to the variation in concentration of individual replisome protein components. We have identified the concentration ranges of various replisome components within which the coordinated synthesis can be supported. The concentration effect on the strand synthesis was most significant for gp45 among the three holoenzyme components, suggesting that gp45 is the limiting factor in holoenzyme assembly. This result is also in agreement with the distributive nature of gp45 during the repetitive lagging strand synthesis. The effects of gp41, gp61, and gp59 were very similar on both strand synthesis as well as on the coupling between leading and lagging strand, providing kinetic evidence for a 1:1:1 stoichiometry of these three proteins during primosome assembly. Our determined protein concentration ranges generally agree with those found in vivo, suggesting that optimal system equilibria are achieved during coordinated replication.
Blocking lagging strand synthesis by omitting gp61 or rNTPs from the reaction mixture did not inhibit the leading strand synthesis during early stage of the reaction (0 -30 s). Neither the rate of dNMP incorporation nor the rate of leading strand fork movement was affected. The inhibition became significant after 1 min when the reaction proceeded into the steady-state phase. We suggest that the accumulation of ssDNA in the reaction traps replication proteins, in particular gp32, and that this trapping can lead to the collapse of the active replisomes.