Insights into Okazaki Fragment Synthesis by the T4 Replisome

Background: The T4 replisome duplicates the lagging DNA strand through discontinuous Okazaki fragments. Results: A third polymerase is not involved in repetitive Okazaki fragment synthesis. Conclusion: The T4 replisome recycles the lagging-strand polymerase but looses the clamp in each Okazaki fragment cycle through a collision or signaling pathway. Significance: Elucidation of the behavior of polymerase, clamp and clamp loader for Okazaki fragment synthesis broadens our understanding of coordinated DNA replication. In this study, we employed a circular replication substrate with a low priming site frequency (1 site/1.1 kb) to quantitatively examine the size distribution and formation pattern of Okazaki fragments. Replication reactions by the T4 replisome on this substrate yielded a patterned series of Okazaki fragments whose size distribution shifted through collision and signaling mechanisms as the gp44/62 clamp loader levels changed but was insensitive to changes in the gp43 polymerase concentration, as expected for a processive, recycled lagging-strand polymerase. In addition, we showed that only one gp45 clamp is continuously associated with the replisome and that no additional clamps accumulate on the DNA, providing further evidence that the clamp departs, whereas the polymerase is recycled upon completion of an Okazaki fragment synthesis cycle. We found no support for the participation of a third polymerase in Okazaki fragment synthesis.

Polymerases together with other replication components organize into a multiprotein complex, the replisome, to collaboratively duplicate the genomic DNA of all living organisms (1)(2)(3)(4). Possessing the fundamental set of protein constituents common to eukaryotic counterparts, the T4 bacteriophage DNA replisome has long served as a model for understanding the DNA replication process. The current working model of the T4 system faithfully portrays the mode of action of replisomes in general. At the front of a moving fork, the hexameric helicase gene product 41 (gp41) 2 , encircling the lagging strand template, separates the two parental strands for duplication by the replication core (5)(6)(7)(8)(9)(10). The core, comprised of a leading-and lagging-strand holoenzyme, each containing a gp43 polymerase and gp45 sliding clamp, the latter loaded on the DNA by the gp44/62 clamp loader, replicates the DNA (5, 7, 10 -13). Unlike the continuous advance of leading-strand synthesis in the direction of fork progression, the lagging strand is extended in the opposite direction through 1-to 2-kb-long, discrete Okazaki fragments (OFs) initiated repetitively by extending a pentaribonucleotide primer synthesized by the gp61 primase complexed to the helicase (6,8,14). Lastly, gp32 single-stranded DNA-binding protein facilitates lagging-strand synthesis by stabilizing the single-stranded intermediates and removing DNA secondary structure (15,16).
Lines of evidence from a number of DNA replication systems support the concept that leading and lagging strands are duplicated at the same net rate despite the disproportionate enzymatic reactions between the "non-stop" leading-strand synthesis and the start and stop lagging-strand synthesis (16 -21). To explain the coupling, Alberts et al. (17) proposed the "trombone model" to illustrate how polymerases on two strands coordinate to copy both strands simultaneously. In this textbook model, lagging-strand polymerase, upon completion of an OF synthesis cycle, disengages from one clamp, transiently releases from the DNA but remains within the replisome, and cycles rapidly to the next primer to begin a new cycle. Polymerases on both strands act such that the nascent lagging strand folds back to form a "trombone loop" during each cycle.
Polymerase recycling, a key element in this elegant model, has been validated by several landmark biochemical studies: 1) lagging-strand synthesis is resistant to polymerase dilution in ensemble studies and in single-molecule investigations (22,23); 2) OF size distribution is independent of polymerase concentration (17,24); and 3) leading-and lagging-strand polymerases are all bound within a replisome via protein-protein interactions (25,26).
It is generally accepted that, concomitant to the recycling process of the lagging-strand polymerase in rounds of OF synthesis, clamps are left behind as the polymerase hops from one OF to initiate the next at a RNA primer and associates with a new clamp loaded from solution. In support of this, T4 gp45 and Escherichia coli ␤ clamps have been shown to be dissocia-tive replisomal components and are recruited from solution in each round of OF cycle (27,28). Moreover, the accumulation of multiple clamps on DNA as rounds of OF synthesis proceed has been confirmed in in vitro replication for the E. coli replisome (29,30). However, it has yet to be established whether gp45 dissociates or remains with an OF during lagging-strand replication in the T4 system.
Recent efforts to further define the stoichiometry of the replisome and the highly dynamic lagging-strand synthesis have uncovered several unexpected behaviors of the polymerase and clamp during OF synthesis. On one hand, Lia et al. (31) observed that the E. coli Pol IIIs within a replisome exchange with pool polymerases in living E. coli cells and have proposed that the exchange only occurs on the lagging strand for OF synthesis, an observation in conflict with the prevailing evidence for a recycling lagging-strand polymerase. In addition, two trombone loops at a moving fork in the T4 system were observed with a low frequency in EM studies, possibly implicating an additional "third" polymerase recruited from solution for lagging-strand synthesis (9). A third polymerase in the E. coli replisome for lagging-strand replication has also been reported (32,33). On the other hand, in opposition to the hypothesis of clamp accumulation on the lagging strand during its replication, Leake et al. (33) showed that E. coli tightly regulates the stoichiometry of ␤ clamps and that only two to three ␤ dimers are present at the replication loci. Furthermore, van Oijen et al. (34) reported that the same ␤ clamp is utilized for ϳ10 rounds of OF synthesis by recycling in a ␤-depleted solution.
These and other new findings have prompted us to revisit the recruitment process of gp43 and examine the fate of gp45 on DNA during lagging-strand synthesis in the T4 system. We used a circular DNA substrate of 1.1 kb with only one priming site (5Ј-GTT-3Ј) in rolling circle reactions by the T4 replisome to enable quantitative analysis of the size distribution of the OFs. We partitioned the OFs into two populations: one derived from a collision mechanism and the other from signaling. We demonstrated that the distribution of OF lengths showed dependence on both gp44/62 and gp32 levels but was independent of the concentration of the exonuclease-deficient gp43 polymerase (herein referred to as gp43(exo-)). The signaling population was sensitive to the level of the gp44/62 clamp loader. By pull-down and quantification of gp45 bound on DNA after the replication reaction, we found no accumulation of gp45 on DNA. Only one clamp, most likely the one associated with the leading-strand polymerase, remained.
Preparation of Rolling Circle Amplification Substrates-The Tailed Replicative Form II with a 5Ј-biotinylated and Cy3-labeled flap (Bio/Cy3 TRFII) was synthesized essentially as reported previously with a few modifications (27). In brief, primer annealing was performed in a 1.5-ml centrifuge tube with 50 nM M13mp18 (7249-bp) single-stranded circular chromosome and 100 nM primer of 70-nt [5Ј-BioTEG-T(iCy3)-T(39)-TGC GCT TAA TGC GCC GCT ACA GGG CGC GTA C-3Ј], which was heated in a water bath at 85°C for 3 min and slowly cooled to room temperature. The annealed singlestranded DNA was converted to the double-stranded form in a reaction with 150 nM gp43(exo-) and 100 M dNTP set in complex buffer (20 mM Tris (pH 7.5), 150 mM potassium acetate, 10 mM magnesium acetate) with 2.5% dimethyl sulfoxide at 37°C for 1 h. After extraction with phenol/chloroform twice, the DNA product was further purified by NeutrAvidin agarose resin to remove free Cy3 dye and non-biotinylated substrate. Neutral agarose gel electrophoresis of the resulting DNA substrate showed quantitative conversion to the doublestranded form. Preparation of the pONE-Nick substrate was carried out as described elsewhere (35).
Rolling Circle Amplification Reactions Titrated with the T4 Proteins-A standard rolling circle replication reaction in complex buffer contained 3 nM pONE-Nick; 200 nM gp43(exo-); 200 nM gp44/62; 200 nM trimeric gp45; 4 M gp32; 200 nM gp41; 200 nM gp61; 200 nM gp59; 100 M CTP, GTP, and UTP; 1 mM ATP; and 100 M dNTPs. In the titrations, concentrations of the titrating protein can be found in the figure legends. Other components remained the same as those in the standard replication reaction. Reactions were allowed to proceed for 7 min for replisome assembly before the addition of 10 Ci of [␣-32 P]dCTP. Aliquots of the reaction mixture were quenched with an equal volume of 500 mM EDTA at 5, 10, and 15 min and analyzed by 0.7% alkaline-agarose gel electrophoresis. The percentages of the OF populations generated in the rolling circle reactions with the pONE-Nick substrate were calculated on the basis of the area integrations of signal intensity of each peak corrected with their sizes, which are multiples of 1.1 kb. The resolution of the alkaline-agarose gel allowed for the analysis of OF up to 8.8 kb in length.
OF Signaling/Collision Analysis-The percentage of the OFs terminated by the signaling and collision mechanism was calculated from integration of the peak areas of OFs as described previously (35). In brief, integration of all peak area above the base line after size corrections yielded the total intensity for all OF products. Percent collision is given by the ratio of total intensity for the peaks at (n ϩ 1) ϫ 1.1 kb to total intensity, whereas percent signaling was calculated from the ratio of the total intensities of the areas at intermediate lengths to the total intensity.
Stoichiometry of Clamp/DNA in the Rolling Circle Replication Reactions-The reactions were performed in complex buffer with a volume of 100 l at 37°C containing 5 nM Neu-trAvidin-blocked Bio/Cy3 TRFII; 100 nM each gp43 (exo-), gp44/62 and Cy5-gp45 (V163C); 200 nM each gp41 and gp59; 4 M gp32; 1 mM ATP; and 100 M each of the other NTPs and dNTPs in the presence (both leading-and lagging-strand syntheses) or in the absence (leading-strand synthesis only) of 200 nM gp61. Conditions of the control reactions absent of replication can be found in the figure legends. Aliquots of the reaction mixture were extracted at 5, 10, and 15 min after the 7-min replisome assembly period and quenched with EDTA. Then, biotin polystyrene beads were incubated with the quenched reaction mixture for 2 min at 37°C to pull down the DNAclamp complex. After removal of unbound proteins by washing the beads with 5 ml of complex buffer, the complex was eluted by treating the beads with 100 l buffer containing 2 mM biotin. To generated calibration curves of the concentration versus fluorescence intensities, standard 5ЈBio/Cy3 TRFII and Cy5-gp45 (V163C) samples were blotted alongside the eluant onto a PVDF membrane using a vacuum manifold device (Bio-Rad). The membrane was scanned on a Typhoon laser scanner (GE Healthcare). The amounts of the TRFII (DNA) and clamp pulled down, and the stoichiometry of gp45 to DNA ([gp45]/ [DNA]) were measured and analyzed by ImageQuant (GE Healthcare).

Size Distribution and Pattern of the OFs Produced in T4 DNA
Replication on pONE-Nick Substrate-The circular pONE-Nick substrate shown in Fig. 1A allows primer synthesis by gp61 at the single primary recognition site (5Ј-GTT-3Ј) when only ATP and CTP are used in the replication reactions. Upon termination of an OF, initiation of the next cycle by using this priming site generates a minimal 1.1-kb fragment, corresponding to the size of the shortest distance between two adjacent utilized priming sites. Delayed initiation of the next OF by skipping the "n " number of priming sites would yield an OF of (n ϩ 1) ϫ 1.1 kb.
As reported elsewhere and as shown in Fig. 1, B and C, OFs are distributed in distinct populations with lengths approximately in multiples of 1.1 kb in the standard reaction on the substrate with excess T4 replisomal proteins (35). The overall size distribution remained largely unchanged in the 15-min reaction time course, as indicated by the narrow error bar.
The 1.1-kb fragments account for 46% of the total fragments in accordance with the previous report for the reaction with 1-fold higher T4 protein levels ( Fig. 1D) (35). This percentage represents the primer utilization efficiency in the standard reaction, assuming that a primer is synthesized each time the priming site is passed.
Effects of the Dissociative T4 Replisomal Proteins on the Size Distribution of the OF Populations-M13mp18 duplex (TRFII) and 70-mer minicircle substrates have been used extensively in investigating the molecular details of lagging-strand synthesis.
Using the two substrates, it was found that levels of dissociative proteins, including gp61, gp44/62, gp45, and gp32, tend to affect the overall length distribution of the OFs in the T4 DNA replication reaction to various extents (13,15).
To prove whether the pONE-Nick substrate would display the expected size shift of OFs as the result of dissociative T4 replisomal protein level changes, we carried out replication reactions on this substrate, titrating the levels of gp44/62 and gp32, respectively.
As shown in Fig. 2A, we titrated the T4 replication reactions on 3 nM pONE-Nick substrate by increasing gp44/62 from 0 -150 nM (left panel) and from 6 to 24 nM (right panel). Except for the reaction excluding gp44/62 that displayed no replication, leading-strand synthesis was not notably affected by the gp44/62 level. In contrast, lagging-strand synthesis showed a clear dependence on the gp44/62 concentration, displaying more lagging-strand synthesis and a shift to shorter OFs as the levels of gp44/62 increased. Fig. 2B shows the quantitative analysis of the OF length distribution as gp44/62 concentration increased. At 24 nM, the majority of the OFs ranged from 3.3 to 5.5 kb, consistent with the results from an earlier study (27). As the amount of gp44/62 increased, shorter OFs (Յ 2.2 kb) gradually became dominant in the size distribution. Percentages of the 1.1-kb OFs increased from Ͻ 10% at 24 nM gp44/62 and reached a plateau at ϳ46% at ϳ90 nM gp44/62 (Fig. 2C). In the next section, we present an analysis of the mechanistic origins of the OFs through signaling and collision.
Next, we examined the effect of gp32 concentration on DNA synthesis and the distribution of OFs (Fig. 3A). As shown, both leading-and lagging-strand syntheses were strongly dependent on the concentration of gp32 in the replication reaction. The increasing amount of total DNA synthesis as the concentration of gp32 increased demonstrates the active role of gp32 in assembling the T4 replisome on DNA and the corresponding increase in the amount of functional replisome formed in the reaction. As the gp32 concentration increased, a slight shift in the OF size distribution from 2.2 kb to 1.1 kb occurred (Fig. 3B), reminiscent of the effect of gp32 on the length of the OFs in replication on the minicircle substrate (15). The percentage of the 1.1-kb OFs showed a moderate increase from 28% to 46% and reached a plateau at around 1.5 M of gp32 (Fig. 3C).
Size Distribution of the OFs Is Independent of gp43(exo-) Concentration-We tested the effect of polymerase concentration on the size distribution of the OFs to understand the mechanism(s) of recycling polymerase in lagging-strand synthesis.
We first performed gp43(exo-) concentration titration experiments at a constant concentration of 100 nM gp44/62, as shown in Fig. 4A. The higher signal intensities and, therefore, more synthesis of both strands observed as the polymerase concentration increased gradually are most likely due to the increased amount of active replisome formed, given that the apparent K m  of gp43 measured in a replication reaction is 40 -80 nM (19). Regardless of the concentration increase of gp43(exo-), the distribution of the OF lengths did not change (Fig. 4B). In the polymerase concentration range tested, the percentage of the 1.1-kb fragment remained constant at ϳ45%, a value comparable with the percentage measured in a standard reaction where the polymerase concentration is 200 nM (Fig. 4C).
We also titrated replication reactions with increasing concentrations of gp43(exo-) but at constant lower (18 nM) gp44/62 levels (Fig. 4D). Under these low gp44/62 conditions, uncoupling between leading-and lagging-strand synthesis was notable, indicated by the loss of the 1.1-kb OFs and the majority of the OFs at lengths of Ͼ 4.4 kb. If gp43(exo-) was recruited from solution to initiate an OF cycle, an increase in the gp43(exo-) level would be expected to shift the distribution to shorter OF lengths. As shown in Fig. 4D, titration of gp43(exo-) from 30 to 120 nM (left panel) or from 6 to 30 nM (right panel) did not appear to affect the size distribution of the OFs.
Collision versus Signaling to Initiate OFs-OF lengths of (n ϩ 1) ϫ 1.1 kb result from the extension of a primer through n ϩ 1 replication cycles and are the consequence of the collision mechanism where replication termination occurs because of the lagging-strand holoenzyme encountering the 5Ј end of the OF synthesized previously. Inspection of the gel in Fig. 1, B and C, shows the accumulation of OFs of intermediate lengths as illustrated in Fig. 5A (data from Fig. 1). We attributed the fraction of these OFs to the operation of a signaling mechanism where replication termination by the lagging-strand holoenzyme occurs prematurely. The percent collision versus signaling was calculated from integration of the peak areas where percent collision is given by the ratio of the total intensity for peaks at (n ϩ 1) ϫ 1.1 kb to the total peak intensity for all areas above the base line, as illustrated in Fig. 5A. In this data set, the % collision is 36.6 Ϯ 1.5%. Signaling is 63.3 Ϯ 1.5% (Fig. 5B). The fraction of (n ϩ 1) ϫ 1.1 kb OF from each mechanism is shown  . Quantitative analysis of the collision versus signaling mechanism for Okazaki fragment termination. A, the Okazaki fragments produced by the signaling or collision mechanism under standard replication conditions. The Okazaki fragments resolved on the alkaline-agarose gel (data from Fig. 1B) at lengths of (n ϩ 1) ϫ 1.1 kb were attributed to operation of the collision mechanism, whereas those at intermediate lengths were signaling products (left panel). Integration of the total peak areas above the corrected base line for size differences yielded the total intensities for all the Okazaki fragments (right panel). Percent collision and percent signaling were given by the ratios of the intensities of the peaks for the Okazaki fragments from each mechanism to the total intensities of all Okazaki fragments. B, percent collision and signaling of the Okazaki fragments produced under standard replication conditions. In this data set, the percent collision and percent signaling are 63.3 Ϯ 1.5% and 36.6 Ϯ 1.5%. C, analysis of the fraction of (n ϩ 1) ϫ 1.1-kb Okazaki fragments from each mechanism. The mean percentage Ϯ S.D. for each Okazaki fragment was calculated from the analyses of the time points at 5, 10, and 15 min. JULY Fig. 5C. The above analysis was carried out on the data for titrations with the clamp loader and the polymerase. For the gp44/62 titration, percent signaling increases from 45% to 64% with increasing levels of the clamp loader with a commensurate decrease in the percent collision (Fig. 6A). For the gp43(exo-) titration, the ratio of collision to signaling (34.2 Ϯ 1.1% to 65.7 Ϯ 1.1%) is unchanged as a function of the polymerase level (Fig. 6B).

Insights into Okazaki Fragment Synthesis by the T4 Replisome
Stoichiometries of the gp45 Clamp in the T4 Replication Reaction-A wealth of information on the molecular details of gp45 loading and association with gp43 in the context of a T4 holoenzyme has been obtained from studies on short forked substrates (12,13,36,37). However, the behavior of gp45 after the departure of gp43 during repetitive OF synthesis remains to be investigated. To this end, we performed pull-down assays of rolling circle DNA replication products and quantified the amount of gp45 associated with the DNA to obtain the stoichiometry of gp45 on the DNA under various reaction conditions. We first tested the ratio of [gp45]/[DNA] in the clamp loading reaction in the presence and absence of gp43(exo-), which serves as a control to determine whether the circular TRFII template allows multiple loading reactions thereby trapping multiple gp45 on the DNA (Fig. 7A). We confirmed that Ͼ 70% of the total DNA was recovered and that nonspecific binding of the labeled gp45 to the biotin polystyrene beads was negligible (supplemental Fig. S1). In the absence of gp43(exo-), a [gp45]/ [DNA] ratio of 0.07 Ϯ 0.03 was observed, suggesting that only a small fraction of gp45 remained bound to the DNA, as shown in Fig. 7A. We next performed a clamp loading reaction in the presence of gp43(exo-), allowing for holoenzyme formation. Following quantification of the pull-down assays, we obtained a higher ratio of 0.52 Ϯ 0.06, in good agreement with the reported ϳ70% active replisome assembly on the same DNA substrate (24). These results suggested that a loaded gp45 species is prone to falling off the DNA but can be stabilized by associating with gp43 in the form of a holoenzyme on the DNA. We also tested whether gp45 would accumulate on DNA as a result of repetitive OF initiation during replication, giving a higher than 1:1 stoichiometry of [gp45]/[DNA]. As shown in Fig. 7B and supplemental Fig. S2, the comparable [gp45]/ [DNA] ratio was attained from the replication reactions in the absence or presence of gp61, allowing for leading-strand synthesis only or both leading-and lagging-strand syntheses, respectively. Reactions omitting gp61 had a 0.6 Ϯ 0.2 [gp45]/ [DNA], whereas reactions with all T4 proteins had a ratio of 0.8 Ϯ 0.2. By fluorescence-imaging the flow stretched, pulleddown DNA products, we also confirmed that the majority of the pulled down DNA templates were extended to 50-to 100-kblong duplexes during the first 5 min of the reaction with both leading-and lagging-strand syntheses, whereas the singlestranded DNA products from the leading strand-only reaction remained coiled under the stretching force (Ͻ 1.5 piconewton) generated by flow (supplemental Fig. S3). Together, these results suggest that no clamp proteins from discontinuous lagging-strand synthesis accumulate on the DNA during T4 replication and that the clamps pulled down in our assay were mainly associated with the leading-strand holoenzyme.

DISCUSSION
Lagging-strand synthesis progresses via the formation of discontinuous OFs derived from sequential steps that feature priming, clamp loading, primer hand-off and holoenzyme assembly, primer extension, and, finally, the termination of a synthetic cycle by polymerase departure. Clearly, the successful production of OFs demands elaborate actions by a number of replisomal proteins. A common strategy in studying the biomolecular dynamics of this elegant process is to analyze the length of the OF products in protein dilution experiments (22,27). The rationale behind this strategy is that dilution of dissociative proteins acts to lengthen the OFs synthesized because of their recruitment rates from solution being concentration-dependent and diffusion-limited. By the same reasoning, varying the concentration of processive proteins should not affect the size of OFs synthesized. With that approach in the T4 system, accessory proteins, including gp44/62, gp45, gp32, and gp61, are dissociative proteins recruited from solution for laggingstrand synthesis, whereas the gp43s on both strands are processive and remain within the replisome once assembled (17,22,24,27).
The use of the 1.1-kb circular replication template, pONE-Nick, has at least two advantages over the M13mp18 duplex and minicircle substrates in the analysis of the size distribution of OFs. First, by limiting the priming site frequency to one per 1.1 kb and thereby precisely controlling the distance between two successive priming sites, replication reactions on this substrate produce OFs with sizes in multiples of 1.1 kb that can be well resolved on an alkaline-agarose gel, and each fragment can be weighted quantitatively. The quantitative analysis of the OF size distribution as the protein level is varied enables improved accuracy in the assessment of the processive and dissociative properties of replisomal proteins that would otherwise be obscured in the qualitative analysis of OF length changes. Sec-ond, the size of an OF reflects its initiation pattern. A 1.1-kb fragment results from the initiation of the next OF by utilizing the priming site directly downstream. Skipping priming sites would yield longer OFs in multiples of 1.1 kb. It is noteworthy that the width of an OF peak on the gel is possibly the consequence of size variation complicated by the two competing mechanisms in terminating an OF cycle, signaling, and collision. Although both mechanisms operate in the T7 and E. coli systems in equal prevalence, the contribution of the collision pathway was estimated to be slightly higher than the signaling pathway from T4 EM data (55% versus 45%) and Monte Carlo simulation on the basis of the rolling circle replication reactions on substrates with a high priming site frequency of ϳ60 bp/site (9,38,39).
The coordination of leading-and lagging-strand syntheses in the T4 system is conditional, depending on the presence of accessory proteins, including gp44/62 and gp32 (22). It is well documented that the levels of gp44/62 affect the lagging-strand incorporation rate and OF lengths and that gp32, via its interactions with many other T4 replication proteins, is needed for proper assembly and functioning of the T4 replisome (15,40,41). In this study, we found that the collision frequency is ca. 36% and that signaling is ca 64% for replication reactions on the pONE-Nick substrate under the standard reaction conditions. The higher observed percent signaling than the reported ϳ45% is likely a result of the low priming site frequency of one site per 1.1 kb. Also, the distribution between the signaling and collision pathways and the OF lengths are sensitive to the level of the clamp loader (Fig. 6A), shifting to increased signaling and shorter OF lengths with increased levels of gp44/62. Increasing levels of gp32 act to increase the level of total DNA synthesis. This collectively agrees with an earlier finding and is consistent with the hypothesis that the short five-oligonucleotide primer may be retained on the template in a clamp-clamp loader-gp32 complex that is associated with polymerase recycling to initiate OF synthesis (15).
We further confirmed that the lagging-strand polymerase is in fact processive and recycles during repetitive lagging-strand initiation in the T4 system. This conclusion is supported by the following experimental results. First, the titration of gp43(exo-) across a 15-fold concentration range does not perturb the overall distribution of the OF lengths. The stable percentage of the 1.1-kb fragments agrees with a priming site utilization efficiency independently of the gp43(exo-) concentration. Second, we ruled out the possibility of recruiting a polymerase from solution to commence a new fragment in the event that laggingstrand polymerase recycling is delayed. When the concentration of gp44/62 was limiting, titration of increasing concentrations of gp43(exo-) did not shift the OFs to shorter lengths, implying that recruitment of a pool polymerase to the T4 replisome for initiating a new cycle did not occur. Collectively, our data from the quantitative analysis of the OF lengths support the polymerase recycling mechanism, and the participation of an exogenous polymerase in lagging-strand synthesis, if extant, is rare, with too small of a probability to be observed in our ensemble experimental setup.
There are observations of apparent polymerase exchange from recent fluorescence studies on the E. coli replisome and the EM data of the T4 replication products (9,31). GFP-labeled Pol III is shown to dynamically release from and be recruited to an active replisome in living cells. The authors proposed that Pol III exchanges during OF synthesis and that the recruitment and release of Pol III occurs via fluctuations in the binding affinity of Pol III to the subunit of the ␥ clamp loader complex. However, a possible alternative explanation is that an exogenous Pol III is recruited for filling in single-stranded DNA gaps between OFs, caused by the signaling mechanism as the replisome travels. The ␤ clamps left behind on the DNA anchor exogenous Pol IIIs at gaps between OFs, and the extra Pol IIIs are spatially indistinguishable from those within the replisome. The analogous interaction between polymerase and the clamp loader is absent in the T4 or T7 system. Further tests are needed to confirm the validity and generality of this exchange mechanism (1,3,42). On the other hand, the EM results of the moving fork structure and protein stoichiometries of the T4 system show that ϳ6% of the extended DNA molecules have two trombone loops or three single-stranded DNA regions flanking two duplex regions at the replication fork, as expected for two lagging-strand polymerases acting simultaneously on two OFs. Intriguingly, of this small fraction of DNA molecules, only ϳ30% have three polymerases within the replisome, whereas more than half have only one or no polymerase (9). Therefore, the observed two-loop T4 fork with three polymerases may be interpreted as an unlikely intermediate state with a very low probability of occurrence (Ͻ2%) during lagging-strand replication.
After confirming the recycling of the lagging-strand polymerase in the T4 system, we focused our attention on the fate of gp45 on the DNA following the departure of the recycled polymerase. We probed the stoichiometry of [gp45]/[DNA] under various conditions. It has been discovered recently that two relatively equal populations of gp45, gp44/62, exist on a short forked DNA substrate blocked on both ends with offrates of 0.107 and 0.0023 per second, with the former applicable to this substrate (43). The low stoichiometry of stably loaded gp45 in the absence of gp43 observed in this study is consistent with our inability to capture a DNA-gp45 complex whose halflife is approximately 10 s. In the presence of the polymerase, the half-life for clamp dissociation is increased to approximately 10 min, enabling capture of the clamp in the pull-down experiment. The observed stoichiometry ratio of [gp45]/[DNA] under the condition where the holoenzyme is present on both strands and the leading strand alone is less than 1. This low value stems from an estimated 70% replisome assembly on this template and from examination of the EM study of the T4 replisome executing rolling circle replication where only approximately 40% of the lagging-strand holoenzyme was found in the replisome (9,24). Consequently, in the case of the leading-strand holoenzyme alone, an estimated ratio is 0.7 compared with the measured ratio of 0.65. For both leading-and lagging-strand holoenzymes, the estimated ratio is 0.7 ϩ 0.4 ϭ 1.1 compared with the measured ratio of 0.8, which supports our conclusion that clamps do not accumulate on the lagging strand after the polymerase is recycled.
Ensemble FRET studies of labeled gp45-gp43-DNA complexes found that gp45 and gp43 depart simultaneously from a short forked DNA (43). The release of gp45 from the polymerase occurs in solution because of the low affinity of gp45 for gp43 in the absence of DNA (44). Thus, within the context of the replisome, the holoenzyme partitions differently to retain the polymerase and lose the clamp protein.
In conclusion, we propose a revised model of the behavior of the lagging-strand holoenzyme components during repetitive OF synthesis in T4 on the basis of the results from this study (Fig. 8). In this model, the lagging-strand holoenzyme synthesizes OFs via the trombone loop intermediate because it is coupled to the leading-strand polymerase. The termination of an OF occurs through either the collision or signaling mechanism. In the collision pathway (Fig. 8, Route 1), the clamp and polymerase of the lagging-strand holoenzyme disengage simultaneously from an OF as a consequence of the collision with the 5Ј end of the previous OF. During this process, the polymerase The collision pathway (Route 1) involves the following steps: 1) release of the lagging-strand holoenzyme from the DNA after colliding with the previous Okazaki fragment and then release of the clamp from the polymerase in the absence of DNA; 2) lagging-strand synthesis is on hold until primer synthesis and handoff to a clamp-clamp loader complex; and 3) the polymerase recycles to the new primer-clamp-clamp loader complex to initiate a new cycle. The signaling pathway (Route 2) involves the following steps: 1) primer synthesis by the primosome at the moving fork and subsequent handoff to the clamp-clamp loader complex; 2) simultaneous release of the lagging-strand holoenzyme from a progressing Okazaki fragment and initiation of a new fragment by holoenzyme reassembly through polymerase recycling to the new primer-clamp-clamp loader complex. In either process, the laggingstrand polymerase would remain within the replisome, and exogenous polymerases from the pool do not participate in Okazaki fragment synthesis. remains within the replisome, whereas the clamp releases from the polymerase. Lagging-strand replication remains idle until the reassembly of the lagging-strand holoenzyme, which occurs through sequential steps, including primer synthesis, primer handoff to the clamp-clamp loader complex, and migration of the same polymerase to the primer-clamp-clamp loader complex at the moving fork.
In the signaling pathway (Fig. 8, Route 2), successful and timely capture of a newly synthesized primer at the fork by the clamp-clamp loader complex serves as the signal for premature termination of OF synthesis, which generates a single-stranded DNA gap between two adjacent OFs. The source of this signal is consistent with the finding that percent signaling, as well as lagging-strand synthesis, is a function of the clamp loader concentration. The termination of an OF by signaling and reassembly of the lagging-strand holoenzyme take place simultaneously as a concerted process, involving disengagement of the laggingstrand holoenzyme from the ongoing OF, causing the collapse of the trombone loop and polymerase recycling to the primerclamp-clamp loader complex. Considering the ϳ12-s half-life of the primer-clamp-clamp loader complex and the replication rate of 250 bp/s by the T4 replisome, multiple primers or perhaps primer-clamp-clamp loader complexes might be present within a trombone loop during an OF cycle (14). From a geometrical standpoint, it is most likely that the newly formed primer-clamp-clamp loader complex at the fork serves as an effective signal and accommodates the recycled polymerase for the lagging-strand holoenzyme reformation. Similar to the collision pathway, a third polymerase or exogenous polymerases would not be involved in OF synthesis, and clamps do not associate with completed OFs in the signaling pathway. We note that this is in sharp contrast to lagging-strand replication reported for the E. coli replisome (32). In that case, a third polymerase has been invoked to participate in lagging-strand synthesis. Furthermore, the degree of signaling is independent of the clamp loader concentration, leading to the proposal that torque in the DNA created by replication causes the recycling of the lagging-strand polymerase (45). Obviously, nature has at least two ways to solve this aspect of the replication problem.